The Prognostic Value of Platelet-to-Lymphocyte Ratio in Urological Cancers: A Meta-Analysis

The relationship of platelet-to-lymphocyte ratio (PLR) and survival in urological cancers remained inconsistent in previous studies. Therefore, we performed a meta-analysis to assess the prognostic significance of PLR in patients with urological cancers. A literature search was performed in the PubMed, Embase, and Web of Science up to July, 2017 and study quality was obtained using the Newcastle-Ottawa Scale. To estimate the association of PLR and overall survival (OS) and other survival outcomes in urological cancers, we used pooled hazard ratios (HRs). Subgroup analyses were conducted on different ethnics, sample sizes and cut-off values. 20 high quality studies involving 7562 patients with urological cancers were included in this meta-analysis. High pretreatment PLR was significantly associated with poor OS in patients with urological cancers (pooled HR = 1.58). Elevated PLR was also correlated with other survival outcomes. However, we found that PLR was significantly relevant to the OS of patients with different types of urological cancers except bladder cancer (BCa, HR = 1.16, 95%CI: 0.96–1.41). In conclusion, elevated PLR was negatively related to the OS of patients with urological cancers, except in BCa. However, more large scale prospective studies with high quality are required in the future.

Urological cancer contains major malignancies with high morbidity and mortality worldwide 1 . Prostate cancer (PCa) is the most prevalent cancer among western men, accounting for 19% new cancer cases and third leading cause of cancer-related death among American men in 2017 2 . Bladder cancer (BCa) represents the ninth most common tumor with approximately 2.7 million people suffering from it on a global scale 3,4 . While Renal cell carcinoma (RCC) takes up for a proportion of 2-3% in overall adult cancers 3 . Upper urinary tract urothelial carcinoma (UTUC) is relatively rare but aggressive. Despite great progress of treatment, the prognosis and clinical outcome of urological cancers remains unsatisfactory because of local recurrence or distal metastasis. The survival rate and survival time may be increased by risk stratification and optimal treatment at early stage. Hence, it is crucial to seek useful personalized biomarkers to estimate patient prognosis.
Accumulating evidence demonstrates that inflammation plays a critical role in tumor development and progression 5,6 . In patients with cancer, systemic inflammation is likely to affect the tumor micro-environment and promote tumor growth, which means poor outcome 7 . It is well acknowledged that systemic inflammation response can be characterized by the changes of peripheral blood cell amounts. Numerous studies have revealed that blood-based biomarkers show great potential in urological cancer prognosis, such as neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP), lymphocyte-to-monocyte ratio (LMR) and platelet-to-lymphocyte ratio (PLR) [8][9][10] . Platelet and lymphocyte counts are easily acquired during routinely blood tests in clinical laboratories worldwide. Thus, PLR is a cheap and objective parameter to potentially help doctors assessing patient prognosis. A couple of meta-analyses have testified the prognostic impact of PLR in malignancies like breast cancer and lung cancer 11,12 .
During the past few years, several studies explored the prognostic significance of PLR in patients with urological cancers. However, some studies have drawn contradictory conclusions. To our knowledge, no published meta-analysis have investigated PLR and urological cancer prognosis. The aim of the current study was to quantitatively and comprehensively summarize the available evidence on the prognostic value of elevated PLR and Quality assessment. While there was a small variation in the methodological quality of included studies, all of the included studies were judged relatively high quality according to the NOS assessment tool, with scores from 6 to 8. Fig. 2a Figure 1).

PLR and survival in patients with urological cancers. As displayed in
Then we performed further analyses based on the type of urological cancers. 6 cohorts investigated PLR and OS in patients with RCC (Fig. 3a). The pooled HR was 1.69 (95%CI: 1.18-2.43, P = 0.003). While six studies provided HRs of PLR and OS in patients with PCa (Fig. 3b). Their pooled HR was 1.77 (95%CI: 1.43-2.20, P < 0.001) and no heterogeneity existed (I 2 = 0%, P = 0.631). However, other 3 cohorts reported PLR and OS in patients with BCa (Fig. 3c), whose pooled HR was 1.16 (95%CI: 0.96-1.41, P = 0.124), with low heterogeneity existed (I 2 = 30.4%, P = 0.238). Only 2 cohorts explored PLR and OS in patients with UTUC (Fig. 3d), whose pooled HR was 1.69 (95%CI: 1.16-2.48, P = 0.007). No heterogeneity was found in UTUC group (I 2 = 0%, P = 0.794).  Subgroup analysis and Meta-regression. There was slight heterogeneity (Fig. 2a, I 2 = 53.9%), so we conducted subgroup analysis to seek more information ( Table 2). In epidemiological studies, ethnicity difference was usually recognized as a critical source of heterogeneity. Notably, the I 2 values of Asian and non-Asian were 62.7% and 0.0% respectively. From another aspect, there were also no heterogeneity when cut-off value of PLR > 150 (I 2 = 0.0%). When we conducted subgroup on sample size, heterogeneity existed when sample size > 254 (I 2 = 52.2%) or ≤254 (I 2 = 56.3%). The pooled HR values and their 95%CI in each subgroup analysis were demonstrated in Table 2 with corresponding forest plots shown in Fig. 4. To sum up, the pooled HRs indicated that high PLR was significantly associated with poor OS in each subgroup.
To seek possible sources of heterogeneity, we also conducted meta-regression. On account of insufficient data on other parameters such as tumor grade or stage, we choose cancer types, PLR cut-off, sample size, follow up time as covariates to estimate between-study variance. The result of this model (P = 0.86) showed no potential association between these covariates (cancer types: P = 0.367, PLR cut-off: P = 0.626, sample size: P = 0.254, follow up time: P = 0.563) and pooled HR. Sensitivity analysis. In order to gauge the stability of the results, we conducted sensitivity analysis by removing one study in sequence to see if a single study could have significant impact on the pooled HRs for survival. The results were not significantly altered by removing anyone of the included studies ( Fig. 5a: OS, Fig. 5b: CSS, Fig. 5c: PFS, Fig. 5d: DFS).  Notably, this correlation of high PLR and poor OS was also applicable to RCC, PCa and UTUC, but not in BCa (HR = 1.16, 95%CI: 0.96-1.41). Only 3 studies available for meta-analysis on PLR and OS in patients with BCa, so it should be interpreted cautiously.
Subgroup analysis divided by ethnicity, sample sizes and PLR cut-off value did not significantly change the main results. Taken all the above results into consideration, we believed that PLR could serve as an reliable marker in urological prognostication. PLR, an parameter which is reproducible, convenient and low cost, can be available from blood routine test in daily clinical practice. Additionally, PLR changes followed by tumor changes during anticancer process are also crucial to know for its application as an indicator of treatment efficacy.
Our findings about PLR are in accordance with previous reported other cancers such as breast cancer, lung cancer and colorectal cancer [38][39][40] . Systemic inflammatory plays a crucial role in tumor progression at almost every  single step including initiation, progression, and metastasis 41 . But the underlying mechanism that PLR influence the survival of patients with urological cancers remains largely unknown. Several hypotheses have been put forward to explain this phenomenon. Platelet-derived cytokines, such as platelet derived growth factor (PDGF), vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) have been found to regulate and promote tumor angiogenesis, then further accelerate tumor aggressiveness 42,43 . Platelet can also release microparticles that help to guard tumor cells from the elimination of natural killer 44 . On the contrary, the reduction of lymphocytes was correlated with poor prognosis, as indicated in a previous study 45 . Lymphocytes, like T-lymphocytes who are able to secret interleukin-4 and -5 in tumor microenvironment, have an anti-tumor activity by inducing cytotoxic cell death and inhibiting tumor proliferation 46,47 . In conclusion, the platelet to lymphocyte ratio could be regarded as a positive correlative marker with worse cancer prognosis theoretically.
Although this is the first meta-analysis concerning PLR and urological cancer prognostication, several limitations are still needed to be addressed. Firstly, though we didn't restricted language, all the 20 studies included were published in English. We found no relevant studies published in Chinese qualified to the inclusion criteria, but we also failed to identify articles in other languages, which might lead to little language bias. Secondly, although sensitivity analysis supported the stability of our results and no publication bias found, the number of studies was relatively small in different cancer types (6 for RCC, 6 for PCa, 3 for BCa, 2 for UTUC) and in some subgroups (6 for non-Asians). Thus we wish to emphasize that the results should be cautiously interpreted. Thirdly, heterogeneity among studies was found, which was probably because of the relatively small sample sizes and the retrospective property of study design. Fourthly, we lack the PLR data of patients with comprehensive different urological cancer stages (localized, advanced or metastasis) at present. Large scale statistics about PLR response during different treatment strategies such as radiotherapy, chemotherapy and immunotherapy are also insufficient. So further large prospective clinical trials are urgently required to verify the prognostic value of PLR in patients with urological cancers in the future.
In summary, this meta-analysis suggested that elevated PLR was negatively related to survival of patients with urological cancers, except in BCa. However, further high quality studies with large sample size should be conducted to validate this paper's results.

Methods
Search strategy. This meta-analysis was conducted under the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 48

Study Selection
Inclusion and exclusion criteria. Studies meeting the following criteria were considered eligible:1. clinical cohort evaluated the prognostic accuracy of PLR in urological cancers; 2. studies compared PLR with other prognosis models and reported survival outcomes like OS and PFS; 3. reported original hazard ratio (HR) with 95% confidence interval (95%CI) or HR could be extracted from sufficient information; 4. articles with the most complete information and the largest cohort if there were multiple studies by the same author or institute.
The exclusion criteria were: 1. repeated publications; 2. studies reporting on less than 20 patients; 3. experimental laboratory articles, animal studies, letters or review articles.
Assessment of study quality. Two investigators(DY.L. and XY.H.) independently reviewed all relevant articles and judged the methodology quality of potential studies using Newcastle-Ottawa Quality Assessment Scale (NOS) assessment tool, including selection, comparability and outcome 49 . A study was considered high quality if the NOS score ≥7. When disagreements occurred, the two reviewers reached consensus by involving a third author(HX.D.).

Data extraction.
We extracted the following variables from each study: first author's name; publication year; study design; country or region of the study; type of urological cancer; cut-off value of PLR; sample size; age, sex, out-come assessment and risk estimates, follow up time and HRs with 95%CI. If the HRs of both univariate and multivariate analysis for the same comparison were available, we only used the latter for analysis. We also used univariate or multivariate HRs separately to test methodological sensitivity. If the HR and 95%CI were not displayed directly, they were estimated from Kaplan-Meier curves 50 . If necessary, the corresponding author was contacted for further information.
Statistical analysis. HRs with 95%CI were pooled using a meta-analysis to access the strength of PLR to survival endpoints such as OS, CSS, PFS and DFS. All the PLR values from 20 included studies were categorical variables in survival analysis. The Cochrane Q test was used to determine the heterogeneity among studies. A P value < 0.10 indicated heterogeneity. Inconsistency (I 2 ) was also calculated to evaluate heterogeneity. AnI 2 value > 50% was considered significant heterogeneity. The fixed-effect model (inverse variance method) was used to calculate pooled results when no heterogeneity existed among included studies, otherwise, a random-effect model (DerSimonian and Laird method) was used with the weights inversely proportional to the variance of log hazard ratio of each trial 51,52 . To find reasons of heterogeneity among studies, we conducted subgroup analysis in ethnic difference, different cut-off value and sample size respectively. When the log-rank statistical value was maximum on receiver operating characteristic curve, the cut-off value of PLR was decided. The median value Scientific RepoRts | 7: 15387 | DOI:10.1038/s41598-017-15673-2 of sample size was chosen to divide subgroup. Meta-regression was also performed by using cancer type, PLR cut-off, sample size and follow up time as covariates. To test the reliability of the main outcomes in our analysis, sensitivity analysis was performed by removing one single study in turn. Egger's and Begg's tests with funnel plots were used to test publication bias. P value > 0.05 indicated no potential publication bias. Kaplan-Meier curves were read by Engauge Digitizer version 9.8 (http://markummitchell.github.io/engauge-digitizer/). We used Stata 12.0 software (Stata Corporation, College Station, TX, USA) to conduct all the statistical analyses. A two-sided P value less than 0.05 was considered statistically significant. Data availability. In our current meta-analysis, all original data analyzed were derived from published articles. All data generated during this study are included in the present article.