Subarachnoid hemorrhage: risks of aneurysm rupture and delayed cerebral ischemia. Nicolien de Rooij

Transcription

1 Subarachnoid hemorrhage: risks of aneurysm rupture and delayed cerebral ischemia Nicolien de Rooij

2 Cover Robert Kanters, Ridderprint Layout Renate Siebes, Proefschrift.nu Printed by Ridderprint, Ridderkerk ISBN Nicolien de Rooij All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage or retrieval system, without permission in writing from the author. The copyright of the articles that have been accepted for publication or that have already been published, has been transferred to the respective journals.

3 Subarachnoid hemorrhage: risks of aneurysm rupture and delayed cerebral ischemia Subarachnoïdale bloedingen: risico s op aneurysma ruptuur en secundaire ischemie (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. G.J. van der Zwaan, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op dinsdag 26 februari 2013 des middags te 2.30 uur door Nicolien Karen de Rooij geboren op 26 mei 1982 te Gouda

4 Promotor: Prof.dr. G.J.E. Rinkel Co-promotor: Dr. C.J.M. Frijns Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully acknowledged.

5 CONTENTS Chapter 1 General introduction 7 PART I Incidence of subarachnoid hemorrhage and risk of aneurysm rupture Chapter 2 Chapter 3 Incidence of subarachnoid hemorrhage: a systematic review with emphasis on region, age, gender and time trends Configuration of the circle of Willis, direction of flow, and shape of the aneurysm as risk factors for rupture of intracranial aneurysms PART II Delayed cerebral ischemia following subarachnoid hemorrhage Chapter 4 Chapter 5 Chapter 6 Secondary infarction in single or in multiple vascular territories: two different entities following subarachnoid hemorrhage? Delayed cerebral ischemia after subarachnoid hemorrhage: a systematic review of clinical, laboratory and radiological predictors Early prediction of delayed cerebral ischemia after subarachnoid hemorrhage: development and validation of a practical risk chart Chapter 7 General discussion 105 Nederlandse samenvatting 123 Dankwoord (Acknowledgements) 129 About the author 133

6

7 1 General introduction

8 Chapter 1 General introduction Subarachnoid hemorrhage (SAH) from a ruptured intracranial aneurysm accounts for about 5% of all strokes. Because it occurs at a relatively young age and has a high case fatality, in the general population the loss of productive life years from SAH is as large as that from cerebral infarction, the most common type of stroke. 1,2 Important risk factors are a familial preponderance, hypertension, smoking and alcohol abuse. 3 Intracranial aneurysms are present in 3% of the general population, 4 and only a minority will rupture and cause an SAH. 5 Since preventive treatment is associated with considerable risks of complications, 6 knowledge on risk factors for rupture is crucial to determine which aneurysms should undergo preventive treatment or warrant more frequent follow up. Currently few risk factors for rupture have been identified: gender, age, world region, size and site of the aneurysm, and possibly previous rupture of another aneurysm. 5,6 Delayed cerebral ischemia (DCI) is one of the main complications after SAH. Up to know, the exact cause of DCI remains unknown. Although cerebral vasospasm has been indicated as the main cause of DCI, 7 vasospasm and DCI do not always go hand in hand, which shows that vasospasm is not the only or essential factor for the development of DCI DCI is a major contributor to the high case fatality and morbidity of SAH. About 30% of the SAH patients develop DCI. 11 Established predictors of DCI are large amount of subarachnoid blood detected on CT imaging and poor clinical condition on admission, but many others are reported. Despite all research on DCI, it is not possible to predict which SAH patients will develop DCI. OUTLINE OF THE THESIS Part I: Incidence of SAH and the risk of aneurysm rupture Chapter 2 concerns a systemic review on the incidence of SAH, describing the influence of age, gender en region. Furthermore, this review emphasizes the time trend of occurrence of SAH over the past 45 years. Chapter 3 describes new anatomical risk factors that may contribute to the challenging decision making whether preventive treatment is necessary, and whether follow-up is warranted in patients with an intracranial aneurysm. 8

9 General introduction Chapter 1 Part II: Delayed cerebral ischemia after SAH Chapter 4 affects the issue whether DCI may have several patterns, and if these patterns represent pathophysiologically different disease entities or different degrees of severity of the same vascular process. Chapter 5 is a systemic review on easy available predictors of DCI present on admission. In Chapter 6 we analysed and compared multiple predictive models for development of DCI, using the results of Chapter 5 as basis. This last chapter demonstrates a practical risk chart of prediction of DCI immediately on admission at the hospital. REFERENCES 1. Feigin VL, Lawes CM, Bennett DA, et al. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: Johnston SC, Selvin S, Gress DR. The burden, trends, and demographics of mortality from subarachnoid hemorrhage. Neurology 1998; 50: Feigin VL, Rinkel GJE, Lawes CM, et al. Risk factors for subarachnoid hemorrhage. An updated systematic review of epidemiological studies. Stroke 2005; 36: Vlak MH, Algra A, Brandenburg R, Rinkel GJ. Prevalence of unruptured intracranial aneurysms, with emphasis on sex, age, comorbidity, country, and time period: a systematic review and meta-analysis. Lancet Neurol 2011; 10: Wermer MJ, van der Schaaf I, Algra A, Rinkel GJ. Risk of rupture of unruptured intracranial aneurysms in relation to patient and aneurysm characteristics: an updated meta-analysis. Stroke 2007; 38: Wiebers DO, Whisnant JP, Huston J3rd, et al. Unruptured intracranial aneurysms: natural history, clinical outcome, and risks of surgical and endovascular treatment. Lancet 2003; 362: Treggiari-Venzi MM, Suter PM, Romand JA. Review of medical prevention of vasospasm after aneurysmal subarachnoid hemorrhage: a problem of neurointensive care. Neurosurgery 2001; 48: Aralasmak A, Akyuz M, Ozkaynak C, et al. CT angiography and perfusion imaging in patients with subarachnoid hemorrhage: correlation of vasospasm to perfusion abnormality. Neuroradiology 2009; 51: Dankbaar JW, Rijsdijk M, van der Schaaf I, et al. Relationship between vasospasm, cerebral perfusion, and delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Neuroradiology 2009; 51: Rabinstein AA, Friedman JA, Weigand SD, et al. Predictors of cerebral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2004; 35: Roos YB, de Haan RJ, Beenen LF, Groen RJ, Albrecht KW, Vermeulen M. Complications and outcome in patients with aneurysmal subarachnoid haemorrhage: a prospective hospital based cohort study in the Netherlands. J Neurol Neurosurg Psychiatry 2000; 68:

10 Chapter 1 General introduction 12. Adams HP, Jr, Kassell NF, Torner JC, Haley EC, Jr. Predicting cerebral ischemia after aneurysmal subarachnoid hemorrhage: influences of clinical condition, CT results, and antifibrinolytic therapy. A report of the Cooperative Aneurysm Study. Neurology 1987; 37: Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980; 6: van Gijn J, Kerr RS, Rinkel GJ. Subarachnoid haemorrhage. Lancet 2007; 369:

11 PART I Incidence of subarachnoid hemorrhage and risk of aneurysm rupture

12

13 2 Incidence of subarachnoid hemorrhage: a systematic review with emphasis on region, age, gender and time trends Nicolien K. de Rooij, Francisca H. H. Linn, Jacob A. van der Plas, Ale Algra, Gabriël J. E. Rinkel Journal of Neurology, Neurosurgery and Psychiatry 2007; 78:

14 Chapter 2 Incidence of SAH ABSTRACT Objective: To update our 1996 review on the incidence of subarachnoid hemorrhage (SAH) and assess the relation of incidence with region, age, gender and time period. Methods: We searched for studies on the incidence of SAH published until October The overall incidences with corresponding 95% confidence intervals were calculated. We determined the relationship between the incidence of SAH and determinants by means of univariate Poisson regression. Results: We included 51 studies (33 new), describing 58 study populations in 21 countries, observing 45,821,896 person-years. Incidences per 100,000 person-years were 22.7 (95% CI 21.9 to 23.5) in Japan, 19.7 (95% CI 18.1 to 21.3) in Finland, 4.2 (95% CI 3.1 to 5.7) in South and Central America, and 9.1 (95% CI 8.8 to 9.5) in the other regions. With age category 45 to 55 years as the reference, incidence ratios increased from 0.10 (95% CI 0.08 to 0.14) for age groups younger than 25 years to 1.61 (95% CI 1.24 to 2.07) for age groups older than 85 years. The incidence in women was 1.24 (95% CI 1.09 to 1.42) times higher than in men; this gender difference started at age 55 years and increased thereafter. Between 1950 and 2005, the incidence decreased by 0.6% (95% CI 1.3% decrease to 0.1% increase) per year. Conclusions: The overall incidence of SAH is approximately 9 per 100,000 person-years. Rates are higher in Japan and Finland and increase with age. The preponderance of women starts only in the sixth decade. The decline in incidence of SAH over the past 45 years is relatively moderate compared with that for stroke in general. 14

15 Incidence of SAH Chapter 2 INTRODUCTION Subarachnoid hemorrhage (SAH) from a ruptured aneurysm accounts for approximately 5% of all strokes. Because it occurs at a young age and has a high case fatality, the loss of productive life years in the general population from SAH is as large as that from cerebral infarction, the most common type of stroke. 1,2 Important risk factors are a familial preponderance, hypertension, smoking and alcohol abuse. 3 In 1996, we performed a systematic review on the incidence of SAH between 1960 and In that review, the incidence of SAH had remained stable at around 8 per 100,000 person-years over 35 years. An interesting finding was the high incidence in Finland in comparison with other European and American populations. In a small subset of studies, gender specific incidences were given, which indicated a higher incidence in women. Since the publication of that review, many new incidence studies have been reported, including regions that were not represented in the first review. The incidence for stroke in general has declined over the past decade, and this has been attributed to a declining proportion of people who smoke and to better detection and treatment of hypertension. 1 As smoking and hypertension are also risk factors for SAH, a similar decline in the incidence of SAH could be expected. We updated the previous review with new information and assessed regional differences in SAH incidence, as well as differences in incidence according to age, gender and time period. METHODS Methods of literature search, inclusion criteria for studies and diagnostic criteria for SAH were essentially the same as in the previous overview. 4 To update the review, we searched for population based studies on the incidence of SAH by performing a MEDLINE search from 1993 until October (Keywords: stroke or subarachnoid h(a)emorrhage together with epidemiology, population or incidence.) In addition, we searched the reference lists of all relevant publications, searched for related articles given on MEDLINE and checked the citation list of all references found, including those from the previous version of the review. This method of cross checking was continued until no further new studies were found. The list of references thus found was compared with the personal database of references from another author (GJER) to check if references had been missed by the (retrospective) PubMed search (which was not the case). This personal database has been prospectively built by daily search of PubMed over the past years by means of the following terms subarachnoid hemorrhage [All Fields] OR aneurysm [All Fields] OR arteriovenous malformation [All Fields] OR perimesencephalic [All Fields] OR subarachnoid haemorrhage [All Fields] OR aneurysm*. 15

16 Chapter 2 Incidence of SAH Two authors (NKR and JAP) reviewed all eligible studies independently and completed a data extraction form. These forms included items regarding design of the study, study population, case finding and diagnostic criteria of SAH. The inclusion criteria were: (1) prospective design; (2) study population is representative of the population in general; (3) upper age limit for the study not below 75 years and lower age limit not above 25 years; (4) for studies about stroke in general, SAH should be considered as a separate entity; (5) results include or at least allow calculation of the overall crude incidence of SAH; (6) the majority of cases were reviewed by the study investigator; (7) case finding methods include at least involvement of all hospitals in the region, and either involvement of general practitioners or reviewing death certificates during the study period; and (8) diagnostic criteria include at least lumbar puncture or autopsy in the pre-ct era, or in case the proportion of patients investigated with CT was lower than 90%. In the event of disagreement in the data extraction forms, the article was re-read by another author (GJER or FHHL) and discussed until agreement was achieved. Excellent case finding was defined as involvement of all hospitals in the region as well as involvement of general practitioners and reviewing death certificates during the study period. Excellent diagnostics was defined as more than 90% of SAH patients had undergone CT. We used incidence rates relating to the entire population, without adjustment for age or sex. Authors were contacted for missing data on crude incidence of SAH if necessary. To assess geographical differences, we compared studies by region. In addition, we extracted gender and age specific incidence rates for those studies that provided sufficient data. Data analysis For each of the selected studies, the overall incidence was computed if necessary. Ninety-five per cent confidence intervals were calculated with Poisson methods. We determined the relationship of the incidence of SAH with region, age, gender and time period by means of univariate Poisson regression. Incidences by region were calculated with the subset of studies from the specific area. Relationship of incidence of SAH with age and gender was analyzed using demographics of the study populations, and age and gender specific incidences of SAH were calculated with the subset of studies that provided sufficient data. Time trend was analyzed using midyear of the study, taking into account regional differences. Multivariate Poisson regression was used to assess the independent contribution of age, gender and time trend to SAH incidence. To examine the influence of design of the study, we selected a subset of studies with excellent case finding and excellent diagnostic criteria for sensitivity analysis. 16

17 Incidence of SAH Chapter 2 RESULTS Literature search The literature search resulted in 42 new studies (Figure 2.1) Thirty-three were relevant for overall analysis. The remaining nine studies were not included in the overall analyses because only incidences of limited age categories were provided. 6,14 16,19,28,32,34,38 These nine studies were included only for analysis on age specific incidences. Eleven authors were contacted for missing information; in four cases the information was retrieved. 14,15,39,45 Together with the 18 investigations from the previous review, studies were used in the total analysis. As four studies provided incidences for 2, 3 or 4 periods or areas, 17,27,39,55 the number of study periods and study regions that we analyzed was 58, of which 39 were new. The studies covered populations in 21 countries with 45,821,896 person-years of observation. Calculated incidences, case finding methods and diagnostic criteria from all of the included study periods and regions are summarized in Table 2.1. Table 2.2 represents the nine studies describing study populations with limited age categories. Included: 18 studies from previous review (19 study periods) Electronic search: 1,407 studies 212 selected studies for abstract screening 91 studies selected for detailed evaluation full paper 42 studies selected based on detailed evaluation full paper 33 relevant studies for total analysis (39 study periods) 51 studies for total analysis (58 study periods) Excluded: 1,195 studies based on title screening Excluded: 121 studies based on abstract screening Excluded: 49 studies based on data extraction full paper For partial analysis: 9 studies due to restricted study population Figure 2.1 Flowchart of literature search on population-based studies on subarachnoid hemorrhage. 17

18 Chapter 2 Incidence of SAH Table 2.1 Incidence, case-finding methods, and diagnostic criteria of subarachnoid hemorrhage in newly identified study periods and regions* Study population Region Midyear of study Patient years Nr of patients Incidence per 100,000 person years (95% CI) Additional Case-Finding methods % of patients with CT Additional diagnostic criteria Rochester 39 USA , ( ) adhjm 0 AB Rochester 39 USA , ( ) adhjm 0 AB Espoo 17 Finland , ( ) ae 0 B Rochester 39 USA , ( ) adhjm 27 ABD Espoo 17 Finland , ( ) ae 11 B Copenhagen 46 Denmark , ( ) ak 47 ABE Izumo city,27 Japan , ( ) a 99 # ABC Rochester 39 USA , ( ) adhjm 85 ABD Finland 17 Finland , ( ) aeh 60 B Izumo city,26 Japan , ( ) ai 100 # BE Asturias 37 Spain , ( ) b 70 Ahmadi 29 Kuwait , ( ) ab 100 A Novosibirsk 35 Russia , ( ) abehjm 0 ## BC Auckland,31 New Zealand ,890, ( ) ae 82 ABC Belluno 22 Italy , ( ) abefjk 90 AB Sweden north 42 Sweden ,212, ( ) abkh 87 ABC L Aquila 21 Italy , ( ) abcefj 89 AB Shimokita,24 Japan , ( ) ai 100 # AC Izumo City 25 Japan , ( ) ai 98 ABC Malmo 45 Sweden ,674, ( ) abde 89 AB Izumo City,27 Japan , ( ) ai 98 ABC Perth 7 Australia , ( ) abdfj >78 BC Sweden south 40 Sweden ,140, ( ) ai 100 ABC Melbourne 5 Australia , ( ) bdg 91 AB London 11 Great Britain , ( ) abefj 88 AB 18

19 Incidence of SAH Chapter 2 Study population Region Midyear of study Patient years Nr of patients Incidence per 100,000 person years (95% CI) Additional Case-Finding methods % of patients with CT Additional diagnostic criteria Vibo Valentia 20 Italy , ( ) abdejk 96 B Dijon 43 France , ( ) abdeh 96 Valle d' Aosta 44 Italy , ( ) abdej 97 AB Erlangen 10 Germany , ( ) abdejk 96 D Kumamoto,23 Japan ,300,000 2, ( ) bij 100 # AC Martinique 30 Caribbean , ( ) abeijk 93 A Scotland 36 Great Britain , ( ) abhfl 91 B Portugal north 33 Portugal , ( ) abdefhikm 97 B Orebro 41 Sweden , ( ) abdefkm 84 AB Tartu 13 Estonia , ( ) abei 92 B Iqueque 9 Chile , ( ) abdefgh 91 AB Tbilisi 18 Georgia , ( ) aehijl 78 # A Barbados 8 Caribbean , ( ) abcdefh 96 B Oxford 12 Great Britain , ( ) abefk 98 AB * Studies listed in ascending order of midyear of data collection and are additional to those in the previous review. Studies based primary on SAH, in contrast with general stroke studies. Case-finding methods. For inclusion involvement of all hospitals in the region necessary and at least a or b. a = death certificates; b = general practitioners; c = rehabilitation; d = nursing homes; e = regular search; f = review radiology requests; g = media attention (campaign/newspaper); h = outpatient clinics, health centres; i = sudden deaths, very early death; j = emergency, ambulance, on call medical services; k = ICD-codes; l = door-to-door, home visit, social services, phone calls; m = autopsy rapports. # Studies providing the proportion of CT use in SAH patients exclusively, in contrast with % of CT in patients with stroke in general. Studies not providing the exact proportion of patients with CT exclusively, but only proportion of patients investigated with CT, autopsy or MRI. ## CT was available after 1992, and 1992 all patients were diagnosed with lumbar puncture or autopsy. Additional diagnostic criteria, besides CT. For inclusion at least A or B necessary in pre-ct era or when CT percentage was below 90%. A = Lumbar puncture; B = autopsy; C = angiography; D = MRI; E = surgery. Proportion of patients investigated with CT or diagnostic criteria unknown, but inclusion after discussion among authors of this review. 19

20 Chapter 2 Incidence of SAH Table 2.2 Incidence, case-finding methods and diagnostic criteria of subarachnoid hemorrhage in newly identified studies describing study populations with limited age categories* Study population Region Midyear of study Patient years Nr of SAH patients Incidence per 100,000 person years (95% CI) Additional Case-Finding methods % of patients with CT Additional diagnostic criteria Restriction of study population: age (years) Oyabe 28 Japan , ( ) abj > 25 Novosibirsk 34 Russia , ( ) aehjm ## AB 25 to 74 Turku 15 Finland ,249, ( ) abeam AB > 25 FINMONICA,16 Finland ,863, ( ) aeh 84 # ABC 25 to 74 FINSTROKE 14 Finland ,933, ( ) aehk 86 AB 25 to 74 Arcadia 19 Greece , ( ) abeh 82 AB > 20 Manhattan 38 USA , ( ) befikl 99 AC > 20 Innhered 32 Norway , ( ) abdegk 88 BD > 15 ACROSS, 6 Australia, N. Zealand ,916, ( ) aefk 90 # ABC > 15 * Studies listed in ascending order of midyear of data collection and are additional to those in the previous review. Studies based primary on SAH, in contrast with general stroke studies. Case-finding methods. For inclusion involvement of all hospitals in the region necessary and at least a or b. a = death certificates; b = general practitioners; c = rehabilitation; d = nursing homes; e = regular search; f = review radiology requests; g = media attention (campaign/newspaper); h = outpatient clinics, health centres; i = sudden deaths, very early death; j = emergency, ambulance, on call medical services; k = ICD-codes; l = door-to-door, home visit, social services, phone calls; m = autopsy rapports. # Studies providing the proportion of CT use in SAH patients exclusively, in contrast with % of CT in patients with stroke in general. Studies not providing the exact proportion of patients with CT exclusively, but only proportion of patients investigated with CT, autopsy or MRI. ## CT was available after 1992, and 1992 all patients were diagnosed with lumbar puncture or autopsy. Additional diagnostic criteria, besides CT. For inclusion at least A or B necessary in pre-ct era or when CT percentage was below 90%. A = Lumbar puncture; B = autopsy; C = angiography; D = MRI; E = surgery. Proportion of patients investigated with CT or diagnostic criteria unknown, but inclusion after discussion among authors of this review. 20

21 Incidence of SAH Chapter 2 Region There was wide variation in SAH incidence, ranging from 2 to 25 per 100,000 person-years, with most regional incidences between 7 and 13 per 100,000 person-years. We defined all countries other than Japan, Finland and South or Central America as the reference group. Overall incidences were 9.1 (95% CI 8.8 to 9.5) per 100,000 person-years in the reference group (42 studies); 22.7 (95% CI 21.9 to 23.5) in Japan (seven studies); 19.7 (95% CI 18.1 to 21.3) in Finland (six studies); and 4.2 (95% CI 3.1 to 5.7) in South and Central America (three studies) (Figure 2.2). The incidence in Japan was 2.5 (95% CI 2.4 to 2.6) times higher than that of the reference region and in Finland 2.2 (95% CI 2.0 to 2.4) times higher, whereas the incidence in South and Central America was 2.2 (95% CI 1.6 to 2.9) times lower. Age The mean age of the study population was mentioned in 37 studies, and univariate Poisson regression analysis was performed for this subset of studies. In populations with a mean age Region Reference countries South and Central America Finland Japan Incidence Figure 2.2 Incidence of subarachnoid hemorrhage by region. Incidences per 100,000 person-years, with corresponding 95% CI between lines. All countries other than Japan, Finland and South and Central America were pooled in a reference group. Overall incidences were 9.1 (95% CI 8.8 to 9.5) in the reference group (42 studies); 22.7 (95% CI 21.9 to 23.5) in Japan (7 studies); 19.7 (95% CI 18.1 to 21.3) in Finland (6 studies); and 4.2 (95% CI 3.1 to 5.7) in South and Central America (3 studies). Agespecific incidences by region reveal the same trend for Japan and Finland (see text). 21

22 Chapter 2 Incidence of SAH of 35 years, calculated incidence was 8.6 (95% CI 8.0 to 9.2), and for every year of increase in mean age, the incidence was 1.06 times higher (95% CI 1.05 to 1.07). Twenty studies, including the nine studies with only age specified subsets of the population, reported separately on incidences per age group. 5,6,8 10,16,18 20,22,27,36,45,49,50,52,54,57,60,64 The overall incidence of these 20 studies was 13.9 (95% CI 13.3 to 14.5) per 100,000 person-years. In this subset, incidence increased with age: taking age 45 to 55 years as the reference category, incidence ratios increased from 0.10 (95% CI 0.08 to 0.14) for age, 25 years, to 1.61 (95% CI 1.24 to 2.07) for >85 years (Table 2.3). For Japan, incidences per age decade were given in two studies. Based on these two studies, increase in age specific incidence seemed to be steeper in Japan than in other regions, ranging from 0.56 (95% CI 0.18 to 1.75) per 100,000 person years for age, 25 years to 7.96 (95% CI 5.33 to 11.88) for >85 years. For Finland, no age specific incidence per age decade was available for analysis, and for South and Central America, numbers were too small to provide reliable estimates. Age adjusted incidences per 100,000 person-years in Japan varied from 21 (95% CI 18 to 24) to 23 (95% CI 19 to 28), and in Finland from 14 (95% CI 10 to 19) to 30 (95% CI 22 to 40). 17,58 From studies in South and Central America, age adjusted incidence was given in only one study (4; 95% CI 2 to 6 per 100,000 person-years), which was also adjusted for sex. 9 Table 2.3 Incidence of subarachnoid hemorrhage per age category in 20 studies Age (years) Incidence per 100,000 person-years (95% CI) Incidence ratio (95% CI) Ratio women/men (95% CI) < ( ) 0.10 ( ) 1.36 ( ) 25 to ( ) 0.40 ( ) 0.67 ( ) 35 to ( ) 0.52 ( ) 0.65 ( ) 45 to ( ) reference 0.91 ( ) 55 to ( ) 1.27 ( ) 1.15 ( ) 65 to ( ) 1.30 ( ) 1.26 ( ) 75 to ( ) 1.34 ( ) 1.50 ( ) > ( ) 1.61 ( ) 0.84 ( ) 22

23 Incidence of SAH Chapter 2 Gender Gender distribution was provided in 37 studies. Univariate Poisson regression analysis showed that for each additional percent of women, the incidence became 1.07 times higher (95% CI 1.04 to 1.10). Furthermore, 18 studies reported incidences for men and women separately. 5,9,10,17,18,20,22,27,29,35,44,45,47,49,50,52,54,57 The overall incidence in this subset of studies was 10.5 (95% CI 9.9 to 11.2) per 100,000 person-years; the incidence for men was 9.2 (95% CI 8.4 to 10.2) and for women 11.5 (95% CI 10.6 to 12.6). Thus the incidence in women was 1.24 (95% CI 1.09 to 1.42) times higher than in men. Separate women men ratios per region were 1.26 (95% CI 1.03 to 1.52) for the reference region, 1.16 (95% CI 0.95 to 1.42) for Japan, 1.58 (95% CI 1.08 to 2.30) for Finland and 0.89 (95% CI 0.32 to 2.47) for South and Central America. Age and gender In the 37 studies that reported mean age of the study population and gender distribution, mean age and proportion of women were analyzed by multivariate analysis. After adjustment for age, incidence increased by a factor of 1.03 (95% CI 0.99 to 1.06) for each additional percentage point of women in the study population. After adjustment for gender, incidence increased by a factor of 1.06 (95% CI 1.05 to 1.07) for each additional year. Furthermore, incidences were reported separately for women and men by age category in 16 studies. 5,6,9,10,16,18-20,22,27,45,49,50,52,54,57 In this subset of studies, the women men ratio ranged from 0.65 (95% CI 0.51 to 0.82) to 1.50 (95% CI 1.07 to 2.10). In the age group 25 to 45 years, incidence was significantly higher in men than in women, but in the age group 55 to 85 years, incidence was significantly higher in women than in men (Table 2.3, Figure 2.3). Time trend Midyear of the study was analyzed by univariate and multivariate analysis for evaluation of a time trend. Because studies in Japan and Finland were confined to more years, analyses on time trend were performed for the reference region only. During the observation period, incidence decreased by a factor of (95% CI to 1.001) per year in the reference region after adjustment for gender and age. When all 42 studies were analyzed from the reference region without adjustment for age and gender, the rate ratio was (0.997 to 1.004) for year-to-year annual change, thus showing no decrease in incidence. In the subset of studies that reported on study periods after 1990 and that provided exact proportions of patients investigated by CT, the rate ratio for use of CT on reported incidence was (95% CI to 1.002) in the 23

24 Chapter 2 Incidence of SAH Incidence (per ,000 person years) Age categories per 10 years Figure 2.3 Incidence of subarachnoid hemorrhage by age and gender. reference region (n=14), (95% CI to 1.022) in Japan (n=4) and (0.821 to 1.13) in South and Central America (n=3). Thus after 1990 there was no obvious relation between the use of CT and reported incidence of SAH. To exclude the influence of percentage of CT use, separate analyses were performed including only studies after For this subset of 24 studies after 1990 from the reference region, results were essentially the same after adjustment for age and gender (Table 2.4). Sensitivity analysis The criterion for excellent case finding was met by 33 studies (20 new, 13 from the previous review) and the criterion for excellent diagnostics by seven studies (five new, two from the previous review). If we combine both excellent case finding methods and excellent diagnostic Table 2.4 Annual time trend of incidence of subarachnoid hemorrhage in the reference region* Number of studies Incidence ratio (95% CI) Incidence ratio (95% CI) adjusted for gender and age Reference region ( ) ( ) Reference region after ( ) ( ) * All countries other than Japan, Finland and South and Central America. Incidence ratio represents coefficient for year-to-year annual change. 24

25 Incidence of SAH Chapter 2 criteria, none of the studies fulfilled these criteria. Therefore, we were unable to perform a sensitivity analysis with excellent studies. DISCUSSION We found that wide variation exists in the incidence of SAH. The overall incidence of SAH was approximately 9 per 100,000 person-years but varied significantly by region, with doubled rates in Japan and Finland and far lower rates in South and Central America. The incidence was higher in women and increased with age. The gender distribution varied with age. At young ages, incidence was higher in men, while after the age of 55 years, the incidence was higher in women. The incidence of SAH has probably decreased slightly over the past 45 years. Several factors may contribute to the higher incidence in Finland and Japan, but the extent of their contributions remains speculative. In Japan and Finland, a higher risk of rupture of intracranial aneurysms is described. 65 Genetic factors may also play an important role in both Japan and Finland. The relatively older age in Japan may be another explanation. Global statistics report the Japanese as being the oldest population in the world, with a median age of 43 years in However, this older age cannot entirely explain the high incidence, because age specific incidences were also higher in Japan than in the reference population. Another explanation may be better case finding, but case finding in the studies from Japan was not more exhaustive than in other regions. Five of the seven studies did not describe regular contacts with general practitioners, and none mentioned contacting rehabilitation facilities or nursing homes as a case finding method. However, the majority of studies from Japan examined instances of sudden death more extensively than studies from other regions. Most studies from Japan used in addition to autopsy, neuroimaging of patients who had died suddenly or during transportation to the hospital. Probably more patients dying early after SAH were detected by scrutinizing all of these events, which increased the incidence of SAH compared with studies in which such instances of sudden death were not examined in this way. However, sudden death accounts for only 12% of all SAH patients 67 ; more extensive examination of patients dying early may contribute to, but cannot entirely explain, the higher incidence in Japan. The proportion of patients in whom the diagnosis of SAH was confirmed by CT scanning was almost 100% in Japan. However, a large proportion of patients investigated by means of CT does not lead to a higher incidence. In our previous review, we found a higher percentage of CT use to be associated with a lower incidence of SAH and in recent studies we found no relation between the proportion of patients 25

26 Chapter 2 Incidence of SAH investigated by means of CT and the reported incidence. The greater use of neuroimaging in Japan is therefore unlikely to be an explanation for the high incidence rates reported in Japan. Age adjusted incidences were also higher in Finland than in the reference region. In Finnish studies, the proportions of patients in whom the diagnosis of SAH was confirmed by CT were low (varying between 0% and 60%). If we apply the rate ratio for proportion investigated by CT on incidence found in the previous version of the review, and if we assume a hypothetical 100% proportion of patients investigated by CT, the incidence of SAH would be 10.6 (95% CI 8.9 to 12.5) in Finland, which is still higher than the incidence in the reference region. Thus the low proportions of CT in Finnish studies do not entirely explain the higher incidences found. Case finding methods in Finnish studies were not more exhaustive compared with other studies, thereby not increasing the incidence found. Other explanations for the high incidence in Finland include high prevalence of smoking and hypertension, 68 and heavy episodic alcohol abuse. 69 The low incidences in South and Central America can perhaps be explained in part by the relatively young mean age of people in these regions. Reported mean ages in the study populations varied between 25 and 35 years, whereas for the reference population this mean age was 37 years. However, the age adjusted incidence given in one study was also lower than in the reference region. 9 Thus other factors are likely to be involved in the lower incidence in this region. No differences in case finding methods were noted, but access to hospitals in these regions may be less than in other regions. Another explanation might be racial differences, although in so studies the incidence of SAH in black populations was higher in comparison with white populations. 70 In summary, none of these explanations can completely explain the regional differences, and other factors are likely to be involved. The higher incidence of SAH in women was found in the previous version of our review but the age dependent gender difference is a new finding. While previous literature describes a peak incidence in the sixth decade, 71 some recent studies found a continuous increase with age, or an age dependent gender difference. 6 The current review confirms these observations from some individual studies. The reasons for the overall higher incidence in women are not clear, but hormonal factors (including use of hormone replacement therapy) are a possible explanation. 72,73 Our finding that the preponderance of women starts only after the sixth decade further supports this suggestion. Although several studies have reported a statistically significant decline in stroke of approximately 2% per year over the past two decades, 12,74,75 it is still uncertain if the reduction in cardiovascular risk factors has also translated into a reduction in the incidence of SAH. Our study found a 26

27 Incidence of SAH Chapter 2 decrease in incidence of 0.6% per year, which is modest compared with the decline in stroke in general. In our analysis, the influences of region, age, gender and improved diagnostic criteria by CT were taken into account. In our previous review, we found that the apparent decline in the incidence of SAH until 1990 was entirely explained by the increasing proportion of patients investigated by CT. 4 In this update, we found that in studies performed after 1990, the proportion of patients investigated by means of CT was no longer significantly related to incidence in any region. The most likely explanation is that after 1990, almost all hospitalized patients were investigated using CT. Thus the contrast between studies with small proportions investigated by CT (with over reporting of SAH) 76 and studies with large proportions investigated by CT has disappeared. The time trend found in our study is therefore not explained by percentages of CT use for confirmation of diagnosis of SAH. The small magnitude of the decline in incidence of SAH may in part be explained by the stronger influence of genetic factors in SAH than in stroke in general. 77 However, genetic factors explain only 10% of SAH, and most cases are attributed to smoking, hypertension and excessive use of alcohol. 77 Perhaps the reduction in risk factors is more effective in older people (where most stokes in general occur) than in younger people (who are most at risk of SAH), but we have no data to support this hypothesis. It seems contradictory that the incidence of SAH decreased over time, although the overall incidence in our update was higher than the incidence found in our previous review. However, by updating the review, we included five new studies in the reference region published after 1993 presenting data from before These five studies had a combined incidence of 10.4 per 100,000 person-years, which is higher than the overall incidence from the studies that had been included in the previous version of the review. The net result is that the incidence of all studies (including the newly found ones) for the observation period from the previous review ( ) has increased compared with that in the previous review. This effect in part explains the paradox of higher incidence in the current review despite declining incidence over time. Furthermore, we found the decrease in incidence only after adjustment for gender and age. Thus the increased incidence in the updated version of the review may be explained in part by inclusion of study populations with higher ages in the more recent years. The number of population based studies (51) and number of person-years (45,821,896) included in this review was large and therefore overall estimates are precise. Subgroup analyses according to region, age, gender and time trend were based on smaller numbers of studies and personyears. Nevertheless, even for these analyses, CI values were narrow. This current review also included data from additional parts of the world compared with the previous version; only African, South Asian and Chinese populations were not represented. 27

28 Chapter 2 Incidence of SAH Our study shows that the incidence of SAH has declined over the past decades, although to a lesser extent than that of stroke in general. Moreover, incidence continues to increase until older age, is higher in women than in men only after the fifth decade and varies considerably per region. Further studies should address the reasons for the relative moderate decline in incidence of SAH, the higher incidence in women only after the fifth decade and the regional differences in SAH incidence. The answers to these questions will probably provide further clues to the etiology of SAH. REFERENCES 1. Feigin VL, Lawes CM, Bennett DA, et al. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2: Johnston SC, Selvin S, Gress DR. The burden, trends, and demographics of mortality from subarachnoid hemorrhage. Neurology 1998; 50: Feigin VL, Rinkel GJE, Lawes CM, et al. Risk factors for subarachnoid hemorrhage. An updated systematic review of epidemiological studies. Stroke 2005; 36: Linn FHH, Rinkel GJE, Algra A, et al. Incidence of subarachnoid hemorrhage: role of region, year, and rate of computed tomography: a meta-analysis. Stroke 1996; 27: Thrift AG, Dewey HM, Macdonell RA, et al. Incidence of the major stroke subtypes: initial findings from the North East Melbourne Stroke Incidence Study (NEMESIS). Stroke 2001; 32: ACROSS study. Epidemiology of aneurysmal subarachnoid hemorrhage in Australia and New Zealand. Stroke 2000; 31: Jamrozik KD, Broadhurst RJ, LaiN, et al. Trends in the incidence, severity, and shortterm outcome of stroke in Perth, Western Australia. Stroke 1999; 30: Corbin DO, Poddar V, Hennis A, et al. Incidence and case fatality rates of firstever stroke in a black Caribbean population: the Barbados Register of Strokes. Stroke 2004; 35: Lavados PM, Sacks C, Prina L, et al. Incidence, 30-day case-fatality rate, and prognosis of stroke in Iquique, Chile: a 2-year community-based prospective study (PISCIS project). Lancet 2005; 365: Kolominsky-Rabas PL, Sarti C, Heuschmann PU, et al. A prospective communitybased study of stroke in Germany the Erlangen Stroke Project (ESPro): incidence and case fatality at 1, 3, and 12 months. Stroke 1998; 29: Wolfe CD, Rudd AG, Howard R, et al. Incidence and case fatality rates of stroke subtypes in a multiethnic population: the South London Stroke Register. J Neurol Neurosurg Psychiatry 2002; 72: Rothwell PM, Coull AJ, Giles MF, et al. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet 2004; 363:

29 Incidence of SAH Chapter Vibo R, Korv J, Haldre S, et al. First-year results of the third stroke registry in Tartu, Estonia. Cerebrovasc Dis 2004; 18: Sivenius J, Tuomilehto J, Immonen-Raiha P, et al. Continuous 15-year decrease in incidence and mortality of stroke in Finland. The FINSTROKE Study. Stroke 2004; 35: Immonen-Raiha P, Sarti C, Tuomilehto J, et al. Eleven-year trends of stroke in Turku, Finland. Neuroepidemiology 2003; 22: Jakovljevic D, Sivenius J, Sarti C, et al. Socioeconomic inequalities in the incidence, mortality and prognosis of subarachnoid hemorrhage: the FINMONICA Stroke Register. Cerebrovasc Dis 2001; 12: Numminen H, Kotila M, Waltimo O, et al. Declining incidence and mortality rates of stroke in Finland from 1972 to Results of three population-based stroke registers. Stroke 1996; 27: Tsiskaridze A, Djibuti M, van Melle G, et al. Stroke incidence and 30-day casefatality in a suburb of Tbilisi. Results of the first prospective population-based study in Georgia. Stroke 2004; 35: Vemmos KN, Bots ML, Tsibouris PK, et al. Stroke incidence and case fatality in southern Greece. The Arcadia Stroke Registry. Stroke 1999; 30: Di Carlo A, Inzitari D, Galati F, et al. A prospective community-based study of stroke in Southern Italy: The Vibo Valentia Incidence of Stroke Study (VISS). Methodology, incidence and case fatality at 28 days, 3 and 12 months. Cerebrovasc Dis 2003; 16: Carolei A, Marini C, Di Napoli M, et al. High stroke incidence in the prospective community-based L Aquila registry ( ). First year s results. Stroke 1997; 28: Lauria G, Gentile M, Fassetta G, et al. Incidence and prognosis of stroke in the Belluno province, Italy. First-year results of a community-based study. Stroke 1995; 26: Hamada J, Morioka M, Yano S, et al. Incidence and early prognosis of aneurysmal subarachnoid hemorrhage in Kumamoto Prefecture, Japan. Neurosurgery 2004; 54: Ohkuma H, Fujita S, Suzuki S. Incidence of aneurysmal subarachnoid hemorrhage in shimokita, Japan, from 1989 to Stroke 2002; 33: Inagawa T, Takechi A, Yahara K, et al. Primary intracerebral and aneurysmal subarachnoid hemorrhage in Izumo City, Japan. Part 1: Incidence and seasonal and diurnal variations. J Neurosurg 2000; 93: Inagawa T, Tokuda Y, Ohbayashi N, et al. Study of aneurysmal subarachnoid hemorrhage in Izumo City, Japan. Stroke 1995; 26: Inagawa T. Trends in incidence and case fatality rates of aneurysmal subarachnoid hemorrhage in Izum City, Japan, between and Stroke 2001; 32: Morikawa Y, Nakagawa H, Naruse Y, et al. Trends in stroke incidence and acute case fatality in a Japanese rural area : the Oyabe study. Stroke 2000; 31: Abdul-Ghaffar NU, el Sonbaty MR, el-din Abdul-Baky MS, et al. Stroke in Kuwait: a three-year prospective study. Neuroepidemiology 1997; 16:

30 Chapter 2 Incidence of SAH 30. Smadja D, Cabre P, May F, et al. ERMANCIA: Epidemiology of Stroke in Martinique, French West Indies: Part I: methodology, incidence, and 30-day case fatality rate. Stroke 2001; 32: Truelsen T, Bonita R, Duncan J, et al. Changes in subarachnoid hemorrhage mortality, incidence, and case fatality in New Zealand between and Stroke 1998; 29: Ellekjaer H, Holmen J, Indredavik B, et al. Epidemiology of stroke in Innherred, Norway, 1994 to Incidence and 30-day case-fatality rate. Stroke 1997; 28: Correia M, Silva MR, Matos I, et al. Prospective community-based study of stroke in Northern Portugal: incidence and case fatality in rural and urban populations. Stroke 2004; 35: Feigin VL, Nikitin YP, Bots ML, et al. A population-based study of the associations of stroke occurrence with weather parameters in Siberia, Russia ( ). Eur J Neurol 2000; 7: Feigin VL, Wiebers DO, Nikitin YP, et al. Stroke epidemiology in Novosibirsk, Russia: a population based study. Mayo Clin Proc 1995; 70: Syme PD, Byrne AW, Chen R, et al. Community-based stroke incidence in a Scottish population: the Scottish Borders Stroke Study. Stroke 2005; 36: Caicoya M, Rodriguez T, Lasheras C, et al. Stroke incidence in Asturias, Rev Neurol 1996; 24: Sacco RL, Boden-Albala B, Gan R, et al. Stroke incidence among white, black, and Hispanic residents of an urban community: the Northern Manhattan Stroke Study. Am J Epidemiol 1998; 147: Brown RD, Whisnant JP, Sicks JRD, et al. Stroke incidence, prevalence and survival. Secular trends in Rochester, Minnestota, through Stroke 1996; 27: Nilsson OG, Lindgren A, Sta hl N, et al. Incidence of intracerebral and subarachnoid haemorrhage in southern Sweden. J Neurol Neurosurg Psychiatry 2000; 69: Appelros P, Nydevik I, Seiger A, et al. High incidence rates of stroke in Orebro, Sweden: Further support for regional incidence differences within Scandinavia. Cerebrovasc Dis 2002; 14: Stegmayr B, Eriksson M, Asplund K. Declining mortality from subarachnoid hemorrhage. Changes in incidence and case fatality from 1985 through Stroke 2004; 35: Wolfe CD, Giroud M, Kolominsky-Rabas P, et al. Variations in stroke incidence and survival in 3 areas of Europe. European Registries of Stroke (EROS) Collaboration. Stroke 2000; 31: D AlessandroG, Bottacchi E, Di Giovanni M, et al. Temporal trends of stroke in Valle d Aosta, Italy. Incidence and 30-day fatality rates. Neurol Sci 2000; 21: Khan FA, Engstrom G, Jerntorp I, et al. Seasonal patterns of incidence and case fatality of stroke in Malmo, Sweden: the STROMA study. Neuroepidemiology 2005; 24: Truelsen T,Gronbaek M, Schnohr P, et al. Stroke case fatality in Denmark from1977 to 1992: the Copenhagen City Heart Study. Neuroepidemiology 2002; 21:

31 Incidence of SAH Chapter Bamford JM, Sandercock PAG, Dennis MS, et al. A prospective study of acute cerebrovascular disease in the community: the Oxfordshire community stroke project Incidence, case fatality rates and overall outcome at one year of cerebral infarction, primary intracerebral and subarachnoid haemorrhage. J Neurol Neurosurg Psychiatry 1990; 53: Longstreth WT Jr, Nelson LM, Koepsell TD, et al. Clinical course of spontaneous subarachnoid hemorrhage: a population-based study in King County, Washington. Neurology 1993; 43: Jerntrop P, BerglundG. Stroke registry inmalmo, Sweden. Stroke 1992; 23: Anderson CS, Jamrozik KD, Burvill PW, et al. Determining the incidence of different subtypes of stroke: results from the Perth community stroke study Med J Aust 1993; 158: Bonita R, Thomson S. Subarachnoid hemorrhage: epidemiology, diagnosis, management and outcome. Stroke 1985; 16: Herman B, Leyten AC, van Luijk JH, et al. Epidemiology of stroke in Tilburg, the Netherlands. The population-based stroke incidence register: 2. Incidence, initial clinical picture and medical care, and three-week case fatality. Stroke 1982; 13: Ricci S, Celani MG, La Rosa F, et al. A community-based study of incidence, risk factors and outcome of transient ischaemic attacks in Umbria, Italy: the SEPIVAC study. J Neurol 1991; 238: Tanaka H, Ueda Y, Date C. Incidence of stroke in Shibata, Japan: Stroke 1981;12: Terent A. Increasing incidence of stroke among Swedish women. Stroke 1988; 19: Sarti C, Tuomilehto J, Salomaa V, et al. Epidemiology of subarachnoid hemorrhage in Finland from 1983 to Stroke 1991; 22: Giroud M, Milan C, Beuriat P, et al. Incidence and survival rates during a twoyear period of intracerebral and subarachnoid haemorrhages, cortical infarcts, lacunes and transient ischaemic attacks. The Stroke Registry of Dijon: Int J Epidemiol 1991; 20: Sivenius J, Heinonen OP, Pyörälä K, et al. The incidence of stroke in the Kuopio area of east Finland. Stroke 1985; 16: Aho K, Fogelhom R. Incidence and early prognosis of stroke in Espoo-Kauniainen area; Finland in Stroke 1974; 5: Hansen BS, Marquardsen J. Incidence of stroke in Frederiksberg, Denmark. Stroke 1977; 8: Gross CR, Kase CS, Mohr JP, et al. Stroke in South Alabama: incidence and diagnostic features a population based study. Stroke 1984; 15: Norrving B, Lo wenheim P. Epidemiology of stroke in Lund-Orup, Sweden, Incidence of first stroke and age-related changes in subtypes. Acta Neurol Scand 1988; 78: Epstein L, Rishpon S, Bental E, et al. Incidence, mortality, and case-fatality rate of stroke in northern Israel. Stroke 1989; 20: Jorgensen HS, Plesner AM, Hubbe P, et al. Marked increase of stroke incidence in men between 1972 and 1990 in Frederiksberg, Denmark. Stroke 1992; 23:

32 Chapter 2 Incidence of SAH 65. Wermer MJ, van der Schaaf IC, Algra A, et al. Risk of rupture of unruptured intracranial aneurysms in relation to patient and aneurysm characteristics: a meta-analysis. Stroke 2007; 38: United Nations Populations Division, DESA. World Population Prospects: The 2006 Revision population ageing. publications/wpp2006/wpp2006_ageing.pdf (accessed 4 october 2007). 67. Huang J, van Gelder JM. The probability of sudden death from rupture of intracranial aneurysms: a metaanalysis. Neurosurgery 2002; 51: Stegmayr B, Asplund K, Kuulasmaa K, et al. Stroke incidence and mortality correlated to stroke risk factors in the WHO MONICA Project. An ecological study of 18 populations. Stroke 1997; 28: Makela P, Fonager K, Hibell B, et al. Episodic heavy drinking in four Nordic countries: a comparative survey. Addiction 2001; 96: Kissela B, Schneider A, Kleindorfer D, et al. Stroke in a biracial population: the excess burden of stroke among blacks. Stroke 2004; 35: van Gijn J, Rinkel GJE. Subarachnoid haemorrhage: diagnosis, causes and management. Brain 2001; 124: Longstreth WT, Nelson LM, Koepsell TD, et al. Subarachnoid hemorrhage and hormonal factors in women. A population-based case-control study. Ann Intern Med 1994; 121: Mhurchu CN, Anderson CS, Jamrozik KD, ACROSS study, et al. Hormonal factors and risk of aneurysmal subarachnoid hemorrhage. An international population-based, case-control study. Stroke 2001; 32: Vibo R, Korv J, Roose M. The Third Stroke Registry in Tartu, Estonia: decline of stroke incidence and 28-day case-fatality rate since Stroke 2005; 36: Pajunen P, Paakkonen R, Hamalainen H, et al. Trends in fatal and nonfatal strokes among persons aged 35 to > or = 85 years during in Finland. Stroke 2005; 36: van Gijn J, van Dongen KJ. Computed tomography in the diagnosis of subarachnoid haemorrhage and ruptured aneurysm. Clin Neurol Neurosurg 1980; 82: Ruigrok YM, Buskens E, Rinkel GJE. Attributable risk of common and rare determinants of subarachnoid hemorrhage. Stroke 2001; 32:

33 3 Configuration of the circle of Willis, direction of flow, and shape of the aneurysm as risk factors for rupture of intracranial aneurysms Nicolien K. de Rooij, Birgitta K. Velthuis, Ale Algra, Gabriël J. E. Rinkel Journal of Neurology 2009; 256;

34 Chapter 3 Anatomical risk factors for rupture of aneurysms ABSTRACT Objective: Improved knowledge on risk factors for rupture of intracranial aneurysms may lead to more tailored aneurysm management. We studied whether configuration of the circle of Willis, direction of flow towards the aneurysm, and shape of the aneurysm are risk factors for rupture. Methods: We reviewed CT angiograms of 126 patients with 75 ruptured and 75 unruptured aneurysms, matched for site of the aneurysm, gender and age of the patient, and year of CT angiogram. For the characteristics studied, we calculated odd ratios (ORs) with corresponding 95% confidence intervals (CIs) for risk of rupture. Configuration of the circle of Willis (incompleteness, asymmetry or dominance) was analyzed on a per site basis. Non-spherical shape was subdivided into elliptical (oval and oblong) and multilobed. In additional analyses, we adjusted for size by means of multivariable logistic regression. Results: Flow straight into the aneurysm (OR 2.0; 95% CI 1.0 to 4.1) and non-spherical shape (OR 2.8; 95% CI 1.5 to 5.5) were associated with rupture. Both elliptical shape, with increasing ORs for oval (OR 1.8; 95% CI 0.8 to 4.0) to oblong shape (OR 6.2; 95% CI 1.9 to 21), and multilobed shape (OR 4.1; 95% CI 1.2 to 14) were associated with rupture. These ORs decreased after adjustment for size. Configuration of the circle of Willis was not associated with a strong risk of rupture; moderate risk could not be excluded. Conclusion: Direction of flow into the aneurysm and nonspherical (both elliptical and multilobed) shape may contribute to the risk of rupture, but are related to aneurysm size and may warrant more frequent follow-up. 34

35 Anatomical risk factors for rupture of aneurysms Chapter 3 INTRODUCTION Intracranial aneurysms are present in 2% of the general population, 1 but only a minority will rupture and cause a subarachnoid hemorrhage (SAH). 2 Since preventive treatment is associated with considerable risks of complications, 3 knowledge on risk factors for rupture is crucial to determine which aneurysms should undergo preventive treatment or warrant more frequent follow-up. Currently few risk factors for rupture have been identified. Characteristics associated with increased risk of rupture are female gender, age, world region, size and site of the aneurysm, and possibly previous rupture of another aneurysm. 1-3 Asymmetry and incompleteness of the circle of Willis could be risk factors for rupture, because these variations of the anatomy may lead to increased flow at the contralateral site. 4 Asymmetry of the proximal segment of the anterior cerebral artery (ACA1) is more often found in patients with ruptured aneurysms of the anterior communicating artery (AcomA) than in patients with ruptured aneurysms at another site. 5 Incompleteness of ACA1 has also been associated with unruptured AcomA aneurysms. 6 However, the relationship between configuration of the circle of Willis and risk of rupture of an aneurysm is unknown. Direction of flow has been described as possible hemodynamic risk factor for formation of aneurysms, 7 and might be a risk factor for rupture as well. The risk of rupture may be increased in aneurysms with multilobed or elliptical shape, because the distribution of wall tension is more heterogeneous compared to spherical shape. We studied whether configuration of the circle of Willis, direction of flow into the aneurysm, and shape of the aneurysm are risk factors for rupture. METHODS Study design and patient selection At the University Medical Center Utrecht, CT angiography (CTA) is routinely performed since 1995 in all patients in whom conventional CT confirms the presence of SAH. Screening patients for unruptured intracerebral aneurysms is mostly done with MR angiography (MRA), CTA usually being reserved for patients with a preference for CT, claustrophobia or other contraindication for MRA such as a pacemaker. Data from all patients admitted with SAH and all patients screened for aneurysms are entered prospectively into a database. 35

36 Chapter 3 Anatomical risk factors for rupture of aneurysms We reviewed CTAs from patients with ruptured intracranial aneurysms and patients with unruptured aneurysms admitted between January 2000 and January Patients with unruptured aneurysms were retrieved from two sources: first, patients with SAH and multiple aneurysms from the SAH database, and second patients with aneurysms detected at screening. We included only patients with aneurysms of the AcomA, internal carotid artery (ICA) bifurcation, middle cerebral artery (MCA), posterior communicating artery (PcomA) and basilar artery (BA). In case patients had more than one unruptured aneurysm, the patient was included for each unruptured aneurysms separately. In patients with SAH and multiple aneurysms, the ruptured aneurysm was identified on the basis of the pattern of hemorrhage on conventional CT scan, or data from the operation. Patients with multiple aneurysms and an inconclusive pattern of hemorrhage on CT and endovascular treatment were excluded from the analysis. We also excluded patients with CTA of poor quality. For each unruptured aneurysm we retrieved a patient with a ruptured aneurysm from the SAH database as comparison. We matched for site of the aneurysm, gender and age of the patient, and year of CTA. Site of the aneurysm was the initial matching factor; if no site match was found, the patient was excluded. Thereafter, matching was done in the order of the factors listed above. For matching according to age, the patient with least age difference was chosen, with preference for a younger age for patients with ruptured aneurysm, to reduce the effect of age as risk factor. In total we retrieved data from 126 patients with 75 ruptured and 75 unruptured aneurysms. Sixty-two patients with 75 unruptured aneurysms were included, of which 49 aneurysms (38 patients) were from the SAH database and 26 aneurysms (24 patients) from the familial screening database. Eleven patients had two unruptured aneurysms and one had three. The match set with 75 ruptured aneurysms originated from 75 patients, of whom 11 patients also had an unruptured aneurysm and were included in the patient group with unruptured aneurysms as well. We used both a single slice (Tomoscan AVE) and 16-multislice (MX 8000 IDT) CT scanner of Philips Medical Systems. The gantry was angled to the orbitomeatal line starting just above the posterior arch of C-1 with a 160 mm field of view and 512 matrix. For the single slice scanner we used the following protocol: 60 1-second rotations of 1 mm collimation, pitch 1,1 mm slice thickness with a reconstruction interval of 0.5 mm, and 140 kv/125 ma. For the 16-multisclice CTA we used mm collimation, pitch 0.938, rotation time 0.75 s, slice thickness 1 mm with a reconstruction interval 0.5 mm and 120 kv/ 200 mas. A testbolus of 15 ml nonionic contrast (Iopromide, Ultravist [300 mg iodine/ml], Schering) was given to determine the scan delay. Using a power injector (CT 9000 Digital Injection System; Liebel Flarsheim, Cincinnati, 36

37 Anatomical risk factors for rupture of aneurysms Chapter 3 OH), 90 ml of nonionic contrast material was injected intravenously at a rate of 3 ml/s for the single slice scanner and 70 ml contrast for the 16-slice scanner (50 ml at 5 ml/s and 20 ml at 3 ml/s). The CT angiographic source image data were transferred to an offline computer workstation (Easy Vision and Extended Brilliance Workspace; Philips Medical Systems) for interactive viewing and postprocessing. Maximum-intensity projection images were created using a 2-cm-thick slab that was angled and rotated separately to the carotid artery circulation and the vertebrobasilar artery circulation according to a set protocol. Data extraction The following clinical features were retrieved from the SAH database: gender, age, history of hypertension, smoking status, alcohol use, family history, and history of coronary artery disease or stroke other than SAH. All CTAs were reviewed on five items (see below) by the same observer (NKR), who was supervised by an experienced neuroradiologist (BKV). Definitions of variables Configuration of the circle of Willis The following seven vessels were reviewed: AcomA, both ACA1s, both PcomAs, and both proximal segments of the posterior cerebral artery (PCA1s). Diameters of 0.1 mm could be measured on CTA, but subdivision according to presence or absence of arteries, and spherical, oval or oblonged shape of the aneurysm was not based on such small differences in diameter. Vessels were assessed as present when the diameter was 1 mm, and as hypoplastic or absent when <1 mm or not visible. If the diameter was measured between 0.8 and 1.0 mm, supervision by an experienced neuroradiologist (BKV) was done. Incompleteness of the A1 segments of the ACA was defined as one of ACA1s being hypoplastic or absent. Asymmetry of the A1 segments of the ACA was defined as difference in diameter between both ACA1s of >33%, which includes patients with an incomplete A1 segments of the ACA. Incompleteness of the proximal segment of the PCA or PcomA was defined as respectively one of the PCA1s or PcomAs being hypoplastic or absent. Dominance of the posterior circulation was defined as fetal type (when the diameter of the PcomA was >10% larger than that of the PCA1, including incomplete PCA1) or normal type (diameter PCA1 >10% larger than PcomA, including incomplete PcomA). 10 In case 10% difference was measured, no dominance was defined. 37

38 Chapter 3 Anatomical risk factors for rupture of aneurysms Flow direction into the aneurysm For assessment of the direction of the blood flow into the aneurysm two angles from the parent vessel were compared: the angle with the aneurysm (A) and the angle with the main branching vessel (B) (Figures 3.1a and b). The angle most approaching 180 is the pathway for dominant blood flow, where the bloodstream has the least resistance. The angles were measured between lines in the center of the parent and branching vessels. The lines were interactively placed on maximum intensity projection (MIP) images, which were rotated to make certain that lines were always exactly in the center of a vessel in all directions. Flow into the aneurysm was defined as curved when the angle with the aneurysm (A) was sharper than the angle with the main vessel (B), and as straight when the angle with the main vessel (B) was sharper than the angle with the aneurysm (A). In case these two angles were similar (less than 5 difference), flow was considered equivalent. Shape of the aneurysm The shape of the aneurysm was defined as spherical when the maximal width of the aneurysm was similar to its length (>80 to 100%) (Figure 3.1c). Shapes other than spherical were defined as non-spherical, with subdivision into elliptical or multilobed shape. Elliptical shape was divided into oval (width between 50 and 80% of length) and oblong (width less than 50% of length). Aneurysms with two or more lobes were assessed separately and defined as multilobed. Size of the aneurysm Size was defined by maximal length and divided into 5 groups: a) less than 2 mm; b) 2 to 5 mm; c) 6 to 10 mm; d) 11 to 15 mm; and e) more than 15 mm. Analysis For each of the possible risk factors for rupture, we calculated odd ratios (ORs) with corresponding 95% confidence intervals (CIs). The configuration of the circle of Willis was analyzed for relevant sites of aneurysms only. Thus, we analyzed the configuration of the A1 segments of the ACA for AcomA aneurysms, and the configuration of posterior circulation for PcomA (on the ipsilateral side of the aneurysm only) and BA aneurysms. Because shape was considered a categorical variable, analyses were performed according to ascending ellipticity (oval and oblong) and lobulation as well. In the primary analyses we considered size to be a result of the risk factor (i.e. configuration of the circle of Willis, flow direction and shape) studied. In additional analyses we considered size as a potential confounder, and adjusted for size in multivariable logistic regression. 38

39 Anatomical risk factors for rupture of aneurysms Chapter 3 (A) (B) Spherical Oval Oblong Multilobed Width >80% Width 50 80% Width <50% of Length of Length of Length Elliptical (C) Non-spherical Figure 3.1 (A) Flow direction into the aneurysm: Two angles from the parent vessel were compared: A = the angle with the aneurysm; B = the angle with the main branching vessel. (B) CTA of two matched patients with an MCA aneurysm. Upper: unruptured aneurysm in which the angle with the aneurysm is sharper compared with the angle with the main branch. Lower: ruptured aneurysm in which the angle with the aneurysm is larger compared with the angle with the main branch. (C) Shape of the aneurysm. 39

40 Chapter 3 Anatomical risk factors for rupture of aneurysms RESULTS Baseline characteristics of all patients with 150 aneurysms are listed in Tables 3.1 and 3.2. All patients were matched for site of the aneurysm, and therefore none of the patients was excluded. Matching for gender was successful for all patients but one. For 56 of the 75 pairs (75%), the age difference between a case and control was less than 4 years and for 58 of the 75 pairs (77%) difference for date of CTA was less than 2 years. Sizes of the aneurysms were larger in the group of ruptured aneurysms. Studied risk factors Incompleteness of the A1 segments of the ACA was present in 7 of 22 (32%) aneurysms of the AcomA, and in 9 of 128 (7%) aneurysms at other sites. Although the prevalence of incompleteness was higher in AcomA aneurysms in comparison with other sites, its presence was not associated with a higher risk of rupture of AcomA aneurysms (OR 0.7; 95% CI 0.1 to 4.0; Table 3.3). Asymmetry of the A1 segments of the ACA showed similar results. Fetal type (PcomA > PCA1) was present in 7 of 20 (35%) aneurysms of the PcomA, and in 24 of 130 (18%) aneurysms at other sites. The OR for fetal type and rupture of PcomA aneurysms was 1.6 (95% CI 0.2 to 9.9; Table 3.3). The OR for normal type (PCA > PcomA) and rupture was 0.6 (95% CI 0.1 to 4.1). Incompleteness, asymmetry or dominance of the circle of Willis at any of the locations studied was not associated with rupture on a per site basis (Table 3.3). Because in these analyses the number of aneurysms per site is relatively small, corresponding 95% CIs were broad. Table 3.1 Baseline characteristics of patients with intracranial aneurysms Index Ruptured (n=75) Unruptured (n=75) Women 57 (76%) 58 (77%) Mean age in years (range) 52 (18 79) 54 (26 81) Hypertension (n=141 * ) 16 (22%) 21 (31%) History coronary disease (n=137 * ) 1 (1%) 7 (11%) History stroke other than SAH (n=141 * ) 4 (5%) 6 (9%) Current smoking (n=108 * ) 39 (65%) 32 (67%) Excessive alcohol abuse (n=96 * ) 10 (18%) 6 (15%) Familial aneurysm (n=116 * ) 5 (9%) 31 (50%) * n<150 due to unknown data Matching factors Difference in family history is due to family screening for unruptured aneurysms 40

41 Anatomical risk factors for rupture of aneurysms Chapter 3 Table 3.2 Site and size of the aneurysm of patients with intracranial aneurysms Site aneurysm Size aneurysm (mm) Ruptured (n=75) Unruptured (n=75) Anterior communicating artery All sizes <2 2 to 5 6 to (15%) (15%) Internal carotid artery bifurcation All sizes <2 2 to 5 6 to (7%) (7%) Middle cerebral artery All sizes <2 2 to 5 6 to (56%) (56%) Posterior communicating artery All sizes <2 2 to 5 6 to (13%) (13%) Basilar artery All sizes <2 2 to 5 6 to (9%) (9%) Matching factors Flow straight into the aneurysm was associated with ruptured aneurysms (OR 2.0; 95% CI 1.0 to 4.1; Table 3.4). Flow was equivalent in six patients. Separate analyses by site all revealed ORs larger than 1.0. In additional analyses the overall OR decreased to 1.2 (95% CI 0.5 to 2.7) after adjustment for size. Analyses of non-spherical shape as risk factor are summarized in Table 3.5. Overall, nonspherical shape was associated with ruptured aneurysms (OR 2.8; 95% CI 1.5 to 5.5). After adjustment for size this OR decreased to 2.0 (95% CI 1.0 to 4.3). Analyses for ascending ellipticity showed an increasing OR for rupture from 1.8 (95% CI 0.8 to 4.0) for oval shape to 6.2 (95% CI 1.9 to 21) for oblong shape. After adjustment for size these ORs were 1.9 (95% CI 0.8 to 4.5) for oval and 2.9 (95% CI 0.8 to 11) for oblong shape. Multilobed shape was also associated with ruptured aneurysms when compared with single-lobed aneurysms (OR 2.7; 95% CI 0.8 to 9.1). This association was stronger when compared with spherical aneurysms only (OR 4.1; 95% CI 1.2 to 14). After adjustment for size this latter OR decreased to 1.7 (95% CI 0.4 to 6.7). 41

42 Chapter 3 Anatomical risk factors for rupture of aneurysms Table 3.3 Configuration of the circle of Willis in patients with ruptured and patients with unruptured aneurysms on per site basis Ruptured Unruptured OR (95% CI) Acom (n=22) (n=11) (n=11) Incompleteness of A1 segments of ACA 1 3 (27%) 4 (37%) 0.7 ( ) Asymmetry of A1 segments of ACA 2 7 (64%) 8 (73%) 0.7 ( ) BA (n=14) (n=7) (n=7) Incompleteness P1 segments of PCA 3 0 (0%) 1 (17%) not estimable Asymmetry of P1 segments of PCA 4 0 (0%) 2 (29%) not estimable Pcom (n=20) Incompleteness PcomA ipsilateral 5 P1 segment of PCA > PcomA ipsilateral 6 (n=10) 6 (60%) 6 (60%) (n=10) 7 (70%) 7 (70%) PcomA > P1 segment of PCA ipsilateral 7 4 (40%) 3 (30%) 1 One ACA1 segment is absent or hypoplastic 2 Difference in diameter between both ACA1s of >33% 3 One PCA1 segment is missing or hypoplastic 4 Dominance of PCA1 or PcomA is different left and right from the BA aneurysm 5 PcomA ipsilateral to the aneurysm is absent or hypoplastic 6 Normal type of dominant PCA1 ipsilateral to the aneurysm 7 Fetal type of dominant PcomA ipsilateral to the aneurysm 0.6 ( ) 0.6 ( ) 1.6 ( ) Table 3.4 Flow straight into the aneurysm analyzed in patients with ruptured and patients with unruptured aneurysms* Site Ruptured (n=71) straight flow Unruptured (n=73) straight flow OR (95% CI) Straight versus curved PcomA (n=20) 4 (40%) 2 (20%) 2.7 ( ) ICA bif (n=10) 3 (60%) 3 (60%) 1.0 ( ) MCA (n= 82) 35 (85%) 27 (66%) 3.0 ( ) AcomA (n=18) 5 (63%) 6 (60%) 1.1 ( ) BA (n= 14) 7 (100%) 7 (100%) not estimable Total (n=144) 54 (76%) 45 (62%) 2.0 ( ) * Direction of flow into the aneurysm as a possible risk factor is explained in text and Figure 3.1 n<150 because flow was equivocal in six patients 42

43 Anatomical risk factors for rupture of aneurysms Chapter 3 Table 3.5 Non-spherical shape of the aneurysm in patients with ruptured and patients with unruptured aneurysms Shape Ruptured n=75 Unruptured n=75 OR (95% CI) for different non-spherical shapes (1) Spherical 29 (39%) 48 (64%) Reference (2) Total non-spherical* 46 (61%) 27 (36%) 2.8 ( ) (2a) Elliptical 36 (48%) 23 (31%) 2.6 ( ) (2a) I: Oval 21 (28%) 19 (25%) 1.8 ( ) (2a) II: Oblong 15 (20%) 4 (5%) 6.2 (1.9 21) (2b) Multilobed 10 (13%) 4 (5%) 4.1 (1.2 14) * Total non-spherical shape = elliptical and multilobed shape together Elliptical shape = oval and oblong shape together DISCUSSION Flow straight into the aneurysm and non-spherical shape (including both elliptical and multilobed) are more commonly observed in ruptured aneurysms compared with unruptured aneurysms. These variables may be associated with an increased risk of rupture, but are related to aneurysm size. The finding that non-spherical shape is a contributing risk factor for rupture is further strengthened by the analysis showing an increasing risk with increasing deviation of shape from spherical. We found no statistically significant differences in configuration of the circle of Willis between patients with ruptured and unruptured aneurysms. Some previous studies have assessed shape of the aneurysm as a risk factor for rupture comparing ruptured and unruptured aneurysms. In a small study on 27 patients with aneurysms on CTA before operation, several shape factors appeared to be more effective than size in discriminating between ruptured and unruptured aneurysms. 8 The shape factors studied were aspect ratio (a ratio of depth to neck width of the aneurysm), undulation (including irregularities of the wall and lobulation), ellipticity and non-sphericity. Other studies found irregular multilobed appearance to be more common in ruptured aneurysms compared with unruptured aneurysms In none of these studies, however, cases and controls were matched for age, gender and site of the aneurysm. Since age and gender of patients, and site of the aneurysm are risk factors for rupture, 2 lack of matching for these factors may introduce bias. In our study we matched for these factors and included a large number of patients with unruptured aneurysms to obtain precise and reliable estimates of the risk of aneurysm characteristics studied. 43

44 Chapter 3 Anatomical risk factors for rupture of aneurysms To our knowledge, the relation between direction of blood flow into the aneurysm and risk of rupture has not been studied before on angiograms. A mathematical simulation study presented a comprehensive analysis of fluid flow in curved arteries and arterial bifurcations and the relation of these hemodynamic factors with intracranial aneurysm formation, growth and subsequent rupture. 7 Pressures and shear stresses developing along the outer wall of curved arteries and at the apex of arterial bifurcations create a hemodynamic state that promotes aneurysm formation. 7 Flow direction will be straight into an aneurysm that develops along the outer wall of curved arteries or at the apex of an arterial bifurcation because the angle with the main branching vessel is relatively sharp (Figure 3.1a). Our study suggests that the hemodynamic state on curved arteries with flow direction straight into the aneurysm also influences rupture of the aneurysm. Our study confirms findings from a previous study, showing that incompleteness or asymmetry of the ACA1 is much more common in patients with AComA aneurysms than in patients with aneurysms at other sites. 5 Similarly, we found that fetal type (PcomA > PCA1) is more common in patients with PcomA aneurysms, than in patients with aneurysms at other sites, which is in accordance with previous literature. 6 Because we did not find statistically significant differences in configuration of the circle of Willis between ruptured and unruptured aneurysms, our results suggest that the configurations studied are not strong risk factors for rupture. However, the number of aneurysms per site in our study is too small to exclude moderate or small risks of rupture associated with configuration of the circle of Willis. The distribution of aneurysm site in our study differs from that in patients with SAH. In our study, which started with retrieving patients with unruptured aneurysms, we had more MCA and less AcomA aneurysms than in series of patients with ruptured aneurysms. Similar proportions of MCA and AcomA aneurysms were found in previous studies on unruptured aneurysms. 3,11 Moreover, MCA aneurysms are more common in family screening. We did not match cases and controls for size of the aneurysm. In our study we analyzed the hypothesis that morphological characteristics of cerebral arteries and aneurysms are important in the etiological pathway of aneurysm rupture. According to this hypothesis, size can be considered as intermediate in the pathway between development and aneurysm rupture. An aneurysm with flow straight into the aneurysm, non-spherical shape, or both, has a higher risk of enlarging, and thus larger size, and thereby a higher risk of rupture. For the morphological factors studied, size can be considered a result of the risk factor. We anticipated that after adjusting for size the effect of our determinants of interest on the risk of rupture would disappear, which they did when we performed these additional analyses. 44

45 Anatomical risk factors for rupture of aneurysms Chapter 3 The series of unruptured aneurysms consisted of both patients with single (discovered on family screening) and additional (discovered by SAH from another site) aneurysms. Differences between these groups cannot be excluded, but baseline characteristics (besides family history) were comparable. We were not able to match on family history of SAH since reliable information was lacking for many patients with a ruptured aneurysm. Family history at the bedside of these patients is not reliable, 12 and we did not perform in depth assessments of family history in this sample. Moreover, family history is no established risk factor for rupture. 2 Flow dynamics in the aneurysm are difficult to measure on CTA. We assumed, besides direction of flow into the aneurysm measured by the branching angles, other morphological factors are likely to play a role in the flow dynamics, for example neck/dome ratio and asymmetry of the aneurysm. To avoid multiple testing in the sample size of our study, these latter factors were not taken into account in our study. Besides shape and direction of flow another potential risk factor for rupture is irregularity of the aneurysm wall (i.e. blebs or nipples). We were not able to analyze these risk factors, because single-slice or 16-slice CTA was performed in our study, while 64-slice CTA or 3D rotational DSA is necessary for reliable assessment of minor irregularities of the vessel wall. Because in our study we used single and 16-multislice CTA, it is conceivable that our findings would have been clearer when we had measured morphological factors with 64-multislice CTA. Emphasis in the current analysis was on the role of morphological characteristics of cerebral arteries and aneurysms in the risk of aneurysm rupture. In our opinion we now may conclude that such characteristics are important in the etiological path. Whether morphological characteristics contribute independently in aneurysm rupture risk stratification should be assessed in new studies in which these characteristics are determined prospectively. REFERENCES 1. Rinkel GJE, Djibuti M, Algra A, van Gijn J. Prevalence and risk of rupture of intracranial aneurysms: a systematic review. Stroke 1998; 29: Wermer MJ, van der Schaaf I, Algra A, Rinkel GJ. Risk of rupture of unruptured intracranial aneurysms in relation to patient and aneurysm characteristics: an updated meta-analysis. Stroke 2007; 38: Wiebers DO, Whisnant JP, Huston J 3rd, et al. Unruptured intracranial aneurysms: natural history, clinical outcome, and risks of surgical and endovascular treatment. Lancet 2003; 362: Hendrikse J, van Raamt AF, van der Graaf Y, et al. Distribution of cerebral blood flow in the circle of Willis. Radiology 2005; 235:

46 Chapter 3 Anatomical risk factors for rupture of aneurysms 5. Velthuis BK, van Leeuwen MS, Witkamp TD, et al. Surgical anatomy of the cerebral arteries in patients with subarachnoid hemorrhage: comparison of computerized tomography angiography and digital subtraction angiography. J Neurosurg 2001; 95: Horikoshi T, Akiyama I, Yamagata Z, et al. Magnetic resonance angiographic evidence of sex-linked variations in the circle of Willis and the occurrence of cerebral aneurysms. J Neurosurg 2002; 96: Foutrakis GN, Yonas H, Sclabassi RJ. Saccular aneurysm formation in curved and bifurcating arteries. AJNR Am J Neuroradiol 1999; 20: Raghavan ML, Ma B, Harbaugh RE. Quantified aneurysm shape and rupture risk. J Neurosurg 2005; 102: Beck J, Rohde S, El Beltagy M, et al. Difference in configuration of ruptured and unruptured intracranial aneurysms determined by biplanar digital subtraction angiography. Acta Neurochir (Wien) 2003; 145: San Millan RD, Yilmaz H, Dehdashti AR, et al. The perianeurysmal environment: influence on saccular aneurysm shape and rupture. AJNR Am J Neuroradiol 2006; 27: Wermer MJ, van der Schaaf IC, Velthuis BK, et al. Yield of short-term follow-up CT/MR angiography for small aneurysms detected at screening. Stroke 2006; 37: Bromberg JEC, Rinkel GJE, Algra A, et al. Validation of family history in subarachnoid hemorrhage. Stroke 1996; 27:

47 PART II Delayed cerebral ischemia following subarachnoid hemorrhage

48

49 4 Secondary infarction in single or in multiple vascular territories: two different entities following subarachnoid hemorrhage? Nicolien K. de Rooij, Catharina J. M. Frijns, Birgitta K. Velthuis, Gabriël J. E. Rinkel Journal of Neurology 2011; 258:

50 Chapter 4 Single and multiple infarctions after SAH: two different entities? ABSTRACT Objective: The pathogenesis of secondary infarctions (SI) after aneurysmal subarachnoid hemorrhage (SAH) is poorly understood. To assess whether SI in single (SSI) or multiple (MSI) vascular territories represent different disease entities, we compared clinical profiles of patients with these patterns of SI. Methods: CT and MRI examinations of 448 patients were reviewed for new infarctions within 28 days after SAH, and categorized into SSI or MSI. Only patients with adequate follow-up imaging excluding any new infarctions were included for analysis (269 patients). Procedure related infarctions were excluded. Odds ratios (ORs) with corresponding 95% confidence intervals (CI) were calculated for patients with SSI or MSI versus patients without SI to analyze differences in demographic characteristics, vascular risk factors, disease-related characteristics and treatment modalities. Results: Thirty-six patients had SSI, 53 MSI and 180 no SI. ORs in MSI patients were >1.5 times higher compared with ORs in SSI patients for multiple vascular risk factors (MSI 5.4 [2.3 to 13] versus SSI 1.2 [0.5 to 2.8]), poor clinical condition on admission (MSI 4.6 [2.4 to 8.9] versus SSI 2.4 [1.1 to 5.2]), initial loss of consciousness (MSI 2.6 [1.3 to 5.3] versus SSI 1.1 [0.5 to 2.3]) and large amounts of intraventricular blood (MSI 2.9 [1.4 to 5.8] versus SSI 1.5 [0.7 to 3.2]). In multivariate analysis ORs remained higher in MSI for presence of multiple vascular risk factors (MSI 1.9 [1.2 to 2.9] versus SSI 1.1 [0.8 to 1.7]) and initial loss of consciousness (MSI 3.0 [1.0 to 8.9] versus SSI 1.6 [0.6 to 4.0]). Conclusions: Our findings suggest that SSI and MSI after SAH are not distinct disease entities. MSI was related to the same characteristics as SSI but to a larger extent, specifically to the presence of multiple vascular risk factors, initial loss of consciousness, larger amounts of intraventricular blood, and poor clinical status on admission. 50

51 Single and multiple infarctions after SAH: two different entities? Chapter 4 INTRODUCTION Delayed cerebral ischemia (DCI) is a major cause of poor outcome after subarachnoid hemorrhage (SAH). Although cerebral vasospasm has been indicated as the main cause of DCI, 1 vasospasm and DCI do not always go hand in hand, which shows that vasospasm is not the only or essential factor for the development of DCI. 2-4 Other factors that may contribute to DCI besides large vessel vasospasm are inflammatory processes, intracranial microthrombosis, cortical spreading depolarisations and impaired cerebral autoregulation. 5-7 DCI may have several patterns. From a subdivision based on number and sites of infarctions on cerebral computed tomographic (CT) scans after SAH, two common patterns were distinguished. 8 These two common patterns were on the one hand a single cortical infarct and on the other multiple cortical and/or deep cerebral infarcts. It is unclear whether these secondary infarct (SI) patterns represent pathophysiologically different disease entities or different degrees of severity of the same vascular process. We assessed whether secondary infarctions in single (SSI) or multiple (MSI) vascular territories represent different disease entities by comparing the clinical profiles of the patients with SSI or MSI with those of patients without infarction. We analyzed demographic characteristics, vascular risk factors, disease specific characteristics, and treatment modalities as determinants for development of SI. METHODS Patients and data From the prospectively collected database of all consecutive patients with SAH admitted to the University Medical Center Utrecht, we included all patients with an aneurysmal SAH who were admitted within 3 days after onset between January 1999 and June The diagnosis of SAH was based on a positive CT scan or xantochromia of the cerebrospinal fluid. Presence of an aneurysm had to be confirmed by means of CT, MR or catheter angiography. For inclusion in the current study at least one follow-up scan, performed more than 24 h after the initial CT scan, was required. For every patient we recorded data on gender, age, current smoking, history of hypertension, previous cardiovascular events (including stroke, myocardial infarction and peripheral vascular disease), hypercholesterolemia, diabetes mellitus, family history of cardiovascular events, and 51

52 Chapter 4 Single and multiple infarctions after SAH: two different entities? use of salicylic acid and statins at the time of admission. The clinical condition on admission was assessed with the World Federation of Neurological Surgeons scale and dichotomized in good (WFNS I to III) or poor (IV or V). 9 We also recorded loss of consciousness at onset, and modality and timing of intervention. All patients were treated according to our standard protocol, including absolute bed rest, compressive stockings, intensive monitoring on our medium or intensive care unit until aneurysm treatment, nimodipine 360 mg daily orally, and intravenous administration of fluids aiming at normovolemia for 3 weeks. Initial CT scans were evaluated for amount of blood, hydrocephalus and pre-existing infarction. The amount of cisternal and ventricular blood was assessed according to the method described by Hijdra et al. 10 and dichotomized at their median values. The presence of acute hydrocephalus was quantified and adjusted for age by means of the bicaudate index (BCI) divided by the corresponding upper limit per age group. 11 Acute hydrocephalus was defined as an age-adjusted relative BCI of >1. Follow-up scans were evaluated for new ischemic lesions. Lesions caused by extraventricular drains, preexisting infarcts and hypodensities around a hematoma or in the vicinity of the operation area were not considered new infarctions. Only infarctions unrelated to angiography or aneurysm treatment were defined as spontaneous infarction. Infarctions were considered to be related to treatment if new neurological symptoms occurred directly after aneurysm treatment, and an infarct was visible in the territory of the parent vessel of the aneurysm on CT within 48 h after treatment. In case the time of development of infarction was uncertain (for example if a new infarction was present on CT >48 h after treatment and/or the neurological state of the patient directly after treatment could not be assessed due to sedative medication), an expert opinion was made by two of the authors, who were not aware of the risk factor profile. In this expert opinion the report of the operation or endovascular treatment was taken into account. In case of disagreement, the infarction was not counted as spontaneous infarction. To investigate if differentiation between edema and infarction could reliably be made in the presence of an intracerebral hematoma (ICH), we performed an interobserver study between three observers (two senior neurologists and one senior radiologist) in 25 patients with an additional ICH and a surrounding hypodensity. This study showed a large interobserver variability with kappa values around 0.50 (95% CI 0.06 to 0.96). Since subdivision in SSI and MSI in these cases could not reliably be made, patients with an ICH and a surrounding hypodensity were excluded. The location(s) of spontaneous infarctions were determined by using validated arterial territory maps We only included infarctions that developed within 28 days after SAH and were 52

53 Single and multiple infarctions after SAH: two different entities? Chapter 4 confirmed on a late follow-up scan. SSI was defined as infarction in one arterial territory and MSI as infarctions in more than one territory. Adequate follow-up (late CT or MRI follow-up between 28 days and 6 months after SAH) is necessary because within the first 28 days SSI could develop into MSI, and NSI could develop into SSI or MSI. Because new (silent) infarctions in other territories had to be excluded for a valid subdivision, we excluded patients with a single infarction or without any infarction who had no late follow-up scan. All scans were reviewed for presence or absence of hypodensities by one author (NKR). The first 35 scans were also reviewed independently by a senior vascular neurologist (CJMF). No disagreement was noticed with respect to presence or absence of hypodensities. All scans with hypodensities were independently reviewed and subdivided into SSI and MSI by NKR and CJMF. In case of disagreement, consensus was reached in a meeting of all authors, including another senior vascular neurologist and a senior radiologist. Data analysis All patients with adequate radiological follow-up (SSI, MSI and NSI) were included for analysis. Odds ratios (ORs) for MSI or SSI compared with NSI were calculated with corresponding 95% confidence intervals (CI) using univariable logistic regression. We decided to perform an indirect comparison between MSI and SSI as primary analysis to increase the statistical power of the study. Furthermore, an indirect comparison reveals both the relative differences in ORs between MSI and SSI, and the absolute ORs for MSI and SSI separately. Additionally, we performed a direct comparison. All patient specific risk factors and disease specific characteristics were analyzed, except for the risk factors found in less than 10% of the patients. The total number of the vascular risk factors was analyzed as well: age above the median value, male gender, hypertension, previous cardiovascular disease, positive family history of vascular disease and current smoking. We defined high number of vascular risk factors as any number above the median number. Multivariate logistic regression was performed with all variables of which the association with MSI was more than 1.5 times higher or lower than the association with SSI. RESULTS Of 828 consecutive patients with aneurysmal SAH admitted to our hospital within 3 days after onset of the hemorrhage during the study period, we excluded 222 patients with incomplete scans. In 182 of these 222 patients no follow-up scan was made because of good outcome 53

54 Chapter 4 Single and multiple infarctions after SAH: two different entities? (n=65), transfer to another hospital (n=7), or early death (n=110), and in 40 patients the initial or follow-up CT scan could not be retrieved. We excluded 149 patients because a large ICH with a surrounding hypodensity was present and subdivision in SSI and MSI could not reliably be made. Nine patients, in whom no consensus for presence or timing of infarction could be reached, were also excluded. An infarction related to treatment was recorded in 76 patients (17%). Eighty-nine of the 269 included patients with adequate follow-up developed spontaneous infarctions: 36 patients in a single vascular territory and 53 patients in multiple territories. In 180 patients no infarction was present (Figure 4.1). Consecutive SAH patients n = 828 Exclusion: no follow up CT or MR available for review n = 222 Exclusion: large intracerebral hemorrhage n = 149 Patients fulfilling inclusion criteria n = 457 Exclusion: no consensus about presence infarction n = 2 Exclusion: uncertain if infarction was present within 28 days n = 7 Patients for analysis n = 448 Treatment related infarctions were excluded * Spontaneous SI n = 111 ( 24%) No SI n = 337 ( 76%) Probable Single SI n = 22 Probable No SI n = 157 Multiple SI n = 53 Single SI n = 36 No SI n = 180 Figure 4.1 Selection of spontaneous single and multiple secondary infarction (SI) within 28 days after subarachnoid hemorrhage (SAH). * All treatment related infarctions were excluded. Depending on the presence or absence of other infarctions these patients were included in the SSI, MSI or the NSI group. All gray marked patients had adequate radiological follow-up (late follow-up scan between 28 days and 6 months after onset of the SAH) and were included for analyses. SSI and NSI were defined as probable if there was no late CT or MRI follow-up to exclude new infarctions in other territories. 54

55 Single and multiple infarctions after SAH: two different entities? Chapter 4 Table 4.1 Baseline characteristics of 269 included patients Number of patients * No secondary infarction n=180 Single secondary infarction n=36 Multiple secondary infarction n=53 Patient characteristics Women (%) (72) 26 (72) 33 (62) Median age in years (range) (18 83) 51 (26 76) 54 (29 77) Hypertension (%) (24) 8 (30) 14 (29) History of cardiovascular events (%) (15) 3 (9) 8 (16) Family history of cardiovascular disease (%) (14) 9 (25) 9 (17) Diabetes (%) (5) 0 (0) 3 (6) Hypercholesterolemia (%) (4) 3 (8) 3 (6) Current smoking (%) (54) 20 (67) 18 (67) Use of salicylic acid (%) (7) 5 (15) 6 (14) Use of statins (%) (7) 3 (8) 4 (9) SAH characteristics Poor WFNS (IV, V) on admission (%) (21) 14 (39) 29 (55) Any loss of consciousness at onset (%) (37) 12 (39) 27 (61) Median amount of blood on initial CT Cisternal blood (range 0 30) Intraventricular blood (range 0 12) (0 30) 2 (0 12) 26 (0 30) 2 (0 8) 26 (0 30) 3.5 (0 12) Acute hydrocephalus (%) (41) 15 (50) 22 (50) External ventricular drainage (33) 16 (44) 19 (36) Site of the aneurysm (%) Posterior communicating artery Internal carotid artery Middle cerebral artery Anterior communicating artery Pericallosal artery Basilar artery Vertebral artery Other Treatment of the aneurysm (%) No treatment Coiling Clipping (19) 6 (3) 23 (13) 77 (43) 9 (5) 13 (7) 12 (7) 5 (3) 4 (2) 75 (42) 101 (56) 10 (28) 1 (3) 9 (25) 9 (25) 0 (0) 3 (8) 1 (3) 3 (8) 0 (0) 16 (44) 20 (56) 13 (25) 5 (10) 4 (8) 20 (38) 1 (2) 2 (4) 4 (8) 4 (8) 19 (36) 17 (32) 17 (32) Poor outcome (GOS 1 or 2) at discharge (%) (1) 2 (6) 28 (60) WFNS = World Federation of Neurological Surgeons Scale; GOS = Glasgow Outcome Scale. * Number of patients is <269 in case of unknown data. These risk factors were recorded as present if they were mentioned as present in patient s history, otherwise as absent or unknown. Acute hydrocephalus was defined as an age-adjusted relative BCI of > 1. 55

56 Chapter 4 Single and multiple infarctions after SAH: two different entities? Baseline characteristics of these 269 patients are presented in Table 4.1. The median number of vascular risk factors was two. History of hypercholesterolemia, diabetes mellitus, use of salicylic acid and use of statins were recorded in less than 10% of patients, and were not used in the analyses. Results of the indirect comparison (patients with MSI or SSI versus NSI) are shown in Table 4.2. ORs were 1.5 for both MSI and SSI in five characteristics: positive family history of cardiovascular events, smoking, poor WFNS, and high amount of both cisternal and ventricular blood. The association with MSI was more than 1.5 times stronger than with SSI in eight characteristics: male gender, high age, history of cardiovascular events, presence of >2 vascular risk factors (including these former three variables), poor WFNS, initial loss of consciousness and high amount of ventricular blood. There were no factors of which the association with SSI Table 4.2 Differences in clinical and radiological data between patients with single and multiple secondary infarctions (SI) compared with patients without SI Number of patients OR (95% CI) Single SI (n=36) vs No SI (n=180) OR (95% CI) Multiple SI (n=53) vs No SI (n=180) Patient characteristics Male gender ( ) 1.5 ( ) Age 54 years (median) ( ) 1.0 ( ) History of cardiovascular disease ( ) 1.1 ( ) History of hypertension ( ) 1.3 ( ) Family history of cardiovascular events ( ) 1.5 ( ) Current Smoking ( ) 1.7 ( ) More than 2 vascular risk factors ( ) 5.4 (2.3 13) SAH characteristics Poor WFNS (IV or V) ( ) 4.6 ( ) Any loss of consciousness at onset ( ) 2.6 ( ) Amount of cisternal blood 25 (median) ( ) 1.5 ( ) Amount of ventricular blood 2 (median) ( ) 2.9 ( ) Acute hydrocephalus ( ) 1.4 ( ) External ventricular drainage ( ) 1.2 ( ) Clipping vs coiling of the aneurysm ( ) 0.7 ( ) WFNS = World Federation of Neurological Surgeons Scale. Number of following 6 risk factors: age above the median value, male gender, hypertension, previous cardiovascular disease, positive family history of vascular disease and current smoking. High number of vascular risk factors was defined as any number above the median of 2 risk factors. Factors in multivariate analyses. 56

57 Single and multiple infarctions after SAH: two different entities? Chapter 4 was more than 1.5 times stronger than with MSI. In the multivariate analysis the association with MSI remained statistically significant for initial loss of consciousness (OR 3.0; 95% CI 1.0 to 8.9) and presence of >2 vascular risk factors (OR 1.9; 1.2 to 2.9) (Table 4.3). In the direct comparison (MSI versus SSI) we found essentially the same associations, however with wider 95%CIs because of the smaller numbers of patients. Two associations reached statistical significance in the univariable analysis: presence of >2 vascular risk factors (OR 4.7; 1.5 to 14) and any loss of consciousness at onset (OR 2.5; 1.0 to 6.5). In the multivariate analysis of the direct comparison the presence of >2 vascular risk factors remained significantly associated with MSI (OR 1.8; 1.0 to 3.1). Table 4.3 Multivariate analysis of associations of patients and SAH characteristics with single or multiple secondary infarctions (SI) compared with patients without SI Number of patients OR (95% CI) Single SI (n=36) vs No SI (n=180) OR (95% CI) Multiple SI (n=53) vs No SI (n=180) More than 2 vascular risk factors ( ) 1.9 ( ) Poor WFNS (IV or V) ( ) 1.8 ( ) Any loss of consciousness at onset ( ) 3.0 ( ) Amount of ventricular blood 2 (median) ( ) 1.1 ( ) WFNS = World Federation of Neurological Surgeons Scale. Number of the following 6 risk factors: age above the median value, male gender, hypertension, previous cardiovascular disease, positive family history of vascular disease and current smoking. High number of vascular risk factors was defined as any number above the median. DISCUSSION Our study showed that all patients with secondary infarction had a similar clinical profile. Both infarct patterns (single or multiple vascular territories) were related to characteristics representing poor initial condition after SAH and presence of vascular risk factors when compared with patients without infarction. However, patients with MSI had an obviously stronger association with these risk factors than patients with SSI, with odds ratios being higher for MSI than for SSI regarding the presence of multiple vascular risk factors, poor neurological condition on admission, initial loss of consciousness, and large amounts of intraventricular blood. These results suggest that SSI and MSI after SAH are not distinct disease entities, but consequences of increasing severity of the same risk factors, and therefore different degrees of the same pathophysiological process. 57

58 Chapter 4 Single and multiple infarctions after SAH: two different entities? A previous study distinguished five distribution patterns of infarctions after SAH (based on location and number of infarctions) and analyzed factors that determined their occurrence. The authors suggested that single cortical or multiple cortical and/or deep infarctions are two common types of SI, and may represent different pathophysiological mechanisms. 8 That conclusion was based on associations of occurrence of multiple infarctions with history of diabetes, early hydrocephalus and requirement of external ventricular drainage (univariable analysis). Requirement of external ventricular drainage was the only independent predictor in their multivariate analysis. In contrast to our methods, this previous study investigated a small population of 56 patients, directly comparing patients with SI in single or multiple vascular territories, without excluding treatment related infarctions and with limited CT follow-up time. Trends found for associations of multiple infarctions with higher age, poor clinical condition on admission and presence of vascular risk factors corresponded with our findings. Because of the limited statistical power with 56 patients, relevant factors such as initial WFNS score may have been undetected. To reduce this risk of type II error, we made an indirect comparison that included 269 patients. In our study acute hydrocephalus was assessed as BCI adjusted for age, which we feel is a more accurate measure than BCI alone, and found no strong association between MSI and hydrocephalus. The factors that proved to be the most important determinants in our study (presence of initial loss of consciousness and presence of multiple vascular risk factors) were not studied in the previous study. We studied a large population of 269 SAH patients, 89 of whom developed spontaneous infarctions after SAH. One of the strengths of our study is the meticulous elimination of possible sources of confounding data due to equivocal assessment of spontaneous SI. All CT scans were independently reviewed by two authors and a specialist team in case of disagreement. We performed an interobserver study of the assessment of hypodense lesions at the location of intracerebral hematomas. Because distinction between perihematomal edema and (hemorrhagic) infarction proved difficult to make, we excluded patients with ICH and surrounding hypodense lesions in the comparison between patients with SSI and MSI. For similar reasons we also excluded patients with hypodensities in the operation field. In a study on pathogenesis of DCI, it is essential to include only infarctions of which it is completely sure that they are unrelated to a procedure. For that reason we considered infarction as procedure related, if there was a chance, not only if it was proven to be procedure related. For analysis we included only patients with adequate radiological follow-up. The disadvantage of this elimination of several sources of bias is that we had to exclude large numbers of patients. However, because the original sample we studied was very large, we were still able to perform sound multivariate analyses. 58

59 Single and multiple infarctions after SAH: two different entities? Chapter 4 As this was a retrospective study, we had some missing data on the characteristics studied. Also, there was no prospective protocol for standardized CT scanning at predetermined time points and patients had scans at varying time points during their stay in our hospital. Since we analyzed two radiological patterns of SI, clinical symptoms of secondary ischemia without an infarct on CT or MRI were not taken into account. MRI was not performed in the acute phase of SI and was performed only in a minority of our patients for follow-up of endovascular treatment of the aneurysms. Follow-up by MRI was mainly performed in patients without infarction, and more in SSI than in MSI. Therefore, the fact that MRI is more sensitive is not likely to have biased our results. Previous literature found that focal or diffuse distribution of vasospasm by TCD or angiogram failed to reliably predict the subsequent pattern of brain infarction. 8 The aim of our study was not to assess the relation between vasospasm and infarction, but to investigate the existence of two different entities of SI. We defined SI on noncontrast CT or MRI and had no full information on presence or absence of vasospasm in all patients. Though vasospasm may contribute to the development of SI, infarction and vasospasm do not always go hand in hand. Obviously, in observational studies SI is a more important outcome event than the presence of arterial spasm by angiography or TCD. 14,15 Our data give no information whether generalized vasospasm was more present in patients with MSI, or focal vasospasm in patients with SSI. However, other authors found no differences in occurrence or extent of vasospasm in patients with different single or widespread infarct patterns. 8 Therefore, we feel that this is not an important shortcoming of our study. In this study the presence of more than two vascular risk factors was associated with development of SI. We could not find previous studies with similar findings. Since the relationship between vascular risk factors and SI was not the primary aim of the current study, no definitive conclusions can be drawn from this finding. Future studies should focus on the identification of additional clinical risk factors of SI and unravel the association between vascular risk factors and SI after SAH. Identification of risk factors may improve our understanding of the pathogenesis of SI and help to tailor preventive treatment to the risk profile of individual patients. In conclusion, our findings suggest that SSI and MSI do not represent two different disease entities. We found that MSI is related to the same characteristics as SSI but to a larger extent, specifically to presence of multiple vascular risk factors, initial loss of consciousness, larger amounts of intraventricular blood and poor clinical status on admission. Our results underline the importance of clinical risk factors for occurrence and extent of secondary infarction after SAH. 59

60 Chapter 4 Single and multiple infarctions after SAH: two different entities? REFERENCES 1. Treggiari-Venzi MM, Suter PM, Romand JA. Review of medical prevention of vasospasm after aneurysmal subarachnoid hemorrhage: a problem of neurointensive care. Neurosurgery 2001; 48: Aralasmak A, Akyuz M, Ozkaynak C, et al. CT angiography and perfusion imaging in patients with subarachnoid hemorrhage: correlation of vasospasm to perfusion abnormality. Neuroradiology 2009; 51: Dankbaar JW, Rijsdijk M, van der Schaaf I, et al. Relationship between vasospasm, cerebral perfusion, and delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Neuroradiology 2009; 51: Rabinstein AA, Friedman JA, Weigand SD, et al. Predictors of cerebral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2004; 35: Dreier JP, Major S, Manning A, et al. Cortical spreading ischaemia is a novel process involved in ischaemic damage in patients with aneurysmal subarachnoid haemorrhage. Brain 2009; 132: Pluta RM, Hansen-Schwartz J, Dreier J, et al. Cerebral vasospasm following subarachnoid hemorrhage: time for a new world of thought. Neurol Res 2009; 31: Vergouwen MD, Vermeulen M, Coert BA, et al. Microthrombosis after aneurysmal subarachnoid hemorrhage: an additional explanation for delayed cerebral ischemia. J Cereb Blood Flow Metab 2008; 28: Rabinstein AA, Weigand S, Atkinson JL, Wijdicks EFM. Patterns of cerebral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2005; 36: Report of World Federation on Neurological Surgeons committee on a universal subarachnoid hemorrhage grading scale. J Neurosurg 1988; 68: Hijdra A, Brouwers PJAM, Vermeulen M, van Gijn J. Grading the amount of blood on computed tomograms after subarachnoid hemorrhage. Stroke 1990; 21: Hasan D, Vermeulen M, Wijdicks EFM, et al. Management problems in acute hydrocephalus after subarachnoid hemorrhage. Stroke 1989; 20: Tatu L, Moulin T, Bogousslavsky J, Duvernoy H. Arterial territories of human brain: brainstem and cerebellum. Neurology 1996; 47: Tatu L, Moulin T, Bogousslavsky J, Duvernoy H. Arterial territories of the human brain: cerebral hemispheres. Neurology 1998; 50: Frontera JA, Fernandez A, Schmidt JM, et al. Defining vasospasm after subarachnoid hemorrhage: what is the most clinically relevant definition? Stroke 2009; 40: Vergouwen MD, Vermeulen M, van Gijn J, et al. Definition of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage as an outcome event in clinical trials and observational studies: proposal of a multidisciplinary research group. Stroke 2010; 41:

61 5 Delayed cerebral ischemia after subarachnoid hemorrhage: a systematic review of clinical, laboratory and radiological predictors Nicolien K. de Rooij, Gabriël J. E. Rinkel, Jan Willem Dankbaar, Catharina J. M. Frijns Stroke 2013; 44:43-54.

62 Chapter 5 Predictors of delayed cerebral ischemia after SAH ABSTRACT Objective: Established predictors of delayed cerebral ischemia (DCI) after aneurysmal subarachnoid hemorrhage (SAH) are large amounts of extravasated blood and poor clinical condition on admission. The predictive value of other factors is uncertain. Methods: We searched MEDLINE ( ) for clinical, laboratory and radiological predictors routinely available within 72 hour after SAH. The studies were categorized according to methodological quality. Crude data and effect estimates (odds ratio (OR), hazard ratios (HR) and risk ratio) with 95% confidence intervals (CI) were extracted, (re-)calculated and pooled if possible. For every potential predictor we assessed all effect estimates on consistency (point estimates in equal direction) and clinical relevance (size and 95% CI). Results: Fifty-two studies on 33 potential predictors were included. There was strong evidence ( 3 high quality studies) for a higher risk of DCI in smokers (pooled OR 1.2; 95% CI 1.1 to 1.4), and moderate evidence (2 high quality studies) for an increased risk in patients with hyperglycemia (OR 3.2; 95% CI 1.8 to 5.8, and HR 1.7; 95% CI 1.1 to 2.5), hydrocephalus (OR 1.3; 95% CI 1.1 to 1.5, and OR 2.6; 95% CI 1.2 to 5.5), history of diabetes (pooled OR 6.7; 95% CI 1.7 to 26) and early systemic inflammatory response syndrome (pooled OR 2.1; 95% CI 1.4 to 3.3). Evidence was limited for increased risk in women (pooled OR 1.3; 95% CI 1.1 to 1.6) and in patients with history of hypertension (pooled OR 1.5; 95% CI 1.3 to 1.7). The evidence on initial loss of consciousness, history of migraine, previous use of SSRIs, hypomagnesemia, low hemoglobin or high blood flow on early transcranial Doppler was also limited. Conclusions: There is strong evidence that smoking is a predictor of DCI. For several other potential predictions the evidence is moderate, limited, or inconsistent. 62

63 Predictors of delayed cerebral ischemia after SAH Chapter 5 INTRODUCTION Delayed cerebral ischemia (DCI) is a major contributor to the high case fatality and morbidity of aneurysmal subarachnoid hemorrhage (SAH). About 30% of the SAH patients develop DCI, 1 but it is difficult to predict which patients. Established predictors of DCI are large amount of subarachnoid blood detected on CT imaging and poor clinical condition on admission. 2-4 Many predictors or risk factors of DCI have been reported, but a systematic overview on suggested predictors and risk factors is lacking. Our aim was to review and summarise the evidence on predictors of DCI in addition to the amount of blood and clinical condition on admission, with a focus on demographic, clinical, laboratory and radiological data that are routinely available on admission or within the first 3 days. METHODS Search strategy and selection criteria A MEDLINE search was performed from 1960 until January, 2012 with the search terms subarachnoid h(a)emorrhage, AND infarction OR isch(a)emia OR isch(a)emic OR vasospasm. Papers published in English, German, French, Italian, or Spanish were reviewed. The reference lists of all included articles were checked for further relevant articles. This method of cross-checking was continued until no further publications were found. One author (NKR) screened all records with respect to title and abstract. For all selected articles two authors (NKR and JWD/CJMF) independently assessed whether the selection criteria, as mentioned in Figure 5.1, were met. In case of disagreement a third author (CJMF/GJER) reviewed the article and the disagreement was resolved by discussion between the three reviewers. Appraisal of the selected articles Two authors (NKR and JWD/CJMF) independently extracted data and assessed methodological quality of all included articles. A level of evidence was assessed for all suggested predictors (hereafter mentioned variables) (Figure 5.1). 63

64 Chapter 5 Predictors of delayed cerebral ischemia after SAH MEDLINE SEARCH 7,311 records 198 articles retrieved for detailed evaluation 52 articles meeting inclusion criteria # (17,496 patients) Data extraction Methodological quality score Two authors independently extracted crude data and effect estimate (odds ratio, risk ratio or hazard ratio) Two authors independently assessed methodological quality (range 0 to 15) (Figure 5.2) Consistency of effect estimates (Figure 5.3) Clinical relevance of effect estimates (Size and 95%CI) High quality (33 studies) score 10 Low quality (19 studies) score <10 LEVEL OF EVIDENCE Depending on the data extraction and the quality score of the studies the level of evidence was assessed for each potential predictor (Figure 5.3) Figure 5.1 Flowchart literature search with inclusion and exclusion criteria. # Inclusion criteria Study population consisted of at least 25 patients with SAH; >85% of whom had a proven aneurysm on imaging. Outcome was symptomatic vasospasm s asm or (delayed) d cerebral infarction/ischemia n ischemia Variable was easily available in clinical practice. ce. Easily available a was defined ed as obtainable able by: history or physical examination ati n general blood tests t electrocardiogram, card o r transcranial r nia Doppler or CT Variable a could be assessed within in 3 days after onset, preferably ra on admission. sio Variable a was assessed before development of DCI Variable a was analysed as primary r aim (not analysed as a covariable) a Effect f estimate (odds d ratio, hazard ratio or risk ratio) including a 95% confidence interval was given or could be calculated. l In studies s on laboratory r measurements ments a p-value was allowed only when crude data were presented Exclusion criteria r i Studies with a primary r diagnostic aim for DCI Studies in which patients t were (pre)selected by presence of vasospasm s asm on imaging. i g. Studies primary r analysing the predictive value of established s predictors r (clinical ical condition io n on admission sio or amount of extravasated blood on imaging) i g) Variables a were excluded if they were influenced directly by a treatment t ent installed after admission sio t 64

65 Predictors of delayed cerebral ischemia after SAH Chapter 5 Data extraction The crude data and effect estimates (odds ratio (OR), risk ratio (RR) or hazard ratio (HR)) were recorded for each variable separately. ORs were the preferred effect estimate. If an OR was not given, it was calculated together with 95% confidence intervals (CI) based on the crude data if possible. Only the data on variables that were the primary aim of a study were retrieved. If studies had multiple definitions on the same variable, we chose the definition of the variable most in line with the other publications to increase the homogeneity of the data. If multiple analyses were performed (or multiple publications were found) on the same study population using different definitions of DCI, we chose those associations with the definition of DCI that included clinical deterioration. Methodological quality score For each article a predefined standardized scoring system was used with 11 criteria on study design and aim, study population and size, outcome measures, and analyses (Figure 5.2). The scoring system was adapted from other systematic reviews on prognostic factors, 5-7 and modified to cover the topic of our review. The quality score was used to assess the hierarchical order of the studies. We predefined high quality (HQ) as a score of 10 out of 15. Level of evidence For each variable separately, we assessed all data on consistency (point estimates in equal of different directions) and clinical relevance (size and 95% CI of the point estimates). Definitions of consistency and clinical relevance are given in Figure 5.3. Based on both their consistency and their clinical relevance all variables were categorised into 3 groups: predictive factors, non-predictive factors, or factors with inconsistent evidence (Figure 5.3). The level of evidence for predictive and non-predictive factors depended on the number of (high quality) studies with a relevant or neutral effect estimate: strong ( 3 HQ-studies), moderate (2 HQ-studies) or limited evidence (<2 HQ-studies). Definitions of relevant and neutral effect estimates are given in Figure

66 Chapter 5 Predictors of delayed cerebral ischemia after SAH Study methods Design Prospective cohort (3) Retrospective from prospective database or trial cohort (2) Retrospective cohort or prospective case control (1) Population Representative group for all SAH patients # Baseline characteristics (age, gender, initial condition) were described for all patients Arrival < 72 hours in hospital* Points Aim Prediction was primary aim of the study DCI was primary endpoint 1 1 Size Number of patients >100 1 Outcome A clinical definition of DCI was reported 1 Data analysis and presentation Univariable technique Multivariable technique Either all crude numbers provided, or odds ratio with 95% CI was calculated (1) Both were provided (2) Multivariable analyses was performed, and the number of predictors studied was less than 1/10 of the number of patients 2 1 Total score <10 = low quality = high quality 15 Figure 5.2 Methodological Quality Score. # A study population was considered not representative if a patient selection was made on gender, age, clinical condition on admission, location of the aneurysm, or treatment method of the ruptured aneurysm. If other exclusion criteria were mentioned, the reviewers checked the baseline table to assess whether a study population was representative. * Articles including only patients with arrival <72 hours in hospital were considered of good quality, because otherwise DCI could have already developed. Analyses Additional to assessment of the level of evidence, we performed a formal meta-analysis if possible. We calculated a pooled OR with corresponding 95% CI for those variables that were assessed in two or more high quality studies that provided crude data and had limited heterogeneity in the definition of the variable. RESULTS Out of 7,311 records, 52 studies totalling 17,496 patients met our inclusion criteria We included 8 prospective cohort studies (1,192 patients), 6 studies using trial cohorts (8,506 66

67 Predictors of delayed cerebral ischemia after SAH Chapter 5 THE LEVEL OF EVIDENCE PREDICTIVE FACTORS (Table 5.2) Consistency All ORs have point estimates in the same direction (all 0.9 or all 1.2), and Clinical relevance At least 1 study with a relevant OR Relevant OR was defined as: OR of 2.0 or 0.5, or a statistically significant OR of 1.2 or 0.9 YES PREDICTIVE FACTORS Strong evidence At least 3 high quality studies with a relevant OR NO Moderate evidence At least 2 high quality studies with a relevant OR NO Limited evidence NO NON-PREDICTIVE FACTORS NON-PREDICTIVE FACTORS (Table 5.3) Consistency All ORs have point estimates around 0 (between 0.5 and 2.0), and YES Clinical relevance Strong evidence At least 3 high quality studies with a neutral OR NO All 95% confidence intervals contain 1, and At least 1 study with a neutral OR Neutral OR was defined as: OR between 0.9 and 1.2 Moderate evidence At least 2 high quality studies with a neutral OR NO NO Limited evidence FACTORS WITH INCONSISTENT EVIDENCE (Table 5.4) If none of the above Figure 5.3 Level of evidence applied for each variable. To assess the level of evidence (strong, moderate or limited) we primarily used results from high quality studies. If less than 2 high quality studies were available, the results from the low quality studies were included. In some studies an OR was not given or could not be calculated. Other effect estimates (risk ratios and hazard ratios) were considered as equivalent. 67

68 Chapter 5 Predictors of delayed cerebral ischemia after SAH patients), 36 retrospective cohort studies (7,666 patients), and 2 case control studies (132 patients). Thirty-three studies (63%) were rated as high quality. 9-13,15,16,18-20,22,28-31,35,36,38,39,41-43,46,48-57 Table 5.1 summarizes the study characteristics and the investigated variables. Altogether, 33 variables were reported. The level of evidence for each variable is shown in Tables 5.2 to 5.4, which also present all effect estimates in univariable and multivariable analyses, and a high quality-pooled OR if possible. Predictive factors (Table 5.2) Strong evidence was found for smoking (pooled OR 1.2; 95% CI 1.1 to 1.4). This was based on 4 consistent HQ-studies, of which 3 found statistically significant ORs, the highest being 4.5. All 4 studies used a definition of DCI which included clinical deterioration. Moderate evidence was found for history of diabetes (pooled OR 6.7; 95% CI 1.7 to 26) and early systemic inflammatory response syndrome (SIRS: pooled OR 2.1; 1.4 to 3.3). Moderate evidence was also found for hyperglycemia on admission (HR 1.7; 95% CI 1.1 to 2.5 and OR 3.2; 95% CI 1.8 to 5.8) and for hydrocephalus based on 2 HQ-studies with relevant effect estimates (OR 1.3; 95% CI 1.1 to 1.5 and OR 2.6; 95% CI 1.2 to 5.5), whereas the third found no difference. Because HRs were given without crude data, no pooled ORs could be calculated for hyperglycemia and hydrocephalus. Limited evidence was found for gender (pooled OR 1.3; 95% CI 1.1 to 1.6) and history of hypertension (pooled OR 1.5; 95% CI 1.3 to 1.7). For both variables evidence was limited because only one HQ-study found a relevant effect estimate. Limited evidence was also found in 6 variables with only one publication (initial loss of consciousness, history of migraine, previous use of selective serotonin reuptake inhibitors (SSRI), hypomagnesemia, high mean blood flow velocity on TCD within the first 48 hours, and low hemoglobin on admission). Non-predictive factors (Table 5.3) Strong evidence was found for absence of an association between DCI and the location of the aneurysm. We found five HQ-studies without a relevant difference in risk of DCI. The only study with a significant OR for lower risk in vertebrobasilar aneurysms was of low quality, including only 17 patients with a vertebrobasilar aneurysm. 21 Moderate evidence was found that previous use of aspirin, alcohol abuse, hypertension on admission, hematocrit, white blood cell count, platelet count and size of the aneurysm are not predictive. For previous use of aspirin, all 4 studies presented data based on history. Using 68

69 Predictors of delayed cerebral ischemia after SAH Chapter 5 Table 5.1 Overview of the 52 included studies and the variables studied, in order of publication date Demographics History Clinical Laboratory test Others Author (reference) Year of publication Country Study design No of SAH patients Representative study population # Diagnostic criterion DCI Methodological score Gender Age Race History of hypertension History of cardiovascular disease History of diabetes History of migraine Previous use of statins Previous use of aspirin Use of cocaine Smoking Alcohol Composite score of atherosclerosis Initial loss of consciousness Adverse physiological conditions * Hemoglobin or hematocrite CRP or white blood cell count Platelet count Magnesium Glucose Hyponatremia Hydrocephalus Location or size of the aneurysm Cardiac changes TCD measurements Other Kale USA Rch 108 Ω C 11 X Zheng USA Rch 124 PS D 9 X Shuie Australia Rch 254 Ω C 12 X Juvela Finland Rch 178 Y D 12 X Alaraj USA Rch 573 Ω C 9 X Ryttlefors Worldwide Tch A 13 X Inagawa Japan Rch 291 S AD 6 X Inagawa Japan Rch 195 S AD 6 X Yousef USA Pch 149 P BC 9 X Magge USA Rch 391 SΩ C 9 X Carrera USA Rch 199 Ω AC 13 X Jeon Korea Rch 94 Ω C 7 X Dumont USA Rch 113 Ω C 12 X Moskowitz USA Rch C 11 X Kruyt Netherlands Rch AB 13 X Table 5.1 continues on next page 69

70 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.1 Continued Demographics History Clinical Laboratory test Others Author (reference) Year of publication Country Study design No of SAH patients Representative study population # Diagnostic criterion DCI Methodological score Gender Age Race History of hypertension History of cardiovascular disease History of diabetes History of migraine Previous use of statins Previous use of aspirin Use of cocaine Smoking Alcohol Composite score of atherosclerosis Initial loss of consciousness Adverse physiological conditions * Hemoglobin or hematocrite CRP or white blood cell count Platelet count Magnesium Glucose Hyponatremia Hydrocephalus Location or size of the aneurysm Cardiac changes TCD measurements Other Dhar USA Rch A 13 X Kerner Germany Pch 170 SΩ C 9 X Dreier Germany CC 72 GW A 7 X Bakker Netherlands Rch B 13 X Kirshnamuthy USA Rch 320 Ω A 12 X McGirt USA Rch C 11 X Naidech USA Rch D 11 X van den Bergh Netherlands Rch 323 Ω B 12 X Schuiling Netherlands Rch 121 Ω AB 10 X Rothoerl Germany Rch 88 S AB 7 X Schuiling Netherlands Rch 136 Ω AB 10 X Badjatia USA Rch 352 G C 12 X Singhal USA Rch 514 Ω C 12 X X X Parra USA CC 60 Ω A 8 X Hirashima Japan Rch 145 S C 9 X Toussaint USA Rch 296 Ω C 11 X 70

71 Predictors of delayed cerebral ischemia after SAH Chapter 5 Demographics History Clinical Laboratory test Others Author (reference) Year of publication Country Study design No of SAH patients Representative study population # Diagnostic criterion DCI Methodological score Gender Age Race History of hypertension History of cardiovascular disease History of diabetes History of migraine Previous use of statins Previous use of aspirin Use of cocaine Smoking Alcohol Composite score of atherosclerosis Initial loss of consciousness Adverse physiological conditions * Hemoglobin or hematocrite CRP or white blood cell count Platelet count Magnesium Glucose Hyponatremia Hydrocephalus Location or size of the aneurysm Cardiac changes TCD measurements Other Collignon USA Rch 58 Ω C 7 X Rabinstein USA Rch 143 Ω D 11 X X X Mac Donald Worldwide Tch 3,567 T A 13 X X X X X X X Howington USA Rch 108 YΩ C 8 X van den Bergh Netherlands Pch AB 14 X Conway USA Rch C 13 X Yoshimoto Japan Rch A 10 X Kremer Germany Rch 40 EP AD 4 X Hop Netherlands Pch 125 Ω AB 14 X X X X X X Charpentier France Rch C 12 X X X X X Weir Europe/USA Tch 3,436 T C 12 X Niikawa Japan Rch 103 AGPS C 6 X X Lasner USA Pch 70 Ω C 12 X X X X X X X X X Fujii Japan Rch BC 11 X X X X Lanzino USA/Canada Tch 457 TΩ C 8 X Kongable USA/Canada Tch 457 TΩ C 8 X Table 5.1 continues on next page 71

72 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.1 Continued Demographics History Clinical Laboratory test Others Author (reference) Year of publication Country Study design No of SAH patients Representative study population # Diagnostic criterion DCI Methodological score Gender Age Race History of hypertension History of cardiovascular disease History of diabetes History of migraine Previous use of statins Previous use of aspirin Use of cocaine Smoking Alcohol Composite score of atherosclerosis Initial loss of consciousness Adverse physiological conditions * Hemoglobin or hematocrite CRP or white blood cell count Platelet count Magnesium Glucose Hyponatremia Hydrocephalus Location or size of the aneurysm Cardiac changes TCD measurements Other Juvela Finland Pch 242 Ω A 12 X Ameriso USA Pch 69 Ω BC 11 X X X Juvela Finland Pch 260 Ω E 10 X Inagawa Japan Rch 138 AGS C 8 X Hijdra Netherlands Tch 176 T AB 11 X SAH = subarachnoid hemorrhage; DCI = delayed cerebral ischemia; CRP = C-reactive protein; TCD = transcranial Doppler; SIRS = systemic inflammatory response syndrome; USA = United States of America. Study design: RCh = retrospective cohort; TCh = trial cohort; PCh = prospective cohort; CC = case control design. # Representative study population: + = yes, the study population is representative for the entire SAH population. A study population was deemed not representative if a predefined patient selection was made on for example gender, age, clinical condition on admission, location or treatment method of the ruptured aneurysm. Ω = not admitted within 72 hours; A = only aneurysms in anterior circulation; E = only patients with endovascular treatment of the aneurysm; S = only patients with surgical treatment of the aneurysm; G = only patients with good clinical condition on admission, or surviving more than 7 days in the hospital; P = only patients with poor clinical condition on admission or high amount of blood on CT; T = patients with contra indication of trial medication were excluded; W = only women included; Y = other selection. Diagnostic criterion DCI: A = clinical deterioration with exclusion of other causes than DCI; B = clinical deterioration with exclusion of other causes than DCI and presence of a compatible new infarction on brain imaging; C = clinical deterioration with exclusion of other causes than DCI and presence of vasospasm on TCD, CT-angiogram or angiography; D = new infarction on brain imaging; E = delayed neurological deficit not further specified. * Adverse physiological conditions = abnormal blood pressure, body temperature or SIRS. 72

73 Predictors of delayed cerebral ischemia after SAH Chapter 5 Table 5.2 Level of evidence for predictive factors for DCI PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis STRONG LEVEL OF EVIDENCE Smoking pros cohort, 22 n= current smoking 42% 1.4 ( ) -- pros cohort, 38 n=70 12 smoking not further defined 64% 4.5 (1.3 16) 4.7 ( ) trial cohort, 56 n= 3, any smoking at time of SAH (long term or intermittently) 61% 1.2 ( ) unknown retro cohort, 35 n= current smoking and/or long term 59% 1.7 ( ) 1.9 ( ) smoking High quality pooled odds ratio 1.2 ( ) MODERATE LEVEL OF EVIDENCE History of diabetes pros cohort, 38 n=70 12 positive 3% 2.6 (0.2 40) unknown retro cohort, 18 n= positive 9% 8.1 (1.6 40) 9.9 (1.2 79) High quality pooled odds ratio 6.7 (1.7 26) Systemic inflammatory response syndrome $ retro cohort, 16 n= positive on admission 50% 2.0 ( ) unknown retro cohort, 57 n= first measured data positive 54% 3.1 ( ) unknown retro cohort, 49 n= positive within 24 hours 72% 1.9 ( ) unknown High quality pooled odds ratio 2.1 ( ) Hyperglycemia retro cohort, 36 n= admission value above median (>6.9 mmol/l) retro cohort, 10 n= admission value >140 mg/dl (RR for every 1 mg/dl increase above 110 mg/dl) -- HR 1.7 ( ) HR 1.4 ( ) 68% 3.2 ( ) RR 0.99 ( ) Table 5.2 continues on next page 73

74 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.2 Continued PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis High quality pooled odds ratio could not be calculated due to missing crude data from HR pros cohort, 32 n=170 9 admission >6.7 mmol/l 59% 0.9 ( ) unknown Hydrocephalus retro cohort, 11 n= increase of 0.1 BCI adjusted for age on admission scan -- HR 1.00 ( ) unknown trial cohort, 39 n=3, hydrocephalus not further defined 42% 1.3 ( ) unknown trial cohort, 20 n= BCI>95 th percentile for age on 19% 2.6 ( ) unknown admission scan High quality pooled odds ratio could not be calculated due to missing crude data from HR LIMITED LEVEL OF EVIDENCE Gender trial cohort, 39 n=3, women 83% 1.3 ( ) 1.2 ( ) retro cohort, 13 n= women* 60% 0.9 ( ) unknown pros cohort, 38 n=70 12 women* 63% 1.6 ( ) unknown retro cohort, 46 n= women* 69% unknown 1.9 ( ) High quality pooled odds ratio 1.3 ( ) trial cohort, 33 placebo n=457 8 women* 35% 1.2 ( ) unknown History of hypertension pros cohort, 22 n= positive 24% 1.2 ( ) -- trial cohort, 39 n=3, positive 33% 1.5 ( ) 1.3 ( ) pros cohort, 38 n=70 12 positive 46% 1.7 ( ) unknown retro cohort, 13 n= use of antihypertensive medication on admission 32% 1.1 ( ) unknown 74

75 Predictors of delayed cerebral ischemia after SAH Chapter 5 PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis High quality pooled odds ratio 1.5 ( ) History of migraine case control, 17 n=72 7 positive (telephone questionnaire) ( ) unknown Previous use of SSRI retro cohort, 52 n= on admission 8% 2.1 ( ) 1.4 ( ) Initial loss of consciousness pros cohort, 22 n= >1 hour at onset (interview with an eyewitness) 46% 4.9 (2.2 11) HR similar as univariate 6.0 (3.0 12) Low hemoglobin retro cohort, 43 n= per g-dl increase in admission hemoglobin ( ) unknown Hypomagnesemia pros cohort, 54 n= <0.70 mmol/l on admission 38% HR 2.4 ( ) HR 1.9 ( ) retro cohort, 14 n=58 7 minimum value in first 3 days (mg/dl) 1.77 vs 1.81 V non significant p-value unknown Transcranial Doppler measurements retro cohort, 12 n= mean blood flow velocity of the middle cerebral artery >90 cm/s, measured within 48 h of SAH onset unknown 2.9 ( ) 2.7 ( ) BCI = bicaudate index; HR = hazard ratio; pros cohort = prospective cohort study; retro cohort = retrospective cohort study; RR = risk ratio; SAH = subarachnoid hemorrhage; SSRI = selective serotonin reuptake inhibitors; trial cohort = cohort study using patients from a trial population. # Studies were presented in order of methodological quality. Ratios are odds ratios, or otherwise marked by HR (hazard ratio) or RR (risk ratio). All calculated odds ratios are marked by. * Recorded OR was from the inverse variable (men instead of women), we recalculated the risk ratio to make it comparable with the other studies. $ SIRS is defined as the presence of at least 2 out of 4 criteria: tachycardia, tachypnea, fever or hypothermia and leukocytosis or leukopenia. V Values in patients with DCI versus no DCI. 75

76 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.3 Level of evidence for predictive factors for DCI NON PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis STRONG LEVEL OF EVIDENCE Location of aneurysm trial cohort, 39 n=3, ICA versus ACA ICA versus MCA ICA versus posterior circulation ICA versus other locations unknown 1.0 ( ) 0.9 ( ) 0.9 ( ) 1.6 ( ) unknown pros cohort, 38 n=70 12 anterior versus posterior circulation 80% 1.0 ( ) unknown retro cohort, 13 n= posterior circulation versus other 17% 0.7 ( ) unknown locations retro cohort, 51 n= posterior versus anterior 24% 0.7 ( ) 0.9 ( ) retro cohort, 46 n= posterior versus anterior 25% unknown 1.1 ( ) retro cohort, 21 n=145 9 vertebrobasilar vs anterior circulation 12% 0.14 ( ) 0.16 ( ) retro cohort, 25 n= categories: AcoA, distal ACA, ICA, MCA, VBA -- non significant p-value (no OR could be calculated) unknown retro cohort, 34 n=40 4 anterior versus posterior circulation 45% 1.8 ( ) unknown MODERATE LEVEL OF EVIDENCE Previous use of aspirin pros cohort, 22 n= last 2 weeks (history) 20% 1.1 ( ) unknown pros cohort, 29 n= last 7 days (history) multiple other analyses 19% retro cohort, 52 n= documented admission medications (history) 1.0 ( ) See text unknown 8% 1.4 ( ) unknown retro cohort, 53 n= questionnaire on admission (history) 10% 1.5 ( ) 1.5 ( ) High quality pooled odds ratio 1.3 ( ) 76

77 Predictors of delayed cerebral ischemia after SAH Chapter 5 NON PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis Alcohol abuse pros cohort, 38 n=70 12 alcohol abuse not further defined 10% 1.0 ( ) unknown pros cohort, 28 n= >300 g/week during last year 19% 0.9 ( ) unknown Hypertension on admission trial cohort, 39 n=3, per 10 mm Hg increase >130 systolic per 10 mm Hg increase >85 diastolic ( ) 1.01 ( ) unknown unknown retro cohort, 13 n= systolic 160 or diastolic 100 mmhg 36% 1.1 ( ) unknown Hematocrit retro cohort, 19 n= admission value T (%) 37.7 vs 39.0 V non significant p-value -- pros cohort, 9 n=69 11 admission value T (%) 38.1 vs 39.0 V non significant p-value -- White blood cell count retro cohort, 19 n= admission value T (10 3 /mm 3 ) 12.9 vs 12.1 V non significant p-value -- pros cohort, 9 n=69 11 admission value T (10 3 /ml) 12.3 vs 12.9 V non significant p-value -- retro cohort, 47 n=88 7 value day 1 T (/nl) 14.5 vs 11.9 V p< retro cohort, 44 n=103 6 value day 0 2 T (10 3 /mm 3 ) 13.3 vs 11.7 V non significant p-value -- Platelet count retro cohort, 19 n= admission value T (10 3 /mm 3 ) 231 vs 243 V non significant p-value -- pros cohort, 9 n=69 11 admission value T (10 3 /ml) 265 vs 262 V non significant p value -- retro cohort, 44 n=103 6 value day 0 2 T (10 3 /mm 3 ) 262 vs 197 V p= Size aneurysm trial cohort, 39 n=3, mm versus <12 mm >25 mm versus <12 mm retro cohort, 51 n= mm versus <5 mm 10 mm versus <5 mm retro cohort, 26 n=195 6 >5 mm versus <5 mm >10 versus <10 mm unknown 0.9 ( ) see text unknown 0.7 ( ) 1.1 ( ) 73% 21% 1.0 ( ) 0.6 ( ) 1.4 ( ) 2.5 ( ) unknown 1.2 ( ) Table 5.3 continues on next page 77

78 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.3 Continued NON PREDICTIVE FACTORS Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariable analysis Odds ratio (95% CI) in multivariable analysis LIMITED LEVEL OF EVIDENCE Race trial cohort, 39 n=3, Afro-American versus Caucasian unknown 1.1 ( ) unknown pros cohort, 38 n=70 12 white versus non-white 74% 1.6 ( ) unknown Composite score of atherosclerosis pros cohort, 22 n= smoking, history of hypertension, and/or cardiovascular disease retro cohort, 55 n= increasing value of modified index of the Framingham study (quartiles) unknown HR 1.4 ( ) unknown -- HR 1.2 ( ) HR 1.5 ( ) HR 1.0 ( ) HR 1.2 ( ) HR 1.5 ( ) HR 1.1 (0.6 21) Fever trial cohort, 39 n=3, >38 ºC on admission unknown 0.9 ( ) unknown D-dimer retro cohort, 19 n= admission value T (μg/ml) 1.7 vs 1.9 V non significant p-value -- C-reactive protein retro cohort, 30 n= admission value (mg/l) 11.7 vs 9.3 V non significant p-value -- retro cohort, 47 n=88 7 value day 1 T (mg/l) 9.14 vs 9.01 V non significant p-value -- Hyponatremia retro cohort, 59 n=124 9 <135 mmol/l in first 3 days 32% 0.8 ( ) 1.4 ( ) High cardiac troponin 1 retro cohort, 27 n=94 7 admission value 0.5 μg/l 22% 1.3 ( ) unknown ACA = anterior cerebral artery; AcoA = anterior communicating artery; HR = hazard ratio; ICA = internal carotid artery; MCA = middle cerebal artery; mm = millimetre; pros cohort = prospective cohort study; retro cohort = retrospective cohort study; trial cohort = cohort study using patients from a trial population; VBA = vertebrobasilar artery. # Studies were presented in order of methodological quality. Ratios are odds ratios, or otherwise marked by HR (hazard ratio). All calculated odds ratios are marked by. T Temporal relation was unclear for some of the described variables, therefore only values on admission were included for this review. V Values in patients with DCI versus no DCI. 78

79 Predictors of delayed cerebral ischemia after SAH Chapter 5 this data, we calculated non-significant ORs between 1.0 and 1.5, two of which had a point estimate between 0.9 and 1.2 (neutral). One study described a protective effect of aspirin based on urine samples, 29 but this was not analysed in the other studies. The pooled OR for previous use of aspirin based on history was non-protective (1.3; 95% CI 0.9 to 1.9). For size of the aneurysm, one study suggested a higher risk in giant aneurysms compared to small ones. 39 However, no other study did a separate analysis for giant aneurysms, therefore this was not applied in Table 5.3. We found limited evidence for absence of an association with DCI for race, composite score of atherosclerosis, fever, hyponatremia, high cardiac troponin I, C-Reactive Protein and D- dimer. Inconsistent evidence (Table 5.4) With respect to age, 6 studies presented contradictory ORs (either younger and older age, or no difference), and the decades associated with peak risks of DCI differed as well. The analyses used in the included studies were too heterogeneous to pool the data. For history of cardiovascular disease both HQ-studies showed non-significant opposite results. For previous use of statins two HQ-studies suggested a protective role (ORs 0.1 and 0.7), whereas another HQ-study showed an opposite non-significant trend. The pooled OR of these HQ-studies was 0.8 (95% CI 0.5 to 1.4). For cocaine a HQ-study found a statistically significant OR >4. In one low quality study we calculated an OR<1, but in another we recalculated the RR of 2.8 into an OR of 9.1. Several EKG changes on admission were examined in a HQ-study showing positive hazard ratios for ST-segment changes, but a low quality study found an opposite (protective) OR for cerebral ischemia. DISCUSSION Strong evidence was found for an increased risk of DCI in smokers. Moderate evidence was found for an increased risk in patients with history of diabetes, hyperglycemia on admission, hydrocephalus, or early systemic inflammatory response syndrome. Furthermore, there is limited evidence based on multiple studies for a slightly increased risk in women and in patients with a history of hypertension. Data are too scarce to draw conclusions on other suggested predictors. In addition, we found strong evidence that location of the aneurysm was not associated with the occurrence of DCI. 79

80 Chapter 5 Predictors of delayed cerebral ischemia after SAH Table 5.4 Factors with inconsistent evidence FACTORS WITH INCONSISTENT EVIDENCE Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariate analysis Odds ratio (95% CI) in multivariate analysis Age trial cohort, 39 n=3, per decade -- peak risk between y with significant p-value y: 1.4 ( ) y: 1.7 ( ) y: 1.7 ( ) y: 1.3 ( ) 70 y: 1.4 ( ) trial cohort, 48 n= a) 65 years b) per decade 11% -- a) 0.9 ( ) b) non-significant p-value (no peak risk were found and no OR could be calculated) peak risk between y with significant p-value y: 1.2 ( ) y: 1.4 ( ) y: 1.3 ( ) y: 0.9 ( ) 70 y: 0.9 ( ) 0.9 ( ) pros cohort, 38 n=70 12 >50 years * 51% 1.6 ( ) unknown retro cohort, 13 n= >50 years 48% 0.5 ( ) RR 0.5 ( ) retro cohort, 31 n= >50 years * 62% 0.4 ( ) Similar retro cohort, 46 n= per 5 year increase -- unknown 1.03 ( ) retro cohort, 40 n=391 9 a) increasing age b) per decade trial cohort, 37 placebo n= a) -- b) peak risk between y with significant p-value (no OR could be calculated) 8 per decade -- peak risk between y with significant p-value (no OR could be calculated) retro cohort, 24 n=138 8 per decade -- peak risk >70 y with nonsignificant p-value (no OR could be calculated) ( )

81 Predictors of delayed cerebral ischemia after SAH Chapter 5 FACTORS WITH INCONSISTENT EVIDENCE Study design (Reference), no of patients Methodological Quality Score (range 0 15) # Definition variable Percentage of patients with variable Odds ratio (95% CI) in univariate analysis Odds ratio (95% CI) in multivariate analysis History of cardiovascular disease pros cohort, 22 n= positive history for stroke or myocardial infarction pros cohort, 38 n=70 12 positive history for coronary artery disease 9% 0.7 ( ) unknown 4% 5.4 (0.6 51) unknown Previous use of statins retro cohort, 52 n= documented admission medications 7% 1.6 ( ) 1.4 ( ) retro cohort, 41 n= documented admission medications (medication was used at least 1 month) 13% 0.1 ( ) 0.1 ( ) retro cohort, 42 n= documented admission medications 8% 0.7 ( ) non-significant p-value High quality pooled odds ratio 0.8 ( ) case control, 45 n=60 8 admission medications 33% 0.2 ( ) unknown Use of cocaine retro cohort, 15 n= recent use (mainly history) 6% 4.1 ( ) 4.1 ( ) retro cohort, 8 n=573 9 last 72 hours (history) or urine 5% 0.6 ( ) non-significant p-value retro cohort, 23 n=108 8 in last 24 hours (urine) 33% 9.1 (3.6 23) similar, see text Elektrokardiogram changes retro cohort, 50 n= ST-segment depression on admission ST-segment elevation on admission pros cohort, 58 n=149 9 QTc prolongation, arrhythmias, ischemia, left ventricle hypertrophy 14% 8% HR 2.4 ( ) HR 2.1 ( ) ( ) for ischemia unknown unknown unknown HR = hazard ratio; pros cohort = prospective cohort study; retro cohort = retrospective cohort study; RR = risk ratio; trial cohort = cohort study using patients from a trial population; y = years. # Studies were presented in order of methodological quality. Ratios are odds ratios, or otherwise marked by HR (hazard ratio) or RR (risk ratio). All calculated odds ratios are marked by. * Recorded OR was from the inverse variable (<50 years instead of >50 years), we recalculated the effect estimate so it was comparable with the other studies. 81

82 Chapter 5 Predictors of delayed cerebral ischemia after SAH We could not find other systematic reviews on this topic. One narrative review on prediction of vasospasm after SAH stated that large amount of extravasated blood was the only consistently demonstrated predictor for DCI. 60 Apart from lack of a systematic appraisal of the literature, other differences were the focus on cerebral vasospasm instead of DCI. Our study has some limitations. Many studies we included had a retrospective design or small study population. For the topic of this review there is a risk of publication bias, since studies with negative results are not always published. To limit this bias we also included articles from national and small journals. There was heterogeneity between studies with respect to the definitions of predictors, definitions of DCI and the analyses used in the included studies. Therefore it was not possible to perform a formal meta-analysis on all variables. In case of limited heterogeneity and available crude data we calculated a pooled OR using only the high quality studies. For several variables however, data were too heterogeneous to pool. For example in the studies on age, some studies calculated ORs for separate decades, some dichotomized age at 65 years, some at 50, whereas others analyzed age as a continuous variable per 5 years increase. In such instances it is impossible to pool the data. We therefore chose a systematic review with classification of the level of evidence. For age the results were conflicting (both positive and negative associations). Therefore, age was classified as factor with inconsistent evidence. Our method has the limitation that a variable is immediately categorized as inconsistent evidence as soon as one effect estimate points in the opposite direction. However, if the results in a majority of studies are in line, such a variable could still be relevant. Another limitation is the inclusion of variables that are routinely available on admission or within the first 3 days. Other assessments, such as findings from echocardiography, might be predictive but fall beyond the scope of the present review. We assessed prediction based on crude numbers and unadjusted effect estimates. Multivariable analyses (if mentioned) were adjusted very heterogeneously, and thus could not be compared or recalculated. The independency of predictors could therefore not be assessed in this review setting. We feel that the restriction to routinely available predictors is one of the strengths of our study, because the results can be applied in a wide range of countries by clinicians treating SAH patients. Another strength is that two thirds of the included studies fulfilled our predefined criteria for high methodological quality. Mainly on the basis of these high methodological quality studies we ranked our findings into an estimated level of evidence. For several of the described variables more research is needed to assess whether or not they are predictors of DCI; these include early SIRS, hyperglycemia on admission and acute hydrocephalus. Because these variables are all frequently present in SAH patients, they could be of importance on top of the known predictors. The evidence for early SIRS and hyperglycemia 82

83 Predictors of delayed cerebral ischemia after SAH Chapter 5 may be underrated in our review due to our time limit of 3 days since onset of SAH. We excluded several positive reports and analyses on hyperglycemia and SIRS because the measurements were (partly) measured after 3 days of onset. Though history of diabetes is not frequently present in SAH patients, 61 its high pooled OR makes it interesting for further research as well. The predictive independency of history of diabetes additional to admission hyperglycemia was not analysed in all studies, though diabetes was suggested to be independent of hyperglycemia in one study. 18 Other striking effect estimates were found for initial loss of consciousness, history of cardiovascular disease and use of cocaine. Furthermore history of migraine, previous of use of SSRIs, low hemoglobin and hypomagnesemia appear worthwhile to analyse in future. Although the level of evidence for previous use of statins was classified as inconsistent, we feel that it could possibly be of value, because the protective influence was in line in two high quality studies and one low quality study. The opposite OR in the fourth study might be explained by the abrupt discontinuation of statins after admission. Abrupt withdrawal may lead to rebound effects. 52 The effect of statin therapy on DCI is now tested in a large trial in the UK. (Simvastatin in Aneurysmal Subarachnoid Haemorrhage (STASH) a Multicentre Randomised Controlled Clinical Trial.) We included only few data on TCD measurements and hyponatremia, although both are frequently studied. Most studies on TCD included data beyond the initial 3 days after the SAH or used TCD as a diagnostic and not a prognostic tool. For hyponatremia we had to exclude most studies because analyses were on late-onset-hyponatremia, or because of an uncertain temporal relation with development of DCI. No further research is necessary on prediction of DCI using the location of the aneurysm, since we found strong evidence for absence of an association. Although not categorised as strong evidence, based on the findings in HQ-studies we also feel that previous use of aspirin is unlikely to be predictor of great importance. Though one included study on previous use of aspirin suggested a large protective effect (based on data of urine samples), 29 none of the other studies found a protective effect. In conclusion, smoking is an established predictor of DCI, in addition to amount of subarachnoid blood and clinical condition on admission. For other predictors more evidence is needed before they can be used in clinical practice. Future research should preferably consist of representative patient populations and take into account the temporal relation with clinical DCI. Finally, the predictive value and the independency of predictors should be assessed to create a valid clinical prediction model for development of DCI in patients with SAH. 83

84 Chapter 5 Predictors of delayed cerebral ischemia after SAH ACKNOWLEDGEMENTS We would like to thank M.D.I. Vergouwen (Utrecht Stroke Center, Department of Neurology and Neurosurgery, the Netherlands) for his comments on the manuscript. REFERENCES 1. Roos YB, de Haan RJ, Beenen LF, et al. Complications and outcome in patients with aneurysmal subarachnoid haemorrhage: a prospective hospital based cohort study in the Netherlands. J Neurol Neurosurg Psychiatry 2000;68: van Gijn J, Kerr RS, Rinkel GJ. Subarachnoid haemorrhage. Lancet 2007;27;369: Adams HP, Jr, Kassell NF, Torner JC, Haley EC, Jr. Predicting cerebral ischemia after aneurysmal subarachnoid hemorrhage: influences of clinical condition, CT results, and antifibrinolytic therapy. A report of the Cooperative Aneurysm Study. Neurology 1987;37: Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980;6: Post B, Merkus MP, de Haan RJ, Speelman JD. Prognostic factors for the progression of Parkinson s disease: a systematic review. Mov Disord 2007;22: Altman DG. Systematic reviews of evaluations of prognostic variables. BMJ 2001;323: Borghouts JA, Koes BW, Bouter LM. The clinical course and prognostic factors of non-specific neck pain: a systematic review. Pain 1998;77: Alaraj A, Wallace A, Mander N, et al. Outcome following symptomatic cerebral vasospasm on presentation in aneurysmal subarachnoid hemorrhage: coiling vs. clipping. World Neurosurg 2010;74: Ameriso SF, Wong VL, Ishii H, et al. Hematogenous factors and prediction of delayed ischemic deficit after subarachnoid hemorrhage. Stroke 1992;23: Badjatia N, Topcuoglu MA, Buonanno FS, et al. Relationship between hyperglycemia and symptomatic vasospasm after subarachnoid hemorrhage. Crit Care Med 2005;33: Bakker AM, Dorhout Mees SM, Algra A, Rinkel GJ. Extent of acute hydrocephalus after aneurysmal subarachnoid hemorrhage as a risk factor for delayed cerebral infarction. Stroke 2007;38: Carrera E, Schmidt JM, Oddo M, et al. Transcranial Doppler ultrasound in the acute phase of aneurysmal subarachnoid hemorrhage. Cerebrovasc Dis 2009;27: Charpentier C, Audibert G, Guillemin F, et al. Multivariate analysis of predictors of cerebral vasospasm occurrence after aneurysmal subarachnoid hemorrhage. Stroke 1999;30: Collignon FP, Friedman JA, Piepgras DG, et al. Serum magnesium levels as related to symptomatic vasospasm and outcome following aneurysmal subarachnoid hemorrhage. Neurocrit Care 2004;1:

85 Predictors of delayed cerebral ischemia after SAH Chapter Conway JE, Tamargo RJ. Cocaine use is an independent risk factor for cerebral vasospasm after aneurysmal subarachnoid hemorrhage. Stroke 2001;32: Dhar R, Diringer MN. The burden of the systemic inflammatory response predicts vasospasm and outcome after subarachnoid hemorrhage. Neurocrit Care 2008;8: Dreier JP, Kremer C, Lammers G, et al. Migraine and delayed ischaemic neurological deficit after subarachnoid haemorrhage in women: a case-control study. Eur J Neurol 2007;14: Dumont T, Rughani A, Silver J, Tranmer BI. Diabetes mellitus increases risk of vasospasm following aneurysmal subarachnoid hemorrhage independent of glycemic control. Neurocrit Care 2009;11: Fuji Y, Takeuchi S, Sasaki O. Serial changes hemostasis in aneurysmal subarachnoid hemorrhage with special reference to delayed ischemic neurological deficits. J Neurosurg 1997;86: Hijdra A, van Gijn J, Nagelkerke N, et al. Prediction of delayed cerebral ischemia, rebleeding and outcome after aneurysmal subarachnoid hemorrhage. Stroke 1988;19: Hirashima Y, Kurimoto M, Hori E, et al. Lower incidence of symptomatic vasospasm after subarachnoid hemorrhage owing to ruptured vertebrobasilar aneurysms. Neurosurgery 2005;57: Hop JW, Rinkel GJE, Algra A, van Gijn J. Initial loss of consciousness and risk of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Stroke 1999;30: Howington JU, Kutz SC, Wilding GE, Awasthi D. Cocaine use as a predictor of outcome in aneurysmal subarachnoid hemorrhage. J Neurosurg 2003;99: Inagawa T. Cerebral vasospasm in elderly patients with ruptured intracranial aneurysms. Surg Neurol 1991;36: Inagawa T. Site of ruptured intracranial saccular aneurysms in patients in Izumo City, Japan. Cerebrovasc Dis 2010;30: Inagawa T. Size of ruptured intracranial saccular aneurysms in patients in Izumo City, Japan. World Neurosurg 2010;73: Jeon IC, Chang CH, Choi BY, et al. Cardiac troponin I elevation in patients with aneurysmal subarachnoid hemorrhage. J Korean Neurosurg Soc 2009;46: Juvela S. Alcohol consumption as a risk factor for poor outcome after aneurysmal subarachnoid hemorrhage. BMJ 1992;304: Juvela S. Aspirin and delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. J Neurosurg 1995;82: Juvela S, Kuhmonen J, Siironen J. C-reactive protein as predictor for poor outcome after aneurysmal subarachnoid haemorrhage. Acta Neurochir (Wien ) 2012;154: Kale SP, Edgell RC, Alshekhlee A, et al. Age-Associated Vasospasm in Aneurysmal Subarachnoid Hemorrhage. [published online ahead of print June 28,2011]. J Stroke Cerebrovasc Dis Kerner A, Schlenk F, Sakowitz O, et al. Impact of hyperglycemia on neurological deficits and extracellular glucose levels in aneurysmal subarachnoid hemorrhage patients. Neurol Res 2007;29:

86 Chapter 5 Predictors of delayed cerebral ischemia after SAH 33. Kongable GL, Lanzino G, Germanson TP, et al. Gender-related differences in aneurysmal subarachnoid hemorrhage. J Neurosurg 1996;84: Kremer C, Groden C, Hansen HC, et al. Outcome after endovascular treatment of Hunt and Hess grade IV or V aneurysms: comparison of anterior versus posterior circulation. Stroke 1999;30: Krishnamurthy S, Kelleher JP, Lehman EB, Cockroft KM. Effects of tobacco dose and length of exposure on delayed neurological deterioration and overall clinical outcome after aneurysmal subarachnoid hemorrhage. Neurosurgery 2007;61: Kruyt ND, Roos YW, Dorhout Mees SM, et al. High mean fasting glucose levels independently predict poor outcome and delayed cerebral ischaemia after aneurysmal subarachnoid haemorrhage. J Neurol Neurosurg Psychiatry 2008;79: Lanzino G, Kassell NF, Germanson TP, et al. Age and outcome after aneurysmal subarachnoid hemorrhage: why do elder patients fare worse? J Neurosurg 1996;85: Lasner TM, Weil RJ, Riina HA, et al. Cigarette smoking-induced increase in the risk of symptomatic vasospasm after aneurysmal subarachnoid hemorrhage. J Neurosurg 1997;87: Macdonald RL, Rosengart A, Huo D, Karrison T. Factors associated with the development of vasospasm after planned surgical treatment of aneurysmal subarachnoid hemorrhage. J Neurosurg 2003;99: Magge SN, Chen HI, Ramakrishna R, et al. Association of a younger age with an increased risk of angiographic and symptomatic vasospasms following subarachnoid hemorrhage. J Neurosurg 2010;112: McGirt MJ, Blessing R, Alexander MJ, et al. Risk of cerebral vasopasm after subarachnoid hemorrhage reduced by statin therapy: A multivariate analysis of an institutional experience. J Neurosurg 2006;105: Moskowitz SI, Ahrens C, Provencio JJ, et al. Prehemorrhage statin use and the risk of vasospasm after aneurysmal subarachnoid hemorrhage. Surg Neurol 2009;71: , discussion. 43. Naidech AM, Drescher J, Ault ML, et al. Higher hemoglobin is associated with less cerebral infarction, poor outcome, and death after subarachnoid hemorrhage. Neurosurgery 2006;59: Niikawa S, Hara S, Ohe N, et al. Correlation between blood parameters and symptomatic vasospasm in subarachnoid hemorrhage patients. Neurol Med Chir (Tokyo) 1997;37: Parra A, Kreiter KT, Williams S, et al. Effect of prior statin use on functional outcome and delayed vasospasm after acute aneurysmal subarachnoid hemorrhage: a matched controlled cohort study. Neurosurgery 2005;56: Rabinstein AA, Friedman JA, Weigand SD, et al. Predictors of cerebral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2004;35: Rothoerl RD, Axmann C, Pina AL, et al. Possible role of the C-reactive protein and white blood cell count in the pathogenesis of cerebral vasospasm following aneurysmal subarachnoid hemorrhage. J Neurosurg Anesthesiol 2006;18: Ryttlefors M, Enblad P, Ronne-Engstrom E, et al. Patient age and vasospasm after subarachnoid hemorrhage. Neurosurgery 2010;67:

87 Predictors of delayed cerebral ischemia after SAH Chapter Schuiling WJ, de Weerd AW, Dennesen PJ, et al. The simplified acute physiology score to predict outcome in patients with subarachnoid hemorrhage. Neurosurgery 2005;57: Schuiling WJ, Algra A, de Weerd AW, et al. ECG abnormalities in predicting secondary cerebral ischemia after subarachnoid haemorrhage. Acta Neurochir (Wien ) 2006;148: Shiue I, Arima H, Hankey GJ, Anderson CS. Location and size of ruptured intracranial aneurysm and serious clinical outcomes early after subarachnoid hemorrhage: a population-based study in Australasia. Cerebrovasc Dis 2011;31: Singhal AB, Topcuoglu MA, Dorer DJ, et al. SSRI and statin use increases the risk for vasospasm after subarachnoid hemorrhage. Neurology 2005;64: Toussaint LG, III, Friedman JA, Wijdicks EF, et al. Influence of aspirin on outcome following aneurysmal subarachnoid hemorrhage. J Neurosurg 2004;101: van den Bergh WM, Algra A, Berkelbach van der Sprenkel JW, et al. Hypomagnesemia after aneurysmal subarachnoid hemorrhage. Neurosurg 2003;52: van den Bergh WM, Algra A, Elias R, Rinkel GJE. Extent of atherosclerosis and prognosis of patients with aneurysmal subarachnoid haemorrhage. Acta Neurochir (Wien ) 2006;148: Weir BK, Kongable GL, Kassell NF, et al. Cigarette smoking as a cause of aneurysmal subarachnoid hemorrhage and risk for vasospasm: a report of the Cooperative Aneurysm Study. J Neurosurg 1998;89: Yoshimoto Y, Tanaka Y, Hoya K. Acute systemic inflammatory response syndrome in subarachnoid hemorrhage. Stroke 2001;32: Yousef K, Crago E, Kuo CW, et al. Predictors of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage: a cardiac focus. Neurocrit Care 2010;13: Zheng B, Qiu Y, Jin H, et al. A predictive value of hyponatremia for poor outcome and cerebral infarction in high-grade aneurysmal subarachnoid haemorrhage patients. J Neurol Neurosurg Psychiatry 2011;82: Harrod CG, Bendok BR, Batjer HH. Prediction of cerebral vasospasm in patients presenting with aneurysmal subarachnoid hemorrhage: a review. Neurosurgery 2005;56: Adams HP, Jr, Putman SF, Kassell NF, Torner JC. Prevalence of diabetes mellitus among patients with subarachnoid hemorrhage. Arch Neurol 1984;41:

88 Chapter 5 Predictors of delayed cerebral ischemia after SAH 88

89 6 Early prediction of delayed cerebral ischemia after subarachnoid hemorrhage: development and validation of a practical risk chart Nicolien K. de Rooij, Jacoba P. Greving, Gabriël J.E. Rinkel, Catharina J.M. Frijns Submitted for publication in Stroke (in revision).

90 Chapter 6 Risk chart for prediction of DCI after SAH ABSTRACT Objective: To develop and validate a risk chart for prediction of delayed cerebral ischemia (DCI) after aneurysmal subarachnoid hemorrhage (SAH) based on admission characteristics. Methods: For derivation of the risk chart we studied data from 371 prospectively collected consecutive SAH patients with a confirmed aneurysm admitted between 1999 and For its validation we similarly studied 255 patients admitted between 2007 and The predictive value of admission characteristics was tested in logistic regression models with DCI related infarction as primary outcome. Procedure-related infarctions were not included. Performance of the models was tested by discrimination and calibration. Based on these models, a risk chart was developed for application in clinical practice. Results: The strongest predictors were clinical condition on admission, amount of blood on CT (both cisternal and intraventricular) and age. A model that combined these four predictors had an AUC of 0.63 (95% CI 0.57 to 0.69). This model improved little by including current smoking and hyperglycemia on admission (AUC 0.65; 95% CI 0.59 to 0.71). The risk chart predicted risks of DCI related infarction varying from 12 to 61%. Both low risk (<20% risk) and high risk (>40% risk) were predicted in approximately 20% of the patients. Validation confirmed that the discriminative ability was adequate (AUC 0.69; 95% CI 0.61 to 0.77). Conclusions: Absolute risks of DCI related infarction can be reliably estimated by a simple risk chart that includes clinical condition on admission, amount of blood on CT (both cisternal and intraventricular) and age. 90

91 Risk chart for prediction of DCI after SAH Chapter 6 INTRODUCTION Delayed cerebral ischemia (DCI) after aneurysmal subarachnoid hemorrhage (SAH) often leads to infarctions that are a major contributor to the high case fatality and morbidity of SAH, and occurs in about a third of the SAH patients. 1 Established predictors of DCI are large amounts of subarachnoid blood and poor clinical condition on admission; 2-4 other potential predictors are current smoking, diabetes, presence of hyperglycemia on admission, hydrocephalus on admission, or early systemic inflammatory response syndrome. 5 Previous prediction models for DCI after SAH are impractical in daily practice or not confined to data present at admission, and most are not validated. 6-8 Furthermore, a risk prediction chart does not exist. We aimed to develop and validate prognostic models based on admission characteristics to predict the risk of DCI related infarction, and to construct a risk chart that can easily be used in clinical practice. Additionally, we also evaluated the risk chart for clinical deterioration due to DCI. METHODS Study population All patients were derived from the prospectively collected cohort of SAH patients admitted to our hospital. The development cohort originated from a previously described study population of patients admitted between January 1999 and June Three-hundred-seventy-one patients met the following inclusion criteria: 1) subarachnoid hemorrhage confirmed by CT or lumbar puncture; 2) aneurysm proven by means of CT/MR- or catheter angiography; 3) admitted within three days after onset; 4) at least one follow-up scan, performed more than 24 hour after the initial scan; 5) initial CT scan available for review; 6) absence of a large intracerebral hemorrhage with surrounding hypodensity; 7) survival of the first 4 days after onset of SAH. The validation cohort consisted of 255 patients from the same prospectively collected cohort admitted between June 2007 to December The inclusion criteria were similar, except that patients with a large intracerebral hemorrhage were not excluded. Delayed cerebral ischemia The primary outcome measure in both cohorts was DCI related infarction. 9,10 This was assessed by two authors independently (NKR,CJMF) and defined as new spontaneous ischemic lesions on at least one follow-up scan. Only spontaneous infarctions (i.e. not related to clipping or 91

92 Chapter 6 Risk chart for prediction of DCI after SAH coiling of the ruptured aneurysm) within 28 days after SAH were included. Lesions caused by extraventricular drains, preexisting infarcts and hypodensities around a hematoma or in the vicinity of the operation area were also not considered new infarctions. Infarctions were considered to be related to treatment if new neurological symptoms occurred directly after aneurysm treatment, and an infarct was visible in the territory of the parent vessel of the aneurysm on CT within 48 h after treatment. In case the time of development of infarction was uncertain (i.e. if a new infarction was present on CT >48 h after treatment and/or the neurological state of the patient directly after treatment could not be assessed due to sedative medication), an expert opinion was made by two of the authors, who were not aware of the predictor profile. In this expert opinion the report of the operation or endovascular treatment was taken into account. In case of disagreement, the infarction was not counted as spontaneous infarction. In the validation cohort, we additionally assessed clinical deterioration due to DCI as secondary outcome measure. It was defined as decreased Glasgow coma scale (GCS) of at least 2 points lasting more than 2 hours, or a new focal deficit with exclusion of other causes (rebleed, hydrocephalus, epilepsy, metabolic or infectious causes), and was assessed by the same two authors independently. 10 Data collection The following admission data were recorded: age, gender, current smoking status, history of hypertension, history of vascular disease (including stroke, myocardial infarction and peripheral vascular disease), diabetes mellitus, presence of any loss of consciousness, clinical condition on admission, amount of blood on CT, presence of enlarged ventricles on CT, and blood levels for glucose and hemoglobin. The clinical condition on admission was recorded using the World Federation of Neurological Surgeons Scale (WFNS). 11 Amount of cisternal blood was dichotomized into thin blood (modified Fisher scale 0 to 2) and thick blood (modified Fisher scale 3 to 4). 12 Intraventricular blood was dichotomized at the median of the Hijdra scale. 13 The presence of hydrocephalus was assessed using the size of the lateral ventricles on the initial CT, and adjusted for age by means of the bicaudate index divided by the corresponding upper limit per age group. 14 An age-adjusted relative bicaudate index of >1 was regarded as presence of hydrocephalus on CT. Model development We selected predictors that could easily be determined within the first few hours of admission in all SAH patients. For consecutive models, we presented groups of predictors according to 92

93 Risk chart for prediction of DCI after SAH Chapter 6 the level of evidence in previous literature. Three prognostic models were defined. Model I was based on the two established predictors clinical condition on admission (WFNS scale) and amount of extravasated blood on CT (Modified Fisher scale and Hijdra scale) together with patient demographics (age and gender). In Model II candidate predictors were added of strong or moderate evidence based on a previous systematic review (current smoking, admission glucose level, hydrocephalus on initial CT). 5 In Model III, candidate predictors with limited evidence in our systematic review were added (history of hypertension, initial loss of consciousness, hemoglobin). 5 Restricted cubic spline functions and graphs were used to determine whether continuous variables (age, glucose, hemoglobin) could be analysed as linear terms or required transformation. 15 Missing values of patient characteristics were imputed by means of regression imputation. Logistic regression analysis was performed with DCI related infarction as outcome variable. All candidate predictors were included in a multivariable logistic regression model (irrespective of their univariable association with DCI) and were excluded step by step if the Wald test had a p-value above Model performance We evaluated both discrimination and calibration of the three models. The discriminative performance was described by an area under the receiver operating characteristic curve (AUC) with a corresponding 95% confidence interval (CI). Calibration was assessed with the Hosmer and Lemeshow test and visually with a calibration plot, plotting the observed outcomes versus predicted risks over quintiles of risks. Internal validation We internally validated our model with bootstrapping techniques where in each bootstrap sample the entire modelling process was repeated to correct for overestimation. 16 This resulted in a shrinkage factor for the regression coefficients. 17 The bootstrap procedure was also used to assess the AUC corrected for over-optimism. The corrected AUC may be considered as an estimate of discriminative ability expected in future similar patients. Model presentation Based on these risk prediction models we constructed a risk chart displaying absolute risks of DCI related infarction in SAH patients according to the absence or presence of the independent predictors. Also, we classified our patients into three risk groups of DCI related infarction: 93

94 Chapter 6 Risk chart for prediction of DCI after SAH Table 6.1 Baseline characteristics of the participants in the development and validation cohort Development cohort Validation cohort Year of admission to our hospital Jan 1999 June 2007 June 2007 Dec 2009 Number of patients Mean number of patients per year (range) 44 (33 67) 102 (86 99) DCI related infarction 110 (30%) 52 (20%) Clinical deterioration caused by DCI not registered 57 (22%) Women 261 (70%) 185 (73%) Age: median years (range) 55 (18 85) 56 (15 87) History of hypertension 102 (27%) 68 (27%) History of vascular disease 53 (14%) 53 (21%) History of diabetes 16 (4%) 10 (4%) Current smoking 193 (52%) 140 (61%) $ Glucose on admission: median (range), mmol/l 7.4 ( ) 7.2 ( ) Clinical condition on admission: WFNS scale I: Glasgow coma scale (GCS) 15 II: GCS without focal deficit III: GCS with focal deficit IV: GCS 7 12 V: GCS 3 6 Amount of blood: Modified Fisher 0 and 1: no/thin SAH and no IVH 2: thin SAH with IVH 3: thick SAH and no IVH 4: thick SAH with IVH Amount of blood: Hijdra scale Hijdra Cisternal, median (range) Hijdra Ventricles, median (range) Site of the aneurysm Posterior communicating artery Internal carotid artery Medial cerebral artery Anterior communicating artery Pericallosa artery Basilar or vertebral artery Posterior inferior cerebral artery Other Treatment of the aneurysm No treatment Coiling Clipping Other treatment (stent, embolization) 149 (40%) 88 (24%) 30 (8%) 53 (14%) 51 (14%) 29 (8%) 18 (5%) 111 (30%) 213 (57%) 24 (0 30) 2 (0 12) 88 (24%) 18 (5%) 49 (13%) 143 (39%) 10 (3%) 44 (12%) 11 (3%) 8 (2%) 47 (13%) 142 (38%) 182 (49%) 0 (0%) 107 (42%) 65 (26%) 20 (8%) 40 (16%) 23 (9%) 28 (11%) 17 (7%) 54 (21%) 156 (61%) 23 (0 30) 3 (0 12) 57 (22%) 17 (7%) 43 (17%) 99 (39%) 7 (3%) 14 (6%) 11 (4%) 7 (3%) 8 (3%) 131 (51.5%) 113 (44.5%) 3 (1%) GCS = Glasgow Coma Scale; WFNS = World Federation of Neurological Surgeons Scale; SAH = subarachnoid hemorrhage; IVH = intraventricular hemorrhage. $ Missing data in 10% of the patients. 94

95 Risk chart for prediction of DCI after SAH Chapter 6 low risk (<20%), average risk (20 to 40%) and high risk (>40%) and represented the risk classification in a colour scheme. Validation We evaluated both discrimination and calibration of the risk chart in the validation cohort, and assessed the number of patients with low or high risk of DCI related infarction. In addition, we performed similar analyses with the second outcome measurement (clinical deterioration due to DCI) to assess whether our risk chart was also useful for prediction of clinical DCI. RESULTS The baseline data of the development and validation cohorts are presented in Table 6.1. DCI related infarction occurred in 110 patients (30%) of the development cohort and 52 patients (20%) of the validation cohort. In the validation cohort clinical deterioration due to DCI occurred in 57 patients (22%). Prognostic models The multivariable models for prediction of DCI related infarction are presented in Table 6.2. The strongest predictors were clinical condition on admission (WFNS), amount of cisternal and intraventricular blood on CT, and age. For the combination of these four predictors (model I), the AUC after correction for optimism was 0.63 (95% CI 0.57 to 0.69). Adding smoking and hyperglycemia (model II) contributed little, the AUC after correction for optimism was 0.65 (95% CI 0.59 to 0.71). Model III had no additional predictive value. ROC curves and calibration plots for model I and II are given in Figure 6.1. Hosmer and Lemeshow tests were non-significant (p=0.377 in model I and p=0.548 in model II). Risk chart Because the first model had almost the same discriminatory performance as the second model (AUC 0.63 versus 0.65 with overlapping 95% CIs), we developed a risk chart on the basis of the first model with four instead of six variables. The risks of developing DCI related infarction ranged from 12% in an older patient with good clinical condition on admission (WFNS I), and no or thin amount of cisternal and intraventricular SAH on the initial CT, to 61% in a 95

96 Chapter 6 Risk chart for prediction of DCI after SAH Table 6.2 Multivariable predictors of delayed cerebral ischemia after subarachnoid hemorrhage in three predictive models Odds ratio (95% confidence interval) Model I # Model II # Demographic characteristics Young age (< median of 55 years) 1.37 ( ) 1.33 ( ) Gender X X Established predictors Clinical condition on admission WFNS I WFNS II III WFNS IV WFNS V Reference 1.37 ( ) 1.52 ( ) 3.18 ( ) Reference 1.28 ( ) 1.31 ( ) 2.29 ( ) Thick amount of cisternal blood on CT 1.77 ( ) 1.61 ( ) Thick amount of intraventricular blood on CT 1.56 ( ) 1.42 ( ) Candidate predictors Current smoking 1.48 ( ) Hyperglycemia (per mmol/l) 1.10 ( ) Enlarged ventricles (hydrocephalus) X Crude AUC 0.66 ( ) 0.69 ( ) AUC adjusted for optimism $ 0.63 ( ) $ 0.65 ( ) $ AUC = area under the receiver operating curve; CT = computed tomographic; WFNS = World Federation of Neurological Surgeons Scale; X = Factor is excluded because the Wald test had a p-value above # Both models are after adjustment for overfitting by shrinkage (regression coefficients were shrunk by 16% in the first model, and 21% in the second model). Thick amount of cisternal blood on CT defined as in the Modified Fisher Scale, score 3 4, focal or diffuse thick SAH. Thick amount of intraventricular blood defined as intraventricular Hijdra scale of at least 3 (> median). $ Adjusted for optimism with bootstrapping techniques. younger patient with poor clinical condition on admission (WFNS V), and large amounts of cisternal and intraventricular blood on the initial scan (Figure 6.2). A low risk of DCI related infarction was predicted in 87 patients (23%), an average risk in 203 patients (55%) and a high risk in 81 patients (22%). Validation The AUC for the risk chart in the validation group was 0.69 (95% CI 0.61 to 0.77) for DCI related infarction (Figure 6.3A), and 0.66 (95% CI 0.58 to 0.74) for clinical deterioration due to DCI. Outcomes in the validation cohort were systematically better than those predicted (Hosmer and Lemeshow test was p=0.003, Figures 6.3B). For clinical deterioration due to DCI 96

97 Risk chart for prediction of DCI after SAH Chapter 6 A. B. 100 Actual probability: observed DCI (%) Model I Model II Predicted probability (%) Figure 6.1 Discrimination and calibration plot for both predictive models (development cohort). A: The ROC curves show the discrimination of both predictive models in the development cohort. B: The plot shows the calibration (actual outcome versus predicted outcome) analyzed in 5 equal groups of both predictive models. 97

98 Chapter 6 Risk chart for prediction of DCI after SAH the overestimation of the risk chart was slightly lower than for DCI related infarction (data not shown). A low risk of DCI related infarction was predicted in 55 patients (22%), an average risk in 158 patients (62%) and a high risk in 42 patients (16%) (Table 6.3). AMOUNT OF CISTERNAL BLOOD ON INITIAL CT 1 THIN SAH (Modified Fisher scale 0 2) THICK SAH (Modified Fisher scale 3 4) I 12% 17% 19% 27% II/III 15% 22% 24% 33% IV 17% 24% 26% 36% V 29% 39% 42% 54% 55 years WFNS 2 AGE I 15% 22% 24% 33% <55 years II/III 20% 28% 30% 41% IV 22% 30% 33% 43% V 36% 47% 50% 61% THIN THICK THIN THICK AMOUNT OF INTRAVENTRICULAR BLOOD ON INITIAL CT 3 <20%, low risk 20 40%, average risk >40%, high risk Figure 6.2 Predicted probabilities of DCI after SAH for each combination of the main four independent predictors present on admission. 1 Based on the Modified Fisher scale. Thick clot = Modified Fisher 3 and 4. 2 World Federation of Neurological Surgeons Scale (WFNS) represents the clinical condition on admission. WFNS I = Glasgow coma scale (GCS) 15; WFNS II/III = GCS (with or without focal deficit); WFNS IV = GCS 7 12; and WFNS V = GCS Dichotomized at the median of the intraventricular Hijdra scale. Thick blood = intraventricular Hijdra scale of at least 3. In other words, in case at least 1 of the 4 ventricles was completely filled with blood; or in case at least 3 of the 4 ventricles contained a spot/sedimentation of blood; or at least 1 of the 4 ventricles was partly filled with blood and another contained sedimentation. 98

99 Risk chart for prediction of DCI after SAH Chapter 6 A. B. Actual probability: observed DCI Risk Chart in Validation Cohort Predicted probability (%) Figure 6.3 Discrimination and calibration plot of the risk chart in the validation cohort. A: The ROC curve shows the discrimination of the risk chart (Model I) in prediction of DCI related infarction in the validation cohort. B: The plot shows the calibration of the risk chart (actual outcome versus predicted outcome) in the validation cohort for DCI related infarction. For clinical deterioration due to DCI, the numbers of observed DCI were slightly higher in the first three groups. 99

100 Chapter 6 Risk chart for prediction of DCI after SAH Table 6.3 Prediction and actual occurrence of delayed cerebral ischemia related infarction in low and high risk patients in both cohorts Group Predicted risk Development cohort # Validation cohort # Mean predicted n Observed Mean predicted n Observed 1 <20% (low risk) 2 20 to 40% (average risk) 3 >40% (high risk) 17.1% % 16.7% % 27.9% % 29.5% % 47.4% % 47.0% % # According to model I. Total 29.6% (12 61%) % 29.6% (12 61%) % DISCUSSION We developed a practical risk chart to predict absolute risks of DCI in individual patients with SAH, based on four predictors that can easily be retrieved on admission. The model categorizes around 20% of the patients to have a low risk (<20% risk) and another 20% to have a high risk of DCI (>40% risk). Validation confirmed that the discriminative ability was adequate (AUC 0.69). The largest amount of prognostic information consisted of a set of four predictors: clinical condition on admission, amount of cisternal and amount of intraventricular blood on CT, and age. A recent study found exactly the same predictors (good clinical condition on admission, small amounts of extravasated blood and older age) to be associated with a low risk of DCI. 8 However, since the authors chose to predict 100% absence of DCI risk, the cut off points needed for that prediction applied to only 12 (4%) of their 307 patients. Another study reported thickness of clot, high flow on transcranial Doppler (TCD), clinical condition (GCS<14) and ruptured aneurysm in the anterior and carotid circulation as main predictors. 6 The definition of DCI used was ambiguous (including clinical symptoms like headache, stiff neck or low grade fever as symptoms of DCI), and it was not possible to make a prediction on admission, because also TCD measurements until day 5 were included in the model. In a small series of 68 patients a high AUC of 0.90 was found, but this model included technical demanding factors (i.e. Lindegaard ratio using cerebral blood flow evaluation with Xenon clearance technique) that is not commonly available in SAH patients. 7 In yet another study, the authors designed an artificial neural network. 18 From the 15 variables included in their 100

101 Risk chart for prediction of DCI after SAH Chapter 6 model, the majority was similar to our study variables, but they also added variables that were not yet available on admission, like elevated TCD velocities until day 5, aneurysm treatment modality and ventricular drainage. They found an extremely high AUC of Up to now the approach of designing an artificial neural network is controversial due to its proneness to overfitting. 19 Thus, to our knowledge, we developed the first risk chart for prediction of DCI using easily retrievable data available on admission. Though the discriminative performance of the risk chart was validated adequately, a limitation of our study is that the calibration shows better outcomes in the validation cohort than predicted. This overestimation is caused by the lower incidence of DCI in the validation cohort, which we believe is largely due to two main differences between the two cohorts. The first difference is the change in policy of follow-up scanning over time in our hospital, with more patients in good clinical condition undergoing follow-up imaging in recent years because of assessment of aneurysm occlusion. This larger proportion of patients with follow-up imaging despite an uneventful clinical course has influenced the rate of patient inclusion, and has inevitably decreased the proportion of patients with DCI over time. The second important difference between the both cohorts concerns the treatment of the ruptured aneurysms. In the validation cohort more patients were coiled, and only 3% of patients were left with untreated aneurysms (Table 6.1). Several studies showed that coiling is associated with a lower risk of DCI compared with clipping. 20,21 Interobserver variability as an explanation for the lower percentage of DCI related infarction is not likely since in both cohorts it was assessed by the same two authors using exactly the same definition. A second limitation is that the AUC of below 0.70 may appear to be low. However, besides accuracy of discrimination (high AUC), the practical value of a model depends on other items, such as the potentials for extrapolation, relevance of the outcome, and usability of the model. 22 The Framingham risk model, for instance, discriminates only moderately in certain (sub)populations with an AUC of little over 0.70, but is nonetheless widely used. 23 One of the strengths of our study is that we developed several models in a large cohort of SAH patients, and also validated our findings. Analyses were performed using up to date methods for development of prognostic models, using backward elimination, shrinkage factors and correction for over-optimism. We used predictors that are well known from the literature. 5 Finally, we developed a simple practical risk chart that can easily be used on admission to predict absolute risks of DCI in individual patients with SAH. Before application of the risk chart in clinical practice, it should be externally validated in another hospital setting, preferably in another health care system. 101

102 Chapter 6 Risk chart for prediction of DCI after SAH REFERENCES 1. Roos YB, de Haan RJ, Beenen LF, et al. Complications and outcome in patients with aneurysmal subarachnoid haemorrhage: a prospective hospital based cohort study in the Netherlands. J Neurol Neurosurg Psychiatry 2000; 68(3): Adams HP, Jr, Kassell NF, Torner JC, Haley EC, Jr. Predicting cerebral ischemia after aneurysmal subarachnoid hemorrhage: influences of clinical condition, CT results, and antifibrinolytic therapy. A report of the Cooperative Aneurysm Study. Neurology 1987; 37(10): Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980; 6(1): van Gijn J, Kerr RS, Rinkel GJ. Subarachnoid haemorrhage. Lancet 2007; 369(9558): de Rooij NK, Rinkel GJ, Dankbaar JW, Frijns CJ. Delayed cerebral ischemia after subarachnoid hemorrhage: a systematic review of clinical, laboratory and radiological predictors. Stroke 2013; 44: Qureshi AI, Sung GY, Razumovsky AY, et al. Early identification of patients at risk for symptomatic vasospasm after aneurysmal subarachnoid hemorrhage. Crit Care Med 2000; 28(4): Gonzalez NR, Boscardin WJ, Glenn T, et al. Vasospasm probability index: a combination of transcranial doppler velocities, cerebral blood flow, and clinical risk factors to predict cerebral vasospasm after aneurysmal subarachnoid hemorrhage. J Neurosurg 2007; 107(6): Crobeddu E, Mittal MK, Dupont S, et al. Predicting the lack of development of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Stroke 2012; 43(3): de Rooij NK, Frijns CJ, Velthuis BK, Rinkel GJ. Secondary infarction in single or in multiple vascular territories: two different entities following subarachnoid hemorrhage? J Neurol 2011; 258(12): Vergouwen MD, Vermeulen M, van Gijn J, et al. Definition of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage as an outcome event in clinical trials and observational studies: proposal of a multidisciplinary research group. Stroke 2010; 41(10): Report of World Federation of Neurological Surgeons Committee on a Universal Subarachnoid Hemorrhage Grading Scale. J Neurosurg 1988; 68(6): Frontera JA, Claassen J, Schmidt JM, et al. Prediction of symptomatic vasospasm after subarachnoid hemorrhage: the modified fisher scale. Neurosurgery 2006; 59(1): Hijdra A, Brouwers PJ, Vermeulen M, van Gijn J. Grading the amount of blood on computed tomograms after subarachnoid hemorrhage. Stroke 1990; 21(8): Hasan D, Vermeulen M, Wijdicks EF, et al. Management problems in acute hydrocephalus after subarachnoid hemorrhage. Stroke 1989; 20(6): Royston P, Moons KG, Altman DG, Vergouwe Y. Prognosis and prognostic research: Developing a prognostic model. BMJ 2009; 338:b Altman DG, Vergouwe Y, Royston P, Moons KG. Prognosis and prognostic research: validating a prognostic model. BMJ 2009; 338:b

103 Risk chart for prediction of DCI after SAH Chapter Harrell FE, Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996; 15(4): Dumont TM, Rughani AI, Tranmer BI. Prediction of symptomatic cerebral vasospasm after aneurysmal subarachnoid hemorrhage with an artificial neural network: feasibility and comparison with logistic regression models. World Neurosurg 2011; 75(1): Tu JV. Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes. J Clin Epidemiol 1996; 49(11): Dumont AS, Crowley RW, Monteith SJ, et al. Endovascular treatment or neurosurgical clipping of ruptured intracranial aneurysms: effect on angiographic vasospasm, delayed ischemic neurological deficit, cerebral infarction, and clinical outcome. Stroke 2010; 41(11): Dorhout Mees SM, Kerr RS, Rinkel GJ, et al. Occurrence and impact of delayed cerebral ischemia after coiling and after clipping in the International Subarachnoid Aneurysm Trial (ISAT). J Neurol 2012; 259(4): Moons KG, Altman DG, Vergouwe Y, Royston P. Prognosis and prognostic research: application and impact of prognostic models in clinical practice. BMJ 2009; 338:b Liao Y, McGee DL, Cooper RS, Sutkowski MB. How generalizable are coronary risk prediction models? Comparison of Framingham and two national cohorts. Am Heart J 1999; 137(5):

104 Chapter 6 Risk chart for prediction of DCI after SAH 104

105 7 General discussion

106 Chapter 7 General discussion The studies reported in this thesis focus on the incidence of aneurysmal subarachnoid hemorrhage (SAH), the risk of intracranial aneurysm rupture, and the risk of delayed cerebral ischemia (DCI) following SAH. In this chapter the main results of this thesis are summarised, put into perspective, and implications for research and clinical practice are outlined. SAH caused by the rupture of an intracranial aneurysm accounts for approximately 5% of all strokes. Patients with SAH are of particular interest within the subgroup of stroke patients because SAH occurs at a relatively young age (mean age of 55 years) and has a high case fatality. The loss of productive life years from SAH in the general population is as large as that from cerebral infarction, the most common type of stroke. 1 Patients with SAH still have a poor prognosis: almost 40% of all patients die, and of the surviving patients about 20% remain dependent for daily activities. A major cause of this poor outcome is delayed cerebral ischemia (DCI). DCI is one of the main complications after SAH and develops in about 30% of the patients, but up to now, it is not possible to predict easily and accurately which patients will be affected. The aim of our research was to gain insight in both the incidence and the risk of aneurysm rupture, and in the risk of DCI. The first part of this thesis focuses on the demographical differences in the incidence of aneurysmal SAH, the time trend of SAH incidence, and the etiological pathway of aneurysm rupture by analysing new anatomical risk factors. By generating new insights on aneurysm rupture, we intend to reduce the number of SAH patients in future. In the second part of this thesis we focussed on the pathophysiology and early prediction of DCI. By generating new insights on pathophysiology and better prediction of DCI, we intend to improve outcome after SAH in future. PART I Incidence of subarachnoid hemorrhage and risk of aneurysm rupture INCIDENCE OF SAH Chapter 2 emphasizes new insights on incidence of SAH. Besides factors as role of region, age and gender, our study also concerned the time trend of SAH incidence. For stroke in general, the incidence has declined over the past decade, which has been attributed to a decreasing proportion of people who smoke and to better detection and treatment of hypertension

107 General discussion Chapter 7 Because smoking and hypertension are risk factors for SAH as well, we hypothesized a similar decline in the incidence of SAH. We updated a previous review with a meta-analysis on prospective population based studies. 3 Our update included 51 studies (33 new) from more than 20 different countries, observing more than 45,000,000 person years. We found that a wide variation exists in the incidence of SAH. The overall incidence of SAH was approximately 9 per 100,000 person years but varied significantly by region. Our update also included data from additional parts of the world compared with the previous version; only African, South Asian and Chinese populations were not represented. INCIDENCE OF SAH: INFLUENCES OF REGION A main finding was the statistically significantly doubled incidences in Japan (22.7 per 100,000 person-years) and Finland (19.7), whereas in South and Central America the incidence was twice as low (4.2) compared with other regions. We discussed the possible influences of differences with respect to demographics, use of CT for diagnosis and method of case finding in these countries compared to other regions. However, we state that none of these explanations can completely explain the regional differences. Explanations for the high incidence in Finland include high prevalence of smoking and hypertension, 4 and heavy episodic alcohol abuse.5 Moreover, recently it has been found that it is not the prevalence of aneurysms, 6 but the risk of rupture that is higher in Japan and Finland. 7 We speculate that genetic factors may also play an important role. 8 Recently it was found that the combination of genetics and smoking results in a much higher risk of SAH, and a gene-environment interaction with smoking is suggested. 9 This cumulative effect could play a role in Japan and Finland. INCIDENCE OF SAH: INFLUENCES OF AGE AND GENDER Another main finding was that incidence of SAH increases with age. Moreover, we confirmed the suggestion that the overall incidence is higher in women, but we also found that the gender distribution varies with age. We found that the preponderance of women starts only after the sixth decade: at younger ages the incidence is higher in men, whereas after the age of 55 years, the incidence is higher in women. The reasons for the overall higher incidence in women are not clear, but hormonal factors and postmenopausal status are a possible explanation A recent systematic review described an increased risk of SAH for postmenopausal compared 107

108 Chapter 7 General discussion to premenopausal women, pointing to an effect of female hormone levels on the risk of SAH, but it remains unclear by which mechanism. 13 Their meta-analysis found no evidence for an increased risk for current use of combined oral contraceptives. Another possible explanation for the gender gap in SAH incidence is that risk implications of currently recognized risk factors for SAH such as cigarette smoking, excessive use of alcohol, hypertension and familial preponderance are stronger in women than in men. A prospective cohort study showed that compared with the risk in nonsmokers, the risk of SAH was higher in current-smoking women than in men. 14 INCIDENCE OF SAH: RELATIVELY MODERATE DECREASE OVER TIME We found that the incidence of SAH has decreased by 0.6% (95% CI between 1.3% decrease to 0.1% increase) per year between 1950 and While the incidence of stroke in general decreased with approximately 2% per year over the past two decades, the decline in incidence of SAH over the past 45 years is relatively moderate. The influences of region, age, gender and improved diagnostic criteria by CT were taken into account in our analyses on time trend. The small magnitude of the decline in incidence of SAH may in part be explained by the stronger influence of genetic factors in SAH than in stroke in general. 18 Although genetic factors explain only 10% of SAH, and most instances are attributed to smoking, hypertension and excessive use of alcohol, 18 a gene-environment interaction with smoking could play a role in the different decline in SAH incidence. 9 Lessons learned There is a wide variation in the incidence of SAH. The overall incidence of SAH is approximately 9 per 100,000 person-years, but varies in specific regions. The incidence rates are doubled in Japan and Finland, and far lower in South and Central America. The incidence is higher in women and increases with age. The gender distribution varies with age. At young ages, the incidence is higher in men, whereas after the age of 55 years it is higher in women. The incidence of SAH has probably decreased slightly over the past 45 years. 108

109 General discussion Chapter 7 Implications for future research Further studies should address the reasons for the relatively moderate decline in incidence of SAH, the higher incidence in women after the fifth decade and the regional differences in SAH incidence. The answers to these questions will probably provide further clues to the etiology of SAH. Because in Finland and Japan the incidence is proven to be far higher, we suggest that genetic, environmental and perhaps also dietary factors play an important role in the pathophysiology of SAH. The higher incidence of SAH in older women suggests a relation with hormones and postmenopausal status. Research into hormonal and genetic factors may contribute insights into the etiology and pathogenesis of development of aneurysms and occurrence of SAH. Since joint effects are a possible explanation, cumulative effects of gender, genetics and/or risk factors such as smoking, hypertension and alcohol abuse are of relevance for future research. RISK OF ANEURYSM RUPTURE: ANATOMICAL RISK FACTORS This part focuses on new findings in the etiological pathway of intracranial aneurysm rupture, with emphasis on anatomical differences visible on radiological imaging. Improved knowledge on factors that contribute to the risk of rupture may lead to more tailored aneurysm management. Three percent of the general population harbors an intracranial aneurysm, and in recent years the number of incidentally discovered unruptured intracranial aneurysms is rising with the increasing quality and availability of non-invasive imaging techniques. 6,19 In all patients with an incidental aneurysm, a clinical decision has to be made whether to treat the aneurysm or not. Therefore, the risk of rupture needs to be balanced against the risks of treatment. 20 To determine which aneurysms should undergo preventive treatment or warrant more frequent follow-up, knowledge on risk factors for rupture is crucial. Currently few risk factors for rupture have been identified: gender, age, world region (Japan and Finland), size, site and type of the aneurysm, and possibly previous rupture of another aneurysm. 7 Up to now, risk prediction for aneurysm rupture in the individual patient is poor. In Chapter 3 we analyzed whether morphological characteristics of cerebral arteries and aneurysms are factors in the etiological pathway of aneurysm rupture. We therefore reviewed CT angiograms of matched patients with ruptured and unruptured aneurysms. For all morphological characteristics studied, we calculated odd ratios (ORs) with corresponding 95% confidence intervals for risk of rupture. Configuration of the circle of Willis was analyzed on a per site basis. In additional analyses, we adjusted for size by means of multivariable logistic regression. 109

110 Chapter 7 General discussion ANEURYSM RUPTURE: NO ASSOCIATION WITH CONFIGURA- TION OF THE CIRCLE OF WILLIS Our first question was whether configuration (asymmetry and incompleteness) of the circle of Willis is important in the etiological pathway of aneurysm rupture. We know these variations of the anatomy may lead to increased flow at the contralateral site. 21 and that asymmetry of the proximal segment of the anterior cerebral artery is more often found in patients with ruptured aneurysms of the anterior communicating artery than in patients with ruptured aneurysms at another site. 22 Incompleteness of the anterior cerebral arteries (no visualization of a unilateral A1 segment) is more often found in patients with unruptured anterior communicating artery aneurysms than in patients with unruptured aneurysms at another site or patients without aneurysms. 23 The relationship of these configurations of the circle of Willis and risk of rupture, however, was unknown up to now. We found no statistically significant differences for configurations of the circle of Willis between patients with ruptured or unruptured aneurysms. This suggests that the configurations studied (asymmetry and incompleteness of the circle of Willis) are not strong risk factors for rupture, though the number of aneurysms per site in our study is too small to exclude moderate or small associations with rupture. ANEURYSM RUPTURE: FLOW INTO THE ANEURYSM AND AN- EURYSM SHAPE Our second question was whether straight direction of flow into the aneurysm plays a role in the risk of aneurysm rupture. This was already described as a possible hemodynamic risk factor for formation of aneurysms. A mathematical simulation study suggests that pressures and shear stresses developing along the outer wall of curved arteries and at the apex of arterial bifurcations create a hemodynamic state that promotes aneurysm formation, and funded the relation of fluid flow in curved arteries and arterial bifurcations with intracranial aneurysm formation, growth and subsequently rupture. 24 Aneurysms that develop along the outer wall of curved arteries or at the apex of an arterial bifurcation, have a relatively sharp angle with the main artery or branches, and the flow direction will be straight into the aneurysm. Our study shows that flow straight into the aneurysm was more commonly observed in ruptured aneurysms compared with unruptured aneurysms, with a statistically significant OR of 2.0. This suggests that the hemodynamic state of curved arteries with flow direction straight into the aneurysm influences the risk of rupture of the aneurysm. To our knowledge, the relation between direction of blood flow into the aneurysm and risk of rupture has not been studied before on angiograms. 110

111 General discussion Chapter 7 Our third question was whether the risk of rupture was increased in aneurysms with multilobed or elliptical shape (non-spherical), since the distribution of wall tension is more heterogeneous compared with spherical shape. Our study showed that non-spherical shape was more commonly observed in ruptured aneurysms compared with unruptured aneurysms, with a statistically significant OR of 2.8. The increasing risk of rupture found for increasing deviation of shape from spherical, further strengthens our finding that non-spherical shape contributes to the risk of rupture. Shape of the aneurysm was studied before as a risk factor for rupture by other authors. In a small study on 27 patients with aneurysms, several shape factors appeared to be more effective than size in discriminating between ruptured and unruptured aneurysms. 25 The shape factors studied were aspect ratio (a ratio of depth to neck width of the aneurysm), undulation (including irregularities of the wall and lobulation), ellipticity and non-sphericity. Other studies found irregular multilobed appearance to be more common in ruptured aneurysms compared with unruptured aneurysms. 26,27 In none of these studies, however, cases and controls were matched for age, gender and site of the aneurysm. Since age and gender, and site of the aneurysm are risk factors for rupture, 7 lack of matching for these factors may introduce bias. Our cases and controls were matched for site of the aneurysm, gender and age of the patient, and year of CT angiogram. Moreover, we included a large number of patients with unruptured aneurysms to obtain precise and reliable estimates of the risk of aneurysm characteristics studied. A limitation is that we did not match for size of the aneurysm. Our aim was to analyze whether morphological characteristics of cerebral arteries and aneurysms play a role in the etiological pathway of aneurysm rupture. According to this hypothesis, size can be considered as intermediate in the pathway between development and aneurysm rupture. An aneurysm with flow straight into the aneurysm, non-spherical shape, or both, has a higher risk of enlarging, and thus larger size, and thereby a higher risk of rupture. For the morphological factors studied, size can be considered a result of the risk factor. We anticipated that after adjusting for size the effects of our determinants of interest on the risk of rupture would disappear, which they did when we performed these additional analyses. Therefore we conclude that these variables may be associated with an increased risk of rupture, but are related to aneurysm size. Besides shape and direction of flow there are other potential aneurysm risk factors for rupture. For example, irregularity of the aneurysm wall (i.e. blebs or nipples) and flow dynamics in the aneurysm (neck/dome ratio and asymmetry of the aneurysm). With improving vascular imaging methods, we assume that more morphological factors will be found to play a role in the flow dynamics, and subsequently in the rupture rate of intracranial aneurysms. 28,29 111

112 Chapter 7 General discussion Lessons learned Direction of flow into the aneurysm and non-spherical (both elliptical and multilobed) shape are morphological aneurysm characteristics that are important in the etiological pathway of rupture of intracranial aneurysms: both are more commonly observed in ruptured aneurysms compared with unruptured aneurysms. Direction of flow into the aneurysm and non-spherical (both elliptical and multilobed) shape may contribute to the risk of rupture, but are related to aneurysm size. Patients with an aneurysm with direction of flow into the aneurysm and/or non-spherical (both elliptical and multilobed) shape may warrant more frequent follow-up. Configuration of the circle of Willis was not associated with a strong risk of rupture; moderate risk could not be excluded Implications for future research In theory, the best method to confirm our results on risk of rupture would be new studies to assess a risk stratification in which these morphological characteristics are determined prospectively. However, because it would take many years to follow up the untreated patients with unruptured aneurysms (only a minority of these aneurysms will rupture in the first 5 years), this method is not feasible in practice. Therefore, we suggest a different prospective approach in patients with aneurysms that (initially) are not considered for preventive treatment. In these patients it would be interesting to investigate whether direction of flow into the aneurysm and non-spherical shape can predict growth or changing (instable) appearance of the aneurysm on follow-up imaging, and subsequently rupture. Furthermore, it should be investigated whether other aneurysm characteristics that influence flow dynamics play a role in aneurysm rupture as well. 112

113 General discussion Chapter 7 PART II Delayed cerebral ischemia following subarachnoid hemorrhage The lack of insight into the pathogenesis of DCI is challenging. Despite many years of research, the exact cause of DCI remains unknown. Although cerebral vasospasm has been indicated as the main cause of DCI, vasospasm is not the only or essential factor for the development of DCI. There is also a lack of knowledge on prediction of DCI. One third of the patients will develop DCI, but it is not possible to predict which patients. In the literature different definitions have been used to define DCI. Frequently, cerebral vasospasm is used in the definition of DCI. To avoid inconsistencies in definitions of DCI, in 2011 a multidisciplinary research group recommended two different definitions: cerebral infarction, after exclusion of procedure-related infarctions and clinical deterioration caused by DCI, after exclusion of other potential causes of clinical deterioration. 30,31 The authors concluded that vasospasm should be reserved for angiographic arterial narrowing only, and should not be used in the definition of DCI. We agree with this statement for several reasons. To begin with, DCI was independently related to poor outcome when cerebral infarction as seen on neuroimaging was included in the definition, whereas with use of radiographic definitions of vasospasm (either with angiography or TCD) no relation was found with poor outcome. 32 Additionally, vasospasm visible on radiological examinations is often seen in SAH patients, but does not always lead to DCI. 32,33 Moreover, we know that DCI can occur in the absence of vasospasm. 32,34 For a valid definition of cerebral infarction, the study group also proposed to exclude both infarctions present on CT or MR between 24 and 48 hours after aneurysm closure, and hypodensities on CT attributed to other causes (surgical clipping, endovascular treatment, ventricular catheter or intraparenchymal hematoma). 30 In the review study described in Chapter 5, we were confronted with the fact that not all studies used similar definitions for DCI. Due to this and other heterogeneities, in most instances it was not possible to pool the data in a formal meta-analysis. We had to use a different approach to summarize the level of evidence. In our studies on DCI we used as definition cerebral infarction due to DCI after exclusion of procedure-related infarctions, which is in line with the proposal of the multidisciplinary research group (Chapter 4 and 6). All CT scans were independently reviewed by two authors and a specialist team in case of disagreement. Lesions caused by extraventricular drains and hypodensities around a hematoma or in the vicinity of the operation area were not considered new infarctions, and all treatment related infarctions were meticulously excluded. In Chapter 6 we also used clinical deterioration due to DCI as 113

114 Chapter 7 General discussion secondary outcome to test the reliability of our results when using the radiological definition. In this part of the thesis we investigated the possibility of two pathophysiologically different infarct patterns of delayed cerebral ischemia (Chapter 4). Furthermore, we reviewed the evidence of all early and easily available predictors of DCI suggested in the past decades (Chapter 5). Next, based on the results in Chapter 5, we investigated the predictive values of the candidate predictors in several multivariable logistic models. We developed and validated the best model for prediction of DCI, and turned the model into a practical risk chart to be used by clinicians treating patients with SAH (Chapter 6). TWO TYPES OF INFARCTION: DIFFERENT DEGREES OF THE SAME PATHOPHYSIOLOGICAL PROCESS Chapter 4 affects the issue whether DCI may have several patterns, and if these patterns represent pathophysiologically different disease entities or different degrees of severity of the same vascular process. In this chapter, cerebral infarction due to DCI is mentioned secondary infarction (SI). Previous research suggested two common patterns of secondary infarctions. 35 We hypothesized that differentiation between two infarct patterns could lead to new clues in the pathophysiology of DCI. In this study, CT and MRI examinations of SAH patients were reviewed for new infarctions within 28 days after SAH, and categorized into secondary infarctions in single (SSI) or multiple (MSI) vascular territories. Only patients with adequate follow-up imaging were included for analysis. We compared demographic characteristics, vascular risk factors, disease-related characteristics and treatment modalities of patients with these patterns of SI with patients without infarction, to assess whether they represent different disease entities. The main finding of chapter 4 was that patients with both infarct patterns (single or multiple vascular territories) had a similar clinical profile. We found that MSI is related to the same characteristics as SSI but to a larger extent, specifically to presence of multiple vascular risk factors, initial loss of consciousness, larger amounts of intraventricular blood and poor clinical status on admission. This suggests that SSI and MSI do not represent two different disease entities, but are consequences of increasing severity of the same risk factors, and therefore are different degrees of the same pathophysiological process. Moreover, our results underline the importance of clinical risk factors for occurrence and extent of secondary infarction after SAH. 114

115 General discussion Chapter 7 Lessons learned In DCI, we can distinguish between infarctions in single and multiple cerebral vascular territories, but these are not distinct disease entities. Both seem to be different degrees of the same pathophysiological process. Clinical risk factors are important for both occurrence and extent of radiological DCI after SAH. EVIDENCE ON EARLY PREDICTORS OF DCI: CURRENT SMOKING Chapter 5 discusses the evidence on the many predictors or risk factors of DCI that have been reported. Up to now, the only established predictors of DCI were large amounts of subarachnoid blood and poor clinical condition on admission This review systematically summarizes all the evidence on other early available predictors. We searched MEDLINE (1960 January 2012) for clinical, laboratory and radiological predictors routinely available within 72 hour after SAH. Out of 7,311 records, 52 studies totalling 17,496 patients met our inclusion criteria. All 52 studies were categorised according to methodological quality, and 63% were rated as high quality. Crude data and effect estimates were extracted, (re-) calculated and pooled if possible. For every potential predictor we assessed all effect estimates on consistency (point estimates in equal direction) and clinical relevance (size and 95% CI) and translated the results into a level of evidence. The main finding of this study is the strong evidence (4 consistent high quality studies) for a higher risk of DCI in smokers, with a pooled OR of 1.2 (95% CI 1.1 to 1.4). Furthermore, for 4 factors we found at least 2 high quality studies with consistent results (moderate evidence), implying an increased risk in the patients with hyperglycemia on admission, enlarged ventricles on the initial CT scan (hydrocephalus), history of diabetes or early systemic inflammatory response syndrome (SIRS). Evidence was limited for increased risk in women and in patients with history of hypertension, initial loss of consciousness, history of migraine, previous use of SSRIs, hypomagnesemia, low hemoglobin, or high blood flow on early TCD. In addition, we found strong evidence that location of the aneurysm was not associated with the occurrence of DCI. Although not categorised as strong evidence, we also feel that previous use of aspirin is unlikely to be a predictor of great importance. Though one included study on previous use of aspirin suggested a large protective effect, none of the other 3 studies found a protective effect. 115

116 Chapter 7 General discussion Because there was heterogeneity between studies with respect to the definitions of predictors, definitions of DCI and the analyses used in the included studies, it was not always possible to perform a formal meta-analysis on all variables. An important strength of the review is that all studies were categorised according to methodological quality, and two thirds of the included studies fulfilled the predefined criteria for high methodological quality. Mainly on the basis of these high methodological quality studies we ranked our findings into an estimated level of evidence. We feel that the restriction to routinely available predictors is one of the strengths of our study, because the results can be applied by clinicians treating SAH patients in a wide range of countries. The independency of predictors could not be assessed in this review setting. Lessons learned Smoking is a predictor of DCI (strong evidence), in addition to amount of subarachnoid blood and clinical condition on admission. For several suggested predictors more research is needed to assess whether or not they are predictors of DCI. Early SIRS, hyperglycemia on admission, enlarged ventricles on CT (hydrocephalus), and history of diabetes may prove to have additional predictive value. Location of the aneurysm is not a valid predictor. Previous use of aspirin is unlikely to be a predictor of importance. Implications for future research Whereas evidence on the predictive value of current smoking was found to be strong, for other suggested predictors more evidence is needed before they can be used in clinical practice. This future research should preferably consist of representative patient populations and take into account the temporal relation with clinical DCI. We suggest more research on the following factors to assess whether or not they are predictors of DCI: early SIRS, hyperglycemia on admission and enlarged ventricles on CT. Because these variables are all frequently present in SAH patients, they could be of importance on top of the known predictors. Though history of diabetes is not frequently present in SAH patients, the high pooled OR (>6), makes it interesting for further research as well. For gender, age, and several other potential predictors evidence was limited or inconsistent, though a predictive value could not be excluded. No further research is necessary on prediction of DCI using the location of the aneurysm. 116

117 General discussion Chapter 7 A CLINICAL RISK CHART ON DCI: LOW AND HIGH RISK PATIENTS In Chapter 6 we developed and validated a clinical prediction model for development of DCI in patients with SAH. The ultimate goal was to generate a simple model with a strong predictive value for development of DCI, that can be used for decision making regarding intensity of monitoring and treatment in patients with SAH. For this study we analysed three prognostic models with increasing complexity based on the level of evidence in the literature. The first model included the two established predictors (clinical condition on admission and amount of subarachnoid blood on CT) and patient demographics. In the second model, candidate predictors with strong or moderate evidence were added (current smoking, blood glucose, enlarged ventricles on CT (hydrocephalus) and history of diabetes mellitus), and in the third model, also candidate predictors with limited evidence were added (history of hypertension, initial loss of consciousness, hemoglobin). Performance of the models was tested by discrimination and calibration. In Chapter 6, the main result was the development of a risk chart based on a model that combined clinical condition, amount of cisternal blood, amount of intraventricular blood, and age. According to the absence or presence of these four predictors that can easily be retrieved at admission, the risk chart displays absolute risks of DCI in SAH patients varying from 12 to 61%. The model categorizes around 20% of the patients to have a low risk (<20% risk) and another 20% to have a high risk of DCI (>40% risk). Validation in another patient population from more recent years confirmed that the discriminative ability was adequate (AUC 0.69), but outcomes were systematically better. The second result of our study was that our model improved little by including current smoking and hyperglycemia. However, though smoking and hyperglycemia were not included in our risk chart, both have an independent predictive value for DCI and increase the risk of DCI in the individual patient. The clinical value of our prediction model depends on reproducibility and consequences for clinical practice. The next step is validation of the risk chart in another hospital. At present, in few patients with low risk of DCI and postponement of treatment of the aneurysm, the risk chart could probably aid in the decision to treat with tranexaminic acid. Currently, tranexaminic acid is not routinely used in SAH patients because of the increased risk of DCI, despite its protection against rebleeds. However, recently is has been suggested that patients with good clinical condition on admission (WFNS I) did not show the adverse effects of tranexaminic acid on DCI increase, whereas the number of rebleeds was reduced. 39 The risk chart could aid to select patients at low risk for DCI. In patients with a high risk of DCI, more intensive and longer lasting monitoring could be considered. Although we may not be there yet, and 117

118 Chapter 7 General discussion additional validation is needed, this practical risk chart may bring us a step closer in the process of accurate prediction of which patients may develop DCI. Lessons learned The best model to predict DCI consists of 4 predictors present on admission and easy to assess: clinical condition, amount of cisternal blood, amount of intraventricular blood, and age. Low or high risk of DCI can be reliably predicted by a simple risk chart including these 4 predictors. Current smoking and hyperglycemia on admission contribute to a higher risk of DCI. Implications for future research Before application of the risk chart in clinical practice, our risk chart should be validated in another external hospital setting. Moreover, in future, risk prediction on DCI could be refined by including other potential predictors (such as early systemic inflammatory response syndrome). REFERENCES 1. Johnston SC, Selvin S, Gress DR. The burden, trends, and demographics of mortality from subarachnoid hemorrhage. Neurology 1998; 50(5): Feigin VL, Lawes CM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol 2003; 2(1): Linn FHH, Rinkel GJE, Algra A, van Gijn J. Incidence of subarachnoid hemorrhage: role of region, year, and rate of computed tomography: a meta-analysis. Stroke 1996; 27(4): Stegmayr B, Asplund K, Kuulasmaa K, et al. Stroke incidence and mortality correlated to stroke risk factors in the WHO MONICA Project. An ecological study of 18 populations. Stroke 1997; 28(7): Makela P, Fonager K, Hibell B, et al. Episodic heavy drinking in four Nordic countries: a comparative survey. Addiction 2001; 96(11): Vlak MH, Algra A, Brandenburg R, Rinkel GJ. Prevalence of unruptured intracranial aneurysms, with emphasis on sex, age, comorbidity, country, and time period: a systematic review and meta-analysis. Lancet Neurol 2011; 10(7):

119 General discussion Chapter 7 7. Wermer MJ, van der Schaaf IC, Algra A, Rinkel GJE. Risk of rupture of unruptured intracranial aneurysms in relation to patient and aneurysm characteristics: a meta-analysis. Stroke 2007; 38(4): Ruigrok YM, Rinkel GJ. From GWAS to the clinic: risk factors for intracranial aneurysms. Genome Med 2010; 2(9): Woo D, Khoury J, Haverbusch MM et al. Smoking and family history and risk of aneurysmal subarachnoid hemorrhage. Neurology 2009; 72(1): Harrod CG, Batjer HH, Bendok BR. Deficiencies in estrogen-mediated regulation of cerebrovascular homeostasis may contribute to an increased risk of cerebral aneurysm pathogenesis and rupture in menopausal and postmenopausal women. Med Hypotheses 2006; 66(4): Longstreth WT, Nelson LM, Koepsell TD, van Belle G. Subarachnoid hemorrhage and hormonal factors in women. A population-based case-control study. Ann Intern Med 1994; 121(3): Mhurchu CN, Anderson CS, Jamrozik KD, ACROSS study, et al. Hormonal factors and risk of aneurysmal subarachnoid hemorrhage. An international population-based, case-control study. Stroke 2001; 32(3): Algra AM, Klijn CJ, Helmerhorst FM, et al. Female risk factors for subarachnoid hemorrhage: A systematic review. Neurology 2012; 79(12): Lindekleiv H, Sandvei MS, Njolstad I, et al. Sex differences in risk factors for aneurysmal subarachnoid hemorrhage: a cohort study. Neurology 2011; 76(7): Pajunen P, Paakkonen R, Hamalainen H, et al. Trends in fatal and nonfatal strokes among persons aged 35 to > or =85 years during in Finland. Stroke 2005; 36(2): Rothwell PM, Coull AJ, Giles MF, et al. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet 2004; 363(9425): Vibo R, Korv J, Roose M. The Third Stroke Registry in Tartu, Estonia: decline of stroke incidence and 28-day case-fatality rate since Stroke 2005; 36(12): Ruigrok YM, Buskens E, Rinkel GJE. Attributable risk of common and rare determinants of subarachnoid hemorrhage. Stroke 2001; 32(5): Gabriel RA, Kim H, Sidney S, et al. Ten-year detection rate of brain arteriovenous malformations in a large, multiethnic, defined population. Stroke 2010; 41(1): Wiebers DO, Whisnant JP, Huston J 3rd, et al. Unruptured intracranial aneurysms: natural history, clinical outcome, and risks of surgical and endovascular treatment. Lancet 2003; 362(9378): Hendrikse J, van Raamt AF, van der Graaf Y, et al. Distribution of cerebral blood flow in the circle of Willis. Radiology 2005; 235(1): Velthuis BK, van Leeuwen MS, Witkamp TD, et al. Surgical anatomy of the cerebral arteries in patients with subarachnoid hemorrhage: comparison of computerized tomography angiography and digital subtraction angiography. J Neurosurg 2001; 95(2):

120 Chapter 7 General discussion 23. Horikoshi T, Akiyama I, Yamagata Z, et al. Magnetic resonance angiographic evidence of sex-linked variations in the circle of willis and the occurrence of cerebral aneurysms. J Neurosurg 2002; 96(4): Foutrakis GN, Yonas H, Sclabassi RJ. Saccular aneurysm formation in curved and bifurcating arteries. AJNR Am J Neuroradiol 1999; 20(7): Raghavan ML, Ma B, Harbaugh RE. Quantified aneurysm shape and rupture risk. J Neurosurg 2005; 102(2): Beck J, Rohde S, El Beltagy M, et al. Difference in configuration of ruptured and unruptured intracranial aneurysms determined by biplanar digital subtraction angiography. Acta Neurochir (Wien) 2003; 145: San Millan RD, Yilmaz H, Dehdashti AR, et al. The perianeurysmal environment: influence on saccular aneurysm shape and rupture. AJNR Am J Neuroradiol 2006; 27(3): Qian Y, Takao H, Umezu M, Murayama Y. Risk analysis of unruptured aneurysms using computational fluid dynamics technology: preliminary results. AJNR Am J Neuroradiol 2011; 32(10): Xiang J, Natarajan SK, Tremmel M, et al. Hemodynamic-morphologic discriminants for intracranial aneurysm rupture. Stroke 2011; 42(1): Vergouwen MD. Vasospasm versus delayed cerebral ischemia as an outcome event in clinical trials and observational studies. Neurocrit Care 2011; 15(2): Vergouwen MD, Vermeulen M, van Gijn J, et al. Definition of delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage as an outcome event in clinical trials and observational studies: proposal of a multidisciplinary research group. Stroke 2010; 41(10): Frontera JA, Fernandez A, Schmidt JM, et al. Defining vasospasm after subarachnoid hemorrhage: what is the most clinically relevant definition? Stroke 2009; 40(6): Dankbaar JW, de Rooij NK, Velthuis BK, et al. Diagnosing delayed cerebral ischemia with different CT modalities in patients with subarachnoid hemorrhage with clinical deterioration. Stroke 2009; 40(11): Dankbaar JW, Rijsdijk M, van der Schaaf IC, et al. Relationship between vasospasm, cerebral perfusion, and delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage. Neuroradiology 2009; 51(12): Rabinstein AA, Weigand S, Atkinson JL, Wijdicks EF. Patterns of cerebral infarction in aneurysmal subarachnoid hemorrhage. Stroke 2005; 36(5): Adams HP Jr, Kassell NF, Torner JC, Haley EC Jr. Predicting cerebral ischemia after aneurysmal subarachnoid hemorrhage: influences of clinical condition, CT results, and antifibrinolytic therapy. A report of the Cooperative Aneurysm Study. Neurology 1987; 37(10): Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. Neurosurgery 1980; 6(1): van Gijn J, Kerr RS, Rinkel GJ. Subarachnoid haemorrhage. Lancet 2007; 369(9558):

121 General discussion Chapter Roos Y. Antifibrinolytic treatment in subarachnoid hemorrhage: a randomized placebo-controlled trial. STAR Study Group. Neurology 2000; 54(1):

122 Chapter 7 General discussion 122

123 Nederlandse samenvatting

124 Nederlandse samenvatting Introductie De aneurysmatische subarachnoïdale bloeding (SAB) is een ernstige vorm van beroerte. Anders dan de meeste vormen van beroerte, komt de SAB voor bij relatief jonge mensen. De gemiddelde leeftijd is 55 jaar. Zowel de kans op sterfte (40%) als de kans op blijvend letsel is bij een SAB zeer groot. Dit promotieonderzoek heeft zich er daarom op gericht de aandoening en één van de belangrijke complicaties ervan (zuurstofgebrek in de hersenen) beter te voorspellen en te begrijpen, om zo in de toekomst hopelijk de sterfte en blijvend letsel van deze patiënten te verminderen. Een SAB is een bloeding tussen de hersenvliezen. Een aneurysmatische SAB wordt veroorzaakt door het openbarsten (ruptuur) van een bolvormige uitstulping (aneurysma) van één van de slagaders in de hersenen. In de algehele bevolking heeft ongeveer 3% een aneurysma in het hoofd, maar niet bij iedereen zal een SAB optreden. Tot op heden blijft het moeilijk om te voorspellen bij welke patiënten het aneurysma zal ruptureren, en zo zal leiden tot een SAB. De slechte prognose na een SAB wordt veroorzaakt door de hersenschade ten gevolge van de bloeding zelf, maar ook door diverse complicaties die kunnen ontstaan in de eerste dagen tot weken na een SAB. Eén van deze complicaties is zuurstofgebrek in de hersenen waardoor een blijvend herseninfarct kan ontstaan. Dit wordt secundaire ischemie genoemd. Tot nu toe is de ontstaanswijze van secundaire ischemie (grotendeels) onduidelijk en is het onvoorspelbaar bij welke patiënten deze complicatie zal optreden. Deel 1 van dit proefschrift beschrijft een analyse over hoe vaak SAB wereldwijd voorkomt en welke patiënten met een aneurysma een verhoogd risico hebben op het krijgen van een SAB. In deel 2 wordt onderzocht welke factoren de kans op secundaire ischemie kunnen voorspellen. DEEL 1 Incidentie van de subarachnoïdale bloeding, en het risico op ruptuur van een intracranieel aneurysma Hoofdstuk 2 beschrijft een grote meta-analyse [1] die is uitgevoerd om te analyseren hoe vaak SAB voorkomt (incidentie). De nieuwe inzichten in dit hoofdstuk over invloed van regio, leeftijd en geslacht op de incidentie brengen ook nieuwe inzichten in de ontstaanswijze van SAB. [1] Bij een meta-analyse worden de data van meerdere onderzoeken over hetzelfde fenomeen met elkaar gebundeld. Daarover wordt een statistische analyse gedraaid om zodoende met een grotere mate van zekerheid een conclusie te kunnen trekken. 124

125 Nederlandse samenvatting We includeerden in totaal 52 studies over de incidentie van SAB, en berekenden een gemiddelde incidentie van 9,2 per patiëntjaren. Uit de resultaten kan echter geconcludeerd worden dat de incidentie sterk varieert. Er zijn grote verschillen per regio, geslacht en leeftijd. Zo blijkt dat de incidentie van SAB tweemaal zo hoog is in Japan (gemiddeld 22,7 per patiëntjaren) en Finland (gemiddeld 19,7) als in andere delen van de wereld. Ook blijkt dat de bloeding op jonge leeftijd vaker voorkomt bij mannen, maar na de leeftijd van 55 jaar juist vaker bij vrouwen. Bekende risicofactoren voor een SAB zijn erfelijkheid, een hoge bloeddruk, roken en overmatig alcoholgebruik. De verschillen in leeftijd en geslacht die werden aangetoond in dit hoofdstuk suggereren dat er een hormonale invloed is op het krijgen van een SAB. De grote verschillen tussen de landen suggereren dat naast erfelijkheid ook bepaalde gezondheids- en eetgewoontes of een combinatie hiervan een belangrijke rol kunnen spelen. De onderliggende oorzaak van een SAB is een intracranieel aneurysma. Een intracranieel aneurysma komt bij 3% van de bevolking voor, maar gelukkig ruptureert niet ieder aneurysma. Als bij patiënten bij toeval een aneurysma wordt gevonden, dient te worden beoordeeld of dit preventief moet worden behandeld. Om het risico op een SAB uit te sluiten kan besloten worden tot clippen (operatief afsluiten van het aneurysma door de neurochirurg) of coilen (opvullen van het aneurysma met coils door de radioloog die deze coils inbrengt met behulp van een katheter via de liesslagader). De ingreep zelf brengt echter ook risico s met zich mee. Daarom is de beslissing tot het al dan niet behandelen van een aneurysma afhankelijk van het individuele risico van de patiënt op ruptuur van het aneurysma. Momenteel is de kennis beperkt om dit individuele risico op een ruptuur te voorspellen. Bij de klinische beslissing om een aneurysma wel of niet preventief te clippen dan wel te coilen wordt tot nu toe gekeken naar de grootte en de plaats van het bij toeval gevonden aneurysma, en de leeftijd en gezondheid van de patiënt. In Hoofdstuk 3 worden diverse anatomische factoren onderzocht als mogelijke nieuwe risicofactoren voor een aneurysmaruptuur: asymmetrie en onvolledigheid van de bloedvataanleg van de cirkel van Willis (de slagaders aan de onderkant van de hersenen en hun onderlinge verbindingen), en de vorm van het aneurysma en stroomrichting vanuit het bloedvat in het aneurysma. Bij het onderzoek zijn de anatomische configuraties vergeleken tussen twee patiëntgroepen: een groep met geruptureerde aneurysmata en de groep met ongeruptureerde aneurysmata, waarbij de patiënten waren gematched op leeftijd, geslacht, plaats van aneurysma en jaar van het CT-angiogram. De analyses toonden dat elliptische of meerlobbige vorm van het aneurysma en de directe stroomrichting vanuit het bloedvat in het aneurysma geassocieerd zijn met ruptuur. Voor configuratie van de cirkel van Willis werd geen statistisch significant verschil gevonden. 125

126 Nederlandse samenvatting DEEL 2 Secundaire ischemie na een subarachnoïdale bloeding Bij een derde van de SAB-patiënten ontstaat tijdens hun opname in het ziekenhuis secundaire ischemie. Door het zuurstofgebrek in de hersenen treedt bij patiënten een gedaald bewustzijn of focale uitval op. Vaak is er sprake van verlammingsverschijnselen aan arm of been, een taalstoornis (afasie) of treden veranderingen op in het geheugen of het karakter. Secundaire ischemie vergroot de kans op sterfte en blijvend letsel na een SAB. Er zijn meerdere bloedvaten die de hersenen van zuurstof voorzien. Elk bloedvat voorziet een deel van de hersenen en heeft daarmee zijn eigen verzorgingsgebied. Sommige patiënten met secundaire ischemie krijgen één herseninfarct (in één verzorgingsgebied), anderen krijgen meerdere herseninfarcten (in meerdere verzorgingsgebieden). In Hoofdstuk 4 wordt de vraagstelling onderzocht of dit twee verschillende patronen van secundaire ischemie zijn. Voor dit onderzoek werden patiënten met een SAB op basis van het aantal infarcten op de follow-up-scans ingedeeld in drie groepen: geen infarct, infarct in één verzorgingsgebied, of infarcten in meerdere verzorgingsgebieden. De hypothese was dat de twee patronen een andere ontstaanswijze hebben. Dit werd geanalyseerd door de risicoprofielen van de twee groepen patiënten met infarct(en) te vergelijken met de risicoprofielen van patiënten zonder herseninfarct. Uit het onderzoek bleek dat zowel de patiënten met één infarct als de patiënten met meerdere infarcten een vergelijkbaar risicoprofiel hadden, alleen was dit profiel bij de patiënten met meerdere infarcten uitgebreider. Geconcludeerd kan worden dat het geen verschillende vormen van secundaire ischemie zijn, en dat er alleen een verschil is in ernst. De ontstaanswijze van secundaire ischemie is voor het grootste deel onopgehelderd, en tot nu toe kunnen we niet voorspellen bij wie secundaire ischemie ontstaat. Twee in het algemeen bekende voorspellers voor secundaire ischemie zijn een grote hoeveelheid subarachnoïdaal bloed op de CT-scan van de hersenen en de slechte klinische conditie van de patiënt bij binnenkomst in het ziekenhuis. Het exacte individuele risico van een patiënt met deze predictoren is echter onbekend. Hoofdstuk 5 beschrijft een grote literatuurstudie naar het bewijs voor andere mogelijke predictoren van secundaire ischemie bij SAB-patiënten. Alleen factoren die in de eerste drie dagen na binnenkomst in het ziekenhuis beschikbaar zijn en gemakkelijk verkregen kunnen worden werden geïncludeerd. Tweeënvijftig studies voldeden aan de inclusiecriteria en deze werden allen beoordeeld op methodologische kwaliteit. In totaal werden 33 verschillende factoren onderzocht. Van alle factoren werd per studie de effectmaat op ontstaan van secundaire ischemie vastgesteld en bij voorkeur (her)berekend tot odds ratio. Aan de hand van deze resultaten werden de 33 factoren ingedeeld in predictief, non-predictief of inconsistent. 126

127 Nederlandse samenvatting Vervolgens werd de mate van bewijs geanalyseerd (het aantal studies met hoge methodologische kwaliteit en de relevantie van hun resultaten) en per onderzochte factor werd een level of evidence toegekend (sterk bewijs, matig bewijs of weinig bewijs). In deze literatuurstudie kwam naar voren dat actief roken een predictor is voor secundaire ischemie (sterk bewijs). Verder zijn hyperglykemie in het bloed bij binnenkomst, het hebben van diabetes, en het hebben van een vroege systemische ontstekingsreactie (SIRS, twee van de volgende vier criteria: koorts of hypothermie; leukocytose of leukopenie; tachycardie; tachypneu) potentiële voorspellers (matig bewijs). Voor lokatie van het aneurysma werd bewezen dat het geen predictor is (sterk bewijs). Voor vele andere onderzochte factoren was er weinig of inconsistent bewijs. In Hoofdstuk 6 zijn de patiënten van het Universitair Medisch Centrum Utrecht op de diverse predictoren onderzocht volgens verschillende modellen waarbij ook de voorspellende waarde van leeftijd en geslacht werd geanalyseerd. Vanuit één van die modellen werd een praktische risicokaart gemaakt. De vier belangrijkste voorspellers waren leeftijd, de hoeveelheid bloed op de CT in de subarachnoïdale cisternen, de hoeveelheid bloed op de CT in de ventrikels (hersenkamers) en de klinische conditie van de patiënt bij binnenkomst in het ziekenhuis gemeten volgens de WFNS score (World Federation of Neurological Surgeons scale). Actief roken en hyperglykemie in het bloed bij binnenkomst hadden daarnaast ook een onafhankelijke bijdrage aan het voorspellen van secundaire ischemie. Twee onderzochte modellen gaven een bijna vergelijkbare mate van voorspelling. Dit onderzoek was specifiek gericht op het toepasbaar maken van de onderzoeksresultaten in de praktijk. Bij het maken van de risicokaart werd daarom bewust gekozen voor het eenvoudigste model met slechts vier factoren in plaats van een meer complex model met zes factoren. Aanwezigheid van de volgende vier factoren werd geassocieerd met een hoger risico: leeftijd onder 55 jaar, grote hoeveelheid bloed op de CT in de cisternen, grote hoeveelheid bloed op de CT in de ventrikels, en slechte klinische conditie van de patiënt bij binnenkomst. In hetzelfde onderzoek werd de risicokaart gevalideerd in een recente patiëntenpopulatie van het Universitair Medisch Centrum Utrecht. De volgende stap is dit ook te doen in een externe populatie. De risicokaart geeft de neuroloog bij opname van een SAB-patiënt in het ziekenhuis een concreet getal voor het risico op secundaire ischemie (absoluut risico tussen de 12% en 61%) en kan gebruikt worden als richtlijn bij beslissingen over de intensiteit van monitoring, starten van mobilisatie en het nemen van preventieve maatregelen. 127

128 128 Nederlandse samenvatting

129 Dankwoord

130 Dankwoord Een proefschrift schrijven dat doe je nooit alleen. Graag wil ik daarom mijn dank uitspreken naar iedereen met wie ik heb samengewerkt en van wie ik steun heb gehad tijdens mijn periode als promovenda. Een aantal mensen wil ik in het bijzonder noemen. Mijn grote dank voor prof. dr. G.J.E. Rinkel. Beste Gabriël, inmiddels kennen we elkaar al een flink aantal jaren. Jouw begeleiding en stimulans tijdens mijn onderzoeksstage zullen mij altijd bijblijven en waren het begin van deze onderzoekscarrière. Veni, Vidi, Vici, was je toespraak tijdens mijn Buluitreiking. Ik waardeer het enorm dat je altijd in mij geloofd hebt, en ik heb waanzinnig veel van je geleerd. Je rode pen maakte het altijd beter. In het laatste traject was je druk als hoofd van de afdeling, maar de inhoud van je advies was er niet minder om. Ook al ga ik qua carrière nu een andere richting op, ik hoop dat we in de toekomst zeker nog gaan samenwerken. Grote dank ook voor de begeleiding van dr. C.J.M. Frijns. Beste Rinie, niet alleen ben je een fantastisch goede neuroloog met veel kennis, je bent ook een hele lieve persoonlijkheid. Met jou overlegde ik (niet overdreven) wel duizenden scans. Beiden bestudeerden we alles met meer dan volle overtuiging. Door jouw precisie en gedrevenheid om casuïstiek te doorgronden hebben we samen data verzameld waar ik met de volle 100% achter durf te staan. Ik ben jouw eerste promovenda, dank voor je betrokken begeleiding, er zullen er vast nog meer volgen. Beste medeauteurs, dank voor jullie waardevolle bijdrages. Prof. dr. A. Algra, dr. J.W. Dankbaar, dr. J.P. Greving, dr. F.H.H. Linn, drs. J.A.P. van der Plas en dr. B.K. Velthuis. Beste Ale, dank voor al je hulp. Je hebt zoveel kennis en weet de moeilijke statistiek voor je collega-artsen begrijpelijk te maken. Het is fantastisch om te zien hoe je alles combineert en zoveel mensen inspireert. Beste Jan Willem, bedankt voor het samen opzetten van DECIDE en het samen schrijven van diverse artikelen. Beste Jacoba, zonder jou geen voorspellingen, hartelijk dank voor de samenwerking, al je uitleg en heldere statistische input. Beste Cisca, ik bewonder nog altijd hoe jij een systematic review hebt geschreven in een tijd dat artikelen niet elektronisch beschikbaar waren en alles per fax gecommuniceerd werd, leuk dat ik je werk mocht voortzetten. Beste Birgitta, bedankt voor je radiologische inzicht, begeleiding en helder commentaar op de artikelen. Beste prof. dr. J. van Gijn en prof. dr. J.H.J. Wokke, bedankt voor alles wat ik van jullie heb geleerd tijdens de opleiding. Beste Paut, Dorien, Marrit, Monique, Ans en alle anderen van het trialbureau, iedereen met wie ik veel heb gedeeld, bedankt voor de leuke samenwerking! Beste kamergenoten. Als onderzoeker ben ik als student al begonnen met Stijntje, en later als AGIKO met Sefanja. Jullie zijn me beiden dierbaar en ik hoop dat het contact ook na de 130

131 Dankwoord promoties zal blijven. Vele uren heb ik doorgebracht in het van Geuns en daarbij lief en leed gedeeld met wisselende van Geuns-matties! In diverse samenstellingen was de 8e verdieping altijd een vertrouwd en gezellig honk! Naast hard werken, werd er ook veel gelachen en besproken. Ik had de tijd niet willen missen met Annette, Aysun, Charlotte van Asch, Charlotte Cremers, Dennis, Esther, Ingeborg, Joanna, Lieza, Stijntje en Suzanne. Annette en Charlotte van Asch, wat een mooie combinatie was dat, zal het niet snel vergeten meiden. Lieza, ik heb goede herinneringen aan onze lunches samen. Dennis, je zorgde altijd voor gezelligheid en heerlijke nuchtere humor! Daarbij zorgde je ervoor dat de deur tot de wetenschap altijd wijd openstond ;-) Thanks dat jullie zulke leuke collega s zijn! Aan alle arts-assistenten van de neurologie, dank voor alle mooie momenten samen. Ik vind het leuk met jullie te hebben samengewerkt en hoop jullie allen in de toekomst nog vaak tegen te komen. Lieve Anne, Anneriek, Christel, Claire, Kim, Hannemeis, Simone en Suze 3x3 is voor ons nog altijd blauw, al sinds 2000 een fantastische jaarclubband waar menigeen jaloers op mag zijn. Jullie zijn topmeiden en ik ben blij met jullie! Als enige dokter hebben jullie ervoor gezorgd dat ik een brede kijk op de wereld heb gehouden. En Anne jij verdient een ereplaats, bedankt voor je fantastische kunde in de taal. Lieve Anne, Annelies, Alette, Emily, Joepe, Ofke en Sunna altijd maken we tijd voor borrels, etentjes, feestjes, vakanties! Wat een heerlijke mix aan dokters zijn jullie! Iedereen z n eigen specialisatie, maar logischerwijs toch dezelfde herkenbare avonturen en belevenissen in het ziekenhuis. Laten we 24 livin the good life nog lang in stand houden, en lief, leed en plezier blijven delen. Lieve oud-huisgenoten, bedankt voor de leuke tijd in Utrecht, het was altijd fijn thuiskomen! Lieve Joepe en Ofke, dat voelt fijn en geruststellend Twee slimme en mooie meiden staan straks naast me om me bij te staan. Fantastisch dit met jullie te delen! Met jullie aan m n zij komt alles goed, en is het ook nog supergezellig! Thanxxx voor al jullie inbreng en hulp! Lieve familie, als jongste kon ik bij iedereen altijd meer dan terecht voor goed advies! Nu (zoveel jaren later) kan ik dat gelukkig nog steeds, en is het uitwisselen van adviezen wederzijds geworden. Onze familie staat als een stevige rots vol geborgenheid en gezelligheid. Lieve papa, de onderzoek-drive heb ik zeker weten van jou. Jij had mijn geaccepteerde publicaties al eerder op Pubmed gezien dan ikzelf. Altijd volop geïnteresseerd om te horen met welk onderzoek ik bezig ben. Fijn dat ik altijd op jullie steun kan rekenen en jullie achter mij staan, welke stap ik ook zet. Lieve mama, dank voor alle fijne momenten en gesprekken. Het is leuk om iemand te 131

132 Dankwoord hebben op wie je zoveel lijkt, want daardoor begrijp je me vaak als geen ander. Ik kan me geen lievere ouders wensen! Lieve Bert-Jan en Gerdine, mijn grote broer en zus, jullie geven mij een trouwe familiebasis. Een basis waarin we alledrie onszelf kunnen zijn. Allemaal op onze eigen manier dragen we bij aan die band en basis die er al van jongs af aan is. Later vulde Nadia en Joost dat gevoel extra aan, en sinds kort is er ook het geluk van Bastiaan, Joris en Amy om samen te delen. Ik hou van jullie allemaal! Lieve Amie, geen knuffel is groot genoeg om ook jou te bedanken. Je toont me hoe leuk en belangrijk het is om jezelf te ontwikkelen, ergens voor te staan en te gaan, en daarvan te genieten. Lang geleden heb jij een steen verlegd, en gezorgd dat je kinderen en je (achter) kleinkinderen de kansen tot onderwijs en ontwikkeling krijgen die je ons zo gunt. Ik geniet van al die mogelijkheden, elke dag, zo ook bij het schrijven van dit proefschrift. Ik ben trots je kleindochter te zijn! Allerliefste Wouter, ik hou zoveel van jou! Jouw briljante sprankelende liefde, je oprechtheid, je steun, je nieuwe slimme ideeën, je directheid, je andere kijk op dingen ze brengen het beste in mezelf naar boven. Leven met jou is heerlijk en bijzonder. Dank je wel dat ik zo op je kan bouwen, met alles wat ik doe! Je maakt woorden altijd tot daden. En dat met de nodige humor, want je grappen en je creativiteit maken mij altijd aan het lachen. Ook in moeilijkere situaties weet jij te relativeren en kunnen we genieten van het moois dat we samen hebben. Ik kijk uit naar een toekomst met jou! 132

133 About the author