Conall T. Morgan, MD, MRCPCH, Cedric Manlhiot, BSc, Brian W. McCrindle, MD, MPH, and Anne I. Dipchand, MD, FRCPC

http://www.jhltonline.org Outcome, incidence and risk factors for stroke after pediatric heart ation: An analysis of the International Society for Heart and Lung Transplantation Registry Conall T. Morgan, MD, MRCPCH, Cedric Manlhiot, BSc, Brian W. McCrindle, MD, MPH, and Anne I. Dipchand, MD, FRCPC From the Labatt Family Heart Centre, Department of Pediatrics, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada. KEYWORDS: heart ation; mortality; pediatric; stroke; congenital heart disease; outcome; mechanical support BACKGROUND: In the registry of the International Society for Heart and Lung Transplantation (ISHLT), cerebrovascular accidents are the fifth most common cause for mortality after pediatric heart ation (PHTx), but details are lacking in the literature. The purpose of this analysis of the ISHLT registry was to determine the prevalence, risk factors and outcomes of stroke after PHTx. METHODS: Data from the ISHLT registry (1998 to 2010) were used to identify all patients whose primary ation was performed at o18 years of age. Of the 10,441 s reviewed, 9,837 primary s and 604 res were analyzed. RESULTS: Three hundred thirty-three (3%) patients had a stroke after PHTx; 54% were male, median age at PHTx was 6 years (0 to 17 years), and 44% had a diagnosis of congenital heart disease (CHD). Freedom from stroke was 99% at 1 month, 97% at 5 years, 95% at 10 years and 91% at 20 years post- PHTx. After a stroke, survival at 1 month, 1 year and 5 years was 83%, 69% and 55%, respectively. Multivariable independent risk factors for stroke included a primary diagnosis of congenital heart disease [hazard ratio (HR) 1.4 (1.1 to 1.7), p ¼ 0.01], previous stroke [HR 4.5 (3.2 to 6.2), p o 0.001], history of aborted sudden death [HR 1.5 (1.1 to 2), p ¼ 0.01], ventricular assist device [HR 1.5 (1.1 to 2.2), p ¼ 0.03] or extracorporeal membrane oxygenation [HR 1.7 (1.2 to 2.2), p ¼ 0.01], post-operative dialysis [HR 3.3 (2.3 to 4.7), p o 0.001], infection requiring antibiotics before discharge [HR 1.9 (1.4 to 2.5), p o 0.001], pacemaker implantation [HR 1.6 (1 to 2.5), p ¼ 0.04] or drug-treated hypertension [HR 1.4 (1.1 to 1.8), p ¼ 0.003] during follow-up. CONCLUSIONS: Stroke after pediatric heart ation is associated with increased mortality. Congenital heart disease and mechanical support both portend greater risk, in addition to markers of increased pre- and post- medical acuity. J Heart Lung Transplant 2016;35:597 602 r 2016 International Society for Heart and Lung Transplantation. All rights reserved. Reprint requests: Anne I. Dipchand, MD, FRCPC, Labatt Family Heart Centre, Department of Pediatrics, The Hospital for Sick Children, University of Toronto, 555 University Avenue, Toronto, ON M5G 1X8, Canada. Telephone: þ416-813-7500. Fax: 416-813-7547. E-mail address: anne.dipchand@sickkids.ca 1053-2498/$ - see front matter r 2016 International Society for Heart and Lung Transplantation. All rights reserved. http://dx.doi.org/10.1016/j.healun.2016.01.1226 Pediatric heart ation continues to grow worldwide since the first ation in 1967. Congenital heart disease (CHD) represents the most common indication for ation, followed closely by cardiomyopathy. 1 The first year after ation has the highest risk of mortality, with graft failure and technical issues being the

598 The Journal of Heart and Lung Transplantation, Vol 35, No 5, May 2016 most common cause of death (30%), followed by multiorgan failure (16%), infection (14%) and rejection (12%). 1 Importantly, cerebrovascular events represent the fifth most common cause of mortality in the early post-operative period (10%). 1 The incidence of stroke in children is estimated at 2.6 per 100,000 per year. 2 Patients with CHD are at higher risk for acute ischemic stroke. 3 Almost one third of pediatric patients with acute ischemic stroke have CHD. 4 Risk factors for stroke in this group include a prothrombotic condition, a mechanical valve and an acute illness at the time of the stroke. 5 The risk of peri-operative stroke increases in those patients with CHD who are older at the time of surgery and in those who undergo reoperation or have a longer duration of cardiopulmonary bypass. 4 Overall, the frequency of stroke after surgery for CHD is low, occurring in just over 0.5% of cases, 4 but has been reported to be as high as 3%. 6 In those patients who have had a stroke, 25% occur after a surgical or catheter-based procedure. 7 Recurrence of acute ischemic stroke in patients with CHD occurs in up to 30%. 5,8,9 Although approximately one third of pediatric patients after stroke remain asymptomatic, it is a major source of morbidity and mortality, with just over a third having a seizure disorder or a persistent neurologic deficit and 20% dying after acute ischemic stroke. 9 The burden of stroke in the pediatric heart population is not well known. According to the registry of the International Society for Heart and Lung Transplantation (ISHLT), cerebrovascular accidents represent the fifth most common cause of death after pediatric heart ation (PHTx). The purpose of this analysis of the ISHLT registry was to determine the prevalence, risk factors and outcomes of stroke after PHTx. Methods Study design This study was a retrospective cohort analysis of ISHLT registry s performed from 1988 to 2010. The ISHLT registry is an international longitudinal voluntary database that collects data from the time of and throughout follow-up. Study population Heart recipients whose primary was performed at o18 years of age were included in our study data. Those patients who had any stroke, irrespective of etiology, were the focus of our analysis. The comparison group included all ations, both primary and reations, within the registry over the same time period. Patients who underwent multiorgan ation were excluded from the analysis. Data collection Risk factors for stroke were divided into patients characteristics, pre- characteristics, characteristics, post complications and complications during follow-up. Outcomes were evaluated at the time of yearly follow-up. Statistical analysis Data are described as mean with standard deviation or as median with the 5th and 95th percentiles and frequencies. Variables were excluded when there were no events in the patient population, when there was a high level of missing data (470%), and wherever rates were o1% or variables were considered not relevant to the analysis. Time-related events (morbid events and survival) were modeled in multiphase parametric hazard models that decompose risk over time in up to 3 additive, overlapping phases of risk (descriptively labeled as early, constant and late, but actually representing different mathematical functions to model specific patterns of event distribution over time). The analysis was performed using the HAZARD procedure for SAS available from http://www.clevelandclinic.org/heartcenter/hazard. A complete mathematical description and validation of the procedure has been published. 10 Associations between patients characteristics and hazard of stroke were first tested at the univariable level. Associations with p o 0.10 were then included in a multivariable regression model with backward selection of variables to obtain a final model. Bootstrap bagging resampling (500 random samples) was used to determine the internal validity of the multivariable risk models using a reliability threshold of 50% (i.e., variable selected in at least 50% of the random samples). Mean imputation was used to account for missing variables. All statistical analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC). Results Patients characteristics Patients characteristics are summarized in Table 1. Univariable risk factors for stroke are summarized in Table 2 and multivariable risk factors are summarized in Table 3. Patients with stroke were more likely to have had: CHD [hazard ratio (HR) 1.31 (1.05 to 1.62), p ¼ 0.02], although the proportion with cardiomyopathy vs CHD was similar in the stroke cohort (46% vs 44%); a pre- history of aborted sudden death [HR 1.63 (1.25 to 2.13), p o 0.001]; previous stroke [HR 6.12 (4.13 to 9.06), p o 0.001]; hospitalization before ation [HR 1.52 (1.17 to 1.97), p ¼ 0.002]; been hospitalized for 431 days (38% vs 29%, p ¼ 0.04); an infection requiring antibiotics before ation [HR 2.08 (1.57 to 2.74), p o 0.001]; a need for dialysis before ation [HR 2.46 (1.15 to 5.24), p ¼ 0.02]; and received a blood transfusion before ation [HR 2.88 (1.06 to 1.46), p ¼ 0.007]. At the time of ation, patients with stroke were more likely to require inotropes [HR 1.26 (1.00 to 1.59), p ¼ 0.05], mechanical ventilation [HR 2.00 (1.56 to 2.56), p o 0.001] or mechanical circulatory support in the form of a ventricular assist device (VAD) [HR 2.17 (1.55 to 3.04), p o 0.001] or extracorporeal membrane oxygenation (ECMO) [HR 2.84 (1.96 to 4.11), p o 0.001]. Reation [HR 1.76 (1.17 to 2.66), p ¼ 0.007] and later year of ation was also associated with stroke [HR 1.11 (1.08 to 1.13)]. Risk factors that remained significant for stroke on multivariable analyses included CHD [HR 1.35 (1.08 to 1.69), p ¼ 0.009], a history of aborted sudden death [HR 1.48 (1.11 to 1.97), p ¼ 0.007], later year

Morgan et al. Stroke After PHTx 599 Table 1 Patients Demographics and Pre-ation Characteristics N Value (%) Recipient age (years) 10,441 o1 year 2,499 (24%) 1 10 years 3,608 (35%) 11 18 years 4,176 (40%) 418 years 159 (2%) Male gender 5,963 (57%) Primary diagnosis 10,441 Cardiomyopathy 5,508 (53%) Dilated cardiomyopathy 4,162 Hypertrophic cardiomyopathy 234 Myocarditis 442 Restrictive cardiomyopathy 481 Other cardiomyopathy 189 Congenital heart disease 3,873 (37%) Reation 563 (5%) Other 497 (5%) Admission status immediately 3,336 before ation Admitted day of 1,314 (39%) or before Admitted o31 days 1,040 (31%) Admitted 431 days 982 (29%) Medical condition at time of 7,096 Not hospitalized 2,358 (33%) Inpatient but not in ICU 900 (13%) In ICU 3,838 (54%) History of anti-arrhythmics or 5,490 1,104 (20%) amiodarone use History of symptomatic 5,494 117 (2%) cerebrovascular disease Cerebrovascular event before 5,006 187 (4%) History of dialysis before 5,042 81 (2%) History of drug-treated systemic 5,473 769 (14%) hypertension Infection requiring IV therapy 5,046 879 (17%) 2 weeks before Aborted sudden death before 5,190 1,187 (23%) Patient on life support at time of 6,819 ECMO 407 (6%) Inotropes 3,153 (46%) PGE 381 (6%) Ventilator 1,503 (22%) VAD 636 (9%) ECMO, extracorporeal membrane oxygenation; ICU, intensive care unit; PGE, prostaglandin estradiol; VAD, ventricular assist device. [HR 1.09 (1.07 to 1.12), p o 0.001], previous stroke [HR 4.47 (3.22 to 6.19), p o 0.001] and need for either VAD [HR 1.51 (1.05 to 2.19), p ¼ 0.03] or ECMO [HR 1.70 (1.15 to 2.19), p ¼ 0.007]. Peri- and post-operative characteristics There were no relevant donor factors that impacted the risk of stroke in this large cohort of PHTx recipients. Patients with stroke had a longer ischemic time per 60 minutes [HR 1.13 (1.04 to 1.22), p ¼ 0.004] and an overall longer postation length of stay per 10 days [HR 1.08 (1.06 to 1.10), p o 0.001]. After ation, they were more likely to have required dialysis before discharge [HR 5.2 (3.77 to 7.19), p o 0.001], cardiac reoperation before discharge [HR 3.12 (2.22 to 4.38), p o 0.001], or infection requiring antibiotics before discharge [HR 2.68 (2.09 to 3.44), p o 0.001]. During follow-up, patients with stroke Table 2 Univariable Risk Factors Associated With Any Stroke Univariable risk factors HR (95% CI) p-value Patient factors Year of ation 1.11 (1.08 1.13) o0.001 (per year) Diagnosis of congenital heart 1.31 (1.05 1.62) 0.02 disease Reation 1.76 (1.17 2.66) 0.007 History of stroke before 6.12 (4.13 9.06) o0.001 Drug-treated infection before 2.08 (1.57 2.74) o0.001 Aborted sudden death before 1.63 (1.25 2.13) o0.001 Dialysis before 2.46 (1.15 5.24) 0.02 Blood transfusion before 2.88 (1.06 7.81) 0.04 Hospitalization before 1.52 (1.17 1.97) 0.002 Lower cardiac output before 1.25 (1.06 1.46) 0.007 (liters/min) Support at the time of : inotropic support 1.26 (1.00 1.59) 0.05 Support at the time of 2.00 (1.56 2.56) o0.001 : ventilation Support at the time of 2.17 (1.55 3.04) o0.001 : VAD Support at the time of 2.84 (1.96 4.11) o0.001 : ECMO Operative factors Duration of hospital stay 1.08 (1.06 1.10) o0.001 (per 10 days) Duration of ischemic time 1.13 (1.04 1.23) 0.004 (per 60 minutes) Post- factors Post- cardiac 3.12 (2.22 4.38) o0.001 reoperation before hospital discharge Post- dialysis before 5.20 (3.76 7.20) o0.001 hospital discharge Post- drug-treated 2.68 (2.09 3.44) o0.001 infection before hospital discharge Pacemaker implantation during follow-up 1.55 (0.98 2.44) 0.06 Any drug-treated hypertension 1.64 (1.28 2.10) o0.001 during follow-up Any hospitalization for infection during follow-up 1.53 (1.20 1.95) 0.001 ECMO, extracorporeal membrane oxygenation; HR, hazard ratio; VAD, ventricular assist device.

600 Table 3 Multivariable Model of Risk Factors Associated With Any Stroke The Journal of Heart and Lung Transplantation, Vol 35, No 5, May 2016 Multivariable risk factors Rel (%) HR (95% CI) p-value Patient factors Year of ation (per year) 100% 1.09 (1.07 1.12) o0.001 Diagnosis of congenital heart disease 81% 1.35 (1.08 1.69) 0.009 History of stroke before 100% 4.47 (3.22 6.19) o0.001 Aborted sudden death before 88% 1.48 (1.11 1.97) 0.007 Lower cardiac output before (liters/min) 84% 1.23 (1.06 1.44) 0.007 Support at the time of : VAD 68% 1.51 (1.05 2.19) 0.03 Support at the time of : ECMO 74% 1.70 (1.15 2.19) 0.007 Post- factors Post- cardiac reoperation before hospital discharge 96% 2.02 (1.40 2.92) o0.001 Post- dialysis before hospital discharge 100% 3.29 (2.30 4.72) o0.001 Post- drug-treated infection before hospital discharge 100% 1.91 (1.44 2.53) o0.001 Pacemaker implantation during follow-up 61% 1.58 (1.01 2.47) 0.04 Any drug-treated hypertension during follow-up 92% 1.41 (1.12 1.78) 0.003 ECMO, extracorporeal membrane oxygenation; HR, hazard ratio; Rel, reliability in bootstrap samples; VAD, ventricular assist device. were more likely to have needed a pacemaker [HR 1.55 (0.98 to 2.44), p ¼ 0.06], re-hospitalization for infection after discharge [HR 1.53 (1.20 to 1.94), p ¼ 0.001] and treatment for drug-related hypertension [HR 1.64 (1.28 to 2.10), p o 0.001]. At multivariable analyses, post- need for dialysis before discharge remained the strongest independent risk factor for stroke [HR 3.29 (2.30 to 4.72), p o 0.001]; other strong predictors were cardiac reoperation before discharge [HR 2.02 (1.40 to 2.92), p o 0.001] and post-ation infection requiring antibiotics before discharge [HR 1.91 (1.44 to 2.53), p o 0.001]. At followup, the need for a pacemaker [HR 1.58 (1.01 to 2.47), p ¼ 0.04] and drug-treated hypertension [HR 1.41 (1.12 to 1.78), p ¼ 0.003] remained as independent risk factors for stroke. Overall outcomes Freedom from stroke in PHTx recipients was 99% at 1 month, 97% at 5 years, 95% at 10 years and 91% at 20 years. Figure 1 depicts survival after a stroke post; survival at 1 month, 1 year and 5 years was 83%, 69% and 55%, respectively, which compared unfavorably to patients without stroke. disease are known to be at greater risk for stroke than the general population. Hoffman et al 11 reported an overall incidence of 132 of 100,000 per year, 16-fold higher than the reported incidence in the general pediatric population of between 2 and 8 of 100,000 per year. 12,13 In our population, not surprisingly, a diagnosis of congenital heart disease was a strong independent risk factor for stroke. Previous stroke was the greatest independent risk factor for an additional stroke in our population. Stroke recurrence risk in the CHD population is reported to be as high as 27%, but lower (around 20%) if the primary stroke was not procedurally related. 5 Recurrence of stroke in the non-chd pediatric population is rare, 14 so CHD and the need for repeated surgery or catheter-based interventions may confer an ongoing risk for stroke in this population. Patients with single-ventricle circulations or mechanical circulatory support, such as VAD or ECMO, are at even greater risk of stroke. Reported rates of stroke while on a VAD range between 28% and 34%. Almond et al 15 undertook a large multicenter study of 204 children supported by a Berlin Heart EXCOR device. Stroke was found in 33% of patients and was the leading cause of death. Thromboembolic strokes outnumbered hemorrhagic stroke by more than double. In 2012, Fraser et al 16 prospectively compared a cohort of 24 patients with a VAD with a Discussion This analysis from the ISHLT registry has included the largest cohort of pediatric patients with cerebrovascular events/stroke after heart ation. Stroke after ation is associated with decreased short- and long-term survival with risk factors for stroke that reflect sicker patients both pre- and during the acute post admission. Post- morbidities related with stroke included renal dysfunction, arrhythmia, hypertension and infection. There is a paucity of data in the literature regarding the outcomes and risk factors for stroke in the PHTx population; thus, our study has provided valuable insight into this rare phenomenon. Children with congenital and acquired heart Figure 1 Survival post- after a stroke.

Morgan et al. Stroke After PHTx 601 historical control group undergoing ECMO. Stroke occurred in 29% in both groups of patients. In our population, patients on ECMO and patients requiring a VAD at the time of were more likely to have a stroke. The need for pre- and post-operative dialysis is an interesting predictor for stroke after ation. Both dialysis after ation and pre-existing renal dysfunction have been well described as predictors for increased short- and long-term mortality in the heart population, 1,17 and the phenomenon of decreased renal function over time after heart ation was not unexpected. 18 In this study, we did not find pre-existing renal dysfunction (abnormal pre creatinine) to be a predictor for stroke. The need for pre- and post-operative dialysis may be a surrogate marker for a cohort of patients who are sicker after and therefore at greater risk of either ischemic or hemorrhagic stroke. There have been no published studies or case reports of the association with dialysis and stroke in the pediatric population. Therefore, this finding would prompt greater neurologic surveillance of this group of patients. Longer hospital stay, pre- need for inotropes and mechanical ventilation, longer bypass times, mechanical circulatory support such as ECMO or VAD, greater need for reoperation before discharge and longer dependence on inotropy post-ation, like dialysis, are all indicators for prediction of stroke. This likely represents a population of patients who are more complex and more unwell before and certainly after ation. Patients who require antibiotics for infection before discharge are twice as likely to have a stroke. We are not sure why this is so, but systemic inflammation leading to a pro-thrombotic state may be a contributor. 19,20 Similarly to the aforementioned predictors for stroke, infection may represent a sicker group of patients who have spent a longer time in hospital and therefore at greater risk of nosocomial infection. Hypertension after ation and need for pacemaker insertion are not unexpected risk factors for stroke. Hypertension is a common morbidity after heart ation, reported in as many as 25% of recipients at any given time post-, with a cumulative incidence as high as 60%. 21 It is easy to postulate that hypertension that is sub-optimally controlled could predispose to stroke. Similarly, the need for a pacemaker likely implies the presence of arrhythmia, which is a risk factor for thromboembolic events and subsequent stroke. We found that stroke is a relatively uncommon morbidity in the heart population, with 95% of patients free from stroke at 10 years. Although stroke in this population tends to be non-fatal, it does lead to increased short- and long-term mortality. As highlighted earlier, this may represent a cohort of patients with increased medical acuity and complexity before, and a longer, more complicated post- course. Study limitations Our study has all the inherent limitations expected of a retrospective, voluntary, multi-institutional investigation, including incomplete data, variations in reporting of adverse events and different patient data analyzed at different timepoints. The type of stroke, whether embolic or hemorrhagic, was not known and thus makes it harder to identify applicable risk factors. However, this is the largest cohort of pediatric heart patients and its value outweighs its limitations. Pediatric stroke is a rare phenomenon, so further knowledge about this high-risk population would be beneficial. In conclusion, stroke after PHTx is a relatively uncommon yet important finding, because the presence of stroke is associated with worse short- and long-term survival. Patients with congenital heart disease and those with previous stroke are at greater risk for stroke. Disclosure statement The authors have no conflicts of interest to disclose. References 1. Dipchand AI, Kirk R, Edwards LB, et al. The Registry of the International Society for Heart and Lung Transplantation: sixteenth official pediatric heart ation report 2013; Focus theme: age. J Heart Lung Transplant 2013;32:979-88. 2. Schoenberg BS, Mellinger JF, Schoenberg DG. Cerebrovascular disease in infants and children: a study of incidence, clinical features, and survival. Neurology 1978;28:763-8. 3. Gardner TJ, Horneffer PJ, Manolio TA, et al. Stroke following coronary artery bypass grafting: a ten-year study. Ann Thorac Surg 1985;40:574-81. 4. Domi T, Edgell DS, McCrindle BW, et al. Frequency, predictors, and neurologic outcomes of vaso-occlusive strokes associated with cardiac surgery in children. Pediatrics 2008;122:1292-8. 5. Rodan L, McCrindle BW, Manlhiot C, et al. Stroke recurrence in children with congenital heart disease. Ann Neurol 2012;72:103-11. 6. Fallon P, Aparicio JM, Elliott MJ, et al. Incidence of neurological complications of surgery for congenital heart disease. Arch Dis Child 1995;72:418-22. 7. Mackay MT, Wiznitzer M, Benedict SL, et al. Arterial ischemic stroke risk factors: the International Pediatric Stroke Study. Ann Neurol 2011; 69:130-40. 8. Ganesan V, Prengler M, Wade A, et al. Clinical and radiological recurrence after childhood arterial ischemic stroke. Circulation 2006; 114:2170-7. 9. Lanthier S, Carmant L, David M, et al. Stroke in children: the coexistence of multiple risk factors predicts poor outcome. Neurology 2000;54:371-8. 10. Blackstone EH, Naftel DC, Turner ME. The decomposition of timevarying hazard into phases, each incorporating a separate stream of concomitant information. J Am Stat Assoc 1986;81:615-24. 11. Hoffman JL, Mack GK, Minich LL, et al. Failure to impact prevalence of arterial ischemic stroke in pediatric cardiac patients over three decades. Congenit Heart Dis 2011;6:211-8. 12. Mallick AA, Ganesan V, Kirkham FJ, et al. Childhood arterial ischaemic stroke incidence, presenting features, and risk factors: a prospective population-based study. Lancet Neurol 2014;13:35-43. 13. Fullerton HJ, Wu YW, Zhao S, et al. Risk of stroke in children: ethnic and gender disparities. Neurology 2003;61:189-94. 14. Fullerton HJ, Wu YW, Sidney S, et al. Risk of recurrent childhood arterial ischemic stroke in a population-based cohort: the importance of cerebrovascular imaging. Pediatrics 2007;119:495-501. 15. Almond CS, Morales DL, Blackstone EH, et al. Berlin Heart EXCOR pediatric ventricular assist device for bridge to heart ation in US children. Circulation 2013;127:1702-11.

602 The Journal of Heart and Lung Transplantation, Vol 35, No 5, May 2016 16. Fraser CD, Jaquiss RDB, Rosenthal DN, et al. Prospective trial of a pediatric ventricular assist device. N Engl J Med 2012;367: 532-41. 17. Kavarana MN, Savage A, O Connell R, et al. Composite risk factors predict survival after ation for congenital heart disease. J Thorac Cardiovasc Surg 2013;146:888-93. 18. Feingold B, Zheng J, Law YM, et al. Risk factors for late renal dysfunction after pediatric heart ation: a multi-institutional study. Pediatr Transplant 2011;15:699-705. 19. Lippi G, Favaloro EJ, Montagnana M, et al. C-reactive protein and venous thromboembolism: causal or casual association? Clin Chem Lab Med 2010;48:1693-701. 20. Järvisalo MJ, Harmoinen A, Hakanen M, et al. Elevated serum C- reactive protein levels and early arterial changes in healthy children. Arterioscler Thromb Vasc Biol 2002;22:1323-8. 21. Roche SL, O Sullivan JJ, Kantor PF. Hypertension after pediatric cardiac ation: detection, etiology, implications and management. Pediatr Transplant 2010;14:159-68.