Preoperative lymphocyte-to-monocyte ratio improves the prediction of survival outcomes after radical cystectomy for bladder cancer


 Purpose

To determine the prognostic significance of preoperative lymphocyte-to-monocyte (LMR) in bladder cancer (BCa) patients undergoing radical cystectomy (RC), and to validate the prognostic benefit provided by LMR compared to the models relying on the clinicopathological factors alone.

Materials and Methods

Retrospective analysis of the 342 BCa patients undergoing RC at our institution from 2004 to 2017 was performed to evaluate the prognostic significance of the LMR. Overall survival (OS) and cancer-specific survival (CSS) was assessed by the Kaplan-Meier method. Cox regression models identified risk factors for survival outcomes. Two nomograms were developed based on the basal models to predict the OS and CSS at 1, 3 and 5 years after RC. The accuracy of the nomograms was assessed with receiver operating characteristics (ROC) curves and concordance-index. Decision curve analyses (DCA) were performed to identify the net benefit by the nomograms.

Results

Excellent long-term survival outcomes of patients were associated with higher LMR level patients. The median survival time for higher LMR level patients was 98.8 months in OS and over 120 months in CSS. In Cox regression multivariate analysis, preoperative LMR, as a continuous variable, is an independent survival outcome predictor ( p <0.001). The addition of LMR to standard model significantly improved its discrimination for prediction of OS by 5.8% and CSS by 5.4% (both p <0.001). Moreover, as shown in DCA, the use of the nomogram including LMR would incur a net benefit over the base models for predicting OS and CSS at 1, 3 and 5 years.

Conclusions

Elevated preoperative LMR among BCa patients undergoing RC is independently associated with significantly better OS and CSS. Moreover, the increase in predictive accuracy after the inclusion of LMR in multiparametric prediction tools is significant. Therefore, LMR may be useful in preoperative patient risk stratification to help patient counseling and clinical decision making.


Introduction
Bladder cancer (BCa) represents the second most common genitourinary tract malignancy worldwide, with an estimated 80,470 new cases and 17,670 cancer-related deaths in 2019 [1]. Radical cystectomy (RC) continues to be the standard treatment for non-metastatic muscle-invasive bladder cancer (MIBC) and high-risk non-muscle-invasive bladder cancer (NMIBC) [2], while the 5-year mortality rates for BCa patients after RC are as high as 38-46% [3][4]. Furthermore, despite multimodal treatment approaches, survival outcomes following RC have not changed over the decades [5]. Thus, improvement in the accurate preoperative risk stratification is suggested currently.
Risk assessment and estimating survival after RC are essential for patient counseling and treatment decision making. The American Joint Committee on Cancer (AJCC) TNM staging system represents the most commonly used prediction tool [6]. Recently, several models for BCa patient assessment were created using clinicopathological features and have been shown to be more accurate at prediction than the TNM staging system [7]. A classical study identified variables such as age, pathologic TNM staging system, pathologic tumor grade, presence of lymphovascular invasion (LVI) at RC, as well as administration of adjuvant therapy (chemotherapy or radiotherapy) to be predictive of survival outcomes after RC, which composed Bladder Cancer Research Consortium (BCRC) models [8].
However, the majority of predictive variables are post-operative pathologic features, and the accuracy of clinical staging in BCa remains poor [9]. Thus, novel prognostic markers are required to facilitate appropriate patient counseling preoperatively.
The systemic inflammatory response is an important prognostic factor in human cancer development and progression [10], and is particularly relevant in BCa, as which is a highly immunogenic malignancy [11]. Several studies have described lymphocyte-to-monocyte ratio (LMR) as an independent prognostic factor for survival outcomes after RC [12][13][14][15][16][17]. However, the question of whether LMR can improve the prognostic accuracy of established predictors of BCa outcome remains unanswered [17], which can be solved by measurement of incremental predictive accuracy [18] and decision curve analysis (DCA) [19]. This study was therefore designed to determine the prognostic significance of preoperative LMR in BCa patients undergoing RC and to validate the prognostic benefit provided by LMR compared to the models relying on the clinicopathological factors alone. were excluded for the following reasons: 12 due to pathologic cell type other than urothelial carcinoma, 4 due to distant metastasis disease at the time of RC, 8 due to postoperative 30-day death and 11 due to conditions before RC that could affect blood cell lines (systemic inflammatory disease: n = 7, malignant lymphoma: n = 2, neoadjuvant chemotherapy: n = 2). This resulted in 342 BCa patients eligible for statistical analysis.
All patients received laparoscopic RC. Extension of pelvic lymph node dissection (PLND) and type of urinary diversion was at the surgeon's discretion. As for urinary diversion, 170 cases have received ureterocutaneostomy, 122 cases undergone ileal conduit, and 50 cases received orthotopic neobladder. A standard PLND was performed in the majority of patients, except for 12 cases where extended PLND was performed.

Study variables
Study variables were extracted from the database and included age, pathologic TNM staging system, pathologic tumor grade, presence of LVI, administration of adjuvant therapy (chemotherapy or radiotherapy), and LMR, according to the BCRC models [8].
RC pathologies were rereviewed by staff pathologists with expertise in genitourinary pathology for tumor stage, tumor grade, presence of LVI. BCa were staged according to the criteria in the seventh edition of the AJCC staging manual. Tumor grade was assigned according to the 2004 World Health Organization grading system. LVI was defined as the presence of tumor cells nest within an endothelium-lined space without underlying muscular walls [20].
Routine full blood counts were collected within 30 days of RC as part of the routine preoperative clinical assessment. Clinical notes were reviewed to rule out any signs or symptoms of infection around the time of blood test. Associations with patient survival outcomes were evaluated with LMR analyzed as a continuous variable, and the cut-off point of LMR was determined as the median value 3.27.

Follow-up
Due to the retrospective nature of the study, there was not a standardized follow-up. Generally, the patient received clinical and radiological follow-up based on the routine protocol used at our institution. This consisted of quarterly follow-up for the first 2 years postoperatively, and then semiannually follow-up for an additional 2 years, with annual follow-up after that. The study outcomes were cancer-specific survival (CSS) and overall survival (OS) from the time of treatment initiation.

Statistical analysis
Descriptive statistics focused on medians and interquartile ranges (IQRs) for continuously coded variables. Frequencies and proportions were reported for categorical variables. Chi-square and independent-sample Mann-Whitney U tests were used to compare the statistical significance of differences in respectively proportions and means. CSS and OS were estimated using the Kaplan-Meier method. Survival was compared between patients with an LMR < 3.27 and ≥ 3.27 with the logrank test.
Cox proportional-hazards model was used to evaluate the association of LMR with survival outcomes, controlling for clinicopathological variables. Specially, the TNM models included pathologic T and N stage, while the BCRC models consisted of age, pathologic T and N stage, tumor grade, presence of LVI, and adjuvant therapy. Based on the BCRC models, we generated two prognostic nomograms additionally included LMR for CSS and OS. To test the discrimination ability of the models, we used Harrell's concordance (C) index [21]. One thousand bootstraps resamples were used for internal validation, and calibration plots were generated to explore nomogram performance [22].
The predictive accuracy of three models at 1, 3 and 5 years was calculated using time-dependent receiver operating characteristic curve (ROC)-derived area under the curve (AUC) estimates [23]. The predictive abilities of the three models were compared using Kang's method [24]. Finally, to identify the net increase in the proportion of cases identified by the new models, we calculated DCA at 1, 3 and 5 years [19].
All statistical analysis was done using R (the R Foundation for Statistical Computing, Vienna, Austria, version 3.6.1). Statistical significance was considered with 2-sided p < 0.05.

Clinicopathological variables of patients
The descriptive characteristics of the 342 BCa patients are shown in Table 1. The median patient age was 68 years (IQR 59-75 years), with a median preoperative LMR level of 3.27 (IQR 2.42-4.25).

Independent prognostic factors for OS
We then investigated the association of preoperative LMR with OS outcomes on multivariate Cox regression analysis controlling for other clinicopathological features (Table 2). We found that LMR (evaluated as a continuous variable) was independently associated with a significantly better OS (Hazards ratio [HR]: 0.555, p < 0.001). That is, for every 1-unit increase in the LMR, there was a 44.5% decreased risk of all-cause death. Furthermore, except for LMR, patient age, pathologic tumor stage and lymph node status were all significantly associated with OS by multivariate analysis (Table 2).

Prognostic nomogram for OS
We then used multivariate regression coefficients to generate prognostic nomogram for OS ( Fig. 2A),

Independent prognostic factors for CSS
We next assessed by multivariate analysis the association of preoperative LMR with CSS outcomes, controlling for other clinicopathological features (Table 3). We found that LMR (evaluated as a continuous variable) remained independently associated with a significantly better CSS (HR: 0.579, p < 0.001). That is, for every 1-unit increase in the LMR, there was a 42.1% decreased risk of cancerspecific death. Meanwhile, the pathologic tumor stage and lymph node status were likewise significantly associated with CSS by multivariate analysis (Table 3).

Prognostic nomogram for CSS
We used multivariate regression coefficients to generate prognostic nomogram for CSS (Fig. 4A), which showed C-index of 0.809, while the TNM-Model was 0.743 (95% CI: 0.704-0.782, p < 0.001) and the BCRC-Model was 0.755 (95% CI: 0.715-0.795, p < 0.001). The calibration plot for the probability of CSS showed an optimal agreement between the prediction by nomogram and actual observation at 1, 3 and 5 years after RC (Fig. 2B).

Discussion
This study retrospectively investigated the relevant data from 342 BCa patients treated with RC to evaluate a relationship between preoperative LMR and survival outcomes. We found that decreased preoperative LMR was associated with advanced pathologic tumor stage and lymph node status at the time of RC, as well as the decreased probability of OS and CSS. In multivariate analysis, preoperative LMR remained as an independent prognostic factor for OS and CSS after controlling for clinicopathological features. Patients with increased LMR had better outcomes.
Lymphocyte and monocyte counts, as part of the preoperative blood test, are cheap and easily obtained in the clinic, reflecting therefore, an optimal candidate for a diagnostic and prognostic biomarker. Furthermore, in our opinion, LMR collected before all the treatment (e.g., cystoscopy), without any influence, was preferred. LMR, at this time point, merely reflected the complex interplay between patient frailty, competing comorbidities, and locally advanced disease [25]. Despite the fact that many investigators have suggested that high pretreatment LMR predicts good clinical outcomes in patients with several malignancies [26,27], there is still no consensus on the ideal cut-off value.
Thus, we evaluated LMR as a continuous variable in the Cox regression analysis. However, to facilitate the clinical use in daily routine, we use the median value of LMR as the cut-off value to develop the nomograms.
The association between increased preoperative LMR and survival outcomes is complex and yet to be elucidated. Bladder cancer is frequently associated with chronic or recurrent urinary tract inflammation and systemic inflammation [11,28]. In turn, through the oncogenic change, the tumor microenvironment activates the adaptive immune response, which induces cancer-promoting inflammation to promote proliferation, progression and metastasis [29]. Monocytes play a central role in the production of proinflammatory cytokines, such as monocyte chemoattractant protein-1, that can contribute to cancer initiation and promotion [30]. Additionally, high tumor-associated macrophage (TAM) counts have been found correlated with poor survival and poor response to treatment [31]. However, lymphocytes are essential in antitumor reactions through the induction of tumor cell apoptosis [32] and by mediating antibody-dependent cell-mediated cytotoxicity [33]. In studies for BCa, the number and function of lymphocyte were both found to be lower in invasive disease than in controls and superficial carcinoma [31,34]. Therefore, a low LMR reflects both a decreased, lymphocyte-mediated, antitumor immune response and a heightened inflammation and immune dysfunction mediated by monocyte [35].
Up to now, five previous studies address the role of pretreatment LMR for survival outcomes in BCa patients undergoing RC. Temraz et al. found that based on cut-off value of 2.81, patients with lower level LMR had shorter OS (2.7 vs. 6.0 years, p = 0.020) in a 68 BCa patients cohort. There was no multivariable analysis in this study [12]. Zhang [16]. In these two studies, some relevant confounders (e.g., LVI, adjuvant chemotherapy/ radiotherapy and/or concomitant systemic inflammatory disease) were not evaluated. In a series by Yoshida et al., preoperative LMR showed higher predictive accuracy for OS than other inflammatory markers did (p = 0.033), and interestingly, perioperative LMR changes is significantly associated with OS (HR: 5.70, p < 0.001) and CSS (HR: 4.53, p < 0.001) [14,15]. However, there were no studies to evaluate whether LMR can improve the prognostic accuracy of established predictors of BCa outcome. The largest multicenter study to date, conducted by D'Andrea et al., confirmed that LMR independently predicted CSS and OS, but the discrimination of the new model increased by adding LMR but was not significant [17]. Therefore, we utilized the C-index comparison, time-dependent ROC and DCA to validate the prognostic benefit provided by LMR.
Several of our findings are noteworthy. Firstly, we reported excellent long-term survival outcomes of patients with higher LMR level patients. The median survival time for higher LMR level patients was 98.8 months in OS and over 120 months in CSS. And in Cox multivariate analysis, preoperative LMR, as a continuous variable, is an independent survival outcome predictor, avoiding the trouble to determine the ideal cut-off value [35]. Similar to D'Andrea et al. [17], we added the LMR in the standard models to validate the improvement of prognostic accuracy. The inclusion of LMR in our nomogram resulted in significant improvements of OS and CSS prediction at all determined time points (p < 0.01). In addition, as shown in DCA, the use of the nomogram, including LMR, would incur a net benefit over the base models for predicting OS and CSS. Therefore, adding LMR in the basal models is a promising tool to predict OS and CSS in BCa patients after RC and guide adjuvant treatment decisions.
There are many limitations in our study, mainly related to its retrospective and single-institutional nature, which could result in selection bias and differences from other geographical regions and institutions. Therefore, in the future, a large prospective cohort and multicenter study are needed to investigate the role of LMR in survival outcomes of BCa patients after RC. In this cohort, although the calibration plot for the probability of survival outcomes showed an optimal agreement at all determined time points, we didn't perform the external validation. The predictive accuracy of our nomogram in other cohorts is needed to study in the future. Finally, whether the role of LMR in survival outcomes for BCa can be applied to patients with non-urothelial carcinoma of BCa remains to be determined.

Conclusions
Decreased preoperative LMR is associated with advanced tumor stage among BCa patients undergoing RC. After controlling for clinicopathological factors, increased LMR remains associated with significantly better OS and CSS. Moreover, the increase in predictive accuracy after the inclusion of LMR in multiparametric prediction tools is significant. In addition, LMR is inexpensive and readily available as part of routine preoperative testing. Therefore, LMR may be useful in preoperative patient risk stratification to help patient counseling and clinical decision making.

Ethics approval and consent to participate
This study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee (No. M2018183).

Consent for publication
All authors are consent for publication.

Availability of data and material
The anonymized data used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that there are no conflicts of interest.

Authors' contributions
JL and YH planned and designed the study. HB and ZJQ collected the data. HB and YY analyzed and interpreted the patient data. GLW and LLM commented on drafts of the paper. HB was a major contributor in writing the manuscript. All authors read and approved the final manuscript.