2015 - Sex-Specific Differences and Predictors of Severe Radiation-Induced Lymphopenia in Glioblastoma
Presenter(s)
M. Berger1, C. Friedes1, L. C. Linkowski1, C. Hollawell1, I. Messing1, M. D. Riina1, J. Sunnerberg1, G. Kurtz1, J. F. Dorsey1, M. Nasrallah2, A. Nabavizadeh3, S. Mohan3, C. Davatzikos3, M. Alonso-Basanta1, N. Amankulor4, C. Jackson4, D. O'Rourke4, S. J. Bagley5, H. G. Hubbeling1, and E. S. Lebow1; 1Department of Radiation Oncology, University of Pennsylvania, Philadelphia, PA, 2Department of Pathology, University of Pennsylvania, Philadelphia, PA, 3Department of Radiology, University of Pennsylvania, Philadelphia, PA, 4Department of Neurosurgery, University of Pennsylvania, Philadelphia, PA, 5Department of Medicine, Division of Hematology/Oncology, University of Pennsylvania, Philadelphia, PA
Purpose/Objective(s): Severe radiation-induced lymphopenia (sRIL) is independently associated with worse outcomes in glioblastoma (GBM). Female sex is the strongest clinical risk factor for development of sRIL; yet, the mechanism remains unclear, and mitigation strategies in this high-risk population have not been well established. We evaluated clinical and treatment factors associated with sRIL and whether the effect of radiotherapy (RT) modality on absolute lymphocyte count (ALC) varies by sex.
Materials/Methods: We retrospectively identified 832 patients > 18 years old with biopsy-confirmed WHO 2021 grade 4 GBM, defined by isocitrate dehydrogenase wildtype status and associated molecular features, treated with RT at a single institution (2014-2024). Complete blood counts within 30 days before RT and 120 days after RT were collected. sRIL was defined as CTCAE v5.0 grade 3+ (<0.5 × 109 cells/L). Target and whole brain dosimetry were abstracted. Missing laboratory data was imputed using multiple imputation by chained equations and predictive mean matching. Multivariable (MV) logistic regression model was used to assess factors associated with sRIL, with prespecified sex-stratified MV and interaction analyses.
Results: Of the 832 patients, the majority were male (61%), MGMT unmethylated (61%), and received conventionally fractionated (73%) photon RT (57%) with concurrent temozolomide (94%). 48% received steroids during RT. Patients with sRIL had higher V20 Gy, lower pre-RT ALC, and more steroid use (all p <0.001). On MV modeling, female sex (OR 2.42, 95%CI 1.72-3.41, p <0.001) increased sRIL risk, whereas higher pre-RT ALC (OR 0.66, 95%CI 0.53-0.82, p <0.001) and no steroid use (OR 0.68, 95%CI 0.49-0.95, p<0.023) were protective against sRIL. RT modality, brain V20Gy, and target volume were not associated with sRIL on MV modeling. In sex-stratified analyses, baseline ALC remained strongly associated with sRIL in both sexes, and RT modality was not significant. On predicted risk curves, risk of sRIL among females was similar between modalities at low pre-RT ALC but diverged at higher ALC, with lower predicted sRIL risk for proton RT. These patterns were observed in visualization-based models, and formal interaction testing did not demonstrate statistically significant effect modification by sex (p=0.063).
Conclusion: Female sex, lower baseline ALC, and steroid use during RT were independently associated with higher risk of sRIL. Although irradiated brain volume is a consistently identified modifiable risk factor for sRIL, and proton RT reduces volume of irradiated brain, this did not translate into lower sRIL risk. On sex-stratified analysis, patterns observed in visualization-based models suggest females without baseline lymphopenia may benefit from proton RT. Further prospective studies are warranted to investigate whether proton RT can mitigate sRIL in these select patient populations.