EXTEND
ForOligometastatic solid tumors, 1-5 metastases, on standard systemic therapy
HR 0.54
95% CI 0.41-0.72, p<0.001
TL;DRPFS HR 0.54 (0.41-0.72), p<0.001 for MDT added to SOC across 6 oligometastatic baskets; RT delivered 98% of MDT.
The prostate-excluded HR 0.60 (0.40-0.89) is the number that matters for an RT reader: the benefit survives removal of the two prostate baskets, where MDT is already routine. RT delivered 98% of MDT (370/379), so this is a radiotherapy result, though dose and fractionation are not reported in source.
In a patient with 1-5 metastases from pancreas or a non-breast, non-kidney histology already on standard systemic therapy, this supports discussing MDT as an addition rather than a deferral; it does not resolve the breast or kidney question, where the baskets were inconclusive.
RT delivered 98% of MDT (370/379 metastases), so the all-basket HR 0.54 and the prostate-excluded HR 0.60 (0.40-0.89) are radiotherapy effect sizes. The prostate-excluded analysis is the one that moves practice beyond the population where MDT is already routine, though dose, fractionation and target volume are not reported in source.
MDT was added on top of standard systemic therapy rather than substituted for it, so nothing here supports deferring or de-escalating systemic treatment. The ctDNA findings (detectable at enrollment with shorter PFS, clearance at 3 months with better survival) are the medonc-relevant signal, pointing toward a molecular rather than lesion-count definition of who to refer.
13 details 5 trials watching
Multicenter randomized phase II basket trial, 6 histology baskets with basket-specific stratification and powering. Accrual 2018 to 2023, median follow-up 53 months.
Patients with 1-5 metastases on standard-of-care systemic therapy, allocated to breast, pancreas, kidney, two prostate baskets, or an "Other" basket. 521 screened, 350 randomized, 334 analyzed per protocol (MDT+SOC n=166; SOC n=168).
Radiotherapy delivered 98% of MDT (370/379 metastases), so this is effectively a radiotherapy trial. Dose, fractionation, technique and target-volume definition are not reported in source.
Primary: PFS, pre-specified in the per-protocol set at three levels (within each basket, across all baskets, and across all baskets excluding the prostate baskets). Exploratory: ctDNA and immune profiling.
All-basket PFS HR 0.54 (95% CI 0.41-0.72), p<0.001; excluding prostate, HR 0.60 (95% CI 0.40-0.89). Superiority in pancreas, prostate and "Other"; breast and kidney inconclusive. No per-basket effect sizes reported in source.
SABR-COMET randomized 99 patients across mixed histologies; STOMP and ORIOLE were prostate-only and smaller. EXTEND's contribution is scale plus histology resolution: it keeps a benefit when the prostate baskets are removed, which the prostate-only trials could not address.
The primary analysis is per-protocol rather than ITT, and an unblinded PFS endpoint in a trial that ablates the very lesions being measured favors the intervention arm. The "Other" basket is a mixed-histology pool, so its superiority signal is the hardest of the three to carry into a single phase III.
The prostate-excluded HR 0.60 is the trial's most load-bearing number, since it shows the pooled result is not simply the prostate literature reasserting itself. The ctDNA correlations (detectable at enrollment with worse PFS and survival; clearance at 3 months with better survival) point at a biological rather than anatomic definition of oligometastasis, which is the more interesting question EXTEND raises without settling.
CONSORT flow
Randomized phase II, per-protocol primary, histology-specific signals explicitly framed as hypothesis-generating for phase III. Consistent with SABR-COMET / STOMP / ORIOLE direction.
- Does MDT-driven PFS benefit translate to overall survival n=340 · primary completion 2026-11 · OLIGAMI: randomised MDT after 12wk systemic, breastactive Stereotactic Ablative Radiotherapy for Comprehensive Treatment of Oligometastatic (1-3 Metastases) Cancer Phase NAn=330 · primary completion 2030-11 · SABR-COMET-style RCT with OS as primary, 1-3 mets
- Can ctDNA select oligometastatic patients for MDT n=60 · primary completion 2027-12 · SABR cohort tracking ctDNA dynamics as biomarker
- Confirmatory phase III in pancreas oligometastatic disease recruiting Stereotactic Body Radiotherapy in Patients With Rare Oligometastatic Cancers (OligoRARE) Phase NAn=200 · primary completion 2028-08 · phase III SBRT+SOC, OS endpoint, pancreas among sitesactive SENECA: First Line metaStatic pancrEatic caNcer Primary and Distant (if Oligometastatic) lEsion direCted rAdiotherapy Phase 3n=108 · primary completion 2030-01 · randomised phase 3 SBRT vs chemo alone, 1L met pancreas
📚 Sources · 📄 1 paper
Abstract
The longer read
The most useful thing EXTEND does is separate two claims that the oligometastasis literature has been quietly conflating. The first is that ablating a small number of visible metastases delays progression. The second is that the delay is a property of oligometastatic biology in general rather than of prostate cancer in particular, where PSMA imaging, a sensitive biochemical endpoint, and a systemic therapy landscape full of deferrable options all conspire to make metastasis-directed therapy look better than it might elsewhere. Nearly every randomized dataset preceding this one either was prostate-only or, in SABR-COMET's case, was small enough that histology-level inference was not on the table. By pre-specifying an all-baskets-excluding-prostate analysis, EXTEND put that objection in the protocol rather than leaving it to correspondence, and the answer, HR 0.60 with a confidence interval that clears 1, is the finding a reader should carry forward.
How much confidence that deserves depends heavily on how one reads the endpoint. PFS in an unblinded ablative trial rewards the intervention arm structurally, not just through assessor bias: the lesions treated are the lesions being watched, and removing them removes the most likely sites of the next radiographic progression event. That mechanism can generate a real hazard ratio that does not translate into survival, and the trial reports no overall survival result in this analysis. Nothing here resolves whether MDT is deferring an event or deferring the appearance of an event. A reader who found SABR-COMET's survival signal persuasive will see EXTEND as consistent; a reader who found it fragile will find nothing here to change their mind, because EXTEND was not built to answer that question.
The per-protocol primary is the second thing to weigh. Sixteen of 350 randomized patients fall out, which is a small fraction, but the choice matters more in a trial where one arm receives an additional procedure: the patients most likely to deviate from protocol are the ones whose disease declared itself before MDT could be delivered, and excluding them shifts the comparison in a predictable direction. The magnitude of that shift is almost certainly smaller than the observed effect, but it is not zero, and it is one of several reasons the pooled estimate should be read as the upper end of what a phase III might reproduce.
The basket-level results are best treated as what the authors say they are, a menu for phase III design rather than a set of histology-specific answers. Pancreas is the surprise worth arguing about, since the prevailing assumption has been that a disease with such a high rate of occult systemic spread has little to gain from local ablation. If that signal replicates, it forces a rethink of the anatomic model of oligometastasis. Breast and kidney being inconclusive is less informative than it sounds, since basket-level powering in a trial this size makes inconclusive the expected result for anything but a large effect. The "Other" basket, meanwhile, is superiority in a pool defined by exclusion, which is the least portable finding on the page.
What may outlast the efficacy result is the correlative work. Detectable ctDNA at enrollment tracking with worse PFS and survival, and clearance at 3 months tracking with better survival, suggests the field's operating definition of oligometastasis, a count of visible lesions, is a proxy for something measurable in blood. Whether ctDNA can select patients prospectively is the trial that should follow this one.