MROQC ADT Practice Patterns
ForIntact high-risk M0, N0-1 prostate cancer, definitive RT
TL;DRGuideline-concordant ≥18mo ADT intended in 67.0% of high-risk M0/N0-1 pts receiving definitive RT; ARPI recommended in 23.2% of STAMPEDE-eligible pts post-publication.
Facility-level variability in ≥18mo ADT recommendations persisted after adjusting for cN1, grade group and PSA (P<.0001), so treating site, not only tumor features, shapes duration. Among STAMPEDE M0-eligible pts, ARPI recommendations reached only 23.2% after publication, a gap a department can audit directly in its own high-risk RT cohort.
Grade group drove concordant ADT recommendations most (GG5 OR 9.45, GG4 OR 6.23), and site-level variability survived adjustment (P<.0001). For an RT department the benchmark is concrete: 67.0% intended ≥18mo ADT and 23.2% ARPI uptake in STAMPEDE M0-eligible pts after publication.
9 details
Prospective statewide quality-consortium data (MROQC), N=553 across 26 centers, June 2020 to November 2024. Multivariable analysis of predictors, a mixed-effects model with treatment site as random intercept, and ARPI adoption compared before vs after STAMPEDE M0 publication.
Intact high-risk M0, N0-1 prostate cancer treated with definitive RT: cT3/4 13.3%, cN1 19.9%, GG 4-5 75.0%, PSA ≥20 ng/mL 40.0%. 27.9% met STAMPEDE M0 criteria (≥2 of cT3/T4, GG 4-5, PSA ≥40 ng/mL, or cN1).
Primary: intended guideline-concordant ADT (GC-ADT, ≥18 months) per 2022 AUA/ASTRO. Also assessed: ARPI recommendation in STAMPEDE-eligible pts and facility-level variability.
91.3% recommended any ADT, 67.0% guideline-concordant. ARPI recommendations in STAMPEDE-eligible pts rose from 0% to 23.2% after publication; facility variability persisted on MVA (P<.0001).
| Factor | OR (95% CI) |
|---|---|
| cN1 | 2.94 (1.44-5.99) |
| GG4 | 6.23 (2.85-13.62) |
| GG5 | 9.45 (4.46-20.06) |
| PSA ≥40 ng/mL | 3.64 (1.22-10.87) |
Benchmarks practice against the 2022 AUA/ASTRO 18-36 month recommendation and STAMPEDE M0 ARPI intensification criteria. Single-state data, and the source offers no comparison with other practice settings.
The abstract gives no reasons for sub-18mo recommendations (comorbidity, pt preference, toxicity), so non-concordance is not necessarily poor care. The ARPI comparison is an uncontrolled before-after design, and no OR is reported for cT3/4.
The strongest predictors of a concordant recommendation were grade group (GG5 OR 9.45) and cN1, consistent with shorter courses being offered when high-risk status rests on other features. Residual site variability points to local norms beyond the modelled case mix, though unmeasured covariates could contribute.
Practice-patterns study of intended ADT recommendations; no efficacy or outcome endpoint, so nothing here confirms or challenges SOC.
- Delivered vs intended ADT duration in routine definitive RT practice
- Drivers of facility-level variation in ADT duration recommendations
- Barriers to ARPI intensification in STAMPEDE M0-eligible pts
📚 Sources · 📄 1 paper
Abstract
The longer read
The headline figure, two thirds of high-risk pts recommended at least 18 months of ADT with definitive RT, reads easily as a quality failure. The abstract supports that reading less than it first appears to. The endpoint is intended duration at the time of RT, a recommendation rather than a delivered course, and the reasons behind a shorter recommendation are not reported. Long-course ADT in older men carries real cardiometabolic, skeletal and fatigue costs, and part of the non-concordance in any registry will be deliberate: comorbidity, competing mortality, or a pt who declines after an informed discussion. A dataset that cannot separate those decisions from habit measures the upper bound of the quality gap, not its true size.
The multivariable pattern tells you more. Grade group 5 (OR 9.45), grade group 4 (OR 6.23), PSA ≥40 ng/mL (OR 3.64) and cN1 (OR 2.94) were each tied to a concordant recommendation. The likely corollary is that pts whose high-risk status rests on PSA below the ≥40 threshold or on T-stage, with lower grade, are the ones most often offered shorter courses. That matches how many radiation oncologists treat high-risk as a mixed category in which grade dominates, while the guideline benchmark used here makes no such distinction. The abstract reports no estimate for cT3/4, and the reference categories are not stated, so how much T-stage alone moves the recommendation cannot be judged from the source.
The persisting facility-level variability (P<.0001) is the finding with the most operational weight. A mixed-effects model with site as a random intercept suits the question, and the result says that where a pt is treated predicts the recommendation beyond the case-mix variables in the model. The usual caveat applies: unmeasured case mix, such as comorbidity burden or referral patterns, can look like a site effect, and the abstract does not list the full covariate set.
The ARPI result is the sharper practice signal. STAMPEDE M0 defined an intensification population, and 27.9% of this cohort met its criteria. Recommendations rose from 0% before publication to 23.2% after. A zero baseline is expected because the data did not yet exist, so the comparison documents adoption rather than departure from an established standard. Uptake below a quarter could reflect access and cost, reluctance to extrapolate from a trial enrolled under older staging, or uncertainty about how modern imaging interacts with the eligibility criteria. The source tests none of these explanations.
What the study leaves open is whether any of this changes outcomes. It does not follow pts, report delivered duration, or link concordance to biochemical or metastatic endpoints. Its use is as a benchmark: a department can set its own high-risk RT cohort against 67.0% concordance and 23.2% eligible ARPI uptake, then ask whether its non-concordant cases are documented decisions or defaults.