Stereotactic Radiosurgery versus Whole-brain Radiation Therapy for Patients with 4-15 Brain Metastases: A Phase III Randomized Controlled Trial

Stereotactic radiosurgery (SRS) has replaced whole-brain radiation (WBRT) as the standard care for patients with 1-3 brain metastases (BMs). However, in patients with 4 or more BMs the role of upfront SRS remains controversial because of lack of randomized evidence. In this phase III randomized controlled trial (NCT01592968), adult patients with 4-15 untreated non-melanoma BMs were stratified by histology, number of lesions, baseline Hopkins Verbal Learning Test - Revised Total Recall (HVLT-R TR) score, extracranial disease, KPS, and age, and randomly assigned to SRS or WBRT. Memantine was encouraged in the WBRT arm after publication of RTOG 0614. Prior SRS to 1-3 BMs with at least 3 months (mos) interval was allowed. Test of neurocognitive function (NCF) were completed at baseline and longitudinally including tests of learning and memory (HVLT-R), verbal fluency (Controlled Oral Word Association (COWA), and processing speed and executive function (Trail Making Test Parts A [TMTA] and B [TMTB]). Primary endpoints were HVLT-R TR and local control (LC) at 4 mos. Secondary endpoints included additional NCF tests, overall survival (OS), distant brain failure, toxicity, and time to systemic therapy. The trial was terminated early due to slow accrual. Between 09/2012 and 09/2019, 72 patients were randomized to SRS (n = 36) or WBRT (n = 36) with an estimated median follow-up time of 6.6 mos (range 0.2- 69.8). Median number of BMs at enrollment was 8. Thirty-one patients were evaluable for HVLT-R TR at 4 mos. In the WBRT arm, 62% of patients received memantine. In the primary endpoint analysis, relative to baseline, the 4 mo HVLT-R TR standardized z-score increased by +0.21 (standard error [SE] 0.27) for SRS-treated patients and declined by -0.74 (SE 0.36) for WBRT-treated patients (p = 0.041). Based on the Clinical Trial Battery Composite (the mean z-score for the HVLT-R, COWA, TMT), at 4 mos the NCF of patients in the SRS arm improved on average +0.23 (SE 0.14) and the NCF of patients in the WBRT arm declined on average -0.73 (SE 0.35), p = 0.008. Median OS time was 10.4 mos for SRS and 8.4 mos for WBRT (p = 0.45). Preliminary analyses of LC at 4 mos demonstrated 100% LC rate for SRS and 95.5% for WBRT (p = 0.53), and median time to distant brain failure (DBF) was 4.3 mos for SRS vs.18.1 mos for WBRT (p = 0.09). LC and DBF results are being independently confirmed by the trial radiologist presently. In non-melanoma patients with 4-15 brain metastases, SRS was associated with reduced risk of neurocognitive deterioration relative to WBRT without compromising OS. These results provide level 1 evidence to support the use of SRS in patients with 4-15 brain metastases amenable to SRS.