NRG-GU005 (quality of life)
ForLocalized intermediate-risk prostate cancer, median age 68, no ADT specified
Bowel 33% vs 46% at 1 yr
p=0.002; 2yr bowel/UIO primary PRO endpoint not reported in source
TL;DRFewer MCID declines with SBRT at 1yr bowel (33% vs 46%, p=0.002) and sexual (34% vs 44%, p=0.026).
Reported via UroToday →
The QoL separation is domain-specific, not global: bowel and sexual at 1yr, urinary incontinence at 2yr, with no longitudinal effect in sexual or hormonal. Rectal manipulation (SpaceOAR 55%) and the 38.78 Gy PTV max cap gate transfer, since the bowel and GU signals came from a spacer-heavy, urethra-constrained delivery.
In localized intermediate-risk prostate cancer choosing between 5-fraction SBRT and moderate hypofractionation, this supports SBRT on patient-reported bowel, sexual, and continence grounds; it does not speak to high-risk disease, nodal coverage, or oncologic non-inferiority, which the trial's primary endpoint carries.
Domain-specific, not global: bowel and sexual at 1yr, incontinence at 2yr, with no longitudinal sexual or hormonal effect. The 38.78 Gy PTV max cap and 55% SpaceOAR use gate transfer, since the favorable GU and bowel profile came from a urethra-constrained, spacer-heavy delivery, not from five fractions alone.
Also covered Aug 14
11 details
Randomized, non-blinded phase III, 1:1, N=698 (MH-IMRT 345, SBRT 353), stratified by Gleason score, PSA, and rectal manipulation. The trial's oncologic primary endpoint sits elsewhere; this analysis reports the patient-reported secondary endpoints.
Localized intermediate-risk prostate cancer, median age 68 (IMRT) and 69 (SBRT). Two patients were ineligible (one high-risk, one PSA out of window). Baseline EPIC domains were balanced across arms, all p≥0.093.
SBRT 36.25 Gy in 5 fractions delivered 2-3 per week, PTV expansion 5mm except 3mm posteriorly and anteriorly, PTV max capped at 38.78 Gy unless the urethra was visualized and contoured (acceptable variation 43.5 Gy). MH-IMRT 70 Gy/28 fx or 60 Gy/20 fx, PTV expansion 8mm except 5mm posteriorly. CTV was prostate ± 1cm proximal seminal vesicles. Rectal manipulation was common: SpaceOAR in 55% overall.
EPIC-26 at baseline, 12 and 24 months. MCID thresholds: >5 points urinary irritative/obstructive, >6 urinary incontinence, >10 sexual, >4 bowel and hormonal. Individual MCID rather than group mean scores was the analytic unit, with an exploratory longitudinal linear model adjusted for baseline score, arm, stratification factors, T-stage, age, and race.
The arm-level MCID and toxicity comparisons are tabulated above. Longitudinal modeling of urinary incontinence gave a least square mean difference of 2.91 (95% CI 0.85-4.97, p=0.0058) favoring SBRT, while sexual and hormonal domains showed no significant treatment effect.
| Domain / timepoint | SBRT | MH-IMRT | p |
|---|---|---|---|
| Bowel, 1 yr | 33% | 46% | 0.002 |
| Sexual, 1 yr | 34% | 44% | 0.026 |
| Urinary incontinence, 2 yr | 26% | 35% | 0.023 |
| Event | SBRT | MH-IMRT | p |
|---|---|---|---|
| Treatment-related G≥3 GU | 0.6% | 2.5% | 0.04 |
| Rectal hemorrhage, any grade | 10.5% | 17.3% | 0.01 |
| Fatigue, any grade | 39.2% | 50.8% | 0.0025 |
Investigator-reported toxicity favored SBRT across the board: grade ≥3 GU 0.6% vs 2.5% (p=0.04), any-grade rectal hemorrhage 10.5% vs 17.3% (p=0.01), any-grade fatigue 39.2% vs 50.8% (p=0.0025). The GU finding is the one that most often runs the other way in ultrahypofractionation series, so it is worth reading as the trial's own answer to the pre-trial toxicity concern.
PACE-B is the other large randomized SBRT vs moderate hypofractionation comparison, and both trials show lower urinary incontinence decline with SBRT despite differing MCID definitions. Prior patient-level meta-analysis had suggested less clinically meaningful bowel and urinary irritative/obstructive decline at two years with SBRT, and the 1-year bowel result here is directionally consistent.
Differential attrition ran against the IMRT arm: 22 IMRT patients (6.4%) died or withdrew before year 1 versus 4 SBRT patients (1.1%), and 22 IMRT patients never received the assigned RT after withdrawal versus 1 in the SBRT arm. Completion was 79.9% (IMRT) and 84.0% (SBRT) at year 1. The domain-by-timepoint pattern (bowel and sexual at 1yr, incontinence at 2yr) is the kind of scattered significance that multiplicity should temper.
The trial was designed when the open question was whether five fractions cost the patient something. The PRO answer is that it does not, and the investigator-reported toxicity answer is that it may cost less. What the QoL data cannot do is settle whether SBRT is oncologically non-inferior, which is the primary endpoint and is not reported in this source.
CONSORT flow
Prespecified PRO secondary analysis of a phase III trial, non-blinded with patient-reported endpoints; aligns with PACE-B rather than establishing a new position. Oncologic primary endpoint not reported here.
- Oncologic non-inferiority of SBRT vs MH-IMRT in this trial
- Whether bowel benefit holds without rectal spacer
- Durability of QoL separation beyond 2 years
📚 Sources · 📄 1 paper
Abstract
The longer read
The framing that makes this analysis useful is the one Dr. Yu opened with: at the time NRG-GU005 was designed, moderate hypofractionation was the settled standard after RTOG 0415 and the concern about five fractions was toxicity, not efficacy. The trial was therefore built to answer a question that has partly answered itself in the interim, since PACE-B reported first and much of the field has already moved. What this dataset adds is a second independent, randomized, North-American-cooperative-group answer to the same toxicity worry, which matters more than a confirmatory analysis usually does because the pre-trial prior ran the other way.
The direction of the result is the striking part. Ultrahypofractionation was expected, if anything, to cost urinary function, and the acute GU signal in earlier SBRT series was the specific reason for caution. Here the GU comparison runs in SBRT's favor on both the patient-reported and the investigator-reported channel: fewer incontinence MCID declines at two years, a positive longitudinal treatment effect on that domain, and a lower rate of treatment-related grade 3 or higher GU events. Convergence between two different measurement systems is the strongest internal evidence in the analysis, because they fail in different ways. A non-blinded PRO endpoint is vulnerable to a patient who knows they finished in two weeks feeling better about their outcome; investigator-graded rectal hemorrhage and grade 3 GU events are not vulnerable in the same way.
That non-blinding is still the reason to hold the domain-level results loosely rather than the overall read. Significance lands on bowel and sexual at one year and on incontinence at two, and it does not land on sexual or hormonal in the longitudinal model. A scattered pattern across five domains and two timepoints is what multiplicity produces when a real but modest effect is measured many ways, and the honest reading is that the direction is consistent rather than that any single domain-timepoint cell is established. The incontinence result is the exception, because it is the one supported by both the MCID analysis and the longitudinal model, and it is also the one PACE-B independently reproduced.
Two delivery details decide whether this transfers. Rectal manipulation was used in roughly three in five patients, dominated by SpaceOAR at 55%, so the bowel comparison is a spacer-heavy one and a practice that does not routinely place a spacer should not assume the same bowel separation. The SBRT protocol also capped PTV maximum dose at 38.78 Gy unless the urethra was visualized and contoured, which is a real constraint rather than a formality, and it is plausible that the favorable GU profile is partly a product of that constraint rather than of five fractions as such. A center delivering SBRT without urethral awareness is not delivering the arm that produced this result.
The attrition asymmetry is worth naming without overreading. Twenty-two IMRT patients withdrew before receiving radiotherapy against one in the SBRT arm, and early death or withdrawal before the year-1 assessment was 6.4% versus 1.1%. Patients who decline a 28-fraction course after randomization are unlikely to be a random sample, and the direction of that bias is not obvious. Finally, none of this speaks to the oncologic question. The QoL analysis is a secondary endpoint of a trial whose primary endpoint is disease control, and a favorable PRO profile is only actionable once the efficacy comparison holds.