Mark Buyyounouski of Stanford University will present the five-year results of NRG Oncology GU003 on September 27 at 1:40 pm ET in the Grand Ballroom, as the second presentation of the ASTRO 2026 Clinical Trials session. The analysis extends follow-up of the only randomized comparison of hypofractionated and conventionally fractionated radiotherapy to the prostate bed. Its interpretation will depend less on the results themselves than on the precision with which the trial's original design is remembered.
Design and primary findings
Between July 2017 and July 2018, NRG-GU003 screened 298 men and randomly assigned 296 following radical prostatectomy: 144 to hypofractionated postprostatectomy radiotherapy (HYPORT), 62.5 Gy in 25 fractions of 2.5 Gy, and 152 to conventionally fractionated treatment (COPORT), 66.6 Gy in 37 fractions of 1.8 Gy. Eligible patients had either an undetectable PSA below 0.1 ng/mL with margin-negative pT3 or margin-positive pT2 disease, or a detectable PSA of 0.1 ng/mL or higher with pT2 or pT3 disease. Randomization was stratified according to baseline Expanded Prostate Cancer Index Composite (EPIC) score across four tiers and according to androgen-deprivation use, which was permitted for up to six months. Nodal irradiation was prohibited.
The coprimary endpoints were the change from baseline in the EPIC genitourinary and gastrointestinal domain scores at 24 months, with noninferiority margins of minus 5 and minus 6 respectively, derived from NRG Oncology/RTOG 0415. The trial was powered for these endpoints alone.
Both were met. At 24 months, the mean genitourinary change score was minus 5.01 with HYPORT and minus 4.07 with COPORT, and the corresponding gastrointestinal scores were minus 4.17 and minus 1.41; both comparisons rejected the null hypothesis of inferiority. Gastrointestinal symptoms at the completion of radiotherapy were substantially worse in the hypofractionated group, minus 15.52 as against minus 7.06, but this difference had resolved by six months and did not recur at 12 or 24 months. The findings were presented at the 2021 ASTRO plenary session and published in JAMA Oncology in 2024, with a correction appearing in April of that year. NRG subsequently characterized the shorter schedule as an acceptable practice standard, and adoption followed.
The limits of what was demonstrated
Three features of the trial constrain the inferences that can be drawn from it, and each becomes more consequential as the schedule enters wider use.
First, the trial was not powered for disease control. At a median follow-up of 2.1 years among censored patients, the two-year rate of biochemical failure, defined as a PSA of 0.4 ng/mL or higher and rising, was 12 percent with HYPORT and 8 percent with COPORT (P = 0.28); local failure was 0.7 percent and 0.8 percent respectively. With 296 patients and follow-up of this duration, the absence of a statistically significant difference carries little information about whether a clinically meaningful difference exists. A nonsignificant comparison of two failure curves does not constitute a demonstration of oncologic noninferiority, and the investigators did not claim that it did.
Second, the numerical direction of the biochemical comparison favored the conventional schedule. This is unlikely to represent a true effect, but it is worth noting that it runs contrary to the radiobiologic rationale. At an assumed low alpha/beta ratio for prostate cancer, 62.5 Gy delivered in 2.5 Gy fractions is modestly dose-escalated relative to 66.6 Gy in 1.8 Gy fractions, and the original analysis accordingly tested whether the hypofractionated schedule was superior for disease control. That null hypothesis was not rejected.
Third, the surgical bed differs from the intact gland in ways that bear directly on the duration of follow-up required. The high-dose volume contains the vesicourethral anastomosis and a surgically altered bladder neck. Urethral stricture and progressive incontinence in this setting characteristically become manifest beyond 24 months. The acute toxicity observed in GU003 was gastrointestinal and transient; late genitourinary events are the more plausible expression of increased fraction size, and they are precisely the events that a two-year endpoint is not positioned to capture. Six patients in the hypofractionated group experienced grade 3 cystitis, of whom one reported urinary function as a major problem on EPIC, a discordance that illustrates why clinician-graded and patient-reported toxicity should be examined together rather than interchangeably.
What the five-year analysis should be examined for
The genitourinary trajectory between 24 and 60 months is the principal result. The relevant question is whether the two arms remain superimposed or begin to diverge, and whether cumulative clinician-graded grade 2 or higher genitourinary events are reported alongside the patient-reported data.
Attrition in patient-reported outcomes warrants equal scrutiny. EPIC compliance was complete at baseline and approximately 83 percent at two years. It will be lower at five. Loss to questionnaire follow-up is not random with respect to symptom burden, and the handling of missing data will determine how much confidence the late quality-of-life comparison can support.
The framing of the disease-control analysis will also merit attention. Five-year curves are considerably more informative than two-year curves, but the confidence intervals will remain wide. Comparable outcomes should be described as reassuring rather than confirmatory. If they are instead characterized as establishing noninferiority, the evidentiary standard will have been adjusted retrospectively to accommodate a practice change that has already occurred.
Finally, the trial enrolled both adjuvant and early salvage patients, populations with materially different failure rates. Whether outcomes are reported separately affects the transferability of the results to contemporary practice, in which adjuvant treatment has largely been supplanted by early salvage.
Generalizability
GU003 accrued in 2017 and 2018, before PSMA PET imaging entered routine restaging. A proportion of these patients would today be found to harbor disease beyond the prostate bed and would be managed differently.
The exclusion of nodal irradiation is a more substantial limitation. GU003 provides no evidence regarding hypofractionation of a pelvic nodal volume. This is directly relevant to the presentation immediately preceding it in the same session, since NRG Oncology/RTOG 0534 established a benefit for elective pelvic nodal irradiation in the salvage setting. The combination now in common use, a 25-fraction schedule delivered to the prostate bed and pelvic nodes together, has support for each component separately and none for the combination. That gap deserves explicit acknowledgment rather than tacit extrapolation.
Assessment
The adoption of a 25-fraction schedule on the basis of a two-year patient-reported endpoint was a defensible decision. The evidence for moderate hypofractionation in the intact prostate was substantial, the burden of a seven-week course is real, and the trial met its prespecified endpoints. The difficulty is one of sequence rather than of judgment: the confirmatory data are arriving after the practice change, and no mechanism exists for reversing that change should the longer follow-up prove less reassuring than the shorter.
Expected findings
We anticipate that five-year genitourinary and gastrointestinal patient-reported outcomes will remain clinically comparable between the two schedules, without late separation sufficient to alter the two-year conclusion. We anticipate that biochemical control will likewise remain comparable, and consider it plausible that the small numerical difference favoring conventional fractionation will persist without approaching statistical significance.
A late divergence would be the finding of consequence: a sustained separation in biochemical control, or an accumulating severe genitourinary signal after 24 months. Either would reopen a question that most clinicians now regard as settled, and either would bear directly on the design of the five-fraction prostate bed regimens currently under prospective investigation.
We will report on the presentation from ASTRO 2026.
This is the second article in the RadOncToday series previewing the Clinical Trials and Plenary sessions at ASTRO 2026 over the three weeks before the meeting.