The Value of Platelet-to-Lymphocyte Ratio as a Prognostic Marker in Cholangiocarcinoma: A Systematic Review and Meta-Analysis

Simple Summary Platelet-to-lymphocyte ratio has shown prognostic value in several malignancies; however, its role in cholangiocarcinoma remains to be determined. Therefore, we conducted a systematic review and meta-analysis of the currently available literature. Overall, our analysis revealed that a high platelet-to-lymphocyte ratio before treatment is associated with an impaired long-term oncological outcome. Further, our results indicate that this assumption was not influenced by the used treatment modality (surgical vs. non-surgical), PLR cut-off values, study population age, or sample size of the included studies. Thus, an elevated pretreatment platelet-to-lymphocyte ratio has valid prognostic value for cholangiocarcinoma patients. Abstract The platelet-to-lymphocyte ratio (PLR), an inflammatory parameter, has shown prognostic value in several malignancies. The aim of this meta-analysis was to determine the impact of pretreatment PLR on the oncological outcome in patients with cholangiocarcinoma (CCA). A systematic literature search has been carried out in the PubMed and Google Scholar databases for pertinent papers published between January 2000 and August 2021. Within a random-effects model, the pooled hazard ratio (HR) and 95% confidence interval (CI) were calculated to investigate the relationships among the PLR, overall survival (OS), and disease-free survival (DFS). Subgroup analysis, sensitivity analysis, and publication bias were also conducted to further evaluate the relationship. A total of 20 articles comprising 5429 patients were included in this meta-analysis. Overall, the pooled outcomes revealed that a high PLR before treatment is associated with impaired OS (HR = 1.14; 95% CI = 1.06–1.24; p < 0.01) and DFS (HR = 1.57; 95% CI = 1.19–2.07; p < 0.01). Subgroup analysis revealed that this association is not influenced by the treatment modality (surgical vs. non-surgical), PLR cut-off values, or sample size of the included studies. An elevated pretreatment PLR is prognostic for the OS and DFS of CCA patients. More high-quality studies are required to investigate the pathophysiological basis of the observation and the prognostic value of the PLR in clinical management as well as for patient selection.


Introduction
Cholangiocarcinoma (CCA) is the second most common primary liver tumor, accounting for 5 to 30% of all primary liver malignancies. It originates from mutated epithelial cells of the hepatic bile ducts [1,2]. With respect to the anatomical location, CCA can be divided

Literature Search
This systematic review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the ID CRD42021271435 and was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) criteria. PubMed and Google Scholar were systematically searched. The following full-text terms were searched: "Lymphocytes" OR "Platelet-to-lymphocyte ratio" AND "Cholangiocarcinoma (CCA)" OR "Biliary tract cancers (BTC)". The Boolean operator "OR" was used to combine all expressions of cases including abbreviation, while "AND" was used to include lymphocytes and PLR in conjunction with CCA in the search. The search period of the electronic database was from January 2000 to October 2021. During the literature search, no proximity operators were used. Two authors (LD and JB) conducted two independent literature searches in this systematic review, both using the same strategy. No additional papers were chosen after the reference list and citation search were completed. There was no search for unpublished literature.

Inclusion and Exclusion Criteria
Studies concerning the prognostic role of the PLR in CCA were the first choice for inclusion. Further criteria for selection included data on overall survival (OS) or disease-free survival (DFS) for evaluation and pre-treatment determination of the PLR. Exclusion criteria were (1) no access to the full text for quality assessment and data extraction; (2) review, case report, comment, or editorial; (3) non-English studies.

Statistical Analysis
The hazard ratio (HR) and its 95% confidence interval (CI) were used to assess the association between the PLR and the prognosis of CCA. If the relevant data was not directly reported, it was extracted using survival data from Kaplan-Meier curves by Engauge Digitizer version 12.1, as described previously [19]. RevMan version 5.4 (Cochrane Collaboration, London, UK) was used to merge the results of the studies. Statistical heterogeneity between trials was assessed by a Chi-squared test and suggested to be significant when I 2 > 50% and/or p < 0.05. A fixed-effects model was used when no heterogeneity was detected among studies, while a random-effects model and subgroup analysis were preferred when variance existed. Subgroup analysis was conducted to explore and explain the heterogeneity among the results of different studies. To determine the stability of the overall treatment effects, sensitivity analyses were performed. Therefore, we excluded one study at a time to ensure that no single study would be solely responsible for the significance of any result. Funnel plots were performed to evaluate publication bias [20].

Quality Assessment of Studies
The quality of the selected studies was systematically evaluated by 2 reviewers (DL and JB) using the Newcastle-Ottawa scale [21]. The Newcastle-Ottawa scale is composed of three parameters of quality-selection, outcome assessment, and comparability. Each study was subsequently scored from 0 to 9 points, with higher values indicating better quality.
Engauge Digitizer version 12.1, as described previously [19]. RevMan version 5.4 (Cochrane Collaboration, London, UK) was used to merge the results of the studies. Statistical heterogeneity between trials was assessed by a Chi-squared test and suggested to be significant when I 2 > 50% and/or p < 0.05. A fixed-effects model was used when no heterogeneity was detected among studies, while a random-effects model and subgroup analysis were preferred when variance existed. Subgroup analysis was conducted to explore and explain the heterogeneity among the results of different studies. To determine the stability of the overall treatment effects, sensitivity analyses were performed. Therefore, we excluded one study at a time to ensure that no single study would be solely responsible for the significance of any result. Funnel plots were performed to evaluate publication bias [20].

Quality Assessment of Studies
The quality of the selected studies was systematically evaluated by 2 reviewers (DL and JB) using the Newcastle-Ottawa scale [21]. The Newcastle-Ottawa scale is composed of three parameters of quality-selection, outcome assessment, and comparability. Each study was subsequently scored from 0 to 9 points, with higher values indicating better quality.

Study Characteristics and Quality Assessment
An overview of the included publications is provided in Table 1. All studies were retrospective cohort studies published over the last 6 years. A total of 5429 patients were eligible for analysis, of which 4453 individuals underwent liver resection, while 976 were treated non-surgically. Each study reported survival and the PLR using a variety of approaches. A subset of 12 studies investigated iCCA, 6 studies focused on eCCA, and 2 studies comprised both tumor entities. All studies reported a correlation between OS and the PLR, and 8 studies further reported a correlation between DFS and the PLR. The Newcastle-Ottawa scale ranged from 6 to 9, indicating an overall good quality of the methodology of the included studies (Table 2).

Correlation between the PLR and OS of CCA Patients
Eight studies identified the PLR as an independent predictor for impaired OS in patients with CCA [22,28,29,33,34,36,38,41], while the PLR was not prognostic for OS in twelve studies [23][24][25][26][27][30][31][32]35,37,39,40]. The combined analysis of all twenty publications showed that the PLR values higher than the defined cut-off values predicted a worse OS (HR = 1.14, 95% CI = 1.06-1.24, p < 0.01) with high heterogeneity (I 2 = 55%, p < 0.01, Figure 2). Table 2. Quality of included cohort studies evaluated by modified Newcastle-Ottawa scale. The quality of the included studies was assessed under six items of Hayden et al. [21] All included translational studies reporting oncological outcomes were evaluated in accordance with the Newcastle-Ottawa scale. The maximum score of the scale is nine points, with studies being categorized as low (0-3 points), moderate (4-6 points), or high quality (7-9 points).

Ref.
Author Selection Comparability Outcomes Quality Score [22] Zhao

Correlation between the PLR and DFS of CCA Patients
Five cohort studies showed that the PLR was an independent indicator of poor DFS in patients with CCA [28,29,31,32,36], whereas three publications detected no significant relationship between the PLR and DFS [22,23,36]. The pooled analysis of all eight studies revealed that a higher PLR was associated with worse DFS (HR = 1.57, 95% CI = 1.19-2.07, p < 0.01) with high heterogeneity (I 2 = 76%, p < 0.01, Figure 3).

Figure 2. Forest plot of the correlation between PLR and OS in CCA patients.
A random-effects model was used to estimate the relationship between PLR and OS. OS, overall survival; PLR, platelet-to-lymphocyte ratio.

Figure 3. Forest plot of the correlation between PLR and DFS in CCA patients.
A random-effects model was used to estimate the relationship between PLR and DFS. DFS, disease-free survival; PLR, platelet-to-lymphocyte ratio.

Subgroup Analyses of Correlation between the PLR and OS of CCA Patients
As stated above, significant heterogeneity was observed in the HR of the OS for the PLR (I2 = 55%, p < 0.01, Figure 2) and the DFS for the PLR (I2 = 76%, p < 0.01, Figure 3). Considering the limited number of studies of DFS, we exclusively explored potential causes of the heterogeneity of the OS by subgroup analyses, focusing on cancer type, specific treatment, PLR cut-off values, sample size, and age.
First, we analyzed the significance of a high PLR with respect to OS for patients according to different cancer types, including CCA (combined analysis of iCCA and eCCA), iCCA, and eCCA. While statistical heterogeneity was observed in the eCCA subgroup (I2 = 60%, p = 0.03), with no significant correlation to OS, heterogeneity was not found in the

Subgroup Analyses of Correlation between the PLR and OS of CCA Patients
As stated above, significant heterogeneity was observed in the HR of the OS for the PLR (I2 = 55%, p < 0.01, Figure 2) and the DFS for the PLR (I2 = 76%, p < 0.01, Figure 3). Considering the limited number of studies of DFS, we exclusively explored potential causes of the heterogeneity of the OS by subgroup analyses, focusing on cancer type, specific treatment, PLR cut-off values, sample size, and age.
First, we analyzed the significance of a high PLR with respect to OS for patients according to different cancer types, including CCA (combined analysis of iCCA and eCCA), iCCA, and eCCA. While statistical heterogeneity was observed in the eCCA subgroup (I2 = 60%, p = 0.03), with no significant correlation to OS, heterogeneity was not found in the CCA subgroup (I2 = 0%, p = 0.44) or the iCCA subgroup (I2 = 35%, p = 0.12, Table 3, Figure S1A).

Sensitivity Analyses of Correlation between the PLR and Prognosis of CCA Patients
We adopted a random-effects model in the sensitivity analyses, deleting each study in each turn, to further determine the robustness of the prognostic role of the PLR in the DFS and OS of CCA. As shown in Figures 4 and 5, the results of heterogeneity were changed when the studies of Zhang Z.Y. et al. [24] and Huh G. et al. [28] were deleted from the data set. The I 2 decreased from 76 to 0% and 55 to 47%, respectively, but still displayed an unfavorable prognostic effect for DFS (HR = 1.66, 95% CI = 1.40-1.97, p < 0.01) and OS (HR = 1.11, 95% CI = 1.03-1.19, p < 0.01). These results indicate a significant contribution of these studies to the high heterogeneity regarding the outcome measures and further support the robustness of the PLR as a prognostic factor of DFS and OS.

Publication Bias
The results from the funnel plot analysis ( Figure S2) demonstrated that asymmetry was not present and, therefore, no publication bias affecting the HRs could be displayed. changed when the studies of Zhang Z.Y. et al. [24] and Huh G. et al. [28] were deleted from the data set. The I 2 decreased from 76 to 0% and 55 to 47%, respectively, but still displayed an unfavorable prognostic effect for DFS (HR = 1.66, 95% CI = 1.40-1.97, p < 0.01) and OS (HR = 1.11, 95% CI = 1.03-1.19, p < 0.01). These results indicate a significant contribution of these studies to the high heterogeneity regarding the outcome measures and further support the robustness of the PLR as a prognostic factor of DFS and OS.

Publication Bias
The results from the funnel plot analysis ( Figure S2) demonstrated that asymmetry was not present and, therefore, no publication bias affecting the HRs could be displayed. changed when the studies of Zhang Z.Y. et al. [24] and Huh G. et al. [28] were deleted from the data set. The I 2 decreased from 76 to 0% and 55 to 47%, respectively, but still displayed an unfavorable prognostic effect for DFS (HR = 1.66, 95% CI = 1.40-1.97, p < 0.01) and OS (HR = 1.11, 95% CI = 1.03-1.19, p < 0.01). These results indicate a significant contribution of these studies to the high heterogeneity regarding the outcome measures and further support the robustness of the PLR as a prognostic factor of DFS and OS.

Publication Bias
The results from the funnel plot analysis ( Figure S2) demonstrated that asymmetry was not present and, therefore, no publication bias affecting the HRs could be displayed.
Of note, our subgroup analysis further demonstrated that the unfavorable effect of the PLR is independent from different treatment types, including palliative or curative therapy, the sample size, and the PLR cut-off value, while we observed no statistical significance for patients younger than 60 years. The independence of the prognostic value from the used treatment is particularly interesting, as the genuine oncological outcome of palliative compared to curative treatment is hardly comparable in CCA. While the median OS in the palliative setting is usually less than 12 months with systemic therapy, 5-year survival rates higher than 50% are reported in distinct subgroups of CCA patients after curativeintent surgical therapy [52][53][54]. This observation indicates that the PLR could be closely associated with the individual tumor biology and, therefore, could predict the outcome irrespective of the standard of care treatment at each oncological stage. Interestingly, Zheng et al. conducted a similar meta-analysis for hepatocellular carcinoma and also identified a high PLR as an independent risk factor for OS and DFS in HCC patients, both within curative and palliative settings [50].
Further, our subgroup analysis failed to detect a statistically significant association between OS and the PLR in eCCA patients with marginally non-significant values (HR = 1.37, 95% CI = 0.93-2.03, p = 0.11). However, it must be noted that the number of included eCCA studies was limited (n = 6) and comprised perihilar as well as distal cholangiocarcinoma, translating to significant heterogeneity (I 2 = 60%, p = 0.03. Figure S1A, Table 3) between the included studies. In addition, eCCA is characterized by a high degree of infectious complications due to recurrent cholangitis, which might be the cause of a dismal long-term outcome [55,56]. Considering the primary association of the PLR with oncological survival, it can be assumed that the predictive value of the PLR might be mitigated by the impaired outcome due to septic events.
Chronic inflammation is thought to play a major role in up to 15% of cancer cases around the world [57]. It is generally known that the systemic inflammatory response plays a key role in carcinogenesis and patient survival. Systemic inflammation is primarily reflected by changes in blood parameters and can be determined by the number of several cell components (neutrophils, lymphocytes, monocytes, and platelets) using standard clinical thresholds [58]. Tumor cells have been demonstrated to excrete platelet-stimulating factors, which promote primary tumor growth, invasion, and metastasis through a variety of pathways [59]. As a result, the platelet count in the peripheral blood is suggested to be an indirect predictor of tumor activity [60,61]. Moreover, the presence of antitumor lymphocytes in the peripheral blood, particularly CD8+ T cells, would indicate tumor suppression activity [62]. The ratio of platelets and lymphocytes could, therefore, be an indicator of antitumor activity, prognosis, and/or response to treatment.
However, the detailed mechanism behind the prognostic significance of the PLR in cancer is still unknown. Cancer and inflammatory cells interact reciprocally in experimental studies in terms of extracellular matrix remodeling, angiogenesis, and metastatic preparation [63]. An elevated PLR indicates the activation of transcription factors of an inflammatory response, for example, the signal transducer and activator of transcription 3 (STAT3), hypoxia-inducible factor 1a (HIF1a), and nuclear factor-kB (NF-kB) [64,65]. These transcription factors result in the secretion of pro-inflammatory cytokines that also promote tumor growth, such as TNF-a, IL-1β, and IL-6 [66,67]. In addition, cancer-related inflammation plays a role in epithelial-mesenchymal transition (EMT), angiogenesis, cell proliferation and survival, tumor-cell migration, invasion, and metastasis, as well as treatment response [63]. NF-κB was found to be overexpressed in CCA tissues and the inhibition of NF-κB action significantly enhanced cell apoptosis and reduced cell growth, which suggests NF-κB as a potential molecular target for CCA therapy [68]. Yang et al. observed that STAT3 overexpression promotes metastasis in iCCA and correlates negatively with surgical outcomes [69]. In addition, Yu et al. found that HIF1a could activate the lncRNA H19-mediated miR-612/Bcl-2 pathway to promote cholangiocarcinoma [70]. An elevated PLR might, therefore, also be a surrogate for the activity of transcription factors associated with cancer progression in CCA. This is also supported by an increasing number of studies focusing on the relationship between the PLR and tumor characteristics. For example, in cervical cancer and colorectal cancer, a higher PLR was associated with a higher rate of lymph node metastasis [71,72].
It is well known that escaping from immune surveillance is a hallmark feature of tumorigenesis and cancer progression. Recently, immunotherapy, for example, immune checkpoint inhibitors (ICIs), chimeric antigen receptor (CAR) T-cells, and tumor vaccines emerged as novel treatment modalities for malignancies [73]. However, the response rates to immunotherapy are quite different in CCA compared to other solid tumors due to the spatial heterogeneity of biliary tract cancer itself [74]. In fact, there is a lack of reliable predictive biomarkers, which is a main obstacle in the use of immunotherapies in CCA [75]. Thus, the identification of such biomarkers identifying subgroups would facilitate clinical management in CCA significantly.
A recent meta-analysis of 1845 patients with non-small cell lung cancer (NSCLC) comprising 21 studies that included treatment with three ICIs found that a high PLR was associated with poor OS and PFS [76]. Another meta-analysis from Xu et al. evaluated 12 trials including 1430 cancer patients and observed the detrimental impact a high PLR had on the efficacy of ICIs (HR for OS = 2.0) [77]. Diem et al. also reported that an elevated PLR was associated with shorter OS and PFS and lower response rates in NSCLC patients treated with nivolumab [78]. Hence, the PLR might serve as an easily accessible prognostic marker for the response to immunotherapy. Nevertheless, Zer et al. detected no significant link between the baseline PLR and immunotherapy efficacy in CCA. Therefore, more studies are required to evaluate the PLR as a predictive biomarker for monitoring therapy success [79].
As with all meta-analyses with limited available literature, our study has certain limitations. All included studies were retrospective analyses in nature, leaving a potential selection bias in the published data. In addition, a variety of methodologies was used and, most importantly, different PLR cut-off levels. These different cut-offs impede the routine application of the parameter in clinical management and warrant further research. As some studies did not report HRs and CIs in detail, these variables had to be extrapolated from the survival curves in four studies [24][25][26][27][28][29][30][31]37,39]. Further, the available data were unfortunately not sufficient to investigate the association between the PLR and tumor clinicopathological characteristics.

Conclusions
In light of the above, we were able to demonstrate that an elevated pretreatment PLR is predictive for the prognosis of CCA patients. Large-scale prospective cohort studies are warranted to confirm the independent prognostic effect of the PLR on CCA.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/cancers14020438/s1, Figure S1: Stratified forest plots of the association between the PLR and OS in CCA patients. Figure

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.