Lymphocyte percentage as a valuable predictor of prognosis in lung cancer

Abstract Lymphocytes and neutrophils are involved in the immune response against cancer. This study aimed to investigate the relationship between lymphocyte percentage/neutrophil percentage and the clinical characteristics of lung cancer patients, and to explore whether they could act as valuable predictors to ameliorate lung cancer prognosis. A total of 1312 patients were eligible to be recruited. Lymphocyte percentage and neutrophil percentage were classified based on their reference ranges. Survival curves were determined using Kaplan–Meier method, and univariate and multivariate cox regression analyses were performed to identify the significant predictors. Decision curve analysis was used to evaluate the clinical benefit. The results of both training and validation cohorts indicated that lymphocyte percentage exhibited high correlation with clinical characteristics and metastasis of lung cancer patients. Both lymphocyte percentage and neutrophil percentage were closely associated with survival status (all p < 0.0001). Low lymphocyte percentage could act as an indicator of poor prognosis; it offered a higher clinical benefit when combined with the clinical characteristic model. Our findings suggested that pretreatment lymphocyte percentage served as a reliable predictor of lung cancer prognosis, and it was also an accurate response indicator in lung adenocarcinoma and advanced lung cancer. Measurement of lymphocyte percentage improved the clinical utility of patient characteristics in predicting mortality of lung cancer patients.

However, the main limitation of LDCT screening is generation of false-positive results, since it is unclear whether all lesions detected in asymptomatic participants will develop significant symptoms, and affect long-term outcomes, suggesting that LDCT may be potentially harmful in large-scale screening programmes. 7,8 Thus, more effective and low-cost strategies need to be developed to consider patient acceptability, and assess the prognosis of lung cancer patients.
Circulating biomarkers in plasma and serum, which usually appear prior to imaging changes, can serve as indicators of tumour progression and predictors of prognosis. 9,10 Identifying reliable markers to better select patients for currently available and upcoming approaches, such as immunotherapy, will greatly assist clinical decisionmaking. A variety of biomarkers, including carcinoembryonic antigen (CEA), cytokeratin 19 fragments (CYFRA 21-1), carbohydrate antigen (CA)125, CA199 and lactate dehydrogenase (LDH), have been identified to be associated with lung cancer prognosis. [11][12][13][14][15] Inflammation is one of the hallmarks of cancer and plays a pivotal role in the modulation of tumour microenvironment. It can also highly influence tumorigenesis and tumour progression. 16,17 Different cells are known to be involved in the immune response against cancer, making the process dynamic and balanced. 18 Lymphocytes and neutrophils are easy to measure and may provide a more convenient strategy for the study of cancer-related inflammation. Neutrophil-to-lymphocyte ratio (NLR) has been evaluated in a variety of cancers, but its prognostic role remains controversial, which may explain the reason why it has not been incorporated into clinical practice. [19][20][21][22] For other haematological parameters, previous study has demonstrated that preoperative lymphocyte count is associated with node-negative non-small-cell lung cancer (NSCLC) prognosis. 23 An elevated neutrophil count has been shown to be a predictor of poor survival in metastatic melanoma. 24 Overall changes in lymphocytes and neutrophils with regard to inflammation and the immune state may be expressed as lymphocyte percentage (LY%) and neutrophil percentage (NEUT%).
Peripheral LY% reflects leukocytosis more directly than NLR does, the relative decrease of lymphocytes results in the diminishment of the immune response and increases the risk of cancer, and it has been reported to predict the survival more accurately than peripheral lymphocyte count in colorectal cancer. 25 However, it was less considered in previous studies regarding the prognostic values of LY% and NEUT% in large cohort of patients with lung cancer.
In this study, a retrospective analysis was performed to investigate the relationship between LY%/NEUT% and clinical characteristics of lung cancer patients, and to evaluate whether LY% and NEUT% could be used for improving prediction of patient outcome.

| Ethics statement
This study was approved by the Medical Ethics Committee and Institutional Review Board of West China Hospital. Informed consent was obtained from all patients before study. All methods used in this study were performed following the approved protocols.

| Patients
This study included 1312 patients diagnosed with lung cancer in West China Hospital from 2008 to 2014. On account of the differences in the levels of haematological indicators, 270 patients were excluded in advance due to preoperative treatment or history of cancer, while 38 patients were excluded owing to insufficient data of survival, or information regarding LY% or NEUT% in peripheral blood was not available ( Figure 1). All clinical information was extracted from the medical records after lung cancer confirmed by biopsy. Information regarding metastasis was obtained using whole-body CT scan, bone scan, lymph node biopsy and fibreoptic bronchoscopy. Survival status was determined on the last followup day, and the overall survival time was defined as the length of time between the lung cancer confirmation date and the date of death or last follow-up, which was done by visits or telephone inquiries. count to white blood cell count, and NEUT% was defined as the percentage of neutrophil count to white blood cell count. In several analyses of cancer prognosis, the cut-off value of LY% was defined as 20%, 28,29 and its normal level was considered to be 20%-50% in a study of lung cancer, 30 which was in line with the clinical criteria of West China Hospital, and based on the diagnostic criteria and clinical experience, the reference range of NEUT% was defined as 40%-75%.

| Statistical analysis
Continuous variables were presented as median (range). Categorical variables were presented as percentage (%). Chi-square test was performed to determine the statistical significance of categorical data. For survival analysis, the log-rank test was used for univariate and multivariate cox regression analysis. To measure effects of variables on survival, statistical significance was expressed as hazard ratio (HR), at 95% confidence interval (CI). Survival curves were constructed using the Kaplan-Meier method. With regard to clinical utility, the net benefit was measured by decision curve analysis

| Patient Characteristics
A total of 1312 lung cancer patients were randomly divided into two cohorts (Table S1)

| Correlation of LY% and NEUT% with clinical characteristics in all lung cancer patients
The association between LY% and the clinicopathologic characteristics of lung cancer patients was investigated (Table 1). In the training cohort, 365 cases were identified to be within the reference range,  (Table S2).

| Correlation of LY% and NEUT% with clinical characteristics in different histological subtypes
For the purpose of treatment, lung cancer was classified as SCLC and NSCLC, ADC and SCC accounted for more than 80% of NSCLC cases. 32,33 Hence, classification analyses of LY% and NEUT% in ADC, SCC and SCLC were performed.   (Table S3). in SCLC (Table S4).

TA B L E 2 (Continued)
F I G U R E 2 Kaplan-Meier curves for overall survival according to LY% (A) and NEUT% (B) in training and validation cohorts of all lung cancer patients. ****p < 0.0001. LY%: lymphocyte percentage; NEUT%: neutrophil percentage

| LY% and NEUT% were associated with overall survival of lung cancer
The overall survival time of patients in training and validation cohorts was evaluated using Kaplan-Meier survival curves. As shown in Figure 2, low LY% was strongly correlated with poor survival status (both cohorts, p < 0.0001), and NEUT% showed worse overall survival in higher level (both cohorts, p < 0.0001).
After stratification based on histology, significant differences were found in each subtype (Figure 3). Patients with lower LY% ex-

| LY% could serve as a valuable predictor of prognosis in lung cancer
Univariate and multivariate cox regression models were introduced to measure prognostic predictors of lung cancer patients. The univariate analysis revealed that aberrant levels of LY% and NEUT% conferred unfavourable prognosis (both p = 0.000). Sex, stage, smoking status, differentiation and metastasis were associated with prognosis in all lung cancer patients ( Figure 5A). A multivariate regression analysis was conducted for 8 variables with statistically significant differences (p < 0.1) in univariate analysis. The HR increased to 1.550 (95% CI: 1.332-1.804, p = 0.000) in low LY% group, compared with the reference, suggesting that low LY% could serve as an important predictor of poor prognosis for lung cancer patients.
Moreover, age older than 60 years (p = 0.018), advanced stage (III:   To ensure the reliability of results, patients in this study were randomly assigned to either training or validation cohort, and the relationship between LY% and clinicopathological factors was investigated first. In addition to some slight inconsistencies regarding age and metastatic sites, the results of two cohorts illustrated that LY% was strongly correlated with other clinical parameters, as well as bone, liver and pleural metastasis in all lung cancer patients.

| Clinical utility of LY% integration model in lung cancer prognosis
Survival curves showed that low LY% had an obvious correlation with unfavourable survival status, and based on these findings, a multivariate cox regression analysis was performed to evaluate the prognostic value of LY%; it was then confirmed to be a significant predictor of prognosis in patients with lung cancer. In addition, the incorporation of LY% resulted in better predictive performance of clinical characteristic model on patient outcome. Neutrophil plays a complex role in inflammation within the tumour, and its elevated count has been demonstrated to have prognostic value in NSCLC. 44 In this study, the role of NEUT% in lung cancer was examined as well. In spite of its relevance to clinical characteristics and survival outcomes, it could not make prediction for lung cancer prognosis in the multivariate analysis.
Stratification analyses were also performed, and it was interesting to note that NEUT% could serve as prognostic factor in both SCLC and non-metastatic patients. Regarding ADC and SCC subtypes, as well as adverse characteristics of advanced stage, undifferentiation and metastasis, LY% remained a more valuable predictor.
The current study shed light on peripheral lymphocyte percentage was significantly associated with prognosis of lung cancer patients. The percentage of lymphocyte has the potential to be a surrogate indicator of disease outcome and a stratification factor in clinical trials. While the results were derived based on the reference range of our clinical criteria, using the appropriate cut-off value for the corresponding population is the best course of action. A ROC analysis might be useful to determine the optimal cut-off level.
Hence, further studies are required before it can be established as a validated prognostic marker. NLR is a well-established predictor for patients with malignancy, which also is what mainly determines the LY%. In the subsequent analysis, the comparison for prognostic prediction potential of NLR and LY% will be conducted. In addition, we aim to figure out whether these two indicators can be combined as a more effective index to reflect the prognosis of patients with lung cancer. This study, conducted in a relatively large number of patients, obtained detailed clinical information to allow for extensive miscellaneous adjustments. However, data on the dynamic changes in lymphocyte and neutrophil percentages during tumour progression, and around the treatment period, were not available.
Furthermore, the mechanism of complex association between inflammatory cells and tumour microenvironment has not been established, while the imbalance of lymphocyte and neutrophil ratio may provide insight into tumour progression and prognosis of individuals with lung cancer. We believe that the interaction or regulation of lymphocytes and neutrophils in lung cancer is worth studying and considering.
Evidence from the study supported the idea that pretreatment lymphocyte percentage effectively predicted the prognosis of lung cancer patients, and it was also an accurate response indicator in ADC and advanced lung cancer. Based on the results obtained, the integration of lymphocyte percentage with clinical characteristic model benefited the prognostic prediction of the disease.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.