Prognostic significance of platelet-to-lymphocyte ratio in urothelial carcinoma patients: a meta-analysis

Background The prognostic value of pre-treatment platelet-to-lymphocyte ratio (PLR) in patients with urothelial carcinoma (UC) remains controversial. Therefore, this meta-analysis aimed to identify the prognostic impact of PLR on UC. Methods The PubMed, Embase, Web of Science, and Cochrane Library databases were systematically searched. Hazard ratios (HRs) with 95% confidence intervals (CIs) were used to summarize the correlations between PLR and overall survival (OS), progression-free survival (PFS), disease-free survival (DFS), and cancer-specific survival (CSS). Odds ratios (ORs) with 95% CIs were used to measure the association between PLR and tumor clinicopathological factors. Results The meta-analysis included 15 studies published from 2015 to 2019 with a total of 5354 patients. Overall, a high PLR was correlated to poorer PFS (HR = 1.81, 95% CI 1.28–2.56, p = 0.001) and DFS (HR = 1.09, 95% CI 1.31–2.16, p < 0.001) but not poor OS (HR = 1.23, 95% CI 0.95–1.59, p = 0.124) or CSS (HR = 1.000, 95% CI 0.998–1.002, p = 0.919) in UC. In addition, an elevated PLR was correlated with patient age > 65 years (OR = 1.72, 95% CI 1.25–2.38, p = 0.001) and hypertension (OR = 1.48, 95% CI 1.01–2.18, p = 0.046). However, no significant association was observed between PLR and sex (OR = 0.79, 95% CI 0.56–1.14, p = 0.206) or diabetes (OR = 1.29, 95% CI 0.77–2.15, p = 0.333). Conclusions Our results demonstrated a significant correlation between elevated PLR and poor prognosis in UC. The prognostic role of PLR may help guide the management and prognostication of UC patients.


Background
Urothelial carcinomas (UCs) are the fourth most prevalent tumors [1]. Upper tract urothelial carcinomas (UTUC) are tumors derived from the urothelium along the urinary tract [2]. UTUCs are rare, accounting for only 5-10% of all UCs [3,4], while bladder cancer (BC) accounts for 90% of all UCs. Sixty percent of UTUCs are diagnosed at the invasive stage, and peak incidence is observed in patients aged 70-90 years [5]. Regardless of the tumor location in the upper urinary tract, radical nephroureterectomy (RNU) with bladder cuff resection is considered the standard treatment for most UTUC patients [5]. Although an adequate surgical treatment, the 5-year cancer-specific mortality remains high, ranging from 20% to 30% [6]. Seventy-five percent of BC patients are diagnosed with non-muscle-invasive bladder cancer (NMIBC), which has a high risk of recurrence. Various prognostic factors such as p53 protein, nuclear factor-kB, and osteopontin have been investigated in UC, but the prognostic efficiency remains unsatisfactory [2]. Therefore, it is important to identify reliable and effective prognostic biomarkers to aid UC prognostication and treatment.

Search strategy
The PubMed, Embase, Web of Science, and Cochrane Library electronic databases were systematically searched to identify relevant studies. The following terms were used in the literature search: "platelet lymphocyte ratio", "PLR", "platelet to lymphocyte ratio", "urothelial carcinoma", "urothelial cancer", "bladder cancer", "bladder tumor", "upper urinary tract cancer", "upper tract urothelial carcinoma", and "UTUC". The last search was updated on September 16, 2019. The reference lists of relevant articles were also examined for additional potential inclusions. This meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [33]. Ethical approval and informed consent were waived because all studies included in this meta-analysis were previously published.

Inclusion criteria and exclusion criteria
The inclusion criteria were (1) pathologically or histologically confirmed diagnosis of UC; (2) studies evaluating the correlation between PLR and overall survival (OS), progression-free survival (PFS), disease-free survival (DFS), and/or cancer-specific survival (CSS); (3) defined PLR cut-off value; (4) preoperative blood cell counts; (5) hazard ratio (HR) and 95% confidence interval (CI) provided or able to be calculated from the available information; and (6) studies published as full-text in English. The exclusion criteria were (1) case reports, reviews, meeting abstracts, or letters; (2) studies with overlapping or duplicate data; and (3) studies without sufficient or usable data.

Data extraction
Two investigators (Y.B and Y.W) independently reviewed all candidate studies, and any disagreements were resolved by discussion with a third investigator (X.L). The following information was extracted from each eligible study: first author, year of publication, country, study period, number of patients, sex, age, treatment, cut-off value, ethnicity, survival analysis, and HRs of PLR for OS, PFS, DFS, and CSS with their 95% CIs.

Quality assessment
Study quality was evaluated using the Newcastle-Ottawa scale (NOS) [34]. The NOS consists of three parts: selection, outcome, and comparability. The scores range from 0 to 9, and studies with NOS scores ≥ 6 are considered to be high-quality studies.

Statistical analysis
This meta-analysis was conducted using Stata 12.0 (Stata Corp, College Station, TX, USA). Heterogeneity among studies was estimated using Cochran's Q test [35] and Higgins' I-squared statistics [36]. A random-effects model was used for studies with significant heterogeneity (I 2 > 50% or Ph < 0.10). Otherwise, a fixed-effects model was used. A pooled HR > 1 with 95% CI not overlapping 1 (p < 0.05) indicated worse OS, PFS, DFS, and CSS for a high PLR in UC. Subgroup analyses were performed to investigate the factors influencing the prognostic function of PLR. The correlation between PLR and clinicopathological factors were measured by pooled odds ratios (ORs) and 95% CIs. Publication bias was determined using Begg's funnel plots and Egger's linear regression tests. p-values < 0.05 were considered statistically significant.

Discussion
The present study comprehensively searched relevant databases and collected data from 15 cohort studies containing 5354 patients. The pooled results suggested that an elevated PLR predicted an inferior PFS and unfavorable DFS in UC patients. The subgroup analysis showed that the prognostic value of PLR for DFS and PFS was not influenced by tumor type. Moreover, the pooled data also indicated that a high PLR was associated with patient age > 65 years and hypertension. To our knowledge, ours is the first meta-analysis to investigate the prognostic ability of PLR in patients with UC. As PLR is non-invasive and easily accessible, it has the potential to guide clinical decision-making.
Recent studies using meta-analytic methods also focused on the association between PLR and prognosis for various types of cancer [17]. Lin [38]. Another work showed the prognostic value of PLR for worse OS (HR = 1.38, 95% CI 1.19-1.62, p < 0.001) and poor RFS or PFS (HR = 1.55, 95% CI 1.27-1.88, p < 0.001) in patients with cholangiocarcinoma [39]. The findings of previous studies were in line with those of the current   [40]. Platelets also facilitate the proliferation of ovarian cancer cells in a transforming growth factor-β1 (TGF-β1)-dependent manner [41]. Moreover, platelets can directly contact tumor cells and secret a series of cytokines including platelet-derived growth factor (PDGF), TGF-β, and prostaglandin (PG) E2, which can enhance the epithelial-mesenchymal transition (EMT) of tumor cells [42,43]. In contrast, lymphocytes play important roles in anti-tumor immune responses. Intraepithelial CD3 + and CD8 + tumor-infiltrating T lymphocytes (TILs) were strongly associated with improved PFS and DFS in ovarian cancer patients [44]. Lymphocytes and interferon (IFN) gamma can collaborate to select to tumor cells to reduce immune surveillance [45]. CD8 + TILs have been associated with good prognosis in various cancers [46]. Therefore, evaluation of PLR is useful and convenient to predict clinical outcomes in patients with UC.
The present study had several limitations. First, the included studies were all retrospective, which may have caused a selection bias in the meta-analysis. Second, only four studies provided data on the association between PLR and clinical features. The sample size was too small. Third, we extracted pooled HRs and 95 CIs from eligible studies but not individual patient information. Fourth, it is hard to normalize PLR because blood counts may vary at different sites, which may cause variability in the index values. Therefore, additional large-scale prospective studies are warranted to confirm our findings.