Six-factor-based stratification system for enfortumab vedotin monotherapy in pre-treated patients with metastatic or locally advanced urothelial carcinoma

Abstract

Purpose

This study aimed to identify prognostic factors and develop a prognostic stratification system for progression-free survival (PFS) and overall survival (OS) in patients with locally advanced or metastatic urothelial carcinoma (la/m UC) treated with enfortumab vedotin (EV).

Methods

We retrospectively analyzed 387 patients with la/m UC who received EV monotherapy and evaluated the prognostic factors using the Cox proportional hazards regression model. We then used the significant factors to derive a system for predicting prognosis.

Results

The median PFS and OS were 7.6 and 14.6 months, respectively. Multivariable analysis revealed six significant factors common to both PFS and OS: poor Eastern Cooperative Oncology Group performance status (≥ 1), presence of aberrant histology (histological subtypes or divergent differentiation), presence of bone metastasis, anemia (hemoglobin < 10 g/dL), high C-reactive protein levels (> 0.5 mg/dL), and hypercalcemia (corrected calcium > 10.2 mg/dL). Based on the number of positive factors, patients were categorized into three groups: 147 patients with none or one factor (low-risk group), 92 patients with two factors (intermediate-risk group), and 148 patients with three or more factors (high-risk group). The log-rank test indicated significant differences in both PFS and OS among the three groups, with C-indices of 0.634 and 0.700 for PFS and OS, respectively.

Conclusions

Six factors were identified as common prognostic factors for both PFS and OS in la/m UC patients receiving EV monotherapy. The newly developed prognostic risk stratification system based on these factors is clinically useful, particularly for predicting OS.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00345-026-06765-8.

Introduction

In 2021, the EV-301 trial demonstrated the oncological benefit of enfortumab vedotin (EV) monotherapy over standard chemotherapy, establishing its use for patients with locally advanced or metastatic urothelial carcinoma (la/m UC) [1]. Subsequently, 24-month findings from the EV-301 trial confirmed the progression-free survival (PFS) and overall survival (OS) benefits for EV against single-agent chemotherapy [2]. Consequently, major guidelines recommend EV monotherapy as an effective later-line therapy for patients who have not previously received EV plus pembrolizumab combination therapy [3, 4].

Accurately predicting oncological outcomes and stratifying the prognostic risk groups of a systemic therapy are essential for determining and optimizing treatment strategies; however, existing data on these factors are limited and often derived from a comparatively small number of patients [5–13]. Therefore, we conducted a multicenter retrospective study comprising a total of 482 la/m UC patients who were treated with EV monotherapy, aiming to determine the prognostic factors and develop a system for predicting clinical outcomes.

Patients and methods

Patients

This study conformed to the Declaration of Helsinki and was approved by the Institutional Review Board of the Graduate School of Medicine and Faculty of Medicine, The University of Tokyo (approval number: 2024246NI-(2)), as well as the ethics committees of all participating institutions. An opt-out method was used to obtain informed consent from the patients because of the retrospective nature of the study. EV was administered at a dose of 1.25 mg/kg on days 1, 8, and 15 of a 28-day cycle. The dosage was reduced according to the institutional policy or the physicians’ discretion, considering the patient’s condition.

Clinical data were retrospectively collected from 27 participating institutions. Briefly, this cohort comprised consecutive 482 patients who were diagnosed with la/m UC and received EV monotherapy between March 2021 and October 2025. This dataset contained information on the patients’ clinical characteristics, laboratory data, and treatment outcomes, including survival data, which were collected in November 2025. The following clinicopathological data of the patients were obtained from their medical records at each hospital: age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS), body mass index (BMI), primary cancer site, histology of the tumor, prior chemotherapy, prior immune-oncology (I-O) therapy, the site of metastasis at the initiation of EV monotherapy, hemoglobin level, albumin level, estimated glomerular filtration rate (eGFR), neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP) level, and corrected calcium (cCa) level. Collectively, we evaluated 387 patients in this cohort after excluding 95 patients with insufficient laboratory data and histopathological details.

Statistical analysis

Statistical analyses were performed using SPSS v28.0 (IBM Corp., Chicago, IL, USA), EZR v1.52 (Saitama Medical Center, Jichi Medical University), and R v4.1.3 (R Foundation for Statistical Computing, https://www.r-project.org/); P-values < 0.05 were considered significant. The follow-up period was defined as the time from EV initiation to either death or final observation. PFS was calculated as the length of time from EV initiation to radiologically diagnosed progression according to Response Evaluation Criteria in Solid Tumors v1.1 or overall death according to the Kaplan–Meier method. OS was defined as the length of time from EV initiation to overall death according to the Kaplan–Meier method. The prognostic impact of clinicopathological factors was assessed by univariable and multivariable analyses using the Cox proportional hazards regression model. The NLR cutoff values were determined using the Youden index of OS. From these results, we derived a risk classification system to predict the PFS and OS, and calculated the corresponding C-indices. Internal validation of the Cox proportional hazards model was performed using bootstrap resampling with 1,000 repetitions. Model discrimination was assessed using the optimism-corrected C-index.

Results

The clinicopathological characteristics of the patients are shown in Table 1. The median age was 74 years, and 72.6% of patients were male. The incidence of associated upper urinary tract urothelial carcinoma (UTUC) was 53.8% and ECOG PS ≥ 2 was 16.5%. The proportion of histological subtypes or divergent differentiation, collectively referred to here as aberrant histology [14], was 14.2%. Although 80.6% of patients had an eGFR of 60 mL/min/1.73m² or less at the initiation of EV monotherapy, more than half of the patients had previously received a cisplatin-based regimen. The median number of previous systemic therapies was two (range, 1–5). Regarding prior I-O drug use, which included overlapping cases, the most common I-O agent was pembrolizumab (79.3% of the prior I-O cases). Conversely, the proportion of nivolumab administration was only 10.6% due to the timing of its clinical introduction; nivolumab was approved as a neoadjuvant therapy in 2022 in Japan. The incidence of metastasis at the initiation of EV was 67.4%, 43.7%, 24.0% and 22.7% in lymph node, lung, bone and liver, respectively. Regarding the laboratory data, anemia (Hb < 10 mg/dL), high CRP level (≥ 0.5 mg/dL), and hypercalcemia (cCa ≥ 10.2 mg/dL) were observed in 33.6%, 66.4%, and 15.0%, respectively. In the supplementary Table 1, we compared the baseline characteristics between the included 387 patients to excluded 95 patients.

Table 1.

Clinicopathological characteristics

Variable	N = 387
Age (median, interquartile range) (years)	74, (67–79)
Sex, no. (%)
Male	281 (72.6)
Female	106 (27.4)
ECOG PS, no. (%)
0	183 (47.3)
1	140 (36.2)
≥ 2	64 (16.5)
BMI (median, interquartile range) (kg/m2)	22.1 (19.9–24.3)
Primary site, no. (%)
Bladder	179 (46.3)
Renal pelvis/ureter	159 (41.1)
Both the bladder and renal pelvis/ureter	49 (12.7)
Histology, no. (%)
Pure UC	332 (85.8)
UC with aberrant histology	55 (14.2)
Resection of the primary tumor, no. (%)
Yes	234 (60.5)
No	153 (39.5)
Prior chemotherapy, no. (%)
Cisplatin-based regimen	226 (58.4)
Carboplatin-based regimen	155 (40.0)
Other	6 (1.6)
Prior immune checkpoint inhibitor, no. (%) *
Pembrolizumab	307 (79.3)
Avelumab	176 (45.5)
Nivolumab	41 (10.6)
Number of previous therapies, no. (%)
≤ 2	313 (80.9)
≥ 3	74 (19.1)
Metastatic site, no. (%)
Lymph node	261 (67.4)
Lung	169 (43.7)
Bone	93 (24.0)
Liver	88 (22.7)
Hemoglobin (g/dL), no. (%)
≥ 10	257 (66.4)
< 10	130 (33.6)
Albumin (mg/dL), no. (%)
≥ 4.0	93 (24.0)
< 4.0	294 (76.0)
eGFR (mL/min), no. (%)
≥ 60	75 (19.4)
< 60	312 (80.6)
NLR, no. (%)
< 5	231 (59.7)
≥ 5	156 (40.3)
CRP (mg/dL), no. (%)
< 0.5	130 (33.6)
≥ 0.5	257 (66.4)
cCa (mg/dL), no. (%)
< 10.2	329 (85.0)
≥ 10.2	58 (15.0)

ECOG PS, Eastern Cooperative Oncology Group performance status; BMI, body mass index; eGFR, estimated glomerular filtration rate; NLR, neutrophil-to-lymphocyte ratio; CRP, C-reactive protein; cCa, corrected calcium

*Including duplicate cases

The median follow-up period was 9.7 months (interquartile range, 4.7–16.8). During the observation period, disease progression occurred in 286 patients (73.9%), and overall death occurred in 204 patients (52.7%). The median PFS and OS were 7.6 months (95% CI: 6.4–8.7) and 14.6 months (95% CI: 12.3–16.8), respectively.

To investigate the prognostic factors for PFS and OS, Cox proportional hazard analysis was performed on 387 patients. Multivariable analysis identified the following poor prognostic factors for PFS: poor ECOG PS, history of primary site resection, presence of aberrant histology, bone metastasis, liver metastasis, anemia, high CRP level, and hypercalcemia (Table 2). In addition, the following prognostic factors were significant for OS: poor ECOG PS, presence of aberrant histology, prior administration of avelumab, bone metastasis, anemia, high NLR, high CRP level, and hypercalcemia (Table 3). Collectively, six common prognostic factors were determined: poor ECOG PS, presence of aberrant histology, bone metastasis, anemia, high CRP level, and hypercalcemia. To assess potential multicollinearity among these six factors, Spearman’s correlation coefficients were calculated; no pair demonstrated an absolute correlation coefficient > 0.7 (Supplementary Table 2). Based on the number of positive factors, patients were categorized into three groups: 147 patients with none or one factor (38.0%, low-risk group), 92 patients with two factors (23.8%, intermediate-risk group), and 148 patients with three or more factors (38.2%, high-risk group). Because the hazard ratios ranged from 1.31 to 2.07 (a maximum difference of 1.58-fold), all factors were assigned equal weight. The median PFS and OS were 14.1, 7.9, and 6.7 months and 32.3, 16.2, and 8.1 months in the low-risk, intermediate-risk, and high-risk groups, respectively. The Kaplan–Meier curves obtained according to the risk stratification are shown in Fig. 1, and the log-rank test indicated significant differences in both PFS and OS among the groups. The C-indices of the risk stratification model were calculated to be 0.634 for PFS and 0.700 for OS. We performed internal validation using bootstrapping, and the optimism-corrected C-indices were 0.635 for PFS and 0.700 for OS.

Table 2.

Cox proportional hazards regression analysis for overall survival

Variable	Univariable analysis		Multivariable analysis
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)
< 70	Reference	0.47	-	-
≥ 70	0.90 (0.68–1.20)
Sex
Female	Reference	0.34	-	-
Male	1.08 (0.83–1.25)
ECOG PS
0	Reference	< 0.001	Reference	0.007
≥ 1	1.74 (1.32–2.31)		1.50 (1.12–2.00)
BMI (kg/m2)
< 22	Reference	0.007	Reference	0.16
≥ 22	0.68 (0.52–0.90)		0.81 (0.61–1.09)
Primary site  Lower only  Upper/ Lower & Upper	Reference 1.12 (0.85–1.47)	0.44
Histology
Pure UC	Reference	0.001	Reference	0.001
UC with aberrant histology	1.35 (1.14–1.61)		1.88 (1.31–2.70)
Primary site resection
No	Reference	0.67
Yes	1.06 (0.80–1.41)
Prior Cisplatin use
No	Reference	0.28
Yes	1.08 (0.94–1.24)
Prior Avelumab use
Yes	Reference	0.030	Reference	0.039
No	1.40 (1.03–1.90)		1.39 (1.02–1.91)
Number of prior therapies
≤ 2	Reference	0.19	-	-
≥ 3	1.24 (0.90–1.71)
Lymph node metastasis
No	Reference
Yes	1.09 (0.81–1.47)	0.55
Lung metastasis
No	Reference	0.80	-	-
Yes	1.04 (0.79–1.37)
Bone metastasis
No	Reference	< 0.001	Reference	0.043
Yes	1.85 (1.37–2.49)		1.38 (1.01–1.88)
Liver metastasis
No	Reference	0.016	Reference	0.19
Yes	1.45 (1.07–1.98)		1.24 (0.90–1.71)
Hemoglobin (g/dL)
≥ 10	Reference	< 0.001	Reference	< 0.001
< 10	2.42 (1.83–3.19)		1.88 (1.39–2.54)
Albumin (mg/dL)
≥ 4.0	Reference	< 0.001	Reference	0.63
< 4.0	1.90 (1.32–2.73)		0.90 (0.60–1.37)
eGFR (mL/min)
≥ 60	Reference	0.29
< 60	0.83 (0.60–1.17)
NLR
< 5	Reference	< 0.001	Reference	0.013
≥ 5	2.44 (1.85–3.22)		1.48 (1.09–2.01)
CRP (mg/dL)
< 0.5	Reference	< 0.001	Reference	0.002
≥ 0.5	3.20 (2.25–4.54)		1.92 (1.27–2.88)
cCa (mg/dL)
< 10.2	Reference	< 0.001	Reference	0.048
≥ 10.2	2.43 (1.74–3.40)		1.44 (1.00–2.08)

ECOG PS, Eastern Cooperative Oncology Group performance status; BMI, body mass index; UC, urothelial carcinoma; eGFR, estimated glomerular filtration rate; NLR, neutrophil-to-lymphocyte ratio; CRP, C-reactive protein; cCa, corrected calcium

Fig. 1.

a Kaplan–Meier curves of progression-free survival, and the P-values calculated among the three risk groups using the log-rank test. b Kaplan–Meier curves of overall survival, and the P-values calculated among the three risk groups using the log-rank test

Discussion

EV monotherapy afforded a median PFS of 5.55 months and a median OS of 12.91 months in the long-term analysis of the EV-301 trial [2]. Thus, it is recommended as a later-line therapy in global guidelines [3, 4]. EV has been used in clinical practice worldwide, and several real-world retrospective studies have been reported. In comparatively large cohort studies, the median PFS and OS were 4.8–6.8 months and 10.0–14.5 months, respectively, comparable to the results of the present study [5–8]. However, definitive prognostic factors have not been established due to inconsistent findings among studies and insufficient sample sizes for robust multivariable analysis. To address this, we identified independent prognostic factors and developed a risk stratification system for EV monotherapy in patients with la/m UC.

Compared with the EV-301 trial, this study included older patients, a higher incidence of UTUC, and poorer ECOG PS. Additionally, the proportion of patients treated with more than three lines of prior therapy was higher in the present study [1].

As detailed in Tables 2 and 3, we identified six significant factors common to both PFS and OS: poor ECOG PS (≥ 1), presence of aberrant histology, presence of bone metastasis, anemia (hemoglobin < 10 g/dL), high CRP level (> 0.5 mg/dL), and hypercalcemia (cCa > 10.2 mg/dL). An ECOG PS greater than 0 and a hemoglobin level less than 10 g/dL are also components of the Bellmunt risk score, a well-known prognostic factor for previously treated UC patients [15]. Although the Bellmunt risk score includes the presence of liver metastasis as a prognostic factor, it did not reach significance for OS in this study. Notably, the subgroup analysis of the EV-301 trial revealed a lower hazard ratio in patients with liver metastasis [2]. Additionally, many previous studies have indicated that liver metastasis does not significantly affect prognosis in this setting [5, 8, 10, 16]. Conversely, the presence of bone metastasis was a significant prognostic factor in this study, aligning with the univariable analyses of several previous reports [8–10]. While most previous studies did not identify histology as a significant prognostic factor [5, 6], a study specifically focused on histology reported shorter PFS [17]. Our large cohort study confirmed the significance of aberrant histology, including histological subtypes and divergent differentiation in both PFS and OS. This may be due to the mechanism of EV, which targets Nectin-4; research suggests that Nectin-4 is expressed less in UC with divergent differentiation than in pure UC [18]. Furthermore, numerous studies have identified high CRP levels as a significant prognostic factor, with reviews indicating a cutoff value of 0.5 mg/dL [19], consistent with the present study. CRP is an accessible and objective risk factor, making this cutoff value clinically relevant. Moreover, hypercalcemia emerged as a significant prognostic factor, as we have previously observed [16]. In patients with malignant tumors, hypercalcemia is induced by the production of parathyroid hormone-related peptides, osteolytic cytokines, and excess 1,25-dihydroxyvitamin D [20]. While hypercalcemia is an established prognostic factor in metastatic renal cell carcinoma [21], it has rarely been evaluated in UC patients. However, the case reports of UC-induced hypercalcemia [22, 23] and the results of this study suggest that hypercalcemia is a useful prognostic factor for EV monotherapy. In the subgroup analysis of EV-301 trial, more than three lines of prior systemic therapy group did not show statistical preference of EV compared to chemotherapy [2]. Although the number of patients was limited, the present study described the lines of prior systemic therapy did not influence the efficacy of EV monotherapy (Tables 2 and 3).

Using the six significant prognostic factors common to both PFS and OS, we developed a simple and objective risk stratification system. The establishment of a novel risk stratification system based on factors significantly associated with both PFS and OS has been reported previously [24, 25]. This approach may enhance the clinical applicability of the risk stratification system because using separate systems composed of different prognostic factors for PFS and OS would complicate its use in clinical practice. Specifically, we calculated cumulative risk scores for each patient, allowing for stratification into three distinct groups (low-, intermediate-, and high-risk). We observed significant differences in both PFS and OS among the three groups (Fig. 1). Furthermore, the optimism-corrected C-index of 0.700 for OS is considered acceptable for clinical prognostic models. The C-indices of the present risk classification were higher than those of the Bellmunt risk score, which were 0.667 for OS and 0.612 for PFS. In addition, we presented PFS and OS separately for patients with bladder cancer alone (Supplementary Fig. 1) and for those with UTUC alone or both bladder cancer and UTUC (Supplementary Fig. 2). Significant differences among the three risk groups were observed, particularly for OS. The optimism-corrected C-indices were 0.641 for PFS and 0.689 for OS in patients with bladder cancer alone, and 0.625 for PFS and 0.710 for OS in patients with UTUC alone or both bladder cancer and UTUC.

There are several limitations to this study. First, the clinical characteristics of future candidates for EV monotherapy may differ from those of the patients included in the present study, considering that guidelines have recently recommended EV plus pembrolizumab therapy as a first-line treatment. The evolving role of EV monotherapy is as a later-line treatment for patients who are ineligible for checkpoint inhibitors, those who have received platinum-based chemotherapy followed by avelumab maintenance, and those who experience recurrence after radical surgery despite perioperative systemic therapy consisting of platinum-based chemotherapy and a checkpoint inhibitor. Second, follow-up schedules and EV treatment protocols were not standardized because the data were collected retrospectively from multiple institutions. Third, the new risk stratification system is prognostic rather than predictive; therefore, its clinical utility for treatment selection and its cost-effectiveness were not evaluated. Next, as shown in Supplementary Table 1, the excluded group had poorer PS, a higher incidence of UTUC, and a lower proportion of patients who underwent surgery for the primary tumor. In addition, this six-factor model was derived from a Japanese (single-ethnicity) patient population. These may limit its generalizability. Finally, the findings presented in this study have not been validated using independent data from external cohorts.

In conclusion, we identified ECOG PS, histology, bone metastasis, anemia, high CRP, and hypercalcemia as significant prognostic factors for both PFS and OS in la/m UC patients receiving EV monotherapy. Our newly developed risk stratification system based on these six factors is clinically useful, particularly for predicting OS in this population.

Table 3.

Cox proportional hazards regression analysis for progression-free survival

Variable	Univariable analysis		Multivariable analysis
	HR (95% CI)	P-value	HR (95% CI)	P-value
Age (years)
< 70	Reference	0.18	-	-
≥ 70	0.85 (0.67–1.08)
Sex
Female	Reference	0.15	-	-
Male	1.10 (0.97–1.25)
ECOG PS
0	Reference	< 0.001	Reference	0.029
≥ 1	1.58 (1.25–2.00)		1.31 (1.03–1.67)
BMI (kg/m2)
< 22	Reference	0.19	-	-
≥ 22	0.86 (0.68–1.08)
Primary site
Lower only	Reference	0.66	-	-
Concomitant with upper	1.05 (0.84–1.33)
Histology
Pure UC	Reference	< 0.001	Reference	< 0.001
UC with aberrant histology	1.76 (1.29–2.39)		2.07 (1.50–2.84)
Primary site resection
No	Reference	0.033	Reference	0.035
Yes	0.78 (0.61–0.98)		1.29 (1.02–1.64)
Prior Cisplatin use
No	Reference	0.56
Yes	1.04 (0.92–1.16)
Prior Avelumab use
Yes	Reference	0.61	-	-
No	1.07 (0.84–1.36)
Number of prior therapies
≤ 2	Reference	0.078	-	-
≥ 3	1.278 (0.97–1.68)
Lymph node metastasis
No	Reference	0.82	-	-
Yes	1.03 (0.81–1.32)
Lung metastasis
No	Reference	0.18	-	-
Yes	1.17 (0.93–1.48)
Bone metastasis
No	Reference	< 0.001	Reference	0.019
Yes	1.59 (1.23–2.06)		1.39 (1.06–1.82)
Liver metastasis
No	Reference	0.005	Reference	0.007
Yes	1.46 (1.12–1.89)		1.46 (1.11–1.92)
Hemoglobin (g/dL)
≥ 10	Reference	< 0.001	Reference	0.004
< 10	1.86 (1.46–2.35)		1.46 (1.13–1.90)
Albumin (mg/dL)
≥ 4.0	Reference	0.008	Reference	0.768
< 4.0	1.47 (1.11–1.96)		0.95 (0.69–1.32)
eGFR (mL/min)
≥ 60	Reference	0.21
< 60	0.84 (0.63–1.11)
NLR
< 5	Reference	< 0.001	Reference	0.275
≥ 5	1.66 (1.32–2.09)		1.16 (0.89–1.50)
CRP (mg/dL)
< 0.5	Reference	< 0.001	Reference	0.044
≥ 0.5	2.01 (1.55–2.62)		1.38 (1.01–1.88)
cCa (mg/dL)
< 10.2	Reference	< 0.001	Reference	0.004
≥ 10.2	2.06 (1.52–2.78)		1.62 (1.17–2.25)

ECOG PS, Eastern Cooperative Oncology Group performance status; BMI, body mass index; UC, urothelial carcinoma; eGFR, estimated glomerular filtration rate; NLR, neutrophil-to-lymphocyte ratio; CRP, C-reactive protein; cCa, corrected calcium

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Project development: Yuto Matsushita, Taku Naiki, Satoru Taguchi, Takahiro Yasui, Haruki Kume, Teruo Inamoto. Data collection: Yuto Matsushita, Taku Naiki, Satoru Taguchi, Gaku Ishikawa, Yu Nakamura, Jimpei Miyakawa, Takashi Nagai, Tomoyuki Kaneko, Hiroki Kubota, Hayato Hoshina, Jun Kamei, Aya Niimi, Yuta Yamada, Daisuke Obinata, Shigenori Kakutani, Masashi Shiozaki, Kento Ozawa, Michio Noda, Daichi Hirata, Yuichi Kitagawa, Yukihiro Yoshimi, Nozomi Sugiyama, Nobuhiko Shimizu, Yohei Tsubouchi, Keigo Sato, Kosuke Uchida, Shinya Watanabe, Jun Takahashi, Toshiharu Morikawa, Yoichi Mori, Takashi Murata, Taro Murata, Taketo Kawai, Takehiro Tanaka, Yukari Uemura, Tappei Takada, Sayuri Takahashi, Toshiki Ito, Masao Nagata, Hideyo Miyazaki, Satoshi Kurokawa, Yukio Yamada, Ryo Sato, Atsushi Otsuka, Yusuke Sato, Hiroaki Nishimatsu, Yukihiro Umemoto, Hiroshi Furuse, Yoshiyuki Akiyama, Yutaka Enomoto, Satoru Takahashi, Tetsuya Fujimura, Tohru Nakagawa, Hiroshi Fukuhara. Data analysis: Yuto Matsushita, Taku Naiki, Satoru Taguchi. Manuscript writing: Yuto Matsushita. Manuscript editing: Taku Naiki, Satoru Taguchi, Takahiro Yasui, Haruki Kume, Teruo Inamoto. Approval of the final manuscript: all authors.

Funding

Open Access funding provided by The University of Tokyo. No funding was received.

Data availability

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

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical approval

The study protocol was reviewed and approved by the Institutional Review Board of the Graduate School of Medicine and Faculty of Medicine, The University of Tokyo (approval number: 2024246NI-(2)), as well as that of each participating institution.

Consent to participate

Informed consent was obtained from participants or deceased patients’ families, using the opt-out method because of the retrospective nature of the study.