The comparator does not correspond to the clinical decision.

All patients in PROTEUS received perioperative ADT; the trial evaluated only the incremental contribution of apalutamide to that backbone. As the investigators acknowledge, perioperative ADT itself represents an intensification beyond current guideline recommendations, under which most of these men would undergo prostatectomy alone, with adjuvant or salvage therapy reserved for relapse. The trial, therefore, addresses whether apalutamide adds to ADT, not whether the combined strategy is superior to standard care. The comparison against surgery alone is being conducted separately and is not part of the present registrational result. This is a defensible design—a pathological endpoint is uninterpretable without systemic therapy in both arms, and an active control preserves blinding—but it has consequences that shift outcomes in a more favorable direction. Because both arms are castrated, the toxicities of ADT are common to both, thus narrowing the difference between the two groups, misleading the reader to think there is no added toxicity to their “standard-of-care”. Meanwhile, in reality, surgery alone can already have a great negative impact on quality of life, let alone adding the toxicity of ADT and apalutamide to this.

The pathological co-primary endpoint: prognostic, but not a basis for treatment decisions

This is the crux of the trial’s interpretation. The composite of pathological complete response or minimal residual disease (pCR/MRD) was markedly more frequent with apalutamide—roughly nine times as frequent—and within PROTEUS the presence of minimal residual disease was strongly associated with subsequent metastasis-free survival. This establishes the endpoint as prognostic. It does not establish it as a basis for clinical decision-making, and the two are not the same.

A prognostic marker stratifies patients by risk; a surrogate endpoint makes the stronger claim that a treatment effect measured on the marker reliably predicts a treatment effect on the clinical outcome. The inference required to justify pCR/MRD as a co-primary efficacy endpoint is the latter, and it has not been demonstrated. But there is a more immediate, and arguably more important, limitation. Pathological response is observable only after treatment, in the prostatectomy specimen; it cannot exist in a patient who has not received neoadjuvant systemic therapy, and it cannot be predicted before treatment is given. The trial offers no way to identify, in advance, the minority of men—fewer than one in ten—who will achieve minimal residual disease. The favorable outcomes of that subgroup are therefore recognized only in retrospect, once the treatment decision has been made and its toxicity incurred. The whole cohort is committed to a year of ADT plus apalutamide so that a small, post-hoc-identifiable fraction may benefit. That the achievement of minimal residual disease is associated with better metastasis-free survival is genuinely encouraging for those patients; it does not inform the patient or the clinician whether this is the optimal treatment for any specific individual.

The consequence is that the endpoint that produced the trial’s most striking numerical result is also the endpoint least able to direct care in real-life setting. It does however teach us post-hoc that reaching pCR and/or MRD has a beneficial impact on MFS, but there is no way to know who will reach pCR/MRD and who will not BEFORE surgery.

The validated endpoint was not met on the basis on which it was validated

Of the two co-primary endpoints, only metastasis-free survival has an established surrogate relationship to overall survival in localized prostate cancer, and that relationship was demonstrated using conventional imaging. In PROTEUS, metastasis-free survival assessed by conventional imaging alone did not reach statistical significance. Significance was achieved only when lesions detected by prostate-specific membrane antigen positron-emission tomography (PSMA-PET) were incorporated—a modality of greater sensitivity that detects metastatic disease earlier and at lower burden than the imaging on which the surrogacy of metastasis-free survival was established, and one introduced into the trial after its inception. The positive primary result therefore rests on a definition of the endpoint whose relationship to survival has not been validated, while the validated definition (metastasis-free survival assessed by conventional imaging, the basis on which ICECaP established its surrogacy for overall survival) yielded a result that did not cross the conventional threshold for significance.

The most patient-relevant benefit

The most clinically meaningful signal in PROTEUS may be the delay in time to subsequent therapy: a median of 33 months, achieved with one year of added treatment and, on current and still immature data, no accompanying difference in overall survival.

Interpreting this benefit requires understanding what therapy-free time means in this disease. High-risk prostate cancer that recurs after prostatectomy is, for most men, managed rather than cured. Once intensified systemic therapy is begun, it is rarely stopped: a patient who responds and then progresses moves through successive lines of treatment, and short of electing best supportive care, he remains on systemic therapy of one form or another for the rest of his life. Each of these treatments carries its own burden. The clinically relevant currency in this setting is therefore not only how long a man lives, but how much of that life precedes the point at which uninterrupted systemic therapy begins. A delay in starting that therapy is a delay in the onset of that burden.

Read this way, the trial’s findings cohere. If overall survival is unchanged but the time to subsequent therapy is extended by years, the principal effect of moving intensification to the front is to redistribute when a man spends his treatment time, not to extend his life. He receives a defined course of therapy early, then enjoys a longer interval free of it, rather than beginning open-ended therapy sooner and remaining on it. For many patients, this is a real and desirable trade, and it may be the most patient-relevant outcome the trial produced.

Conclusion

PROTEUS demonstrates convincingly that apalutamide is active when added to perioperative ADT. Its most tangible benefit may be a substantial delay in the need for subsequent therapy—on current data, time away from treatment rather than additional time alive—which, for many patients, is a meaningful gain in its own right. But the trial’s claims are narrower than its positive framing conveys. The benefit is measured against a comparator that does not represent standard care; it is anchored to a pathological endpoint that is prognostic but cannot guide treatment selection, since it is identifiable only after the fact; and its one survival-linked endpoint, metastasis-free survival, reached significance only when assessed by PSMA-PET—not by the conventional imaging on which its surrogacy for overall survival was actually established. The comparator and the endpoints do not fail independently so much as they reinforce one another, lending an impression of established benefit to findings that, on present evidence, demonstrate biological activity but not yet clinical benefit. Before perioperative androgen-receptor inhibition is adopted into routine practice, it should be shown to improve outcomes that are known to matter, measured against the treatment that patients actually receive. PROTEUS was not designed to provide that demonstration, and its positive result should be read with that limitation firmly in view.

The recently published SAKK 08/14 trial evaluating metformin in metastatic castration-resistant prostate cancer caught our attention because, once again, the results were negative.

While reviewing the study, we realized this was not an isolated finding. Across the prostate cancer disease continuum, randomized trials evaluating metformin have consistently failed to demonstrate meaningful oncologic benefit despite years of biological enthusiasm and promising observational data.

That observation prompted us to put the available randomized evidence together and write an editorial in European Urology Oncology discussing what the metformin story can teach us about drug repurposing in prostate cancer.

The Rise and Fall of the Metformin Hypothesis

For nearly two decades, metformin was one of oncology’s favorite repurposing candidates.
The story sounded compelling. A cheap, globally available diabetes drug appeared to inhibit tumor growth through AMPK activation, mTOR inhibition, and downstream metabolic effects on insulin and IGF-1 signaling. Laboratory studies looked promising. Retrospective analyses hinted at improved cancer outcomes. And because metformin was already widely prescribed and generally well tolerated, enthusiasm spread quickly.

In prostate cancer, the hypothesis became particularly attractive. Could a familiar metabolic drug enhance androgen receptor–targeted therapy? Delay progression? Improve survival?

The latest randomized evidence suggests the answer is probably no.

Published in European Urology Oncology, Gillessen and colleagues reported the SAKK 08/14 trial, also known as IMPROVE, a randomized phase 2 study evaluating enzalutamide plus metformin versus enzalutamide alone in metastatic castration-resistant prostate cancer (mCRPC). The study enrolled 166 patients and tested whether adding metformin could improve disease control at 15 months.

It did not.

Disease control rates were nearly identical between groups (52% vs 56%), with no significant differences in event-free survival, PSA progression, radiographic progression, or overall survival.
SAKK 08/14 is not a practice-changing trial. It is another negative metformin study.
But that is precisely why it matters.

Across localized disease, hormone-sensitive metastatic disease, and now mCRPC, randomized evidence has become increasingly consistent: metformin repeatedly fails to deliver meaningful oncologic benefit despite years of biological enthusiasm and promising observational data.

And at this point, it is becoming difficult to argue that metformin’s anticancer promise was anything more than an observational mirage.

A Decade of Negative Randomized Evidence

The metformin story in prostate cancer has now been tested across almost the entire disease continuum.

The MANSMED trial initially generated optimism by suggesting improved castration-resistant prostate cancer–free survival when metformin was added to androgen deprivation therapy in hormone-sensitive disease. Yet the study was small, and the signal never translated into improved overall survival or stronger secondary endpoints.

Then came the larger studies.

The phase 3 MAST trial (on which I previously wrote here) evaluated metformin in men on active surveillance for low-risk prostate cancer. Again, no meaningful reduction in progression was observed. Intriguingly, obese patients receiving metformin even appeared to experience more pathological progression in a prespecified subgroup analysis.

The STAMPEDE platform then addressed the question in a much larger and methodologically rigorous setting: nearly 1900 men with metastatic hormone-sensitive disease starting androgen deprivation therapy. If metformin truly had clinically meaningful anticancer effects, this was the type of trial where they should have emerged. However, no improvement in overall survival or progression-related outcomes was noted.

In mCRPC, the TAXOMET trial similarly failed to demonstrate benefit when metformin was added to docetaxel. And now SAKK 08/14 extends that same negative pattern to enzalutamide combinations.

At some point, repeated neutral trials stop being disappointing and start becoming informative.

Why Did We Get This Wrong?

The metformin story highlights a recurring problem in oncology: we consistently overestimate the clinical relevance of biologically plausible interventions.

Retrospective studies created the initial momentum. Patients taking metformin often appeared to have better cancer outcomes. But those datasets were deeply vulnerable to confounding. Metformin users frequently differ from nonusers in obesity, diabetes severity, healthcare access, cardiovascular risk, and overall metabolic health.

In retrospect, many of these analyses likely measured patient selection rather than drug efficacy.

This is not unique to metformin. Oncology has repeatedly seen promising repurposed agents fail once subjected to randomized testing. Vitamin D. Statins. Aspirin in several settings. The pattern is familiar: strong mechanistic rationale, enthusiastic observational data, disappointing phase 3 reality.

The deeper issue is that biological plausibility is not the same as therapeutic relevance.

Cancer metabolism is extraordinarily complex. Altering systemic glucose or insulin signaling does not necessarily translate into clinically meaningful tumor control, particularly in advanced prostate cancer, where multiple redundant survival pathways dominate.

The Biomarker Question

To be fair, SAKK 08/14 did generate one potentially interesting signal.
Patients with PTEN-expressing tumors appeared to experience delayed PSA progression with metformin, whereas PTEN-negative tumors did not.
But this finding should be interpreted cautiously.

Post hoc biomarker analyses in negative trials are notoriously unreliable. Oncology is littered with subgroup signals that never replicated. The temptation after a negative study is always the same: search for the subgroup where the drug “works.”

Still, the PTEN observation reinforces an important principle: if repurposed therapies are to succeed, they will likely require biomarker-driven selection rather than empiric treatment of unselected populations. That may ultimately be the most useful lesson from metformin’s long journey in prostate cancer.

What Should Clinicians Take Away?

Clinically, the message is now fairly straightforward.

There is currently no convincing evidence supporting metformin as an anticancer therapy in prostate cancer — not in localized disease, not in hormone-sensitive metastatic disease, and not in castration-resistant disease.

That does not diminish metformin’s value in diabetes management. Nor does it invalidate the broader concept of drug repurposing. But metformin increasingly looks like a mature negative example of how enthusiasm can outpace evidence.

And perhaps that is useful in itself.

Because oncology does not only advance through positive trials. Sometimes progress comes from learning which hypotheses — despite elegant biology and years of optimism — simply do not translate into meaningful benefit for patients.

A long-awaited positive trial

The phase 3 KEYNOTE-905 (EV-303) trial reports something the field has been waiting for: a clear survival benefit with perioperative systemic therapy in patients with muscle-invasive bladder cancer who are ineligible for cisplatin or decline it. This is a population for which, until now, surgery alone has largely remained the default.

Patients were randomized to receive perioperative enfortumab vedotin plus pembrolizumab — administered both before and after surgery — or surgery alone, with adjuvant nivolumab permitted later during the trial. The results are striking. Event-free survival shows a hazard ratio of 0.40, with a 2-year rate of 74.7% versus 39.4%. Overall survival is also improved, with a hazard ratio of 0.50. Pathologic complete response rates are substantially higher. There is early and sustained separation of the curves.

Taken at face value, this is a positive and likely practice-changing trial.

The real clinical question

Once the headline numbers are set aside, a more nuanced question emerges. What exactly is the trial comparing?

The key clinical issue is not whether enfortumab vedotin plus pembrolizumab works. Its activity in metastatic urothelial cancer is already well established. The relevant question is whether this regimen needs to be given upfront, in the perioperative setting, or whether it could be reserved for the time of recurrence without compromising outcomes. In other words, is earlier truly better than later?

The design of EV-303 suggests that it could inform this question. In reality, it does not.

What happened after progression?

The most important limitation lies in what happened after progression. Among patients in the control arm who developed recurrent or metastatic disease and received systemic therapy, most were treated with chemotherapy. Only a minority received enfortumab vedotin, and almost none received the combination of enfortumab vedotin and pembrolizumab — the very regimen under investigation.

In practical terms, most patients in the control arm were never exposed to the study drug at any point during their disease course, despite it being standard-of-care after progression.

This is not a sequencing comparison — but it raises a question

This has direct consequences for interpretation. The trial was not designed to compare perioperative versus deferred use of enfortumab vedotin plus pembrolizumab, and it successfully demonstrates superiority over surgery alone with heterogeneous subsequent therapy.

However, the pattern of post-progression treatment raises an important and unexpected question. Among control patients treated for recurrent or metastatic disease, only one received the combination of enfortumab vedotin and pembrolizumab, despite this regimen being described as first-line standard therapy in the metastatic setting.

At the same time, many of these patients received chemotherapy, despite having been classified as cisplatin-ineligible before surgery. In some cases, this may be clinically plausible—for example, if renal impairment related to tumor-associated obstruction improved after cystectomy. However, the trial does not provide sufficient detail on the reasons for initial ineligibility or on changes in eligibility over time.

This creates an internal inconsistency: patients appear to become eligible for chemotherapy after progression, yet rarely receive the study regimen for which they were previously eligible. Without further clarification, it remains unclear whether this reflects changes in clinical status, access to therapy, or treatment selection patterns.

Although the trial was not intended to address treatment sequencing, the limited use of enfortumab vedotin plus pembrolizumab after progression restricts any indirect inference about deferred use.

A smaller but telling inconsistency

A second, more subtle issue concerns the definition of the cisplatin-ineligible population. The trial relies on Galsky criteria, which include an ECOG performance status of 2 or higher. However, there is a small discrepancy between the number of patients classified as ECOG 2 in the main baseline table and those listed as ECOG 2 among the reasons for cisplatin ineligibility.

The difference is minor in absolute terms, but it suggests that performance status as a baseline variable and as a criterion for ineligibility may not have been entirely aligned, or there is a problem with the data itself, which should not happen in a trial of this magnitude. Given that ECOG status is both prognostic and central to eligibility, this lack of clarity is not entirely trivial.

What the trial shows — and what it does not

None of these points undermine the main finding of the trial. EV-303 is a strong study with a clear signal. It will change practice.

But it does not fully answer the question of sequencing, and it leaves important aspects of patient selection and post-progression management insufficiently explored.

I’m looking forward to the full manuscript of the sister trial EV-304 in which the patient population is not limited to cisplatinum-inelegible patients. If the control group in this trial also lacks exposure to EV-pembro after progression, this would be a strong signal of bias.

Bottom line

Enfortumab vedotin plus pembrolizumab is effective in this setting. Whether it must be given perioperatively, or could be reserved for relapse without compromising outcomes, remains unresolved.

A convincing result—at first glance

The final overall survival results of the PEACE-3 trial suggest that adding radium-223 to enzalutamide improves outcomes in mCRPC with bone metastases. At first glance, the message seems straightforward: combine early, treat aggressively, improve survival. But that interpretation rests on an assumption that the trial does not actually test.

The real clinical question: timing

Radium-223 is not a novel intervention in this setting. Since ALSYMPCA, we already know it prolongs survival in patients with bone-predominant mCRPC. The clinically relevant question is therefore not whether radium-223 works, but when it should be used. Should it be incorporated upfront in combination with an androgen receptor pathway inhibitor, or reserved for later in the disease course? PEACE-3 appears to address this question, but on closer inspection, it does not.

What happened after progression?

The trial compares upfront combination therapy versus enzalutamide alone, yet provides limited insight into what happens after progression. While post-progression treatments are reported, the use of radium-223 in the control arm is not clearly described.
What is reported is telling: chemotherapy accounts for roughly three-quarters of subsequent treatments, while the category of “other therapies” remains small. Given that radium-223 is not explicitly listed, the most plausible interpretation is that relatively few patients in the control arm ultimately received it, despite being a life-prolonging treatment option.

Exposure versus timing

If that is the case, the trial does not primarily compare early versus delayed radium-223. Instead, it compares patients who received radium-223 at some point with those who likely never received it at all. This distinction matters because the observed overall survival benefit—numerically compelling, with a hazard ratio around 0.69—may therefore reflect the impact of radium-223 exposure itself, rather than the timing of its administration.

In other words, PEACE-3 may be telling us that radium-223 should be used, not necessarily that it should be used early.

A delayed effect in the survival curves

The shape of the survival curves reinforces this uncertainty. The curves cross at approximately 18 months, followed by a delayed separation.

This pattern suggests a time-dependent treatment effect, with limited early benefit and a more pronounced impact later in the disease course. Such dynamics are consistent with the biology of bone-targeted therapies and raise a plausible alternative explanation: that the benefit of radium-223 emerges over time and might not depend on immediate upfront administration.

Why this distinction matters

This is not a trivial nuance. Upfront combination therapy commits all patients to early exposure, with added toxicity and reduced flexibility in sequencing. A strategy that reserves radium-223 for selected patients with persistent bone-predominant disease could, in theory, capture much of the same benefit while avoiding unnecessary treatment in others. PEACE-3 does not allow us to distinguish between these approaches.

What PEACE-3 shows—and what it doesn’t

None of this diminishes the importance of the trial. PEACE-3 clearly supports the role of radium-223 in mCRPC and demonstrates that combining it with enzalutamide is feasible. But it leaves a critical clinical question unresolved: does earlier use improve outcomes—or does simply ensuring that patients receive radium-223 at some point drive the benefit?

Bottom line

PEACE-3 confirms that radium-223 matters.
It does not establish that earlier is better.
And until sequencing is directly tested, that distinction should not be ignored.

Trial design in brief

This was a randomized, double-blind, placebo-controlled phase II study evaluating oxybutynin for ADT-related hot flushes in men with prostate cancer.
Patients experiencing a predefined minimum number of weekly hot-flush episodes were randomized to placebo, low-dose oxybutynin, or higher-dose oxybutynin for six weeks.
The primary endpoint was reduction in hot-flash score (a composite of frequency and severity), with secondary endpoints including patient-reported bother, quality-of-life measures, and tolerability.
Participants kept daily symptom diaries, and outcomes were largely patient-reported. In total, 88 patients were accrued across 15 centers over approximately 25 months.

The design is methodologically sound for a symptom-control question: placebo control, blinding, and patient-centered endpoints are all appropriate.
At the same time, the heavy reliance on diaries and questionnaires inevitably introduces both compliance challenges and expectation effects.

The appeal of small, practical studies

There is real value in asking pragmatic questions with cheap, generic drugs. Oxybutynin is widely available, familiar to clinicians, and generally well tolerated. The investigators deserve credit for conducting a randomized, placebo-controlled trial in a domain where enthusiasm, funding, and prestige are typically limited.
Symptom-control research rarely attracts headlines, yet it often has a more immediate impact on patients' daily well-being than incremental survival gains.

The accrual paradox

One operational detail is hard to ignore: the pace of recruitment.
Eighty-eight patients over roughly 25 months across 15 centers equates, under the simplifying assumption of equal contributions, to approximately one patient per center per month.
For a toxicity that is nearly universal in one of the most common cancers, it is slow.

Several explanations are possible: eligibility thresholds, diary burden, staffing limitations, and competing trials all play a role.
Still, it is difficult not to consider the structural reality that generic, inexpensive drugs lack commercial momentum. Without industry infrastructure or marketing pressure, enrollment depends on motivated investigators and already stretched research teams. The result is prolonged timelines even for clinically relevant questions.
Another reason to applaud the investigators for persevering despite slow accrual.

The placebo effect is not a footnote

Perhaps the most striking finding is not only the efficacy signal with oxybutynin, but the magnitude of the placebo response.

Before unblinding, 45% of patients in the placebo arm indicated they would choose to continue the “study drug.” That is substantial. It highlights how expectation, attention, and structured follow-up can meaningfully influence patient-reported vasomotor symptoms.

Hot-flush research is inherently susceptible to placebo influence, which raises the evidentiary bar for any intervention and underscores the importance of rigorous blinding and control arms.

Take-home message

This trial is a reminder of two parallel truths.
First, simple, low-cost symptom-control interventions are worth studying, even if they lack commercial appeal.
Second, patient-reported endpoints for vasomotor symptoms exhibit substantial placebo components, necessitating careful interpretation.

Oxybutynin emerges as a plausible, accessible option, but the broader lesson may be about how—and how slowly—we generate evidence for the everyday toxicities that shape patients’ quality of life.

The long-term outcomes of the HYPO-RT-PC trial were recently published in The Lancet Oncology.

The headline result is clear: ultra-hypofractionated radiotherapy demonstrates noninferiority to conventional fractionation for long-term oncologic control.
The subheadline is less comfortable: the absolute burden of late urinary and bowel toxicity is high, and the way toxicity was captured likely underestimates what patients actually experience.

This post focuses on what lies beyond the primary endpoint—how toxicity was measured, what may be missing, and why the absolute numbers warrant greater attention in daily counselling.

Trial design

• Multicentre, randomised, open-label phase 3 non-inferiority trial.

• Population: Men with localised, predominantly intermediate-risk prostate cancer eligible for external-beam radiotherapy; only a small minority were high-risk. No androgen-deprivation therapy was allowed

• Intervention:

  • Ultra-hypofractionated radiotherapy — 42.7 Gy in 7 fractions
    (high dose per fraction, short overall treatment time).

• Comparator:

  • Conventional fractionated radiotherapy — 78.0 Gy in 39 fractions
    (standard daily schedule over ~8 weeks).

• Randomisation: 1:1 allocation across multiple centres; treatment was not blinded.

• Primary endpoint: Failure-free survival, a composite including biochemical failure (PSA rise), clinical recurrence, start of salvage androgen-deprivation therapy, or prostate-cancer–related death.

• Key secondary endpoints: Late genitourinary (GU) and gastrointestinal (GI) toxicity, overall survival, and other oncologic outcomes.

In essence, the study asked whether delivering prostate radiotherapy in seven large fractions could achieve cancer control not meaningfully worse than the traditional 39-fraction schedule, while also tracking long-term side effects.

Outcomes

From a trial-design perspective, the non-inferiority margin (4%) was appropriate, and the primary endpoint was met. Ultra-hypofractionation does what it set out to do: similar cancer control with fewer treatment sessions.

However, once you step away from comparative statements and look at absolute long-term morbidity, the picture is different:

• ~30% grade ≥2 genito-urinary (GU) toxicity at 10 years

• ~14% grade ≥2 gastro-intestinal (GI) toxicity at 10 years

• ~7% grade 3 GU toxicity at 10 years (supplementary data)

Overview of the RTOG scoring system

Grade 3 GU events are rarely trivial. They often imply invasive management—repeated catheterisation, endoscopic procedures, or hospital-level care. This is not a “borderline symptom” signal; it represents severe morbidity in roughly 1 in 14 patients a decade after treatment. The key point: schedule shortening does not eliminate serious late urinary harm.

Why the real toxicity rate may be higher

Two methodological features nudge the cumulative incidence downward.

A. Physician-reported toxicity

Late GU/GI events were recorded by physicians using RTOG criteria, frequently via telephone follow-up. Physician scoring is known to be conservative for moderate symptoms (urgency, frequency, minor bleeding), especially when contact is brief or remote.
Moreover, the trial does not state who the “physician” is.
Urologists and radiation oncologists view urinary symptoms through different clinical lenses: urologists may recognise and document moderate morbidity earlier, whereas radiation oncologists may interpret similar complaints as expected post-treatment effects.
The unblinded design further increases the risk of reporting bias for treatment-related side effects if the physician is a radiation oncologist, since they will be biased towards underreporting of toxicity.
This creates a systematic difference in detection thresholds, primarily affecting grade 2 toxicity, whereas grade 3 events are less susceptible because they typically require procedures.
No conclusions can be drawn because the trial does not report the physician's subspecialty; however, this should be kept in mind, as it may influence the toxicity outcome.

B. Censoring at metastatic progression

Patients were censored for late toxicity at the time of regional or distant metastasis.
In other words, patients who are progressive no longer contribute to the cumulative incidence numbers.
The assumption is that those who progress would have had a toxicity profile similar to those who do not—an assumption that is methodologically convenient but clinically uncertain.
Probably, there will again be no difference in toxicity between both arms concerning the patients who progress; however, it is possible that these are more frail patients who will experience more toxicity in the long term. This assumption would lead to an underestimation of the long-term toxicity.

Open-label design and management discretion

The trial was open-label. As discussed above, this may have an impact on the reporting of toxicity.
However, a clear example of a related signal of open-label bias appears in the management of biochemical recurrence.
There was no strict protocol-defined PSA threshold for initiating salvage androgen-deprivation therapy, and the reported median PSA at ADT start was numerically higher in the ultra-hypofractionated arm (9.2 ng/mL vs 8.4 ng/mL).
This does not prove bias, but it suggests that clinician discretion—potentially influenced by knowledge of treatment allocation—may have delayed salvage therapy in one arm, since time to initiation of ADT was included in the composite primary endpoint. In other words, try to delay ADT initiation as long as possible, and you might get a better outcome in the intervention arm.

What we still do not know

Patient-Reported Outcomes (PROs) are pending.
This is not a minor add-on. PROs will likely reshape the interpretation of grade 2 toxicity, where patient experience and clinician scoring often diverge the most. If physician-reported rates already approach one-third for GU toxicity, PRO data may reveal an even larger symptomatic burden—or, less likely, reassure us that many clinician-graded events were clinically negligible. Until PROs are available, any statement about the true lived impact of late toxicity remains provisional.

We have PROs regarding this trial, including a 6-year follow-up.

A scoring system different from the RTOG toxicity scale is used.
However, as shown in the red square, approximately 40% of patients report urinary tract problems, and approximately 15% report moderate to severe problems.

We expect this number to increase with longer follow-up, given the physician-reported evolution in cumulative RTOG toxicity.

Practical take-home for counselling

• This trial proves that ultra-hypofractionation is a valid alternative to conventional fractionation for cancer control, with clear benefits for patients.
  If radiotherapy is the treatment of choice for a patient, we should aim towards ultra-hypofractionation.

• The absolute long-term toxicity is substantial, with severe urinary morbidity around 7% at 10 years and moderate-to-severe GU symptoms in roughly one-third of patients.
  A systematic review published in 2022 showed a pooled RTOG ≥ 2 incidence of 17% at 5-year follow-up. We now have prospective data showing double-trouble at 10 years of follow-up

• Measurement methods probably undercall symptoms, and censoring rules may further lower reported cumulative incidence.

• For me, this substantial toxicity is particularly relevant for younger men and those with long life expectancy, where a decade-scale horizon is clinically meaningful.

Bottom line:

Ultra-hypofractionation is convincingly non-inferior to conventional fractionation for cancer control, but the long-term burden of urinary and bowel morbidity after radiotherapy remains substantial. These prospective data therefore confirm the view that, in men with a long life expectancy, radiotherapy should not automatically be the default local treatment and that surgery will often be the preferred option.
This is not to suggest that surgery is without harm—prostatectomy carries clear risks of incontinence and erectile dysfunction—but rather that the long-term side-effect profile of radiotherapy deserves equal weight in treatment selection and patient counselling.

The recent publication of the 23-year results of the European Study of Prostate Cancer Screening (ERSPC) has generated distinctly divided reactions. Some interpret the findings as confirmation that PSA-based screening has lost its purpose; others view the results more favorably.
In this piece, I aim to put the ERSPC data into proper context and highlight how the urological community has already evolved its practice to improve outcomes.

Trial Overview

The ERSPC was launched in 1993 to assess whether population-based PSA screening reduces prostate cancer mortality. Men aged 50–74 were randomized to either PSA screening or no screening invitation. In most centers, a PSA ≥3 ng/mL triggered transrectal ultrasound-guided biopsy. The primary endpoint was prostate cancer–specific mortality; secondary endpoints included incidence.

A total of 162,236 men in a pre-specified core age group (55–69 at randomization) were included: 72,888 in the screening arm and 89,348 in the control arm.

Key Results (Core Age Group)

Incidence of prostate cancer

• 14% (screening)

• 12% (control)

Prostate cancer mortality

• 1.4% (screening)

• 1.6% (control)
  → Risk ratio: 0.87 (13% relative risk reduction in the screening group)

• Absolute risk reduction: 0.22%

• Number needed to invite: 456

• Number needed to diagnose: 12
  → To prevent one prostate cancer death

Interpreting the Data

At face value, an absolute mortality reduction of 0.22% over 23 years is underwhelming. For a random man aged 55–69, PSA screening reduces the risk of prostate cancer death by just 0.22 percentage points. Even among men diagnosed via screening, the benefit translates to only a 13% relative reduction in prostate cancer mortality.

As Vinay Prasad would say, if you can fit your laser pointer between the curves, you can give the plenary session. One would need a very small laser pointer and an extremely steady hand to get on the podium...

Viewed this way, the results seem discouraging—almost humiliating for practicing urologists.
But this interpretation is incomplete. These data require context, and fortunately, plenty is available.

Contextualizing the ERSPC

The trial design was intentionally simple:

1. Invite men to have a PSA test;

2. biopsy if PSA exceeds a pre-defined threshold;

3. diagnose and treat accordingly.

This reflects a classic model of population screening: a cheap, minimally invasive test that flags those needing further work. However, such simplicity comes with major limitations—especially the risks of overdiagnosis and overtreatment.
Luckily, much has changed since 1993.

1. Contamination of the Control Arm

In the ERSPC, 83% of the screening arm underwent at least one PSA test—a reasonable participation rate compared with mammography-based screening in breast cancer.
However, if we really want to isolate the effect of a screening test, we need to ensure the test is not performed on the control group. And in the case of ERSPC (as in another prostate-cancer screening trial), the assumption that the control group received no screening does not hold. This kind of opportunistic screening in the control group is what we call contamination, and it was high:

• Sweden: up to 50%

• Finland: 63%

• Spain: 66%

If half of the control group underwent PSA screening anyway, interpreting the trial as “screening vs. no screening” becomes impossible. This inevitably obscures the true effect size.

2. PSA Alone Is No Longer the Gatekeeper

In the ERSPC, a PSA ≥3 ng/mL led directly to biopsy.
This fast-track pathway is outdated, prone to overdiagnosis, and no longer reflective of real practice. Today, PSA is contextualized.

PSA density

PSA must be interpreted relative to prostate volume: we know that the bigger the prostate, the higher the PSA can be. PSA density reduces unnecessary biopsy referrals and improves risk stratification. This alone significantly reduces overdiagnosis.

Clinical examination

The ERSPC did not incorporate digital rectal examination (DRE) by a trained urologist. A low PSA density and negative DRE provide strong reassurance that biopsy is unnecessary.

Confirmatory PSA

Current guidelines recommend repeat PSA testing to confirm an elevated PSA; the ERSPC did not.

MRI before biopsy

This is the biggest step forward.
Prostate MRI has a sensitivity of ~91% for clinically significant disease. In other words, with its high negative predictive value, a negative MRI is a very reliable tool to avoid biopsy.
MRI also allows targeted biopsy rather than systematic sampling—further minimizing overdiagnosis of low-risk disease.

Together, PSA density, repeat testing, DRE, and MRI have reshaped prostate cancer diagnostics, and none of these refinements are represented in ERSPC.

3. Safer Biopsy Techniques

In ERSPC, all biopsies were performed using a transrectal approach, which necessitated antibiotic exposure and, despite prophylaxis, posed meaningful risks of infection and sepsis. The shift to transperineal biopsy has eliminated the need for antibiotics while simultaneously reducing the risk for urosepsis from 0.9% to 0.1%.
Thus, even when biopsies are necessary, their safety profile has improved.

4. Overtreatment Remains a Central Concern

Screening should avoid treating cancers that would never cause harm. In prostate cancer, overtreatment carries significant morbidity such as urinary incontinence, erectile dysfunction, lower urinary tract symptoms, bowel dysfunction, treatment-related mortality (rare but real), fatigue, hot flushes, etc.
While multiple attempts are being made to reduce the risk of treatment-induced morbidity and mortality by implementing new surgical & radiational techniques, and the development of new drugs, the best way to reduce the risk of complications is to reduce the number of patients we unnecessarily expose to treatment.
In other words, we need to avoid treating indolent disease.
This is where the concept of active surveillance has transformed outcomes.
Active surveillance prevents unnecessary radical therapy for low-risk tumors and delays treatment to a point where the tumor is evolving, in ideal circumstances, without compromising outcome.

What the ERSPC Tells Us About Prostate Cancer

Despite its limitations, ERSPC is reassuring in one regard: prostate cancer is largely indolent. The cumulative prostate cancer mortality at 23 years was 1,4% in the screening group and 1,6% in the control group, while cumulative other-cause mortality at 23 years was 49% in each of the two groups.
In other words, men are far more likely to die of something other than prostate cancer.
Screening should therefore be applied only to men with good life expectancy and discontinued as overall health declines.

Future Directions

Reducing overdiagnosis

Although MRI shows excellent sensitivity and a strong negative predictive value for clinically significant prostate cancer, it remains too time- and resource-intensive to serve as an initial screening tool for the general population. Nevertheless, the development of biparametric (instead of multiparametric) MRI protocols and growing implementation of artificial intelligence have the potential to increase accessibility, streamline acquisition and interpretation, and ultimately lower costs.

Reducing overtreatment

Active surveillance will continue to play a central role in minimizing overtreatment. Although currently limited to low-risk and carefully selected intermediate-risk cases, there are gains to be made by increasing the implementation of active surveillance in its current settings, as a realistic potential to further expand the indications.

Together, these developments signal a future in which early detection is paired with carefully modulated intervention—seeking not only to detect high-risk cancer earlier but to treat only when clinically meaningful.

Conclusion

Although we must acknowledge that the ERSPC demonstrates only a marginal reduction in prostate cancer–specific mortality, its findings must be interpreted in light of substantial contamination of the control group and a study design that no longer reflects contemporary diagnostic pathways.

If the major concerns surrounding population-based screening are overdiagnosis and overtreatment, the ERSPC confirms that these concerns are real in prostate cancer.

Yet the field has changed dramatically since the trial was conceived. PSA is no longer used in isolation; MRI and PSA-based risk stratification have reshaped the diagnostic landscape, reducing unnecessary biopsies and improving the detection of only clinically significant disease. At the same time, the widespread adoption of active surveillance has helped ensure that men with indolent tumors are not exposed to the morbidity of radical treatment.

Ultimately, in my opinion, the ERSPC should not be viewed as an indictment of prostate cancer screening, but rather as a reminder of how essential it is to approach screening thoughtfully.
Modern screening must be targeted to men with sufficient life expectancy and must be coupled with careful diagnostic triage and restraint in treatment. Prostate cancer remains a common yet often indolent disease; our task is not merely to detect it, but to do so in a way that preserves the balance between benefit and harm.

Introduction

This is a long one, but bear with me, it’s necessary given the trial's significant impact!

IMvigor011 is one of the most discussed urology papers of 2025 — and for good reason. Published in The New England Journal of Medicine, the trial reports significant improvements in disease-free survival and overall survival with adjuvant atezolizumab in muscle-invasive bladder cancer.

What distinguishes IMvigor011 is not the drug, but the strategy.
Rather than treating patients based solely on clinicopathologic risk, the trial limits intervention to those with molecular evidence of residual disease, as identified by circulating tumor DNA (ctDNA).
The intent is explicit: escalate treatment only for patients at high biological risk, while sparing others unnecessary toxicity.
Exactly what we need in clinical practice!

That intent deserves both recognition and careful scrutiny.

However, careful scrutiny is always needed, even in “hot” trials.
The trial's selective nature, the timing of the intervention, and the patients who never reached randomization are central to interpreting the results.

This Beyond the Abstract review examines what IMvigor011 truly tested, why it succeeded where IMvigor010 failed, how robust the efficacy signal appears under sensitivity analyses, and what the trial reveals about the limits of post-cystectomy risk stratification in an aggressive disease.

Trial Design — Simple in Concept, Selective in Practice

IMvigor011 tested a straightforward idea: treat only patients with molecular evidence of residual disease after cystectomy.

Patients with muscle-invasive bladder cancer who had undergone radical cystectomy and had no radiographic evidence of disease entered a surveillance phase. During this period, they underwent serial ctDNA testing for up to one year after surgery.

• Patients who remained ctDNA-negative received no adjuvant therapy.

• Patients who became ctDNA-positive at any point, while still radiographically disease-free, became eligible for randomization.

These ctDNA-positive patients were randomized in a 2:1 ratio to receive atezolizumab or placebo every four weeks for up to one year. Disease-free survival (DFS) from the time of randomization was the primary endpoint, with overall survival (OS) as a key secondary endpoint tested hierarchically.

This design deliberately enriches for patients at the highest biological risk — a strength that also shapes how the results must be interpreted.

Results — Robust Signals in a Narrow, High-Risk Window

Among 761 patients enrolled in the surveillance phase, 379 developed ctDNA positivity during follow-up. Of these, 250 were randomized, while 95 (25%) were excluded because they developed radiographic recurrence before randomization.

This attrition is not incidental.
It reflects the aggressive natural history of muscle-invasive bladder cancer and indicates that the randomized population represents a biologically and temporally selected subset of ctDNA-positive patients.

Within this selected cohort, adjuvant atezolizumab significantly improved DFS compared with placebo (median 9.9 vs 4.8 months; HR 0.64). The Kaplan–Meier curves separate early — within the first two months after randomization — and then largely run in parallel.

An early split followed by parallel curves raises legitimate concerns about informative censoring, particularly disappointment-driven dropout in the placebo arm.
To explore this, I asked to digitize the published curves and performed sensitivity analyses (using the BREAKING-ICE app) adjusting for potential informative censoring (huge shout-out!). Reassuringly, the hazard ratio remained stable after adjustment, suggesting that the DFS benefit is not an artifact of early attrition or censoring patterns.

How did I get to this sensitivity analysis? I’ll explain at the bottom!

But more important than a PFS benefit, the trial is positive for overall survival (HR 0.59), which is far less susceptible to censoring bias.
This is very encouraging, though necessarily provisional.

The Control Arm in Context — An Unspoken Tension

Enrollment in IMvigor011 began in May 2021 and continued until November 2024, a period during which adjuvant immunotherapy entered routine practice. Adjuvant nivolumab was FDA-approved in August 2021 for high-risk muscle-invasive urothelial carcinoma regardless of PD-L1 status, and EMA-approved in May 2022 for patients with PD-L1 expression ≥1%.

Despite this, the control arm of IMvigor011 remained placebo throughout accrual.

It is not possible to determine precisely how many patients in IMvigor011 would have been eligible for adjuvant nivolumab. The trial used the VENTANA SP142 assay and reported PD-L1 status as <5% versus ≥5% immune-cell expression, which does not map directly to the EMA cutoff of ≥1%. However, at least 35% of patients in each randomized group had PD-L1 expression ≥5%, meaning a substantial proportion would almost certainly have met EMA eligibility, and an even larger proportion would have been eligible under FDA labeling.

Nivolumab was approved solely on the basis of a disease-free survival benefit, without a demonstrated overall survival advantage. Nonetheless, for much of the trial’s conduct, many high-risk patients would have had access to adjuvant immunotherapy outside the trial, rather than observation or placebo.

This context is not discussed in the IMvigor011 manuscript. While it does not invalidate the trial, it matters for interpretation. IMvigor011 does not compare ctDNA-guided atezolizumab with ctDNA-agnostic adjuvant nivolumab, nor does it address whether ctDNA-negative but clinicopathologically high-risk patients should forgo approved therapy.

These questions fall outside the sponsor’s scope, but they are unavoidable in clinical practice — and they belong in any complete reading of the trial.

Why IMvigor011 Worked When IMvigor010 Failed

IMvigor011 did not succeed because atezolizumab suddenly became more effective. It succeeded because patient selection fundamentally changed.

IMvigor010 randomized patients immediately after cystectomy based on clinicopathologic risk. In that less selected population, adjuvant atezolizumab failed. Subsequent retrospective analyses showed that the benefit was confined to patients who were ctDNA-positive — a signal diluted by the inclusion of many patients already cured by surgery.

IMvigor011 operationalized that lesson prospectively. By requiring ctDNA positivity before randomization, the trial enriched for patients with true minimal residual disease and avoided unnecessary treatment of low-risk patients. The result is not proof that adjuvant immunotherapy works broadly, but proof that biology-guided escalation can succeed where non-selective strategies fail.

Adjuvant Therapy — or Early Salvage by Another Name?

IMvigor011 is frequently labeled an adjuvant trial, but that description deserves nuance.

Treatment was not initiated from a disease-free state at a fixed postoperative time point. Instead, therapy began after molecular relapse had already occurred, but before radiographic progression. In practical terms, IMvigor011 evaluates early salvage therapy triggered by molecular progression, rather than classical adjuvant treatment.

This distinction helps explain both the trial's success and its limits. DFS is measured from randomization — not from cystectomy — and therefore reflects control of molecularly evident disease rather than prevention of recurrence from a truly disease-free state. The OS signal is therefore the more compelling endpoint, but also the one requiring the most caution.

Who Is Missed — and What That Says about Timing

The same design that strengthens IMvigor011 also exposes its blind spots.

A substantial proportion of ctDNA-positive patients progressed to radiographic recurrence before treatment could begin. These patients represent a biologically aggressive subgroup for whom ctDNA-triggered post-operative intervention did not create a therapeutic window.

Importantly, these patients were not only missed by adjuvant strategies — they were likely overtreated with cystectomy alone. Early relapse suggests that surgery was never sufficient for disease control, raising the broader question of whether meaningful risk stratification must occur before cystectomy, not only after.

It would be particularly informative to know whether these excluded patients, on average, entered surveillance later within the allowed 6–24 week post-cystectomy window. If aggressive cases were disproportionately enrolled later, this would support the hypothesis that the issue is not ctDNA selection itself, but delayed intervention, and that adjuvant treatment may need to start earlier for some patients.

This remains hypothetical, but it is a testable and clinically relevant question that the published data do not address.

What Data Are Still Needed Before Practice Changes

IMvigor011 provides proof of principle, not a finished clinical algorithm.

First, outcomes of the excluded ctDNA-positive patients with early radiographic recurrence are essential to understand who is being systematically missed and why.

Second, longer follow-up is required. The OS signal is encouraging, but remains immature. In muscle-invasive bladder cancer, survival curves can converge late due to differences in subsequent therapy upon progression, making durability of benefit critical.

Finally, IMvigor011 does not address how ctDNA-guided strategies will perform in an evolving perioperative landscape that increasingly includes neoadjuvant and perioperative immunotherapy and antibody–drug conjugates.

Conclusion

IMvigor011 is a strong trial. Its core idea — to restrict adjuvant immunotherapy to patients with molecular evidence of residual disease — directly addresses the problem of overtreatment after cystectomy. Within this biologically enriched population, the results are reassuring: disease-free survival improves, an early overall survival signal is observed, and sensitivity analyses support the robustness of the findings.

The trial’s success, however, is inseparable from its selectivity. A substantial fraction of ctDNA-positive patients progressed before treatment could begin, highlighting both the aggressiveness of the disease and the limits of post-cystectomy intervention. These patients were likely overtreated with cystectomy alone, suggesting that meaningful risk stratification may need to occur earlier, potentially before surgery.

IMvigor011 was also conducted in a changing therapeutic landscape. During much of its accrual, adjuvant nivolumab was an approved option for high-risk patients, meaning that some participants would have had access to immunotherapy outside the trial rather than placebo. This context, not discussed in the original manuscript, matters for how the results are translated into practice.

Taken together, IMvigor011 should not be viewed as a replacement for existing adjuvant strategies, but as convincing evidence that biology-guided escalation can succeed where broad approaches fail. It moves the field forward while making clear that timing, patient selection, and evolving standards of care remain central challenges.

Addendum, for those who want the real deep-dive

What about the sensitivity analyses?

Since the authors don’t provide the censoring data, we have to estimate it.
For this, I look at the PFS curves and focus on the first interval at 2 months.

You can see my calculations at the bottom right of this picture.

At 2 months, the Atezolzumab is around 80%; in other words, there is a 20% drop, or 20% of patients either experienced the event or were censored. We start with 167 patients, and at 2 months, 145 patients remain at risk. This means 22 (167-145), or 13% of patients experienced the event. However, since 20% of patients are lost, this means about 7% are censored.
We can do the same calculations for the placebo arm, accounting for a drop of around 33%. Using the same calculations, we estimate that approximately 16% of patients are censored.

Now, for the sensitivity analysis, I increased the amount of censoring relative to what I estimated. In other words, in the sensitivity analysis, I censored more patients than I estimated were actually censored, and I increased the censoring time from 2 months to 12 months.

Even with this overestimation of censoring, the HR holds strong at 0,69 - p=0,017

In other words, even if there was informative censoring, it would probably not have influenced the results!

We already had the results of KEYNOTE-921 (pembrolizumab + docetaxel), but now there is no more escaping reality thanks to the publication of CheckMate 7DX (nivolumab + docetaxel) in the Lancet Oncology this month.

The setup: correct question, correct control, correct population

Both trials enrolled patients with mCRPC who had progressed on androgen deprivation therapy and at least one androgen receptor pathway inhibitor (ARTA). Both were chemotherapy-naïve in the mCRPC setting. Both used docetaxel plus prednisone as the control arm, which is exactly what these patients receive in real life. There is no straw-man comparator here.

This matters because when a negative trial uses the right control arm, the result is informative. These studies were not undercut by design flaws that would excuse failure.

Primary endpoints: flat, overlapping, and unrecoverable

• In CheckMate 7DX, nivolumab added nothing to docetaxel. Median radiographic progression-free survival was 9.4 vs 8.7 months (HR 0.96, p=0.59), and overall survival was 18.7 vs 18.9 months (HR 1.09, p=0.36). The Kaplan–Meier curves never meaningfully separate, and the hazard ratios hover around 1.0 throughout follow-up.

KM curve for PFS: complete overlap, no significant difference in early censoring (green square).

KM curve for OS: complete overlap, you can forget about the tail of the curve, even if it’s in favor of the placebo.

• KEYNOTE-921 tells the same story with a different PD-1 inhibitor. At the interim analysis, rPFS failed to meet the prespecified boundary despite a numerically favorable HR (0.85). At final analysis, overall survival was 19.6 vs 19.0 months (HR 0.92, p=0.17). No statistical success, no clinical signal.

Secondary endpoints: nowhere to hide

When primary endpoints fail, authors often pivot to secondary outcomes. That strategy fails here.

Across both trials:

• PSA response rates were virtually identical between arms.

• Objective response rates in measurable disease were the same, with no improvement in depth or durability of response.

• Time to PSA progression, time to pain progression, and time to subsequent therapy showed no meaningful differences.

• Toxicity increased modestly but consistently with PD-1 addition, including immune-mediated events and treatment-related deaths, without any offsetting benefit.

There is no secondary endpoint that plausibly changes the interpretation.

Subgroups and biomarkers: reproducible disappointment

Both trials performed extensive subgroup and biomarker analyses—PD-L1 expression, disease burden, prior ARPI exposure, visceral metastases. The result is monotony. Forest plots center on unity. Occasional numerically favorable subgroups (e.g., PD-L1–positive patients in KEYNOTE-921) fail interaction testing and collapse under multiplicity and post-hoc selection.

If a therapy only works when sliced thin enough, it does not work.

The uncomfortable implication

Taken together, these trials are not just two failures. They are a replicated negative experiment. Different sponsors, different antibodies, different geographies—same outcome. The hypothesis that taxane chemotherapy “primes” prostate cancer for PD-1 blockade is no longer untested. It has been tested at scale and rejected.

The problem is not trial execution. It is not patient selection within the current framework. It is the assumption that unselected mCRPC is meaningfully sensitive to PD-1 inhibition when layered onto docetaxel. And it is clearly not.

The authors of CheckMate 7DX, or at least the funded writers, try to save the trial by hoping nivolumab would be beneficial in a subgroup of patients.
Personally, I don’t buy it. For me, if the trial is completely negative in an all-comer population, there won’t be a big effect size from one or the other subgroup without at least a signal in the curves.

Bottom line

Both CheckMate 7DX and KEYNOTE-921 are unequivocally negative. No improvement in progression-free survival. No improvement in overall survival. No rescue by secondary endpoints. No convincing subgroup. Just more toxicity layered onto standard chemotherapy.

At this point, repeating this strategy in unselected mCRPC is not perseverance—it is inertia.

This discussion focuses exclusively on the HRR-deficient cohort (cohort 2) of TALAPRO-2. These data directly led to FDA approval of talazoparib in mCRPC in the HRR-deficient population. Expansion of the label for talazoparib/enzalutamide to non-HRR gene-mutated patients was declined.

Trial design: large, rigorous, and fit for purpose

TALAPRO-2 is a large, international, randomized, double-blind phase 3 trial with a two-stage design. Patients with first-line mCRPC were initially enrolled into an unselected cohort, followed by a prospectively defined HRR-deficient cohort.

Patients in the HRR-deficient cohort were randomized 1:1 to:

• talazoparib + enzalutamide, or

• placebo + enzalutamide,

with ongoing androgen deprivation therapy in both arms.

In this cohort:

• Radiographic progression-free survival (rPFS) was the primary endpoint, and

• Overall survival (OS) was a prespecified, alpha-protected key secondary endpoint.

This matters. The OS result was not exploratory and was not rescued by post-hoc subgroup analyses. It was embedded in the statistical framework from the start.

Endpoints and control arm: appropriate and defensible

The choice of rPFS as the primary endpoint is standard in first-line mCRPC, and blinded independent central review strengthens internal validity. More importantly, OS testing was hierarchically incorporated into the design.

The control arm — enzalutamide alone — was appropriate at the time the trial was initiated and remains defensible. This was not a weakened comparator, and the enrolled population reflects routine first-line mCRPC practice.

On crossover: a justified absence, not a design flaw

The lack of mandated crossover in TALAPRO-2 should not be interpreted as a methodological weakness. Crossover is informative when a drug has already demonstrated efficacy in later-line therapy, and the scientific question is whether earlier use improves outcomes over sequencing. That was not the case here.

Talazoparib had never been shown to confer a survival benefit in later-line mCRPC, and there was therefore no validated downstream strategy to cross over to. Mandating crossover would have introduced an unproven intervention into the control arm.

The headline result: a real and meaningful OS benefit

In the HRR-deficient cohort, talazoparib plus enzalutamide produced a statistically significant and clinically meaningful improvement in overall survival, with a hazard ratio of approximately 0.62 and a median OS gain of more than one year.

OS curve in patients with any HRR gene alteration - HR 0,62 - Median difference in OS of 14 months

This is not a marginal effect. It is the clearest OS benefit reported so far with a PARP–ARPI combination in mCRPC and provides a solid evidentiary basis for FDA approval.

There seems to be some additional early censoring in the control arm during the first year (probably disappointment-driven), however, not enough to change the outcome.

Not all HRR alterations behave the same

The survival benefit is not uniform across HRR alterations. The largest and most robust effect is seen in patients with BRCA1/2 alterations (HR ~0.50). In contrast, patients with non-BRCA HRR alterations show a smaller and statistically nonsignificant OS signal.

B: OS in patients with BRCA1/2 gene laterations - C: OS in patients with non-BRCA1/2 alterations

Toxicity: the unavoidable cost of intensification

Upfront combination therapy comes with substantial haematological toxicity, particularly anaemia and neutropenia, leading to frequent dose interruptions, reductions, transfusions, and treatment modifications.

This is especially relevant in a first-line setting, where patients are often asymptomatic and functioning well at the time of treatment initiation. Intensification shifts a significant toxicity burden into an early disease phase, raising legitimate concerns about downstream treatment feasibility.

Post-progression therapy: reassuring data, and credit to the authors

One strength of TALAPRO-2 is the availability of post-progression treatment data (supplementary appendix).

Despite early haematological toxicity, access to cytotoxic chemotherapy after progression appears similar between arms:

• 71% of patients receiving subsequent therapy in the talazoparib arm were treated with chemotherapy.

• 69% of patients in the placebo arm who received subsequent therapy received chemotherapy.

These data argue against the idea that upfront toxicity meaningfully compromised access to docetaxel or cabazitaxel and weaken a common alternative explanation for the OS benefit.

PARP inhibitors after progression: the missing comparison

In contrast, exposure to PARP inhibitors after progression in the placebo arm was surprisingly low. Only 31 of 113 patients who received subsequent therapy received a PARP inhibitor, despite FDA approval of olaparib for HRR-altered mCRPC midway through trial accrual.

The most plausible explanation is limited access outside the United States, reflecting the trial's global nature. Nevertheless, this has important implications for interpretation.

Had a substantially larger proportion of patients in the control arm received a PARP inhibitor upon progression, TALAPRO-2 would have come much closer to answering a different — and clinically central — question: whether upfront treatment intensification is necessary, or whether sequencing therapies could achieve similar survival while sparing a subset of patients the added haematological toxicity of early combination treatment.

Because post-progression PARP exposure was limited, TALAPRO-2 convincingly establishes the benefit of early PARP use, but stops short of definitively resolving the intensification-versus-sequencing debate.

Quality of life: numerically reassuring, analytically fragile

Quality-of-life analyses show no significant differences between arms, with numerically longer time to deterioration in the intervention arm.

Given the clear excess of adverse events and the absence of censoring data for the completion of the QoL questionnaires, this finding should be interpreted cautiously. Patients with the most significant toxicity are most likely to stop completing questionnaires, thus the information in the KM-plot may mainly harbor the data from patients who experienced the least side effects. This potentially inflates apparent QoL outcomes

As such, QoL data are best considered inconclusive.

Bottom line

Focusing on the HRR-deficient cohort — the basis for FDA approval — TALAPRO-2 is a well-designed trial that convincingly answers its primary question. Upfront talazoparib plus enzalutamide improves overall survival in HRR-deficient mCRPC compared to Enzalutamide alone.

However, toxicity is real and abundant; quality-of-life data are difficult to interpret, and the optimal timing of PARP inhibition remains unresolved (upfront versus delayed).

Among everything covered this year, one study clearly stands out. The trial evaluating Disitamab vedotin plus Toripalimab stands out as the strongest paper of 2025 for me. Not because it was flashy, but because it was methodologically solid: a large and clinically meaningful effect size, transparent and disciplined reporting, and a trial design that aligns closely with the clinical question being asked. In a landscape crowded with underpowered studies and endpoint gymnastics, this one respected both the data and the reader.

It’s also important to be explicit about what this newsletter is—and isn’t. Not every urology or oncology paper published in 2025 was covered yet. Selectivity is unavoidable, and since this is a project in addition to my clinical activity, time is limited. That said, the goal going forward is clear: to stay on top of every major publication in 2026 and continue filtering signal from noise, even when the signal is uncomfortable or disappointing. Don’t hesitate to contact me if you have an idea; guest posts are welcome!

On a more personal note, starting Beyond the Abstract Urology has been one of the most rewarding professional decisions I’ve made this year. I’m genuinely grateful to everyone who reads, subscribes, and challenges interpretations. The only way to grow knowledge on critical appraisal is through interaction with like-minded people and learn from other viewpoints.

A special thank-you goes to Timothée Olivier from The Oncology Shot, whose work and discussions have been particularly influential for me this year—especially around the often-misunderstood concept of censoring, and why getting it wrong quietly undermines so many oncology trials.

2025 was a year of learning more than celebrating. That’s fine. The standards are higher now—and they won’t be lowered in 2026.

High-risk non–muscle-invasive bladder cancer (NMIBC) is treated with Bacillus Calmette–Guérin (BCG) induction followed by BCG maintenance. This approach lowers recurrence, delays progression, and is required by all major guidelines. Event-Free Survival (EFS) is an appropriate clinical endpoint because high-grade recurrence and persistent carcinoma in situ often lead to cystectomy.

CREST investigated whether adding subcutaneous sasanlimab (a programmed cell death protein 1 inhibitor) to BCG could improve outcomes in BCG-naive patients with high-risk NMIBC.

2. Trial design (brief overview)

CREST was a randomized, open-label phase 3 trial that assigned patients 1:1:1 to three treatment strategies:

• Arm A: Sasanlimab + full BCG
  Sasanlimab combined with BCG induction and BCG maintenance per standard-of-care schedules.

• Arm B: Sasanlimab + BCG induction only
  Sasanlimab combined with BCG induction without BCG maintenance

• Arm C: Full BCG alone
  The control arm receiving BCG induction and BCG maintenance, representing established standard of care.

The protocol originally defined two co-primary endpoints:

1. Event-Free Survival (EFS) for Arm A versus Arm C, and

2. Event-Free Survival (EFS) for Arm B versus Arm C.

Later in trial conduct, the Arm B comparison was downgraded to a secondary endpoint without explanation. Investigator-assessed EFS became the sole primary endpoint.

3. Understanding what it means to call CREST “a positive trial”

On the surface, CREST is labeled positive because the revised primary endpoint—investigator-assessed EFS for sasanlimab + full BCG (Arm A) versus full BCG alone (Arm C)—met statistical significance.

However, this interpretation fails to consider two fundamental issues.

A. The original protocol specified two co-primary endpoints, and no reason for demotion was provided

The protocol lists both:

1. Event-Free Survival (EFS) for Arm A versus Arm C, and

2. Event-Free Survival (EFS) for Arm B versus Arm C

as co-primary endpoints.

Only later was the Arm B comparison demoted to a key secondary endpoint.

The manuscript gives no explanation for this change—no methodological, operational, or statistical justification. This silence is important because:

• Arm B uses an induction-only BCG regimen, which is below standard of care.

• Arm B ultimately fails to show benefit (hazard ratio 1.16; 95% CI 0.87–1.55).

• Under a co-primary framework, failure of either comparison means the trial is not positive.

Thus, CREST is “positive” only under the revised analysis plan, not the one originally used to justify the study.

B. Blinded Independent Central Review (BICR) does not confirm significance

The primary endpoint relies on unblinded investigator assessment in an open-label setting. When independently assessed through Blinded Independent Central Review (BICR):

• The hazard ratio remains similar (0.75),

• But the confidence interval crosses 1 (95% CI 0.52–1.06),

• And significance is not achieved (one-sided P = 0.0517).

If central review had been the primary analytic method, CREST would again be negative.

4. Open-label design and vulnerability to bias in carcinoma in situ

Carcinoma in situ (CIS) assessments depend on cystoscopy interpretation, cytology, and biopsy decisions. After the scheduled post-induction confirmation biopsy, further biopsies are performed only at investigator discretion.

Inflammation and BCG-related erythema are common and visually indistinguishable from CIS. In an unblinded setting, subconscious bias is plausible:

• In patients receiving sasanlimab, ambiguous findings may be attributed to inflammation, leading to delayed biopsy, thus elongating EFS.

• In patients receiving standard BCG, the same findings may trigger earlier biopsy.

Because biopsy timing determines the recording of EFS events, such subconscious differences could modestly delay events in Arm A or accelerate them in Arm C. The reported difference in CIS complete response is only about 5%, easily within the range of bias introduced by small variations in biopsy decisions. I want to highlight that this is a hypothetical problem that I cannot prove, and that it is a structural limitation of an open-label design, not investigator misconduct.

5. Toxicity, treatment discontinuation, and the possibility of informative censoring

Sasanlimab markedly increases toxicity:

• Grade ≥3 treatment-related adverse events: 29.1% (Arm A) and 21.8% (Arm B) vs 6.3% (Arm C).

• Any-grade immune-related adverse events: ~43–47% in sasanlimab-containing arms vs 1.4% in Arm C.

Treatment withdrawal due to adverse events was highly imbalanced:

• 131 patients discontinued treatment for toxicity in Arm A versus 37 patients in Arm C.

Although treatment discontinuation does not automatically censor EFS, the Statistical Analysis Plan (SAP) anticipates “informative censoring due to imbalanced dropouts” and requires a sensitivity analysis for this scenario. That analysis is not reported. Given the magnitude of toxicity-driven withdrawal, non-random censoring is a legitimate concern and could subtly inflate the EFS benefit.

6. Ethical and scientific concerns related to omission of BCG maintenance

Guidelines universally mandate BCG induction followed by BCG maintenance for high-risk NMIBC. The CREST protocol acknowledges this. Maintenance reduces recurrence and progression compared with induction alone.

Despite this, one-third of participants were randomized to a regimen lacking maintenance. No compelling rationale is provided. The protocol cites only an “expected improvement” with sasanlimab; the article cites a desire to “reduce the burden of BCG treatment.” Neither is acceptable justification in a phase 3 trial intended to support regulatory approval.

A valid development sequence would involve:

1. First demonstrating that sasanlimab improves outcomes on top of full BCG standard of care, then

2. Testing whether BCG maintenance can be safely reduced in a separate study.

CREST reverses this sequence. The results confirm that omission of BCG maintenance is harmful: Arm B performed worse than standard-of-care BCG alone.

7. Sponsor-controlled classification of immune-related adverse events

While investigators record all adverse events, immune-related adverse events (irAEs) are identified by a combination of prespecified terms and sponsor-led medical review. The final classification—whether an AE is considered immune-mediated—rests with sponsor clinicians.

Given the high rates of immune-related toxicity and the regulatory relevance of irAEs, sponsor-controlled adjudication weakens transparency in the assessment of safety.

8. Missing endpoints essential for clinical interpretation

Several clinically meaningful outcomes are absent from the report:

• Number and timing of radical cystectomies

• Disease-specific survival

• Patterns of post-progression therapy

Without these data, it is unclear whether the modest improvement in EFS corresponds to meaningful bladder preservation or simply delayed detection of events or altered timing of cystectomy. Overall survival is neutral at the current follow-up.

9. Conclusion

CREST shows a modest improvement in Event-Free Survival for sasanlimab added to full BCG, but this signal is not confirmed by blinded review and depends on a mid-trial endpoint change whose rationale is not provided.
The trial’s open-label design, subjective biopsy pathways, substantial toxicity, and missing analyses of informative censoring further weaken confidence in the result.

Most importantly, CREST knowingly randomized patients to a regimen lacking guideline-required BCG maintenance, exposing them to inferior treatment without a compelling scientific justification.

In its current form, CREST does not offer sufficiently robust, ethical, or clinically meaningful evidence to support the routine use of sasanlimab in high-risk non–muscle-invasive bladder cancer.

The trial was a large, global, double-blind, randomized phase III study. PTEN deficiency was centrally defined using an immunohistochemistry cutoff of ≥90% absence of cytoplasmic PTEN staining. Patients were randomized to

• Capivasertib + abiraterone + prednisone + ADT or

• Placebo + abiraterone + prednisone + ADT
  

Crossover was not allowed, as Capivasertib has not demonstrated clinical benefit in the later-line setting.

The primary endpoint was investigator-assessed radiographic progression-free survival (rPFS), and overall survival (OS) was a key secondary endpoint.

At the primary analysis, rPFS favored capivasertib:
median 33.2 vs 25.7 months, HR 0.81, 95% CI 0.66–0.98, p = 0.034.

Interim OS analyses, however, showed no statistically significant benefit (HR 0.90, 95% CI 0.71-1.15, p = 0.401).

At face value, this appears promising. Yet a closer look reveals a picture that is far from practice-changing.

A fragile PFS result, much smaller than the trial anticipated

The reported hazard ratio (HR) for rPFS was 0.81, with a 95% confidence interval of 0.66–0.98. The effect size is substantially smaller than the expected HR=0.70, for which the study was powered, and statistical significance is achieved only because the upper boundary of the confidence interval (0.98) narrowly excludes 1.0. This is a fragile result and far from the robust effect the trial design anticipated.

Put simply:

• The trial shows statistical significance only because the CI barely excludes 1.0.

• The effect size is modest, smaller than expected, and nowhere near the magnitude needed to justify widespread upfront use.

Complicating interpretation, the control arm performed unusually poorly: median rPFS with abiraterone was 25.7 months here, compared with 33 months in LATITUDE’s unselected high-risk population. The authors acknowledge this discrepancy. Whether this reflects differences in patient biology, regional variations, imaging frequency, or other factors is unclear.
What we can say is that if the control arm had performed better, as in LATITUDE, there might not have been a PFS benefit at all.

Toxicity that will matter enormously in real practice

Capivasertib’s toxicity profile is significant, particularly considering this is a front-line therapy for otherwise treatment-naïve patients.

• Diarrhea in 51.9% of patients

  • 14% grade ≥2, meaning 4–6 additional stools per day

  • 31% grade 1, still requiring treatment and affecting daily life

• Hyperglycemia in 38%, including cases of diabetic ketoacidosis (one fatal)

• Rash in 35%, with 12% grade ≥3

These toxicities occurred early and often, leading to high rates of dose interruptions (62.8%) and treatment discontinuations (18.3%).

And these numbers come from a carefully selected population where the majority of patients were ECOG 0. In real-world settings, where ECOG 2 patients are common, tolerability will almost certainly be worse.

PTEN loss as a biomarker: not ready for prime-time

The trial explored stricter PTEN cutoffs (≥95%, ≥99%, 100%). Hazard ratios were numerically more favorable at higher cutoffs; however, the difference is modest to say the least:

Forest plot of rPFS for each subgroup, HR goes from 0,75 to 0,68 with increasing PTEN score

For OS, the picture is even clearer: no subgroup shows a significant effect, and hazard ratios are nearly identical across PTEN thresholds.

Censoring: the missing data that could explain the entire PFS result

The trial does not report censoring distributions for rPFS:

• no cumulative censoring curves,

• no breakdown of reasons for censoring,

• no timing of censoring,

• no sensitivity analyses.

This is a critical omission. When an experimental arm has substantial early toxicity—leading to treatment withdrawal and potential early censoring—and the control arm has minimal toxicity, the risk of informative censoring becomes real (patients who experience no side effects might expect they are on a placebo). This is especially suspicious because of the >50% risk of diarrhea when treated with Capivasertib.

At the same time, placebo recipients who experience very few side effects may suspect they are on a placebo and withdraw early to receive another ARPI outside the trial (“disappointment censoring”).
Reason for this? If they don’t get the potential benefit from the interventional drug, they might opt for an ARPI (enzalutamide or apalutamide) that lacks the need for concomitant prednisone use.

This combination can produce exactly the pattern seen here:

• a modest PFS advantage,

• no OS benefit,

• curves that nearly overlap,

• and wide uncertainty around the clinical relevance.

If the sponsor ever seeks reimbursement based solely on PFS, full censoring transparency will be essential when we use this drug in the upfront setting.

Post-progression therapy: another gap undermining OS interpretation

The publication provides no information on what therapies patients received after progression. In a global mHSPC study spanning 32 countries with uneven access to established mCRPC therapies, this omission makes the OS result impossible to contextualize.

Without knowing:

• who received docetaxel,

• who received second-line ARPIs,

• who received cabazitaxel, PARP inhibitors, radioligand therapy,

• how sequencing differed across arms,

…we cannot determine whether survival equality reflects a true lack of benefit or imbalances in downstream treatment.

A cost profile that is completely out of proportion to the benefit

Capivasertib is extraordinarily expensive:

• UK: £5,850 per month (~£70,000/year)

• Belgium: €6,890 per month (~€82,680/year)

Given modest PFS benefit, no confirmed OS improvement, significant toxicity, and unclear long-term benefit, there is no economic justification.

Ethical considerations: testing first-line without late-line evidence

One feature of the trial was the absence of cross-over, which is methodologically sound because Capivasertib has not yet proven a benefit in a later-line prostate cancer setting. Although the lack of a crossover design is ethically correct, the immediate implementation in the first-line setting may be questioned.

Instead of proving benefit where patients have exhausted established therapies, the drug was moved directly into first-line mHSPC, one of the healthiest, longest-surviving prostate cancer populations.

This reverses the usual ethical progression of oncology drug development, where new agents are vetted in later lines before exposing large, fit populations to toxicity and cost.

All of this is based on the observation that blocking the androgen receptor might enhance PI3K/AKT pathway activity through crosstalk, thereby increasing reliance on this AKT-driven proliferative mechanism.

Conclusion: not a game changer at all

CAPItello-281 reports a statistically significant but clinically modest PFS effect, achieved with a fragile confidence interval and an unexpectedly weak control arm. It shows no OS benefit, has substantial toxicity, lacks censoring or post-progression data, and relies on a biomarker that appears more prognostic than predictive. Its first-line positioning is difficult to justify, its cost is extremely high, and its real-world tolerability is uncertain.

If the final OS analysis remains negative, capivasertib should not be used in the upfront setting. The modest rPFS benefit does not outweigh the toxicity, financial burden, and methodological gaps. At present, nothing in CAPItello-281 supports capivasertib as a meaningful addition to first-line therapy for PTEN-deficient mHSPC.

This study appeared almost silently in The Lancet, with surprisingly little noise on X/Twitter or at the latest European oncology meetings. No viral threads. No panel debates. No “biopsy revolution” headlines. But it deserves them. Because, despite the quiet release, the design is strong, the question is clinically relevant, and the conclusions are both easy to interpret and potentially practice-changing.

What SEXTANT-PRO asks is simple:

In the MRI era, do we still need 12 systematic cores — or can we safely do less?

Trial Overview (Brief Summary)

The SEXTANT-PRO trial is a single-center, prospective, non-inferiority randomized controlled trial comparing MRI-Targeted Biopsy (TB) plus 6-core systematic biopsy (SB) (TB + 6SB) versus classical TB plus 12-core systematic biopsy (TB + 12SB) in biopsy-naïve men with a single MRI-visible lesion (PI-RADS ≥3). The primary endpoint was the detection of clinically significant prostate cancer (csPCa).

Key Findings

• csPCa detection was 54.3% (TB + 6SB) vs 54.8% (TB + 12SB) — virtually identical.

• RD (rate difference) = –0.5%, with a 95% CI: –9.2% to +8.3% → comfortably above the non-inferiority threshold, I will come back on this point below.

• Procedure time, number of cores, cost, pain, and complication burden were all lower in the 6-core strategy.

• Crucially, there was minimal upgrading, and underdiagnosis of csPCa was rare in both groups, supporting oncological safety when reducing systematic cores — but only in MRI-positive patients.

Interpretation – Three Main Points

1. Non-Inferiority Design: Appropriate and Clinically Meaningful

In a non-inferiority trial, the investigators must predefine a non-inferiority margin — this margin represents the maximum acceptable difference by which the new strategy can perform worse than the standard, while still being considered clinically acceptable.

This is crucial:

A trial may be statistically positive for non-inferiority, but still clinically unacceptable if the margin is set too wide.

In other words, the margin reflects judgment rather than pure statistics. If authors allow a very broad margin (e.g., –20%), the trial might technically prove “non-inferiority,” yet the new strategy could still be substantially worse in clinical practice. Therefore, the choice of margin must be scientifically justified and clinically defensible before the trial begins.

The non-inferiority approach is well-suited here — the study asks a pragmatic question:

“Can we safely do less without compromising detection of clinically significant disease?”
In such settings, proving non-inferiority is powerful, because once you show that reduced sampling does not harm detection, every subsequent endpoint naturally favors the less-invasive strategy (time, cost, pain, workflow, histopathology load, etc.).

The rate difference (RD) represents the absolute difference in csPCa detection between the two groups (TB + 6SB minus TB + 12SB). A 95% confidence interval (CI) was calculated around this RD; non-inferiority was established when the lower bound of this CI remained above –15%, indicating that—even in the worst-case statistical scenario—the 6-core approach is not more than 15% inferior to the standard 12-core scheme.

• RD reflects the absolute difference in csPCa detection between groups.

• The chosen non-inferiority margin of –15% is arguably high, but since the lower CI bound was only –9.2%, the result remains statistically solid and clinically acceptable.

• Therefore, the study convincingly shows that less can be safely considered enough in MRI-visible disease.

2. More Than “Non-Inferior” — The 6-Core Strategy Is Operationally Superior

Across secondary outcomes, the 6-core approach consistently outperformed the 12-core scheme:

Thus, non-inferiority in oncologic yield, but superiority in procedural metrics. This aligns with the principle:

Once safety is proven, efficiency becomes the new benchmark.

3. Concern for Underdiagnosis – Addressed but Must Be Framed Carefully

A key anxiety when reducing systematic cores is the risk of missing clinically significant lesions or under-grading tumors. This is addressed in the following table, which I edited to make it clearer:

This table compares the Grade Group (GG) from MRI-targeted biopsy (x-axis) with the final GG after adding systematic biopsy (y-axis).

• X-axis (columns):
  → Grade Group from MRI-targeted biopsy alone (MRI-TB)
  → Shows what we would have concluded if no systematic cores were added.

• Y-axis (rows):
  → Grade Group after combining MRI-TB + systematic biopsy (SB)
  → Represents the final biopsy result used to guide treatment.

So every cell in the grid shows how many patients were reclassified after adding systematic cores — either upgraded, downgraded, or unchanged.

Why Only the Red Zones Truly Matter

A red cell means:

The MRI-targeted biopsy alone underestimated the tumor — and systematic sampling corrected it.

This reflects potential underdiagnosis — and is the key reason people are reluctant to reduce SB cores. But in this trial, that risk did not increase when reducing from 12 to 6 cores.

Blue zones represent downgrades — SB finds a lower grade than MRI-TB.
But in modern practice, this does not influence treatment, because:

The highest Grade Group determines management — not the lowest.

This means downgraded cores have little to no clinical weight in MDTs, treatment planning, or risk stratification.

Conclusion — A Clinically Safe Simplification

For MRI-positive patients with a single lesion, routine use of 12-core systematic sampling may no longer be required. The 6-core strategy did not compromise detection of clinically significant prostate cancer and did not increase clinically meaningful upgrading. These results support a simplified biopsy pathway where MRI guides targeting and systematic sampling are scaled down, rather than discarded.

Once oncologic safety is preserved, the advantages of core reduction become clear:

• Shorter procedure time

• Less discomfort and anxiety for the patient

• Lower complication risk (bleeding, pain, urinary symptoms)

• Reduced pathology workload and cost

• More streamlined workflow in busy biopsy clinics

The evidence does not yet support reducing systematic cores in MRI-negative men.
But for MRI-visible disease, less is more.

Why This Trial Matters

It must be one of the most common questions I get in the outpatient clinic:

“Are there no pills to prevent me from getting prostate cancer?”

And once low-risk disease has been diagnosed, the question quickly becomes:

“What can I do to prevent progression?”

For more than a decade, metformin has been one of the brightest hopes among repurposed metabolic agents—a familiar, inexpensive drug with plausible anticancer effects based on epidemiology and preclinical work. If it were going to help anywhere, the active surveillance space seemed like the right setting.

But if there was ever any hope for metformin to delay progression, that hope is now buried with the publication of the MAST study.

The MAST Trial

The design of the Metformin Active Surveillance Trial (MAST) was commendably straightforward:

• Population: Low-risk localized prostate cancer

• Randomization: 1:1 to metformin (850 mg BID) vs placebo

• Design: Multicenter, double-blind

• Exclusion: Type 1 or 2 diabetes

• Assessments: Confirmatory biopsy at 18 and 36 months, or biopsy for cause—mirroring real-world active surveillance

• Primary endpoint: Time to progression (therapeutic or pathologic)

One notable omission from the protocol is the lack of routine prostate MRI, something that has become standard in contemporary AS pathways.

A total of 408 patients were randomized:

• 203 to placebo

• 205 to metformin

The cohort was overwhelmingly White (93.6%), limiting extrapolation to other racial groups—especially relevant given the higher prostate cancer risk and different tumor biology in men of African ancestry.

Results

At a median of 36 months, there was no difference in progression between arms.
Among 144 patients who progressed:

• 70 were in the metformin group

• 74 in the placebo group

The hazard ratio for progression with metformin was essentially null: HR 1.09
Time-point analyses reveal similarly overlapping curves, with no indication of a clinically relevant benefit.

The BMI Signal

A prespecified subgroup analysis stratified by BMI revealed an unexpected finding.
Among patients with a BMI ≥ 30, those receiving metformin had increased pathologic progression compared with placebo. This signal was not observed in men with a BMI < 30.

It is challenging to determine whether this interaction is due to biology, chance, or the unique metabolic and immune microenvironment associated with obesity.
Nevertheless, it’s a clear signal: don’t use metformin to slow progression, especially not in obese patients with low-risk prostate cancer.

Conclusion

Taken together with the previously negative data on adding metformin to ADT in metastatic hormone-sensitive disease, this trial likely marks the end of the road for metformin in prostate cancer.

• It does not reduce the risk of progression in men with low-risk disease on active surveillance.

• And in obese patients, outcomes may be worse when treated with metformin.

So when patients ask what they can do to prevent developing clinically significant prostate cancer, the answer remains simple:

Maintain a healthy lifestyle, and stay fit for treatment if you need it in the future.

With this, we can bury the hope for metformin in prostate cancer.
And that, too, is progress.
Strong randomized science can—and should—close chapters as confidently as it opens them.

💡 It’s rare to see a clinical trial that delivers both spectacular results and solid design.
The phase 3 RC48-C016 study, evaluating disitamab vedotin (DV) plus toripalimab in HER2-expressing advanced urothelial cancer, is one of those uncommon cases.
Not only are the efficacy results extraordinary, but the trial’s methodology withstands careful scrutiny — a combination we seldom encounter in oncology.

Thanks for reading Beyond The Abstract: Urology! Subscribe for free to receive new posts and support my work.

A Trial That Gets the Fundamentals Right

RC48-C016 enrolled 484 patients with HER2 IHC ≥ 1+ urothelial carcinoma who had not received prior systemic therapy for metastatic disease.
Participants were randomized 1:1 to receive disitamab vedotin (DV) + toripalimab every two weeks or gemcitabine + platinum (cisplatin or carboplatin) every three weeks.

At the time of study initiation in mid-2022, gemcitabine–platinum chemotherapy was the accepted global standard of care.
The enfortumab vedotin–pembrolizumab combination, now considered the standard of care for first-line treatment, was only approved in December 2023, long after enrollment began.
Moreover, in China, where the trial was conducted, DV + toripalimab was only reimbursed for this indication starting in January 2025, ensuring that patients in the control group did not have off-trial access to the experimental therapy during the study period.

“This was not an outdated control arm — it reflected real-world practice at the time of trial initiation.”

A Clinically Meaningful Endpoint

The co-primary endpoints — progression-free survival (PFS) and overall survival (OS) — were both clinically robust and clearly prioritized.
The results were remarkable:

• Median PFS nearly doubled (13.1 vs 6.5 months, HR 0.36, P < 0.001).

• Median OS doubled (31.5 vs 16.9 months).

Benefits were consistent across prespecified subgroups, including patients with low (IHC 1+) HER2 expression.
Such a magnitude of effect is rarely seen in urothelial cancer.

Early Censoring: Two Plausible Explanations

One of the most intriguing aspects of the RC48-C016 data is the pattern of early censoring in the PFS curves.
Visual inspection suggests a higher rate of censoring in the chemotherapy arm, particularly during the first few months of treatment.

Estimated visually from Kaplan–Meier curves.

Two explanations are possible — and both are reasonable.

1️⃣ Toxicity-driven censoring.
The first hypothesis is that more patients in the control arm discontinued chemotherapy early due to toxicity.
Those censored early tend to be frailer, meaning the remaining at-risk group is biased toward fitter patients.
This dynamic would actually underestimate the true PFS benefit of DV + toripalimab — since censored, more vulnerable control-arm patients would likely have progressed earlier had they stayed on treatment.

2️⃣ Disappointment-driven withdrawal.
A second, equally plausible explanation is patient disappointment.
Because the study was open-label, some patients randomized to chemotherapy may have left early, hoping to find immunotherapy elsewhere.
Such non-random withdrawal could inflate censoring rates in the control arm — but its effect on PFS would move in the opposite direction compared to toxicity-driven censoring.
If disappointed patients left early, the control arm would disproportionately retain frailer patients with fewer treatment options, while the fitter, more motivated ones sought treatment elsewhere.
This would tend to worsen the apparent PFS in the control arm, rather than underestimate the benefit of the experimental combination.

Both mechanisms are plausible.
Yet, given the trial’s setting, broad access to immunotherapy outside the study was limited — enfortumab vedotin was not available in China during the trial, and DV + toripalimab was not reimbursed until 2025.
While some well-resourced patients might have sought private treatment, the majority would still have had access only to chemotherapy.

Access to Post-Progression Therapy

Another subtle but important detail lies in post-progression care.
Only 77% of patients in the control arm who progressed on chemotherapy received subsequent immunotherapy — surprisingly low in a study testing an immune-based regimen.
Given that all participants were fit enough for inclusion, this figure suggests real-world limitations in treatment access.

Access was also relatively modest in the experimental arm, reflecting broader constraints in availability and reimbursement.
Such limitations can amplify overall survival differences between treatment arms: when few patients receive effective therapy after progression, the group with longer PFS — here, DV + toripalimab — naturally enjoys a greater OS advantage.

HER2 Biology and Broader Mechanism

Roughly one in five patients had HER2 IHC 1+ tumors, yet they appeared to benefit as much as those with higher expression.
This raises the question of whether DV + toripalimab acts strictly as a HER2-targeted regimen, or whether it represents a broader synergy between the antibody–drug conjugate and immune checkpoint blockade.

Global Context and Control Arm Performance

The main limitation of the trial is its limited global applicability.
All patients were enrolled in China, where access to subsequent therapies and health system structures differ from Western practice.
Until multinational validation is available, the global generalizability of these results remains to be established.

Why This Trial Stands Out

Despite these nuances, RC48-C016 remains an example of solid trial conduct.
The control arm was appropriate, the endpoints were meaningful, the design was ethically sound, and post-protocol therapy was clearly reported.
The magnitude of benefit is extraordinary, yet the results remain methodologically credible — a balance that’s increasingly rare in oncology trials.

Takeaway

The RC48-C016 trial of disitamab vedotin plus toripalimab achieves what few studies manage:
truly impressive efficacy supported by a design that earns confidence rather than skepticism.

The discussion around censoring and post-progression therapy shows how even robust trials are influenced by patient behavior, system constraints, and access.

Replication in multinational cohorts will tell us how far these findings reach.
For now, RC48-C016 sets a new benchmark for both clinical impact and trial design integrity in urothelial cancer.

Poly(ADP-ribose) polymerase (PARP) inhibitors have reshaped the management of metastatic castration-resistant prostate cancer (mCRPC) with homologous recombination repair (HRR) defects, particularly in patients with BRCA1/2 mutations. Yet resistance is inevitable.

The AMPLITUDE trial tested whether introducing PARP inhibition earlier—combining niraparib with abiraterone acetate and prednisone (AAP) in metastatic castration-sensitive prostate cancer (mCSPC)—could improve disease control and delay progression in HRR-deficient tumors.

Trial design

• Design: Phase 3, randomized, double-blind, placebo-controlled

• Population: 696 men with mCSPC and deleterious germline or somatic HRR alterations

• Intervention: Niraparib (PARP inhibitor) + AAP vs placebo + AAP

• Primary endpoint: Investigator-assessed radiographic progression-free survival (rPFS)

• Key subgroups:

  • BRCA subgroup (n = 387)

  • HRR effector subgroup (n = 456)

  • Intention-to-treat (ITT) population (n = 696)

Results

Primary endpoint (investigator-assessed rPFS) positive in favor of niraparib:

• BRCA subgroup: HR = 0.52 (95% CI 0.37–0.72; p < 0.0001); median not reached vs 26 months

• HRR effector subgroup: HR = 0.57 (95% CI 0.42–0.77; p = 0.0003)

• ITT population: HR = 0.63 (95% CI 0.49–0.80; p = 0.0001)

These findings were consistent across prespecified subgroups and confirmed by blinded central review.

Kaplan-Meier estimates for PFS - a: BRCA subgroup / b: HRR subgroup / c: ITT subgroup

Secondary endpoints:

• Time to symptomatic progression: HR = 0.50 (95% CI 0.36–0.69; p < 0.0001)

• Overall survival: Immature, HR = 0.79 (95% CI 0.59–1.04); not statistically significant

Kaplan-Meier estimates for Overall survival - c: BRCA subgroup / d: ITT subgroup

Safety:
Grade 3–4 adverse events occurred in 75% of the niraparib + AAP group, compared with 59% in the AAP-alone group.

• Most common: anemia (29% vs 5%) and hypertension (27% vs 18%)

• 25% required transfusion for anemia

• Serious adverse events: 39% vs 28%

• Treatment-emergent deaths: 14 vs 7

• Quality of life (FACT-P) showed an initial decline in the niraparib arm, with recovery only after several cycles.

Patient-reported outcomes for Quality of Life: blue line shows initial worse QoL in the niraparib arm.

Critical perspective

While AMPLITUDE achieved its primary endpoint, several aspects of its design and interpretation limit its immediate clinical impact.

1. Lack of crossover — a scientific and ethical flaw:
   The trial did not allow crossover to niraparib for patients in the control arm who progressed, despite OS benefit of PARP inhibitors in later disease stages*.
   This is not only an ethical concern. It also prevents answering the crucial sequencing question: is there true value in giving niraparib early, or could we safely reserve it for progression? If crossover was included in the trial, we could look at the difference in OS between patients who received niraparib immediately and those who received it upon progression.

   
   * Admittedly, the OS benefit is compared to the ARPI switch, which is a weak comparator (PROfound trial).
   

2. No overall survival benefit — and toxicity matters:
   There is currently no OS benefit, and if we intend to move a toxic therapy earlier in the treatment course—thereby exposing far more patients—we must demand an OS benefit, not just improved PFS.
   PFS alone is insufficient to justify a shift in standard practice with added toxicity.

3. Serious adverse events and quality of life:
   The combination carries a substantial toxicity burden, with far more serious adverse events and a measurable initial decline in quality of life. For a first-line therapy, this trade-off is nontrivial and clinically relevant.

4. Potential bias through differential censoring:
   Detailed censoring data were not reported, but inspection of the Kaplan–Meier curves suggests more early censoring in the control arm (cfr number of vertical ticks). This could reflect disappointment-driven withdrawal: patients perceiving a lack of side effects may suspect they’re on placebo and drop out early. Those who leave are often healthier and better connected, leaving behind more frail participants—potentially skewing the control curve unfavorably.
   This remains a hypothesis, but it deserves verification with full censoring data.

   

   Kaplan-Meier estimates for PFS in the ITT group; the curves in the green box show more vertical ticks on the red line (control arm), suggesting increased censoring due to disappointment.

5. Post-progression therapy imbalances:
   Only 56% and 72% of patients in the experimental and control arms, respectively, received any life-prolonging therapy after progression.
   These figures are unusually low for mCSPC and likely reflect limited drug access in low- and middle-income countries participating in the trial.
   Such disparities can dilute overall survival differences and complicate interpretation.

Unresolved mysteries

1. Investigator-assessed versus blinded central PFS:
   It is difficult to understand why investigator-assessed rPFS was selected as the primary endpoint, while blinded central review was designated as the secondary endpoint. If the trial already funds an independent radiologic review, why not make that the primary measure? If investigator-assessed outcomes appear stronger than the blinded readout, this paradoxically highlights potential bias among trial investigators.

2. The post-progression data raise several questions, but the one I can’t get my head around is the fact that 11% of patients in the niraparib arm received another PARP inhibitor (olaparib) after progression. This makes no biological sense, so why would a patient opt for this, or would the treating physician advise this?

Conclusion

AMPLITUDE demonstrates a statistically significant improvement in investigator-assessed rPFS with niraparib plus AAP in HRR-deficient mCSPC, confirming biological activity in BRCA1/2-altered disease.

However, the trial’s lack of crossover, the absence of an overall survival benefit, substantial toxicity, and early decline in quality of life make it challenging to support early PARP inhibition as standard of care.
Without evidence that upfront niraparib leads to better survival than its use after progression, exposing all patients to its toxicity cannot be justified.

In essence: AMPLITUDE shows biological promise but falls short of clinical justification. Until an overall survival advantage—or a crossover design—demonstrates that early PARP inhibition is truly superior, niraparib should not be used in the hormone sensitive setting outside of a clinical trial.

Trial Design

The EMBARK trial enrolled men with biochemical recurrence (BCR) after definitive local therapy — radical prostatectomy, radiotherapy, or both — without evidence of disease on conventional imaging (i.e., no metastases on CT or bone scan). This means no PSMA, and at enrollment, patients were not receiving androgen deprivation therapy (ADT).

Patients were randomized into three groups:

1. Enzalutamide + leuprolide

2. Enzalutamide monotherapy

3. Placebo + leuprolide

Importantly:

• Treatment was suspended if PSA dropped below 0.2 ng/mL, and restarted upon PSA relapse to a predefined level.

• No crossover was allowed; patients on placebo + leuprolide could not receive enzalutamide within the trial.

• No PSMA PET was used — appropriate, as the trial began in 2015. I will not get into the discussion of we should or should not use PSMA in this setting.

Outcomes

The primary endpoint was metastasis-free survival (MFS) in the combination arm compared with leuprolide alone. This endpoint had already been reported as positive.

At 5 years:

• Enzalutamide + leuprolide: 87.3% (95% CI, 83.0–90.6)

• Leuprolide alone: 71.4% (95% CI, 65.7–76.3)

• Enzalutamide monotherapy: 80.0% (95% CI, 75.0–84.1)

The combination significantly improved MFS (HR for metastasis or death, 0.42; 95% CI, 0.30–0.61; p<0.001).

In the newly reported data, overall survival (OS) was also longer with the combination:

At 8 years:

• Enzalutamide + leuprolide: 78.9% (95% CI, 73.9–83.1)

• Leuprolide alone: 69.5% (95% CI, 64.0–74.3)

• Enzalutamide monotherapy: 73.1% (95% CI, 67.6–77.9)

The hazard ratio for death was 0.60 (95% CI, 0.44–0.80; p<0.001).

There were no new safety signals, and previous analyses showed that combination therapy did not negatively affect health-related quality of life. Interestingly, sexual-activity–related quality of life was better with enzalutamide monotherapy than with leuprolide alone.

Why was this trial conducted?

Enzalutamide is already established across multiple stages of prostate cancer: mCRPC, non-metastatic CRPC, and hormone-sensitive metastatic disease. While the sponsor will frame EMBARK as an effort to further improve survival and quality of life, it’s clear that commercial strategy also plays a major role.

Each time a drug moves earlier in the disease continuum, the treatable population expands — and so does the market. With the metastatic settings already covered, the next frontier was obvious: non-metastatic biochemical recurrence. Moreover, this niche currently lacks competition from other androgen receptor pathway inhibitors (ARPI) like apalutamide (Erleada) or darolutamide (Nubeqa), giving enzalutamide a potential first-mover advantage.

That said, credit where it’s due — the decision to allow treatment suspension for patients achieving deep PSA responses (<0.2 ng/mL) deserves acknowledgment. This approach offers patients a treatment break, despite providing no financial upside for the sponsor.

Critical Reflections

The OS benefit reported here is important — especially since these are asymptomatic, non-metastatic patients. Starting systemic therapy in this context requires clear justification: if we compromise quality of life, which we will with hormonal therapy, it must be in exchange for a tangible survival benefit.

However, two key concerns remain:

1. Post-progression treatment data are missing.
   Perhaps the most significant limitation — and one that has become frustratingly common — is the absence of detailed post-progression treatment data.
   Once patients progress on their assigned arm, we need to know what therapies they received and if there is an inbalance in post-protocol treatments between the arms.

   Without these details, the observed OS advantage becomes much harder to interpret.
   The absence of this data always raises eyebrows. Sponsors know it’s essential for interpretation, yet often withhold it — either because it weakens the conclusion, or because it wasn’t properly collected. Neither explanation is reassuring.

2. No crossover.
   Patients in the leuprolide-alone arm who progressed could not receive enzalutamide within the trial — even though enzalutamide is a well established treatment option in the castration-resistant setting.
   This design element virtually guarantees an OS advantage for the experimental arm — not necessarily because the drug is more effective in an earlier line, but because the control group was denied optimal post-progression therapy.
   
   Geographic disparities compound this problem. EMBARK enrolled patients across 17 countries, including three (Brazil, Poland, and Slovakia) where reimbursement for ARPIs remains limited.
   In those regions, patients progressing in the control arm may have been limited to chemotherapy or best supportive care.
   If a significant number of patients refuses chemotherapy, this would inevitably depress survival in the control arm, exaggerating the relative OS benefit.

   And since we have no data on post-protocol care (cfr above) this remains speculation.

Conclusion

The EMBARK trial confirms that enzalutamide plus ADT significantly improves both metastasis-free and overall survival in men with high-risk biochemical recurrence after local therapy. Clinically, it’s a meaningful advance — but scientifically, some caution is warranted.

Before we rush to change practice, we should demand transparent post-progression data and consider the implications of restricting crossover in trials of life-prolonging therapies.

Ultimately, EMBARK extends the reach of enzalutamide into yet another disease state — but whether it truly transforms outcomes, or simply expands the drug’s territory, depends on how we interpret the data beyond the abstract.

Trial Summary

POTOMAC investigated the use of Durvalumab (a PD-L1 inhibitor) in BCG-naive, high-risk non–muscle-invasive bladder cancer.
Patients were randomized into three arms:

1. BCG for 2 years (control group)

2. Durvalumab for 1 year + BCG for 2 years

3. Durvalumab for 1 year + BCG induction only

The primary endpoint was disease-free survival (DFS).
Key secondary endpoints included:

• Proportion of patients alive and disease-free at 2 years

• Overall survival (OS) at 5 years

• Safety and tolerability

• Health-related quality of life

Results

A total of 1,018 patients underwent randomization.
For DFS assessment, patients were followed with cystoscopy, cytology, and CT urography every 3 months until month 36, then every 6 months.

1. Disease-Free Survival

• The risk of recurrence or death was 32% lower in the Durvalumab + BCG induction and maintenance arm than in the control arm:
  HR 0.68 (95% CI 0.50–0.93); p = 0.015.

• The estimated proportion of patients alive and disease-free at 24 months was:

  • 86.5% (95% CI 82.2–89.8) with Durvalumab + BCG induction and maintenance

  • 81.6% (95% CI 76.9–85.3) with BCG alone

So, a statistically significant — but modest — difference.
Before we do the deep dive, let’s look at the rest of the results.

2. Durvalumab + BCG Induction Only

No difference in DFS compared to BCG induction + maintenance:
HR 1.14; p = 0.35.

3. Overall Survival

No significant difference between Durvalumab + BCG (I & M) and BCG alone:
HR 0.80 (95% CI 0.53–1.20).

4. Adverse Events

As expected, patients receiving Durvalumab experienced more adverse events, consistent with its known toxicity profile.
The authors describe side effects as tolerable and manageable.

Authors’ Conclusions

According to the investigators:

• Adding Durvalumab to BCG (I & M) improves disease-free survival.

• There is no detriment to overall survival.

• The safety profile is consistent with known data for each therapy; side effects were tolerable and manageable.

What They’re Not Telling Us

Now we go “Beyond the Abstract”

Let’s look more closely at the DFS Kaplan–Meier curve — there’s a story in the censoring pattern, let me know if you agree or not.

• Red arrows: numbers between brackets, they represent the number of patients being censored.

• Green square: lots of censoring here (vertical ticks), which is normal and what we call “late censoring”. Basically, there is a lot of censoring near the end of the curve (or follow-up period) because a significant number of patients joined the trial later, meaning they are still in the trial without having presented the event of interest (disease recurrence). In other words, they are censored because we don’t know what will happen to them later in the curve.

• Blue square: In the early part of the curve (where the lines split), there appear to be more censoring events in the Durvalumab arm.
  This is confirmed by the numbers at the red arrows: significantly more early censoring in the Durvalumab group.

Why does this matter?
Early censoring often reflects patients who drop out due to toxicity. Frail patients are more likely to discontinue the treatment or skip follow-up cystoscopies and CT scans. That means the healthier patients remain in the Durvalumab arm’s analysis, introducing bias and artificially improving DFS results.

You can see a similar pattern in early censoring in the Durvalumab + BCG induction only vs. control curves, confirming the hypothesis that there is increased censoring due to toxicity.

Spinning the OS Result

The authors frame the absence of an overall survival difference as good news:

“There is no detriment to overall survival with the addition of Durvalumab to BCG.”

Let’s be clear — being happy that a drug didn’t kill more patients is not a win.
Durvalumab causes more side effects, carries a significant financial burden, and does not improve overall survival. That’s the real takeaway.

Unanswered Questions

The paper provides no details on:

• What happened to patients after progression

• Whether there were differences in radical cystectomy rates

• Any delay in progression to metastatic disease

• Subsequent treatments — particularly whether patients who received Durvalumab (a PD-L1 inhibitor) were later treated with another checkpoint inhibitor

These post-progression details are crucial to understanding the broader impact of Durvalumab on overall survival.

Conclusion

Durvalumab + BCG induction and maintenance increases DFS modestly compared to BCG alone — but the effect size is likely inflated by early censoring due to toxicity.

Importantly, there is no overall survival benefit, despite increased toxicity and cost.
And remember: the trial only included ECOG 0–1 patients — in real life, the toxicity bias could be even more pronounced.

So while AstraZeneca celebrates that Durvalumab didn’t increase mortality, we should ask the tougher question:
Why expose patients to a toxic, expensive drug that doesn’t help them live longer?

• What is it all about?

The recently published CONTACT-02 trial in The Lancet Oncology evaluated the combination of cabozantinib plus atezolizumab versus an androgen receptor pathway inhibitor (ARPI) switch in patients with metastatic castration-resistant prostate cancer (mCRPC) who had progressed on a prior ARPI.

The trial reported a statistically significant improvement in progression-free survival (6.3 vs 4.2 months; HR 0.65), but no difference in overall survival (14.8 vs 15.0 months).

Despite these modest results, the authors concluded that the combination “could be a useful option” for this patient population. I believe that interpretation overstates the clinical relevance of the data.

• A modest gain, at a high cost

While statistical significance is often mistaken for clinical progress, the absolute benefit in progression-free survival—barely two months—was modest, particularly in a disease setting where treatment goals extend beyond short-term disease control.

This limited efficacy came with substantial toxicity: nearly half of the patients in the cabozantinib–atezolizumab arm experienced grade 3 adverse events, and the regimen inevitably adds considerable financial burden.

Without a survival advantage, it’s difficult to justify the claim that this represents a meaningful new option for patients.

• The problem with the control group

Perhaps more concerning than the marginal efficacy is the choice of control arm.

Patients enrolled in CONTACT-02 had high-risk disease biology, given their rapid progression to castration resistance and the presence of visceral or nodal metastases. In such a context, an ARPI switch—after progression on a prior ARPI—is a weak comparator.

Cross-resistance between ARPIs such as abiraterone and enzalutamide is well documented, and clinical guidelines already discourage serial ARPI use.

Instead, patients previously treated with a single ARPI should have received docetaxel, and those who had already received both an ARPI and docetaxel should have been treated with cabazitaxel.

The CARD trial (NEJM 2019)—published well before CONTACT-02 began—provided level 1 evidence that cabazitaxel is superior to a second ARPI in this setting.

The fact that 98% of patients in the CONTACT-02 control arm ultimately received chemotherapy underscores that its inclusion as a comparator was not only feasible but also appropriate (Supplementary Table S6).

The authors justified the ARPI switch by citing real-world data and the absence of direct head-to-head evidence between docetaxel and ARPI switch.

However, real-world practice reflects accessibility and physician preference—not efficacy.

Repeating flawed comparators across trials does not validate the design; it amplifies its limitations.

• The bigger picture

Taken together, CONTACT-02 demonstrates a narrow benefit in a highly selected population, at the cost of toxicity and without extending life.

The suboptimal control group undermines the interpretability of the findings.

While exploratory, this trial should not alter clinical practice at this moment.

Instead, it serves as a reminder that trial design must align with contemporary standards of care.

• What the authors said in reply

In their published reply, the investigators defended the ARPI switch as an appropriate control, citing prior study designs, mechanistic arguments, and concerns about taxane toxicity.

They also highlighted a 35% reduction in the risk of progression or death (HR 0.65) as a meaningful outcome, and pointed to an exploratory overall survival signal among patients with liver metastases.

However, these arguments are unconvincing.

1. The reliance on earlier study precedents does not justify perpetuating a weak control, especially when superior options have long been validated.

2. The subgroup finding in liver metastases—while interesting—is exploratory and statistically fragile.

And while toxicity may have been “manageable,” its frequency and severity cannot be ignored when the regimen fails to improve survival.

Final thoughts

CONTACT-02 reflects a recurring issue in prostate cancer trials: designs that prioritize regulatory endpoints over clinical relevance.

When modest PFS gains are achieved at the expense of toxicity, cost, and interpretability, we must ask whether such results truly advance patient care.

The conversation sparked by this trial—and the subsequent exchange of correspondence—highlights the need for stricter alignment between trial design, evidence-based comparators, and meaningful outcomes that actually impact how we treat our patients.

Leave a comment