Case-Based Radiosurgery Society Practice Guideline: SRS for Brain Metastases
TL;DRUp to 15 lesions, ≤3 cm and ≤14 cc are SRS-suitable. Fractionate 2-4 cm lesions (27 Gy/3 fx or 30 Gy/5 fx). Above 4 cm, surgery comes first.
RTOG 9508N107C/RTOG 1270
The fractionation advice is driven by toxicity more than control. In 4-14 cc lesions, radionecrosis was 23.1% single-fraction vs 7.3% multifraction (P = .003). Above 14 cc and 3 cm, neither LC nor necrosis differed significantly, which is the guide's case for moving to surgery. Margin is the other lever: going from 0 to 2 mm triples V12.
In an intact 2-4 cm metastasis where resection is not an option, this supports 27 Gy/3 fx or 30 Gy/5 fx over single-fraction SRS. It does not extend to classical leptomeningeal disease or lesions >6 cm.
Fractionation is a toxicity lever: in 4-14 cc lesions, radionecrosis was 23.1% single-fraction vs 7.3% multifraction (P = .003). Margin is the other lever, since 0 → 2 mm triples V12. That is the basis for a 1 mm PTV whenever setup confidence allows.
Most cited studies support giving systemic therapy concurrently with SRS without a washout. BRAF inhibitors (intratumoral bleeding) and erlotinib with WBRT plus SRS are the flagged exceptions. Per ASCO-SNO-ASTRO, deferring SRS for asymptomatic mets on CNS-active therapy is a conditional, multidisciplinary call.
Resection is first-line above 4 cm. The size and volume thresholds (≤3 cm, ≤14 cc) mark where SRS alone stops being the default. Preop SRS reported 5.8% meningeal disease vs 16% to 21% postop, but only in initial studies, and LC, distant brain failure and OS were similar.
11 details 5 trials watching
Case-based practice guideline sponsored by the Radiosurgery Society, built around two cases: 7 NSCLC brain mets treated with 20 Gy/1 fx, and a resected 3.5 cm cerebellar colon met given 27 Gy/3 fx postop. The source states no panel voting, evidence grading or agreement metric.
Suitable: 1-15 lesions, diameter ≤3 cm, volume ≤14 cc. Cautionary: >15 lesions, 3-6 cm or >14 cc, SCLC, nodular LMD. Unsuitable: >6 cm or classical LMD.
<2 cm: 20-24 Gy/1 fx; 2-4 cm: 27 Gy/3 fx or 30 Gy/5 fx; >4 cm: surgery, with multifraction SRS if inoperable. Intact-lesion target is GTV on contrast T1 with a 1-2 mm PTV (1 mm whenever possible) and no CTV. Postop target is the cavity CTV with a PTV typically 2 mm, delivered within 4 wk of surgery.
Single-fraction brain V12 of 5, 10, >15 cc carries ~10%, 15%, 20% symptomatic RN. For reirradiation the guide cites total brain EQD2 <120 Gy, cumulative cord EQD2 ≤70 Gy and at least 5 months between courses.
The guide builds on RTOG 9508 (WBRT + SRS vs WBRT) and on N107C/RTOG 1270 in the postop setting, which made SRS alone standard for 1-4 mets. It extends SRS beyond that population to up to 15 lesions, larger lesions via fractionation, and preoperative SRS.
The 15-lesion boundary rests on a prospective observational study and an MD Anderson trial terminated early and published only as an abstract. The preop SRS LMD advantage comes from initial studies, and the guide itself reports similar LC, distant brain failure and OS for pre- and postoperative SRS.
Volume may matter more than count: on MVA, tumor volume, KPS and histology predicted OS but lesion number did not. Integral brain dose is offered as the practical ceiling, where a 9-, 177- or 42-tumor burden yields an 8 Gy whole-brain dose depending on size. The guide calls SRS in SCLC an acceptable option but not yet the default.
| Category | Lesion number | Metastasis size | Specific scenarios |
|---|---|---|---|
| Suitable | 1-15 | Diameter ≤3 cm, volume ≤14 cc | n/a |
| Cautionary | >15 | Diameter 3-6 cm, volume >14 cc (operative management preferred) | SCLC, nodular leptomeningeal disease |
| Unsuitable | n/a | >6 cm | Classical leptomeningeal disease |
| Lesion size | Recommended approach |
|---|---|
| <2 cm | 20-24 Gy in 1 fx (ASTRO) |
| 2-4 cm | 27 Gy in 3 fx or 30 Gy in 5 fx; staged SRS an alternative |
| >4 cm | Surgery; multifraction over single-fraction SRS if inoperable |
| Group | 1-yr LC SF vs MF | Radionecrosis SF vs MF |
|---|---|---|
| A: 4-14 cc | 77.6% vs 92.9% (p not legible in source) | 23.1% vs 7.3%, P = .003 |
| B: >14 cc, >3 cm | 77.1% vs 79.2%, P = .76 | 11.7% vs 6.5%, P = .29 |
| Organ | 1 fx | 3 fx | 5 fx |
|---|---|---|---|
| Brain | V12 <5-10 cc; V10 <12 cc | V18 <26 cc; V21 <21 cc; V24 <16.8 cc | V25 <16 cc; V28.8 <7 cc; V30 <10.5-30 cc |
| Brain stem (not medulla) | Dmax <15 Gy; V10 <0.5 cc | Dmax <23.1 Gy; V15.9 <0.5 cc | Dmax <31 Gy; V23 <0.5 cc |
| Spinal cord and medulla | Dmax <12.4-14 Gy; V10 <0.35 cc | Dmax <20.3-22.5 Gy; V15.9 <0.35 cc | Dmax <25.3-28 Gy; V22 <0.35 cc |
| Cochlea | Dmax <9 Gy | Dmax <14.4 Gy | Dmax <22 Gy |
| Optic pathway | Dmax <10 Gy; V8 <0.2 cc | Dmax <17.4-20 Gy; V15.3 <0.2 cc | Dmax <25 Gy; V23 <0.2 cc |
- Does preop SRS lower leptomeningeal disease vs postop SRS in prospective trials? active Comparing the Addition of Radiation Either Before or After Surgery for Patients With Brain Metastases Phase 3n=237 · primary completion 2027-03 · phase 3 SRS before vs after resection, 1-4 BMrecruiting Trial of Preoperative Radiosurgery Versus Postoperative Stereotactic Radiotherapy for Resectable Brain Metastases Phase NAn=200 · primary completion 2028-12 · preop SRS vs postop SRT, 12-mo LMD incidence
- Role of SRS vs WBRT in SCLC brain metastases n=340 · primary completion 2027-08 · phase 3 SRS/SRT vs HA-WBRT, ≤10 BM in SCLC
- SRS vs WBRT for more than 15 brain metastases active Hippocampal Sparing Whole Brain Radiation Versus Stereotactic Radiation in Patients With 5-20 Brain Metastases: A Phase III, Randomized Trial Phase NAn=196 · primary completion 2024-11 · randomized HA-WBRT vs SRS for 5-20 BMn=126 · primary completion 2026-12 · SRS alone vs SRS+WBRT, 5-30 BM, observational
📚 Sources · 📄 1 paper
The longer read
This is a practice guide rather than a graded guideline: two cases, a suitability table, a dose-constraint table and narrative recommendations, with no stated panel process, evidence grading or agreement metric. Its value is in writing down what high-volume SRS programs already do. Its weak points sit where the field is moving fastest.
The lesion-count ceiling of 15 is the clearest example. The authors base it on a 2014 multi-institutional prospective observational study, which found 5 to 10 lesions (total volume <15 cc) noninferior to 2 to 4, and on an MD Anderson trial of SRS vs WBRT in 4 to 15 metastases that was terminated early and exists only as an abstract. That is a thin base for a hard 'suitable' boundary. The paper itself concedes that volume may outrank count, citing an MVA in which tumor volume, KPS and histology predicted OS while lesion number did not. A reader treating 12 small lesions and a reader treating 4 bulky ones should take more from the volume argument and the whole-brain dose estimate than from the number in Table 1.
The fractionation advice is better grounded, and its strongest data concern toxicity, not control. In the cited meta-analysis of 24 studies and 1887 lesions, multifraction SRS cut radionecrosis from 23.1% to 7.3% (P = .003) in 4-14 cc lesions. For the largest group, neither LC (77.1% vs 79.2%, P = .76) nor necrosis (11.7% vs 6.5%, P = .29) separated. That pattern supports the guide's move to surgery above 3 to 4 cm: fractionation buys safety in mid-size lesions but does not rescue control in the biggest ones. The HyTEC estimate of 1-year LC 69% to 75% single-fraction vs approximately 80% fractionated for 2.1 to 4 cm lesions points the same way.
The preoperative SRS section deserves the most caution. The 5.8% meningeal disease rate, against 16% to 21% postoperatively, comes from what the authors call initial studies. They also report that pre- and postoperative SRS have shown similar LC, distant brain failure, OS and need for salvage WBRT. The LMD advantage is a hypothesis for the ongoing prospective trials to test. Postoperative SRS remains the option with level 1 evidence, a point Table 3 makes explicitly.
The concurrent systemic therapy guidance (no routine washout) rests on a systematic review and retrospective series. BRAF inhibitors (intratumoral bleeding) and erlotinib given with WBRT plus SRS are the named exceptions. The survival gap of 41.6 vs 21.5 months with concurrent therapy is confounded by which patients were fit enough to receive it and should not be read as a treatment effect. Endorsing SRS in SCLC on a 9-study meta-analysis while declining to make it the default is the right calibration. Here the guide is candid that the prospective data are missing.