onc brain

The longer read

The efficacy half of this paper is the less contested half. A pooled 2-year local control of 89% lands where RTOG 0813 did for central and ultra-central tumors, and the BED dose-response on 1-year control echoes the long-standing BED10 ≥100 Gy principle. Heterogeneity above 90% makes the point estimate soft, but no plausible reweighting moves local control into a range that would argue against SBRT on efficacy grounds. The live question for ultra-central disease has always been whether the airway tolerates it, and that is where the paper should be read.

On toxicity, the 4% pooled G5 rate looks reassuring next to HILUS's 15%, but the two numbers are not measuring the same thing. HILUS was prospective, with mandated reporting, and it prescribed to a 67% isodose line with PTV margins up to 15 mm, so the central PTV ran near 150% and the main bronchus median Dmax reached 97 Gy. The retrospective series that drive the pooled estimate have the opposite bias: events are ascertained after the fact, attribution of a terminal hemoptysis to treatment versus progression is discretionary, and 60% of studies had median follow-up under two years in a toxicity whose timing extends past that. Publication bias was detected on every toxicity endpoint. A reader should treat 4% as a floor for ultra-central SBRT delivered in experienced centers, not as the expected rate.

The more durable contribution is the dosimetric synthesis. The meta-regression found no relationship between PBT Dmax and airway toxicity, and the authors are right that this reflects how few studies reported usable dosimetry rather than an absence of effect. The argument instead rests on convergence across individual series: fatal hemoptysis in Tekatli, Wang, Mihai and HILUS clusters at airway doses well above the 133-150 Gy BED3 constraint proposed, and the Fig 3 mapping places events disproportionately near the trachea and main bronchi. That is a reasonable basis for a conservative planning target, but it is expert inference from case-level data, not a dose-response model, and HILUS's substructure finding (4% vs 18% G5 hemoptysis by distance from the main bronchi) suggests a single PBT constraint is too coarse.

The risk-factor layer may matter more in clinic than the dose number. Endobronchial extension, antiplatelet or anticoagulant exposure, VEGF inhibitors near treatment, and ILD recur across the fatal cases. These are identifiable before simulation, and the guideline's practical move is to shift those patients toward 60 Gy/15 with tighter constraints rather than to abandon RT. The VEGF hazard ratio comes from one retrospective series and carries wide uncertainty, but the direction is consistent with the abdominopelvic SBRT experience.

What this does not settle: whether 60 Gy/8 and 60 Gy/15 differ in toxicity at matched airway dose, whether any constraint holds for tumors abutting the trachea itself, and whether adaptive MR-guided delivery changes the calculus. For the result to be wrong in the reassuring direction, the retrospective series would need to be undercounting fatal events by a factor near the gap with HILUS, which is not implausible. SUNSET and LUNGTECH are the prospective tests.