Patient Factors Predicting Intracranial Progression: Optimizing MRI Surveillance Interval Following Brain Metastases SRS

Abstract

Stereotactic radiosurgery (SRS) is standard treatment for limited brain metastases (BM) patients (pts). Serial MRIs are crucial to identify asymptomatic progression (AP) salvageable with further SRS. Optimal MRI intervals are poorly understood. We conducted a multicenter study to evaluate patient and disease factors associated with AP and symptomatic progression (SP) timing post-SRS. Consecutive BM pts treated with SRS and ≥ 1 follow-up imaging study from 1/1/02-6/30/17 were identified in 3 institutional databases. Pt charts and intracranial (IC) imaging were reviewed for the presence/absence of progression and symptoms. Overall survival (OS) rates were estimated using the Kaplan-Meier method. Cumulative incidence (CuI) functions were used to estimate AP and SP risks overall and differences as grouped by histology. Fine-Gray regression models were applied to estimate the subdistribution hazards of IC progression associated with pt demographics and disease factors with a significance threshold of p<0.05. We identified 485 evaluable BM pts treated with SRS. Lung (46.4%), breast (20.2%), and melanoma (18.6%) were the 3 most common primaries. 219 (45.2%) pts had prior WBRT and 109 (22.5%) had prior surgery for BMs. Median number of treated BM was 1 (range: 1-7). At the time of SRS, 320 (66.0%) pts had extracranial (EC) disease (61% active, 29% controlled, and 10% unknown status). Median OS was 11.7 months (95% CI 10.8-13.1). 309 (64%) pts experienced IC progression [117 (37.9%) SP and 192 (62.1%) AP]. CuI and 95% CI of SP was 11% (8-14) at 4 months and 26% (22-31) at 1 year, compared to 15% (12-19) and 37% (33-43) for AP, respectively. Median time to any progression and AP after SRS was 8.1 (7.2-9.3) and 24 months (18.4-40.9), respectively. Pts with adenocarcinoma (AC) were least likely to have SP (p=0.011), with 8% (5 – 13) developing SP at 6 months compared to 25% (17 – 37) with melanoma and 17% (11 – 25) with other histologies. Histology was not associated with AP (p=0.25). In regression analysis, increased SP risk was associated with melanoma vs AC [HR: 2.06 (1.27 – 3.34)], other histologies vs AC [1.59 (1.05 – 2.41)], and prior WBRT [1.46 (1.02 – 2.10)]. Shorter time to initial metastases (p=0.04), no prior WBRT [HR: 1.45(1.07 – 1.95)], and unknown EC disease status [2.06 (1.30 – 3.27) compared to uncontrolled status] were associated with increased AP hazard. AP was more common than SP and different variables were associated with progression types. Histology was associated with SP, but not AP. WBRT was associated with increased SP risk and decreased AP risk. Time to initial metastases and EC disease were associated with AP, but not SP. The SP risk plateaued sooner than AP risk (near 20 vs near 40 months), highlighting the need for more individualized MRI intervals. With expanded data, we will model altered imaging frequency post-SRS.