Vasculopathy in Pediatric Craniopharyngioma Patients Treated with Surgery and Proton Radiotherapy

As much as 40% of pediatric brain tumor patients will experience varied levels of Vasculopathy (VS), however, few predictive factors have been described. Here we describe the type and timing of VS and explore the relationship between treatment modality and the timing, location, and distribution of VS. Pediatric patients with Craniopharyngioma (N = 94) were enrolled and treated with proton radiotherapy (PRT) and surgery at a single institution. Pre- and post-treatment imaging, cumulative physical and biological PRT dose maps, clinical details, and measures of dyslipidemia were evaluated. MR and MRAs were evaluated for pre- and post-PRT VS, VS type [Cavernoma, Aneurysm, Stenosis (VSN), Cerebral Microbleed (CMB), Infarction, Dilated Perivascular Space (DPVS)], VS workup [Digital Subtraction Angiography (DSA), Arterial Spin-labelled (ASL) Perfusion)], VS location, and VS severity. VS events were segmented and related to their corresponding normal brain region, and vascular territory. Forty-seven (50%) patients were found to have 154 instances of confirmed VS of varying severity with a median time to event of 3.41 years 95%, CI 3.08-3.88. Twenty-two percent (N = 21) of patients had ≥1 pre-existing radiographic instances of VS preceding PRT (N = 11 VSN, N = 7 Aneurysm, N = 3 Infarction, N = 2 Cavernoma, N = 1 DPVS) and 26.6% (N = 25) had some form of dyslipidemia following surgical evaluation. VS post-RT was present in 46 (48.9%) patients with the most common being VSN (N = 24), CMB (N = 4), DPVS (N = 3), and Cavernoma (N = 3). Post-RT VS were clinically significant in 9.5% (N = 9), severe in 4.2% (N = 4), and compensated in 2.1% (N = 2). Aspirin was recommended for potential flow-limiting stenosis in 10.6% (N = 10) patients. ASL and DSA were performed in 13 (13.8%) and 11 (11.7%) cases respectively. Revascularization procedures were required in 4 (4.2%) patients. The most common brain regions affected were the orbitofrontal cortex, temporal lobe, brainstem, and basal forebrain. The most commonly affected vascular territories were the Internal Carotid Artery (a.) (N = 14), Ant. Cerebral a. (N = 13), Middle Cerebral a. (N = 9), and Anterior Choroidal a. (N = 11) territories. The median linear energy transfer and physical dose for VSN/DPVS events were 2.66 keV/μm (2.41-2.92) and 47.9 CGE (34.5-47.9) respectively. While clinical details such as age, weight, and comorbidities were not predictive of VS, an increased frequency of VSN, DPVS, infarction events along the spatial distribution of the operative corridor and high-PRT dose were observed. VS often precedes PRT necessitating appropriate baseline imaging. Surgery type and extent are interrelated to the risk for PRT-induced VS. While the spatial PRT dose distribution approximated most vascular injury events, it was not all-inclusive. Spatial modeling of biological and physical PRT may offer insights into therapy-related VS.