onc brain

The longer read

The useful claim here is narrower than the abstract's framing and more actionable than its size suggests. Eleven patients with two events cannot establish that 40 Gy in 15 fractions controls an in-breast recurrence as well as the hyperfractionated schedules that generated the existing re-RT evidence base. What the series can establish, and does, is that a once-daily moderately hypofractionated partial-breast re-RT plan is deliverable inside conventional organ-at-risk constraints in patients who have already had a full course of whole-breast irradiation. Heart mean dose came in at 1.44 Gy against a 3.2 Gy objective, ipsilateral lung V16Gy at 7.74% against 15%, and PTV V95% at 98.36% with zero volume above 107%. That is a planning result, and planning results generalize better than 11-patient outcome estimates do.

The comparison that matters is RTOG 1014, which delivered 45 Gy at 1.5 Gy twice daily and reported 7% late grade 3 toxicity with no grade 4 or 5 at 5.5 years, and the Janssen cohort of 83 patients at 45 Gy / 1.8 Gy daily with a 15% local recurrence rate at 35 months. Both are larger and both have longer follow-up. Neither is a randomised comparison either. The practical objection to the RTOG schedule was never its toxicity, it was that twice-daily attendance over three weeks is a real burden on a patient population that is, by construction, older and already through one full course of breast radiotherapy. If a daily 15-fraction schedule reaches the same place, that is the finding, and the dosimetric table is the evidence for it rather than the survival curves.

Several things should hold a reader's confidence back. Follow-up at a median of 41 months is short for in-breast recurrence after re-RT, and both events here occurred inside 15 months, which says nothing about the late failures that a 5-year read would capture. The cohort is heavily selected: age at least 50, unifocal disease confirmed on three imaging modalities, tumour under 2 to 3 cm, and a mandated interval of at least 48 months from primary treatment, which in practice was a median of 107 months between courses. A recurrence that late is a biologically different animal from one at 24 months, and the local control figure should not be carried across to a shorter-interval relapse. The retrospective design invites exactly the selection bias the authors name.

Two reporting problems deserve mention because they affect how much weight the numbers bear. The confidence intervals as printed (46-62%, 52-65%, 47-63%) do not contain their own 86% point estimates, so they cannot be read as stated. And no plan sum with the first course was possible, since the original dosimetry was unavailable. That is the single most important missing number in a re-RT paper: cumulative dose to skin, chest wall, heart and lung is what determines whether the second course is safe, and the clean second-course table cannot substitute for it. The unblinded cosmesis scoring, where 3 of 11 were rated poor while the abstract records cosmesis as good, is a smaller version of the same problem.

What would make this wrong is a late-toxicity signal emerging past four or five years, most plausibly fibrosis progressing in the tumour-bed volume where the two courses overlap, or an in-breast failure rate that drifts toward the 15% Janssen reported once follow-up matures. Neither is excluded by this data. The reasonable position is that daily hypofractionated PBI re-RT is worth planning and worth enrolling on a prospective study, not that it is established.