MROQC ADT Practice Patterns
ForIntact high-risk M0/N0-1 prostate ca planned for definitive RT
TL;DR67.0% of 553 high-risk pts got guideline-concordant ADT (≥18mo); ARPI use 23.2% among STAMPEDE-eligible post-publication.
The deviation is concentrated where the guideline is weakest-anchored: cN1 (OR 2.94) and GG4-5 (OR 6.23 / 9.45) drove ≥18mo recommendations, so the third of pts NOT getting guideline-concordant ADT are largely single-factor high-risk. Facility remained a predictor after case-mix adjustment (P<.0001), which points at practice culture, not patient selection.
For a man with single-factor high-risk localized disease (GG4-5 or PSA ≥20 alone) heading to definitive RT, this frames how much of the ≥18mo ADT recommendation is guideline versus local habit; it does not address post-prostatectomy salvage or M1 disease.
GC-ADT tracked the features that also define STAMPEDE M0 eligibility (GG4 OR 6.23, GG5 OR 9.45, cN1 OR 2.94), so the non-concordant third is enriched for single-factor high-risk disease. Facility stayed predictive after case-mix adjustment (P<.0001), making department habit, not patient selection, the target.
ARPI intensification reached only 23.2% of STAMPEDE M0-eligible pts after publication, up from 0%. With 27.9% of this high-risk RT population eligible, the referral and co-management pathway for adding an ARPI to definitive RT plus ADT is the bottleneck, not the evidence.
9 details
Prospective observational analysis within the Michigan Radiation Oncology Quality Consortium (MROQC), a statewide quality collaborative. 26 centers, accrual June 2020 to November 2024, N=553. Intended ADT duration and ARPI use were collected prospectively at the treatment decision, not abstracted retrospectively.
Intact (non-postoperative) high-risk M0/N0-1 prostate cancer planned for definitive radiotherapy. Risk features: cT3/4 13.3%, cN1 19.9%, GG 4-5 75.0%, PSA ≥20 ng/mL 40.0%.
Primary outcome was intended guideline-concordant ADT (GC-ADT, ≥18 months) per the 2022 AUA/ASTRO recommendation of 18-36 months. Secondary analyses covered multivariable predictors of GC-ADT, ARPI adoption before versus after STAMPEDE M0 publication, and facility-level variability via a mixed-effects model with treatment site as random intercept.
91.3% were recommended ADT and 67.0% were guideline-concordant. Among the 27.9% meeting STAMPEDE M0 eligibility, ARPI recommendation went from 0% pre-publication to 23.2% after. Site-level variability remained significant on multivariable analysis (P<.0001).
| Factor | OR (95% CI) |
|---|---|
| cN1 | 2.94 (1.44 to 5.99) |
| GG4 | 6.23 (2.85 to 13.62) |
| GG5 | 9.45 (4.46 to 20.06) |
| PSA ≥40 | 3.64 (1.22 to 10.87) |
The 2022 AUA/ASTRO guideline sets 18-36 months, and STAMPEDE M0 supports ARPI intensification for ≥2 of cT3/T4, GG 4-5, PSA ≥40, or cN1. Both benchmarks are met by a minority here, echoing the long-standing gap between the long-course ADT durations tested in EORTC 22863 and RTOG 9202 era trials and what is actually intended in practice.
Intended duration is a proxy for delivered duration, so the true concordance rate could fall further with early discontinuation. STAMPEDE M0 eligibility was applied to a cohort accrued partly before that trial reported, so the 23.2% post-publication figure reflects an adoption curve still in motion rather than a steady state.
The finding that facility explains variance after adjusting for cN1, grade group, and PSA is the load-bearing result: with case mix accounted for, where a man is treated still moves how long he is recommended ADT. That is a quality-improvement target rather than an evidence gap, and it is the kind of signal a consortium is uniquely built to detect and act on.
Prospective practice-pattern survey of intended treatment, no efficacy endpoint. Documents a care-delivery gap rather than testing whether the guideline duration is right.
- Does intended ADT duration match delivered duration?
- Which facility-level factors drive the residual variability?
- Optimal ADT duration for single-factor high-risk disease
📚 Sources · 📄 1 paper
Abstract
The longer read
The useful way to read this is not as a scolding about guideline adherence but as a measurement of where the guideline's authority is actually thin. The 2022 AUA/ASTRO recommendation of 18-36 months treats high-risk localized disease as one bucket, but the odds ratios here show clinicians are not treating it that way: GG5 carried an OR of 9.45 and GG4 an OR of 6.23 for a concordant recommendation, with cN1 at 2.94 and PSA ≥40 at 3.64. In other words, when a man carries the features that also happen to define STAMPEDE M0 eligibility, long-course ADT is recommended nearly reflexively. The third of the cohort not receiving a concordant recommendation is therefore not a random third; it is enriched for single-factor high-risk disease, where the marginal benefit of 18 months over a shorter course has never been isolated in a trial designed to ask that question. Given that 75.0% of this cohort was GG 4-5 and only 13.3% cT3/4, the non-concordant group is likely dominated by men whose only high-risk feature is grade or PSA.
The facility effect is the more uncomfortable result. A mixed-effects model with site as a random intercept still found significant variability at P<.0001 after multivariable adjustment, which means the case-mix explanation is not available. Two men with the same stage, grade, and PSA get different ADT recommendations depending on which of the 26 centers they walk into. That is the classic signature of a quality gap rather than an evidence gap, and it is precisely what a consortium structure exists to surface. It also complicates any attempt to read the 67.0% figure as clinical nuance: genuine individualization would be expected to track patient factors, comorbidity, and preference, not the building.
The ARPI numbers deserve more caution than the headline suggests. Moving from 0% to 23.2% among STAMPEDE M0-eligible patients is a real change in behavior, but the denominator is small (27.9% of 553) and the observation window straddles publication, so the post-publication rate is a point on an adoption curve, not an equilibrium. Comparing it against ADT concordance is somewhat unfair: eighteen months of ADT has been the recommendation for years, while ARPI intensification in M0 disease was new, expensive, and adds a distinct toxicity burden to men who are already going to spend a year and a half hypogonadal. A 23.2% uptake within months of publication is not obviously slow.
The main thing this study cannot tell you is whether the guideline-concordant two thirds did better. Intent was captured prospectively, but no oncologic endpoint is reported, and intended duration is a decision, not a delivered exposure. Real-world discontinuation for cardiometabolic toxicity, fatigue, or patient preference means actual exposure is almost certainly shorter than intended in some fraction of the concordant group, which would push the true concordance rate below 67.0%. For a radiation oncologist, the actionable read is narrow but genuine: the variation lives in your department's habits as much as in your patients, and the single-factor high-risk patient is where your recommendation is most likely to diverge from your colleague's down the hall.