A Survival Score for Patients Treated with Whole-Brain Radiotherapy plus Simultaneous Integrated Boost for Brain Metastases

Abstract

Different types of radiotherapy (RT) are used for brain metastases including whole-brain RT (WBRT) and local therapies (radiosurgery, fractionated stereotactic RT). WBRT plus simultaneous integrated boost (WBRT+SIB) is another treatment option that was reported to improve intracranial control and overall survival (OS) when compared to WBRT alone. This study aimed to create a score to identify patients who may not benefit from this approach due to poor OS. In 128 patients treated with WBRT+SIB for brain metastases (2014-2021), year of RT, interval from tumor diagnosis to RT, WBRT+SIB regimen [(14 x (2.5 + 0.5) Gy vs. 18 x (2.0 + 0.5) Gy], pre-RT systemic treatment, age, gender, performance score, primary tumor type, N lesions, and extracranial metastases were retrospectively evaluated for OS. Univariate analyses (UVA) were performed with Kaplan-Meier method and log-rank test. Factors that showed significant (p<0.05), a strong trend (p<0.07) or a trend (p<0.14) in UVA were included in a Cox proportional hazards model (MVA). To create the appropriate OS score, three models were created, one including factors significant on MVA, a second including factors with p<0.07 on UVA, and a third one including factors with p<0.14 on UVA. For each factor, 6-month OS rates (%) were divided by 10 and resulting points were added for each patient. Based on patient scores, three prognostic groups were designed for each model. Positive predictive values (PPV) of correct prediction of death ≤6 and OS ≥6 months were calculated. On UVA, KPS >80 (p<0.001) and ≤3 lesions (p = 0.006) were associated with OS, age ≤64 years (p = 0.061) showed a strong trend, and no extracranial metastases (p = 0.128) a trend. On MVA, KPS (HR: 0.40, 95% CI: 0.27-0.61, p<0.001) and N lesions (HR: 1.81, 95% CI: 1.23-2.65, p = 0.002) were significant; age (HR: 1.12, 05% CI: 0-76-1.66, p = 0.57) and extracranial metastases (HR: 1.19, 95% CI: 0.79-1.78, p = 0.40) were not significant. In Model 1 (including only KPS and N lesions), groups were 5, 6-8 and 9 points with 6-month OS rates of 15%, 38% and 57%, respectively. PPVs to predict death and OS were 85% and 57%, respectively. In Model 2 (KPS, N lesions, age), groups were 8, 9-12 and 13 points with 6-month OS rates of 17%, 33% and 75%, respectively. PPVs were 83% and 75%, respectively. In Model 3 (KPS, N lesions, age, extracranial metastases), groups were 11, 12-16 and 17 points with 6-month OS rates of 14%, 34% and 78%, respectively. PPVs were 86% and 78%, respectively. All three models achieved high accuracy regarding prediction of death ≤6 months and can help identify patients who may not be ideal candidates for WBRT+SIB. For prediction of OS ≥6 months, Models 2 and 3 were superior. Since differences between Models 2 and 3 were marginal and correct identification of extracranial metastases (Model 3) may require extensive staging, Model 2 appears preferable.