Elevated pretreatment platelet-to-lymphocyte ratio is associated with poor survival in stage IV non-small cell lung cancer with malignant pleural effusion

A higher platelet-to-lymphocyte ratio (PLR) has a clinical correlation with shorter survival in non-small cell lung cancer (NSCLC). The present study evaluated the association between the PLR and survival in patients with advanced NSCLC with malignant pleural effusion (MPE). Between January 2012 and July 2016, 237 patients with stage IV NSCLC were selected for evaluation. Receiver operating characteristic analysis was used to determine a cutoff for the PLR. Clinicopathological characteristics were compared between the high and low PLR groups, and the role of PLR as a predictive/prognostic maker was investigated. Among the 237 patients, 122 were assigned to the low PLR group and the other 115 to the high PLR group. According to multivariate analysis, male sex, not receiving active anticancer treatment, low hemoglobin level, low albumin level, high C-reactive protein level, and high PLR were identified as significant risk factors for shorter overall survival (OS) (p = 0.010, <0.001, 0.011, 0.004, 0.003, and <0.001, respectively). In the subgroup multivariate analysis of driver mutation-negative NSCLC, high Eastern Cooperative Oncology Group score, not receiving active anticancer treatment, low hemoglobin level, high C-reactive protein level, and high PLR were identified as significant risk factors for shorter OS (p = 0.047, <0.001, = 0.036, = 0.003, and <0.001, respectively). A high pretreatment PLR is independently associated with poor survival in stage IV NSCLC with MPE and in a subgroup of epidermal growth factor receptor and anaplastic lymphoma kinase wild-type NSCLC.

Lung cancer is one of the leading causes of cancer death worldwide 1 . Non-small cell lung cancer (NSCLC) comprises 85% of all lung cancers 2 . Despite recent advances in treatment modalities, such as targeted therapies, the 5-year survival rate of NSCLC is only 16.6% 3 .
Inflammation plays a key role in cancer progression 4 . Several studies have shown that elevated systemic inflammation contributes to angiogenesis, invasion, and metastasis of cancer cells 5,6 . Furthermore, an association between inflammation and poor prognosis has been confirmed in various cancers [7][8][9] . The clinical significance and usefulness of inflammatory markers that can predict prognosis during cancer treatment are increasing 10 .
In lung cancer, inflammatory markers such as C-reactive protein (CRP), neutrophil-to-lymphocyte ratio, Glasgow prognostic score, and platelet-to-lymphocyte ratio (PLR) have been evaluated [11][12][13] . Among them, PLR, which is defined as the absolute platelet count divided by the absolute lymphocyte count, is a widely studied inflammatory marker 10,14 . Its prognostic value has been evaluated in various types of cancer, including colorectal, breast, ovarian, and pancreatic [15][16][17] . In NSCLC, an elevated PLR is associated with poor overall survival (OS) and progression-free survival (PFS) 3,14 . An elevated PLR was also associated with poor prognosis in surgically treated NSCLC patients 13 . Pretreatment PLR was identified as an independent risk factor for brain metastasis of lung adenocarcinoma 10 . Although the predictive and prognostic values of PLR have been evaluated in both advanced (2019) 9:4721 | https://doi.org/10.1038/s41598-019-41289-9 www.nature.com/scientificreports www.nature.com/scientificreports/ lung cancer and lung cancer requiring surgery, PLR has not been assessed specifically in stage IV NSCLC patients with cytologically proven malignant pleural effusion (MPE). MPE is common complication which can occur during the course of treatment in advanced NSCLC 18,19 . Moreover, NSCLC patients with MPE show poor prognosis with a median survival time of less than 12 months 20 , and complain more respiratory symptoms which lead to decreased quality of life 21 . While they show distinct and more unfavorable clinical manifestations, evaluation of prognostic values of PLR in this subgroup of advanced NSCLC is meaningful.
Furthermore, the prognostic and predictive values of PLR in patients with epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) wild-type NSCLC specifically have not been investigated. The evaluation of PLR in a driver mutation-negative subgroup can be clinically important after excluding patients with EGFR and ALK mutations, who are likely to have improved survival 22 . Patients with EGFR mutations are eligible for targeted therapy and thus have a longer median survival compared with those without EGFR mutations 22 . For patients with an ALK gene rearrangement, the ALK tyrosine kinase inhibitor crizotinib is an available treatment modality 23 . In addition, crizotinib therapy improved the survival of patients with ALK-positive NSCLC compared with those not treated with crizotinib 23 .
In this study, we evaluated the predictive and prognostic values of PLR in stage IV NSCLC patients with MPE and characterized the clinical significance of PLR in a wild-type EGFR and ALK NSCLC subgroup.  24 . Inclusion criteria for enrollment were patients with 1) MPE diagnosed either by pleural biopsy that proved malignant or cytologically and 2) all clinical data available from electronic medical record. Exclusion criteria were patients 1) with SCLC, 2) with significant infection, 3) who were on antibiotic treatment at the time of enrollment, and 4) with a concurrent hematologic disease 24 . Clinical and laboratory data. For all enrolled patients, data including sex, age, histology, tumor stage by tumor-node-metastasis stage (AJCC 2009), Eastern Cooperative Oncology Group performance status (ECOG PS), smoking status, and baseline blood chemistry values were collected for all enrolled patients. The PLR was calculated based on the initial pretreatment complete blood count. eGFR and ALK testing. Excluding other uncommon EGFR profiles, EGFR mutations were defined as an exon 21 point mutation or an exon 19 deletion. EGFR genotyping was performed by peptide nucleic acid (PNA)-mediated PCR clamping methods using the PNAClamp TM EGFR Mutation Detection Kit (PANAGENE, Inc., Daejeon, Korea) 24,25 .

Methods
Specimens for FISH acquired from the study centers were prepared using a molecular analysis platform and analyses were performed at Yeouido St. Mary's Hospital Central Molecular Laboratory 24,26 . Utilizing a break-apart probe specific to the ALK locus, the Vysis LSI ALKDual Color Break Apart Probe (Abbott Molecular, Abbott Park, IL, USA), FISH was carried out on formalin-fixed paraffin-embedded (FFPE) tumor specimens. Positive ALK rearrangement was defined as an isolated red signal or a split signal; those with >15% of counted nuclei within tumor cells exhibiting an isolated red signal or split signal were considered as positive cases. A specimen-specific assessment approach was used to minimize technical bias. A total of 100 tumor cells were scored in surgical resection specimens. An ALK FISH split signal rate of <15% was interpreted as negative 26 . Chemotherapy and adverse reactions. Systemic conventional chemotherapies given to the patients comprised pemetrexed, docetaxel, gemcitabine, or paclitaxel combined with cisplatin/carboplatin regimens 24 . Targeted therapies were comprised of erlotinib, gefitinib, and afatinib for positive EGFR mutations and crizotinib for positive ALK translocations. As a first line regimen, the standard recommendation in the study centers was platinum-based doublet chemotherapy regimen, however in some special cases, monotherapy regimen was also considered for treatment.
During the treatment course, all study patients were checked for treatment-related adverse reactions either at the outpatient clinic or during admission. All significant adverse reactions were available from electronic hospital records. Treatment was either delayed or changed to another regimen, when patients experienced grade III or IV adverse reactions. os and pFs. Response Evaluation Criteria in Solid Tumors version 1.1 was used to evaluate responses to cancer treatment 27 . For response evaluation, patients underwent a computed tomography scan after every two cycles of treatment. Treating physicians and independent radiologists evaluated the responses. OS was defined as the time from the date of lung cancer diagnosis to death. PFS was defined as the time duration between lung cancer diagnosis and disease progression after first-line treatment. Patients were considered censored if they died or lost contact during the follow-up period 24 .
Defining cutoff values. The optimal cutoff value for PLR was determined by receiver operating characteristic (ROC) analysis. Survival outcomes were dichotomized by survival (alive or dead). All survival statuses were assessed in October 2017.
For other continuous values entered in the univariate analyses of clinical outcomes, their cutoff points were also calculated from ROC curves. Normal range values were used if the p-value from the ROC analysis was not significant.
www.nature.com/scientificreports www.nature.com/scientificreports/ statistical analysis. All statistical analyses were done using the Statistical Package for Social Sciences software version 18.0 (SPSS Inc., Chicago, IL, USA). Data of continuous variables are shown as means or medians with ranges. The Chi-squared test was performed to compare categorical parameters. Continuous variables were analyzed using two-sided t-tests or the Mann-Whitney U test depending on the distribution status.
Univariate analysis using the Cox regression model was performed to determine the variables significantly associated with OS and PFS. The median OS and PFS are presented as 95% confidence intervals. Survival curves were constructed using Kaplan-Meier analysis. A log-rank test was performed to determine significant differences in the survival outcomes between groups. Statistically significant variables were entered into multivariate analysis using the Cox proportional hazards regression model. A P value less than 0.05 was considered statistically significant in all analyses. ethics statement. The  (range, 11.9-18.9) months. The median PFS after first-line treatment was 6.9 (range, 5.7-8.1) months.
The optimal cutoff value for the PLR, calculated by ROC analysis, was 181.24, with an area under the curve of 0.619 (p = 0.002) (Fig. 1). Clinical parameters were statistically compared between the high and low PLR groups, defined according to this cutoff value. The median age was 70 (range, 32-90) years in the high PLR group and 67 (range, 38-92) years in the low PLR group. The median OS was significantly different between the two groups (p < 0.001): 10.7 (range, 7.0-14.4) months versus 28.3 (range, 19.7-36.9) months in the high PLR versus low PLR group. There was no significant difference in PFS between the two groups. The proportion of males, ECOG PS, histologic features, proportion of ever smokers, tumor-node-metastasis stage, and first-line treatment modalities were not significantly different between the two groups, nor was the proportion of patients who underwent pleurodesis before or during anticancer treatment.
There were no significant differences in the mean white blood cell count, carcinoembryonic antigen, lactate dehydrogenase, or protein levels between the low PLR group and the high PLR group (8807.9 vs 9135; 114.2 vs 511.2; 496.4 vs 512.6; and 6.7 vs 6.6, respectively). The high PLR group showed significantly higher platelet level, PLR, lower albumin and hemoglobin levels than the low PLR group (329,630 vs 269,040, p < 0,001; 284.3 vs 135.0, p < 0.001; 3.6 vs 3.7, p = 0.005; and 12.7 vs 13.7, p < 0.001, respectively). The mean CRP level was higher in the high PLR group, although the difference did not reach statistical significance (p = 0.076). overall survival. Univariate and multivariate analyses of survival are shown in Table 2. The following factors showed a significant association with shorter OS in the univariate analyses: male sex (p = 0.001), squamous cell histology (p = 0.019), not receiving active anticancer treatment (p < 0.001), lower hemoglobin level (p < 0.001), higher platelet count (p = 0.017), lower albumin level (p < 0.001), higher CRP level (p < 0.001), and higher PLR (p < 0.001) (Fig. 2A). These significant variables from the univariate analyses were entered into a multivariate analysis, which revealed the following as independent predictors of shorter OS: male sex (p = 0.010), not receiving active anticancer treatment (p < 0.001), lower hemoglobin level (p = 0.011), lower albumin level (p = 0.004), higher CRP level (p = 0.003), and high PLR (p < 0.001).
Progression-free survival. In univariate analyses, the following variables showed significant associations with PFS: male sex (p = 0.001), history of smoking (p = 0.007), squamous histology (p = 0.023), poor ECOG PS (p = 0.048), not receiving active anticancer treatment (p < 0.001), lower hemoglobin level (p = 0.001), higher platelet count (p = 0.015), lower albumin level (p = 0.043), and higher CRP level (p = 0.039). However, a high PLR did not show a statistically significant association (p = 0.094) (Fig. 2B). All significant variables from the univariate analyses were entered into a multivariate analysis, which revealed that only a lower hemoglobin level was a significant risk factor for shorter PFS (p = 0.010) ( Table 2).

Wild-type eGFR and ALK NsCLC. Among the 142 patients with wild-type EGFR and ALK NSCLC,
Kaplan-Meier analysis showed a significant difference in OS between the PLR groups (p = 0.002) (Fig. 3A). However, no significant difference in PFS was observed (p = 0.053) (Fig. 3B). The median OS was 27.1 months for the low PLR group and 16.2 months for the high PLR group. The median PFS was 8.0 months for the low PLR group and 5.1 months for the high PLR group.
In univariate analyses of PFS, the following variables showed significant associations with PFS: poor ECOG PS (p = 0.021), lower hemoglobin level (p = 0.002), higher platelet count (p = 0.035), and higher CRP level (p = 0.049). However, a high PLR did not show a statistically significant association (p = 0.053). All significant variables from the univariate analyses were entered into a multivariate analysis, which revealed a higher CRP level as the only significant risk factor for shorter PFS (p = 0.048) ( Table 3).

Discussion
The present study showed that a higher PLR was an independent predictor of shorter survival of stage IV NSCLC patients with MPE. This finding is consistent with previous studies that showed that an elevated PLR was an independent predictor of shorter OS of patients with NSCLC 9,13 . Several studies have shown that the PLR has prognostic value in NSCLC; however, the present study evaluated a multicenter lung cancer patient cohort, and to our knowledge, no previous study has evaluated the prognostic value of the PLR in stage IV NSCLC with www.nature.com/scientificreports www.nature.com/scientificreports/ cytologically and pathologically proven MPE. Identifying an association between the PLR and prognosis specific to NSCLC with MPE can be meaningful because NSCLC patients with MPE have distinct clinical characteristics. In lung cancer, MPE is associated with a poor prognosis 28 , and lung cancer patients with MPE have a decreased quality of life due to symptoms of dyspnea 29 .
Another novel finding of the present study is that a high PLR was associated with shorter survival of patients with wild-type EGFR and ALK NSCLC. The PLR, an important inflammatory marker with predictive value in NSCLC 14 , can also be used as a predictor of advanced NSCLC without major driver mutations. Targeted therapies such as EGFR and ALK tyrosine kinase inhibitors are typically used as first-line treatments for advanced NSCLC harboring EGFR and/or ALK mutations. While NSCLC patients with EGFR and ALK mutations show relatively  www.nature.com/scientificreports www.nature.com/scientificreports/ improved survival compared with those with wild-type EGFR and ALK 22,23 , determining the prognostic value of the PLR in wild-type EGFR and ALK lung cancer would be meaningful.
An elevated PLR can be interpreted as a simultaneous increase in the platelet count and decrease in the lymphocyte count. Platelets in the tumor microenvironment release growth factors including thrombospondin, VEGF, platelet factor 4, and platelet-derived growth factor 30 . Promoted by these secreted growth factors, platelets in the tumor microenvironment promote sustained proliferative signals, metastasis, and evasion of immune detection 30,31 . Furthermore, thrombocytosis in the tumor microenvironment is related to chemoresistance 31 . On the other hand, lymphocytes play major anticancer roles 32 , and lymphocytopenia is associated with poor survival in cancer 33 . We assume that thrombocytosis and lymphocytopenia, which occur simultaneously, contribute to a shorter overall survival.
The cutoff of PLR determined in our study was 181.24. A previous study that evaluated the association between chemotherapy response and PLR used a cutoff of 152.6 7 , and another study evaluating surgically treated NSCLC patients used a PLR cutoff of 162 34 . Our cutoff value is relatively high compared with those of previous studies; however, we believe this is due to the different methods of selecting study patients. Thrombocytosis in the tumor microenvironment is associated with increased cancer activity and metastasis 30 , and all of the study patients in our study had stage IV NSCLC. Thus, thrombocytosis may have contributed to the relatively higher cutoff value.