onc brain

The longer read

The number that should move a radiation oncologist here is not the hazard ratio, it is the two-year post-treatment biopsy: 19.6% positive with aglatimagene versus 36.4% with placebo. That is the mechanism, and the DFS curve is downstream of it. Read that way, this trial belongs in a specific and crowded conversation, the one about how much residual intraprostatic disease survives definitive EBRT and what is worth adding to reduce it. The discussant made exactly that framing explicit by putting CAN-2409 on the same slide as RTOG 9408, ASCENDE-RT and FLAME, and the comparison is not flattering in the way a positive phase 3 usually is. ASCENDE-RT cut ten-year local failure from 7.1% to 1.5% with an LDR boost; FLAME cut crude local failure from 7.7% to 2.7% with a 95 Gy simultaneous integrated boost. Those are tools already in the department, already reimbursed, already familiar. The honest question is not whether the viral therapy works but whether its local-control contribution is additive to a boost or substantially overlapping with it, and this trial cannot answer that because no boost strategy was specified in either arm.

The accrual window compounds the problem. Randomisation ran from February 2012 to September 2021, nine and a half years, and the standard of care did not hold still. Conventional 78 Gy was the reference schedule at the start; moderate hypofractionation became routine during, and the protocol permitted 60 Gy in 3 Gy and 70 Gy in 2.5 Gy. PSMA PET arrived and reclassified a meaningful fraction of what was then called high-risk localised disease into occult metastatic disease. The discussant's own STAMPEDE-versus-POP-RT slide made the same point about a different trial: a high-risk cohort staged conventionally is not the high-risk cohort in front of a reader today. A control arm with 86.1-month median DFS is a plausible historical benchmark and an uncertain contemporary one, and every part of the drift runs in the direction of shrinking the measured effect if the trial were repeated now.

Two structural features deserve more weight than they usually get. ADT use was a stratification factor, which means the investigators expected it to matter, yet its distribution and its interaction with the viral therapy are not in the source. Since ADT is itself a potent reducer of two-year biopsy positivity, RTOG 9408 took it from 40% to 20% with four months, the possibility that the aglatimagene effect is largest in the ADT-free stratum and attenuated where hormonal therapy is already delivering local kill is entirely live and entirely untested here. Separately, the primary endpoint is a composite of recurrence or death, and the biopsy-driven local component is doing the work while OS and prostate-cancer-specific mortality show one PCSM event per arm. That is not a negative survival result, it is an uninformative one, and at 50.3 months in an intermediate-and-high-risk localised population it could hardly be otherwise. Long-term follow-up is ongoing, and the trial's real verdict is several years out.

What would have to be true for this to be wrong in the direction that matters? Only that the residual disease being cleared at two years is disease a modern plan would have controlled anyway, through dose escalation, a boost, or hormonal intensification. Nothing in the reported data excludes that. The safety profile, by contrast, is genuinely reassuring: grade 3 or worse events at 8% versus 7%, acute kidney injury the most common in both arms at 2%, no treatment-related deaths, so the added toxicity cost of finding out is low. That asymmetry, real local signal and low toxicity against an unproven increment over available intensification, is the honest read.