KEY POINTS
- This single-institution retrospective cohort included 159 patients with 332 intact brain metastases treated from 2018–2023 with single-fraction linear accelerator SRS: 127 lesions received 20 Gy, 183 received 22 Gy, and 22 received 24 Gy. Median follow-up was 14 months.
- Dose selection was not randomized: higher doses were preferentially used for smaller lesions away from critical structures. Mean GTV was 0.75 cm³ at 20 Gy, 0.33 cm³ at 22 Gy, and 0.41 cm³ at 24 Gy (p<0.001), creating substantial confounding by indication.
- Overall one-year local control was 93.3%. By dose it was 90.1% with 20 Gy, 96.3% with 22 Gy, and 86.5% with 24 Gy (log-rank p=0.032); the 24-Gy cohort contained only 22 lesions and was underpowered for meaningful comparisons.
- After adjustment for GTV, 22 Gy remained associated with higher odds of one-year local control than 20 Gy (adjusted OR 3.04, 95% CI 1.22–7.59; p=0.017). The effect remained significant with cluster-robust standard errors but not with the authors' generalized estimating-equation sensitivity model, reducing confidence in the apparent dose effect.
- Radiation necrosis occurred in 69/332 lesions (20.8%). Crude rates were paradoxically higher after 20 Gy (29.9%) than 22 Gy (14.8%) because the 20-Gy lesions were larger; after adjustment, both prescription dose and GTV remained independently associated with necrosis, with GTV increasing the odds by 40% per doubling of volume (p<0.001).
- ROC-derived GTV thresholds for necrosis were 0.325 cm³ at 20 Gy and 0.203 cm³ at 22 Gy, but these cutoffs were unstable on bootstrap validation and are explicitly hypothesis-generating. Median overall survival was 14 months, with no significant difference across dose groups (p=0.63).
CLINICAL TAKEAWAY
The study supports a dose-response relationship for local control and reinforces tumor volume as a major determinant of radiation necrosis after single-fraction SRS. It does not establish that 22 Gy should routinely replace 20 Gy: treatment selection was strongly confounded by lesion size and location, the 24-Gy cohort was very small, and established dose-volume metrics such as V12 Gy were unavailable.