A new standard for cisplatin-ineligible muscle-invasive bladder cancer: Landmark results from KEYNOTE-905

SUMMARY Cisplatin-based neoadjuvant chemotherapy, followed by radical cystectomy (RC) and pelvic lymph-node dissection (PLND), is the standard of care for muscle-invasive bladder cancer (MIBC), yet 50% of the patients are cisplatin ineligible as per Galsky criteria,1 resulting in poor outcomes with upfront RC + PLND alone. KEYNOTE-905 (EV-303; NCT03924895) is the first phase III randomized trial in cisplatin-ineligible or cisplatin-declining patients with MIBC (cT2–T4aN0M0 or cT1–T4aN1M0). Initiated in 2019 as a two-arm study (perioperative pembrolizumab plus RC + PLND vs. RC + PLND alone), KEYNOTE-905 added a third enfortumab vedotin (EV) plus pembrolizumab arm in 2020. Following discontinuation of the pembrolizumab only arm in 2022, randomization continued 1:1 between EV plus pembrolizumab (P) and surgery alone. The pembrolizumab arm received neoadjuvant pembrolizumab 200 mg IV Q3W for 3 cycles, RC + PLND, then 14 adjuvant cycles. The EV + P arm received neoadjuvant EV 1.25 mg/kg on days 1 and 8 Q3W plus pembrolizumab for 3 cycles, RC + PLND, then 6 adjuvant EV cycles and 14 pembrolizumab cycles. Controls underwent RC + PLND alone. The primary endpoint was event-free survival (EFS), including progression precluding surgery, residual or recurrent disease, metastasis, or death; assessed by blinded independent central review. Key secondary endpoints included overall survival (OS) and pathological complete response (pCR). A total of 170 and 174 patients were enrolled in the EV + P and control arms, respectively. Median ages were 74 and 72.5 years; (Eastern Cooperative Oncology Group) ECOG performance status 2 was present in 12.4% and 14.9%, indicating balanced baseline characteristics. Most patients had T3/T4aN0 disease on TURBT pathology and imaging. EV + P significantly improved EFS, with a median not reached versus 15.7 months in controls (hazard ratio HR 0.40; 95% confidence interval CI 0.28–0.57; P < 0.0001). OS was also improved, with a median not reached versus 41.7 months (HR 0.50; 95% CI 0.33–0.74; P = 0.0002). pCR (pT0N0 on RC + PLND) was significantly higher with EV + P than control (57.1% vs. 8.6%), an absolute improvement of 48.3% (P < 0.000001). In the EV + P arm, pCR was associated with superior EFS (median not reached vs. 41.2 months; HR 0.43). Even without pCR, EFS favored EV + P (26.1 vs. 14.2 months; HR 0.76). Treatment-emergent adverse events (TEAESs) were dominated by pruritus (47.3%) and alopecia (34.7%), whereas anemia and incidental prostate cancer were most common during the surgical phase. Grade ≥3 TEAEs were reported in 71.3% in the EV + P arm, compared with 45.9% in the control arm. Dose reduction of EV due to TEAEs were required in 16.8% of the patients. COMMENTS After 29.1 months of follow-up, the EV-302 trial2 showed that EV + P achieves approximately double the progression-free survival as compared to chemotherapy in both cisplatin-eligible and ineligible patients with untreated locally advanced or metastatic urothelial carcinoma. These findings provide the scientific rationale for KEYNOTE-905, which targets the major unmet need in the cisplatin-ineligible muscle-invasive cohort, a population with historically poor survival. Cisplatin-ineligible patients frequently have substantial comorbidities that confer competing mortality risks, thereby making EFS a clinically appropriate primary endpoint in this population. KEYNOTE-905 is the first phase 3 trial to show significant EFS and OS benefits across all the subgroups, including age, ECOG status, PD-L1 (Programmed Death-Ligand 1) CPS, and tumor stage. Discontinuation of the pembrolizumab only arm did not affect power or validity, as the study was repowered for a 1:1 comparison between EV + P and RC, with preserved alpha control and randomization. Data from the discontinued arm remain descriptive and hypothesis generating. Surgical feasibility was maintained with EV + P, with 87.6% of the patients undergoing RC compared with 89.7% in the control arm. These rates are similar to those reported in the cisplatin-eligible NIAGARA trial.3 The treatment’s safety profile was consistent with known toxicities of EV and pembrolizumab and did not increase the perioperative complications or compromise the ability to undergo RC. To date, CheckMate 2744 led to the FDA approval of adjuvant nivolumab for cisplatin-ineligible patients with high-risk MIBC, based on a significant disease-free benefit; however, only 16.7% of the patients in the control arm received adjuvant nivolumab due to issues with local access and reimbursement. Only 4.7% of the patients in the EV + P arm required subsequent systemic therapy, compared with 26.4% in the control group. Notably, limited access to EV + P upon relapse likely influenced the OS outcomes in the control arm. The findings also raise an important clinical question regarding optimal first-line therapy for metastatic disease following prior perioperative EV + P exposure. Emerging ctDNA-based minimal residual disease assays and molecular biomarkers may help identify patients who can safely omit adjuvant therapy, particularly those with negative ctDNA or pCR.5 Incorporating biomarker-driven strategies in future trials will be the key to refiningperioperative treatment selection and optimizing outcomes for cisplatin-ineligible MIBC. Future trials should integrate these biomarkers to optimize treatment selection and refine perioperative therapeutic strategies. Financial support and sponsorship: Nil. Conflicts of interest: There are no conflicts of interest.