Dose-tailored stereotactic radiosurgery for multiple brain metastases
Brain metastases are the most common intracranial tumor in adults, arising far more frequently than primary brain tumors. Detecting every lesion — including subcentimeter deposits — is the essential first step in SRS planning, since each metastasis will receive its own individually contoured, individually dosed treatment.
Brain metastases occur in an estimated 20–40% of adult cancer patients, making them roughly ten times more common than primary malignant brain tumors. Improvements in systemic therapy have paradoxically increased their incidence: patients now survive long enough for occult micrometastases to become clinically apparent in the brain, a sanctuary site historically less penetrated by systemic drugs.
The three most common primary tumors seeding the brain are lung cancer (especially adenocarcinoma and small-cell histology, accounting for roughly half of all cases), breast cancer (particularly HER2-positive and triple-negative subtypes), and melanoma (which has a disproportionately high propensity for CNS spread relative to its incidence). Renal cell carcinoma and colorectal cancer round out the most frequent sources.
Brain metastases are typically multiple rather than solitary — autopsy and MRI series show that 50–65% of patients present with more than one lesion at diagnosis, which is precisely the clinical scenario single-fraction SRS is designed to address.
Accurate detection and delineation of every metastasis requires volumetric, thin-slice (typically 1 mm or thinner) T1-weighted MRI obtained after intravenous gadolinium contrast. Because metastases disrupt the blood-brain barrier, they enhance avidly, appearing as bright, well-circumscribed nodules — often with surrounding vasogenic edema visible on T2/FLAIR sequences.
Standard clinical MRI with thicker slices (5 mm) can miss lesions under 5–10 mm; dedicated stereotactic protocols with sub-millimeter isotropic acquisition and double- or triple-dose contrast improve sensitivity for the small lesions that are common in patients with more than a handful of metastases. Each detected lesion is individually contoured (gross tumor volume, GTV) by the treating radiation oncologist and neuroradiologist on the planning MRI, fused to a stereotactic CT for coordinate accuracy.
Because SRS treats each metastasis as its own discrete target with a margin-free or near margin-free GTV-to-PTV expansion, missing a lesion on imaging means it simply will not be treated — meticulous thin-slice detection is therefore inseparable from good SRS outcomes.
Two variables captured at this detection stage — the total number of metastases and the diameter of the largest lesion — drive nearly every downstream planning decision: the prescribed dose per lesion (via the size-dose relationship), total beam-on and session time, and the achievable sparing of normal brain.
Historically, more than three or four metastases prompted a default to whole-brain radiotherapy (WBRT). Contemporary practice, informed by trials showing comparable survival and better neurocognitive preservation with SRS alone, now extends single-fraction SRS to well-selected patients with substantially higher lesion counts, provided total treated volume and normal-brain dose constraints remain acceptable.
Single-fraction SRS dose is not one-size-fits-all. The landmark RTOG 90-05 dose-escalation trial established the size-dependent maximum tolerated dose that remains the backbone of modern prescribing: smaller lesions can safely receive higher single-fraction doses, while larger lesions must be dosed lower to keep radionecrosis risk acceptable.
RTOG 90-05, led by Shaw and colleagues and published in the International Journal of Radiation Oncology in 2000, was designed to identify the maximum single-fraction dose tolerable for recurrent previously irradiated primary brain tumors and metastases, stratified by tumor diameter. Dose was escalated in successive cohorts until unacceptable central nervous system toxicity was observed.
The trial established three diameter tiers with distinct maximum tolerated doses: tumors ≤20 mm tolerated up to 24 Gy in a single fraction; tumors 21–30 mm tolerated up to 18 Gy; and tumors 31–40 mm tolerated up to 15 Gy. Lesions larger than 40 mm were generally excluded from single-fraction dosing altogether because of unacceptable toxicity, and are more often managed with fractionated stereotactic radiotherapy or surgical resection followed by SRS to the cavity.
These thresholds remain the most widely cited dosing reference in brain SRS more than two decades later, though many centers now prescribe modestly below the RTOG maxima (for example 20 Gy rather than 24 Gy for small lesions) to further reduce toxicity while preserving high local control.
The inverse relationship between tumor volume and tolerated dose reflects a simple physical reality: larger tumors require larger treated volumes of surrounding normal brain to receive an ablative dose, and normal brain tolerance — not tumor biology — is what ultimately limits the prescription.
The toxicity that constrains SRS dose is radiation-induced brain necrosis — a delayed, often months-to-years-later injury to normal brain tissue within or adjacent to the high-dose region, caused by vascular endothelial damage and downstream inflammatory cascades rather than direct tumor effect.
Radionecrosis risk correlates strongly with both prescribed dose and treated volume. The volume of normal brain receiving 12 Gy or more (V12Gy) is the most widely validated predictor: multiple series show symptomatic radionecrosis risk rises sharply once V12Gy exceeds roughly 5–10 cubic centimeters, with risk continuing to climb as V12Gy grows further. This is precisely why larger lesions — which generate larger V12Gy volumes for any given dose — must be prescribed lower doses to keep V12Gy, and therefore necrosis risk, within an acceptable range.
Management of established radionecrosis includes corticosteroids, bevacizumab (an anti-VEGF antibody that reduces the vascular permeability driving edema), and occasionally laser interstitial thermal therapy or surgical resection for refractory, symptomatic cases.
Because a single patient may harbor metastases spanning a wide range of sizes, the size-dose relationship is applied independently to every lesion — a 5 mm lesion and a 35 mm lesion in the same patient, treated in the same session, will receive markedly different prescriptions (for example 24 Gy versus 15 Gy). This lesion-by-lesion tailoring is a defining feature of multi-target SRS and has no equivalent in whole-brain radiotherapy, where the entire brain receives one uniform, comparatively low dose regardless of individual tumor size.
Modern treatment planning systems compute each lesion's dose-volume histogram independently, and composite plans sum the cumulative normal-brain dose across all treated lesions to confirm the aggregate V12Gy remains within institutional safety limits even when many lesions are treated in a single session.
Modern SRS platforms (Gamma Knife, robotic linac systems, and dedicated cranial linac cones) allow each metastasis to be planned with its own independently optimized isocenter — or several isocenters for larger, irregularly shaped lesions — producing a tightly conformal, steep dose falloff around every target while sparing everything in between.
Unlike whole-brain radiotherapy, which delivers a single field arrangement covering the entire cranial vault, multi-lesion SRS is really a series of discrete micro-plans stitched into one appointment. Each metastasis gets its own isocenter (the point around which radiation beams converge), its own collimator or cone selection sized to the lesion, and its own dose calculation matched to the size-based prescription from the previous stage.
For small, roughly spherical lesions, a single isocenter with a matched circular collimator or multi-leaf-collimator aperture achieves excellent conformality. Larger or irregularly shaped metastases — particularly those with lobulated or ellipsoid geometry — are covered with two, three, or more isocenters of overlapping spherical dose distributions, an approach pioneered on Gamma Knife and now achievable with volumetric-modulated arc therapy (VMAT) on linear accelerators as well.
Because each lesion is planned and delivered sequentially, treating five or eight metastases in one session simply means running five or eight short sub-plans back to back, with total session time scaling roughly with lesion count and aggregate volume rather than remaining fixed as in WBRT.
The defining physical signature of SRS is an extremely steep dose gradient at the target edge — the prescription isodose surface tightly hugs the tumor, and dose falls off by more than 50% within a few millimeters outside it. This is achieved through the convergence of many non-coplanar beams or arcs, each individually weak, that overlap only at the isocenter to produce an ablative dose exactly where intended.
Conformity index (the ratio of the prescription isodose volume to the target volume, ideally close to 1.0) and gradient index (how rapidly dose falls outside the target) are the key plan-quality metrics reviewed for every lesion before delivery. Tight conformality is what allows adjacent normal brain, sometimes only millimeters away, to receive a negligible dose even while the tumor itself receives an ablative single fraction.
A well-planned 15 mm metastasis prescribed 24 Gy can have essentially normal-looking brain tissue just 5–8 mm away receiving under 10% of the prescription dose — a gradient far steeper than anything achievable with conventional fractionated fields.
Patients are immobilized in a rigid stereotactic frame or a precisely fitted thermoplastic mask with image-guided robotic couch correction, allowing sub-millimeter positioning accuracy throughout a session that may last well over an hour when many lesions are treated. Lesions are typically sequenced from a pre-planned order optimized for beam angle efficiency and patient comfort, with intrafraction imaging verifying position before and often during delivery of each sub-plan.
This single-session, multi-target workflow is what makes it practical to treat five, eight, or more metastases with individually tailored dose in one outpatient visit — a throughput advantage that has helped drive the expansion of SRS eligibility to higher lesion counts.
Once every lesion has been treated, the cumulative dose distribution is a scattering of small, tightly conformal high-dose clouds surrounding each metastasis — with the vast majority of brain volume between lesions receiving a dose close to zero. This is the central physical argument for SRS over whole-brain radiotherapy in patients with limited metastatic burden.
Whole-brain radiotherapy delivers a single, spatially uniform dose (classically 30 Gy in 10 fractions or 37.5 Gy in 15 fractions) across the entire intracranial contents, tumor and normal tissue alike, on the rationale of treating both visible and presumed microscopic disease everywhere in the brain simultaneously.
SRS instead treats only the visible, contoured lesions, each with its own small conformal dose cloud. In a patient with, for example, six metastases averaging 12 mm in diameter, the summed high-dose volume across all lesions typically represents well under 5% of total intracranial volume — meaning over 95% of normal brain parenchyma receives a dose close to zero. This sparing is the physical basis for SRS's markedly better preservation of neurocognitive function relative to WBRT.
Three landmark randomized trials reshaped practice away from routine addition of WBRT to SRS. Aoyama et al. (JAMA 2006) randomized patients with 1–4 metastases to SRS alone versus SRS plus WBRT, finding that added WBRT reduced distant intracranial recurrence but did not improve overall survival. Chang et al. (Lancet Oncology 2009) showed that adding WBRT to SRS caused significantly greater neurocognitive decline at 4 months (learning and memory function) despite better intracranial control, prompting early trial closure on cognitive safety grounds. JCOG0901 (Japan) similarly demonstrated preserved neurocognitive function with SRS alone.
The consistent finding across these trials is a trade-off: SRS alone yields higher rates of new (distant) brain metastases developing over time compared with SRS plus WBRT, but this is not accompanied by a survival penalty and is offset by substantially better preserved cognition and quality of life — and, critically, by the ability to treat any new lesions with repeat SRS rather than committing the patient to whole-brain toxicity up front.
The paradigm shift is not that distant brain failure stopped mattering — it is that salvage SRS for new lesions as they appear achieves comparable disease control to upfront WBRT without paying the neurocognitive price for patients whose disease never recurs.
Historical practice restricted SRS-alone to patients with three or four metastases or fewer, reserving WBRT for higher counts. Prospective and retrospective evidence, most notably Yamamoto et al. (Lancet Oncology 2014, JLGK0901), demonstrated that overall survival was statistically similar for patients with 5–10 metastases treated with SRS alone compared with those having 2–4 metastases, provided total cumulative tumor volume remained modest.
Current guidelines from ASTRO and other societies therefore support extending SRS-alone to well-selected patients with more than four, and in appropriate cases ten or more, brain metastases — with the total treated volume and predicted normal-brain dose (rather than lesion count alone) increasingly recognized as the more clinically meaningful constraint.
Because SRS irradiates only the treated lesions and spares the rest of the brain, patients remain eligible for repeat courses of focused radiosurgery whenever new metastases are detected on surveillance imaging — a repeatability that whole-brain radiotherapy, constrained by cumulative brain dose tolerance, cannot offer.
Patients treated with SRS alone (without upfront WBRT) enter a structured surveillance program, typically with contrast-enhanced brain MRI every 2–3 months for the first one to two years, since this is the period of highest risk for both local recurrence at treated sites and distant new metastases elsewhere in the brain. Surveillance intervals may lengthen thereafter if disease remains controlled.
Each surveillance scan is reviewed both for local control of previously treated lesions (looking for regrowth versus expected post-SRS changes such as transient enhancement) and for any new enhancing lesions elsewhere in the brain, which are contoured and staged exactly as the original cohort was in Stage 1.
Distant brain failure — new metastases appearing outside the originally treated volume — occurs in roughly 40–50% of patients within a year of SRS alone, reflecting ongoing seeding from systemic disease rather than a failure of the original treatment. This is the expected trade-off accepted in exchange for neurocognitive preservation, as established by the trials discussed in Stage 4.
When new lesions are found, options include repeat (salvage) SRS to the new metastases, occasionally combined with continued surveillance of previously treated sites, or transition to WBRT (or hippocampal-avoidance WBRT) if the number or distribution of new lesions becomes unfavorable for further focal treatment. Because SRS deposits its dose in small, spatially separated volumes, the brain retains substantial reserve capacity for additional focused courses — a second or third round of SRS targeting newly discovered lesions rarely overlaps meaningfully with prior treated volumes.
This repeatability is arguably SRS's greatest practical advantage: a patient can, in principle, be treated for new brain metastases every few months for years, something categorically impossible with whole-brain radiotherapy once cumulative brain dose approaches tolerance.
For patients who do eventually require whole-brain treatment — very high lesion burden, leptomeningeal spread, or after exhausting reasonable focal options — hippocampal-avoidance WBRT (HA-WBRT) has become the preferred technique over conventional WBRT. By conformally reducing dose to the bilateral hippocampi, structures critical to memory formation, HA-WBRT (studied in the NRG Oncology CC001 trial) reduces the relative risk of memory decline by roughly a quarter compared with standard WBRT, often combined with the neuroprotective drug memantine.
HA-WBRT does not eliminate the fundamental limitation that whole-brain treatment can typically be delivered only once in a patient's lifetime because of cumulative normal-tissue tolerance — reinforcing why SRS, with its capacity for repeated focal courses, remains the preferred first-line strategy whenever lesion number and distribution make it feasible.