Prognostic value of neutrophil-to-lymphocyte ratio as predictive marker of recurrence and progression in non-muscle invasive bladder cancer: a prospective cohort study in a tertiary care center

¹Department of Urology, Tahar Sfar University Hospital, Monastir, Tunisia ²Faculty of Medicine of Monastir, University of Monastir, Interventional Radiology, Research Laboratory, Monastir, Tunisia

^&Corresponding author
Aymen Sakly, Department of Urology, Tahar Sfar University Hospital, Monastir, Tunisia

Introduction    

Bladder cancer (BC) is the second most common urological malignancy in industrialized nations after prostate cancer [1]. At diagnosis, approximately 75% of BC are non-muscle invasive bladder cancer (NMIBC) [2,3]. This group of tumors is particularly characterized by its prognostic heterogeneity [4]. Recurrence and progression remain major concerns in NMIBC, making accurate prognostic assessment critical for guiding treatment. International associations such as the European Association of Urology (EAU) and the American Urological Association (AUA) have developed scoring systems that incorporate endoscopic data (e.g., tumor number and size) and histopathological characteristics (e.g., stage, grade, and presence of carcinoma in situ (CIS) to predict recurrence [3,5]. Improving recurrence and progression prediction remains a challenge. Identifying novel prognostic factors, particularly non-invasive molecular markers, is an area of active investigation.

Among these, the neutrophil-to-lymphocyte ratio (NLR) has emerged as a potentially valuable biomarker in NMIBC follow-up [6]. This marker reflects the tumor-associated inflammatory microenvironment [7]. Meta-analytic evidence has linked elevated NLR to poorer recurrence-free survival in NMIBC [6] and urothelial carcinoma more broadly [8], though its added value over established risk factors remains debated. The aim of this study is to evaluate the independent prognostic value of preoperative NLR in predicting recurrence-free and progression-free survival in patients with NMIBC. Specifically, this study addressed the following research questions: i) is preoperative NLR independently associated with recurrence-free survival after transurethral resection of bladder tumour (TURBT) for NMIBC? ii) is preoperative NLR independently associated with progression-free survival in this population? iii) what is the discriminative accuracy of preoperative NLR for predicting recurrence and progression?

Methods     

Study design and setting: this is a prospective cohort study conducted in Tahar Sfar University Hospital, Mahdia, Tunisia, including 166 patients who underwent transurethral resection of bladder tumour (TURBT) for newly diagnosed NMIBC between March 2016 and December 2024. The sample size was determined to ensure adequate statistical power for the primary endpoint (recurrence-free survival). Based on previous literature reporting hazard ratios for recurrence in the range of 2.0-2.3 among patients with elevated preoperative NLR [6,8], we assumed an expected hazard ratio (HR) of 2.0 for tumor recurrence in patients with an elevated NLR. Assuming a two-sided alpha error of 0.05, a statistical power of 80%, an expected proportion of 40% of patients presenting with an elevated NLR, and an estimated recurrence rate of 50%, a minimum of 68 recurrence events were required, corresponding to a total sample size of approximately 136 patients. Therefore, our final recruited cohort of 166 patients (yielding 87 recurrence events) provides sufficient statistical power to validate the primary hypothesis. Patients were enrolled prospectively at the time of TURBT, prior to knowledge of their eventual recurrence or progression status, and were followed forward in time until the occurrence of a study endpoint (recurrence and/or progression), death, loss to follow-up, or the common study closure date of 31^st December 2024, whichever occurred first, yielding a median follow-up of 47 months (IQR 24). During this follow-up, all patients were monitored according to the risk-adapted surveillance protocol described below (cystoscopy, urine cytology and, when indicated, uro-CT), applied prospectively and uniformly to all participants. This study was reported in line with the STROBE statement [9].

Participants: all patients who underwent TURBT for newly diagnosed NMIBC during the period of study and who met the inclusion criteria were enrolled. Patient flow through the study is summarized in Figure 1.

Exclusion criteria: patients followed for another synchronous or metachronous neoplasia or for autoimmune disease or immune deficiency, patients with a history of pelvic radiotherapy, chemotherapy, or immunotherapy prior to TURBT, patients with a hematological disorder in the complete blood count (CBC) corresponding to an active infection or malignant hematopathy, and patients who received pre-operative blood transfusion were excluded from the study. Patients lost to follow-up were censored at the date of their last clinical evaluation.

Data sources and measurement: for all eligible patients, demographic, clinical, biological, histopathological, and follow-up data were collected. Before TURBT, biological analyses were performed within 7 days before TURBT in our local laboratory. A CBC from peripheral venous samples was performed for all our patients. NLR was calculated by considering neutrophil and lymphocyte counts. Our population was dichotomized into two distinct groups (low and high NLR) according to the cut-off of 2.5. Data were extracted from a structured SPSS case report form based on patients' computerized medical records, and NLR relied on a systematic preoperative blood count available for all patients, ensuring data completeness throughout the recruitment period. After TURBT, a second look was performed within 6-8 weeks in accordance with EAU guidelines and the NMIBC risk calculator [3].

Additionally, patients with intermediate or high-risk NMIBC underwent intravesical bacillus calmette-guerin (BCG) consisting of an induction protocol of six weekly instillations, followed by a monthly protocol instillation for either one or three years. Postoperative surveillance protocols were determined according to risk stratification. In patients with high-risk disease, follow-up included endoscopic evaluation, urine cytology, and uro CT-scan every three months during the first two years, every six months from years two to five, and annually thereafter. Tumor stage was determined using the TNM 2017 classification system, while histologic grading was based on 2004/2016 World Health Organization (WHO) criteria [3]. Cystoscopy and urine cytology were performed during follow-up. During follow-up, recurrence and progression were assessed as defined below (see study endpoints).

Study endpoints: our two endpoints were recurrence and progression, defined as follows. Recurrence was operationally defined as the first pathologically confirmed tumor relapse in the bladder identified on follow-up cystoscopy and/or urine cytology, regardless of the stage or grade of the recurrent lesion; recurrence-free survival was calculated from the date of TURBT to the date of first confirmed recurrence or last follow-up. Progression was operationally defined as any of the following occurring after TURBT: an increase in T category from CIS or Ta to T1 (lamina propria invasion), development of muscle-invasive disease (≥T2), regional lymph node involvement (N+), or distant metastasis (M1), or an increase in histological grade from low to high [3]; progression-free survival was calculated from the date of TURBT to the date of the first confirmed progression event or last follow-up. This broader definition, chosen to capture early markers of tumor aggressiveness and increase the statistical power of the progression analysis, is expected to inflate the observed progression rate relative to studies restricted to the ≥T2 definition and should be considered when comparing our progression-free survival estimates with the literature.

Statistical analysis: all analyses were performed using SPSS version 26. For descriptive comparisons, qualitative variables were compared between NLR groups using the chi-square test. Quantitative variables were first tested for normality using the Kolmogorov-Smirnov test, then expressed as mean ± SD (normal distribution) or median and IQR (non-normal distribution); NLR values were compared between groups using the independent-samples t-test or the Mann-Whitney U test, as appropriate.

To address the research question: whether preoperative NLR is independently associated with recurrence-free survival-recurrence-free survival was measured as the time from TURBT to the first confirmed recurrence or to censoring. This was compared between the low- and high-NLR groups using the Kaplan-Meier estimator with the log-rank test, and univariate and multivariate Cox proportional hazards regression models were fitted, with NLR (≥2.5 vs <2.5), tumor stage, histological grade, tumor size, focality, and concurrent CIS as covariates. The Cox model was chosen because it estimates hazard ratios while properly accounting for censored follow-up time, which is required for this time-to-event outcome. For both models, the proportional hazards assumption was verified using Schoenfeld residuals.

To address research question: whether preoperative NLR is independently associated with progression-free survival-the same analytic approach (Kaplan-Meier estimation with the log-rank test, followed by univariate and multivariate cox regression using the same covariate set) was applied to progression-free survival, measured as the time from TURBT to the first confirmed progression event or to censoring. For both models, the proportional hazards assumption was verified using Schoenfeld residuals.

To address research question: the discriminative accuracy of preoperative NLR-receiver operating characteristic (ROC) curve analysis was performed separately for recurrence and progression, and the area under the curve (AUC) was used as the measure of discriminative accuracy. Although ROC analysis identified a cohort-derived optimal cut-off of 2.145 (sensitivity 67.8%, specificity 70.9%), the literature-based cut-off of 2.5 was retained for the primary cox models to ensure comparability with prior studies [10-12]. To evaluate the robustness of this choice, a sensitivity analysis re-fitted the multivariate cox models using the 2.145 cut-off, adjusted for the same covariate set. No missing data were identified for the variables included in the primary or sensitivity analyses. A two-sided p-value ≤0.05 was considered statistically significant.

Ethical consideration: our study obtained the favorable opinion of the Ethics Local Committee for Biomedical Research of Tahar Sfar University (EAC 45-16). All patients provided informed written consent with guarantees of confidentiality.

Results     

General characteristics of the study population: the study included 166 patients with a mean age of 65.9 ± 12.4 years. Active tobacco use was reported in 151 patients (90.9%), while occupational exposure to organic solvents, particularly aromatic hydrocarbons, was identified in five cases (3%). The most common presenting symptoms of bladder cancer were hematuria and lower urinary tract symptoms (LUTS), observed in 84.3% and 54.2% of patients, respectively. Clinical, endoscopic, and histopathological characteristics of the population are summarized in Table 1. The mean neutrophil-to-lymphocyte ratio (NLR) was 2.17 ± 0.78. Anemia was observed in 66 patients (39.8%). The study population was divided into two groups: a low NLR group (n = 103; 62%) and a high NLR group (n = 63; 38%).

Survival analyses: over a median follow-up period of 47 months (IQR = 28), 87 patients (52.4%) experienced bladder cancer recurrence, with a median recurrence-free survival (RFS) of 24 months (IQR = 24). Among patients who experienced recurrence, tumor progression was observed in 39 cases (44.8% of recurrences), corresponding to an overall progression rate of 23.5% in the entire cohort. This rate is higher than typically reported in NMIBC series, reflecting the broader progression definition used in this study.

Associations of neutrophil-to-lymphocyte ratio with standard prognostic variables: in univariate analysis, preoperative NLR was not associated with endoscopic characteristics of NMIBC. However, a significant association was found between NLR and both tumor stage and histological grade (p = 0.001) (Table 2).

Association of neutrophil-to-lymphocyte ratio with recurrence of non-muscle invasive bladder cancer: the mean NLR values in the recurrence and non-recurrence groups were 2.46 and 1.84, respectively, with a statistically significant difference (p < 0.001). In univariate Cox proportional hazards regression analysis, elevated NLR (≥ 2.5) was a significant predictor of recurrence-free survival (HR = 1.842; 95% CI: 1.206-2.815; p = 0.015), alongside pT1 stage (HR = 3.434; 95% CI: 2.154-5.474; p < 0.001) and high histological grade (HR = 3.475; 95% CI: 2.159-5.594; p < 0.001). In multivariate cox proportional hazards regression analysis, elevated NLR (≥ 2.5) and high histological grade were identified as independent predictors of recurrence-free survival (HR = 1.427; 95% CI: 1.024-2.136; p = 0.046 and HR = 2.403; 95% CI: 1.316-4.390; p = 0.004, respectively). pT1 stage did not retain independent significance in the multivariate model (HR= 1.859; 95% CI: 0.968-3.568; p = 0.063) (Table 3). The proportional hazards assumption was confirmed for this model (global Schoenfeld test, p = 0.289). The ROC curve analysis yielded an area under the curve (AUC) of 0.727, showing acceptable to moderate discriminative performance of NLR in predicting tumor recurrence. The optimal NLR cut-off value was 2.145, providing a sensitivity of 67.8% and a specificity of 70.9% (Figure 2). Kaplan-Meier analysis showed a significantly shorter median recurrence-free survival in the high NLR group compared with the low NLR group (16 vs. 30 months; p = 0.002) (Figure 3).

Association of NLR with progression of non-muscle invasive bladder cancer: the mean NLR was significantly higher in patients with tumor progression compared to those without progression (2.68 vs. 2.01; p < 0.001). When patients were categorized according to NLR levels (low vs. high), histological progression was significantly associated with a higher preoperative NLR value (p < 0.001). In univariate cox proportional hazards regression analysis, elevated NLR (≥ 2.5) was a significant predictor of progression-free survival (HR = 3.752; 95% CI: 1.891-7.442; p < 0.001), alongside pT1 stage (HR = 12.270; 95% CI: 4.348-34.624; p < 0.001) and high histological grade (HR = 11.734; 95% CI: 4.137-33.283; p < 0.001). In multivariate cox proportional hazards regression analysis, NLR ≥ 2.5 (HR = 2.054; 95% CI: 1.169-4.355; p = 0.047), pT1 stage (HR = 3.555; 95% CI: 1.004-12.588; p = 0.049), and high histological grade (HR = 5.837; 95% CI: 1.782-19.123; p = 0.004) were identified as independent predictors of progression-free survival (Table 4). Given the low events-per-variable ratio (6.5) for this model, these estimates, notably the wide confidence interval for pT1 stage, should be interpreted with caution. The proportional hazards assumption was also confirmed for this model (global Schoenfeld test, p = 0.557). The ROC curve analysis demonstrated acceptable predictive accuracy of NLR for tumor progression, with an area under the curve (AUC) of 0.755 (Figure 4). Kaplan-Meier analysis revealed a significantly shorter progression-free survival in the high NLR group compared with the low NLR group (12 vs. 30 months; p = 0.023) (Figure 5).

Neutrophil-to-lymphocyte ratio cutoff interpretation: results were consistent across cutoffs: neutrophil-to-lymphocyte ratio remained a significant independent predictor of both recurrence-free survival (cutoff 2.5: HR = 1.769; 95% CI: 1.122-2.789; p = 0.014; cutoff 2.145: HR = 2.248; 95% CI: 1.373-3.680; p = 0.001) and progression-free survival (cutoff 2.5: HR = 2.546; 95% CI: 1.229-5.276; p = 0.012; cutoff 2.145: HR = 3.180; 95% CI:1.347-7.510; p = 0.008), supporting the robustness of our findings.

Discussion     

Key results: in this prospective cohort of 166 patients with NMIBC followed over a median of 47 months, recurrence and progression occurred in 52.4% and 23.5% of cases, respectively. In multivariate cox analysis, NLR ≥ 2.5 was an independent predictor of both recurrence-free survival (HR= 1.427; 95% CI: 1.024-2.136; p = 0.046) and progression-free survival (HR = 2.054; 95% CI: 1.169-4.355; p = 0.047), alongside high histological grade and pT1 stage. Receiver operating characteristic analysis demonstrated acceptable to moderate discriminative performance for recurrence (AUC = 0.727) and progression (AUC = 0.755).

Comparison with the literature: in our context, bladder cancer represents the most frequent urological malignancy. This high incidence is closely linked to the widespread prevalence of tobacco use in our country, an established major risk factor that was present in 90.9% of our cohort. In terms of NMIBC, better characterization of recurrence risk is a major challenge for urologists. Recently, there has been increased awareness of the need to improve the performance of existing prognostic models by incorporating additional clinical and biological factors [13]. The use of inflammatory markers, particularly NLR, represents a promising and cost-effective biological approach. The precise mechanisms underlying the prognostic performance of NLR have not yet been fully elucidated [14]. However, there are scientific findings that support the adoption of this biological parameter as a prognostic marker in uro-oncology [15]. Bladder cancer is characterized by a high immunogenicity. This reflects the interaction between the immune system and tumor tissue [7]. Neutrophil-to-lymphocyte ratio represents the balance between host pro-tumor inflammatory response, mediated by neutrophils, and anti-tumor immune response involving lymphocytes [2]. During the acute inflammatory phase, neoplasms produce excessive cytokines, including interleukin 8 (IL8) [16]. This cytokine is a neutrophil-activating factor [17]. Neutrophils are involved in different stages of the carcinogenesis process [18]. They participate in tumorigenesis by releasing proteases, reactive oxygen species, and nitrogen [19]. They also contribute to tumor cell proliferation through the secretion of anti-apoptotic markers and growth factors such as vascular endothelial growth factor (VEGF) [20].

The relationship between NLR and risk of recurrence in NMIBC can be explained by the alteration of anti-tumor immunity, which is essentially lymphocyte dependent [21]. High NLR reflects relative lymphopenia, which may impair the protective anti-tumor immune response and increase the risk of recurrence. These pathophysiological mechanisms support the rationale for incorporating NLR into cancer prognostic models. Several studies support our results. A statistically significant association between NLR and recurrence was found in the umbrella review of Mjaess et al.(p<0.001) [14]. A systematic review published by Vartolomei et al. showed that NLR is effective in predicting tumor recurrence [6]. In the meta-analysis of Suh et al. a significant association between pre-operative NLR and RFS (HR = 2.32) has been established [8]. A prospective study conducted by Getzler et al. found a significant difference in the mean of RFS between the two groups (low and high NLR) (p = 0.007) [11]. In accordance with our study, Mbeutcha et al. adopted the cutoff value of 2.5 [10]. A significant association between high NLR and an increased risk of recurrence was also retained in the multivariate regression analysis (p = 0.013). In contrast, Efiloğlu et al. reported in a case control study that NLR is not affected by NMIBC recurrence [22]. However, some studies have reported conflicting results. Favilla et al. and Kang et al. found no significant association between NLR and recurrence, attributing this discrepancy to the relatively low tumor burden in NMIBC, which may limit systemic inflammatory expression [12,15]. The discrepancy between the ROC-derived cut-off (2.145) and the literature-based threshold (2.5) highlights the lack of standardization of NLR values across studies. Likewise, Zhang et al. emphasized the lack of a standardized cutoff value as a major limitation in validating NLR as an accurate biomarker [2]. Regarding tumor progression, our multivariate analysis identified preoperative NLR ≥ 2.5 as an independent predictor alongside stage and grade. This finding is in accordance with recent studies highlighting the prognostic value of systemic inflammatory markers in NMIBC progression. Huang et al. demonstrated in a meta-analysis that elevated NLR was significantly associated with both recurrence and progression in patients receiving intravesical BCG therapy (HR = 1.72; p = 0.004) [23].

Similarly, Ziani et al. reported that NLR > 2.5 independently predicted progression (HR = 2.91; p = 0.01) in a cohort of 300 NMIBC patients [24]. In addition, Bardowska et al. found that incorporating NLR into the European Organisation for Research and Treatment of Cancer (EORTC) risk model significantly improved its predictive accuracy for progression (AUC = 0.618; p = 0.0047) [25]. Of note, these comparator studies [23,24] likely used the conventional ≥ T2 progression definition; the concordant HR direction therefore reflects convergent evidence of NLR's prognostic value, rather than a direct quantitative comparison given our broader definition. Although the discriminative performance of NLR was moderate (AUC = 0.727 for recurrence and 0.755 for progression), these results remain clinically meaningful. In the context of NMIBC, where risk stratification relies on a combination of clinical and pathological parameters rather than a single biomarker, such performance is considered acceptable. Importantly, NLR is an inexpensive, widely available, and non-invasive biomarker that can be easily integrated into routine clinical practice. In high-risk patients with NLR ≥ 2.5, closer surveillance or earlier consideration of adjuvant intravesical therapy may be warranted, pending prospective validation. Therefore, even a moderate discriminative ability may provide incremental prognostic value when combined with established risk factors such as tumor stage and grade. Furthermore, previous studies have reported comparable AUC values for other inflammatory markers in uro-oncology, supporting the external validity of our findings. This suggests that NLR should not be interpreted as a standalone predictive tool, but rather as a complementary parameter that may enhance existing prognostic models such as EORTC or EAU risk stratification systems.

Strengths and limitations: the main strength of this study consists of its homogeneous population and standardized data collection, allowing for comparison of recurrence and progression outcomes. The evaluation of both recurrence and progression within the same population provides an integrated view of NLR's prognostic value. Nevertheless, the single-center design and the ethnic homogeneity may limit the generalizability of these results. The definition of progression used in our study was broader than the conventional definition limited to muscle-invasive disease (≥ T2). This may have resulted in higher progression rates. The low events-per-variable ratio (6.5) for the progression model raises a risk of overfitting, as reflected by the wide confidence interval for the pT1 stage (HR = 3.555; 95% CI: 1.004-12.588), which barely excludes unity. While the overall direction of our findings is consistent with the literature, these individual estimates should be interpreted with caution. In addition, ROC analysis used a standard (static) approach on censored survival data, which may overestimate discriminative performance compared with time-dependent methods. Finally, BCG compliance (completion of the induction and maintenance instillation schedules) was not systematically recorded and could not be included in the multivariate models; as BCG response is strongly associated with recurrence and progression, this omission represents a potential source of residual confounding that should be addressed in future studies.

Perspectives: future multicentric, prospective studies are warranted to validate optimal cutoff values and confirm the clinical applicability of NLR in risk stratification of NMIBC.

Conclusion     

Elevated NLR was significantly associated with higher recurrence and progression rates in NMIBC. As an easily accessible and cost-effective parameter, NLR could improve current risk stratification models and assist urologists in tailoring follow-up and therapeutic strategies for high-risk patients. The value of NLR lies less in its standalone discriminative power than in its potential to refine existing risk stratification models. Nevertheless, prospective multicentric studies are necessary for the external validation of these results.

Competing interests     

The authors declare no competing interests.

Authors' contributions     

Aymen Sakly, Nejmeddine Jleli, Ahmed Ben Brahim, and Elyes Dimassi: contributed to study concept and design, data collection and analysis. Aymen Sakly and Ines Chouaya drafted the manuscript. Walid Zakhama and Yassine Binous: contributed to reviewing and finalizing the manuscript. All authors reviewed the manuscript for intellectual content and approved the submission. All authors have read and agreed to the final manuscript.

Tables and figures     

Table 1: baseline demographic, clinical, endoscopic and histopathological characteristics of the 166 patients treated by transurethral resection of bladder tumour (TURBT) for newly diagnosed non-muscle-invasive bladder cancer (NMIBC) at the Department of Urology, Tahar Sfar University Hospital (Mahdia, Tunisia), from March 2016 to December 2024 (N=166)

Table 2: univariate associations between preoperative neutrophil-to-lymphocyte ratio (NLR) group (low NLR <2.5, n=103; high NLR ≥2.5, n=63) and standard endoscopic and histopathological prognostic variables, among the 166 patients of the cohort described in Table 1 (Tahar Sfar University Hospital, Mahdia, Tunisia, March 2016-December 2024)

Table 3: univariate and multivariate cox proportional hazards regression analysis of factors associated with recurrence-free survival among the 166 patients of the cohort, including 87 recurrence events, treated by TURBT for NMIBC at Tahar Sfar University Hospital (Mahdia, Tunisia) between March 2016 and December 2024

Table 4: univariate and multivariate cox proportional hazards regression analysis of factors associated with progression-free survival among the 166 patients of the cohort, including 39 progression events, treated by TURBT for NMIBC at Tahar Sfar University Hospital (Mahdia, Tunisia) between March 2016 and December 2024

Figure 1: low diagram of patient selection among individuals treated by transurethral resection of bladder tumour (TURBT) for suspected non-muscle-invasive bladder cancer (NMIBC) at Tahar Sfar University Hospital (Mahdia, Tunisia), from March 2016 to December 2024

Figure 2: receiver operating characteristic curve showing the discriminative accuracy of preoperative neutrophil-to-lymphocyte ratio in predicting tumour recurrence among the 166 patients of the study cohort (area under the curve = 0.727)

Figure 3: kaplan-Meier estimates of recurrence-free survival among the 166 patients of the study cohort, stratified by preoperative neutrophil-to-lymphocyte ratio group using the 2.5 cut-off (low group, n=103; high group, n=63; log-rank p = 0.002)

Figure 4: receiver operating characteristic curve showing the discriminative accuracy of preoperative neutrophil-to-lymphocyte ratio in predicting tumour progression among the 166 patients of the study cohort (area under the curve = 0.755)

Figure 5: kaplan-Meier estimates of progression-free survival among the 166 patients of the study cohort, stratified by preoperative neutrophil-to-lymphocyte ratio group using the 2.5 cut-off (low group, n=103; high group, n=63; log-rank p = 0.023)

References