ISRS Technical Guidelines: SRS for Small Brain Metastases (≤1 cm)
TL;DRSub-mm geometric accuracy, ≤1 mm E2E tolerance, 5% dosimetric accuracy and MRI ≤7 days pre-SRS anchor ISRS physics consensus for ≤1 cm brain mets.
The margin is the lever on these targets: a 1-mm expansion takes a 0.8-cm met from 0.268 to 0.524 cc, and the one randomized comparison (1 vs 3 mm, 80 mets) showed no LC difference with more radionecrosis at 3 mm. Planning-MRI age also gates control: 12-mo local FFP 95% vs 56% at <14 vs ≥14 days.
The margin is the lever on these targets: a 1-mm expansion takes a 0.8-cm met from 0.268 to 0.524 cc, and the one randomized comparison (1 vs 3 mm) showed no LC difference with more radionecrosis at 3 mm. Planning-MRI age also gates control: 12-mo local FFP 95% vs 56% at <14 vs ≥14 days.
10 details 2 trials watching
ISRS physics guideline built on a PRISMA systematic review of PubMed 2009-2020 (81 papers plus 7 international guidelines) and a 10-question survey of 14 committee members, 100% completion.
Scope is SRS for brain metastases ≤1 cm in maximal diameter on Gamma Knife, CyberKnife and SRS-adapted LINACs. Platform mix of included papers: LINAC 21, mixed 21, GK 20, CK 2.
Requires sub-mm geometric accuracy, a ≤1 mm geometric tolerance on system-specific E2E tests, and dosimetric accuracy within 5%. MLC width of no more than 5 mm is called a prerequisite; no CTV margin is usually applied, and PTV margin use is left platform-dependent.
Defers small-field dosimetry and QA tolerances to AAPM TG142, TG198 and TG178 and IAEA TRS 483, and adopts the ICRU 91 GTV definition. The ISRS contribution is the small-target layer: imaging recency, slice thickness, contrast timing and margin trade-offs.
Survey reached no consensus on 3 of 10 questions, so several statements reflect unresolved practice variation. Commissioning, QA program design and platform-specific planning are explicitly out of scope, and several authors carry vendor ties.
The panel frames margin addition or omission as the most controversial variable for small mets: absolute necrosis risk at this size is low, so relative increases may matter less in absolute terms. Cumulative dose across repeat SRS sessions is flagged as an open, clinically relevant problem.
| Domain | Recommendation |
|---|---|
| Imaging | MRI ≤7 days from treatment is a prerequisite for contouring |
| Imaging | Fine-cut MR, ≤1.5-mm slices; planning CT no thicker than MR slices |
| Imaging | 10 to 15 min scan delay after contrast injection |
| Imaging | SRS-optimized sequences; regular MRI QA for geometric distortion |
| Contouring | Contour all targets and OARs; weigh margins per SRS platform |
| Patient treatment | Sub-mm geometric accuracy; immobilization and IGRT chosen to achieve it |
| Patient treatment | Lower energy beams may reduce normal tissue dose |
| Dosimetry | Dosimetric accuracy within 5%; small-field dosimetry per IAEA and AAPM |
| Study | Comparison | Finding |
|---|---|---|
| Randomized, 49 pts / 80 mets, LINAC | 1 vs 3 mm PTV margin | No LC difference; more radionecrosis with 3 mm |
| Retrospective, 93 mets, LINAC | With vs without 2-mm PTV margin | Severe parenchymal complications 7.1% and 19.6%; no LC effect |
| Sneed, retrospective GK, 2200 treatments | ≤1 cm lesions, ≤20 Gy, no margin | 1-yr adverse radiation effect 1% or less |
| Ma et al, GK modeling | Margins 0.5 to 3 mm | Necrosis risk up 6% to 25%; only 1 of 15 lesions <1 cm |
- Does no margin vs 1 mm PTV margin change LC for ≤1 cm mets? recruiting Trial of Differential Margins in Single Isocenter Radiosurgery of Brain Metastases Phase NAn=180 · primary completion 2028-06 · 0 mm vs 2 mm margin, mets ≤4 cm, single isocentern=166 · primary completion 2028-08 · randomized PTV margin omission, intact mets, PFS
- Cumulative normal-brain dose limits across repeat SRS courses
- Does 2.5 vs 5 mm MLC width change outcomes for small targets?
📚 Sources · 📄 1 paper
The longer read
This is a physics document, and it should be read as one. It does not tell a radiation oncologist what dose to give a 7-mm melanoma metastasis or whether to treat it at all; it tells a program what has to be true about its imaging, contouring, immobilization and dosimetry chain before a sub-centimetre target can be treated with confidence. Its value is in compressing a scattered technical literature into a short list of conditions, and in naming where the field has not converged.
The evidence underneath the recommendations is uneven. Of 81 included papers, 57 were single-institution, and the clinical outcome data on margins rest on a handful of studies: one randomized comparison of 1 vs 3 mm PTV margins in 49 pts with 80 mets, a retrospective series of 93 mets, Sneed's Gamma Knife experience of 2200 treatments, and a modeling exercise by Ma et al in which only 1 of 15 lesions was under 1 cm. The randomized trial found no local control difference and more radionecrosis at 3 mm, which supports keeping margins small, but it does not test zero against 1 mm, which is the decision most LINAC programs actually face. The committee's own survey reflects that gap: strong consensus on 5 of 10 questions and none on 3.
The most practice-relevant numbers concern time and geometry rather than dose. A planning MRI 14 or more days old was associated with 12-mo local freedom from progression of 56% against 95% for fresher imaging, and a separate series measured median volume growth of 20% over a median 8-day interval. Both are retrospective, but the direction is mechanistically plausible and the cost of acting on it is low, which is why the ≤7-day imaging recommendation is likely the easiest to adopt. The volume arithmetic for margins is similarly hard to argue with: a 1-mm expansion of a 0.8-cm sphere roughly doubles the treated volume, which for multiple adjacent mets compounds normal-brain dose.
Relative to the AAPM task group reports and IAEA TRS 483, which the document leans on rather than replaces, the ISRS contribution is specificity to target size. The single-isocenter data are a good example: uncertainty within 1 mm inside 4 cm of isocenter and 2 mm beyond is exactly the kind of figure a program treating many small mets with one isocenter needs to weigh when deciding whether a peripheral target needs its own plan or a margin.
What would make the guideline wrong is mostly what it does not measure. If penumbral dose or immune effects compensate for small geometric misses, as the authors speculate, then the sub-mm insistence may be stricter than outcomes require; if cumulative dose across repeat SRS courses proves to drive late toxicity, the single-session framing will understate risk. Vendor ties among several authors are disclosed and worth noting, though the recommendations are largely platform-neutral.