Background
Phase III, randomized controlled trial (MD Anderson). 58 patients with 1–3 newly diagnosed brain metastases randomized to SRS alone vs SRS + WBRT. Enrolled Jan 2001–Sep 2007. Trial stopped early by DSMB based on Bayesian early stopping rules (96% probability that SRS+WBRT causes worse neurocognition at 4 months). WBRT 30 Gy in 12 fractions.
Interventions and follow up
Arm A: SRS alone (SRS dose per protocol based on lesion size)
Arm B: SRS + WBRT (30 Gy/12 fractions)
Primary endpoint: HVLT-R (Hopkins Verbal Learning Test-Revised) total recall decline ≥5 points at 4 month
mFollow up: 4 months (primary)
Arm B: SRS + WBRT (30 Gy/12 fractions)
Primary endpoint: HVLT-R (Hopkins Verbal Learning Test-Revised) total recall decline ≥5 points at 4 month
mFollow up: 4 months (primary)
Results
HVLT-R cognitive decline at 4 months: 24% (SRS alone) vs 52% (SRS+WBRT), posterior probability SRS+WBRT worse = 96%
1-year CNS failure-free: 27% (SRS alone) vs 73% (SRS+WBRT), P=.0003
Deaths at 4 months: 4 (13%) SRS alone vs 8 (29%) SRS+WBRT
1-year CNS failure-free: 27% (SRS alone) vs 73% (SRS+WBRT), P=.0003
Deaths at 4 months: 4 (13%) SRS alone vs 8 (29%) SRS+WBRT
Adverse events
Main adverse events: Grade 3 toxicity: 1 case each (SRS alone: aphasia; SRS+WBRT: seizures/neuropathy/altered consciousness). Grade 4 radiation necrosis: 2 patients in SRS alone arm. WBRT caused acute fatigue, alopecia.
Conclusions
SRS + WBRT caused significantly greater cognitive decline at 4 months compared with SRS alone, as measured by Hopkins Verbal Learning Test. Although WBRT markedly improved CNS disease control, this came at the cost of neurocognition. This trial provided the key evidence supporting omission of upfront WBRT in patients with limited BM receiving SRS.
Key Limitations
Key Limitations: Trial stopped early at N=58 — substantially underpowered to detect differences in OS or other secondary endpoints reliably. Bayesian stopping creates challenges for frequentist interpretation. Higher early mortality in the WBRT arm (29% vs 13% at 4 months) may reflect prognostic imbalance rather than treatment harm. The 4-month cognitive endpoint may miss longer-term benefit of WBRT (reduced intracranial failure) or longer-term harm. HVLT-R is only one cognitive domain.
Clinical Context
Despite small N, this trial had major practice impact because it demonstrated measurable cognitive harm from WBRT on standardized testing. Combined with JROSG 99-1 (no OS benefit from WBRT), it established the rationale for SRS-alone strategies. The larger N0574 (Alliance 2016, N=213) confirmed these findings at scale. Current standard in patients with 1–4 BM and good prognosis: SRS alone with close MRI surveillance; WBRT reserved for poor prognosis.
References