onc brain

The longer read

The headline everyone will take from this analysis is the antagonist-versus-agonist gap, and it is the least defensible number in it. The comparison is crude as described: the presenter calls the reduction "numerically lower" and gives no confidence interval, adjusted hazard ratio or propensity approach, while noting that agonist recipients were older, more often on Medicare and followed for longer. Each of those imbalances pushes the agonist group toward more observed events independent of drug class. Older age raises baseline cardiovascular risk directly. Medicare enrollment tracks age and comorbidity. Longer follow-up, when the outcome is expressed per person-time, should in principle be absorbed by the denominator, but it also places more agonist exposure in earlier calendar years, before the awareness shift the authors themselves describe after 2021. A 37% crude difference produced under those conditions could shrink a great deal once adjusted, and the source gives no way to know.

The antagonist group also is not the drug that generated the original hypothesis. HERO, which the presenter cites as the backdrop, concerned relugolix. Here roughly two-thirds of antagonist exposure was degarelix and only about a third relugolix, and the data stop in May 2024, before the relugolix uptake the presenter acknowledges. The analysis therefore cannot speak to relugolix specifically, and pooling two antagonists with different routes and pharmacology into one class label assumes a class effect that the interviewer explicitly declined to adjudicate. Read honestly, this is consistent with HERO's direction without adding independent evidence for it.

The finding with more practical weight is the preexisting-cardiovascular-disease stratum: about 11 MACE per 100 person-years against roughly four overall. That contrast is less exposed to the confounding that undermines the class comparison, because it describes absolute risk in an identifiable group rather than attributing an effect to a treatment choice. It moves a decision a clinician can act on regardless of which GnRH agent is used: who should have cardiology or primary care involvement when ADT starts. This is where PRONOUNCE enters the argument as the presenter frames it, as evidence that cardiologist involvement matters, and where the forthcoming PCF consensus statement on risk-tiered referral is aimed.

Two further cautions belong with the result. First, MACE is an end-stage composite. The interviewer's point that it misses diabetes, metabolic syndrome, atherosclerosis and renal dysfunction means event rates here understate the cardiometabolic cost of ADT, especially across the multi-year exposure now typical as survival lengthens. Second, the modest post-2021 decline is a calendar-time trend in a claims dataset. Changes in coding, case mix, statin use or general cardiovascular care could produce it as easily as awareness generated by HERO and PRONOUNCE.

For this to change ADT class selection, it would need an adjusted, relugolix-specific estimate with a modern data window and an interval excluding no effect. Until then it supports a narrower claim: men with established cardiovascular disease starting ADT carry roughly threefold MACE risk, and that risk warrants co-management whatever the agent.