University of Birmingham Risk of cerebrovascular disease among 13457 five year survivors of childhood cancer: A population based cohort study

• Users may freely distribute the URL that is used to identify this publication. • Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of private study or non-commercial research. • User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?) • Users may not further distribute the material nor use it for the purposes of commercial gain.

with cranial irradiation, the risk of CVD continues to increase substantially beyond age 50 up to at least age 65. Such survivors should be: counselled regarding this risk; regularly monitored for hypertension, dyslipidaemia and diabetes; advised on lifestyle risk behaviours. Future research should include the recall for counselling and brain MRI to identify subgroups that could benefit from pharmacological or surgical intervention and establishment of a case-control study to comprehensively determine risk-factors for CVD.
cancer survivorship, cohort, epidemiology, late effects 1 | BACKGROUND Survival after childhood cancer has markedly improved over the last few decades with overall 5-year survival in the United Kingdom now exceeding 80%. 1 Although the number of long-term survivors continues to increase, many subgroups of survivors are at risk of developing adverse health conditions many years after treatment. 2,3 Circulatory conditions, including cerebrovascular disease (CVD), are the leading cause of death among ageing survivors. 4 Studies specifically investigating CVD in survivors have reported substantively increased risks among survivors previously treated with cranial radiotherapy, but only up to age 50. [5][6][7][8][9][10][11][12] Beyond age 50, the risk of developing CVD in the general population doubles every 10 years, 13 but it is uncertain how the risk among survivors of childhood cancer, particularly those treated with cranial radiotherapy, develops with increasing age. If the relative risk (RR) remains elevated into ages at which the risk of developing CVD in the general population starts to increase substantially, then a considerable number of survivors could be affected. To our knowledge, this is the first large-scale study to quantify the risks of CVD up to age 65 according to whether survivors were treated with cranial radiotherapy or not.
The principal aim of our study was to determine the long-term risks of hospitalisations due to CVD among 5-year survivors of childhood cancer-particularly among those treated with cranial radiotherapy-through electronic linkage of the British Childhood Cancer Survivor Study (BCCSS) with the national hospital episode statistics (HES) database.

| British Childhood Cancer Survivor Study
The BCCSS is a population-based cohort of 17 980 individuals diagnosed with a childhood cancer between 1940 and 1991 inclusive, before age 15 years, in Great Britain and who survived for at least 5 years from diagnosis. 14 The cohort was identified through the population-based National Registry of Childhood Tumours.

| Hospital episode statistics
HES are a centralised data warehouse maintained by National Health Service (NHS) Digital containing records of inpatient, outpatient and accident and emergency admissions to NHS hospitals within England. 15 Records in HES are classified into hospital episodes which relate to a period of care for a patient under a single consultant. Variables recorded in HES include: an episode start and end date, a primary diagnosis code and 19 additional subsidiary diagnosis codes which may relate to the primary diagnosis or to other coexisting conditions. 16

| Definition of CVD
The primary and 19 subsidiary diagnosis code fields for each inpatient HES record were used to identify CVD related hospitalisations What's New?
Many sub-groups of childhood-cancer survivors have an increased risk of developing adverse health conditions later in life. In our study, the authors found that patients who were treated with cranial irradiation in childhood face a substantially greater risk of cerebrovascular disease after age 50. By age 65, as many as 26% of these patients will have been hospitalised for a cerebrovascular event. These survivors should therefore be counselled regarding lifestyle behaviours to reduce this risk. They should also be regularly monitored for hypertension, dyslipidaemia and diabetes.
(ICD-10:I60-68). If an individual had multiple CVDs, then only the first occurrence was considered except for analyses relating to the mean cumulative count. Survivors with any hospitalisation recorded as "sequelae of CVD" (ICD-10:I69), but no prior CVD recorded in HES were excluded (N = 44); as such survivors most likely have had a prior CVD before the HES database was available for linkage.

| Radiotherapy ascertainment
Information on initial radiotherapy treatment in the form of yes/no has previously been abstracted from medical records for 74% of all survivors in the cohort available for analyses. We assumed that any survivor with a tumour site "brain" or "meninges" and who had been treated with radiotherapy had been exposed to cranial irradiation. Any leukaemia survivor treated with radiotherapy before 1991 was assumed to have received prophylactic cranial irradiation unless information in the medical record unequivocally stated that no radiotherapy was given.
Survivors who had received radiotherapy for a tumour other than a central nervous system (CNS) tumour or leukaemia that was within the neck or head region were considered as having received "head and neck" radiotherapy. For survivors of any other tumour site-regardless of whether they were treated with radiotherapy or not-we assumed that they had not been exposed to cranial irradiation or radiation to the head and neck. Separate analyses were conducted for CNS tumour, leukaemia and head and neck tumour survivors treated with radiotherapy as these were assumed a priori to be at high risk. Analyses for Hodgkin lymphoma survivors treated with radiotherapy were not conducted as it was not possible to accurately determine the site of the radiation. To investigate the simultaneous effect of potential risk factors of CVD, a multivariable Poisson regression model with the log of the expected number of CVD hospitalisations as the offset was used to estimate the RRs. 19 RRs can be interpreted as the ratio of the SHRs adjusted for potential confounders. A similar multivariable Poisson regression model but with the offset being the ln(person-years) and the link function: ln(μ j − d Ã j Þ , where μ j is the observed and d Ã j the expected number of CVDs for stratum j of a relevant factor, was fitted to estimate relative excess risks (RERs). 20 RERs can be interpreted as the ratio of the AERs adjusted for potential confounders. Negative binomial regression was used instead of Poisson regression when the model fit showed signs of overdispersion. A likelihood-ratio test was used to test for linear trend of a factor by comparing the deviance of a model including the factor variable of interest, which was coded using consecutive nonnegative integer values (eg, 0/1/2/3), to the deviance of a model without the factor variable of interest. The cumulative incidence of the first occurrence of being hospitalised for CVD was calculated taking into account the competing risk of death. 21 In addition, the mean cumulative count (MCC) of being hospitalised for a CVD including for any recurrent CVDs was calculated. 22 The MCC can be interpreted as the average number of CVD hospitalisations per survivor. Statistical significance was taken at the 5% level (two-sided test). All statistical analyses were conducted in Stata statistical software version 16 except for the MCC which was conducted in R. 1.6, 3.5). In contrast, the AER increased significantly with increasing attained age (P trend < .001) from 3 (per 10 000 person-years) among those aged <20 years to 45 among those aged ≥60 years. These trends with attained age were confirmed in multivariable analyses (Table S1).

| Head and neck tumour survivors
The 736 survivors who were previously diagnosed with a tumour in the head and neck region and treated with radiotherapy (Table S4) were at 2-fold the expected risk (95% CI: 1.2-4.1; 10 CVD events).

| Previous studies
Recently, the North-American Childhood Cancer Survivor Study (CCSS) 6 reported risks of self-reported CVD up to age 50 with a cumulative incidence of 19.9% for survivors at high risk. A French study demonstrated that 11.3% of survivors treated with high-dose cranial irradiation with doses exceeding 10 Gray to the major intracranial vessels developed CVD by age 45. 8 A Dutch study including 28 strokes found a cumulative incidence of 10.0% for developing stroke by age 45 among survivors treated with cranial irradiation only. 12 The cumulative incidence figures from these studies are consistent with the 13.4% and 16.0% cumulative incidence at age 45 and 50, respectively, for survivors of a CNS tumour treated with cranial irradiation we report here. However, the current study shows that the cumulative incidence of CVD continues to increase substantially beyond age 50 years up to 26.0% by age 65.
The 4-fold overall increased SHR of CVD among all survivors in our study was similar to that observed in the Scandinavian ALiCCS study 25 ; the only other large-scale population-based study investigating long-term risks of CVD hospitalisations. Consistent with our study, the ALiCCS study also demonstrated that the AER of developing CVD increases substantially with attained age, although our study did not report risk estimates by whether survivors had been treated with cranial irradiation.
In our study, CNS tumour survivors treated without cranial irradiation were still at excess risk of CVD, although the risks were much lower than among those treated with cranial irradiation. This is consistent with data reported from the CCSS, 5 although the risk when compared with siblings was 13-fold (95% CI: 4.8-34.5) within the CCSS; much higher than the 3-fold increased SHR we report here.
Notably, the risks among survivors treated without cranial irradiation were mainly high in the first few decades after 5-year survival, but did not appear to increase substantially with increasing attained age.
Several studies among survivors of childhood cancer demonstrated strong dose-response relationships between the cumulative radiation dose to the brain and risk of CVD. 5,[8][9][10][11][12] Intracranial vascular damage-including stenosis constricting blood flow and aneurysms-is common after high-dose cranial radiation involving the major cranial blood vessels, 26 although the exact mechanism by which cranial irradiation may increase the risk of CVD is unclear.
Children's Oncology Group survivorship guidelines 27 28 In the general population, silent CVD is 10 times more prevalent than symptomatic stroke 29 and it is conceivable that the prevalence of silent CVD would be even greater in the survivor population. In a study 30 among 132 paediatric patients treated with brain radiotherapy 41.6% showed microbleeds or cavernomas after a mean follow-up of only 11 years-which is a remarkably large percentage considering the young age of these patients. In studies among individuals from the general population the risk of developing a symptomatic, mainly ischaemic, stroke after a silent brain infarct was 1.5-to 3.3-fold. 31

| Implications for clinical practice
As reported in a recent review, 32 there is suggestive evidence that conditions which predispose to CVD such as hypertension, dyslipidaemia and diabetes are more common 33,34 and tend to develop at younger ages among childhood cancer survivors as compared with siblings or the general population. [35][36][37] Therefore, to avoid the potential for underdiagnosis, it would be prudent that survivors who received cranial irradiation are regularly monitored for such conditions. The review 32 also identified suggestive evidence of lifestyle factors being important in terms of reducing the risk of cardiovascular conditions among survivors, in particular exercise 38 and diet. 33 Smoking and alcohol consumption are risk factors for stroke in the general population. 39,40 Therefore, regular counselling in follow-up clinics in relation to exercise, diet, smoking and alcohol would be prudent from a precautionary perspective. However, most previous work relates to cardiac disease and there is a need for detailed aetiological studies concerning CVD. Health England to whom we offer our profound thanks. The views expressed in this publication are those of the authors and not necessarily those of the funders or collaborators. Finally, thanks to all the cancer survivors whose experience has been used to try and understand the full consequences of childhood cancer and its treatment.

CONFLICT OF INTEREST
The authors declared no potential conflicts of interest.

ETHICS STATEMENT
Our study was approved by the National Research Ethics Committee

DATA AVAILABILITY STATEMENT
Individual patient data are not publicly available due to potential identification of individuals. Access to other anonymised aggregated data may be granted under conditions agreed with the relevant legal and research ethics committees and with appropriate data sharing agreements and permissions from external data providers in place. All outputs are subject to the codes of practice for official statistics. The corresponding author should be contacted to for enquiries relating to potential data access.