Radiation-induced accelerated aging of the brain vasculature in young adult survivors of childhood brain tumors

Abstract Background Cranial radiotherapy may damage the cerebral vasculature. The aim of this study was to understand the prevalence and risk factors of cerebrovascular disease (CVD) and white matter hyperintensities (WMHs) in childhood brain tumors (CBT) survivors treated with radiotherapy. Methods Seventy CBT survivors who received radiotherapy were enrolled in a cross-sectional study at a median 20 years after radiotherapy cessation. The prevalence of and risk factors for CVD were investigated using MRI, MRA, and laboratory testing. Tumors, their treatment, and stroke-related data were retrieved from patients’ files. Results Forty-four individuals (63%) had CVD at a median age of 27 years (range, 16-43 years). The prevalence rates at 20 years for CVD, small-vessel disease, and large-vessel disease were 52%, 38%, and 16%, respectively. Ischemic infarcts were diagnosed in 6 survivors, and cerebral hemorrhage in 2. Lacunar infarcts were present in 7, periventricular or deep WMHs in 34 (49%), and mineralizing microangiopathy in 21 (30%) survivors. Multiple pathologies were detected in 44% of the participants, and most lesions were located in a high-dose radiation area. Higher blood pressure was associated with CVD and a presence of WMHs. Higher cholesterol levels increased the risk of ischemic infarcts and WMHs, and lower levels of high-density lipoprotein and higher waist circumference increased the risk of lacunar infarcts. Conclusions Treating CBTs with radiotherapy increases the risk of early CVD and WMHs in young adult survivors. These results suggest an urgent need for investigating CVD prevention in CBT patients.


Study Population
Consecutive CBT survivors who were diagnosed between 1970 and 2008 and treated with radiotherapy (n = 127) were identified from the registers at 5 university hospitals in Oulu, Kuopio, Turku, Tampere, and Helsinki, where all CBTs were treated in Finland. An invitation letter was sent to all included CBT survivors to participate in a study of late complications. The inclusion criteria were patients who (i) were diagnosed at younger than age 16 years, (ii) received radiotherapy as part of their treatment, (iii) were age 16 years or older at the time of the study, (iv) had undergone 5 or more years of follow-up since therapy cessation, and (iv) had no known progressive brain tumors.
All participants were treated with conventional radiotherapy (n = 70). Patients' records were reviewed regarding the treatment of the primary tumor and occurence of stroke. This is a cross-sectional study in which participants were examined clinically using brain MRI and MRA, and their blood samples were analyzed for the presence of atherosclerotic risk factors during a visit to one of the study centers. Family history of stroke or myocardial infarction at age younger than 50 years was gathered by a questionnaire. Baseline characteristics of participants and nonparticipants are shown in Table 1.
Viewing applications for diagnostic radiology, which comprised picture archiving and communication systems or digital imaging and communications in medicine, were used to evaluate the MRI scans, namely, neaView (Neagen) in Oulu, Sectra Workstation IDS7 version 19.1. 10.3584 (Sectra AB) in Kuopio, and Agfa Impax version 6.6.1.5551 2017 (Agfa Healthcare N.V.) in Helsinki. Radiologists from 3 hospitals evaluated the MRI scans, all of which were reevaluated by M.S.-P.

Cerebrovascular Disease Classification Criteria
CVD was defined as a history of cerebral hemorrhage or TIA, which was obtained from reviews of patients' files or any vascular lesion, excluding WMHs, detected following brain MRI.
Large-vessel disease was diagnosed if the participants had signs of previous ischemic infarct on MRI, or experienced a TIA or presence of large-vessel vasculopathy. Large-vessel vasculopathy was defined as any pathology Neuro-Oncology Practice of the cerebral vessels on MRA. Ischemic infarct was defined according to American Heart and Stroke Associations criteria of brain cell death attributable to ischemia, based on imaging evidence of cerebral focal ischemic injury in a defined vascular distribution. 24 The definition of TIA according to the American Academy of Neurology was used in the present study (transient episode of neurological dysfunction caused by focal brain and spinal cord or retinal ischemia without acute infarction). 25 From the clinical files, the etiology and the date of TIAs and ischemic infarcts were determined.
Small-vessel disease was defined as the presence of lacunar infarcts or focal hemosiderin deposits (FHDs). 22 Given the direct effects of radiotherapy and chemotherapy on the brain, WMHs were not considered signs of smallvessel disease in our cohort. 22,23,26 Perivascular spaces were excluded in the analysis because of the controversy as to whether they should be considered lesions and the poor understanding of the mechanism underlying enlarged perivascular spaces. 22 Lacunar infarcts were defined according to international consensus as round or ovoid, subcortical, 3-mm to 15-mm diameter fluid-filled cavities that were consistent with previous acute small subcortical infarcts or hemorrhages in the territory of one perforating arteriole. 22 International criteria were used to diagnose cerebral microbleeds that were defined as small areas, generally 2 to 5 mm in diameter, but sometimes up to 10 mm in diameter, of signal voids with associated blooming artifacts seen on gradientecho MRI sequences that were sensitive to susceptibility effects. 22 In this study, FHDs were noted within low B-value DWI sequences in the absence of susceptibility-weighted imaging or T2* sequences. We used the international definition of WMH, that is, signal abnormalities of variable sizes in the white matter that show hyperintensity on T2-weighted images without cavitations. 22 WMHs were classified using the Fazekas scale based on their distributions and sizes of the periventricular and deep WMHs ( Table 2; Fig. 1). 27 Mineralizing microangiopathy was diagnosed as signs of calcifications on MRI. The definition of the American Stroke Association was used to diagnose cerebral hemorrhage as a nontraumatic focal collection of blood within the brain parenchyma or ventricular system. 24 Although subdural hematomas may occur spontaneously, they are not included in the definition of stroke. 24 We used Common Terminology Criteria for Adverse Events (CTCAE) classification version 5.0 for reporting the severity of strokes, TIAs, and traumatic and subdural hemorrhages. 28 In this classification, grade 1 is used to describe mild severity of an adverse event, 2 for moderate severity, 3 for severe or medically significant but not immediately life-threatening, 4 for life-threatening consequences, and 5 for death. 28

Laboratory Analyses
Laboratory samples were collected after overnight fasting, and all analyses were performed in the Nordlab at Oulu University Hospital. Plasma glucose, cholesterol, high-density lipoprotein (HDL), low-density lipoprotein, triglyceride, and glycosylated hemoglobin A 1c levels were measured and analyzed using a clinical chemistry system (Advia 1800; Siemens Healthcare GmbH). Serum insulin levels were analyzed using a chemiluminescent immunoassay (Advia Centaur XP; Siemens Healthcare GmbH). One participant's blood samples could not be collected and another individual had not fasted, so these samples were excluded from the analyses. Homeostatic model assessment of insulin resistance index was calculated using fasting glucose and insulin levels for quantifying insulin resistance. 29

Radiation Dose Distribution Analyses
The majority of the patients were treated in the 1980s and 1990s using 2-dimensional treatment planning techniques. The medical physicists analyzed dose distributions using patients' charts, treatment plans, and radiation field images to determine the radiation doses. Nine patients' radiation field images and treatment plans were not available; among them, 4 patients were treated with local radiotherapy and 5 with whole-brain radiotherapy.

Statistical Analyses
Differences between the median values of 2 independent groups were tested using the Mann-Whitney U test. Differences between the median values of 4 independent groups were tested using the Kruskal-Wallis test. The chi-square exact test was used to compare the distribution of categorical variables relative to categorical variables in other groups. For CVDs, only the time before the event occurred was known and interval-censored survival analysis with the EMICM algorithm was used to calculate the cumulative prevalence. Logistic regression analysis was used to calculate the odds ratios (ORs) for the atherosclerotic risk factors and MRI markers; the results are presented as ORs and their 95% CIs. Statistical analyses were performed using IBM SPSS software, version 25 for Windows (IBM Corp) and SAS 9.4 (SAS institute Inc). Graphs were produced using OriginPro 2018 software (OriginLab).

Neuro-Oncology Practice
Ethics Written informed consent was obtained from all enrolled participants or their legal guardians. The study was approved by the institutional review boards at Oulu, Kuopio, Turku, Tampere, and Helsinki University Hospitals, Finland. The research was conducted according to the principles of the Declaration of Helsinki.   13 were lost to follow-up. Two participants did not undergo craniospinal MRI because of vagus nerve stimulator (n = 1) and claustrophobia (n = 1). Two survivors treated with stereotactic radiotherapy were excluded from the analysis. The median age of the participants at the time of imaging was 27.1 years (range, 16.2-43.8 years), and their median age at tumor diagnosis was 8.3 years (range, 1.1-15.7 years). The median interval from the end of the radiotherapy course to imaging was 20.7 years (range, 5.0-33.1 years). Radiotherapy was given between 1980 and 2007. One patient was reradiated year after the first treatment, so that the total dose of radiation was 51.0 Gy. Type 1 neurofibromatosis was diagnosed in one participant. Radiotherapy for glial cell tumors has been reduced over the years; after 2000, only one glial tumor was treated with radiotherapy in this cohort. The median follow-up time was shorter for survivors treated with whole-brain radiotherapy than for those treated with local radiotherapy (16.

Cerebrovascular Disease
CVD was diagnosed in 63% of the participants ( Table 2). The cumulative prevalence of CVD at 20 years of follow-up was 52% (95% CI, 39%-66%, Fig. 2). A total of 44% of the participants had coexisting imaging findings in the brain ( Table 2).

Large-Vessel Disease
Large-vessel disease, which was diagnosed based on the presence of ischemic infarct, TIAs, or large-vessel vasculopathy, was present in 19% of the participants ( Table 2). The cumulative prevalence of large-vessel disease at 20 years of follow-up was 16% (95% CI, 9%-28%, Fig. 2). Overall, 8 individuals had either TIAs (n = 2) or lateoccurring ischemic infarcts (n = 6, Table 2). One perioperative ischemic infarct was excluded from all analyses. In the brain MRI, ischemic infarcts were in the occipital (n = 3), frontal (n = 1), and temporal lobes (n = 1) as well as the temporoparietal area (n = 1), basal ganglia (n = 1), and thalamic area (n = 1). Two ischemic infarcts were detected in 2 individuals; recurrence was seen in 1 participant. In the individual with recurrent stroke, the first ischemic infarct was found in the left occipital lobe, and the recurrence was in the left temporal lobe. No vascular pathology was found on the MRA. Ischemic infarcts were classified as CTCAE grade 1 (n = 3), grade 2 (n = 4), and grade 3 (n = 1). CTCAE classifications for the 2 TIAs were grade 1 and 2. The participant with recurrent ischemic infarcts reported using clopidogrel as an antiplatelet medication at the time of the study; the other individuals with ischemic infarcts or TIAs did not report using antiplatelet therapy. The cumulative prevalence of ischemic infarcts at 20 years of follow-up was 6% (95% CI, 3%-15%, Fig. 2).
MRA detected large-vessel vasculopathy in 6 participants, including stenotic caliber changes in the right middle cerebral artery (MCA, n = 1), vasculopathy in the right MCA with ischemic infarct in the right basal ganglia on MRI (n = 1), and stenoses in the left posterior cerebral artery (n = 1) and the left vertebral artery with a lacunar infarct in the right basal ganglia on MRI (n = 1, Fig. 1, 1A). In one individual, neither the right MCA nor the anterior cerebral arteries could be visualized. One patient had undergone surgery for a left frontal arteriovenous malformation after radiotherapy.

Cerebral Hemorrhage
Cerebral hemorrhage was confirmed in 2 (3%) individuals; both were intracerebral hemorrhages ( Table 2). All hemorrhages occurred more than 1 year after radiotherapy. In addition, 3 traumatic and 2 subdural hemorrhages had been diagnosed. Traumatic cerebral hemorrhages included subdural hemorrhages (n = 2) and contusion hematomas (n = 1). Only one participant with cerebral hemorrhage did not have any other signs of CVD. An ischemic infarct and a subdural hemorrhage coexisted in one individual. CTCAE classification for 2 cerebral hemorrhages were grade 1, as well as for all traumatic and subdural hemorrhages.

Telangiectasia
One participant had a left thalamic slightly T1-and T2-hyperintense lesion, which revealed a brush-like enhancement with gadolinium contrast media that was indicative of telangiectasia.

White Matter Hyperintensities
WMHs quantified by the Fazekas scale were observed in 49% of the participants (Table 2). Periventricular hyperintensities and deep WMHs were detected in 28% and 36% of the individuals, respectively ( Table 2). Fig. 1 panel 2, shows representative MRI scans of WMHs, and Table 2 presents the distribution of the WMHs according to the Fazekas scale. WMHs were seen in survivors followed for more than 10 years. The cumulative prevalence at 20 years of follow-up was 42% (95% CI, 29%-57%, Fig. 2).

Tumor-and treatment-related risk factors
Tumor-and treatment-related risk factors are shown in Supplemental Tables 1 and 2. Total dose of radiation was not associated with CVD, large-vessel disease, smallvessel disease, or strokes. Age at the follow-up visit was associated with large-vessel disease and WMHs. Ventriculoperitoneal shunt was associated with smallvessel disease; more specifically with FHDs (OR 3.76; 95% CI, 1.20-11.79; P = .023). Increasing radiation dose increased the presence of mineralizing microangiopathy (OR, 1.28; 95% CI 1.06-1.54; P = .010). Most of the CVD-associated imaging findings (68%) were located in the regions that received radiation doses 40 Gy or greater. A total of 11% of the microbleeds were found outside the radiation fields (Table 3). Most tumor characteristics and treatment protocols were not associated with Fazekas grade, except deep WMH, which was associated with total radiotherapy dose (P = .035) and body mass index (P = .011, Supplemental table 3).

Atherosclerotic Risk Factors
Male patients were more likely to have small-vessel disease than female patients (OR, 8.38; 95% CI, 2.19-32.04; P = .002). The CVD risk rose 1.03-fold (P = .045); specifically, the large-vessel disease risk increased 1.05-fold (P = .021) for every mm Hg increase in systolic blood pressure. Increasing diastolic blood pressure elevated the large-vessel disease risk 1.07-fold (P = .027) and the lacunar infarct risk 1.08-fold (P = .031). Total cholesterol and low-density lipoprotein levels were associated with the occurrence of ischemic infarct. Lower HDL levels were associated with lacunar infarcts. Participants with WMHs had higher total cholesterol and systolic and diastolic blood pressure (Table 4). Waist circumference was associated with occurrence of lacunar infarcts. Markers of glucose metabolism, diabetes, smoking, or estrogen therapy were not associated with the prevalence of CVD (Supplemental  table 4). A total of 17 survivors had a family member diagnosed with stroke or myocardial infarction at age 50 years or younger. Family history had no association with CVD (Supplemental table 4).
In the logistic regression analysis of atherosclerotic risk factors, systolic blood pressure was associated with CVD and large-vessel disease. Diastolic blood pressure was associated in the multivariable analysis with lacunar infarcts. In logistic regression analysis of atherosclerotic risk factors, WMHs were associated with diastolic blood pressure. In multivariable analysis, lacunar infarcts were associated with HDL and diastolic blood pressure (Supplemental table 5).

Association Between MRI Findings and Stroke
No associations were found between WMHs, FHDs, or mineralizing microangiopathy and strokes (see Supplemental Table 2).

Discussion
Radiotherapy is essential for the treatment of many CBTs, but it causes significant damage to the brain vasculature and can cause early CVD. [1][2][3][4][5][6][7]13,[17][18][19]21,[30][31][32] In this study, most of the long-term CBT survivors who received radiotherapy developed CVD, and the cumulative prevalence of CVD at 20 years of follow-up (52%) was alarmingly high. The rates of ischemic infarct, microbleeds, and lacunar infarct were similar to or higher than those in the general population age 70 years or older, which suggests accelerated aging of the cerebral vasculature. [33][34][35] Atherosclerotic risk factors further aggravated the effects of radiotherapy in CBT survivors. CVD and WMLs are known late effects of irradiation, and their relationships with cognitive impairment are recognized. [2][3][4]6,7,17,20,36 Generating knowledge about the prevalence of and risk factors for CVD in CBT survivors is crucial for disease prevention.
In Finland, the incidence of ischemic infarcts is 6.6 in the age group of 18 to 34 years and 25.8 in the age group of 35 to 44 years per 100 000 people. 37 In this study, the cumulative prevalence of ischemic infarcts at 20 years Neuro-Oncology Practice of follow-up was 6%. The cumulative incidence of stroke was 12% at 30 years of follow-up in the Childhood Cancer Survivor Study. 3 Despite the fact that most cerebral hemorrhages are not classified as strokes, the high prevalence of cerebral hemorrhage shows the fragility of the cerebral vasculature in survivors. 13,24 The high prevalence in this study might be related to the use of systematic MRI screening.
Moyamoya disease has been frequently reported after cranial radiotherapy, but this was not seen in our cohort. 19,32 The sensitivity of MRA is not as high as that associated with conventional angiography in relation to detecting Moyamoya disease, but MRA revealed largevessel vasculopathy in 6 cases. 19 The high prevalence of small-vessel disease might reflect the greater sensitivity of small vessels to the effects of radiation. The high prevalence of FHDs and the association between microbleeds and cognitive impairment in CBT survivors have been recognized. 6,13,17 The use of sensitive MRI techniques, such as susceptibility-weighted imaging or T2* sequences, explains the even higher rates of small-vessel disease and microbleeds in some studies. 13,31,38 Although the number of microbleeds is likely to increase during follow-up both in adult and pediatric brain tumor survivors, some microbleeds may disappear during follow-up. 17,39 Ischemic infarct risk has been associated with higher total radiation doses, especially those applied to the area surrounding the circle of Willis. 2,3,7,30 In this study, almost all CVD imaging findings, except for some microbleeds, were located in the radiation field. This suggests that reducing the dose and the size of the radiation field should be beneficial for brain vasculature. Although it has been taught that proton beam radiotherapy could be less harmful than photon radiotherapy because the radiation does not scatter to surrounding tissues, recent studies have shown large-vessel vasculopathy and high incidence of microbleeds after proton beam radiotherapy. 18,40 However, we did not find a lower rate of CVD in those treated with local radiotherapy, which may be related to fewer glial cell tumors, and thus less local radiotherapy in the survivors with shorter follow-up time. However, even stereotactic radiotherapy reduces cerebral blood flow in the surrounding tissue, which suggests that irradiation may harm the brain vasculature beyond the radiation field. 41 Atherosclerotic risk factors may further increase the risk of radiation-induced CVD in CBT survivors. Hypertension is a well-known risk factor associated with stroke, and its association with lacunar infarcts has been established in the general population. 42,43 In this study, higher BP was associated with CVD, large-vessel disease, and lacunar infarcts. A previous study's findings showed that hypertension was associated with a 4-fold increase in the risk of stroke among CBT survivors. 3 Higher cholesterol levels were associated with ischemic infarcts, but the mean cholesterol levels were not particularly high. The beneficial effect of antihypertensive medication on stroke risk, even in the high-risk population, is well established. 44,45 Balancing cholesterol levels in survivors presenting both vascular insufficiency and early atherosclerosis is challenging. 2 Total cholesterol levels have been inversely associated with hemorrhagic stroke, but directly associated with ischemic strokes in the general population. [46][47][48] Future prevention studies for CVD should consider antihypertensive treatment in CBT survivors.
Although the use of antiplatelet therapy for secondary prevention of stroke is well established, it is not currently known how incidental vascular findings should be managed even in the general population. 8, 24 We found a high prevalence of lacunar infarcts, WMHs, and microbleeds, which are all associated with hemorrhagic strokes in the general population. 24 A high prevalence of cerebral hemorrhagic lesions in CBT survivors suggests vascular fragility. Because survivors of CBT treated with radiotherapy are predisposed both to ischemic and hemorrhagic lesions, treating CVD is particularly challenging in this patient population.
whole-brain radiotherapy, higher blood pressure among CBT survivors was associated with a presence of WMHs in this study. 51 Antihypertensive treatment and lowering blood pressure may even reduce the progression of WML in the general population. 52 In cancer survivors, the etiology behind periventricular and deep WMH may differ. 53,54 Radiation-induced and chemotherapy-induced changes in white matter are typically seen in the periventricular area and are thought to result from demyelination, gliosis, edema, and coagulation necrosis. 54 Deep WMH on MRI may mimic those seen in older patients and those with cerebrovascular risk factors. 53 In the present study, we could not find significant associations supporting this difference.

Neuro-Oncology Practice
techniques. However, the total radiation dose administered to the CNS was relatively homogeneous. This study's strengths are associated with the systematic use of MRI screening on a cohort of consecutive adult CBT survivors who received radiotherapy and the reasonably high participation rate.
In summary, we found an alarmingly high prevalence of CVD as a late complication of cranial irradiation among CBT survivors, and many of the CBT survivors experienced strokes during the follow-up period. Although cranial irradiation remains necessary for the treatment of CBTs, strategies to prevent and treat its late effects on cerebral vasculature are urgently required.