Introduction

Muscle-invasive urothelial carcinoma (MIUC) is a lethal disease with a 5-year survival rate of 50−60%, even in the absence of distant metastases1,2,3,4,5. Neoadjuvant chemotherapy (NAC), radical cystectomy (RC) or nephroureterectomy (RNU), and extended pelvic lymph node dissection have not improved the prognosis of patients with MIUC6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21. Although adjuvant chemotherapy may improve survival22toxic chemotherapy is not feasible in all patients due to advanced age, impaired renal function, and frailty. Disease-free survival (DFS) improved with adjuvant nivolumab in patients with high-risk MIUC in the CheckMate 274 trial23. Nivolumab was approved in Japan on March 28, 2022, as an adjuvant immunotherapy for MIUC. The effects of nivolumab on overall survival (OS) have not been determined because more time is needed. Furthermore, a similar study (IMvigor 010 trial) evaluating the effects of adjuvant atezolizumab after radical surgery failed to show efficacy, leading to questions about the consistency of the results24. Thus, the benefits of adjuvant immune checkpoint inhibitors (ICIs) in clinical practice are unclear25,26. We evaluated trends in the use of perioperative therapy and efficacy of adjuvant ICIs in patients with muscle-invasive bladder carcinoma (MIBC) and upper tract urothelial carcinoma (UTUC) in a real-world practice.

Results

Baseline characteristics

The median patient age was 71.0 years. Table 1 shows the baseline characteristics of the patients, including 1383 patients with MIBC and 1124 patients with UTUC. The eGFR in patients with UTUC was significantly lower than the eGFR in patients with MIBC (P < 0.001). The NAC administration rate in patients with UTUC was 20.5%. Pathological high-risk disease occurred in 46.9% and 40.2% of patients with MIBC and UTUC, respectively (Table 1). Patient treatment journeys are shown in a Dendrogram (Fig. 1A) and a Sankey diagram (Fig. 1B). NAC was administered to 1095 (43.7%) patients and adjuvant therapy was administered to 366 (14.6%) patients. In the NAC and non-NAC groups, 12.1% and 16.5% of patients, respectively, received adjuvant therapy. Figure 1C illustrates the use of adjuvant therapy by institution and time period. Prior to 2019, adjuvant therapy was administered only at selected institutions. However, since 2020, it has been more widely adopted across a broader range of centers (Fig. 1C).

Table 1 Background of patients.
Full size table
Fig. 1
figure 1

Treatment journeys of patients shown as a Dendrogram and Sankey diagram. (A) Display of the treatment journey using a dendrogram. (B) Display of the treatment journey using a Sankey diagram. (C) Annual number of patients receiving adjuvant therapy.

Full size image

Primary outcome: trends in the use of perioperative therapy

The use of neoadjuvant chemotherapy increased from 2005 to 2024, reaching 70.5% in 2024. The use of adjuvant therapy gradually decreased from 2005 to 2021 but increased after the approval of nivolumab in 2022 (Fig. 2A). However, the use of adjuvant therapy increased in patients with pathological high-risk disease (Fig. 2B). The adjuvant therapy usage rate was 30.3% before 2022 but increased 2-fold (61.0%) after 2022 (Fig. 2C). Of the adjuvant therapies, the use of chemotherapy gradually decreased, while the use of ICIs rapidly increased (Fig. 2D). The use of adjuvant ICIs was 67.5% after 2022 (Fig. 2E). After 2022, adjuvant chemotherapy was mainly administered in patients without NAC, and adjuvant ICIs were mainly used for patients with NAC (Fig. 2F). In patients with pathological high-risk UTUC, the use of ICIs in combination with NAC increased rapidly (Fig. 3A). The main regimen for adjuvant therapy changed from chemotherapy (23.0%) to ICIs (54.3%) after 2022 (Fig. 3B).

Fig. 2
figure 2

Trends in perioperative therapy use. (A) Trends in the use of neoadjuvant and adjuvant therapies in all patients with MIBC or UTUC. (B) Trends in the use of adjuvant therapies in patients with pathological high-risk disease. (C) Comparison of adjuvant therapy use in patients with pathological high-risk disease before 2022 and after 2022. (D) Trends in the use of adjuvant therapies (chemotherapy or immunotherapy) in patients with pathological high-risk disease. (E) Comparison of the type of adjuvant therapy in patients with pathological high-risk disease before 2022 and after 2022. (F) Differences in the rate of NAC use between adjuvant chemotherapy and immunotherapy.

Full size image
Fig. 3
figure 3

Trends in the use of perioperative therapy in UTUC. (A) Trends in the use of neoadjuvant and adjuvant therapies in patients with pathological high-risk UTUC. (B) Comparison of the type of adjuvant therapy in patients with pathological high-risk disease before 2020, 2020−2021, and after 2022.

Full size image

Secondary outcomes: effects of adjuvant immunotherapy on oncological outcomes.

Of the 1101 patients with pathological high-risk disease, 766, 268, and 63 patients received no adjuvant therapy, adjuvant chemotherapy, and adjuvant ICIs, respectively (Table 2). The unadjusted DFS and OS were not significantly different between the adjuvant chemotherapy and ICIs (Fig. 4A and B). However, the inverse probability treatment weighting (IPTW)-weighted Cox regression analyses in combination with an average treatment effect on the treated (ATT) method showed significant differences in DFS (P < 0.005) and OS (P < 0.005) between adjuvant chemotherapy and ICIs (Fig. 4C and D).

Table 2 Background of patients with pathological high-risk (pT3-4, ypT2-4, or pN+).
Full size table
Fig. 4
figure 4

The prognostic impact of adjuvant immunotherapy on prognosis. (A) Unadjusted disease-free survival in patients with pathological high-risk MIUC. (B) Unadjusted overall survival in patients with pathological high-risk MIUC. (C) The ATT-based IPTW-adjusted disease-free survival in patients with pathological high-risk MIUC. (D) The ATT-based IPTW-adjusted overall survival in patients with pathological high-risk MIUC.

Full size image

The weights calculated for each case using the IPTW method are shown (Fig. S1A). We evaluated the standardized mean difference (SMD) after IPTW adjustment and observed imbalances between the two groups (Table 3 and Fig. S1B). To address this, we additionally applied the average treatment effect (ATE)–based IPTW method. After ATE adjustment, covariate balance improved for age, sex, and ECOG performance status (SMDs < 0.1). However, some covariates—pT (SMD = 0.34), NAC (SMD = 1.30), and UTUC (SMD = 0.33)—remained imbalanced, indicating that ATE weighting did not fully resolve these differences. Therefore, we applied an average treatment effect on the treated (ATT)–based IPTW approach to estimate the treatment effect specifically within the treated group (Table 3 and Fig. S1C).

Table 3 Covariate balance between the patients with adjuvant chemotherapy and adjuvant ICIs.
Full size table

The difference in the efficacy of adjuvant immunotherapy between MIBC and UTUC is of clinical interest. However, due to the limited number of patients in each subgroup, we present unadjusted DFS and OS outcomes in Fig. 5 as exploratory data. In MIBC, both DFS and OS showed a trend toward improvement (Fig. 5A and B). In contrast, for UTUC, DFS was comparable to that in the adjuvant chemotherapy group (Fig. 5C), while OS showed a trend toward prolongation (Fig. 5D).

Fig. 5
figure 5

Unadjusted DFS and OS by tumor type (MIBC vs. UTUC) in patients receiving adjuvant immunotherapy. (A) Kaplan–Meier curve for DFS in patients with MIBC. (B) Kaplan–Meier curve for OS in patients with MIBC. (C) Kaplan–Meier curve for DFS in patients with UTUC. (D) Kaplan–Meier curve for OS in patients with UTUC. Red lines represent patients treated with adjuvant immune checkpoint inhibitors (ICIs), and blue lines represent those treated with adjuvant chemotherapy. Hazard ratios (HRs) are shown in each panel. It should be noted that no adjustment for baseline covariates was performed in these analyses due to the limited sample size.

Full size image

We also analyzed factors associated with receipt of neoadjuvant and adjuvant therapy. In the multivariable model, male sex, favorable ECOG PS, UTUC, absence of cerebrovascular disease, clinical stage cT3–4 or cN+, and absence of variant histology were independently associated with administration of NAC (Table S1, Fig. S2A). Predictors of adjuvant therapy were younger age, favorable ECOG PS, prior NAC, and higher pathological stage (pT3–4, ypT2–4, or N+) (Table S1, Fig. S2B).

Discussion

Trends in perioperative treatment and the effects of adjuvant immunotherapy associated with the development of new treatment strategies in patients with MIUC were investigated. The CheckMate 274 and IMvigor 010 trials showed inconsistent results24,27highlighting the fragility of evidence and the need to evaluate outcomes from clinical trial to practice28,29,30,31. We confirmed the rapid increase in adjuvant ICI use and observed improved prognoses in patients with MIUC. However, there is an inherent limitation due to the substantially shorter follow-up period for patients treated with adjuvant immunotherapy, as reflected in the Kaplan–Meier curves. The available data were insufficient to allow for a meaningful comparison of oncological outcomes. This limits the ability to draw definitive conclusions regarding the long-term benefits of immunotherapy at this stage. Moreover, outcome disparities between upper- and lower-tract diseases were reported (HR 1.27 vs. 0.66)28. Further research is needed to determine if this disparity can be explained by the small number of UTUC cases32 or whether other factors were involved. Given the clinical importance of assessing the real-world efficacy of adjuvant immunotherapy, we plan to address this issue in an ongoing multicenter study. This future study will provide more valid comparisons, as the control group will be limited to a more recent cohort treated in 2019 or later.

Patient selection for adjuvant therapy should be considered in the context of institutional and temporal factors. There are considerable inter-institutional differences in the use of adjuvant therapy, and treatment policies have also evolved over time within the same institutions, likely in response to emerging clinical evidence. Prior to 2019, adjuvant therapy was administered only at selected centers. However, following the publication of the POUT trial for UTUC33 and the CheckMate 274 trial for MIUC23its use has become more widespread across institutions. Although this study did not directly examine the factors influencing treatment decisions, it is reasonable to speculate that adjuvant therapy was administered to patients with a high risk of recurrence and strong motivation for treatment during each period. Therefore, patient selection criteria may have remained relatively consistent over time, lending support to the validity of the observed outcomes.

Regarding the use of adjuvant ICIs in combination with NAC, our results showed that, in most cases, adjuvant ICIs were administered after NAC. This result is in line with the results of a sub-analysis of the CheckMate 274 trial showing that adjuvant nivolumab may be highly effective after NAC. Interestingly, the use of NAC plus adjuvant ICIs in UTUC is comparable to that in MIBC. Level 1 evidence supporting adjuvant chemotherapy was reported in the POUT study33. Thus, the choice between adjuvant chemotherapy and ICIs for patients with high pathological risk is a dilemma in clinical practice. Of the 42 patients with pathological high-risk UTUC after 2022 in our study, 23 (54.8%) patients were treated with adjuvant chemotherapy but did not receive NAC, 1 patient (2.4%) received adjuvant ICIs without NAC, and 18 (42.9%) patients received adjuvant ICIs plus NAC. These findings indicate a tendency to choose adjuvant chemotherapy without NAC and adjuvant ICIs with NAC for patients with high pathological risk UTUC. Although the observation period was short, recent real-world data indicate that the beneficial effects of adjuvant nivolumab may be higher in patients treated with NAC compared with patients who were not treated with NAC25. Further investigation is needed to determine whether this selection is beneficial.

Several limitations of this study should be acknowledged. First, due to the retrospective study design, we could not control for selection bias and other potential confounders. Second, the statistical analysis may be underpowered due to the small sample size and short observation period (up to April 2024) in patients treated with adjuvant ICIs. Third, there were significant temporal differences between the groups; all patients receiving adjuvant ICIs were treated in recent years, whereas those receiving adjuvant chemotherapy were treated in earlier periods. These historical imbalances may have introduced unmeasured confounding that could not be fully adjusted for, even with IPTW. Fourth, the cohort was drawn from a single national population and did not include patients from multiple countries or diverse ethnic backgrounds, which limits the generalizability of our findings. Fifth, this study included four patients who had been enrolled in clinical trials. Although they were included in the analysis due to the small number of cases, this may have introduced selection bias, as trial participants are generally healthier than unselected real-world patients. Sixth, biomarker data such as PD-L1 expression, tumor mutational burden, and molecular subtypes were not available in this study. Our recent research demonstrated that NAC significantly affects PD-L1 expression, Ki-67 labeling index, and gene expression changes in patients with MIBC34. In addition, inflammatory markers have also been implicated35. These biological factors may influence both the decision to administer immunotherapy and clinical outcomes, potentially resulting in residual confounding. Seventh, treatment-related toxicity, dose intensity of chemotherapy, and immune-related adverse events were not systematically captured, preventing evaluation of risk–benefit profiles and their potential impact on survival. Eighth, an immortal-time bias cannot be entirely excluded because patients had to survive long enough post-surgery to receive adjuvant therapy; although we minimized this risk by landmarking at therapy initiation, some bias may persist. Ninth, central pathological review was not performed, and inter-observer variability in staging or variant histology assignment across centers could have affected risk stratification. Tenth, follow-up protocols were not standardized among institutions, which might have caused differential detection of recurrence and affected disease-free survival estimates. Despite these limitations, this is the first study to show the trends in adjuvant ICI use and the impact on prognoses in patients with MIUC in Japan. Further studies are required to determine the optimal strategies for selecting treatments for patients with MIUC.

In conclusion, the introduction of adjuvant immunotherapy led to increased adjuvant therapy use. Given the short follow-up period and the retrospective design of the study, it is not possible to draw definitive conclusions regarding the efficacy of adjuvant immunotherapy in improving outcomes. Therefore, the findings should be interpreted as showing an association rather than causality, in order to avoid potential misinterpretation.

Patients and methods

Design and ethics statement

We conducted this retrospective, multi-institutional study in accordance with the Declaration of Helsinki. The need for informed consent to participate was waived by an Institutional Review Board by the ethics committee of the Hirosaki University School of Medicine and all participating hospitals. Written informed consent was not obtained from individual patients due to the use of an opt-out approach. This study was performed in accordance with appropriate guidelines and the experimental protocol that was approved by the ethics committee of the Hirosaki University School of Medicine (2023–063–1).

Patient selection, demographics, and surgical procedures

Between January 1994 and April 2024, 2521 patients with MIUC (MIBC and UTUC) but no distant metastases who received RC or RNU at 10 academic centers were enrolled in the study. After excluding 14 patients with insufficient clinical data, 1383 patients with MIBC and 1124 patients with UTUC were included in the study (Fig. 1A). The following variables were collected and analyzed: age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS), estimated glomerular filtration rate (eGFR), clinical stage, pathological stage, the use of adjuvant therapy, the type of adjuvant therapy, the year of treatment, DFS, and OS. The tumor stage and grade were stratified based on the 8th edition of the TNM classification36. Patients with MIBC underwent open or robotic RC, urinary diversion, and standard pelvic lymph node dissection. Patients with UTUC underwent open or laparoscopic RNU, including kidney and ureter removal, and ipsilateral bladder cuff. Regional lymph node dissection was performed at the discretion of the attending physician37.

Neoadjuvant or adjuvant therapy

Neoadjuvant chemotherapy regimens were selected according to our guideline for cisplatin eligibility based on the Galsky criteria38,39. The marginal criteria included ECOG PS 1, eGFR 50–60 mL/min/1.73 m2NYHA class II heart failure, and age > 80 years, and patients with two or more marginal factors were classified as cisplatin ineligible. Indications for NAC included MIBC ≥ T2, UTUC ≥ T2, or cN + disease. NAC cycles were repeated every 21 days for up to four cycles. Adjuvant chemotherapy or ICIs were indicated for patients with a high pathological risk, including pT3−4, ypT2−4, positive surgical margin, or pN+. Two or three cycles of adjuvant chemotherapy (either gemcitabine plus cisplatin, gemcitabine plus carboplatin, or methotrexate, vinblastine, doxorubicin, and cisplatin) were administered to eligible patients, if their postoperative status was suitable for toxic chemotherapy2,40.

Outcomes

Primary outcomes included trends and usage rates of NAC and adjuvant therapy. Treatments with NAC and adjuvant therapy were evaluated using Sankey diagrams. NAC and adjuvant therapy usage rates were assessed for different time periods. Secondary outcomes included the effects of adjuvant therapy on DFS and OS in patients with pathological high-risk (pT3−4, ypT2−4, or pN+) disease. DFS was defined as the time of primary treatment to nonurothelial tumor recurrence. Superficial urothelial recurrence was not included in DFS. OS was defined as the time from the primary treatment to any cause of death.

To explore factors associated with the administration of neoadjuvant or adjuvant therapy, multivariable logistic regression analyses were performed using patient demographics and tumor characteristics.

Statistical analyses

Statistical analyses were performed with BellCurve for Excel v4.07 (Social Survey Research Information Co., Ltd., Tokyo, Japan), GraphPad Prism v7.00 (GraphPad Software, San Diego, CA, USA), R v4.0.2 (R Foundation for Statistical Computing, Vienna, Austria), and Python v3.10.18 (CPython) using the lifelines, NumPy, pandas, seaborn, and matplotlib libraries. SMDs were compared before and after adjustment using a love plot. Groups were compared using the Mann–Whitney U and Fisher’s exact tests. Quantitative variables were expressed as medians with interquartile ranges. The OS rate was estimated using the log-rank test. The effects of adjuvant therapy on DFS and OS were evaluated with background-weighted multivariable Cox regression analyses using the IPTW method. Weighted Cox models incorporating IPTW were fitted to estimate hazard ratios (HRs) with 95% confidence intervals (CIs). Robust standard errors were obtained with the Huber–White sandwich estimator. The proportional-hazards assumption was checked using Schoenfeld residual tests and graphical diagnostics. Stabilized ATT weights were applied and trimmed at the 1st and 99th percentiles to reduce the influence of extreme values. Hazard ratios (HR) with 95% confidence intervals were calculated after controlling for potential confounders, including patient age, sex, ECOG PS, tumor type (UTUC), variant histology, NAC, pT stage, and type of adjuvant therapy.