onc brain

The longer read

The document's real contribution is to treat isolated intraprostatic radiorecurrence as a distinct, potentially curable disease state at exactly the moment PSMA PET is manufacturing the diagnosis faster than the evidence can absorb it. That timing is also the central weakness, and the panel says so plainly: the studies it synthesised were done in men worked up before PSMA PET, whose primary treatment was conventionally fractionated external beam RT, with prior brachytherapy patients excluded outright. Moderate and ultrahypofractionated primary RT is now standard, so the men a reader will be re-treating five years from now are not the men in this evidence base.

The biopsy mandate is the most defensible recommendation, and it rests on a specificity problem rather than a sensitivity one. FORECAST put mpMRI sensitivity at 94% (95% CI 88-98%) against a specificity of 18% (95% CI 7-35%), and Fendler reported an 8% false-positive rate against histopathology, largely in men with post-treatment change. When the intervention on the other side of the imaging finding is a second course of prostate radiation, the false positive is the expensive error. The insistence on both systematic and targeted sampling follows from the FORECAST secondary analysis showing up to 59% of true radiorecurrences would be missed by targeted biopsy alone.

The modality recommendation deserves a more skeptical read. It rests almost entirely on MASTER, a pooled synthesis of 150 largely retrospective studies with acknowledged follow-up asymmetry between older and newer salvage techniques. On efficacy MASTER was flat: adjusted 5-yr recurrence-free survival of 50% to 60%, with no survival difference between any modality and salvage prostatectomy. The preference for reirradiation is therefore a toxicity argument, not a disease-control argument, and readers should carry it that way. The numbers driving it (severe GU 20% after RP against 5.6% after SBRT, 9.6% after HDR and 9.1% after LDR brachytherapy) are adjusted rates from pooled retrospective series with no PSMA PET selection and no assessment of sexual toxicity, which is not a trivial omission in this population.

Where the panel failed to agree is more informative than where it agreed. Short PSA doubling time with early failure, castrate-resistant local recurrence, a second salvage attempt, and prior grade 3 or worse toxicity all produced disagreement. Those are not exotic scenarios; they are a large share of who actually presents. Read honestly, the disagreement argues for trial enrolment or continued imaging surveillance rather than for picking a modality by preference.

The hormonal recommendation is the weakest link and the text is candid about it: no study met the inclusion bar, and the 4 to 6 month LHRH agonist course is extrapolated from de novo intermediate-risk data. The comparator being displaced, hormones alone, was defined by Crook, TOAD and EMBARK, all of which enrolled before PET-based selection and none of which isolated a biopsy-confirmed local-only cohort.

What would have to be true for the framework to be wrong is that PET-detected isolated intraprostatic recurrence usually travels with occult distant disease, in which case local salvage buys toxicity and little else. RTOG 0526 is the shape of that worry in prospective data: a 10-yr local failure rate of 5% alongside disease-free survival of 61% at 5 yr falling to 33% at 10 yr. Local control was not the problem.