onc brain

The longer read

The number that governs how this trial should be read is 89.9% versus 88.4%: nearly nine in ten patients in both arms received radiotherapy that included nodal target volumes. The authors say so themselves, placing SENOMAC alongside AMAROS and OTOASOR rather than Z0011, SINODAR-ONE or POSNOC. This is not a trial of leaving a positive axilla alone. It is a trial of which modality treats it, and the answer is that regional nodal irradiation plus modern systemic therapy substitutes for dissection without a recurrence-free survival cost at five years. A reader who takes the headline as licence to omit both the dissection and the nodal fields is drawing a conclusion the trial did not test.

What SENOMAC adds over its predecessors is population, not direction. Z0011 and AMAROS both pointed this way and both were criticised for enrolling a population easier than the one in clinic: nearly 40% micrometastasis-only, one T3 patient between them, 248 mastectomies (17.4%) in AMAROS and none in Z0011, extracapsular extension either excluded or unreported. SENOMAC restricted entry to macrometastases, allowed extracapsular extension, enrolled T3 disease, and randomised more than a third of its patients to or after mastectomy. Each of those was a live source of hesitation at tumour board, and each is now covered by randomised data pointing the same way as the two trials it was designed to validate. That convergence, more than the hazard ratio itself, is what should move confidence.

The dissection arm is the trial's most instructive feature, because it quantifies what omission leaves in place. Additional non-sentinel-node metastases were found in 403 of 1167 patients (34.5%), reaching 51.3% among those with two macrometastases, and 9.9% of the dissection arm were reclassified pN2 with 3.0% pN3. Recurrence-free survival was nonetheless noninferior. The honest reading is not that this disease was absent but that it was treated, by adjuvant systemic therapy in 65% receiving chemotherapy and 93% endocrine therapy, and by the nodal fields. That framing matters for a radiation oncologist because it makes nodal coverage load-bearing rather than incidental, and the trial cannot yet say at what dose or to which levels: the detailed target-volume and dose review is explicitly still ongoing.

Two things should temper the read. Median follow-up is 46.8 months in a population that is overwhelmingly luminal, the subtype whose recurrences arrive late, so the five-year estimate is early for the biology in question and the primary endpoint of overall survival remains unreported. And the arm-morbidity case, the entire clinical rationale for de-escalating, rests here on questionnaires with no clinical lymphedema measurement and no three-year data yet for the full cohort. AMAROS supplies that piece, with lymphedema roughly twice as common after dissection as after axillary radiotherapy, but SENOMAC has not yet supplied its own.

For the result to be wrong, the non-sentinel disease left in situ would have to convert to clinical recurrence on a longer horizon than the current follow-up covers, which is precisely the failure mode a luminal population invites. The confidence interval helps, since 1.19 sits well below the 1.44 boundary and the event count rather than the missed accrual target drives precision. The ER-positive/HER2-positive subgroup HR of 0.26 (0.07-0.96) should be ignored: 3 events against 10, one of thirteen subgroups, and no mechanism by which forgoing an operation improves outcomes.