2068 - Investigating Prognostic Value of Diffusion Weighted MRI for Patients with High Grade Glioma Undergoing PULSAR Reirradiation
Presenter(s)
G. Gard1, J. Deng1, Y. Li1, Y. Zhang1, Y. Dai1, Y. P. Liao1, B. Zhang2, L. Pari Mitre1, L. Giuliani Schmitt1, M. Dohopolski1, T. Dan1, Z. Wardak1, X. Cai1, M. Youssef3, E. Maher3, T. Patel4, A. R. Patel4, M. Z. Sun4, R. D. Timmerman1, and J. B. De Vis1; 1Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, TX, 2Cincinnati Children's Hospital Medical Center, Division of Biostatistics and Epidemiology, Cincinnati, OH, 3Department of Neurology, University of Texas Southwestern Medical Center, Dallas, TX, 4Department of Neurological Surgery, University of Texas Southwestern Medical Center, Dallas, TX
Purpose/Objective(s):
Patients with recurrent WHO grade 4 gliomas have poor prognosis and few salvage options. Reirradiation can be pursued, but it is unclear which patients benefit most. In newly diagnosed glioblastoma (GBM) patients, diffusion weighted imaging (DWI) and Intravoxel Incoherent Motion (IVIM) parameter fitting predict overall survival. Using these with technologies such as personalized ultrafractionated stereotactic adaptive radiotherapy (PULSAR) may improve patient selection and outcome. In this study, the prognostic role of DWI and IVIM was evaluated. We hypothesized that increases in Apparent Diffusion Coefficient (ADC) and true diffusion (Dt) and decreases in pseudo-diffusion (Dp) and perfusion fraction (Fp) predict improved overall survival (OS).
Materials/Methods:
This was a retrospective study of 38 patients with recurrent WHO grade 4 gliomas treated with PULSAR. Radiation treatments were delivered on a 1.5T MR-LINAC (Elekta Unity), and the imaging protocol included DWI. An in-house developed modular DWI processing pipeline was used to generate mean, median, and interquartile range (IQR) of ADC, Dt, Dp and Fp within the planning target volume (PTV). For statistical analysis, longitudinal measurements of each of these parameters over the treatment course were analyzed with OS using joint modeling. A linear mixed-effects model was first fitted to characterize the longitudinal trajectories of the markers, and a Cox proportional hazards model was specified for OS. The analysis was performed using JM package in R to account for the association between the time-dependent DWI biomarkers and the risk of death.
Results:
Median age of the patients was 53 (IQR: 43 – 61) years, and 16/38 (42%) females were included. Eight patients were diagnosed with grade 4 astrocytoma and 30 with GBM. As shown in Table 1, of all parameters, Dt was most significantly associated with risk of death (p<0.05) with a hazard ratio (HR) of 0.16 for median Dt value.
Conclusion:
Our initial analysis suggests that in patients treated with PULSAR reirradiation, Dt shows promise as a potential imaging biomarker predicting overall survival. However, ongoing analyses are needed to further verify this relationship and identify response values that can dichotomize radiation response. Future work will evaluate the effect of targeted dose-escalation on biological markers and treatment outcomes.
| Table 1. | Overall Survival | ||
| Mean, HR (95% CI) | Median, HR (95% CI) | IQR, HR (95% CI) | |
| ADC | 0.24 (0.04, 1.42), p-value: 0.901 | 0.14 (0.02, 0.91), p-value: 0.039 | 0.45 (0.03, 7.17), p-value: 0.576 |
| Dp | 1.01 (0.97, 1.03), p-value: 0.449 | 1.01 (0.99, 1.03), p-value: 0.420 | 1.03 (0.91, 1.17), p-value: 0.593 |
| Dt | 0.16 (0.03, 0.78), p-value: 0.023 | 0.16 (0.03, 0.74), p-value: 0.020 | 0.002 (0.001, 0.38), p-value: 0.021 |
| Fp | 11.4 (0.25, 506), p-value: 0.210 | 6.48 (0.18, 238), p-value: 0.310 | 29.1 (<0.01, >100), p-value: 0.711 |