Preoperative Prognostic Nutritional Index and Systemic Immune-Inammation Index for Predicting the Long-Term Recurrence-Free Survival of Patients with High-Risk Non-Muscle-Invasive Bladder Cancer After Transurethral Resection of a Bladder Tumour: A Cohort Study

Background The predictive values of preoperative prognostic nutritional index (PNI) and systemic immune-inammation index (SII) for long-term recurrence-free survival (RFS) in high-risk non-muscle-invasive bladder cancer (NMIBC) have not been fully elucidated. Hence, this study aimed to examine the associations between PNI and SII as well as RFS in patients with high-risk NMIBC who received intravesical instillation of Bacillus Calmette–Guerin (BCG) after transurethral resection of bladder tumour (TURBT). Methods We retrospectively collected data on 387 high-risk NMIBC patients between January 2004 and December 2014. Preoperative PNI was calculated as albumin concentration (g/L)+5×total lymphocyte count (10 9 /L), while preoperative SII was dened as neutrophil count (10 9 /L)×platelet count (10 9 /L)/lymphocyte count (10 9 /L). The optimal cut-off values of PNI and SII were determined via a receiver operating characteristic analysis. RFS was evaluated via a Kaplan–Meier analysis. Between-group differences were compared using the log-rank test. Univariate and multivariate Cox regression analyses were performed to assess the predictive values of PNI and SII for RFS. Results Patients were divided into two groups according to the cut-off values of PNI (<50.17 vs ≥ 50.17) and SII (<467.76 vs ≥ 467.76). Kaplan-Meier analyses revealed that low PNI and high SII were associated with poor RFS in patients with high- and higher-risk NMIBC. Furthermore, based on univariate and multivariate Cox regression analyses, PNI and SII were independent predictive factors of RFS in patients with high-risk NMIBC. NY, USA), Sigma Jose, CA, GraphPad Prism version Software CA, were used in statistical analyses. Continuous data the mean standard deviation (SD) or median with interquartile range. Categorical variables are presented as frequencies and percentages and were compared using Pearson’s chi-squared test. Receiver operating characteristic (ROC) curve analysis was then applied to identify the optimal cut-off values of PNI and SII with the highest Youden’s index for predicting the 5-year overall survival (OS). The end point of this retrospective cohort study was the RFS. The Kaplan–Meier method and log-rank test were used to analyse RFS and compare RFS curves. A Cox proportional hazards regression model was used to analyse various independent predictors of postoperative RFS. Univariate analysis was used to assess various factors related to patients’ clinicopathological characteristics. Multivariate analysis was performed for variables with p < 0.1 in univariate analysis. In two-tailed tests, p < 0.05 was considered statistically signicant.

with poor RFS in patients with high-and higher-risk NMIBC. Furthermore, based on univariate and multivariate Cox regression analyses, PNI and SII were independent predictive factors of RFS in patients with high-risk NMIBC.
Conclusion Preoperative PNI and SII are simple, noninvasive, inexpensive and useful tools for predicting long-term RFS among patients with high-risk NMIBC who received intravesical instillation of BCG after TURBT.

Background
Bladder cancer (BC) has been identi ed as the 7 th most frequent cancer in the male population and 11 th when considering both genders [1]. It ranks as the ninth leading cause of cancer deaths in the male population worldwide as BC is more common in men than in women, with respective incidence and mortality rates of 9.6 and 3.2 per 100,000 [2]. The mortality and morbidity of BC vary across countries and regions due to differences in related risk factors, detection and diagnostic practices and availability of treatments. BC is classi ed into non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC) subtypes according to distinct clinical recurrence, progression and prognosis.
Approximately 75% of BC patients present with a disease con ned to the mucosa or submucosa and are eventually diagnosed with NMIBC [3]. The NMIBC subtype can be further distinguished as Ta (noninvasive papillary carcinoma), T1 (tumour invading subepithelial connective tissue) or carcinoma in situ (CIS, ' at tumour'). The European Association of Urology (EAU) Guidelines 2019 de ned the high-risk NMIBC group as follows: (1) T1 tumours; (2) G3 (high-grade [HG]) tumours; (3) CIS and (4) multiple, recurrent and large (>3 cm) TaG1G2/low-grade tumours (all features must be present) [4]. In addition, based on prognostic factors, the highest-risk NMIBC subgroup includes T1G3/HG tumours associated with concurrent bladder CIS; multiple and/or large T1G3/HG and/or recurrent T1G3/HG tumours; T1G3/HG tumours with CIS in the prostatic urethra; some forms of variant histology of urothelial carcinoma; and lymphovascular invasion.
Transurethral resection of bladder tumour (TURBT) is a typical rst line treatment for NMIBC patients.
Meanwhile, low-risk NMIBCs only undergo TURBT, whereas for intermediate-and high-risk NMIBCs, TURBT with Bacillus Calmette-Guerin (BCG) immunotherapy or chemotherapy is the standard treatment.
The clinical behaviour of NMIBC is largely unpredictable, and even patients treated in accordance to EAU recommendations have a heterogeneous prognosis. Often, high-risk NMIBC patients have high rates of disease recurrence and progression to MIBC and BCG treatment failure [5]. As the highest-risk subtype of NMIBC, high-grade T1 (HGT1) BC presents an almost 40% rate of recurrence and 20% of progression at 5 years [6]. Recurrence-free survival (RFS) rates were 54.9%, 55.3% and 60.4% for primary, concomitant and secondary CIS, respectively, at 5-year follow-up. Progression-free survival rates at 5-year follow-up were 65.8%, 72.1% and 77% for primary, concomitant and secondary CIS, respectively [7]. Thus, the strati cation of risk groups based on recurrence and progression is a strategy that attempts to individualise and standardise BC patients.
A patient's nutritional and immune status has an important impact on oncological outcomes [8]. The prognostic nutritional index (PNI) was rst introduced by Onodera et al. [9] in 1984. It was used to evaluate the risk of postoperative complications among patients with gastric cancer according to baseline nutritional status. Growing evidence has shown that a low PNI score is signi cantly associated with major postoperative complications and poor prognosis [10,11]. The systemic immune-in ammation index (SII) was rst used to evaluate hepatocellular carcinoma and was considered a predictor of several solid tumours, particularly prostate cancer and metastatic non-clear cell renal cell carcinoma [12,13]. The preoperative PNI and SII are identi ed as effective tumour biomarkers that have important signi cance in the prognosis of many malignant tumours [14,15]. Considering that the recurrence and progression factors of high-risk NMIBC cannot be precisely identi ed before or after surgery, which often results in missed timeframes for appropriate and effective clinical therapy, an urgent need arises for new markers predicting the recurrence risks of high-risk NMIBC patients. We have found no relevant research reports on PNI or SII for predicting the long-term recurrence risks of high-risk NMIBC patients. Therefore, this study was conducted to evaluate PNI and SII in predicting the long-term RFS of high-risk NMIBC patients who received intravesical instillation of BCG after TURBT.

Patient Selection
We retrospectively collected and examined the medical data of 387 patients who met the high-risk NMIBC group criteria and received intravesical instillation of BCG after TURBT at the Department of Urology of Xuanwu Hospital Capital Medical University, National Clinical Research Centre for Geriatric Diseases from January 2004 to December 2014. Patients who met the following criteria were included in our study: (1) complete clinical and pathological records; (2) conformity to the strati cation of high-risk NMIBC; (3) receipt of BCG therapy after TURBT; (4) absence of autoimmune diseases; (5) absence of cancers in other systems; (6) absence of preoperative radiotherapy and neoadjuvant chemotherapy; (7) absence of distant metastasis before surgery and (8) complete follow-up medical data.

Data Collection and De nitions
Clinicopathological data, including gender, age, smoking status, body mass index (BMI), recurrence, tumour number, tumour size, tumour grade, T stage, lymphovascular invasion (LVI), concomitant CIS and preoperative routine laboratory data, were obtained from patients' medical records. The albumin concentration was obtained using the hepatic function test, while the neutrophil, platelet and lymphocyte counts were determined through a routine blood test performed before breakfast within 1-2 weeks before TURBT. PNI was calculated as albumin concentration (g/L) + 5 × total lymphocyte count (10 9 /L) and SII as neutrophil count (10 9 /L) × platelet count (10 9 /L)/lymphocyte count (10 9 /L).

Follow-up
All 387 patients were regularly followed up with physical examination, blood tests, biochemical tests, urine routine tests and cystoscopy every 3 months for the rst 2 years, every 6 months in the following 3 years and annually thereafter. Computed tomography or contrast-enhanced computed tomography was performed every year to assess recurrence in the bladder and upper urinary tract. BCG was performed after TURBT according to the following schedule: induction therapy (weekly BCG instillation for the rst 6 weeks) and maintenance therapy for up to 3 years (3 weekly maintenance instillations at 3, 6, 12, 18, 24 and 36 months).

Statistical Analysis
The SPSS software version 25 Inc., San Diego, CA, USA) were used in statistical analyses. Continuous data are shown as the mean ± standard deviation (SD) or median with interquartile range. Categorical variables are presented as frequencies and percentages and were compared using Pearson's chi-squared test. Receiver operating characteristic (ROC) curve analysis was then applied to identify the optimal cut-off values of PNI and SII with the highest Youden's index for predicting the 5-year overall survival (OS). The end point of this retrospective cohort study was the RFS. The Kaplan-Meier method and log-rank test were used to analyse RFS and compare RFS curves. A Cox proportional hazards regression model was used to analyse various independent predictors of postoperative RFS. Univariate analysis was used to assess various factors related to patients' clinicopathological characteristics. Multivariate analysis was performed for variables with p < 0.1 in univariate analysis. In two-tailed tests, p < 0.05 was considered statistically signi cant.

The Optimal Cut-off Values of PNI, SII and Other Indexes for Estimating Prognosis
The optimal cut-off values of PNI and SII for predicting 5-year OS were determined by ROC curve analysis. As shown in Figure 1, the areas under the curve for 5-year OS were 0.668 (95% CI: 0.599-0.738) and 0.671 (95% CI: 0.604-0.753) for PNI and SII, respectively. The corresponding optimal cut-off values were 50.17 for PNI and 467.76 for SII based on the maximum Youden index. Similarly, the optimal cut-off values for age, BMI, albumin concentration and lymphocyte, neutrophil and platelet counts were 63.5 years, 22.31 kg/m 2 , 42.19 g/L, 1.78×109/L, 4.36 × 10 9 /L and 235 × 10 9 /L, respectively. Based on the chosen optimal cut-off values, the 387 patients were divided into the following groups: low PNI group (<50.17; n = 116) or high PNI group (≥50.17; n = 271) and low SII group (<467.76; n = 242) or high SII group (≥467.76; n = 145).
Associations between PNI and SII as well as RFS in Patients with High-and Highest-risk NMIBC Patients were divided into two groups based on preoperative PNI (<50.17 vs ≥50.17) and SII (<467.76 vs ≥467.76). RFS was signi cantly lower in the low PNI group than in the high PNI group, and a low SII was correlated with better RFS, as shown in Figures 2A and 2B.
We further examined whether PNI and SII were associated with RFS in patients with highest-risk NMIBC.
Notably, RFS was signi cantly lower in the high PNI group (≥50.17) than in the low PNI group, and a high SII (≥467.76) was associated with poor RFS, as shown in Figures 3A and 3B.

Signi cant Predictors of RFS by Cox Univariate and Multivariate Regression Analyses
The results of univariate and multivariate Cox regression analyses of clinicopathological factors related to RFS are presented in Table 3

Discussion
Urothelial carcinomas are identi ed as malignant tumours that arise from the urothelial epithelium and may involve the bladder and upper urinary tract. As the most common malignancy of the urinary tract, BC accounts for 90%-95% of all urothelial carcinomas characterised by multiple, multifocal recurrences throughout the genitourinary tract. In China, the incidence of BC is increasing in men, although mortality is decreasing in both genders [16]. According to distinct clinical recurrence, progression and prognosis, BC is classi ed into non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC). Recently, most patients with BC have been diagnosed with NMIBC [4], and approximately 53% experience progression to life-threatening MIBC [17]. Given the very high recurrence rate, the prediction of recurrence is particularly important in NMIBC. According to the probabilities of recurrence and progression de ned in the European Organisation for Research and Treatment of Cancer Genitourinary Cancer Group risk scoring system, patients were divided into low-, intermediate-and high-risk groups [4]. Given the extremely high recurrence and progression risk in NMIBC patients, the clinical management of high-risk NMIBC is identi ed to be challenging. Early detection of recurrences and accurate identi cation and prediction of the likely progression of tumours are especially vital. BCG followed by maintenance therapy remains the golden standard of care for high-risk patients [18].
Generally, tumour-related pathological factors such as tissue type, grade, stage and LVI have been used to predict recurrence, progression, metastasis and prognosis in a majority of cancer cases. However, this standard evaluation system ignores demographic and other clinical features, such as immune and in ammatory status, which may contribute to its heterogeneity. Malnutrition is a common comorbidity and a consequence of the increase in in ammatory cytokines associated with cancer and metabolic alterations in most patients with malignant tumours [19]. Cancer malnutrition or cachexia, re ected by hypoalbuminemia and hyperlymphocytosis, is associated with increasing morbidity and mortality in patients with cancer [20,21]. In addition, several studies have shown that malnutrition is associated with recurrence and interferes with the patient's response to cancer treatment [22,23]. Albumin concentration and lymphocyte count, which are the common indicators of nutritional status, are important predictors of nutritional risk and postoperative complications. PNI, which is calculated using serum albumin concentration and absolute lymphocyte count, is considered an essential predictive indicator of different types of human cancers [24,25,26]. As a hallmark of cancer that fosters recurrence, progression and metastasis, tumour-associated systemic in ammation has been extensively reviewed in recent years [27]. Systemic in ammatory response (SIR) markers such as lymphocyte-to-monocyte ratio, platelet-tolymphocyte ratio, neutrophil-to-lymphocyte ratio and derived neutrophil-to-lymphocyte ratio have been used to evaluate in ammatory responses and predict tumour prognosis in several types of malignancies, including urological cancer [28,29]. However, these in ammation-based markers integrated only two types of in ammatory cells. SII, which is based on neutrophil, platelet and lymphocyte counts, was rst used in hepatocellular carcinoma and was considered a predictor of several types of solid tumours [12,13,30]. In this study, we investigated the preoperative nutrition-and in ammation-based factors, PNI and SII, in high-risk NMIBC patients who received intravesical instillation of BCG after TURBT. To the best of our knowledge, this is the rst study to investigate the predictive values of PNI and SII for long-term RFS and the relationships of these indices, gender, age, smoking status, BMI, smoking status and pathological parameters with the RFS outcomes of high-risk NMIBC.
The optimal cut-off values of PNI and SII for predicting RFS outcomes vary across studies. Lee et al. [31] reported a PNI of 45 for follicular lymphoma, whereas Car et al. [32] reported a PNI of 46 as the optimum cut-off value for OS in patients with metastatic colorectal cancer. Additionally, Tang et al. [33] reported an SII of 463.56 for BC with total cystectomy. However, Akan et al. [34] determined an SII of 672.75 as cut-off value for high-risk NMIBC with intravesical instillation of BCG after surgery. Our PNI value was slightly higher than those in previous studies, whereas the SII value was similar to that of Tang et al. [33]. The difference could have been in uenced by the different types of tumours, differences in risk group strati cation and clinicopathological stages for BC, heterogeneous patient status and different statistical methods.
Preoperative nutrition-and in ammation-based indicators are reported to be associated with postoperative complication and tumour prognosis [35]. In our study, low PNI and high SII were to be associated with lower BMI, higher recurrence frequency, higher tumour number, larger tumour size, higher tumour grade, more advanced T stage, more common LVI and concomitant CIS, all indicating more aggressive tumour phenotypes. The results of our study were consistent with those of previously published reports [36,37,38]. However, no associations were found between PNI and gender or smoking status. Furthermore, no relationships between SII and gender, age or smoking status were found. These ndings may have been affected by certain in uential factors, such as the limited number of patients, a bias resulting from patient selection, differences in patients in terms of risk strati cation and pathological stages of BC.
Our study identi ed several noteworthy ndings. Signi cant correlations were made between PNI or SII and RFS in all 387 patients in this study. Lower PNI and higher SII were found to be related to worse RFS, whereas elevated PNI and decreased SII were independently associated with better RFS. In addition, univariate analysis demonstrated that PNI, SII, age, BMI, tumour number, tumour size, tumour grade, T stage, LVI and concomitant CIS were signi cantly associated with RFS. However, some results obtained varied from those of previous studies. No relationship between gender or smoking status and RFS was found in univariate analysis. Moreover, PNI, SII, tumour number, tumour size, tumour grade, T stage, LVI and concomitant CIS were signi cantly associated with RFS in multivariate analysis. Incidentally, we found that older age, which is signi cantly associated with BC, was not an independent predictor of RFS, which was inconsistent with previous studies [37,39]. From a mechanistic view, lower BMI may be related to the patients' low nutritional status and decreased immune response. However, in this study, BMI was not an effective predictor of RFS. Contradictory results have been reported as to whether BMI can predict clinical outcomes for BC [40,41]. These results may be due to biases related to the limited patient pool, varying clinicopathological stages of patients and risk strati cation.
According to the EAU Guidelines 2020, sub-stratifying high-risk group patients and identifying those who are at highest risk of disease progression based on prognostic factors are necessary. An analysis of the highest-risk NMIBC subgroup showed that low PNI and elevated SII were signi cantly associated with poorer RFS. Unfortunately, due to the limited number of patients who met the highest-risk NMIBC group criteria, evaluating the relative predictive factors by univariate and multivariate analyses was insu cient.
The mechanism by which low nutritional status and increased SIR in uence the biological features of tumours was not su ciently understood and thus need further study. The association between low PNI or elevated SII and poor RFS in high-risk NMIBC patients could be elucidated at the physiological level. Patients with low PNI and high SII often have hypoalbuminemia, lymphopenia, neutrophilia and thrombocytosis. Malnutrition, re ected by hypoalbuminemia, creates a favourable microenvironment for tumour recurrence [42]. In addition, a poor nutritional status leads to tumour progression through the suppression of adaptive immunity against cancer cells [43]. Although the causes of tumour-induced lymphocytopenia remain to be elucidated in detail, one potential reason is the enhanced lymphocyte apoptosis and impaired lymphocyte homeostasis in cancer [44]. As part of the in ammatory response, neutrophils play a fundamental role in generating high levels of reactive oxygen species, tumour necrosis factor-α and macrophage migration inhibitory factor [45]. It can also cause the secretion of large amounts of arginase, and nitric oxide, which results in the disorder of T cell activation and the production of vascular endothelial growth factor [46,47]. Elevated platelet counts could also stimulate tumour angiogenesis and protect tumour-related cells from cytolysis, which contributes to tumour recurrence and progression. Evidence suggests that circulating platelet-tumour cell aggregates may favour cancer metastasis development [48]. Furthermore, platelets induce tumour cells to form cytoskeletal connections, thereby promoting cancer migration [49].
This present study had several limitations. First, it was based on retrospective data from a single institution representing one region. Second, this study has a retrospective cohort design, which may have promoted selection bias during patient selection and data collection. Third, the assessment of other reported in ammation-and nutrition-based indicators, such as Glasgow Prognostic Score, Naples Prognostic Score, Controlling Nutritional Status score, Carlson Comorbidity Index, mid-upper arm muscle area, Dietary In ammatory Index and C-reactive protein/albumin ratio were not included in this study. Fourth, although laboratory data were all collected before surgery, the timing of blood withdrawal was variable. Finally, the speci cities of PNI and SII may not be high. Therefore, further prospective studies with external validation are needed to con rm the ndings of this study.

Conclusions
PNI and SII can be potential valuable independent predictors of RFS in high-risk NMIBC patients. Preoperative PNI and SII are simple, non-invasive and inexpensive but could evaluate long-term recurrence in high-risk NMIBC.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Authors' contributions
All authors contributed signi cantly to the work reported, including conception, study design, execution, data acquisition, analysis and interpretation. HB, CJ and TO contributed to the conception and study design, provided administrative and technical support and supervised the study. HB and ZS contributed to the drafting and writing of the manuscript. ZF and XN edited the draft along with contributing to the acquisition, analysis and interpretation of data. TO made critical revisions to the manuscript. The authors read and approved the nal version of the manuscript. Moreover, they agree with the submission, and they are accountable for all aspects of this work.

Ethics approval and consent to participate
This retrospective cohort study was approved by the Ethics Committee of Xuanwu Hospital Capital Medical University, National Clinical Research Centre for Geriatric Diseases (2018-086). All participants provided a written informed consent for the retrieval and utilization of their records in our hospital database. In addition, the study was performed in accordance with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Consent for publication
Informed consent was obtained from all participants or their authorized family members.

Competing interests
The authors received grants from the Beijing Municipal People's Government during the implementation of this study and outside the submitted work. All authors declare that they have no other competing interests. Patients, n (%)     Kaplan-Meier analysis of RFS in highest-risk NMIBC patients according to PNI (A) and SII (B).