Prognostic impact of platelet-to-lymphocyte ratio on diffuse large B-cell lymphoma: a meta-analysis

Background Recently, some studies reported the prognostic value of platelet-to-lymphocyte ratio (PLR) in patients with diffuse large B-cell lymphoma (DLBCL), however, the results varied from different studies. Therefore, we performed a meta-analysis to explore the prognostic value of PLR in DLBCL. Methods A comprehensive literature retrieval was conducted by using PubMed, Embase, Web of Science, the Cochrane Library, the China National Knowledge Infrastructure (CNKI), and Wanfang. Pooled hazard ratio (HR) and 95% confidence interval (CI) were used to evaluate the association of PLR and overall survival (OS) and progression-free survival (PFS). Odd ratios (ORs) and 95% CIs for clinicopathological characteristics were statistically analyzed. Results Eight studies with 1931 patients were included for meta-analysis. The pooled analysis indicated that elevated PLR was significantly associated with poor OS (HR = 1.73, 95% CI 1.29–2.31, p < 0.001), but not PFS (HR = 0.85, 95% CI 0.57–1.27, p = 0.438). Furthermore, elevated PLR was significantly associated with presentation of B symptoms (OR = 2.27, 95% CI 1.29–3.98, p = 0.004), elevated lactate dehydrogenase (LDH) (OR = 2.76, 95% CI 2.05–3.72, p < 0.001), higher tumor stage (OR = 2.22, 95% CI 1.66–2.98, p < 0.001), and Eastern Cooperative Oncology Group (ECOG) performance status (PS) ≥ 2 (OR = 1.71, 95% CI 1.09–2.69, p = 0.019). However, elevated PLR was not significantly correlated with gender, age or cell of origin. Conclusion This meta-analysis revealed that PLR may be an effective and noninvasive biomarker for poor prognosis and aggressive disease characteristics for patients with DLBCL.

significant prognostic marker in various solid tumors including esophageal cancer [7], head and neck squamous cell carcinoma [8], ovarian cancer [9], and breast cancer [10]. Previous retrospective studies also explored the prognostic effect of PLR in DLBCL, whereas the results were inconsistent even contrary [11][12][13][14][15][16]. Therefore, it is necessary to conduct a meta-analysis to comprehensively evaluate the prognostic and clinicopathological role of PLR in DLBCL patients.

Search strategy
We performed the present meta-analysis based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement [17]. The databases including PubMed, Web of Science, Cochrane Library, Embase, CNKI (Chinese), and Wanfang were searched for relevant studies. The search strategies included the combination of MeSH terms and free-text terms: "platelet-to-lymphocyte ratio, " "platelet-lymphocyte ratio, " "PLR, " and "diffuse large B-cell lymphoma or DLBCL". The last search was up to May, 2019. In addition, the references list of relevant researches was examined to identify relevant studies. Ethical approval was not required for this study because all data were from previous published studies.

Selection criteria
The inclusion criteria were following: (1) patients with DLBCL were diagnosed histologically; (2) studies reported the prognostic role of PLR on overall survival (OS) or/and progression-free survival (PFS) or provided sufficient data for calculation [18]; (3) a cut-off value of PLR was identified; (4) studies were published in English or Chinese; (5) the laboratory parameters including the blood counts were assessed prior to the start of first chemotherapy cycle. Studies did not meet the inclusion criteria were removed.

Data extraction and quality evaluation
Two experienced investigators (Y.C. and Z.Z.) independently extracted the data. Any discrepancies were resolved by discussion with the third investigator (H.J.). The main characteristics of eligible studies were: name of first author, year of publication, country, sample size, patients' age, gender, treatment regimen, stage, cut-off value of PLR, study period, and survival outcomes. The quality assessment of included studies were performed according to Newcastle-Ottawa Scale (NOS) [19]. The NOS consists of three parts: selection (0-4 points), comparability (0-2 points), and outcome (0-3 points). The studies of highest quality scored 9 points and studies ≥ 6 points indicated a high quality.

Statistical analysis
All statistical analyses were performed by using Stata version 12 (STATA, College Station, TX). Pooled hazard ratio (HR) and 95% confidence interval (CI) were used to evaluate the association of PLR and OS and PFS. HR and 95% CI were extracted from included studies if reported, or were calculated from Kaplan-Meier curves by Tierney's method [18]. The heterogeneity across studies was assessed by Cochran's Q test and I-squared test [20,21]. A random-effect model was employed when the heterogeneity was significant (p < 0.10 or I 2 > 50%), otherwise, a fixed model was applied. Odd ratios (ORs) and 95% CIs for clinicopathological characteristics were statistically analyzed. Sensitivity was performed by sequential omitting of each study to test the credibility of the results. The potential publication bias was examined by Begg's funnel plot and Egger's test. A two-tailed p-value < 0.05 was defined as statistically significant.

Search results
A total of 49 studies were identified after initial literature search. After duplicate records were removed, 20 studies remained. Seven records were excluded by title/abstract screening and 13 studies were left for full-text evaluation. Five full-text articles were removed for reasons: no available data, not on PLR, and overlapped patients. At last, eight studies [11-16, 22, 23] were included in the current meta-analysis. The detailed process of literature retrieval was depicted in Fig. 1.

PLR and clinicopathological characteristics
The correlation of PLR and several clinicopathological characteristics were analyzed.

Sensitivity analysis
To evaluate the influence of each single study of the pooled results, sensitivity analysis was carried out. As shown in Fig. 4, the pooled HRs were not significantly altered by any individual study, which indicated the stability of the results.

Discussion
Inflammation plays a pivotal role in tumor progression [24]. PLR was widely investigated for prognosis of DLBCL patients with distinct results. Previous studies reported the prognostic value of PLR in DLBCL patients [11-16, 22, 23], whereas the results were inconsistent. For example, some studies [22,23] demonstrated that PLR was a significant prognostic factor for DLBCL patients, whereas other studies failed to find the prognostic value of PLR [11,13,14]. As meta-analysis can aggregate data from a series of studies and make quantitative analysis, therefore, the results of meta-analysis are objective and credible.
In the present meta-analysis, we aggregated data from eight studies with 1931 patients to shed light on this issue. The results showed that a high PLR was a significant prognostic marker for poorer OS (HR = 1.73, 95% CI 1.29-2.31, p < 0.001). Furthermore, the prognostic efficiency enhanced for Asian patients (HR = 1.95, 95% CI 1.34-2.84, p < 0.001) and with a cut-off value > 150 (HR = 1.76, 95% CI 1.25-2.49, p = 0.001). However, PLR was not associated with PFS in DLBCL, regardless of ethnicity, sample size, or cut-off value of PLR. We also found that PLR was significantly correlated to presentation of B symptoms, elevated LDH, higher tumor stage, and ECOG PS ≥ 2. The results suggested that PLR was positively connected with clinical features reflecting high aggressiveness of the disease. Taken together, this study revealed that PLR was a significant prognostic factor for poor OS and invasiveness in DLBCL patients. To our knowledge, this is the first meta-analysis investigating the prognostic and clinicopathological value of PLR in DLBCL. Notably, the eight included studies are retrospective study design and recruited patients with one ethnicity. In the present meta-analysis, we collected the data and conducted subgroup analysis to investigate the prognostic value of PLR in different We also investigated the correlation of PLR and clinical features with enlarged sample size compared with included studies. The current meta-analysis provides more comprehensive and systemic analysis than any single included study. Those factors were strengths of this meta-analysis.
Recent evidence suggests that inflammation response is involved in the processes of tumor angiogenesis, tumor growth, and metastasis [25]. However, the mechanism underlying the correlation between high PLR and poor prognosis in DLBCL patients has not been determined. The increasing of platelet counts and/or decreasing of lymphocyte counts can result in a high PLR. On the one hand, activated platelets were involved in early and advanced stages of tumor angiogenesis [26]. Platelets could secrete various biological molecules to facilitating angiogenesis in tumor microenvironment [27]. In addition, platelets derive transforming growth factor-β1 (TGF-β1), work together with the direct interaction of platelets and tumor cells to activate epithelial-mesenchymal transition (EMT) related pathways and induce subsequent metastasis [28]. On the other hand, lymphocytes exert critical roles in antitumor immune responses. Tumor-infiltrating lymphocytes (TILs) including CD3+ T cells, CD8+ T cells, Th1 CD4+ T cell could inhibit tumor cell proliferation and metastases [29,30]. Therefore, it is reasonable to apply PLR as an easily available immunological parameter to predict survival outcomes in cancer patients.
Previous studies also demonstrated the prognostic value of PLR in various tumors [31]. A recent meta-analysis showed that a high NLR was significantly associated with decreased OS and PFS in ovarian cancer [9]. Another work suggested that higher PLR was an indicator of poor progress in oral cancer [32]. Those findings were in accordance with our results in DLBCL. In addition, in the present meta-analysis, we found that PLR was a significant prognostic factor for OS, especially in Asian patients and PLR > 150. Those results suggest that PLR may have enhanced prognostic role when the cut-off value > 150, which provides implications for clinical use. An elevated PLR was also correlated to aggressive tumor characteristics, which may imply that DLBCL patients with high PLR should be treated with strong therapeutic strategies. However, we did not observe significant prognostic impact of PLR on PFS, which may be explained by the relative short follow-up of PFS, compared to OS. In addition, the results suggested that cell of origin had nonsignificant association with PLR. However, because only two studies were included for analysis, which may lead to the negative results, therefore, more large-scale studies are still needed.
There are several limitations to this study. First, significant heterogeneity was observed in the analysis, although we applied random-effect model accordingly. Because the included studies recruited patients with different ethnicity, disease stage, cut-off values and treatment strategies, which could result in heterogeneity in the meta-analysis. The subgroup analysis showed that significant heterogeneity still exists in different sample size and cut-off values groups. These indicated that the heterogeneity could be inherent among included studies and various cut-off values may be a source of heterogeneity. According to the heterogeneity, we selected corresponding effects model (random effects model or fixed effects model) to pool the data. Second, the cut-off values of PLR were different in included studies, which may influence the distribution of low and high PLR groups and cause heterogeneity. Third, we only included studies published in English and Chinese, therefore, relevant studies published in other languages may be unavailable.

Conclusion
Our study shows that high PLR was a significant prognostic marker for poorer OS in DLBCL. Furthermore, PLR was associated with presentation of B symptoms, elevated LDH, higher tumor stage, and ECOG PS ≥ 2. Considering the above-mentioned limitations, large-scale prospective studies with uniform cut-off value of PLR are needed to validate our findings.