onc brain

The longer read

The question A-DREAM asks is not whether treatment can be interrupted, but whether the modern ARPI era has changed the answer. Intermittent androgen deprivation was tested extensively in the pre-ARPI period and the metastatic result was never reassuring: the large intermittent-versus-continuous trials in metastatic disease could not rule out a survival decrement, and the field largely retreated to continuous therapy in metastatic disease while accepting intermittent approaches in biochemical relapse. What has changed since is the depth of response. A patient sustaining PSA under 0.2 for 18 to 24 months on ADT plus an ARPI is a different biological entity from the 1990s patient whose PSA nadired at 2 on ADT alone, and the entire premise here is that the depth of response, not the interruption itself, selects who is safe.

That premise is exactly what the single-arm design cannot test. The 41% primary result tells you what fraction of a highly selected cohort achieved a composite of two independent things, and the composite hides the more useful information. Disease control off treatment was reasonably common: 57.7% remained treatment-free for the full 18 months. Testosterone recovery was the tighter constraint, reached by 66.7% with a median of 9.0 months, which in a cohort with median age 70 and two or more years of continuous castration is not surprising. The clinical implication is that the two components should be counseled separately. A man deciding whether to interrupt is weighing a plausible chance of a year or more off therapy against a real chance that his testosterone simply does not come back, in which case he has accepted the disease risk of interruption and received none of the quality-of-life return. Nothing in these data predicts which man he is.

On safety, the honest reading is that 26.9 months of median follow-up is too early for the failure mode that matters. The reassuring surface numbers are 15 rPFS events and 4 deaths out of 78, with medians not reached and no comparator to measure them against. The buried signal is more informative: of the 29 patients who resumed ADT plus their original ARPI after meeting protocol criteria, 4 subsequently progressed and came off that ARPI. That is a small number, but it is the mechanism by which an interruption strategy could cost a patient something durable. The implicit safety argument for interruption is that resumption restores control; in four patients it did not, and the follow-up is short enough that the denominator of resumed patients is still accruing risk. A five-year update with a matched continuous-therapy comparison would move confidence far more than a longer single-arm curve will.

For a radiation oncologist the cohort composition is worth noticing even though radiotherapy was not a study variable. Half had prostate radiation as local therapy and nearly a third had radiation to metastatic sites, so the population that proved able to interrupt was one in which local and metastasis-directed treatment had already been used liberally. Whether that history contributed to the low disease burden that permitted interruption, or merely reflects the practice patterns of the enrolling centers, is unanswerable from a single arm. It does, however, mark the obvious next design: metastasis-directed therapy as a deliberate tool to extend time off systemic treatment is a randomizable question, and this cohort describes the population in which to ask it. Until then, A-DREAM supports a conversation, not a protocol change, and the conversation should be about testosterone as much as about cancer.