onc brain

The longer read

This statement mostly codifies practice that changed with IMRT; it does not propose new practice. The split-course chemoradiation of RTOG 9610 and RTOG 9911 carried treatment-related mortality around 8%, and the conformal-era retrospective series that followed reported lower severe and fatal toxicity. The panel's algorithm is built on those later series. Its practical contribution is a sequence a department can adopt and cite. Selection is gated on interval since prior RT, resection status, and organ dysfunction using the Ward RPA classes, technique follows target volume and subsite, and the plan is built against cumulative EQD2 limits. A reader already treating with IMRT to 60-66 Gy without elective nodes will find that most statements confirm current practice.

The main caution is that agreement and evidence are separate axes. The paper reports both, and they should be read separately. Elective nodal omission reached 100% agreement at evidence level 3, and it rests largely on one multi-institutional retrospective cohort of 505 pts. That cohort also supplies the 66 Gy dose statement. A dose gradient in a retrospective reirradiation series is open to confounding by indication: pts with smaller targets, better performance status, and more distance from critical organs are the ones who can be taken to full dose. The 2-year OS spread of 49.3% vs 34.2% vs 30.4% across dose bands likely mixes a dose effect with selection. That fits the moderate 82% agreement and the authors' preference for a dose corridor over a single number.

The IMRT vs SBRT split follows the same logic. In the Vargo multi-institutional comparison, the RPA class II OS gap favoring IMRT (35.4% vs 18.6%) disappeared once SBRT was restricted to 35 Gy or more on small targets. That fits SBRT losing efficacy as volume rises. It fits equally well with the unrestricted SBRT group carrying lower doses and less favorable pts. The 25 cc cut is therefore a defensible working boundary, not a demonstrated threshold. It does not show SBRT fails on a target just above it, only that no one has shown it works there. The caution for larynx and hypopharynx is the firmer constraint, because that toxicity signal is large and specific to subsite, not a between-technique comparison.

The organ-at-risk panel is where the document is weakest, and it says so. Cord and brainstem cumulative limits drew the lowest agreement (70%), recovery factors carry very low evidence, and the carotid thresholds come from retrospective series. Because no-opinion votes counted against agreement, a moderate label may reflect abstention as much as dissent. These numbers work best as planning targets to document deviations against, not as tolerance data.

What the statement leaves open is as useful as what it settles. It does not answer how to sequence reirradiation with first-line checkpoint inhibition, how protons compare with IMRT, or whether hyperfractionation helps outside nasopharynx, and it excludes nasopharyngeal disease entirely. The evidence most likely to move its moderate-consensus statements would be a prospective comparison of SBRT and IMRT stratified by volume, or prospective cumulative-dose toxicity data for serial organs.