2561 - Stereotactic Radiosurgery (SRS) over Whole Brain Radiotherapy (WBRT) Preserves Brain Volume: An Imaging and Radiomic Analysis
Presenter(s)
J. Bae1, L. Torre-Healy2, J. Wu2, L. Pang2, A. Stessin2, A. T. Hsia3, R. Farrell4, K. M. Mani4, P. Prasanna2, and S. Ryu4; 1Department of Biomedical Informatics, Stony Brook University Hospital, Stony Brook, NY, 2Stony Broom Univeersity, Stony Brook, NY, 3Radiation Oncology, Stony Brook University Hospital, Stony Brook, NY, 4Department of Radiation Oncology, Stony Brook University Hospital, Stony Brook, NY
Purpose/Objective(s): WBRT is associated with neuro-cognitive adverse effects, potentially mediated by parenchymal brain volume loss (BVL) after radiation. In this study, we hypothesize that WBRT results in greater whole brain (WB) volume loss than SRS and that radiomic features from pre-treatment MRI may correlate with BVL following treatment.
Materials/Methods: In this retrospective analysis, we identified a cohort of patients who have been treated with WBRT or SRS at our institution with pre-treatment (preTx) baseline MRI and at least one post-treatment (postTx) MRI within 12 months postTx. An auto-contouring software was used to segment WB regions above foramen magnum from T1 MRI sequences. Contour quality was assessed manually and corrected where necessary. WB volumes were calculated and used to determine percent volume changes following treatment. Welch’s t-test was used to assess statistically significant differences in BVL trends over one year of follow up between SRS and WBRT cohorts. Average percent change (APC) in brain volume was assessed between preTx MRI and postTx MRI. APC between WBRT and SRS cohorts were compared using Wilcoxon test. 3D Radiomic features were extracted on preTx T1 MRIs and correlated with BVL.
Results: 143 patients were identified to have pre and postTx MRIs for analysis including 58 WBRT patients (median dose 30 Gy, range: 2-42 Gy) and 85 SRS patients (median dose 18 Gy, range 14-20 Gy). Median time to follow up was 163 days (range 19-544). Overall rates of BVL differed significantly at 1 year postTx between WBRT (-3.14%) and SRS cohorts (-0.77%, p = .002). Patients treated with WBRT experienced an APC of -3.93%, while patients treated with SRS experienced significantly reduced APC of –2.11% (p=.038). There were two patterns of BVL including “rapid” vs “slow” BVL relative to average APC for WBRT-treated patients. Two radiomic features were identified with statistically significant differences in expression between WBRT-treated patients with rapid and slow APC (p<0.05). One of these features was also significantly differentially expressed between SRS-treated patients and rapid APC WBRT-treated patients, correlating with susceptibility to BVL after treatment with either RT modality. The other feature was not significantly different between SRS and WBRT-treated patients, suggesting that it might only correlate with risk of BVL for WBRT and not SRS.
Conclusion: WBRT induces significantly increased BVL when compared with SRS. Further, some patients may experience increased “rapid” rate of BVL which is significantly associated with specific radiomic features on preTx imaging. Future work might examine whether patients with risk for rapid BVL following WBRT can be identified via pre-treatment radiomic imaging biomarkers and perhaps be diverted to alternative treatment options.