ProtecT (cribriform morphology secondary analysis)
ForPSA-screened clinically localized prostate cancer, cribriform status on diagnostic biopsy
TL;DRSecondary analysis: cribriform-negative disease (87% of reviewed cohort) got no significant 15-yr metastasis reduction from early radical treatment vs monitoring.
Reported via UroToday →
For the RT reader this is a selection question, not a technique one: it asks whether cribriform status identifies the ProtecT subgroup that actually earned the metastasis reduction from radical treatment. RT here was EBRT with 3-6mo neoadjuvant ADT, a dated backbone. No subgroup HRs, CIs or event counts appear in the source excerpt.
In PSA-screened clinically localized disease with cribriform-negative grade group 2 on biopsy, this analysis supports treating long-term metastasis risk as comparable to grade group 1; it says nothing about cribriform-positive pts, where guidelines already discourage surveillance.
The decision this moves is whether to offer definitive RT at all, not how to deliver it: cribriform-negative grade group 2 showed metastasis risk equivalent to grade group 1 at 15 years. The RT arm was EBRT with 3-6mo neoadjuvant ADT in a pre-MRI, systematic-biopsy cohort, so transferability to current practice is limited. No subgroup effect sizes in source.
Relevant to how neoadjuvant ADT exposure is weighed against a metastasis benefit that this analysis does not find in the 87% cribriform-negative majority. ProtecT's ADT was 3 to 6 months alongside EBRT. The result argues for a biomarker-gated rather than grade-group-gated treatment decision.
Bears on the counselling conversation before radical prostatectomy in grade group 2: cribriform-negative pts carried the same 15-year metastasis risk as grade group 1, supporting surveillance discussion in that stratum. Cribriform-positive disease is untouched by this read, and guidelines already discourage surveillance there.
9 details
Secondary pathological analysis of the ProtecT randomized trial (active monitoring vs radical prostatectomy vs EBRT with neoadjuvant ADT). Central review of biopsy slides retrieved for 712 of 1,643 randomized pts, with both intention-to-treat and per-protocol analyses, the latter accounting for crossover to radical treatment.
PSA-screened men with clinically localized prostate cancer, the original ProtecT eligibility. Slides were regraded under 2019 ISUP criteria rather than the trial-era 2005 criteria; the presenter reports regrading did not change the picture. Age, PSA and revised Gleason score did not differ significantly across arms in the reviewed subset.
The RT arm was external-beam radiotherapy with 3 to 6 months of neoadjuvant ADT. No dose, fractionation or target volume is given in the source. The presenter notes cribriform-positive disease was more prevalent in this arm, attributed to chance since cribriform status was not a randomization characteristic.
Primary: metastases at 15-year median follow-up, analysed by cribriform status. Multivariable Cox modelling was used to compare metastasis risk across grade groups within the cribriform-negative stratum.
Roughly 13% of the reviewed cohort was cribriform-positive. Among the 87% cribriform-negative, early radical treatment produced no statistically significant reduction in 15-year metastases vs active monitoring, in both ITT and per-protocol analyses. Cribriform-negative grade group 2 carried metastasis risk equivalent to grade group 1.
Slides were available for under half the randomized cohort, so the analysis rests on whichever centres have so far returned material. The source excerpt is a video interview that stops mid-presentation and reports no HRs, CIs or event counts, so the size of the null in the cribriform-negative group cannot be judged from it.
The interest here is the inverse of the usual cribriform argument. Rather than showing cribriform-positive pts do badly on surveillance, the useful signal is that cribriform-negative grade group 2 behaves like grade group 1 over 15 years, which is a de-escalation read. Whether the cribriform-positive stratum shows the reciprocal benefit is not in the source excerpt.
Post-hoc subgroup of a non-stratified variable in 712 of 1,643 pts, with a non-significant negative result and no effect sizes in the source.
- Metastasis outcomes in the cribriform-positive stratum on active monitoring
- Whether cribriform status holds as a marker in MRI-targeted biopsy cohorts
- Reproducibility of binary cribriform calls across pathologists
📚 Sources · 📄 1 paper
Abstract
The longer read
ProtecT's 15-year result left the field with an awkward asymmetry: prostate-cancer-specific mortality was low and indistinguishable across active monitoring, surgery and radiotherapy, yet early radical treatment reduced metastases. That combination only makes sense if the metastasis benefit is concentrated in some identifiable minority rather than spread evenly across a screen-detected cohort. This analysis takes the most guideline-ready candidate marker, binary cribriform status, and asks whether it marks that minority. The framing is the right one, and it is worth noting that no prospective randomized cohort had previously been used to test the cribriform question in men actually managed with monitoring, which is precisely the population the EAU position affects.
The result reported in the source is the cribriform-negative half of the argument, and it is a null: no statistically significant reduction in 15-year metastases with early radical treatment in the 87% of reviewed pts without cribriform morphology, holding in both intention-to-treat and per-protocol analyses. The per-protocol concordance matters more than it might appear, because ProtecT's crossover from monitoring to radical treatment was substantial and any null in ITT alone would be open to a dilution objection. That the two analyses agree removes the easiest way to dismiss the finding.
The accompanying multivariable Cox result is the more provocative claim: cribriform-negative grade group 2 carried the same long-term metastasis risk as grade group 1. If that survives full publication with effect estimates attached, it argues that grade group 2 is not a single entity for counselling purposes, and that the cribriform-negative half of it belongs with Gleason 6 rather than with the intermediate-risk category that currently pushes men toward treatment. That is a de-escalation argument reached from a treatment trial, which is unusual and worth taking seriously.
Several things should hold confidence back. Cribriform status was not a randomization characteristic, so this is a post-hoc stratification of a randomized trial, and the presenter concedes a prevalence imbalance favouring the radiotherapy arm that he attributes to chance. Slides were retrievable for 712 of the 1,643 randomized men, so the analysis rests on a subset defined by which centres have so far returned material, a selection mechanism unlikely to be related to outcome but not demonstrably unrelated to it either. The reassurance offered, that age, PSA and revised Gleason did not differ across arms in the reviewed subset, addresses balance between arms rather than representativeness of the whole trial. And the source here reports no hazard ratios, confidence intervals or event counts, so the width of this null cannot be assessed: with metastasis events sparse in a screen-detected cohort split by a 13% marker, a wide interval consistent with meaningful benefit is entirely plausible.
For a radiation oncologist the practical caution is generational. The ProtecT radiotherapy arm was external-beam radiotherapy with three to six months of neoadjuvant ADT, delivered to men diagnosed by systematic biopsy without MRI, and graded originally under 2005 criteria. Contemporary practice differs on every one of those axes, and the men who would be offered radiotherapy today after MRI-targeted biopsy are not the same men. What the analysis would change, if confirmed with numbers, is not how radiotherapy is delivered but who is offered it at all, and the direction is toward offering it to fewer.