Real-world efficacy and safety of immune checkpoint blockades in advanced urothelial carcinoma with histologic variant subtypes: the pembroblad study

Abstract

Background

Histologic variants (HV) of urothelial carcinoma (UC-V) and non-urothelial carcinomas (NUC) of the bladder and urinary tract are aggressive malignancies with poor prognosis and limited response to platinum-based chemotherapy. Despite their biological particularities and distinct immune microenvironments, these HV subtypes are under-represented in immune checkpoint blockade (ICB) trials.

Methods

PEMBROBLAD is a French national multicenter retrospective study conducted across 24 centers. We included patients with advanced UC-V or NUC who received second-line anti–PD-(L)1 monotherapy after progression on platinum-based chemotherapy between 2016 and 2022. Primary endpoint was overall survival (OS); secondary endpoints included progression-free survival (PFS), objective response rate (ORR) and safety.

Results

We included 129 UC-V and 34 NUC. With a median follow-up was 21.9 months, the median OS was 8.6 [95% confidence interval (CI), 5.3–11.2] months, with 8.5 [95% CI 5.3–10.9] months for UC-V and 10.5[95% CI 3.5–22.8] months for NUC. The median PFS was 2.6 [95% CI 2.1–3.4] months with 2.5 [95% CI 2.0–3.4] months for the UC-V group and 2.6 [2.0–3.4] months for the NUC group. ORR was 31.4% overall (complete response 15.3%, partial response 16.1%). In multivariable analysis, ECOG performance status ≥ 2 was independently associated with both poorer OS (HR 3.30, 95% CI 1.22–8.92; p = 0.018) and PFS (HR 2.32, 95% CI 1.52–3.54; p < 0.001). Pembrolizumab treatment was associated with improved PFS (HR 0.53, 95% CI 0.31–0.81; p = 0.017). Grade ≥ 3 immune-related adverse events occurred in 10.8% of patients, with no treatment-related deaths.

Conclusions

Anti–PD-(L)1 monotherapy shows clinical benefit in advanced UC-V and NUC after platinum failure.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-026-04421-z.

Introduction

Bladder cancer is a significant global public health challenge, with more than 600,000 newly diagnosed cases and more than 200,000 reported deaths annually worldwide [1].

Bladder and urinary tract cancers exhibit various histologic variant (HV) subtypes. Urothelial carcinoma (UC) is the most common form, accounting for approximately 90% to 95% of cases [2]. It can be classified as pure UC, consisting exclusively of dysplastic urothelial cells, or as HV UC (UC-V), which is characterized by the presence of additional histological features. While up to 25% of all UC harbor HV histologies, each individual HV occurs rarely [3]. Among these HV, UC-V with squamous and/or glandular metaplasia is the most frequent, accounting for approximately 20% to 30% of cases [4]. Other HV are less common, such as sarcomatoid UC, nested UC, micropapillary UC, and giant cell UC [5].

Although pure UC, or UC with HV (UC-V), is by far the most common form of bladder and urinary tract cancer, non-urothelial carcinomas (NUC) also exist. NUC encompass diverse HV subtypes, notably squamous cell carcinomas, adenocarcinomas, and neuroendocrine carcinomas, each with distinct treatment responses and prognoses [6, 7]. UC-V and NUC HV subtypes in bladder and urinary tract cancers are associated with poor prognosis and reduced response rates to platinum-based chemotherapy compared to pure UC [2, 8].

However, due to the rarity of these HV and the lack of dedicated studies, available data remain insufficient to establish definitive conclusions. Consequently, no specific treatment guidelines exist, and management is largely extrapolated from UC, despite distinct immunogenetic profiles including variability in PD-L1 expression, lower tumor mutational burden (TMB), and differences in tumor-infiltrating lymphocyte (TIL) levels [9–11].

Immunotherapy, particularly immune checkpoint blockade (ICB) targeting programmed cell-death one (PD-1) or its ligand PD-L1, has revolutionized the management of several solid tumors [12–14]. In advanced UC, anti-PD-(L)1 have become the cornerstone of treatment. In 2017, the KEYNOTE-045 trial showed a survival benefit of pembrolizumab over chemotherapy in the second-line setting [15]. In 2020, the JAVELIN Bladder 100 trial confirmed the role of ICB with avelumab maintenance improving survival after first-line platinum-based chemotherapy [16]. Most recently, in 2024, the EV-302 trial demonstrated superior survival with enfortumab vedotin plus pembrolizumab versus standard chemotherapy in the first-line setting, establishing a new benchmark in advanced or metastatic disease [17]. More recently, the treatment landscape of advanced urothelial carcinoma has further evolved with the results of the CheckMate 901 trial, which demonstrated a survival benefit with nivolumab combined with gemcitabine–cisplatin in the first-line setting[18]. This trial included approximately 150 patients with variant or non-urothelial histologies, representing one of the largest prospective datasets in this population, whereas the EV-302 trial included only about 60 such patients. Although dedicated subgroup analyses according to variant histology were not reported, these studies highlight the increasing use of ICB earlier in the disease course.

However, the efficacy and safety of ICB in the management of UC-V or NUC remain largely unknown. All large phase 3 randomized trials excluded patients with NUC or UCV with predominant non urothelial component[16, 19, 20]. Therefore, no treatment is currently approved for those subgroups.

Due to this lack of data on UC-V and NUC, the French national real-world PEMBROBLAD retrospective study was launched in 2022 to assess the clinical benefit and safety profile of anti-PD-(L)1 monotherapy in the management of advanced UC-V or NUC of the bladder and/or urinary tract in the second-line setting, after platinum-based chemotherapy.

Participants and methods

Study design, and patients

Retrospective data were collected from all consecutive patients with histologically confirmed UC-V or NUC who received second-line anti–PD(L)1 monotherapy following platinum-based chemotherapy, between 2016 and 2022, across 24 French centers.

Eligibility criteria were age ≥ 18 years, histologically confirmed UC-V, regardless the percentage of the non-urothelial component, or NUC and disease progression after platinum-based chemotherapy for advanced disease, or recurrence within 12 months following cisplatin-based adjuvant or neoadjuvant therapy for localized muscle-invasive disease. Histological assessment was performed locally only, without centralized review.

Exclusion criteria included receipt of ICB as maintenance therapy without prior progression after platinum-based chemotherapy, incomplete or missing clinical or pathological data, receipt of anti–PD(L)1 combination therapy, non-metastatic disease at the time of ICB initiation and unconfirmed histological diagnosis.

Study treatment

Anti-PD-(L)1 monotherapies were administered intravenously according to the current recommendations and local guidelines.

Objectives and endpoints

The primary objective of this study was to assess real-world overall survival (OS) of second-line ICB. Secondary endpoints included progression-free survival (PFS), objective response rate (ORR) of both UC-V peer RECIST 1.1 criteria and NUC patients and safety.

Statistical analysis

Data are expressed as median (range) or [95% confidence interval (CI)] for continuous variables, and number (percentage) for categorical variables. OS was defined as the time from initiation of anti–PD(L)1 monotherapy to death from any cause; PFS was defined as the time from treatment initiation to radiological or clinical progression, non-RECIST progression deemed relevant by the treating physician, clinical deterioration, or death, whichever occurred first. Patients alive and progression-free were censored at last follow-up or at the start of subsequent therapy. Median survival times with 95% CI were estimated by Kaplan–Meier; comparisons used the log-rank test. Univariable and multivariable Cox models identified factors associated with OS and PFS, with variables p < 0.20 in univariable analyses included in multivariable models; p < 0.05 was considered significant. Follow-up was estimated using the inverse Kaplan–Meier method. Missing values were handled via an “unknown” category when applicable. ORR was the proportion achieving complete or partial response per RECIST 1.1 with exact 95% CI. Safety was assessed by grade ≥ 3 immune-related adverse events (irAEs) per CTCAE v4.0, graded locally.

All statistical analyses were performed using XLSTAT 2020 Life Sciences software (Addinsoft, Paris, France) and R software (version 4.4.5; R Foundation for Statistical Computing, Vienna, Austria).

Ethics statement

PEMBROBLAD (2022–004) is a multicenter retrospective real-world study promoted by ICANS (Strasbourg Europe Cancer Institute) and conducted in GETUG (Urogenital Tumor Study Group) affiliated centers in France.

The study adhered to the principles outlined in the Declaration of Helsinki (as revised in 2013) and received approval from the French Advisory Committee on Information Processing in Health Research (CCTIRS). The study was approved by the local Ethics Committee with the study approval number 2023-CF025, with compliance to the French policy of individual data protection. All living patients received written information about the research was provided to obtain their non-opposition for participation. For deceased patients, a waiver of information was obtained from the CCTIRS. Patient’s clinical records were retrospectively gathered from electronic files using a de-identified form.

Results

Patient characteristics

Between May 2016 and June 2022, 201 patients diagnosed with UC-V or NUC and treated with anti-PD-(L)1 therapies were identified. Thirty-eight patients were excluded due to incomplete or missing data (n = 8), receipt of combination therapy rather than anti-PD-(L)1 monotherapy (n = 9), localized disease (n = 11), or unconfirmed histological diagnosis (n = 10). Consequently, 163 patients were included in the analysis (Figure S1). The patient characteristics was listed in Table 1. The median follow-up was 21.9 months (95% CI 3.5 to 96.7). The median age was 70 (range 44–89) years, and 96 (72.2%) patients were male. Most patients were current or former smokers (74.3%) and with a Charlson score < 6 (83.7%). Regarding histological type, 79.1% were UC-V while 20.9% were NUC. Within UC-V cases, the most frequent histological HV subtype was squamous cell carcinoma (40.3%), followed by micropapillary (14.7%) and sarcomatoid (11.6%). Most common NUC HV subtypes were squamous cell carcinoma (38.2%), adenocarcinoma (29.4%), and neuroendocrine carcinoma (17.7%). The most frequent metastatic sites were: lymph node (49.1%), lung (28.2%), bone (22.1%), peritoneal carcinomatosis (17.2%) and liver (14.7%). Molecular screening was conducted in 98.2% of patients (N = 160). Among these, 6 tumors (3.7%) harbored ERBB2 amplifications, 3 (1.9%) had FGFR3 alterations, and 25 (15%) carried TP53 mutations. Elevated LDH (Lactate Dehydrogenase) levels were observed in 42% of patients along with high NLR (Neutrophil-to-Lymphocyte Ratio) ≥ 3 was identified in 66.4% of cases.

Table 1.

Characteristics of patients

Characteristics	Total (n = 163)
Age at start of ICB (n = 163)
Median	70
(Range)	(44; 89)
Sex (n = 133)
Female	37 (27.8%)
Male	96 (72.2%)
ECOG-PS (n = 108)
0–1	66 (61.1%)
≥ 2	42 (38.9%)
Tobacco (n = 148)
Non smoker	38 (25.7%)
Smoker	110 (74.3%)
Benzene (n = 162)
Non exposed	154 (95.1%)
Exposed	8 (4.9%)
Charlson score (n = 163)
< 6	123 (83.7%)
≥ 6	24 (16.3%)
Localization (n = 159)
Upper tract	27 (17.0%)
Lower tract	132 (83.0%)
Brain metastases (n = 163)
No	161 (98.8%)
Yes	2 (1.2%)
Lung metastases (n = 163)
No	117 (71.8%)
Yes	46 (28.2%)
Bone metastases (n = 163)
No	127 (77.9%)
Yes	36 (22.1%)
Pleural metastases (n = 163)
No	162 (99.4%)
Yes	1 (0.6%)
Lymph node metastases (n = 163)
No	83 (50.9%)
Yes	80 (49.1%)
Liver metastases (n = 163)
No	139 (85.3%)
Yes	24 (14.7%)
Peritoneal carcinomatosis (n = 163)
No	135 (82.8%)
Yes	28 (17.2%)
Adrenal metastases (n = 163)
No	160 (98.2%)
Yes	3 (1.8%)
Subcutaneous metastases (n = 163)
No	160 (98.2%)
Yes	3 (1.8%)
LDH (n = 100)
N	58 (58%)
> N	42 (42%)
NLR (n = 140)
< 3	47 (33.6%)
≥ 3	93 (66.4%)
TP53 mutation (n = 160)
No	135 (85%)
Yes	25 (15%)
HER2 amplification (n = 160)
No	154 (96.3%)
Yes	6 (3.7%)
FGFR3 alterations (n = 160)
No	157 (98.1%)
Yes	3 (1.9%)
Corticosteroid therapy (n = 163)
No	148 (90.8%)
Yes	15 (9.2%)
Histological (n = 163)
Variants	129 (79.1%)
Pure	34 (20.9%)
Variants (= 129)
Adenocarcinoma	8 (6.2%)
Squamous cell carcinoma	52 (40.3%)
Neuroendocrine	11 (8.5%)
Giant cell carcinomah	1(0.8%)
Lymphoepithelial	2 (1.6%)
Microcystic	1 (0.8%)
Micropapillary	19 (14.7%)
Nested	11 (8.5%)
Plasmacytoid	6 (4.6%)
Pleomorphic	2 (1.6%)
Rhabdoid	1 (0.8%)
Sarcomatoid	15 (11.6%)
Pure (n = 34)
Adenocarcinoma	10 (29.4%)
Squamous cell carcinoma	13 (38.2%)
Neuroendocrine	6 (17.7%)
Giant cell carcinoma	1 (2.9%)
Micropapillary	2 (5.9%)
Sarcomatoid	2 (5.9%)
Immune related adverse events grade 3–5 (CTCAE 4.0) (n = 158)
No	141 (89.2%)
Yes	17 (10.8%)
Colitis	3/17 (17.6%)
Pneumonitis	4/17 (23.5%)
Skin	1/17 (5.9%)
Nephritis	2/17 (11.9%)
Thyroiditis,	3/17 (17.6%)
Rheumatoid arthritis	3/17 (17.6%)
Hypophysitis	1/17 ((5.9%)
First-line treatment (n = 136)
Cisplatin Gemcitabine	27 (19.9%)
Carboplatin Gemcitabine	51 (37.5%)
MVAC (Methotrexate, Vinblastine, Adriamycin, Cisplatin)	37 (27.2%)
Carboplatin VP16	7 (5.2%)
Carboplatin Taxol	6(4.4%)
FOLFOX (Folinic Acid, Fluorouracil, Oxapliplatin)	1 (0.7%)
Paclitaxel Ifosfamide Cisplatin	2 (1.5%)
Gemcitabine Oxaliplatin	3 (2.2%)
Carboplatin only	1 (0.7%)
Cisplatin only	1 (0.7%)

ICB: immune-checkpoint blockade; ECOG-PS: Eastern Cooperative Oncology Group performance status

Among anti-PD-(L)1 agents, 85.5% of patients received pembrolizumab, 12.3% received durvalumab, 1.2% received atezolizumab, and 1.2% received another anti-PD-1 agent. Of note 15 (9.2%) patients were exposed to systemic corticosteroids at time of anti-PD-(L)1 initiation.

Efficacy

The median duration of exposure to anti-PD-(L)1 was 2.4 months. At data cut-off, 119 (73%) patients had died, and tumor progression was observed in 127 (77.9%) cases during follow-up.

In the overall population the median [95% CI] OS was 8.6 [5.3–11.2] months (Fig. 1) and the restricted mean survival time (RMST) at 12 months was 7.0 months (95% CI, 6.2–7.6). The median [95% CI] OS was 8.5 [5.3–10.9] in the UC-V group and 10.5 [3.5–22.8] months in the NUC group (Fig. 2). In univariable analysis, a performance status (PS) ≥ 2 (HR 1.93 [95% CI 1.33–2.82], p = 0.001) and a NLR ≥ 3 (HR 2.04 [95% CI 1.32–3.15], p = 0.001) were both associated with reduced OS (Table 2). In multivariable analysis only PS ≥ 2 was confirmed as independent poor prognostic factors for OS (HR 3.30 [95% CI 1.22–8.92], p = 0.018) (Table 2, Fig. 3).

Fig. 1.

Kaplan–Meier estimates of overall survival in all patients

Fig. 2.

Kaplan–Meier estimates of overall survival according to tumor subtype: urothelial carcinoma with histological variants (UC-V) and non-urothelial carcinomas (NUC)

Table 2.

Univariable and multivariable analyses of overall survival according to clinical and treatment-related factors

Characteristics	Event/N	Univariable		Multivariable
		HR [95%CI]	p-value	HR [95%CI]	p-value
Age at start of ICB   < 65 ans   ≥ 65 ans	41/52 78/111	1.00 0.81 [0.55; 1.17]	0.258	–	–
Sex  Female  Male	25/37 71/96	1.00 0.98 [0.62; 1.55]	0.942	–	–
ECOG-PS  0–1   ≥ 2	69/101 46/56	1.00 1.93 [1.33; 2.82]	0.001	1.00 3.30 [1.22; 8.92]	0.018
Tobacco  Non smoker  Smoker	28/38 79/110	1.00 0.77 [0.50; 1.18]	0.234	–	–
Benzene  Non exposed  Exposed	111/154 6/8	1.00 1.02 [0.45; 2.34]	0.962	–	–
Charlson score   < 6   ≥ 6	107/136 15/21	1.00 0.82 [0.46; 1.46]	0.495	–	–
Localization  Upper tract  Lower tract	21/27 95/132	1.00 0.69 [0.43; 1.12]	0.132	1.00 0.48 [0.14 1.62]	0.237
Brain metastases  No  Yes	118/161 1/2	1.00 0.649 [0.09; 4.65]	0.667	–	–
Lung metastases  No  Yes	89/117 30/46	1.00 0.90 [0.59; 1.36]	0.608	–	–
Bone metastases  No  Yes	87/126 32/37	1.00 1.41 [0.94; 2.12]	0.096	1.00 2.24 [0.66 7.59]	0.194
Pleural metastases  No  Yes	118/162 1/1	1.00 3.74 [0.51; 27.21]	0.293	–	–
Lymph node metastases  No  Yes	61/83 58/80	1.00 0.89 [0.62; 1.28]	0.538	–	–
Liver metastases  No  Yes	99/139 20/24	1.00 1.26 [0.78; 2.05]	0.343	–	–
Peritoneal carcinomatosis  No  Yes	98/135 21/28	1.00 1.29 [0.81 2.07]	0.287	–	–
Adrenal metastases  No  Yes	117/160 2/3	1.00 0.86 [0.21; 3.50]	0.838	–	–
Subcutaneous metastases  No  Yes	116/160 3/3	1.00 1.90 [0.60; 5.99]	0.276	–	–
LDH (n = 100)  N   > N	37/52 35/48	1.00 1.38 [0.87; 2.20]	0.171	1.00 0.78 [0.22; 2.76]	0.705
NLR (n = 140)   < 3   ≥ 3	30/47 60/93	1.00 2.04 [1.32; 3.15]	0.001	1.00 2.10 [0.76; 5.75]	0.150
TP53 (n = 49)  No  Yes	21/24 16/25	1.00 0.56 [0.28; 1.14]	0.108	1.00 0.68 [0.23; 2.01]	0.490
HER2 (n = 49)  No  Yes	33/43 3/6	1.00 0.47 [0.14; 1.53]	0.208	–	–
FGFR3 (n = 49)  No  Yes	35/46 1/3	1.00 0.38 [0.05; 2.80]	0.344	–	–
Corticosteroid therapy (n = 163)  No  Yes	108/148 11/15	1.00 0.97 [0.73; 2.55]	0.325	–	–
Histological (n = 163)  Variants  Pure	98/129 21/34	1.00 0.86 [0.53; 1.38]	0.526	–	–
Histological (n = 163)  Non-sarcomatoid  Sarcomatoid	104/146 15/17	1.00 1.07 [0.62; 1.84]	0.801	–	–
Platinium (n = 134)  Carboplatin  Cisplatin	47/64 53/69	1.00 0.99 [0.67; 1.47]	0.967	–	–
ICB (n = 163)  Other ICB  Pembrolizumab	20/24 99/139	1.00 0.80 [0.49; 1.30]	0.372	–	–

Fig. 3.

Independent prognostic factors associated with overall survival in all patients

When stratified by HV subtype, significant differences were observed. Among UC-V patients, those with nested differentiation was associated with a more favorable OS (HR 0.37 [95% CI 0.00–0.91], p = 0.030) (Table S1). In contrast, within the NUC cohort, squamous cell carcinoma histology correlated significantly with reduced OS (HR 3.28 [95% CI 1.28–8.39], p = 0.013) (Table S2).

In the overall population the median [95% CI] PFS was 2.6 [2.1–3.4] months (Fig. 4). Regarding specific subgroups the median [95% CI] PFS was 2.5 [2.1–3.4] months in the UC-V group and 2.6 [2.0–3.4] months in the NUC group (Fig. 5). In univariable analysis, a PS ≥ 2 (HR 1.79 [95% CI 1.24; 2.60], p = 0.002), and a NLR ≥ 3 (HR 1.67 [95% CI 1.11; 2.51], p = 0.014) were associated with significantly worse PFS (Table S3). In the multivariable analysis, treatment with pembrolizumab versus other ICBs was associated with improved PFS (HR 0.53 [95% CI 0.31; 0.81], p = 0.017), while a PS ≥ 2 was associated with worse PFS (HR 2.32 [95% CI 1.52; 3.54], p = 0.000) (Table 3, Fig. 6).

Fig. 4.

Kaplan–Meier estimates of progression-free survival in all patients

Fig. 5.

Kaplan–Meier curves of progression-free survival according to tumor subtype: urothelial carcinoma with histological variants (UC-V) and non-urothelial carcinomas (NUC)

Table 3.

Univariable and multivariable analyses of progression-free survival according to clinical and treatment-related factors

Characteristics	Event/N	Univariable		Multivariable
		HR [95%CI]	p-value	HR [95%CI]		p-value
Age at start of ICB   < 65 ans   ≥ 65 ans	43/52 84/111	1.00 0.84 [0.58; 1.21]	0.332	–	–
Sex  Female  Male	31/37 76/96	1.00 1.3 [0.85; 1.99]	0.225	–	–
ECOG-PS  0–1   ≥ 2	75/101 47/56	1.00 1.79 [1.24; 2.60]	0.002	1.00 2.32 [1.52; 3.54]	0.000
Tobacco  Non smoker  Smoker	31/38 83/110	1.00 0.88 [0.54; 1.215]	0.305	–	–
Benzene  Non exposed  Exposed	121/154 5/8	1.00 0.64 [0.26; 1.59]	0.340	–	–
Charlson score   < 6   ≥ 6	107/136 15/21	1.00 0.80[0.47; 1.39]	0.436	–	–
Localization  Upper tract  Lower tract	24/27 100/132	1.00 0.68 [0.43 1.07]	0.098	1.00 0.81 [0.50 1.33]	0.410
Brain metastases  No  Yes	125/161 2/2	1.00 1.24 [0.31; 5.00]	0.768	–	–
Lung metastases  No  Yes	91/117 36/46	1.00 1.125 [0.76; 1.66]	0.552	–	–
Bone metastases  No  Yes	96/126 31/37	1.00 1.27 [0.85; 1.91]	0.250	–	–
Pleural metastases  No  Yes	126/162 1/1	1.00 2.94 [0.41; 21.34]	0.285	–	–
Lymph node metastases  No  Yes	70/83 57/80	1.00 0.75 [0.53; 1.06]	0.102	1.00 0.71 [0.48; 1.06]	0.092
Liver metastases  No  Yes	106/139 21/24	1.00 1.53 [0.95; 2.44]	0.079	1.00 1.10 [0.64; 1.87]	0.739
Peritoneal carcinomatosis  No  Yes	104/135 23/28	1.00 1.21 [0.77 1.91]	0.398	–	–
Adrenal metastases  No  Yes	125/160 2/3	1.00 0.76 [0.19; 3.08]	0.699	–	–
Subcutaneous metastases  No  Yes	124/160 3/3	1.00 1.23 [0.39; 3.87]	0.725	–	–
LDH (n = 100)  N   > N	41/52 37/48	1.00 1.14 [0.73; 1.78]	0.560	–	–
NLR (n = 140)   < 3   ≥ 3	34/47 76/93	1.00 1.67 [1.11; 2.513]	0.014	1.00 1.27 [0.79; 2.07]	0.328
TP53 (n = 49)  No  Yes	21/24 16/25	1.00 0.65 [0.33; 1.28]	0.213	–	–
HER2 (n = 49)  No  Yes	34/43 3/6	1.00 1.79 [1.24; 1.007]	0.251	–	–
FGFR3 (n = 49)  No  Yes	35/46 2/3	1.00 0.91 [0.22; 3.83]	0.901	–	–
Corticosteroid therapy (n = 163)  No  Yes	115/148 12/15	1.00 1.21 [0.68; 2.21]	0.525	–	–
Histological (n = 163)  Variants  Pure	101/129 26/34	1.00 0.92 [0.60; 1.42]	0.710	–	–
Histological (n = 163)  Non-sarcomatoid  Sarcomatoid	124/146 15/17	1.00 0.95 [0.56; 1.62]	0.861	–	–
Platinium (n = 134)  Carboplatin  Cisplatin	52/64 53/69	1.00 1.03 [0.70; 1.51]	0.886	–	–
ICB (n = 163)  Other ICB  Pembrolizumab	21/24 106/139	1.00 0.72 [0.45; 1.15]	0.167	1.00 0.53 [0.31; 0.89]	0.017

HR: hazard ratio; ICB: immune-checkpoint blockade; ECOG-PS: Eastern Cooperative Oncology Group performance status; LDH: lactate dehydrogenase level; NLR: neutrophil-to-lymphocyte ratio

Fig. 6.

Independent prognostic factors associated with progression-free survival in all patients

When analyzing by HV subtype, no statistically significant differences were observed among UC-V patients (Table S3). However, in the NUC group, squamous cell carcinoma histology was associated with shorter PFS (HR 2.37 [95% CI 1.05; 5.37], p = 0.039). (Table S4).

Out of the 163 patients included in the study, 137 were evaluable by CT and had available response assessments. Among them, 21 achieved a complete response (CR) and 22 a partial response (PR), resulting in an ORR of 31.4%. Sixteen patients showed stable disease (SD), while 78 experienced progressive disease (PD) as the best response. Among patients achieving an objective response (CR or PR), the median duration of response (DoR) was 12.0 months (95% CI 7.3–15.4).

Safety

During treatment, grade ≥ 3 irAEs occurred in 10.8% of the patients. The main irAEs reported were pneumonitis (4/17), colitis (3/17), thyroiditis (3/17) and arthritis (3/7). No grade 5 irAEs were reported. One patient discontinued treatment due to hypophysitis (Table 1).

Discussion

To our knowledge, the PEMBROBLAD study represents one of the largest real-world cohorts to specifically evaluate the effectiveness of second-line anti–PD(L)1 monotherapy in patients with UC-V and NUC, two subgroups that are often understudied and historically excluded from pivotal ICB trials. Conducted across 24 French centers and including 163 patients treated after progression on platinum-based chemotherapy, this study provides clinically relevant insights into survival outcomes and prognostic factors in this heterogeneous population.

In this cohort, the median OS was 8.6 months and the median PFS was 2.6 months. The ORR was 31.4%, with 15.3% achieving a complete response and 16.1% a partial response. These outcomes appear consistent with prior reports of ICB efficacy in conventional UC, despite the inclusion of more aggressive HV subtypes, particularly NUC, when compared to the results of the Keynote-045 study [19]. Given the very short median PFS observed in our cohort and the nature of responses to ICB, we additionally calculated the RMST up to 12 months for OS. The RMST at 12 months was 7.0 months (95% CI, 6.2–7.6 months), providing a robust measure of survival that does not rely on the proportional hazards assumption and complementing the median OS estimate.

Notably, among patients achieving an objective response, the median DoR was 12.0 months, indicating that despite an overall short PFS, a subset of patients experienced durable clinical benefit, a hallmark of ICB. This observation is in line with the KEYNOTE-045 trial, in which pembrolizumab demonstrated a markedly prolonged median duration of response (29.7 months) compared with chemotherapy (4.4 months), with sustained responses observed in a substantial proportion of patients at long-term follow-up [21].

Importantly, no significant difference in OS or PFS was observed between UC-V and NUC patients overall, although specific HV subtypes such as nested UC-V and squamous NUC emerged as strong prognostic factors for patient’s outcome. Interestingly, when analyzing survival by histologic subtypes, patients with nested UC-V demonstrated a more favorable OS, whereas those with squamous NUC exhibited poorer outcomes. While these findings are based on relatively small numbers and should be interpreted with caution, they are biologically plausible and align with previous reports suggesting that nested variants may retain sensitivity to ICB [22]. In contrast to squamous histology, which is generally associated with more aggressive behavior and lower responsiveness to systemic therapies. Notably, no significant survival advantage was observed for micropapillary UC-V in our cohort, highlighting the heterogeneity of responses among variant histologies and the need for further studies in larger populations to validate these subtype-specific signals [23].

In the multivariable analysis, poor PS (≥ 2) emerged as a strong and independent negative prognostic factor for both OS and PFS. In contrast, treatment with pembrolizumab was associated with prolonged PFS compared to other ICBs. Other variables that showed prognostic value in the univariable analysis, such as the NLR consistent with prior literature [24, 25], did not retain statistical significance in the multivariable model, suggesting potential confounding by other clinical parameters, particularly ECOG PS.

Importantly, the interpretation of our results must be contextualized within the rapidly evolving therapeutic landscape of advanced UC. With the emergence of first-line ICB-based combinations, an increasing proportion of patients with variant histologies are now exposed to ICB earlier in the disease course. Notably, the CheckMate 901 trial included approximately 150 patients with variant or non-urothelial histologies, representing one of the largest prospective datasets in this population, whereas the EV-302 trial included only about 60 such patients. Although dedicated subgroup analyses were not reported, these data suggest that first-line ICB combined with platinum-based chemotherapy, particularly gemcitabine–cisplatin plus nivolumab, should be considered for cisplatin-eligible patients with variant histologies. In current practice, however, many centers still predominantly use enfortumab vedotin plus pembrolizumab in this population, regardless of cisplatin eligibility. Consequently, the future applicability of single-agent ICB in the second-line setting is likely to decrease, as many patients will have already received ICB in the neoadjuvant, maintenance, or first-line metastatic setting.

Nevertheless, our findings remain highly relevant. Demonstrating meaningful single-agent activity of ICB in UC-V and NUC provides critical biological and clinical insights, supporting the intrinsic sensitivity of selected variant histologies to ICB. This information is essential to inform the development of future combination strategies, sequencing approaches, and potential rechallenge paradigms, particularly in rare histologic subtypes where prospective data remain scarce.

In terms of safety, the incidence of grade 3 or 4 irAEs reported remains low (10.8%), compared to those reported in the KEYNOTE‐045 (15.0%). Similarly, the rate of treatment discontinuation due to adverts events was similar: 0.6% in this analysis compared to 5.6% in the KEYNOTE-045. However, it is difficult to draw firm conclusions from retrospective data, as AE reporting and documentation are often underestimated in such settings. No patient in our study died from an AE compared to one patient in KEYNOTE‐024 and 13 patients (2%) in the KEYNOTE‐042 study. These findings suggest the effective management of AEs in routine clinical practice.

Our findings can also be compared to those of the AVENANCE study, a real-world observational analysis evaluating the efficacy of avelumab as first-line maintenance treatment in patients with advanced UC without progression after platinum-based chemotherapy [26]. Although only a small proportion of patients in AVENANCE had NUC or UC-V (7.6% of the cohort), PEMBROBALD draws important parallels with pembrolizumab use in these subgroups. Avelumab demonstrated sustained efficacy in routine clinical settings across all HV subtypes, including UC-V and NUC. Although limited in number, the inclusion of these rare HV highlights AVENANCE’s strength in reflecting everyday clinical diversity and supports our conclusion that ICB remains a relevant option beyond the pure UC context [26].

Landmark advances has been made and had dramatically changed the therapeutic landscape of bladder cancer. The first line enfortumab vedotin and pembrolizumab combination was recently approved following the remarkable results of the phase III EV-302 trial, which demonstrated a doubling of both OS and PFS in patients with advanced or metastatic UC [17]. Preliminary data from the HV subcohort of this trial indicated that patients with squamous or glandular HV (< 50%) exhibited suboptimal responses to this combination therapy, irrespective of the extent of these histologic features. Real-world data form the UNITE trial suggest that the percentage of squamous cell carcinoma is related to response rate to EV without any response for pure SCC. In contrast, other rare HV, such as micropapillary, appeared to maintain favorable responses in the EV 302 trial [27]. However, these observations should be interpreted with caution due to the small number of cases in this subcohort and the fact that the HV represent less than 50% of the whole tumor sample.

Our study has several limitations. First, its retrospective design exposes it to potential selection, information, and unmeasured confounding biases. Missing data for key variables (ECOG-PS, LDH, NLR and molecular data) and the absence of central pathological review, may have limited the power of certain analyses. In addition, RECIST-based radiological assessment was not available for all patients, precluding a centralized and standardized evaluation of tumor response. The small size of molecular and histological subgroups limits the robustness of subgroup comparisons. Furthermore, the absence of centralized pathology review may have introduced heterogeneity in the classification of HV. Finally, although our cohort is among the largest reported to date in patients with UC-V/NUC treated with ICB, prospective independent validation remains necessary to confirm our findings, particularly the prognostic factors identified.

Future studies should move beyond purely histopathological classifications and integrate molecular profiling approaches, including next-generation sequencing, to better characterize these rare and heterogeneous tumors. Emerging data suggest that grouping UC according to basal-like and luminal-like molecular patterns may provide a more biologically relevant framework to predict response to ICB than traditional variant labels alone [28]. Such molecular clustering could help identify distinct immune microenvironment features and refine patient selection for immune checkpoint inhibitor–based strategies. Prospective studies incorporating integrated histo-molecular approaches are therefore warranted in UC-V and non-urothelial carcinomas.

Despite these limitations, the PEMBROBLAD study has several strengths that contribute meaningfully to the growing body of evidence on the efficacy of ICB in UC-V and NUC. This is one of the largest real-world, multicenter studies specifically focused on these rare and often underrepresented HV subtypes of bladder and urinary tract carcinomas. By including data from 24 centers across France, the study captures a broad and diverse patient population, enhancing the generalizability of its findings. Moreover, the study provides valuable real-world insights into the outcomes of pembrolizumab and other anti–PD-(L)1 agents outside the controlled environment of clinical trials. Real-world data are critical for understanding treatment effectiveness in routine practice, where patient characteristics and management strategies often differ from those in pivotal studies. Notably, this study confirms that UC-V remains a HV subtype capable of deriving clinical benefit from ICB therapy, reassuring clinicians that pembrolizumab and related agents remain viable therapeutic options in this context. The study also identifies key clinical determinants of poor outcomes, such as poor baseline performance status, which may help refine patient selection and guide shared decision-making.

Conclusion

In this large real-world cohort, second-line anti–PD-(L)1 therapy showed meaningful activity and a manageable safety profile in patients with advanced UC-V and NUC. Poor performance status was the strongest predictor of poor outcomes, while certain HV subtypes, such as nested UC-V and squamous NUC, were associated with better or worse survival, respectively. These results support the use of ICB in these rare bladder cancer subgroups and highlight the need for prospective validation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Luca Campedel, Mathilde Gout, Fabien Moinard-Butot and Karim Amrane performed material preparation, data collection and analysis. Mathilde Gout, Luca Campedel and Karim Amrane wrote the first draft of the manuscript and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors declare that no funds, grants, or other supports were received during the preparation of this manuscript.

Data availability

The data that support the findings of this study are available from the corresponding author, Karim Amrane, upon reasonable request.

Declarations

Competing interest

KA received honoraria from Astellas, MSD, BMS, and Pfizer. The other authors have no conflicts of interest to declare. MG has no conflicts of interest to declare CD reports occasional engagements for Astellas, BMS, Janssen, MSD and Pfizer FMB has no conflicts of interest to declare HM declares receiving payment or honorarium from Astellas, AstraZeneca, Bayer, Bristol Myers Squibb, Ipsen, Janssen, Merck, MSD, Novartis and Pfizer; travel or other expenses from AstraZeneca, Novartis, Pfizer, Roche; and acting on a scientific steering committee for Curium MB: Financial Interests, Invited Speaker: Janssen, AstraZeneca, Ipsen, Astellas TH has no conflicts of interest to declare PEG has no conflicts of interest to declare DP serves as a consultant/receives honorarium/travel reimbursements from Astellas, Astra Zeneca, Bayer, BMS, MSD, Pfizer MRD has no conflicts of interest to declare CT: Amgen, AstraZenenca, AdaCap, Astellas, Bayer, BMS, Gilead, Ipsen, Johnson & Johnson, Merck, MSD, Novartis, Pfizer, Sanofi, Seagen MC has no conflicts of interest to declare AF declares receiving payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from Astellas, AstraZeneca, AAA, Bayer, Janssen, Novartis, and Pfizer; and support for attending meetings and/or travel from Astellas, AstraZeneca, AAA, Bayer, Janssen, Novartis, and Pfizer OH serves as a consultant/receives honorarium/travel reimbursements from serves as a consultant/receives honorarium/travel reimbursements from ADACAP, Astra Zeneca, Bayer, BMS, Curium, Eisai, Ipsen, Johnson & Jphnson, Merck, MSD, Novartis, Pfizer, Sanofi BA has received research Grants from MSD Avenir; consulting fees or honoraria from Novartis, Astellas, Sanofi, AstraZeneca, BMS, and MSD; and travel accommodation form Janssen, MSD, Pfizer, IPSEN Pharma, Bayer, and Takeda. DB has received honoraria for advisory boards or lectures from Accord Healthcare, Astellas, Astra-Zeneca, Bayer, BMS, Ipsen, Janssen, Merck, Novartis, Pfizer. LD has no conflicts of interest to declare LAP has no conflicts of interest to declare EB has no conflicts of interest to declare PC serves as a consultant/receives honorarium/travel reimbursements from Amgen, Astellas, Astra Zeneca, Bayer, BMS, Eisai, Ipsen, Johnson & Jphnson, Merck, MSD, Pfizer LC serves as a consultant/receives honorarium/travel reimbursements from Amgen, Astellas, Astra Zeneca, Bayer, BMS, Eisai, Ipsen, Johnson & Jphnson, Merck, MSD, Pfizer P.B. received advisory/consulting fees from Astellas Pharma, AstraZeneca, Bayer, Bristol-Myers Squibb, Eisai, Ipsen, Merck, MSD, Pfizer, and Novartis/AdaCaP, research-funding support from Ipsen, AstraZeneca, and BMS.

Ethical approval.

The study conformed to the principles of the Declaration of Helsinki and Good Clinical Practice Guidelines. A local independent Ethics Committee (2023-CF025) approved it.

Consent to participate

Informed consent was obtained from all individual participants included in the study.