Modeling the Risk of Radiation Induced Alopecia in Brain Tumor Patients Treated with Active Beam Proton Therapy

Abstract

Hair follicles are very sensitive to radiation, which may induce temporary or permanent alopecia ultimately impacting patient quality of life (QOL). Our aim was to develop a normal tissue complication probability (NTCP) model for late radiation-induced alopecia (RIA) in patients (pts) treated with scanning beam proton therapy (PT) for brain tumors (BrT). We evaluated 78 consecutive BrT pts undergoing PT by pencil beam scanning (median dose, 50.4 Gy; range: 36–72 Gy) in a retroprospective analysis assessing acute (≤90 days) and late (>90 days) RIA classified according to the CTCAE v.4 scoring system. Median patient age was 56 years (range: 30–84 years). Median follow-up was 4 months (range: 1–19 months). Dose-surface histograms (DSHs) of the body structure were considered as representative of scalp irradiation and were extracted using an in-house developed library for Matlab. To calculate absolute DSH, we extracted the relative complement in the body structure of its 3D erosion defined by a spherical structuring element of radius r= 4 mm (i.e. the mean depth of scalp hair follicle). The following DSH metrics were then extracted: the surface receiving ≥ X Gy (Sx) in step of 5 Gy, the maximum and the mean dose to the scalp. The Sx were normalized using the surface of normal brain tissue as typical scale. Patient and treatment-related characteristics (e.g., age, gender, chemotherapy (CHT), re-irradiation) were analyzed with DSHs metrics. Chi-square/Mann-Whitney tests were employed for univariate statistical analysis. NTCP modeling by multivariate logistic regression was performed and model performance evaluated by Spearman’s Rs coefficient and ROC area. Acute Grade 2 (G2) RIA was found in 38 of 78 (49%) patients at a median time of 36 days (range: 0-58 days) from the end of PT; late G2 RIA was found in 19 of 58 patients (33%) at a median time of 124 days (range: 91-213 days). Clinical variables significantly associated with late G2 RIA were CHT (p= 0.04) and acute G2 RIA (p<0.0001). All Sx values were significantly correlated with the outcome. At the multivariate analysis only S20 survived within the NTCP model (Odds Ratio= 1.163, 95%CI 1.071-1.262; constant= 2.352, Rs= 0.64). Model ROC-AUC was 0.89 (95%CI 0.81–0.97). Using body DSH as representative of scalp irradiation, we derived a logistic NTCP model for late G2 RIA in PBT with a good prediction performance. This preliminary result could help in designing a BPT treatment plan that might minimize the risk of this morbidity ultimately impacting BrT pts QOL.