Circulating tumor DNA–guided adjuvant immunotherapy in muscle-invasive bladder cancer

The rationale for adjuvant therapy after radical cystectomy is straightforward: to eradicate occult micrometastatic disease and reduce recurrence and cancer-specific mortality in patients who appear disease-free on conventional imaging. In muscle-invasive bladder cancer (MIBC), this principle is particularly relevant given the aggressive biology of the disease and the well-recognized limitations of radiographic staging. Historically, adjuvant treatment decisions have relied on pathologic findings at cystectomy, most notably pT stage and nodal status, often in the context of prior neoadjuvant chemotherapy. ¹ While this framework has improved outcomes at the population level, it lacks precision for individual patients. Pathologic stage is an imperfect surrogate for residual systemic disease: some patients with relatively low-stage tumors experience early relapse, whereas others with adverse pathologic features are cured by surgery alone. As a result, stage-based selection inevitably leads to overtreatment of patients who do not require systemic therapy and undertreatment of those who do.

These limitations have driven interest in biologically informed strategies that directly measure residual disease burden. Circulating tumor DNA (ctDNA) has emerged as a compelling marker of molecular residual disease across solid tumors. ² In bladder cancer, multiple retrospective and prospective studies have consistently demonstrated that postoperative ctDNA positivity is strongly associated with recurrence and death. This association is now supported by a recent meta-analysis of more than 1100 patients with MIBC, which reported a four- to six-fold increase in the risk of recurrence, progression, and mortality among ctDNA-positive patients across both pre- and post-treatment settings. ³ Importantly, longitudinal analyses showed that ctDNA clearance over time was associated with substantially improved outcomes, making ctDNA a dynamic biomarker of residual disease rather than a static prognostic factor. ³

In the December 2025 issue of the New England Journal of Medicine, Powles et al. report the results of IMvigor011, the first large, prospective, randomized trial to formally test a ctDNA-guided adjuvant strategy in MIBC. ⁴ In this phase 3, double-blind study, patients without radiographic evidence of disease after radical cystectomy entered a surveillance phase with serial ctDNA testing for up to one year. Patients who were found to be ctDNA-positive at any point during surveillance and met eligibility criteria were randomized in a 2:1 ratio to receive adjuvant atezolizumab or placebo, whereas patients who remained persistently ctDNA-negative were followed with protocol-defined observation without adjuvant therapy. Patients with radiographic recurrence identified prior to treatment assignment were excluded. This design differs from prior adjuvant immunotherapy trials ⁵ by restricting treatment to patients with molecular evidence of residual disease rather than relying on pathologic risk features alone. Rather than asking who looks high risk, IMvigor011 asks who still has biologically active disease.

Among the 250 ctDNA-positive patients who underwent randomization, adjuvant atezolizumab significantly improved disease-free survival compared with placebo (median, 9.9 vs. 4.8 months; hazard ratio for recurrence or death, 0.64). In the prespecified hierarchical analysis, this benefit extended to overall survival (median, 32.8 vs. 21.1 months; hazard ratio for death, 0.59). These findings are particularly notable in a disease setting where demonstrating an overall survival benefit in unselected adjuvant trials has historically been challenging, despite consistent improvements in disease-free survival. The ctDNA-guided strategy likely played a central role in enabling this signal by enriching the study population for patients with true molecular residual disease and reducing dilution of treatment effects from patients already cured by surgery.

One of the most clinically impactful findings of IMvigor011 comes from the persistently ctDNA-negative cohort. Despite the absence of adjuvant therapy, disease-free survival in this group was 95% at one year and 88% at two years. Notably, nearly half of the enrolled population remained persistently ctDNA-negative, defining a large post-cystectomy subgroup with excellent outcomes under observation alone. This has important clinical implications. In the atezolizumab arm, nearly half of treated patients experienced treatment-related adverse events, including grade 3–4 toxicity, and treatment-related deaths, in three patients, were observed. Against this backdrop, the ability to safely spare a substantial proportion of patients from adjuvant therapy is not only cost-effective but may meaningfully reduce treatment-related harm.

Importantly, the ctDNA-negative population was not uniformly low risk. More than half of these patients had pT3 disease, and approximately one-fifth were node-positive, yet outcomes remained favorable. This suggests that ctDNA negativity may represent one of the strongest available indicators of true curative potential after surgery, even in patients with adverse pathologic features. These findings raise the hypothesis that, in selected patients, effective lymph node dissection may be biologically curative; however, this interpretation remains exploratory and warrants further investigation.

IMvigor011 also highlights the dynamic nature of ctDNA as a biomarker. Forty-one percent of ctDNA-positive patients converted from negative to positive on subsequent testing rather than at the initial postoperative assessment. We see this as an important limitation of single time-point testing and reinforces the value of serial monitoring. ctDNA sensitivity appears to increase over time, perhaps reflecting evolving disease biology rather than assay failure and emphasizing that postoperative risk assessment is not static. Had treatment decisions been based solely on the first postoperative ctDNA result, both trial outcomes and clinical conclusions would likely have differed substantially.

An additional point that warrants elaboration relates to the ctDNA assay and trial design. ctDNA was assessed using a personalized, tumor-informed assay that tracks patient-specific somatic variants derived from whole-exome sequencing of the resected tumor, allowing for highly specific longitudinal detection of molecular residual disease. Importantly, randomization occurred only after ctDNA positivity was confirmed, so treatment arms within the ctDNA-positive cohort were expected to be broadly balanced with respect to baseline molecular disease burden, although detailed quantitative baseline ctDNA levels by arm were not reported.

An exploratory analysis examined ctDNA clearance over time and its biologic relevance. Early ctDNA clearance was more common with atezolizumab (25%), although clearance was also observed in a small subset of placebo-treated patients (14%). However, meaningful biologic differences emerged when depth and durability were considered. Clearance of high baseline ctDNA levels and large, sustained quantitative declines occurred exclusively in the atezolizumab arm. In contrast, clearance events in the placebo group were confined to patients with uniformly low pretreatment ctDNA levels, near the assay's limit of detection, and were frequently followed by late reconversion to ctDNA positivity, as reported in the supplementary material.

This dynamic framework provides an important lens through which to interpret earlier trials such as CheckMate 274, in which adjuvant nivolumab produced durable disease-free survival benefit in an unselected high-risk population and, with extended follow-up, has demonstrated emerging overall survival signals, particularly among patients with PD-L1–positive tumors. ⁵ Notably, recent post hoc exploratory analyses incorporating ctDNA further clarify this heterogeneity. ⁵ In CheckMate 274, patients with detectable postoperative ctDNA were at markedly higher risk of recurrence and derived substantial benefit from adjuvant nivolumab (hazard ratio for disease-free survival, 0.35), whereas patients with undetectable ctDNA had excellent outcomes regardless of treatment assignment, with no clear incremental benefit from nivolumab (hazard ratio, 0.99). Such results support that the population-level benefit observed in pathologic risk-based trials may be driven primarily by patients with molecular residual disease, and the potential value of ctDNA-based enrichment strategies.

This trajectory aligns with the broader movement toward personalized oncology. Ongoing studies are extending ctDNA-guided strategies beyond patient selection toward fully adaptive treatment paradigms. The MODERN study, ⁶ for example, is evaluating whether ctDNA can guide not only the need for adjuvant therapy, but also its timing and intensity, by randomizing ctDNA-positive patients to nivolumab alone or nivolumab plus relatlimab and managing ctDNA-negative patients with surveillance and treatment initiation upon molecular relapse. Together with IMvigor011, these efforts signal a shift away from blanket adjuvant treatment toward intervening selectively when molecular residual disease is detected.

In conclusion, IMvigor011 represents a meaningful advance toward truly personalized postoperative management in muscle-invasive bladder cancer. By directly measuring molecular residual disease rather than relying on anatomic surrogates, ctDNA-guided strategies enable rational escalation and de-escalation of adjuvant therapy, reconcile prior inconsistencies in the adjuvant immunotherapy literature, and reduce unnecessary treatment exposure. As ctDNA assays continue to mature, molecular residual disease detection is poised to play an increasingly central role in postoperative decision-making, not only in bladder cancer but across solid tumors. In selected patients with persistently negative ctDNA, these advances also raise the possibility of bladder-sparing approaches, an idea that may become increasingly plausible given the growing efficacy of systemic combination therapies currently being evaluated in the neoadjuvant setting. ⁷