PROSTOX Ultra (SCIMITAR/EXCALIBUR)
ForPost-RP pts receiving 5-fx prostate bed SBRT ± pelvic nodes
TL;DRContinuous AUC 0.814 (SCIMITAR) and 0.824 (EXCALIBUR) for late G2+ GU toxicity after post-RP prostate bed SBRT; recalibrated 0.25 threshold gives pooled AUC 0.748.
Reported via UroToday →
The intact-prostate 0.50 cutoff failed post-op: in EXCALIBUR every pt with toxicity scored below it, while continuous ranking held (AUC 0.814 and 0.824). Clinical factors (pad use, IPSS, time from RP) predicted nothing, so a germline score at 0.25 is the only stratifier here for counseling or planning before prostate bed SBRT.
In a post-prostatectomy pt being considered for 5-fraction prostate bed SBRT, this supports reading a PROSTOX Ultra score against a 0.25 post-op threshold rather than the intact 0.50; it does not inform conventionally fractionated postop RT, which was not studied.
The intact-prostate 0.50 cutoff failed post-op: in EXCALIBUR every pt with toxicity scored below it, while continuous ranking held (AUC 0.814 and 0.824). With clinical factors predicting nothing, a germline score at 0.25 is the only stratifier here for counseling or fractionation choice before prostate bed SBRT.
10 details 1 trial watching
Biomarker evaluation in two prospective post-op SBRT cohorts, SCIMITAR (NCT03541850) and EXCALIBUR (NCT04915508). Continuous and dichotomized discrimination compared; exploratory threshold recalibration targeting sensitivity >=0.65 and specificity >=0.70; AUCs bootstrapped and pooled by fixed-effects meta-analysis.
SCIMITAR enrolled 100 post-prostatectomy pts, 89 evaluable; EXCALIBUR 74 evaluable. Germline DNA from saliva.
SCIMITAR: 30-34 Gy in 5 fractions ± pelvic nodes. EXCALIBUR: 5-fraction SBRT to the prostate bed ± nodes.
Late grade 2+ GU adverse events, with discrimination by continuous AUC and binary AUC at the 0.50 intact threshold and a recalibrated post-op threshold.
Risk ordering preserved: scores higher with toxicity in both cohorts. At 0.50, EXCALIBUR binary AUC <0.50 (threshold misalignment); at 0.25, pooled AUC 0.748. Candidate thresholds spanned ~0.10-0.31.
| Metric | SCIMITAR | EXCALIBUR |
|---|---|---|
| Continuous AUC | 0.814 | 0.824 |
| Binary AUC at 0.50 (intact) | 0.755 | <0.50 |
| Specificity at 0.50 | 0.985 | 0.926 |
| Sensitivity at 0.50 | 0.546 | not reported |
| Binary AUC at 0.25 (post-op) | 0.750 | 0.727 |
| Median score, tox vs no tox | 0.547 vs 0.049, p<0.0001 | 0.338 vs 0.043, p=0.009 |
The intact-prostate PROSTOX Ultra model used a prespecified 0.50 threshold after intact SBRT; post-RP performance was previously unknown. Here the genetic signal transfers but the clinical trigger shifts downward post-op.
The 0.25 cutoff was selected on the same two cohorts it is reported in, so the 0.748 pooled AUC is optimistic until tested externally. EXCALIBUR toxicity rate and nodal-field proportions are not reported in source, limiting judgment of whether field size confounds the signal.
Supports mirSNP germline risk as a continuous predictor that generalizes post-RP, needing context-specific calibration rather than a new model. Whether a high score should change dose, field, or modality is not tested.
Biomarker validation in two small cohorts (89 and 74 evaluable); post-op threshold derived in-sample by exploratory recalibration, untested in an independent cohort or for risk-adapted planning.
- Does the 0.25 post-op threshold hold in an independent cohort?
- Does risk-adapted fractionation lower GU toxicity in high-score pts? active Germline DNA-Based Radiosensitivity Biomarker Influence on Toxicity Following Prostate Radiotherapy, GARUDA Trial Phase NAn=208 · primary completion 2027-12 · germline score guides SBRT vs hypofx; GU tox, intact gland
- Is the signal confounded by pelvic nodal field inclusion?
📚 Sources · 📄 1 paper
Abstract
The longer read
The most useful finding here is not the pooled AUC but the dissociation between continuous and dichotomized performance. A signature trained on intact-prostate SBRT ranked post-prostatectomy pts by GU toxicity risk about as well as one could hope from a germline test (continuous AUC 0.814 and 0.824), yet the threshold carried over from the intact setting produced a binary AUC below 0.50 in EXCALIBUR, with specificity high and sensitivity poor. That pattern is what one expects when the absolute baseline risk shifts between settings while the relative ordering of pts by susceptibility stays intact. Post-RP bladder neck and urethral anastomosis tissue sits in the high-dose region of a prostate bed volume, and the reported median scores in pts with toxicity (0.338 and 0.547) sit well below or barely above the old 0.50 trigger. The biology appears portable; the decision line is not.
That distinction should shape how a radiation oncologist reads the 0.25 threshold. It was chosen by exploratory recalibration within these same two cohorts, from a candidate range of roughly 0.10 to 0.31, and the dichotomized AUCs at 0.25 (0.727 and 0.750, pooled 0.748) are therefore in-sample estimates. Small cohorts with a modest event count, at least in SCIMITAR where 21 of 89 developed late G2+ GU toxicity, give a recalibrated cutoff room to fit noise. The continuous AUCs are the more defensible numbers because they did not depend on any choice made after seeing the data. A prospective cohort applying 0.25 as a fixed rule is the test that would tell us whether the trigger is stable.
The negative clinical-factor result deserves weight. Age, baseline pad use, IPSS and time from prostatectomy to SBRT showed no association with toxicity (all p>0.2). Those are exactly the variables clinicians lean on when counseling about post-op RT, so a germline score that discriminates where they do not is a real gain in principle. What the report does not resolve is whether treatment factors explain part of the signal: both cohorts allowed pelvic nodal coverage, SCIMITAR used a dose range of 30 to 34 Gy in 5 fractions, and the source does not break out toxicity by field or dose. A germline score that partly tracks who received larger fields would look predictive without being so.
For practice, the honest position is that this moves counseling more than planning. A high score before prostate bed SBRT is reasonable grounds for a fuller conversation about late GU risk, and perhaps for weighing conventionally fractionated or moderately hypofractionated postop RT, but nothing here shows that changing fractionation, dose or field in high-score pts lowers their toxicity. The concern that late G2+ GU events limit wider adoption of post-op SBRT is the motivation, and a risk-adaptive trial randomizing high-score pts between ultra-hypofractionated and longer courses is what would turn this from a stratifier into a decision tool. Until then, the result says the test travels across the surgical boundary, provided the reader recalibrates expectations about where the cutoff sits.