Key Points
Question
Is first-line enfortumab vedotin plus pembrolizumab (EV/P) associated with improved clinical outcomes compared with gemcitabine plus cisplatin (G/C) in patients with metastatic urothelial carcinoma in a multicenter retrospective cohort study?
Findings
In this cohort study of 4433 patients, first-line EV/P was associated with significantly longer overall survival, lower risk of early hospitalization, and longer time to next treatment compared with G/C. EV/P had a distinct adverse event profile, with fewer hematologic and gastrointestinal toxic effects but more immune-related adverse effects.
Meaning
In this cohort study, EV/P appeared to be an effective and tolerable first-line therapy for metastatic urothelial carcinoma in routine clinical practice.
This cohort study compares the outcomes associated with first-line enfortumab vedotin plus pembrolizumab and gemcitabine plus cisplatin among patients with metastatic urothelial carcinoma.
Abstract
Importance
Enfortumab vedotin plus pembrolizumab (EV/P) recently emerged as the preferred first-line treatment regimen for metastatic urothelial carcinoma (mUC), but evidence outside trial settings is relatively limited.
Objective
To evaluate outcomes associated with first-line EV/P compared with gemcitabine plus cisplatin (G/C) in patients with mUC.
Design, Setting, and Participants
This retrospective cohort study used the TriNetX database. Adults with mUC initiating first-line EV/P or G/C between September 2007 and September 2025 were included. Cohorts were propensity score matched for comparison by demographic characteristics, comorbidities, kidney function, and location of metastatic disease.
Main Outcomes and Measures
The primary outcome was overall survival (OS). Secondary outcomes included time to next treatment (TTNT), time to hospitalization within 90 days of treatment initiation (EHA), and adverse events (AEs). Outcomes were compared using Kaplan-Meier analysis and log rank test, reporting hazard ratios (HRs) and 95% CIs.
Results
The study included 4433 patients (1841 receiving EV/P and 2592 receiving G/C), with the EV/P group being significantly older (mean [SD] age, 70.6 [10.1] years vs 67.2 [10.6] years; P < .001) and having a higher proportion of Black patients (129 [6.7%] vs 139 [5.3%]; P = .04) and White patients (1445 [76.0%] vs 1614 [61.8%]) but a smaller proportion of Asian patients (145 [7.6%] vs 326 [12.5%]; P < .001). The EV/P group also carried a greater comorbidity burden, including higher rates of heart failure, diabetes, and obesity, while baseline characteristics were well balanced after propensity score matching. First line EV/P was associated with lower risk of death (HR, 0.70; 95% CI, 0.63-0.79), longer median OS (20 vs 11 months), lower risk of EHA (HR, 0.73; 95% CI, 0.64-0.83), and longer TTNT (median, 35 vs 10 months; HR, 0.49; 95% CI, 0.39-0.61). Receipt of EV/P at first line was associated with higher rates of rash, hyperglycemia, and hypothyroidism but fewer cytopenias and gastrointestinal toxic effects.
Conclusions and Relevance
In this retrospective cohort study of patients with mUC, EV/P was associated with longer OS, delayed TTNT, and lower time to EHA compared with G/C. These findings support the broader adoption of EV/P as first-line therapy in routine clinical practice.
Introduction
Metastatic urothelial carcinoma (mUC) is a highly aggressive disease with poor long-term survival. Historically, first-line platinum-based chemotherapy has provided modest benefit, with a median overall survival (OS) of approximately 15 months and limited durability of response.1,2,3 Enfortumab vedotin (EV) and pembrolizumab (P) have each demonstrated survival benefit as monotherapy in previously treated mUC.3,4,5,6,7 The phase 3 EV-302 trial showed superior OS and progression-free survival (PFS) with EV plus P (EV/P) compared with platinum-based chemotherapy, regardless of cisplatin (C) eligibility, leading to US Food and Drug Administration approval of EV/P as a first-line regimen in 2024.6,8,9
Clinical trial populations often exclude patients with poor performance status or significant comorbidities, limiting generalizability to mUC populations outside trial settings.10 In routine practice, outcomes may vary due to broader clinical, institutional, and access-related factors, contributing to the efficacy-effectiveness gap.11 Observational data are therefore essential to assess whether the benefits observed in trials like EV-302 translate to everyday clinical settings.
Evidence for EV/P in patients with mUC outside trial settings remains limited to small retrospective reports.12,13 There exists a need for observational data outside trial settings to evaluate the effectiveness of EV/P and contextualize the findings from clinical trials in routine clinical practice.14 We therefore performed a retrospective cohort study using the TriNetX database to evaluate whether the benefits observed in EV-302 are reproducible in a broader, clinically heterogeneous population of patients with mUC.15,16 We hypothesized that EV/P would be associated with longer OS compared with gemcitabine (G)/C chemotherapy.
Methods
This retrospective cohort study used deidentified electronic health record data from the TriNetX Global Collaborative Network, comprising more than 140 health care organizations across North and South America, the Middle East, Africa, and the Asia-Pacific region.15,16 Because all data were deidentified, institutional review board approval and informed consent were not required. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Exposures
Exposures were defined as receipt of first-line systemic therapy for mUC with either EV/P or G/C. Patients were included if they initiated one of these regimens as their initial treatment following mUC diagnosis. Exposure assignment was based on medication administration records using RxNorm and Healthcare Common Procedure Coding System (HCPCS) codes within the TriNetX platform, with the index date defined as the first documented administration of therapy.
Study Population
Adults (age ≥18 years) with diagnosis of mUC between September 2007 and September 2025 were included. Diagnosis, staging, and treatment exposures were identified using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), HCPCS, RxNorm, and American Joint Committee on Cancer TNM classification codes. Eligible primary tumor sites included the bladder, ureter, and renal pelvis. Patients were included if they initiated first-line systemic therapy with either EV/P or G/C, irrespective of cisplatin eligibility. Details of cohort construction are provided in eTable 1 and eFigure 1 in Supplement 1.
Primary Outcomes
The primary outcome was OS, defined as the interval from initiation of first-line therapy to death from any cause. The index date was the first administration of EV/P or G/C. Death was ascertained from electronic health record–recorded events, and patients were censored at the date of last documented activity in TriNetX. Subgroup analyses of OS were conducted by prespecified baseline characteristics, including age (<80 years vs ≥80 years), race (White vs other groups), sex (female vs male), primary site of disease (upper tract vs lower tract), liver metastasis (yes vs no), site of metastasis (visceral vs lymph nodes only), and kidney function (normal, mild impairment, moderate to severe impairment). These characteristics appear as reported in patients’ electronic health records. TriNetX reports race information as White and other, and no additional information is available.
Secondary Outcomes
Secondary outcomes included treatment sequencing patterns, time to next treatment (TTNT) as a surrogate for PFS, time to early hospitalizations (EHA), and adverse events (AEs). The TTNT encompassed the interval between initiation of first-line systemic therapy and the commencement of any subsequent systemic treatment from a pharmacologically distinct class, as categorized by the Systematic Nomenclature of Medicine and TriNetX-curated Veterans Affairs/Anatomical Therapeutic Chemical codes.17 TTNT has been examined as a reliable surrogate for PFS in multiple solid tumors, showing a strong correlation in published material (medians: R2 = 0.88; hazard ratios [HRs]: R2 = 0.91).17 Hospitalizations within 90 days of initiating first-line systemic therapy were used to define EHA. This timeframe was selected to capture clinically meaningful admissions potentially attributable to acute and subacute treatment-related complications.7,18
We ascertained AEs using ICD-10 and North American Association of Central Cancer Registries diagnostic codes corresponding to established treatment-related toxic effects, with inclusion restricted to those occurring after initiation of first-line therapy to mitigate misclassification of preexisting conditions. Captured events of interest included peripheral neuropathy, hematologic toxic effects (anemia, neutropenia, thrombocytopenia), alopecia, gastrointestinal symptoms (nausea, diarrhea), fatigue, and selected immune-related adverse events (irAEs), such as rash, pruritus, hypothyroidism, and hyperglycemia (eTable 2 in Supplement 1).
Sensitivity Analysis
To test the robustness and generalizability of our findings, we conducted an OS sensitivity analysis including patients treated with G/C or G plus carboplatin (G/P). This allowed for the evaluation of treatment outcomes across platinum-based regimens.
Covariates
To control for confounding, we adjusted for baseline patient demographics (age and race), comorbidities, and tumor-related characteristics. All covariates were derived from structured ICD-10 diagnostic codes available in TriNetX. Comorbidities encompassed cardiovascular conditions (heart failure, diabetes, smoking history, obesity), chronic obstructive pulmonary disease, liver disease, and chronic kidney disease. Kidney function was further stratified based on estimated glomerular filtration rate (eGFR) into less than 60 mL/min/1.73 m2 and 60 mL/min/1.73 m2 or greater. Performance status documentation (Eastern Cooperative Oncology Group [ECOG]), age-related physical debility, and wheelchair dependence were used to document subjective and objective functional status.19 Tumor-related characteristics included the presence of visceral metastases and metastatic lesions involving bone or lymph nodes. Laboratory markers were also included, such as anemia, hypoalbuminemia, and elevated C-reactive protein and lactate dehydrogenase (LDH), as markers for systemic inflammation and prognosis.20
Statistical Analysis
Treatment patterns were assessed using the TriNetX analytics platform, which defined lines of therapy (LOT) based on the sequential initiation of distinct systemic regimens after the index date. Bar graphs and longitudinal sequencing plots were used to illustrate transitions between therapies across LOT 1 to LOT 6.
We performed propensity score matching (PSM) for each pairwise comparison between treatment cohorts to mitigate confounding.18 Patients receiving first-line EV/P were matched 1:1 with those treated with first-line G/C using nearest-neighbor matching and a caliper width of 0.1 SDs of the logit of the propensity score. Matching was based on all covariates described previously.
Descriptive statistics were used to characterize the cohort, with group comparisons conducted using t tests or Wilcoxon rank-sum tests for continuous variables and χ2 tests for categorical variables. Kaplan-Meier analyses with log-rank tests assessed OS, TTNT, EHA, and AEs, with HRs and 95% CIs reported. Multivariable Cox proportional hazards models identified prognostic factors for OS.16 All analyses were conducted within the TriNetX platform in real time using R version 4.5.1 (R Project for Statistical Computing), using a 2-sided significance threshold of P < .05
Results
Baseline Demographic Characteristics
At baseline before PSM, patients receiving EV/P (n = 1841) were significantly older than those treated with G/C (n = 2592), with mean (SD) ages of 70.6 (10.1) years vs 67.2 (10.6) years, respectively (P < .001). Median (IQR) follow-up was 249 (70-428) days for the EV/P group and 293 (0-672) days for the G/C group. Racial distribution differed, with a higher proportion of Black and White patients in the EV/P group compared with the G/C group (Black patients: 129 [6.7%] vs 139 [5.3%]; P = .04; White patients: 1445 [76.0%] vs 1614 [61.8%]; P < .001) and a lower proportion of Asian patients in the EV/P group compared with the G/C group (145 [7.6%] vs 326 [12.5%]; P < .001) (Table).
Table. Baseline Characteristics of Patients Treated With First-Line EV/P vs G/C Before and After Propensity Score Matching.
| Covariate | Before propensity score match | After Propensity Score match | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Patients by treatment, No. (%) | P value | Standard difference | Patients by treatment, No. (%) | P value | Standard difference | ||||
| EV/P (n = 1841) | G/C (n = 2592) | EV/P (n = 1395) | G/C (n = 1395) | ||||||
| Age at index, mean (SD), y | 70.6 (10.1) | 67.2 (10.6) | <.001 | 0.33 | 69.5 (10.3) | 69.5 (9.37) | .9 | <0.01 | |
| Race | |||||||||
| American Indian or Alaska Native | 10 (0.5) | 10 (0.4) | .48 | 0.02 | 10 (0.7) | 10 (0.7) | >.99 | <0.01 | |
| Asian | 145 (7.6) | 326 (12.5) | <.001 | 0.16 | 135 (9.7) | 133 (9.5) | .9 | <0.01 | |
| Black or African American | 129 (6.7) | 139 (5.3) | .04 | 0.05 | 82 (5.9) | 85 (6.1) | .8 | <0.01 | |
| Native Hawaiian or Other Pacific Islander | 0 | 20 (0.8) | <.001 | 0.12 | 0 | 0 | NA | NA | |
| White | 1445 (78.5) | 1614 (61.8) | <.001 | 0.31 | 1018 (72.9) | 1028 (73.7) | .6 | 0.01 | |
| Additional groupsa | 60 (3.1) | 76 (2.9) | .65 | 0.01 | 48 (3.4) | 42 (3.0) | .52 | 0.02 | |
| Unknown race | 119 (6.2) | 435 (16.6) | <.001 | 0.33 | 110 (7.9) | 104 (7.5) | .7 | 0.01 | |
| Comorbidities | |||||||||
| Heart failure | 275 (14.4) | 194 (7.4) | <.001 | 0.17 | 153 (11.0) | 150 (10.7) | .85 | <0.01 | |
| Diabetes | 558 (29.2) | 560 (21.4) | <.001 | 0.20 | 379 (27.2) | 354 (25.4) | .28 | 0.04 | |
| Overweight and obesity | 548 (28.7) | 367 (14.0) | <.001 | 0.36 | 299 (21.4) | 287 (20.6) | .58 | 0.02 | |
| Nicotine dependence | 390 (20.4) | 233 (8.9) | <.001 | 0.32 | 222 (15.9) | 202 (14.5) | .29 | 0.03 | |
| COPD | 296 (15.5) | 309 (11.8) | <.001 | 0.10 | 206 (14.8) | 196 (14.1) | .58 | 0.02 | |
| Cirrhosis of liver | 41 (2.1) | 52 (2.0) | .71 | 0.01 | 28 (2.0) | 24 (1.7) | .57 | 0.02 | |
| Age-related debility | 32 (1.7) | 12 (0.5) | <.001 | 0.11 | 10 (0.7) | 12 (0.9) | .66 | 0.01 | |
| Wheelchair dependence | 18 (0.9) | 10 (0.4) | .01 | 0.06 | 10 (0.7) | 10 (0.7) | >.99 | <0.01 | |
| ECOG PS 0-1 | 78 (4.1) | 44 (1.8) | <.001 | 0.14 | 36 (2.7) | 36 (2.7) | >.99 | <0.01 | |
| ECOG PS ≥2 | 33 (1.7) | 15 (0.6) | <.001 | 0.10 | 14 (1.1) | 14 (1.1) | >.99 | <0.01 | |
| Chronic kidney disease | 845 (44.2) | 613 (23.5) | .09 | 0.04 | 396 (28.4) | 405 (29.0) | .7 | 0.01 | |
| eGFR, mL/min/1.73 m2 | |||||||||
| Mean (SD) | 59.7 (25.8) | 65.9 (28.7) | <.001 | 0.22 | 62 (26.7) | 63.8 (28.1) | >.99 | 0.06 | |
| ≤60 | 1448 (75.7) | 1589 (60.8) | <.001 | 0.32 | 991 (71.0) | 987 (70.8) | .86 | <0.01 | |
| >60 | 1505 (78.7) | 1814 (69.4) | <.001 | 0.21 | 1071 (76.8) | 1096 (78.6) | .25 | 0.04 | |
| Laboratory results | |||||||||
| Hemoglobin <10 g/dL | 1202 (62.8) | 1288 (49.3) | <.001 | 0.19 | 806 (57.8) | 810 (58.1) | .88 | <0.01 | |
| Albumin <3.5 g/dL | 1152 (60.2) | 1367 (52.3) | <.001 | 0.15 | 811 (58.1) | 823 (59.0) | .64 | 0.01 | |
| Lactate dehydrogenase >250 U/L | 287 (15.6) | 408 (15.6) | .57 | 0.01 | 207 (14.8) | 224 (16.1) | .37 | 0.03 | |
| C-reactive protein >10 mg/dL | 270 (14.1) | 369 (14.1) | .99 | <0.01 | 198 (14.2) | 196 (14.1) | .91 | <0.01 | |
| Metastatic disease site | |||||||||
| Bone | 534 (27.9) | 672 (25.8) | .09 | 0.04 | 396 (28.4) | 405 (29.0) | .7 | 0.01 | |
| Liver | 312 (16.3) | 573 (21.9) | <.001 | 0.14 | 265 (19.0) | 261 (18.7) | .84 | <0.01 | |
| Lymph nodes | 1041 (54.4) | 847 (32.4) | <.001 | 0.44 | 638 (45.7) | 654 (46.9) | .54 | 0.02 | |
| Lung | 341 (17.8) | 583 (22.3) | <.001 | 0.10 | 274 (19.6) | 260 (18.6) | .5 | 0.02 | |
Abbreviations: COPD, chronic obstructive pulmonary disorder; ECOG PS, Eastern Cooperative Oncology Group performance status; NA, not applicable.
SI conversion factors: To convert albumin and hemoglobin to grams per liter, multiply by 10; C-reactive protein to milligrams per liter, multiply by 10; lactate dehydrogenase to microkatals per liter, 0.0167.
Any race not otherwise specified.
The EV/P group had a higher comorbidity burden, including more heart failure (275 [14.4%] vs 194 [7.4%]), diabetes (558 [29.2%] vs 560 [21.4%]), and overweight/obesity (548 [28.7%] vs 367 [14.0%]) (all P < .001). Subjective and objective indicators of impaired performance status were also more prevalent. Regarding metastatic patterns, patients receiving EV/P more frequently had lymph node (1041 [54.4%] vs 847 [32.4%]) and bone (534 [27.9%] vs 672 [25.8%]) metastases, while liver (312 [16.3%] vs 573 [21.9%]) and lung (341 [17.8%] vs 583 [22.3%]) metastases were more common in the G/C group. Baseline characteristics were well balanced after PSM, showing standardized differences of less than 0.1. More details about baseline characteristics appear in the Table.
Treatment Patterns
In the EV/P cohort, 1586 patients (99.7%) received the regimen as first-line therapy, indicating strong adherence to intended sequencing. Minor deviations were attributed to data limitations in TriNetX algorithms. For LOT 2, 121 of 225 (53.8%) transitioned to G with C/P, and 42 (18.7%) initiated fibroblast growth factor receptor (FGFR) inhibitors. Subsequent lines (LOT 3 to LOT 6) featured limited use of other regimens, such as dose-dense methotrexate, vinblastine, adriamycin, cisplatin (ddMVAC) (Figure 1A).
Figure 1. Treatments Patterns in Patients With Metastatic Urothelial Carcinoma.
C indicates cisplatin; ddMVAC, dose-dense methotrexate, vinblastine, adriamycin, cisplatin; EV, enfortumab vedotin; FGFR, fibroblast growth factor receptor; G, gemcitabine; and P, pembrolizumab.
In the G/C cohort, 2477 patients (94.3%) received first-line treatment as intended, with 417 of 1142 (36.5%) transitioning to EV/P in LOT 2. Other patients initiated alternative platinum-based regimens or immune checkpoint inhibitors, including avelumab (168 [14.7%]), atezolizumab (87 [7.6%]), and nivolumab (71 [6.2%]). Later lines showed increasing heterogeneity, with minimal FGFR inhibitor use. (Figure 1B).
Primary Outcome
Receipt of EV/P as first-line therapy was associated with a lower risk of all-cause mortality compared with G/C (HR, 0.70; 95% CI, 0.63-0.79). Median OS was longer in the EV/P group, with a median OS of 20 months vs 11 months in the G/C group. By the end of the observation period, the OS probability was 25.71% in the EV/P group and 13.92% in the G/C group (P < .001) (eFigure 2 in Supplement 1). The total number of patients in each cohort for this was 1395, with 500 events in the EV/P cohort and 861 events in the G/C cohort, for a median survival of 19.66 months vs 10.98 months, respectively.
Multivariable Cox regression analysis identified several adverse clinical features significantly associated with increased all-cause mortality, including heart failure (HR, 1.33; 95% CI, 1.14-1.57), wheelchair dependence (HR, 1.80; 95% CI, 1.09-2.81), and age-related physical debility (HR, 2.00; 95% CI, 1.18-3.50). The presence of bone metastases (HR, 1.36; 95% CI, 1.24-1.49) and liver metastases (HR, 1.52; 95% CI, 1.36-1.69) also significantly increased mortality risk. Notably, ECOG performance status of 0 or 1 (HR, 0.20; 95% CI, 0.10-0.38) and overweight or obesity status (HR, 0.86; 95% CI, 0.77-0.97) were independently associated with a lower risk of death (Figure 2). Laboratory markers linked to poorer prognosis included anemia (hemoglobin <10 g/dL; HR, 1.19; 95% CI, 1.05-1.30), hypoalbuminemia (albumin <3.5 g/dL; HR, 1.34; 95% CI, 1.20-1.48), elevated C-reactive protein (CRP >10 mg/L; HR, 1.16; 95% CI, 1.04-1.36), and elevated LDH (LDH >250 U/L; HR, 1.23; 95% CI, 1.12-1.42) (Figure 2).
Figure 2. Multivariable Risk Factors Associated With Overall Survival.
To convert albumin and hemoglobin to grams per liter, multiply by 10; C-reactive protein to milligrams per liter, multiply by 10; lactate dehydrogenase to microkatals per liter, 0.0167. ECOG indicates Eastern Cooperative Oncology Group; HR, hazard ratio.
In our exploratory subgroup analysis (Figure 3), EV/P was associated with a lower risk of all-cause mortality vs G/C across age groups (<80 years: HR, 0.74; 95% CI, 0.62-0.88; ≥80 years: HR, 0.72; 95% CI, 0.55-0.94). Similar associations were observed by sex (female: HR, 0.73; 95% CI, 0.56-0.96; male: HR, 0.74; 95% CI, 0.63-0.87) and primary tumor site (upper tract: HR, 0.76; 95% CI, 0.61-0.95; lower tract: HR, 0.78; 95% CI, 0.66-0.93). Among racial subgroups, lower risk was observed in White patients (HR, 0.77; 95% CI, 0.66-0.90), while the association for patients belonging to additional racial and ethnic groups was not statistically significant (HR, 0.76; 95% CI, 0.55-1.07), although numerically comparable.
Figure 3. Overall Survival by Subgroups in Patients Receiving Enfortumab Vedotin/Pembrolizumab (EV/P) vs Chemotherapy (G/C).
HR indicates hazard ratio.
The treatment effect size was also consistent in patients with or without liver metastasis (with: HR, 0.69; 95% CI, 0.49-0.98; without: HR, 0.68; 95% CI, 0.60-0.80) and by site of metastasis (visceral: HR, 0.73; 95% CI, 0.64-0.83; lymph nodes only: HR, 0.76; 95% CI, 0.60-0.97). Across kidney function categories, HRs also favored EV/P in patients with normal function (HR, 0.59; 95% CI, 0.44-0.79), mild impairment (HR, 0.70; 95% CI, 0.56-0.90), and moderate to severe impairment (HR, 0.74; 95% CI, 0.60-0.92) (Figure 3).
Secondary Outcomes
Patients treated with first-line EV/P had a significantly lower risk of transitioning to subsequent systemic therapies vs those receiving G/C (HR, 0.49; 95% CI, 0.39-0.61). The median TTNT was 35 months in the EV/P group vs 10 months in the G/C group. At the end of follow-up, the probability of patients in the EV/P cohort remaining on initial therapy was 33.03% compared with 25.31% in the G/C group (P < .001) (eFigure 3 in Supplement 1).
First-line treatment with EV/P was associated with a significantly lower risk of EHA compared with the G/C group (HR, 0.73; 95% CI, 0.64-0.83). At day 90, the probability of remaining free of inpatient admission in the EV/P group (67.1%) was lower compared with the G/C cohort (57.93%) (P < .001) (eFigure 4 in Supplement 1).
Patients receiving first-line EV/P had a higher risk of developing of skin rash (HR, 2.77; 95% CI, 2.24-3.42), diarrhea (HR, 1.42; 95% CI, 1.14-1.77), hyperglycemia (HR, 1.62; 95% CI, 1.21-2.16), and hypothyroidism (HR, 1.78; 95% CI, 1.27-2.48) vs the G/C group. In contrast, the EV/P group had a lower risk of fatigue (HR, 0.76; 95% CI, 0.62-0.92), nausea (HR, 0.63; 95% CI, 0.49-0.80), anemia (HR, 0.87; 95% CI, 0.78-0.99), thrombocytopenia (HR, 0.50; 95% CI, 0.41-0.60), and neutropenia (HR, 0.36; 95% CI, 0.27-0.47) (eFigure 5 in Supplement 1).
Sensitivity Analysis
Receipt of EV/P as first-line therapy was associated with a lower risk of all-cause mortality compared with G/P (HR, 0.83; 95% CI, 0.75-0.91). Median OS was longer in the EV/P group, with a median OS of 602 days vs 466 days in the G/P group. By the end of the observation period, the OS probability was 25.01% in the EV/P group and 13.74% in the control group (P < .001) (eFigure 6 in Supplement 1).
Discussion
In this retrospective cohort study using the TriNetX database, first-line EV/P was associated with a 30% lower risk of all-cause mortality vs G/C. A 51% lower risk of transitioning to a subsequent line of therapy was also associated with EV/P, along with a 23% lower risk of time to EHA (within 90 days of therapy initiation). Notably, patients treated with EV/P as first-line therapy experienced a higher risk of certain irAEs and non-irAEs, including skin rash, diarrhea, hyperglycemia, and hypothyroidism. In contrast, the risks of nausea, fatigue, and hematologic toxic effects (anemia, thrombocytopenia, and neutropenia) were higher with G/C.
The observed relative OS benefit with EV/P in this retrospective study mirrors outcomes reported in the phase 3 EV-302 trial.6 Quantitatively, median OS with EV/P was shorter in our cohort (20 months vs 31.5 months in EV-302), likely reflecting the efficacy-effectiveness gap often observed when translating trial results into settings outside trials with broader heterogeneity and comorbidity. These outcomes are consistent with a meta-analysis by Yajima et al,21 which included clinical trials and prospective studies, and reported a pooled 1-year survival rate of 79% in patients with mUC treated with first-line EV/P, a survival advantage associated with higher response rates compared with chemotherapy. Additionally, low hemoglobin (<10 g/dL) and albumin (<3.5 g/dL) were associated with poorer survival, consistent with the UNITE study,2,7 reinforcing their value as prognostic markers outside trial settings.22 Importantly, the survival benefit with EV/P remained consistent in a sensitivity analysis including G/P–treated patients, supporting generalizability across broader treatment populations.
The notable OS advantage of EV/P vs G/C was consistent across most prespecified subgroups, including age, sex, race, primary tumor site, kidney function variations, and metastatic site distribution. Patients with lymph node–only metastasis also demonstrated benefit, consistent with their reported more favorable prognosis.23 These results align with subgroup analyses from EV-302, which demonstrated significant OS benefit regardless of metastatic site distribution.24 Subgroups with confidence intervals approaching or crossing the unity, particularly among smaller populations (eg, other races), likely reflect limited statistical power due to smaller sample size rather than absence of effect. Such findings are a recognized inherent limitation of subgroup analyses and should be considered hypothesis-generating.
Beyond OS, treatment durability and sequencing patterns also favored first-line EV/P. TTNT, a surrogate for PFS, was significantly longer with EV/P, with a 51% lower hazard of transitioning to subsequent therapy compared with G/C, consistent with progression reductions seen in EV-302.6 The median TTNT in our EV/P cohort exceeded both the median PFS of 12.5 months observed in EV-302 trial and the 12.7 months in the retrospective analysis by Jain et al,25 likely reflecting potential differences in censoring, multiple unmeasured confounders, and likely broader clinical discretion in practice, eg, including therapy continuation beyond radiographic progression if there is perceived clinical benefit. At study end, one-third of patients receiving EV/P remained on first-line therapy, similar to EV-302 (32.6%). However, crossover rates were lower in routine practice (7.6% from EV/P to Cis, 18.6% from G/C to EV/P) compared with the EV-302 reported rates (24.9% and 58.6%, respectively).6 Retrospective studies by Li et al26 and Mathew Thomas et al27 highlight that, outside trial settings, many patients with mUC do not receive subsequent lines of therapy, often due to systemic barriers, such as limited access to novel treatments, reimbursement issues, or clinical deterioration necessitating hospice care.
In this cohort, first-line EV/P was associated with a lower risk of EHA and better 90-day hospitalization-free survival compared to G/C, indicating improved tolerability. These findings align with a network meta-analysis by Yanagisawa et al,28 which reported fewer grade 3 or higher AEs with EV/P than with chemotherapy. While EV/P was associated with more immune and metabolic toxic effects (eg, rash, hyperglycemia, hypothyroidism), patients receiving EV/P had fewer hematologic and gastrointestinal events.6,28 EV-103 further supports these results, with O’Donnell et al7 reporting frequent EV-related neuropathy (60.5%), skin toxic effects (67.1%), and hyperglycemia (14.5%) as well as pembrolizumab-related hypothyroidism (13.2%).
Limitations and Strengths
This study has several limitations. Its retrospective design introduces potential residual confounding, and the inclusion of patients treated since 2007 may reflect outdated treatment contexts. Molecular biomarkers, such as programmed cell death ligand 1 expression, microsatellite instability status, and tumor mutational burden, were not consistently available, limiting biomarker-adjusted analyses. Additionally, TTNT may not fully capture radiographic progression, and AEs identified via administrative codes may underestimate incidence or severity. As the study was conducted on the TriNetX platform, formal testing of the proportional hazards assumption was not possible; visual inspection of the Kaplan-Meier curves demonstrated brief convergence at 1 time point without frank crossing, and therefore the hazard ratio should be interpreted as an average effect estimate across the follow-up period rather than a constant effect size. As TriNetX captures all-cause inpatient encounters, we were unable to differentiate hospitalizations attributable to treatment toxic effects from other causes.
This study offers several strengths. To our knowledge, it is among the first large-scale, multi-institutional analyses of EV/P in routine clinical practice. Use of PSM can help to mitigate confounding by aiming to balance clinical and demographic variables. Moreover, the inclusion of multiple end points, including OS, TTNT, hospitalization, and toxic effects, allowed for a comprehensive assessment of the regimen’s impact in clinical practice.
Conclusions
This retrospective study found that EV/P was associated with meaningful clinical advantages over G/C as first-line therapy for mUC. The combination was associated with longer OS, delayed TTNT, lower risk of EHA, and a distinct toxicity profile favoring tolerability. The persistence of survival benefit among patients treated with G/C further supports the generalizability of these findings across platinum-based regimens. These results reinforce the effectiveness of EV/P in routine clinical practice, even among clinically heterogeneous populations. Given that many patients may not receive subsequent lines of therapy, optimizing first-line treatment remains essential. These data support broader adoption of EV/P in everyday oncology practice.
eTable 1. Diagnostic, Staging, and Treatment Codes for Cohort Identification in Metastatic Urothelial Carcinoma
eTable 2. Adverse Event Definitions by Corresponding Diagnosis Codes
eFigure 1. Flow Diagram of Patient Selection for Analysis of First-Line Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin in Metastatic Urothelial Carcinoma
eFigure 2. Comparative Overall Survival of First-Line Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin in Patients With Metastatic Urothelial Carcinoma
eFigure 3. Kaplan-Meier Estimates of Time to Next Treatment by First-Line Therapy in Metastatic Urothelial Carcinoma
eFigure 4. Kaplan-Meier Estimates of Early Hospital Admission Within 90 Days of First-Line Therapy Initiation for Metastatic Urothelial Carcinoma
eFigure 5. Hazard of Treatment-Related Adverse Events Following First-Line Therapy With Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin
eFigure 6. Overall Survival by First-Line Treatment Group Including Patients Treated With Gemcitabine–Cisplatin/Carboplatin (Sensitivity Analysis)
Data Sharing Statement
References
- 1.Bellmunt J, de Wit R, Vaughn DJ, et al. ; KEYNOTE-045 Investigators . Pembrolizumab as second-line therapy for advanced urothelial carcinoma. N Engl J Med. 2017;376(11):1015-1026. doi: 10.1056/NEJMoa1613683 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Galsky MD, Arija JÁA, Bamias A, et al. ; IMvigor130 Study Group . Atezolizumab with or without chemotherapy in metastatic urothelial cancer (IMvigor130): a multicentre, randomised, placebo-controlled phase 3 trial. Lancet. 2020;395(10236):1547-1557. doi: 10.1016/S0140-6736(20)30230-0 [DOI] [PubMed] [Google Scholar]
- 3.Hoimes CJ, Flaig TW, Milowsky MI, et al. Enfortumab vedotin plus pembrolizumab in previously untreated advanced urothelial cancer. J Clin Oncol. 2023;41(1):22-31. doi: 10.1200/JCO.22.01643 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Heath EI, Rosenberg JE. The biology and rationale of targeting nectin-4 in urothelial carcinoma. Nat Rev Urol. 2021;18(2):93-103. doi: 10.1038/s41585-020-00394-5 [DOI] [PubMed] [Google Scholar]
- 5.Rosenberg JE, Powles T, Sonpavde GP, et al. EV-301 long-term outcomes: 24-month findings from the phase III trial of enfortumab vedotin versus chemotherapy in patients with previously treated advanced urothelial carcinoma. Ann Oncol. 2023;34(11):1047-1054. [DOI] [PubMed] [Google Scholar]
- 6.Powles T, Valderrama BP, Gupta S, et al. ; EV-302 Trial Investigators . Enfortumab vedotin and pembrolizumab in untreated advanced urothelial cancer. N Engl J Med. 2024;390(10):875-888. doi: 10.1056/NEJMoa2312117 [DOI] [PubMed] [Google Scholar]
- 7.O’Donnell PH, Milowsky MI, Petrylak DP, et al. Enfortumab vedotin with or without pembrolizumab in cisplatin-ineligible patients with previously untreated locally advanced or metastatic urothelial cancer. J Clin Oncol. 2023;41(25):4107-4117. doi: 10.1200/JCO.22.02887 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Powles TB, van der Heijden MS, Loriot Y, et al. Enfortumab vedotin plus pembrolizumab in untreated locally advanced or metastatic urothelial carcinoma: 2.5-year median follow-up of the phase III EV-302/KEYNOTE-A39 trial. Ann Oncol. 2025;36(10):1212-1219. doi: 10.1016/j.annonc.2025.05.536 [DOI] [PubMed] [Google Scholar]
- 9.Brave MH, Maguire WF, Weinstock C, et al. FDA approval summary: enfortumab vedotin plus pembrolizumab for locally advanced or metastatic urothelial carcinoma. Clin Cancer Res. 2024;30(21):4815-4821. doi: 10.1158/1078-0432.CCR-24-1393 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Geynisman DM, Broughton E, Hao Y, Zhang Y, Le T, Huo S. Real-world treatment patterns and clinical outcomes among patients with advanced urothelial carcinoma in the United States. Urol Oncol. 2022;40(5):195.e1-195.e11. doi: 10.1016/j.urolonc.2021.11.014 [DOI] [PubMed] [Google Scholar]
- 11.Monti S, Grosso V, Todoerti M, Caporali R. Randomized controlled trials and real-world data: differences and similarities to untangle literature data. Rheumatology (Oxford). 2018;57(57)(suppl 7):vii54-vii58. doi: 10.1093/rheumatology/key109 [DOI] [PubMed] [Google Scholar]
- 12.Aydogdu C, Zschaebitz S, Büttner T, et al. Treating the untended: Real-world insights into enfortumab vedotin plus pembrolizumab for frail metastatic urothelial carcinoma patients excluded or underrepresented in clinical trials (GUARDIANS). J Clin Oncol. 2025;43(16)(suppl):e16548. doi: 10.1200/JCO.2025.43.16_suppl.e16548 [DOI] [Google Scholar]
- 13.Zschaebitz S, Casuscelli J, Büttner C. et al. Enfortumab vedotin plus pembrolizumab in metastatic urothelial carcinoma: First results on outcomes and safety in a German multicenter real-world patient cohort (GUARDIANS). J Clin Oncol. 2025;43(5)(suppl):715. doi: 10.1200/JCO.2025.43.5_suppl.715 [DOI] [Google Scholar]
- 14.Roviello G, Santoni M, Sonpavde GP, Catalano M. The evolving treatment landscape of metastatic urothelial cancer. Nat Rev Urol. 2024;21(10):580-592. doi: 10.1038/s41585-024-00872-0 [DOI] [PubMed] [Google Scholar]
- 15.TriNetX. TriNetX: Real-world data and analytics platform. Accessed May 26, 2026. https://trinetx.com
- 16.Ludwig RJ, Anson M, Zirpel H, et al. A comprehensive review of methodologies and application to use the real-world data and analytics platform TriNetX. Front Pharmacol. 2025;16:1516126. doi: 10.3389/fphar.2025.1516126 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Agapow P, Mulla R, Markuzon N, Ottesen LH, Meulendijks D. Systematic review of time to subsequent therapy as a candidate surrogate endpoint in advanced solid tumors. Future Oncol. 2023;19(23):1627-1639. doi: 10.2217/fon-2022-0616 [DOI] [PubMed] [Google Scholar]
- 18.Grant RC, Moineddin R, Yao Z, Powis M, Kukreti V, Krzyzanowska MK. Development and validation of a score to predict acute care use after initiation of systemic therapy for cancer. JAMA Netw Open. 2019;2(10):e1912823. doi: 10.1001/jamanetworkopen.2019.12823 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Quinn SE, Crandell CE, Blake ME, et al. The correlative strength of objective physical assessment against the ECOG performance status assessment in individuals diagnosed with cancer. Phys Ther. 2020;100(3):416-428. doi: 10.1093/ptj/pzz192 [DOI] [PubMed] [Google Scholar]
- 20.Mori K, Miura N, Mostafaei H, et al. Prognostic value of preoperative hematologic biomarkers in urothelial carcinoma of the bladder treated with radical cystectomy: a systematic review and meta-analysis. Int J Clin Oncol. 2020;25(8):1459-1474. doi: 10.1007/s10147-020-01690-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Yajima S, Hirose K, Masuda H. Enfortumab vedotin with or without pembrolizumab in metastatic urothelial carcinoma: a systematic review and meta-analysis. JAMA Netw Open. 2025;8(3):e250250. doi: 10.1001/jamanetworkopen.2025.0250 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Jindal T, Jiang CY, Alhalabi O, et al. 1988P enfortumab vedotin (EV) + pembrolizumab (P) outcomes outside clinical trials and biomarkers of benefit in patients with advanced urothelial carcinoma: analysis of the UNITE study. Ann Oncol. 2024;35(suppl 2):S1149-S1150. [Google Scholar]
- 23.Makrakis D, Talukder R, Lin GI, et al. Association between sites of metastasis and outcomes with immune checkpoint inhibitors in advanced urothelial carcinoma. Clin Genitourin Cancer. 2022;20(5):e440-e452. doi: 10.1016/j.clgc.2022.06.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.van der Heijden MS, Powles T, Gupta S, et al. Exploratory subgroup analyses of EV-302: a phase III global study to evaluate enfortumab vedotin in combination with pembrolizumab versus chemotherapy in previously untreated locally advanced or metastatic urothelial carcinoma. ESMO Open. 2025;10(8):105544. doi: 10.1016/j.esmoop.2025.105544 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Jain P, Naqvi SAA, Tripathi N, et al. Real-world outcomes of enfortumab vedotin and pembrolizumab in advanced urothelial carcinoma: a multicenter retrospective analysis. Clin Genitourin Cancer. 2025;23(6):102453. doi: 10.1016/j.clgc.2025.102453 [DOI] [PubMed] [Google Scholar]
- 26.Li H, Sura S, Babcock A, et al. Evolving treatment patterns and outcomes among patients with metastatic urothelial carcinoma post-avelumab maintenance approval: insights from The US Oncology Network. ESMO Real World Data Digit Oncol. 2025;7:100110. doi: 10.1016/j.esmorw.2024.100110 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mathew Thomas V, Jo Y, Tripathi N, et al. Treatment patterns and attrition with lines of therapy for advanced urothelial carcinoma in the US. JAMA Netw Open. 2024;7(5):e249417. doi: 10.1001/jamanetworkopen.2024.9417 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Yanagisawa T, Mori K, Matsukawa A, et al; European Association of Urology–Young Academic Urologists Urothelial Carcinoma Working Group (EAU-YAU). Novel combination therapy for platinum-eligible patients with locally advanced or metastatic urothelial carcinoma: a systematic review and network meta-analysis. Cancer Immunol Immunother. 2025;74(3):76. doi: 10.1007/s00262-024-03910-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Diagnostic, Staging, and Treatment Codes for Cohort Identification in Metastatic Urothelial Carcinoma
eTable 2. Adverse Event Definitions by Corresponding Diagnosis Codes
eFigure 1. Flow Diagram of Patient Selection for Analysis of First-Line Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin in Metastatic Urothelial Carcinoma
eFigure 2. Comparative Overall Survival of First-Line Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin in Patients With Metastatic Urothelial Carcinoma
eFigure 3. Kaplan-Meier Estimates of Time to Next Treatment by First-Line Therapy in Metastatic Urothelial Carcinoma
eFigure 4. Kaplan-Meier Estimates of Early Hospital Admission Within 90 Days of First-Line Therapy Initiation for Metastatic Urothelial Carcinoma
eFigure 5. Hazard of Treatment-Related Adverse Events Following First-Line Therapy With Enfortumab Vedotin Plus Pembrolizumab vs Gemcitabine Plus Cisplatin
eFigure 6. Overall Survival by First-Line Treatment Group Including Patients Treated With Gemcitabine–Cisplatin/Carboplatin (Sensitivity Analysis)
Data Sharing Statement
