Clinicopathological and prognostic significance of platelet to lymphocyte ratio in patients with gastric cancer

The present study was aim to investigate the prognostic role of platelet to lymphocyte ratio (PLR) for patients with gastric cancer (GC) using meta-analysis. A total of 13 studies (14 cohorts) with 6,280 subjects were included. By pooling hazard ratios (HRs) and 95% confidence intervals (CIs) and odds ratios (ORs) and 95% CIs from each study, we found that elevated PLR was significantly associated with poorer overall survival (OS) (HR: 1.3, 95% CI: 1.1–1.52, p = 0.001; I2 = 68.5%, Ph < 0.001) but not with poor disease-free survival (DFS) (HR: 1.6, 95% CI: 0.88–2.9, p = 0.122; I2 = 87.8%, Ph < 0.001). Subgroup analysis showed that a high PLR significantly predicted poor OS in Caucasian populations, patients receiving chemotherapy and patients at advanced stage. In addition, the cut-off value of PLR > 160 showed adequately prognostic value. Furthermore, elevated PLR was associated with lymph node metastasis and CEA levels in GC. Our meta-analysis showed that elevated PLR could be a significant prognostic biomarker for poor OS in patients with GC.


INtrODUctION
During the past several decades, although gastric cancer (GC) incidence rates have been declining in most Western countries [1,2], GC still ranks fourth in incidence among all cancers and is the third leading cause of cancerrelated deaths worldwide [3]. GC at early stages is often asymptomatic and most patients are diagnosed when the disease has already advanced. A variety of prognostic factors including tumor histological type, genetic polymorphisms and tumor stage have been reported for GC [4]. Despite of these, the prognosis of GC is poor, with the 5-year survival rate being 20% [5], therefore, more novel and easily available prognostic biomarkers are needed.
The inflammatory responses to cancer have been recognized more than one century ago [6]. In recent years, clear evidence showed that inflammation plays pivotal roles in carcinogenesis and tumor metastasis [7,8]. Laboratory parameters which reflect the status of systemic inflammation, have been investigated as prognostic biomarkers in various cancers. These inflammatory markers include modified Glasgow Prognostic Score (mGPS), neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR) [9,10]. PLR has been suggested as an independent prognostic factor in several solid tumors including colorectal cancer [11], non-small cell lung cancer [12], pancreatic cancer [13] and gastric cancer [14]. Previous evidence suggested that platelets played multiple roles in inflammatory processes, for example, platelets could facilitate neutrophils adhesion to endothelium through releasing chemokines and cytokines [15]. Platelets may also promote tumor progression through facilitation of neoangiogenesis, production of adhesion molecules and increase of early metastatic niches [16,17]. In contrast, lymphocytes are known to hinder tumor cell proliferation and metastasis [18]and mediate antibody-dependent cellmediated cytotoxicity (ADCC) effects [19]. Therefore, there is a biological rationale for using PLR, to measure the systemic host response in gastric cancer to predict Research Paper clinical outcomes. Accumulated studies have reported the association between PLR and survival conditions in GC, however, the results were controversial. For example, Lee et al. [20], found that elevated PLR predicted poor overall survival (OS) in GC patients treated with chemotherapy. However, Jiang et al. [21] did not detect the prognostic value of PLR for GC patients receiving radical resection. As meta-analysis is an effectively analytic approach to pool these controversial findings, we thus conducted a meta-analysis to reveal the prognostic significance of PLR for overall survival (OS) and disease-free survival (DFS) and the associations between PLR and clinicopathological features in patients with GC.

PLr and tumor clinicopathological parameters
To further exploit the impact of PLR on the clinical features in GC, we identified ten clinical factors in GC and extracted the patient amounts in PLR high and PLR low groups regarding each factor. The ten factors could be classified as three categories: first, general information about patients, including gender and age; second, parameters reflecting invasiveness of tumor, including TNM stage, tumor differentiation, depth of invasion, tumor size, lymph node metastasis and distant metastasis; third, specific indexes for GC, including Lauren type, and CEA level. As shown in Table 3, the results demonstrated that high PLR was positively correlated with lymph node metastasis (n = 4, HR = 1.56, 95% CI: 1.33-1.83, p < 0.001) and CEA (n = 2, HR = 1.57, 95% CI: 1.11-2.24, p = 0 .012). Whereas elevated PLR was not shown to be associated with age, gender, TNM stage, tumor differentiation, depth of invasion, tumor size, Lauren type or distant metastasis.

sensitivity analysis
Every single study was moved out and thereafter the pooled data was recalculated to test the stability of the results. The results of the sensitivity analysis were shown in Figure 4. The corresponding pooled HRs did not substantially change, which confirmed the robustness of our results.

Publication bias
Publication bias was assessed by using Begg's test. The results indicated that there was no significant publication bias in OS and DFS (p = 0.08 for OS and www.impactjournals.com/oncotarget p = 0.296 for DFS, respectively; Figure 5). In addition, there was also no significant publication bias for the analyses involving the relationship between PLR and clinical features in GC (Table 3).

DIscUssION
The current study was designed to investigate the prognostic value of elevated PLR for OS and DFS in patients with GC by meta-analysis. Pooled results from 14 cohorts with 6,280 subjects demonstrated that elevated PLR was associated with poor OS. However, PLR had not prognostic role for DFS. Moreover, elevated PLR was correlated with lymph node metastasis and CEA levels in patients with GC.
To the best of our knowledge, this meta-analysis was the first to identify the prognostic role of PLR in GC.
The strong linkage between immune responses and cancer progression was increasingly investigated in the past decade. Cancer related inflammation could attenuate antitumor activity of the host through recruiting immunosuppressive cells such as regulatory T cells (Treg) [32] and myeloid-derived suppressor cells (MDSC) [33,34]. A variety of chemokines and cytokines secreted in the tumor microenvironment could also facilitate tumor progression and metastasis [35]. The mechanisms between carcinogenesis, platelets assembly and lymphocytopenia remained unclear. Notably, recent studies demonstrated that platelet-derived signals were necessary for the recruitment of granulocytes, which could further contribute to the formation of early metastatic niches for     tumor cells [17]. Moreover, platelets could also promote the communication between primary tumor cells and bone remodeling alterations prior to metastasis [36]. These could be the potential reasons for the association between elevated PLR and lymph node metastasis in the present study. In addition, lymphocytopenia and the suppression of lymphocytes activity induced by the systemic inflammatory response impairs the innate cellular immunity [37]. Thus, the combination of blood parameters such as PLR could predict prognosis more accurately. What's more, the value of PLR could be acquired from the routine laboratory tests, which provides clinical implications at a low cost.
Recently, several meta-analyses [38,39] investigated the prognostic value of PLR in various solid tumors. In Templeton et al.'s work including twenty studies, they found that elevated PLR was associated with poor OS in colorectal cancer, hepatocellular carcinoma, ovarian cancer and pancreatic carcinoma [38]. Whereas, only three studies on gastric cancer were included in their meta-analysis and the limited data may restrain the credibility of the results. Another work, reported by Zhou and colleagues [39], showed that high PLR had prognostic efficiency for OS in cancer by pooling data from 26 studies with 13,964 patients. Compared with the previous studies [38,39], the current study including 14 cohorts with 6,280 GC patients was more comprehensive with sufficient data. Moreover, we not only investigated the prognostic value of PLR for OS, but also reported the results for DFS in GC. The associations between PLR and clinicopathological characteristics were also explored. Therefore, our metaanalysis had more specificity for GC population and the adequate data made the results convincing. In the present study, we selected PLR as the study object, which was frequently compared with another blood-derived index, NLR, in included studies [14,20,21,23,[25][26][27][28]31]. NLR was also widely investigated as a prognostic indicator for GC, for it showed significant association with patients survival [20,27]. As the present study was designed to investigate the relationship between PLR and GC, the studies regarding NLR on GC were partially included and a comprehensive conclusion could not be drew according to these studies. Therefore, the prognostic efficiency between PLR and NLR in GC could not be directly compared in this meta-analysis. Interestingly, we noted that several meta-analyses [40][41][42][43] had exploited the prognostic role of NLR in GC. The pooled HRs and 95% CIs of NLR on OS in GC ranged from HR = 1.65 (95% CI: 1.47-1.83)  [42] to HR = 2.16 (95% CI: 1.86-2.51) [41], which were higher than the HR and 95% CI of PLR for OS (HR = 1.3, 95% CI: 1.1-1.52) in the present study. Previous metaanalyses [40][41][42][43] also indicated that NLR was associated with poor DFS in GC, whereas we did not find such correlation of PLR and DFS. These results suggested that NLR might have more powerful prognostic efficiency for poor OS in GC than PLR and could predict shorter DFS in GC when PLR could not. This phenomenon may be due to neutrophils, as the immune cells of the innate system, were more intensely and comprehensively involved in immune responses, compared with platelets, therefore, NLR could be more intensively influenced in GC and be more sensitive than PLR. This possible explanation should be verified in further studies.
There were some limitations need to be addressed in this meta-analysis. First, only three studies investigating the role of PLR for DFS prognostication were analyzed. The sample size was relatively small and subgroup analysis was not performed due to limited data. Second, the cut-off values of PLR were various in the studies, which calls for uniformly used value in further investigations.
In summary, we found that elevated PLR was a prognostic factor for poor OS, but not for DFS in GC patients. Furthermore, a high PLR had more significantly prognostic significance for OS in Caucasians, patients receiving chemotherapy and at advanced stage. Elevated PLR was also associated with lymph node metastasis in GC. Due to the limitations in this study, more large scale studies using uniform cut-off value of PLR are needed to validate our results.

Literature search
A thorough literature searching was conducted in the databases of Pubmed, Embase and Web of Science. The last search was updated to March 2016. The search strategy was as follows: (PLR or platelet to lymphocyte ratio or platelet-lymphocyte ratio) and (gastric cancer or GC or gastric carcinoma or gastric neoplasm or stomach neoplasms). Only studies in English were included. The reference lists were manually retrieved for additional studies. Ethical approval was not required for this study because this was a meta-analysis.

selection criteria
Studies included in the meta-analysis need to meet the following criteria: (1) the value of PLR was acquired from a peripheral venous blood test before treatment, PLR was calculated as the ratio of the platelets to lymphocytes; (2) the diagnosis of GC was pathologically confirmed; (3) HRs and 95% CIs for PLR in OS and (or) DFS were reported, or could be calculated from raw data in the articles; (4) the cut-off value of PLR was reported. The exclusion criteria were: (1) letters, conference abstracts or review articles; (2) animal studies; (3) insufficient data to estimate HRs and 95% CIs; (4) did not present the cut-off value for elevated PLR; (5) not published in English.

Data extraction
Based on a consensus on all items, two investigators (XB,G and XS,G) independently extracted the following information from each study: surname of the first author, year of publication, country, duration of the studies, sample size, treatment methods, stages of the disease, HRs with 95% CIs and the cut-off value of elevated PLR. Disagreement was resolved by consulting the third investigator (M,X).

Quality assessment
The quality of included studies was assessed according to the Newcastle-Ottawa Scale (NOS)(http:// www.ohri.ca/programs/clinical_epidemiology/oxford.asp) by two reviewers (W,G and M,C). The maximum score is 9 points and studies with an NOS score ≥ 6 were considered as high-quality researches.

statistical analysis
HRs and 95% CIs for OS and DFS were directly obtained from each study if available or were calculated from raw data using the method reported by Tierney et al. [44]. When analyzing the relationship between PLR and clinicopathological factors, odds ratios (OR) and 95% CI were combined. The heterogeneity between the studies was estimated with the χ 2 -based Q test and Higgins' Ι 2 statistic. A p-value < 0.1 for the Q-test or Ι 2 > 50% indicated significant heterogeneity, and the random-effects model (DerSimonian and Laird method) was used, otherwise, the fixed-effects model (Mantel-Haenszel method) was applied. Sensitivity analysis was performed by omitting each single study in turn to assess the stability of the pooled results. Begg's funnel plot was carried out to examine the publication bias. Statistical data were analyzed using STATA 12.0 (College Station, TX, USA). P < 0.05 was considered as statistically significant.

cONFLIcts OF INtErEst
The authors declare no conflicts of interest.