Eye Complications and Markers of Morbidity and Mortality in Long-term Type 1 Diabetes

Transcription

1 Acta Ophthalmologica Thesis Doctoral Thesis Eye Complications and Markers of Morbidity and Mortality in Long-term Type 1 Diabetes Jakob Grauslund Faculty of Health Science University of Southern Denmark Department of Ophthalmology Odense University Hospital, Denmark Odense 2011

2 ACTA OPHTHALMOLOGICA 2011 To Vilma, my wonderful daughter Cover image: Retinal photos prepared for vascular diameter analysis (upper pictures) and fractal analysis (lower pictures).

3 ACTA OPHTHALMOLOGICA 2011 Contents List of Papers iv Acknowledgements v Abbreviations vi Introduction 1 Type 1 diabetes 1 Diabetic retinopathy 1 Risk factors of mortality 2 Visual impairment in type 1 diabetes 2 Retinal vascular analysis 3 Noninvasive markers of micro- and macrovascular complications 3 Aims 4 Methods 4 Study population 4 First examination ( ) 4 Second examination ( ) 5 Blindness 5 Cataract surgery 5 Retinal vascular calibre analysis 5 Fractal analysis 6 Blood measurements 6 Results 6 Retinopathy and mortality (Paper I) 6 Long-term incidence of blindness and associated risk factors (Paper II) 7 Cataract surgery and risk factors (Paper III) 8 Retinal vascular calibre and micro- and macrovascular complications (Paper IV) 8 Retinal fractals and diabetes-related complications (Paper V) 8 N-terminal pro-brain natriuretic peptide in type 1 diabetes (Paper VI) 10 Osteoprotegerin as a marker of long-term complications (Paper VII) 10 Discussion 11 Predictors of mortality 11 Long-term incidence of blindness and cataract surgery 11 Retinal vascular calibres and fractals 12 Noninvasive markers in type 1 diabetes 13 Strengths and limitations 15 Perspectives 15 Summary in English 15 Summary in Danish 16 References 17

4 ACTA OPHTHALMOLOGICA 2011 List of Papers This thesis is based on the following papers which will be referred to by their Roman numerals: I Grauslund J, Green A, Sjølie AK (2008). Proliferative retinopathy and proteinuria predict mortality rate in type 1 diabetic patients from Fyn County, Denmark. Diabetologia 51: II Grauslund J, Green A, Sjølie AK (2009). Blindness in a 25-year follow-up of a population-based cohort of Danish type 1 diabetic patients. Ophthalmology 116: III IV V VI VII Grauslund J, Green A, Sjølie AK (2009). Cataract surgery in a population-based cohort of type 1 diabetic patients: long-term incidence and risk factors. Acta Ophthalmologica Scandinavica. doi: /j x Grauslund J, Hodgson L, Kawasaki R, Green A, Sjølie AK, Wong TY (2009). Retinal vessel calibre and micro- and macrovascular complications in type 1 diabetes. Diabetologia 52: Grauslund J, Green A, Kawasaki R, Hodgson L, Sjølie AK, Wong TY (2010). Retinal vascular fractals and micro- and macro-vascular complications in type 1 diabetes. Ophthalmology 117: Grauslund J, Nybo M, Green A, Sjølie AK (2010). N-terminal pro brain natriuretic peptide reflects long-term complications in type 1 diabetes. Scand J Clin Lab Invest 70: Grauslund J, Rasmussen LM, Green A, Sjølie AK (2010). Does osteoprotegerin relate to micro- and macrovascular complications in long-term type 1 diabetes? Scand J Clin Lab Invest 70:

5 ACTA OPHTHALMOLOGICA 2011 Acknowledgements The work presented in this doctoral thesis was carried out at the Department of Ophthalmology, Odense University Hospital in It is a successor to the PhD Thesis Long-term mortality and retinopathy in type 1 diabetes that was accepted in I would like to express my sincere gratitude to the following, without whom this work would not have been possible. Firstly I would like to express my thankfulness to my friend and mentor Anne Katrin Sjølie who inspired and guided me throughout this work. Her dedication to this project helped and inspired me all the way. Likewise Anders Green was always able to find time for guidance and fruitful discussions. I am very thankful for this. A part of this thesis was done at the Centre for Eye Research Australia in Melbourne, Australia. I am very grateful to Tien Wong for giving me this opportunity and for his help throughout the study. I am also very indebted to my friend and mentor Ryo Kawasaki who supervised my stay in Melbourne. Likewise I would not have been able to learn how to grade vascular diameters and fractals without the tremendous help from Lauren Hodgson, Annie McAuley, Yumiko Kawasaki and Ignatios Koukouras. I am very grateful for your friendliness and, in particularly, for the nice weather you arranged for the entire month I was in Melbourne! At the Ocular Epidemiology Reading Center in Madison, Wisconsin, USA, I had the great pleasure of receiving good advice and invaluable suggestions from Ron Klein, Barbara Klein and Stacy Meuer. This was a huge help for which I am very grateful. Back home again, Lars Melholt Rasmussen and Mads Nybo from the Department of Biochemistry, Pharmacology and Genetics, Odense University Hospital were always very helpful. Thanks. From the same department I would also like to thank Charlotte Olsen, Lone Hansen, Kirsten Ulla Bahrt, Anette Tyrsted Mikkelsen and Gitte Primdahl Nielsen who all helped me to sample and store the blood throughout the study. I am indebted to Kelvin Kamp Mortensen, the head of the Department of Ophthalmology, Odense University Hospital, who gave me the time and opportunity to do this work at the department. Likewise, I would like to give my thanks to my colleagues in the daily work. Among these, a special credit has to be given to my friends and colleagues at the department s Research Unit: Karen Bjerg Pedersen, Birgitte Justesen, Majbrit Lind and Flemming Møller were always very helpful and supportive. Thanks. My family and friends provided moral support through all the years. I am very grateful for this. And finally, without the never-ending love and support from my dear wife Julie and our lovely daughter Vilma, I could not have made this thesis. Thank you. I am very grateful for the financial support provided for this study. With regard to this I would like to thank Velux Foundation, Danish Eye Health Society, Institute of Clinical Research at University of Southern Denmark, Sehested Hansen s Foundation, Danish Diabetes Association, Synoptik Foundation, The A.P. Møller Foundation for the Advancement of Medical Science, The Danish Society of Ophthalmology, The Legacy of Teacher Karen Svankjær Yde, The Legacy of Engineer August Frederik Wedel Erichsen, The Legacy of A. and J. Rasmussen and Odense University Hospital. And, of course, I would like to acknowledge each and every patient who participated in this study. I thank you all. Conflicts of interest: None Jakob Grauslund, December 2010

6 ACTA OPHTHALMOLOGICA 2011 Abbreviations ANP: Atrial natriuretic peptide AVR: Arteriolar venular ratio BNP: Brain natriuretic peptide CI: Confidence interval CRAE: Central retinal arteriolar equivalent CRVE: Central retinal vein equivalent CWS: Cotton wool spots DCCT: Diabetes Control and Complications Trial Df: Fractal dimension DME: Diabetic macular oedema DR: Diabetic retinopathy DRS: Diabetic Retinopathy Study ETDRS: Early Treatment Diabetic Retinopathy Study HbA1: Haemoglobin A1 HbA1c: Haemoglobin A1c HR: Hazard ratio IHD: Ischaemic heart disease IRMA: Intraretinal microvascular abnormalities IRIS: International Retinal Imaging Software NPDR: Non-proliferative diabetic retinopathy NT-proBNP: N-terminal pro brain natriuretic peptide OPG: Osteoprotegerin OR: Odds ratio PDR: Proliferative diabetic retinopathy WESDR: Wisconsin Epidemiologic Study of Diabetic Retinopathy

7 ACTA OPHTHALMOLOGICA 2011 Denne afhandling er den 9. november 2010 af Akademisk Råd, Det Sundhedsvidenskabelige Fakultet, Syddansk Universitet antaget til forsvar for den medicinske doktorgrad. Ole Skøtt h.a.dec. Forsvaret finder sted d. 25. marts 2011 kl i Auditoriet, Winsløwparken 25, 5000 Odense C. This thesis has been accepted for defence for the medical doctoral degree on November by the Academic Council, Faculty of Health Science, University of Southern Denmark. Ole Skøtt Dean The defence will take place on March at in Auditoriet, Winsløwparken 25, DK 5000 Odense C, Denmark. Members of the evaluation commitee: Professor, MD, PhD, Massimo Porta, Unit of Internal Medicine 1, University of Turin, Italy. Professor, dr.med., Toke Bek, Dept. of Ophthalmology, Aarhus Hospital, Aarhus, Denmark. Professor, dr.med. Henning Beck-Nielsen, Department of Endocrinology, Odense University Hospital and Clinical Institute, University of Southern Denmark (chairman), Odense, Denmark.

8 Doctoral Thesis Eye Complications and Markers of Morbidity and Mortality in Long-term Type 1 Diabetes Jakob Grauslund 1,2 1 Faculty of Health Science, University of Southern Denmark, Odense, Denmark 2 Department of Ophthalmology, Odense University Hospital, Odense, Denmark ABSTRACT. The incidence of type 1 diabetes is rising all over the world. Furthermore, the increased life-expectancy of type 1 diabetic patients is likely to cause a higher number of diabetes-related micro- and macrovascular complications in the years to come. In order to examine the level of long-term complications in type 1 diabetes as well as potential markers of micro- and macroangiopathy, a population-based cohort of Danish type 1 diabetic patients was examined in order to achieve the following aims: 1. To evaluate diabetic retinopathy as a long-term marker of all-cause mortality in type 1 diabetes (Paper I). 2. To estimate the long-term incidence and associated risk factors of blindness (Paper II) and cataract surgery (Paper III) in type 1 diabetes. 3. To use retinal vascular analyses in order to investigate the associations of long-term micro- and macrovascular complications and retinal vascular diameters (Paper IV) and retinal fractals (Paper V) in type 1 diabetes. 4. To examine N-terminal pro brain natriuretic peptide (Paper VI) and osteoprotegerin (Paper VII) as non-invasive markers of micro- and macrovascular complications in type 1 diabetes. In Paper I it was a major finding that, despite a mean age of only 38.3 years at baseline, 44.7% of the patients died during the 25-year follow-up. Patients who had proliferative retinopathy as well as proteinuria at the baseline examination had a significantly higher mortality. For these, the 10-year survival was only 22.2%. As demonstrated in Paper II, blindness is an important issue in type 1 diabetes. The 25-year cumulative incidence of blindness was 7.5%. Glycaemic regulation and maculopathy at baseline were both identified as risk factors of blindness. Finally, mortality was higher in patients who went blind during the follow-up. Cataract surgery is quite common in type 1 diabetes. In Paper III a 25-year cumulative incidence of 20.8% was found. Adjusted for mortality, this was even higher (29.4%). As compared to patients without diabetes, cataract surgery takes place approximately 20 years earlier in type 1 diabetic patients. Age and maculopathy at baseline were both identified as predictors of cataract surgery. In Paper IV it was demonstrated that patients with retinal arteriolar narrowing were 2.17 and 3.17 times more likely to have nephropathy and macrovascular disease, respectively. This was an important finding that suggests that retinal fundus photos may be used in order to predict the risk of non-ophthalmological complications in type 1 diabetes. Retinal fractal analysis is another way to evaluate the vascular system of the retina. In Paper V we found associations between retinal fractal and microvascular but not macrovascular disease. For instance, patients with lower fractal dimensions were more likely to have proliferative retinopathy (OR 1.45, 95% CI ) and neuropathy (OR 1.42, 95% CI ). NT-proBNP is likely to be a future predictor of diabetes-related complications. In Paper VI higher levels of NT-proBNP were related to nephropathy (OR 5.03, 95% CI ), neuropathy (OR 4.08, 95% CI ) and macrovascular disease (OR 5.84, 95% CI ). These associations should be confirmed in future prospective studies. As opposed to NT-proBNP, osteoprotegerin is less likely to be a predictor of either micro- or macrovascular disease in type 1 diabetes. As demonstrated in Paper VII, even though association between higher levels of OPG and nephropathy were found in an age- and sex-adjusted model (OR 2.54, 95% CI ), this was no longer statistically significant when other factors were taken into account. Overall, it was demonstrated that various complications such as mortality, blindness and cataract surgery were high in type 1 diabetes. Furthermore, retinal arteriolar narrowing, decreased retinal fractals and plasma NT-proBNP were associated with various micro- and macrovascular complications. If confirmed by prospective studies, these modalities may be used in order to identify patients at risk of diabetesrelated complications. This could, ultimately, lead to decreased mortality and morbidity in type 1 diabetic patients. Acta Ophthalmol. 2011: 89 thesis1: 1 19 ª 2011 The Author Acta Ophthalmologica ª 2011 Acta Ophthalmologica Scandinavica Foundation doi: /j x Introduction Type 1 diabetes Type 1 diabetes is an autoimmune disease caused by the destruction of the beta cells of the pancreas that leads to insulin deficiency and chronic hyperglycaemia. Mortality within a few years after the onset of diabetes was inevitable prior to the introduction of insulin by Banting and Best in Since then, the life expectancy of patients with type 1 diabetes has increased, but early mortality still remains an important issue (Moss et al. 1991; Soedamah-Muthu et al. 2006a). The prolonged life expectancy for patients with type 1 diabetes raises another important issue. The chronic load of hyperglycaemia may lead to various micro- and macrovascular complications. Among the former, microvascular changes in the eyes, kidneys and nerves lead to retinopathy, nephropathy and neuropathy, respectively. For the larger arteries, macrovascular complications like ischaemic heart disease (IHD) and stroke have a significant impact on morbidity and mortality (Moss et al. 1991; Laing et al. 2003; Soedamah- Muthu et al. 2004, 2006b). Diabetic retinopathy (DR) Diabetic retinopathy affects almost all patients with type 1 diabetes with duration of diabetes of 15 years or more (Klein et al. 1984c; Grauslund et al. 2009). Diabetic retinopathy is characterized by lesions of the retinal 1

9 microvasculature. Although the exact mechanisms that lead to DR are not completely clear, it seems evident that chronic hyperglycaemia has a pivotal role in the pathogenic alterations of the retinal microvasculature. This was demonstrated in the Diabetes Control and Complications Trial in which intensive insulin therapy led to a 76% decreased risk of incident DR (The Diabetes Control and Complications Trial Research Group 1993). Thickening of the basement membrane of the retinal endothelial cells as well as pericyte loss and increased capillary permeability are among the first physiological changes in the diabetic retina (Kohner 1993; Gardiner et al. 2007). This causes capillary closure and, hence, retinal nonperfusion. To compensate, other capillaries dilate, which leads to microaneurysms the first type of visible lesion caused by DR. When retinal ischaemia progresses, other lesions like cotton wool spots (CWS), venous beading and intraretinal microvascular abnormalities (IRMA) might occur (Kohner 1993). These characteristics are commonly known as nonproliferative diabetic retinopathy (NPDR) as opposed to proliferative diabetic retinopathy (PDR) that may come secondary to further retinal ischaemia. In PDR, new vessels emerge from the retinal vessels. Visual loss may be caused by these vessels as a result of preretinal and vitreous haemorrhage followed by tractional retinal detachment. At any stage, vision can also be impaired by diabetic macular oedema (DME) caused by the breakdown of the blood retinal barrier. Risk factors of mortality Owing to the excess mortality in type 1 diabetes, it is important to be able to identify the high-risk patients. Proteinuria has been identified as an important risk factor (Borch-Johnsen et al. 1985; Borch-Johnsen & Kreiner 1987; Klein et al. 1989b; Rossing et al. 1996; Soedamah-Muthu et al. 2008), and a relative cardiovascular mortality of 37 and 4.2 when compared to the general population has been reported for type 1 diabetic patients with and without proteinuria, respectively (Borch-Johnsen & Kreiner 1987). Poor glycaemic regulation has also been recognized as a risk factor of mortality in type 1 diabetes in some (Moss et al. 1994; Shankar et al. 2007) but not all studies (Muhlhauser et al. 2000). For instance, in an 11-year follow-up of the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR), patients in the highest quartile of glycosylated haemoglobin had a relative risk of 2.42 and 3.28 for all-cause mortality and cardiovascular mortality, respectively, when compared to patients in the lower quartile (Shankar et al. 2007). It is still debatable whether DR is also associated with mortality in type 1 diabetes. Several studies have been performed (Klein et al. 1999; Rajala et al. 2000; Cusick et al. 2005; van Hecke et al. 2005) but long-term data from population-based studies have so far only been presented in a 16-year follow-up of the WESDR (Klein et al. 1999). Klein et al. found PDR to be associated with all-cause mortality in an age- and sex-adjusted model [hazard ratio (HR) 5.53, 95% confidence interval (CI) ] but not in a multivariate model (HR 1.28, 95% CI ). Furthermore, the study demonstrated a higher all-cause mortality as well as a greater risk of IHD for patients with PDR when compared to patients with NPDR. The study indicates an important confounding effect of other risk factors such as proteinuria, glycosylated haemoglobin, hypertension and smoking. It also suggests that PDR is more likely to be a risk factor of mortality in type 1 diabetes than NPDR. The higher risk of mortality for patients with the former may reflect more severe systemic morbidity among patients with PDR. Consequently, additional subgroup analyses are needed to determine this. However, the combined effect of retinopathy and other risk factors on mortality has not been investigated in previous studies. Visual impairment in type 1 diabetes Diabetic retinopathy is the most common reason for visual impairment and blindness for the working-age population of the Western world (Klein & Klein 1995). It has been demonstrated by the Diabetic Retinopathy Study (DRS) that properly timed panretinal photocoagulation decreases the 2-year risk of severe visual loss from 7% to 3% in patients with PDR (The Diabetic Retinopathy Study Research Group 1981). Likewise, for patients with DME, it was shown in the Early Treatment Diabetic Retinopathy Study (ETDRS) that focal photocoagulation halves the 3-year risk of severe visual loss (Early Treatment Diabetic Retinopathy Study Research Group 1985). Despite the implementation of strict glycaemic control as well as the use of laser photocoagulation, visual loss and blindness are still major concerns in type 1 diabetes. In a 14-year follow-up of the WESDR, Moss et al. (1998) found a 2.4% incidence of blindness. However, the WESDR results only included patients who had participated at the baseline examination as well as the 14-year followup. Consequently, data were not included for the patients who became blind and died prior to the follow-up. This is a concern because mortality has been found to be three times higher in blind patients with type 1 diabetes (Sjolie & Green 1987). Furthermore, it was not possible to identify risk factors of blindness in the WESDR 14-year follow-up. Updated long-term incidence data of blindness in type 1 diabetes that accounts for the competing risk of death and also identifies risk factors of blindness are called for. Cataract is another common cause of visual impairment in type 1 diabetes (Janghorbani et al. 2000; Esteves et al. 2008). Among patients with type 1 diabetes of the WESDR, cataract has been identified as the second most common reason for blindness (PDR was the most common) (Klein et al. 1984b). In type 1 diabetes, cataract has been associated with age (Nielsen & Vinding 1984; Klein et al. 1985, 1995; Di et al. 1999; Janghorbani et al. 2000), duration of diabetes (Nielsen & Vinding 1984; Klein et al. 1985; Janghorbani et al. 2000), glycaemic regulation (Klein et al. 1985; Di et al. 1999; Janghorbani et al. 2000; Kato et al. 2001) and DR (Nielsen & Vinding 1984; Klein et al. 1985, 1995; Janghorbani et al. 2000; Kato et al. 2001). However, long-term data on the incidence of cataract surgery as well as associated risk factors in type 1 diabetes are needed in order to facilitate handling of the increased burden 2

10 1DD 0.5DD Zone A Zone B Fig. 1. Left: Fundus photograph centred at the optic disk and used for vascular calibre analyses by the ivan program. Right: Identification of the largest arterioles (red) and venules (blue). See Methods Retinal vascular calibre analysis for description of data analysis. of the disease caused by the rising number of patients with diabetes. Retinal vascular analysis The human vascular tree is available for direct in vivo inspection in the eye. This allows examinations of the retinal microvascular system in different ways. The retinal vessels are part of the systemic microvasculature. Given this and the improved methods of capturing retinal images, investigations into the appearance of the retinal vascular system may be valuable in order to predict diabetes-related complications in other organs. Vascular diameter analysis is the most common method of analysing the retinal vascular system (Fig. 1). In previous studies, generalized arteriolar narrowing has been associated with age and hypertension (Hubbard et al. 1999; Leung et al. 2003b; Wong et al. 2003). For patients with diabetes, associations have been found between retinal vascular diameters and retinopathy (Klein et al. 2004b, 2007; Alibrahim et al. 2006; Cheung et al. 2008; Rogers et al. 2008), nephropathy (Klein et al. 2003, 2007; Wong et al. 2004b) and cardiovascular disease (Klein et al. 2003, 2004a, 2007). For instance, in multivariate analyses of patients with type 2 diabetes from the WESDR, wider arterioles (fourth versus first quartile) were associated with an increased 10-year incidence of DR [odds ratio (OR) 1.78]. Likewise, wider venules were associated with increased 14-year incidence of diabetic nephropathy (OR 2.08) and increased 22-year stroke mortality (HR 1.71) (Cheung et al. 2007; Klein et al. 2007). Conversely, narrow arterioles (first versus fourth quartile) were associated with increased 14-year incidence of lower extremity amputation (OR 2.20), 22-year all-cause mortality (HR 1.18) and 22-year stroke mortality (HR 1.47). So far, most studies on retinal vascular diameters have been carried out in the general population or in patients with type 2 diabetes. Studies to associate retinal vascular calibres with long-term micro- and macrovascular complications in type 1 diabetes are still lacking. Fractal analysis is another method of analysing the retinal vascular system. It is a more global way of evaluating the retinal vasculature in which the features of the vascular tree are summarized into a single parameter. In general, fractal analysis is based on the concept of self-similarity. That is, the parts of a pattern show the overall structure despite changes in magnification. Fractal patterns are common in nature and include snowflakes, leaves, trees and coastal lines. Fractal analyses have been known for some time (Mainster 1990; Daxer 1993a,b; Stosic & Stosic 2006). However, with a new method, International Retinal Imaging Software (IRIS-Fractal (National University of Singapore, Singapore; University of Sydney, Australia; and University of Melbourne, Australia), it is possible to grade each fundus photograph within 5 min when compared to up to 20 hr with earlier methods (Fig. 2) (Mainster 1990). In earlier studies, larger retinal fractals have been associated with early DR as well as retinal neovascularization (Daxer 1993b; Cheung et al. 2009). However, relations between retinal fractals and other micro- and macrovascular complications in type 1 diabetes have not yet been described so far. Such studies are important in order to evaluate retinal fractals as predictors of long-term complications in type 1 diabetes. Fig. 2. Left: Cropped fundus photograph centred at the optic disk. Right: Line tracing provided by the International Retinal Imaging Software (iris). See Methods Fractal analysis for description of data analysis. Noninvasive markers of micro- and macrovascular complications The burden of micro- and macrovascular complications is almost universal in type 1 diabetes. Besides a 97% prevalence of DR for patients with long-term type 1 diabetes (Grauslund 3

11 et al. 2009), we were recently able to report a prevalence of neuropathy, nephropathy (micro- and macroalbuminuria) and macrovascular disease of 52.7%, 33.2% and 21.9%, respectively, among patients with type 1 diabetes from a population-based cohort with a median duration of diabetes of 43 years (Grauslund et al. 2009). Along with duration of diabetes, hyperglycaemia has been identified as the most important risk factor of microvascular complications (The Diabetes Control and Complications Trial Research Group 1993). Despite a strict glycaemic control, some patients will, inevitably, face diabetesrelated complications. It is therefore vital to identify the patients in risk in order optimize treatment. Noninvasive markers of micro- and macrovascular complications are needed to accomplish this. Brain natriuretic peptide (BNP) is synthesized and secreted from the ventricular myocardium in response to myocyte stress and ischaemia (Levin et al. 1998; de Lemos et al. 2003). N- terminal probnp (NT-proBNP) is an inactive fragment that is cleaved from BNP and is considered more stable for analysis (Downie et al. 1999). NTproBNP has been associated with heart failure in non-diabetic patients (Hunt et al. 1997; McDonagh et al. 1998; Gardner et al. 2003), and with nephropathy (McKenna et al. 2001, 2005; Siebenhofer et al. 2003; Tarnow et al. 2005) and mortality (Hovind et al. 2003) in type 1 diabetes. However, relations to other diabetesrelated complications have not been examined in type 1 diabetes. Osteoprotegerin (OPG) is another potential marker of micro- and macrovascular complications in type 1 diabetes. Osteoprotegerin is a known regulatory molecule in bone turnover (Simonet et al. 1997) and is also present in vascular smooth muscle cells (Zhang et al. 2002) and endothelial cells (Malyankar et al. 2000). Besides being associated with coronary heart disease (Jono et al. 2002; Omland et al. 2008), elevated levels of OPG have been described in type 1 diabetes (Galluzzi et al. 2005; Rasmussen et al. 2006). Furthermore, in type 1 diabetic patients with nephropathy, OPG has been associated with mortality (Jorsal et al. 2008), glycaemic regulation, systolic blood pressure, kidney function and cardiovascular disease (Rasmussen et al. 2006). However, relations to retinopathy and neuropathy have not been described. Aims A population-based cohort of Danish patients with type 1 diabetes was examined in order to achieve the following aims: 1. To evaluate DR as a long-term marker of all-cause mortality in type 1 diabetes (Paper I). 2. To estimate the long-term incidence and associated risk factors of blindness (Paper II) and cataract surgery (Paper III) in type 1 diabetes. 3. To use retinal vascular analyses in order to investigate the associations between long-term micro- and macrovascular complications and retinal vascular diameters (Paper IV) and retinal fractals (Paper V) in type 1 diabetes. 4. To examine N-terminal pro-brain natriuretic peptide (Paper VI) and OPG (Paper VII) as noninvasive markers of micro- and macrovascular complications in type 1 diabetes. Methods Study population This thesis is based on the studies of a population-based cohort of patients with type 1 diabetes from Fyn County, Denmark. As of 1 July 1973, insulin prescriptions were used to identify all patients with type 1 diabetes from Fyn County with an onset of diabetes before the age of 30 (Green et al. 1981; Sjolie 1985). At that time, Fyn County had approximately inhabitants and was considered a demographically representative 9% sample of the general Danish population. It was estimated that the patient material was more than 98% complete (Green et al. 1981). Seven hundred and twenty-seven patients were identified. Of these, 413 (56.8%) were men and 314 (43.2%) were women. All patients who were still alive were asked to participate in a first examination in and a second examination in The principles of the Declaration of Helsinki were followed, and written informed consents were obtained from all patients at the follow-up examination. Furthermore, approval was obtained from the local ethics committee. First examination ( ) As of 1 June 1981, 627 of the 727 patients (86.2%) were still alive and living in Denmark. Prior to the examination, 96 patients had died and four had emigrated. The patients still available were invited to a clinical examination that took place between 1 June 1981 and 1 June A total of 577 patients (92.0%) chose to participate. Data were later lost on four patients, which left data available for 573 patients (321 men and 251 women). All examinations were performed by a single examiner [Anne Katrin Sjølie (AKS), Odense University Hospital, Odense, Denmark]. The first examination provides baseline data for the prospective studies of Papers I III. For these papers, patients were followed from the date of the baseline examination until 6 November 2006 or censoring, whichever came first. Patients could be censored owing to death, emigration, unwillingness to provide data to scientific projects or if the event of interest had occurred (Paper I: death, Paper II: blindness, Paper III: cataract surgery). A structured interview was performed as well as a clinical examination, blood and urine sampling and a full ophthalmological examination. Patients were asked about their smoking habits. Current and former smokers were considered to be smokers for the upcoming analyses. Body mass index (BMI) was defined as the body weight divided by the square of the height and expressed in kg m 2. Blood pressure was measured by an Erkameter sphygmomanometer (Morton Medical Ltd, London, UK) on one arm with the patient in sitting position after 10 min of rest. Blood measurements included haemoglobin A 1 (HbA 1 ) made as total HbA 1 with resin 70 (Bio-Rad, Hercules, CA, USA) at 20 C and ph Urine protein was measured from a single-spot urine, and proteinuria was considered present if protein was 0.5 g l. The best-corrected visual acuity was measured for both eyes. Both pupils were dilated using tropicamide 1%, and a slit lamp examination was per- 4

12 formed (Haag-Streit, Wedel, Germany). Ophthalmoscopy was performed and retinopathy was classified (no retinopathy, NPDR or PDR) by a single trained grader (AKS). The level of DR was determined by the worse eye. NPDR was defined as one or more of the following characteristics: microaneurysms, haemorrhages, hard exudates, CWS, venous beading or IRMA. Proliferative DR was present for patients who had newly formed vessels in addition to the aforementioned. Finally, maculopathy was defined as retinal thickening and or hard exudates in the macular area. Second examination ( ) As of 1 March 2007, 320 of the original 727 patients (44.0%) were still alive and living in Denmark. Of these, 208 (65.0%) agreed to participate in a clinical examination carried out between 1 March 2007 and 1 March Patients who had died prior to the second examination had a higher age in 1973 than patients who were still alive. The second examination provides data for the cross-sectional studies of Papers IV VII. Participation in the first examination was not needed to participate in the second examination. All examinations were performed by a single grader [Jakob Grauslund (JG)]. A full medical interview was performed including a history of smoking. Pack-years were calculated with one pack-year defined as 20 cigarettes smoked per day for 1 year. Data on macrovascular disease prior to the examination were provided by the Danish National Patient Registry. Macrovascular disease was defined as a history of one or more of the following conditions before the second examination: stroke, myocardial infarction, angina pectoris, percutaneous coronary intervention, coronary bypass surgery or limb amputation. An Omron M4 (Omron, Matsusaka, Japan) was used to measure blood pressure. The patient was placed in a sitting position after at least 10 min of rest and the mean value of the last two of three measurements were used. A monofilament test (Bailey Instruments Ltd, Manchester, UK) was used to test for neuropathy. A consistent force of 10 g was applied on three different sites on the planar surface of each foot and hand. With the patient unable to see the hand or foot, the pressure was applied until the filament buckled. Neuropathy was considered present for patients who were not able to feel the pressure in at least one of the four locations. Haemoglobin A 1c (HbA 1c ) was measured by cation-exchange chromatography using Tosoh G8 HPLC equipment (Medinor, Brøndby, Denmark). Nephropathy was tested in a single-spot urine sample that was taken at any time during the day. Levels of 0 19, and above 200 mg ml were used for the definition of normoalbuminuria, microalbuminuria and macroalbuminuria, respectively. Fundus photographs were captured in both eyes through dilated pupils. Topcon TRC-NW6S (Topcon, Tokyo, Japan) was used to capture nine 45 colour fields. IMAGEnet was used to auto-mosaic the photographs. Grading protocols were used (Klein et al. 1986, 1989a, 2008) according to the ETDRS adaptation of the modified Airlie House classification of DR (Early Treatment Diabetic Retinopathy Study Research Group 1991a,b). Proliferative DR was considered present for ETDRS-levels The final category was determined by the worse eye. All images were graded at the Ocular Epidemiology Reading Center, Madison, Wisconsin. Grading was performed by two independent, masked graders. When the two graders disagreed, the eye was regraded by a senior grader. If that grader agreed with either of the first two determinations, that result was accepted. If not, the case was referred to the most senior grader at the Reading Center for adjudication (Klein et al. 1986). Inter-observer and intra-observer variations and the validity of the systems have been presented elsewhere (Klein et al. 1984a, 1986, 1989a; Early Treatment Diabetic Retinopathy Study Research Group 1991b). Blindness All data on blindness were provided by the Danish Association of the Blind as of January This is a voluntary organization that is open for all patients who have either 1) a best-corrected visual acuity in the best eye equal to or below 6 60 or 2) eye complications (i.e. visual fields lower than 10 degrees) that subjectively correspond to a best-corrected visual acuity in the best eye equal to or below Because of the various benefits for the members of the organization (i.e. free counselling, discounts and access to special apartments), it is estimated that 88% of all blind patients with type 1 diabetes in Denmark are members of the organization (Sjolie & Green 1987). The cause of blindness for the patients who went blind during the follow-up was found by manually reviewing the patients charts at the Department of Ophthalmology, Odense University Hospital. This was carried out by the author. In the statistical analyses, the cumulative proportion of blindness was adjusted for mortality. Mortalityadjusted incidence rates were calculated as the number of blind patients per 1000 person-years at risk. Cataract surgery The Danish National Patient Registry was used to determine the date of cataract surgery for the patients who were operated during follow-up. In order to adjust for misclassifications, the charts of all patients were reviewed by hand by a single observer (JG). For patients who had had cataract surgery performed in both eyes, the date of the first surgery was used. In a similar way as in the statistical analyses of blindness, the cumulative incidence of cataract surgery was adjusted for mortality. Mortalityadjusted incidence rates were calculated as the number of patients who had had cataract surgery performed in at least one eye per 1000 person-years at risk. Retinal vascular calibre analysis The semi-automatic computer program ivan was used for the measurement of retinal vascular calibres (Knudtson et al. 2003). Because of a well-known high inter-eye correlation (Leung et al. 2003a), only the right eye was graded. Left eye photographs were used if photographs of the right eye had not been captured or were ungradable. For each patient, one disc-centred image was used. The ivan program automatically identified all vessels within an area of one-half to 5

13 one disc diameter from the optic disc (Fig. 1). The grader then chopped the vessels and identified which were arteries and which were veins. Using formulas proposed by Knudtson et al. (2003), the program then measured the diameters of the six largest arteries and veins and summarized these into the central retinal arteriolar equivalent (CRAE) and the central retinal vein equivalent (CRVE). Finally, the arteriolar venular ratio was defined as the ratio of CRAE to CRVE. The arteriolar venular ratio (AVR) is often used as a common measure of retinal vascular calibres in order to minimize the effect of differences in magnifications that have been associated with refractive errors (Wong et al. 2004c). All gradings were made by a single trained grader (JG). In order to determine inter-grader reproducibility, 20 randomly chosen photographs were regraded by a senior grader [Lauren Hodgson (LH), Centre for Eye Research Australia, Melbourne, Australia]. A high reproducibility was found (intraclass correlation 0.96 and 0.99 for CRAE and CRVE, respectively). Fractal analysis The semi-automatic computer program International Retinal Imaging Software (iris-fractal) from National University of Singapore, Singapore, University of Sydney and University of Melbourne, Australia, was used for fractal analysis. A standard protocol was used in which the fractal dimension (D f ) was calculated within a predefined region, 3.5 optic disc radii centred on the optic disc (Fig. 2) (Liew et al. 2008). The line tracing provided by the computer was then checked by the grader, and artefacts such as peripapillary atrophy, choroidal vessels and pigmentary abnormalities were then removed. iris-fractal then used the box-counting method to calculate fractal dimension. The boxcounting method is an established method used to calculate D f of structures such as retinal vasculature that are not perfectly self-similar (Mainster 1990; Stosic & Stosic 2006; Macgillivray et al. 2007). In the case of the analyses of retinal vessel calibres, fractals were graded for the right eye with the left eye as backup when a right eye fractal could not be determined. Reverse analyses (using left eye primarily and right eye as backup) were also performed in post hoc analyses. All grading was performed by a single trained grader (JG), and as with the calibre analyses, the inter-grader reliability was determined by the regrading of 20 randomly chosen photographs by a senior grader (LH). A high inter-grader reliability was found (intraclass correlation coefficient 0.98). Blood measurements In Paper VI, NT-proBNP was measured in blood drawn by venipuncture. Heparin-containing tubes were used and centrifuged for 10 min at 2000 g within 1 hr of collection. Afterwards, the samples were stored at )80 C until analysis. Dedicated agents from the Modular System (Roche Diagnostics, Basel, Switzerland) were used for analysis. Blood drawn by venipuncture was also used for plasma OPG analysis in Paper VII. A sandwich ELISA from R&D systems (Minneapolis, MN, USA) was used with mouse antihuman OPG used as captured antibody and a biotinylated goat antihuman OPG used for detection. We used recombinant human OPG for calibration with an analytical range of the assay way of pg ml. All plasma samples were diluted 1:3 and duplicate measurements were made. The intra-assay coefficient of variation was 3% (Rasmussen et al. 2006). Results Retinopathy and mortality (Paper I) As of November 2006, 51.8% (297 of 573) who participated at the baseline examination in were still alive. The rest had died (44.7%), emigrated (0.3%) or refused to provide data for scientific studies (3.0%). There was no statistically significant difference in mortality between men and women (45.8 versus 43.3%, p = 0.52). Patients were stratified according to the level of retinopathy at the baseline examination: no retinopathy (n = 143), nonproliferative retinopathy (n = 291) and proliferative retinopathy (n = 139). The survival rate was significantly lower (p < ) for patients with PDR (33.8%) than for patients with no retinopathy (62.9%) or NPDR (55.0%). The patients in the three categories also differed according to age, duration of diabetes, smoking, glycaemic regulation, systolic and diastolic blood pressure and rate of proteinuria (see Paper I Table 1). A survival analysis was made in order to determine whether the level of retinopathy per se was a predictor of mortality (see Paper I Table 2). In an age- and sex-adjusted model, HR was 2.04 (95% CI , p < 0.001) for patients with PDR when compared to patients with no DR. On the other hand, NPDR was not associated with increased mortality (HR 1.01, 95% CI , p = 0.97). In order to determine whether other risk Table 1. Multiple logistic regression analysis for the 25-year incidence of blindness according to characteristics at the baseline examination in Characteristics Difference Odds Ratio (95% confidence interval) Age (years) 10 years 1.28 ( ) Duration (years) 10 years 0.71 ( ) Gender Female versus male 0.97 ( ) HbA 1 (%) 1% 1.69 ( ) Proteinuria Present versus absent 0.53 ( ) Smoking Ever versus never smoked 0.54 ( ) Systolic blood pressure (mmhg) 10 mmhg 1.10 ( ) Diastolic blood pressure (mmhg) 10 mmhg 0.79 ( ) Retinopathy Level versus no retinopathy Nonproliferative without maculopathy 1.34 ( ) Proliferative without maculopathy 3.32 ( ) Nonproliferative with maculopathy 6.18 ( ) Proliferative with maculopathy 8.61 ( ) 6

14 Table 2. Cox regression model of risk factors for cataract surgery according to characteristics at the baseline examination in Characteristics Increase difference factors were also associated with increased mortality, these were added to the age- and sex-adjusted model one by one. There was no difference in HR when duration of diabetes, smoking, glycaemic regulation, systolic or diastolic blood pressure or BMI were added. However, when proteinuria was added to the model, the hazard ratio of PDR declined to 1.49 (95% CI , p = 0.054). In a multivariate model, there was no further reduction in the impact of PDR on mortality. It was therefore concluded that PDR and proteinuria were the most important predictors of mortality. In order to evaluate the combined effect of PDR and proteinuria on mortality, patients were stratified into four groups according to the combined level of PDR (present absent) and proteinuria (present absent) at Hazard Ratio (95% confidence interval) Unadjusted Multivariate Age (years) 10 years 2.28 ( ) 1.89 ( ) Duration (years) 10 years 2.16 ( ) 1.16 ( ) Sex Female versus male 0.97 ( ) 0.82 ( ) HbA 1 (%) 1% 0.72 ( ) 0.82 ( ) Proteinuria Present versus absent 1.00 ( ) 1.00 ( ) Smoking Ever versus never smoked 1.46 ( ) 1.26 ( ) Systolic blood pressure 10 mmhg 1.21 ( ) 1.09 ( ) Diastolic blood pressure 10 mmhg 1.05 ( ) 0.94 ( ) Maculopathy Present versus absent 3.08 ( ) 1.89 ( ) Retinopathy Level versus no retinopathy Nonproliferative retinopathy 2.31 ( ) 1.33 ( ) Proliferative retinopathy 4.83 ( ) 1.76 ( ) baseline (Fig. 3). The 25-year survival was significantly lower (p < 0.001) for patients who had both PDR and proteinuria (16.7%) when compared to patients with neither (66.4%), PDR only (48.2%) or proteinuria only (33.3%). It was also demonstrated that the highest risk of mortality for patients with PDR as well as proteinuria was within the first 10 years after the baseline examination. The 10-year survival rate for these patients was only 22.2% when compared to 86.8%, 79.0% and 70.4% for the other groups, respectively. Long-term incidence of blindness and associated risk factors (Paper II) Twenty-four of the 573 patients (4.2%) who participated at the baseline examination in were already blind. When compared to Fig. 3. All-cause mortality rate according to the level of baseline retinopathy and proteinuria. those who were not blind at baseline, patients who were blind were more likely to be women, older, had a longer duration of diabetes, had better glycaemic regulation, a higher level of retinopathy and were more likely to have maculopathy (see Paper II Table 1). Five hundred and forty-seven patients were not registered as blind at the time of the baseline examination. Of these, 41 (7.5%) went blind during the follow-up. According to the charts, diabetes was the leading cause of blindness among 29 of 30 (96.6%) for whom the cause of blindness could be established. There was no statistically significant difference in the 25-year incidence of blindness among men and women (8.0 versus 6.8%, p = 0.61). The median age at the time of blindness was a little higher among women (53.8 years) than men (45.0 years). Mortality was higher (p = 0.02) for those who were registered as blind during follow-up (61.0%) when compared to those who were not (42.1%). Mortality-adjusted incidence rates were calculated. For all patients, the mortality-adjusted incidence rate of blindness was 4.11 per 1000 personyears at risk. The overall 25-year cumulative incidence rate of blindness was 9.5% (95% CI %). There was no statistically significant difference between men (10.1%) and women (8.9%) (see Paper II Fig. 2). Finally, risk factors of blindness were evaluated. In univariate models, maculopathy and level of retinopathy were able to predict blindness, as opposed to gender, HbA 1, proteinuria, smoking, blood pressure and visual acuity (see Paper II Table 2). For instance, the 25-year incidence of blindness was 4.3%, 6.2% and 14.3% for patients who had no retinopathy, nonproliferative retinopathy and proliferative retinopathy at baseline, respectively (p = ). The importance of retinopathy in the development of blindness was confirmed in a multivariate analysis adjusted for age, duration of diabetes, gender, glycaemic regulation, proteinuria, smoking, systolic and diastolic blood pressure (Table 1). When compared to patients with no retinopathy at baseline, OR of blindness was 6.18 (95% CI ) and 8.61 ( ) for patients with maculopathy in 7

15 combination with NPDR and PDR, respectively. Furthermore, blindness was also predicted by the level of HbA 1. OR of blindness was 1.69 (95% CI ) for each 1%-point increase in baseline HbA 1. Cataract surgery and risk factors (Paper III) Of the 573 patients who participated in the baseline examination, cataract surgery had already been performed in at least one eye of 11 patients. Patients who had not already been operated at baseline were younger, had a shorter duration of diabetes, a lower diastolic blood pressure and a lower mortality rate during follow-up when compared to those who had been operated (see Paper III Table 1). Of the 562 patients at risk at baseline, cataract surgery was performed in at least one eye in 117 patients during follow-up, corresponding to a crude 25-year cumulative incidence of cataract surgery of 20.8% and a mortality-adjusted 25-year cumulative incidence of cataract surgery of 29.4% (95% CI %). This corresponds to 117 patients operated in 9841 risk-years or 11.9 per 1000 person-years at risk. There was no statistically significant difference between men and women. For the patients who underwent surgery during followup, the median age and duration at the time of the first surgery was 59.3 and 42 years, respectively. In univariate analyses, cataract surgery was positively associated with age, duration of diabetes, maculopathy and the level of retinopathy at baseline (see Paper III Table 2). A Cox regression was performed in order to identify risk factors of cataract surgery in an unadjusted as well as a multivariate model adjusted for age, duration, sex, glycaemic regulation, smoking, blood pressure, maculopathy and level of retinopathy (Table 2). In the multivariate model, age (HR 1.89 for any 10-year increase, 95% CI ) and maculopathy (HR 1.89, 95% CI ) were both associated with incident cataract surgery. The age-specific prevalence of cataract surgery in any eye was compared with data from the Blue Mountains Eye Study: a population-based study of vision and various eye diseases in an elderly urban population from Blue Mountains in Australia (Mitchell et al. 1997). In the present study, a higher prevalence of cataract surgery was found in all age groups. The highest ratio was found for patients in the age groups of years (ratio 11.9, 22.6 versus 1.9%) and years (ratio 11.7, 42.2% versus 3.6%). Retinal vascular calibre and micro- and macrovascular complications (Paper IV) This was a cross-sectional study of 208 patients. Twenty patients had to be excluded because retinal vessel calibres could not be measured in either eye. Median age and duration of diabetes for the remaining 188 patients were 57.9 and 42 years, respectively. Clinical differences between the patients with gradable photographs and those without are shown in Table 3. In Table 3. Clinical characteristics of patients according to status of vessel grading. Ungradable Gradable p value n Age (years) 62.5 ± ± * Sex (% male) Duration of diabetes (years) 50 ± ± 10 <0.01* HbA 1c (%) 8.0 ± ± Systolic blood pressure (mmhg) 161 ± ± Diastolic blood pressure (mmhg) 77 ± ± History of smoking (%) Proliferative retinopathy (%) <0.01* Microalbuminuria (%) Macroalbuminuria (%) <0.01* Neuropathy (%) * Macrovascular disease (%) <0.01* Continuous data presented as median and interquartile range (25th to 75th centile) and categorical data presented as per cent. *Statistically significant (p < 0.05). general, patients who were excluded were older, had a longer duration of diabetes and were more likely to have proliferative retinopathy, macroalbuminuria and macrovascular disease. Central retinal arteriolar equivalent, CRVE and AVR were associated with age, duration of diabetes, glycaemic regulation, systolic and diastolic blood pressure, and pack-years of smoking in univariate analyses (see Paper IV Table 1). Arteriolar narrowing was associated with poor glycaemic regulation as well as higher systolic blood pressure. Likewise, venular narrowing was also associated with a higher systolic blood pressure, and, furthermore, high CRVE was also related directly to pack-years of smoking. Finally, associations were found between AVR and duration of diabetes as well as pack-years of smoking. Univariate analyses were also performed to examine relations between vascular diameters and diabetesrelated complications (Table 4). It was demonstrated that arteriolar narrowing and AVR were associated with nephropathy and macrovascular disease. No such associations were found for CRVE. Although not part of the original protocol, it was also observed that both vessel calibres were significantly lower in patients with PDR (CRAE: versus lm, p < 0.01, CRVE: versus lm, p < 0.01, AVR: 0.65 versus 0.68, p = 0.02). Besides the univariate analyses, a logistic regression was performed to evaluate the relation between retinal vascular diameters and micro- and macrovascular disease. An age- and sex-adjusted as well as a multivariate model was presented (see Paper IV Table 2). In multivariate analyses, a lower CRAE was statistically significantly associated with nephropathy (OR 2.17, 95% CI per SD decrease in CRAE) and macrovascular disease (OR 3.17, 95% CI per SD decrease in CRAE). The same associations were found for AVR while CRVE remained unrelated to all complications. Retinal fractals and diabetes-related complications (Paper V) Two hundred and eight patients participated in the study. Of these, grading was not possible in either eye of 8