Neutrophil–lymphocyte ratio is prognostic in early stage resected small-cell lung cancer

Ethics statement

The study was conducted in accordance with the guidelines of the Helsinki Declaration of the World Medical Association and with the approval of the national level ethics committee (Hungarian Scientific and Research Ethics Committee of the Medical Research Council, ETTTUKEB-7214-1/2016/EKU), which waived the need for individual informed consent for this retrospective study. The Ethics Committee of Istituto Oncologico Veneto in Padova (Italy) evaluated and approved the study and patients provided informed consent to be signed for the collection, analysis and publication of data, when the investigators have the possibility to collect it from patients, according to Italian data protection authority dispositions. The ethics Committee of the Department of Health in Ugra (Russia) has evaluated and approved the study, which waived the need for individual informed consent for this retrospective study. Clinical information was collected and patients were de-identified. Subsequently, patients cannot be identified either directly or indirectly.

Study population

In this retrospective study, surgically resected and histologically confirmed consecutive SCLC patients evaluated at three centers, between 2000 and 2013 at the National Koranyi Institute of Pulmonology (cohort #1), between 2000 and 2013 at Università degli Studi di Padova (cohort #2), and between 1999 and 2013 at the Surgut District Clinical Hospital (cohort #3) were included (Fig. 1).

Data from all three institutions were measured and graded with comparable ways across the different institutions.

Histologic diagnosis was performed according to the same contemporary pathology guidelines across all three centers. Clinicopathological data collected included gender, age, smoking history, chemo- (CHT) and radiation therapy (RT) treatments, type of operation (segmentectomy, lobectomy, and pneumonectomy), and overall survival (OS) and/or disease free survival (DFS). TNM stage according to the Union for International Cancer Control (7th edition) (Mirsadraee et al., 2012), and patient age at the time of diagnosis was recorded. OS was estimated from the time of diagnosis, until death, or the last available follow-up visit. The study and all treatments were conducted based on the individual institutional guidelines in accordance with the current National Comprehensive Cancer Network guidelines with no major differences across the three centers.

Cut-off for the date of the last follow-up visit included in this analysis was set to April 2017. Blood samples were obtained from routine hematologic control using the same reference values across the three centers. Most studies examining NLR and PLR did not explicitly state the time frame of blood sample analysis prior to surgery, thus we set the time point to be within 30 days prior to surgery, so that the data could be broadly applicable to clinical practice. The NLR and PLR were calculated according to the absolute neutrophil and platelet counts and absolute lymphocyte counts on routine blood tests obtained within 30 days prior to surgery. NLR values were obtained by dividing the absolute neutrophil count by the absolute lymphocyte count, while PLR values were obtained by dividing the absolute platelet count by the absolute lymphocyte count. For patients who received neoadjuvant chemotherapy, blood tests prior to initiation of neoadjuvant chemotherapy were analyzed.

Based on the relatively low case numbers in each individual patients’ cohort and the lack of significant differences in OS and DFS according to cohorts (p = 0.360 and p = 0.744, respectively), we pooled cohorts #1–3 together. There were no significant differences in gender or stage across the three individual cohorts, however, there were significant differences in age, CHT and RT administration (Tables S1 and S2).

Treatment

Patients who underwent complete tumor resection including segmentectomy, lobectomy, and pneumonectomy with mediastinal lymph node dissection were included in this retrospective analysis. Selected patients were treated with postoperative systemic CHT with a platinum-etoposide doublet regimen or with a combination of cyclophosphamide, epirubicin, and vincristine. Selected patients with positive mediastinal lymph node status received preoperative CHT with the aforementioned regimens.

Statistical methods

Statistical analyses were performed only on patients with available PLR and NLR data (n = 155). Since there were no optimal cut-off values reported in resected limited-stage SCLC for NLR and PLR, time-dependent receiver operating curve (ROC) analysis was used to identify the optimal cut-off values for these parameters according to OS and DFS. ROC was applied to determine the optimal cut-off, as the values whose sensitivity and specificity are the closest to the value of the area under the ROC curve and the absolute value of the difference between the sensitivity and specificity values is minimal, as described in other studies (Coffelt et al., 2010; DeNardo, Andreu & Coussens, 2010). Next, patients were divided into “high” and “low” groups according to their preoperative NLR and PLR.

Kaplan–Meier curves and two-sided log-rank tests were used for univariate survival analysis. Two-sided p-values less than 0.05 were considered statistically significant. All variables with p-values less than 0.05 were included in the multivariate analysis. The Cox proportional hazards model was used for univariate and multivariate survival analyses to calculate the hazard ratios (HR) and corresponding 95% confidence intervals (CI). In the univariate analysis, we included gender, age, smoking status, preoperative CHT, adjuvant CHT, thoracic RT, prophylactic cranial irradiation (PCI), stage, and type of operation. For multivariate survival analyses, the Cox regression model was adjusted for significant variables in the univariate analysis. Metric data are shown as median or mean and corresponding range or, in case of OS, as median and corresponding 95% CI. Clinical characteristics of patients with low PLR (LPLR) vs. (HPLR) or low NLR (LNLR) (vs. HNLR) were analyzed by Chi-square test in each individual cohort and then inter-cohort heterogeneity was assessed with the same test. Age (<65 years vs. ≥65 years) was considered as a categorical variable. To address the problem of multiple comparisons between clinicopathological parameters of the individual cohorts, Bonferroni’s correction was applied. Thus, with six comparisons comprising three cohorts according to LNLR (vs. HNLR) or LPLR (vs. HPLR), p-values less than 0.008333 were considered to indicate statistical significance. All p-values were two-sided. Next, we evaluated the associations of preoperative NLR or PLR with clinicopathological characteristics (age, gender, stage, vascular involvement, tumor necrosis, peritumoral inflammation, and tumor grade) using Spearman’s rank correlation which assess linear correlation coefficient as it measures the strength of linear relationship between the variables. The value of linear correlation coefficient (r) varies from −1 to 1 both values inclusive. No linear correlation (r = 0), weak positive linear correlation (0.2 < r ≤ 0.5), moderate positive linear correlation (0.5 < r ≤ 0.8), strong positive linear correlation (0.8 < r <1), a p-value of less than 0.05 was considered to indicate statistical significance. All statistical analyses were performed using the PASW Statistics 22.0 package (SPSS Inc., Chicago, IL, USA).

Clinicopathological characteristics

Initially, we identified a total of 189 patients that underwent surgical resection: 91, 54, and 44 according to cohorts # 1, #2, and #3, respectively. PLR and NLR data were available in n = 155 cases (Fig. 1). Major clinicopathological characteristics of resected SCLC patients in pooled cohorts #1–3 (n = 155) are shown according to PLR and NLR in Table 1. The median age at diagnosis was 58 years. According to AJCC tumor staging, there were 60, 39, and 40 patients with Stage I, II, and III; respectively (16 patients had inaccurate staging data). During the follow-up period, 100 patients received adjuvant chemotherapy. The median preoperative neutrophil, lymphocyte, and thrombocyte counts were 4.621, 2.115, and 241.00; respectively, while NLR and PLR median values were 2.214 and 111.489. The identified cut-off PLR and NLR values for OS were 111.253 (sensitivity: 0.566, specificity: 0.589; Fig. S1A) and 2.258 (sensitivity: 0.545, specificity: 0.661; Fig. S1B), respectively. For DFS, PLR, and NLR value cut-offs of 112.174 (sensitivity: 0.531, specificity: 0.614; Fig. S1C) and 2.254 (sensitivity: 0.551, specificity: 0.649; Fig. S1D) were used, respectively.

Clinicopathological characteristics of resected SCLC patients in cohorts #1, #2, and #3 (n = 189) are shown according to NLR (Table S1) and PLR (Table S2). The 44 cases without NLR/PLR vs. 155 with NLR/PLR have no differences in clinicopathological characteristics. Nevertheless, we found significant differences only in pathological stage according to PLR and NLR in the full cohort. A higher percentage of stage I patients had pre-treatment LNLR.

Prognostic factors for OS and DFS

Univariate survival analysis identified significantly longer OS in patients with lobectomy and segmentectomy (vs. pneumonectomy, median OS, 56.2 vs. 31.7 months, p = 0.015, Fig. 2A), and also in patients with pathological N (pN) stage 0–1 (vs. pN 2, median OS, 64.9 vs. 30.3 months, p = 0.001; Fig. 2C). Moreover, OS was significantly longer in patients with LNLR (vs. HLNR, median OS, 74.8 vs. 44.5 months, respectively, p = 0.003, Fig. 3A), however there was no significant difference in OS according to PLR (LPLR vs. HPLR, median OS, 73.6 vs. 40.4 months, respectively, p = 0.084, Fig. 3C).

Univariate survival analysis also identified a significantly longer DFS in patients with lobectomy and segmentectomy (vs. pneumonectomy, median OS, 55.2 vs. 18.8 months, p = 0.033; Fig. 2B) and in patients with pN stage 0–1 (vs. pN 2, median OS, 55.7 vs. 23.0 months, p = 0.002; Fig. 2D), but there were no significant differences in DFS in patients with LNLR (vs. HNLR median OS, 68.8 vs. 34.9 months, p = 0.051, Fig. 3B) or LPLR (vs. HPLR, median OS, 68.8 vs. 32.8 months, p = 0.086, Fig. 3D).

The Cox multivariate analysis found that pN stage 0–1 (vs. pN 2, HR, 2.07; 95% CI [1.237–3.448]; p = 0.006) is a significant independent prognostic factor for OS, while the type of operation (lobectomy and segmentectomy vs. pneumonectomy, HR, 1.61; 95% CI [0.977–2.666]; p = 0.062) and LNLR (vs. HLNR, HR, 1.36; 95% CI [0.879–2.103]; p = 0.167) had no prognostic impact on OS. Furthermore pN stage 0–1 (vs. pN 2, HR, 2.041; 95% CI [1.241–3.357]; p = 0.005) had a prognostic impact also on DFS while the type of operation (lobectomy and segmentectomy vs. pneumonectomy, HR, 1.48; 95% CI [0.904–2.445]; p = 0.119), and LNLR (vs. HLNR, HR, 1.27; 95% CI [0.833–1.939]; p = 0.226) had no predictive value. Since the prognosis is significantly different for pN2 patients, we performed a univariate and a Cox multivariate analysis including only stage I and stage II patients, and we found that LNLR (vs. HNLR, HR, 1.582; 95% CI [1.010–2.478]; p = 0.045) is a significant prognostic factor for OS, as shown in Table 2. However, the type of operation (lobectomy and segmentectomy vs. pneumonectomy, HR, 1.697; 95% CI [0.974–2.958]; p = 0.062) had no prognostic value for stage I and stage II patients.

The effects of adjuvant CHT on OS according to NLR and PLR are shown in Fig. S2. There were no significant differences in OS of patients with HNLR treated with adjuvant CHT compared to those did not receive adjuvant CHT (median OS, 53.2 vs. 40.4 months, respectively, p = 0.965, log-rank test, Fig. S2A). In contrast, OS of patients with LNLR received adjuvant CHT was significantly longer compared to those did not receive adjuvant CHT (median OS, 113.3 vs. 31.7 months, respectively, p = 0.013, log-rank test, Fig. S2B). We found no significant differences in OS of patients with HPLR that did not receive adjuvant CHT compared to those received adjuvant CHT (median OS, 25 vs. 61.3 months, respectively, p = 0.125, log-rank test, Fig. S2C). Moreover, there were no significant differences in OS of LPLR patients that did not receive adjuvant CHT compared to those treated with adjuvant CHT (median OS, 53.2 vs. 80 months, respectively, p = 0.481, log-rank test, Fig. S2D).

Next, we investigated the associations between preoperative NLR or PLR with clinicopathological characteristics and Spearman’s correlation coefficient was calculated (Table 3). We found a statistically significant weak positive correlation between NLR and tumor stage (r = 0.226, p = 0.007), NLR and age (r = 0.167, p = 0.038), and NLR and PLR (r = 0.035, p < 0.0001). Of note, based on the limited case numbers in each individual cohort, we were not able to use ROC curves to identify individual cut-off values for combined analysis.