Lung cancer is the leading cause of cancer death worldwide as well as in Taiwan. Interleukin-6 (IL-6) is a multifunctional cytokine and has been implicated in tumor progression. This study recruited 245 patients with advanced (Stage 3B/4) nonsmall cell lung cancer (NSCLC) that had received chemotherapy, to evaluate associations between IL-6 and lung cancer-specific survival. Among these subjects, 112 gave blood samples before and 133 after the start of chemotherapy. Plasma IL-6 was measured using an enzyme linked-immunosorbent assay. The 33rd and 66th percentiles of IL-6 concentrations were 2.01 and 25.16 for the 245 patients and were defined as the cutoff points for dividing the patients into low, intermediate and high groups. Kaplan-Meier and Cox proportional-hazard models were used to evaluate the relationship between the IL-6 level and survival time. Results after adjusting for age, sex, smoking history, histologic type and stage of lung cancer revealed a significant relationship. For all patients, the hazard ratio with high IL-6 levels for lung cancer-specific survival was 2.10 [95% confidence interval (CI) = 1.49 − 2.96] compared with low IL-6 levels. The hazard ratio for patients who were recruited before and after the start of chemotherapy was1.25 (95% CI = 0.73 − 2.13) and 3.66 (95% CI = 2.18 − 6.15), respectively. Patients with high circulating IL-6 also responded poorly to chemotherapy. Therefore, a high level of circulating IL-6 was associated with an inferior response and survival outcome in NSCLC patients treated with chemotherapy.
Lung cancer is a major cause of death worldwide.1 In Taiwan, lung cancer ranks third for males and fourth for females in terms of cancer incidence.2 For mortality, however, lung cancer is the leading cause of cancer death among women and the second-leading cause of cancer death among men.3 This reflects poor survival with lung cancer, which is largely due to poor detection of lung cancer in its early stages and ineffective treatment for advanced stages. Only around 12.5% of lung cancer patients survive more than 5 years after diagnosis in Taiwan.2 In past years, investigators have identified genetic markers to predict risk or select therapeutic modalities for lung cancer patients, including a combined 15-gene signature,4 ribonucleotide reductase M1 (RRM1) gene expression,5 epidermal growth factor receptor (EGFR) gene mutation6, 7 and the Kirsten rat sarcoma viral oncogene homolog (KRAS) gene mutation.8, 9 However, other studies have focused on identification of potential genetic markers, such as the gene expression of a combined 5-gene signature,10RRM1 and ERCC1,11 and phosphatase and tensin homologue (PTEN),12 as prognostic factors for patients with lung cancer. The promising leads from these studies, however, have not yet been widely used because the genetic tests require special laboratory support.
Circulating biomarkers have been used as prognostic markers for survival in cancer patients.13 Such markers are easily measured and are able to characterize a subgroup of patients and provide biological targets for therapy.14 Interleukin-6 (IL-6), a pleiotropic cytokine, is one such marker.15 It is produced by various types of normal and cancer cells and is implicated in the stimulation of tumor cell proliferation, malignant transformation and tumor progression.16
In previous studies, an increased level of circulating IL-6 was found to be associated with shorter survival in patients with renal cell carcinoma,17 chronic lymphocytic leukemia,18 gastric carcinoma,19 prostate cancer,20 gastrointestinal cancer21 and breast cancer.22 In lung cancer patients, a high circulating IL-6 level was also reported by two groups of investigators to indicate shorter survival.23, 24 De Vita et al.25 found that patients with metastatic diseases had higher serum IL-6 levels and patients who responded to chemotherapy had lower serum IL-6 levels. However, Yanagawa et al.26 found that serum IL-6 levels were not significantly different in lung cancer patients with or without metastasis. The discrepancy may be due the heterogeneity of the patients and/or the small sample size of the studies. Besides, the numbers of patients in the four aforementioned studies23–26 were too small to control for potential confounders, including age, sex, smoking history, histologic type and stage of disease.
Chemotherapeutic agents have been shown to induce IL-6 production27, 28; hence, it would be important to know whether IL-6 levels can be used as independent prognostic factors for lung cancer survival, especially for patients who have been treated with chemotherapy. In this study, we use data from a large sample size to evaluate the association between circulating IL-6 levels and lung-cancer-specific survival in patients with advanced nonsmall cell lung cancer (NSCLC). We have also adjusted the survival data for several potential confounders, especially for patients who had taken chemotherapy, to obtain a reliable conclusion.
Material and Methods
The Genetic Epidemiological Study of Lung Adenocarcinoma (GELAC) was constructed from September 2002 to April 2011 in Taiwan. Details of the GELAC study have been described previously.29 In brief, lung cancer patients were recruited from six medical centers in northern, central and southern Taiwan. New-onset/prevalent patients with all histological types of lung cancer were enrolled, with the aim of comparing the clinical and etiologic differences between adenocarcinoma and other lung carcinomas. Patients aged 18 years or older with histologically or cytologically confirmed lung cancer were enrolled in this study. Written informed consent was obtained from each study subject. The GELAC study was approved by the institutional review boards of all participating institutes and medical centers (the Research Ethics Committee of the National Health Research Institutes, EC0970303-R2).
A blood sample (30 mL) and questionnaire information were collected from each participant. The structured questionnaires were administered in face-to-face interviews conducted by well-trained nurses. The information requested in the questionnaire included sociodemographic characteristics and history of exposure to various risk factors for lung cancer. The database was established at the National Health Research Institutes in Taiwan through a quality control and double-entry process.
Participants in this study for analysis were patients with lung cancer recruited from September 2002 through September 2004. Patients who declined to participate (N = 259) or had no histological/cytological confirmation (N = 32) were excluded; 950 eligible patients were ultimately included. The sample ascertainment protocol was not strictly limited to patients with new-onset lung cancer, so each patient was labeled with two dates: the date of onset from the medical chart and the date of recruitment. Their blood samples were drawn and questionnaires obtained at the date of recruitment. To test more rigorously for a possible temporal relationship between IL-6 and survival time, only those patients with an interval of <3 months from the date of onset of lung cancer to the date of recruitment were included (N = 514). Since small cell and large cell types have the worst prognosis in lung cancer, two large cell and 55 small cell lung cancer patients were excluded. Patients with early stage (N = 52) and stage 3A (N = 48) were excluded, as were 112 patients whose date of chemotherapy could not be confirmed by chart review. In the end, 245 patients were included in the following analyses (167 with adenocarcinoma, 78 with squamous cell or poorly differentiated cell types). The sixth edition of the American Joint Committee on Cancer (AJCC) staging system was used for the tumor node metastasis (TNM) staging of all patients.
Enzyme-Linked Immunoabsorbent Assay for IL-6
For each volunteer patient, 30 mL of blood was collected in the yellow-topped vacutainer tube with acid citric dextrose (ACD) additive. These vacutainers were sent to the processing laboratory at the coordination center within 24–48 hr after collection and maintained cold chain shipping, followed by centrifuge these vacutainers at 3000 rpm for 15 min at 4°C and then aliquot and stored the plasma cryovials at −80°C until to be used. In this study, the samples were obtained from August 2002 to March 2004 and tested on March 2005. The average storage time was 21.9 months (range 12–31months). No sample was thawed before testing.
Plasma levels of IL-6 were assessed by enzyme-linked immunosorbent assay (ELISA) using the Human Interleukin-6 (hIL-6) ELISA kit (BioSource International, Camarillo, CA). For the sensitivity of Human IL-6 (Hu IL-6), the minimum detectable dose is <2 pg/mL. For the specificity, the following substances were tested and found to have no cross-reactivity: human IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-7, IL-8, IL-10, G-CSF, GM-CSF, IFN-α, IFN-γ, LIF, MIP-1α, MIP-1β, MCP-1, OSM, RANTES TGF-β, TNF-α and TNF-β. For the reproducibility, samples of known Human IL-6 concentration were assayed in replicates of 16 to determine precision within an assay, and samples were assayed eight times in five different assays to determine precision between assays. The %CV is range from 5.1 to 9.3% (as measured by BioSource international, ELISA assays).
ELISAs were performed according to the manufacturer's protocol. According to the protocol of ELISA assay kit, this assay has been calibrated against the WHO reference preparation 89/548 (NIBSC, Hertfordshire, UK, EN6 3QG). One picogram equals 100 mIU. The assays were standardized by the Hu IL-6 standard provided by ELISA kit and run within one month by the same lot of ELISA kits. The absorbance of each well was read at 450 nm by an ELISA reader (SPECTRAmax 340PC384 and SOFTmax PRO software, Molecular Devices, CA) immediately or after 5–10 min; then, the concentrations of unknown samples were estimated from the standard curve. Standard were assayed in replicates between assays and the average %CV was 13.13% in this study, slightly higher than measures by BioSource international.
Computerized tomography (CT) scanning of the chest including liver was performed before and every 3 months during treatment or after completion of chemotherapy, and every 3–6 months thereafter until disease progression or the initiation of subsequent anticancer therapy. Chest X-rays were repeated monthly between each chest CT scan evaluation for the assessment of disease progression. At the clinicians' discretion, radiologic tumor assessment could be repeated early on the basis of clinical need or suspicion of disease progression. The assessment strategies were the same for those who completed therapy and those who did not. The sum of maximal diameters of all measurable tumor lesions was recorded at baseline and after treatment. These measurements were used to calculate the largest percentage of shrinkage or smallest percentage of growth using the baseline assessment as reference. The best response to therapy was categorized according to RECIST 1.0 criteria.30
Lung cancer-specific survival was calculated from the date of diagnosis to the date of death from lung cancer. Dates of death were ascertained through computerized data linked with Taiwan's National Death Certificate Registry from January 1, 2002 to December 31, 2008. Good agreement between the death certificates and coders was found for malignant neoplasms (κ = 0.94).31 Of the 245 lung cancer patients, 220 deaths were due to lung cancer. Patients who died from causes other than lung cancer (N = 14) or who were alive on December 31, 2008 (N = 11), were treated as censored cases, not as survival cases.
Patients were trichotomized based on tertiles of plasma IL-6 levels, where each subject was classified as “low IL-6 level,” “intermediate IL-6 level” or “high IL-6 level.” The tertile cut-off points were 2.01 and 25.16 pg/mL for all patients and were used for the following subgroup analyses. Cumulative cigarette smoking was divided into two levels: never-smokers and ever-smokers (those who had smoked regularly for more than 6 months during their lifetime). Histologic types of lung cancer were dichotomized as adenocarcinoma or squamous/poorly differentiated and Stage 3B or Stage 4, respectively. The baseline characteristics and clinical information of patients were compared across three IL-6 groups, using the Pearson x2 test (for categorical variables), one-way ANOVA (for continuous variables) and the Kruskal-Wallis test (nonparametric version of one-way ANOVA), as appropriate.
All 245 patients were classified into two groups according to whether their date of recruitment was before or after the first day of beginning chemotherapy. The relationships between lung cancer-specific survival and the three IL-6 groups were modeled using the Kaplan-Meier method. Further stratification analyses were conducted by histological type (adenocarcinoma and squamous/poorly differentiated) and chemotherapy status (before and after). Differences in the survival curves between groups were estimated using the log-rank test. In controlling for covariates, Cox proportional hazards regression models were used to evaluate IL-6 levels and lung cancer survival while adjusting for age, sex, smoking history, histological type and clinical stage. The above-mentioned analyses were conducted using SPSS software version 19.0 (SPSS, Chicago, IL). All statistical testing was two-sided, and p values < 0.05 were considered statistically significant.
The mean age of the Stage 3B/4 lung cancer patients in the three groups classified by IL-6 concentration was 62.7, 62.1 and 61.6 years, respectively. The median lung cancer-specific survival time decreased with the increase in the level of IL-6 (p < 0.0001) and was about twice the survival time of the low IL-6 level (17.9 months) compared with the high IL-6 level (9.2 months) groups (Table 1). Those with higher IL-6 levels tended to be males, ever-smokers and with adenocarcinoma and Stage 4, though only differences in cell type and stage achieved statistical significance.
Table 1. Baseline characteristics of lung cancer patients as stratified according to interleukin-6 level1
Results of Kaplan-Meier survival analyses and log-rank tests are shown in Figures 1 and 2. For all lung cancer patients (N = 245), there was a statistically significant difference in lung cancer-specific survival across the three IL-6 level groups (p < 0.0001; Fig. 1a). Such significant differences of survival time were also seen in the adenocarcinoma (N = 167) and squamous/poorly differentiated (N = 78) subtypes (p = 0.0002 and p = 0.04; Figs. 1b and 1c). In stratification analysis by chemotherapy status, the circulating IL-6 level in patients with the date of recruitment after the first day of chemotherapy correlated well with survival time (p < 0.0001; Fig. 2b). However, the influence of circulating IL-6 levels on survival time in patients with the date of recruitment before the first day of chemotherapy (Fig. 2a) did not reach statistical significance.
To test whether the relationship between survival time and IL-6 levels still held after adjusting for potential confounders, Cox proportional hazards regression models were used, with results for all patients and by histological type (Table 2) and chemotherapy status (Table 3) presented. Among all patients, individuals with a high circulating IL-6 level had a statistically worse survival outcome, with a hazard ratio of 2.10 [95% confidence interval (CI), 1.49–2.96], than those with a low IL-6 level (Table 2). Stage 4 was slightly associated with lung cancer-specific survival as well, compared with Stage 3B (hazard ratio = 1.41; 95% CI = 0.99 − 2.01). After stratifying by histological type, similarly, patients with a high circulating IL-6 level had worse survival in both subtypes (hazard ratio = 2.16 in adenocarcinoma and 2.96 in squamous/poorly differentiated). However, the stage effect was significantly associated with lung cancer-specific survival for squamous/poorly differentiated cell (hazard ratio = 2.48; 95% CI = 1.38 − 5.19), but not for adenocarcinoma (hazard ratio = 1.06; 95% CI = 0.67 − 1.67). When stratified by chemotherapy status (Table 3), the effect of circulating IL-6 levels in predicting survival time was significant in patients with the date of recruitment after the first day of chemotherapy, but not in patients that were recruited before chemotherapy, after controlling for the same covariates (hazard ratio = 3.66; 95% CI = 2.18 − 6.15). The hazard ratio (3.66) for patients recruited after chemotherapy was greater than that for all patients (2.10).
Table 2. Multivariate analysis of lung cancer-specific survival for all patients and stratified by histologic type of lung cancer1
Table 3. Multivariate analysis of lung cancer-specific survival stratified by the date of recruitment1
In addition to survival data, we also examined the associations between clinical responses to chemotherapy and IL-6 plasma levels among 133 NSCLC patients (Table 4). We observed a higher proportion of progression in Group 2 (intermediate IL-6 level, 20.5%) and Group 3 (high IL-6 level, 45.7%) patients, compared with Group 1 (low IL-6 Level, 18.8%) patients; however, the statistical difference only existed between Group 1 and Group 3 (p-value = 0.02). Therefore, circulating IL-6 level might have a potential to predict response to chemotherapy.
Table 4. Response to advanced nonsmall cell lung cancer patients treated with chemotherapy as stratified according to interleukin-6 level
The elevated circulating IL-6 level has been associated with decreased cancer survival in several studies.17–24 To the best of our knowledge, ours is the first study investigating the association between circulating IL-6 levels and survival in NSCLC patients, adjusting for potential confounders, especially chemotherapy. With a large sample size, the study has stronger statistical power than others in detecting this association. Through stratification and regression model fitting analyses, adjusting for potential confounders, our results demonstrated that the circulating IL-6 level is a prognostic factor for lung cancer-specific survival in advanced NSCLC patients treated with chemotherapy.
The underlying biological mechanism of the observed effect of the plasma IL-6 level on lung cancer survival remains to be elucidated. Circulating IL-6 might be secreted from immune and stromal cells in response to tumor progression32 or from cancer cells per se33 or both. A high circulating IL-6 level in turn might facilitate tumor cell proliferation and immune invasion.34
In previous studies,35–37 age, gender and smoking history have been suggested to be associated with circulating levels of IL-6. Consistent with the report,35–37 we also observed an increased incidence of high circulating IL-6 levels in male patients and in patients with a smoking history (Table 1). However, results from the Cox regression models for all patients showed a strong relationship between IL-6 and survival time, even when adjusting for age, gender and smoking history (Table 2).
To examine further, whether the influence of circulating IL-6 levels on the survival of lung cancer patients is modulated by histological types, we stratified lung cancer into adenocarcinoma (less related to smoking) and squamous/poorly differentiated cell carcinoma (smoking-related). In both subgroups of patients, we found that levels of circulating IL-6 were still significantly associated with lung cancer-specific survival, which is independent of the smoking effect (Table 2).
When stratified by stage, we found the stage effect was significantly associated with lung cancer-specific survival for squamous/poorly differentiated cell lung cancer (hazard ratio = 2.48; 95% CI = 1.38 − 5.19), but not for adenocarcinoma (hazard ratio = 1.06; 95% CI = 0.67 − 1.67; Table 2). The most likely reason for the observation is that diseases associated with malignant pleural effusion (MPE) were classified as Stage IIIB in the sixth edition of the AJCC staging system and lung adenocarcinoma is the major histological subtype of lung cancers that are associated with MPE. Therefore, a significant number of lung adenocarcinoma patients are understaged using the sixth edition of the AJCC staging system. This drawback has been corrected in the seventh edition of the AJCC staging system.
IL-6 has been implicated in the development of tumor drug resistance.38 Wang et al.27 reported that paclitaxel may upregulate IL-6 expression in human ovarian cancer cells. Poth et al.28 showed that cisplatin treatment induced IL-6 expression in head and neck squamous cell carcinoma and that cisplatin-induced IL-6 expression can contribute to the increase in the tumorigenic potential of cancer cells. In this study, we found that the ability to predict survival was better for patients whose blood samples were obtained after the start of chemotherapy than for those who gave their samples before chemotherapy (Table 3 and Fig. 2). One explanation for the phenomena is that the circulating IL-6 levels of the non-responders to chemotherapy remain high after chemotherapy. Two small studies39, 40 that measured serial serum IL-6 levels in patients with NSCLC before and after chemotherapy demonstrated a decrease in the IL-6 serum level in responders, but not in nonresponders. The results hold with our assumption. The other possibility is that patients whose circulating IL-6 levels had been induced by chemotherapy suffered a poor survival outcome. The observation that tumor-adjacent cells can respond to genotoxic stress by secreting IL-6 and Timp1 to overcome cytotoxicity from doxorubicin41 supports this possibility. Our study, however, is flawed by its retrospective study design. A prospective study designed to collect blood samples serially in an adequate sample size of patients is warranted to clarify whether the postchemotherapy circulating IL-6 level is an independent biomarker to predict the survival of patients with NSCLC.
It is note that this study had limitation about the stability of IL-6 in stored plasma. Because the study was a nationwide cooperation project in Taiwan, the blood samples were shipped by transportation companies with low-temperature facilities to the central laboratory in the National Health Research Institutes, Taiwan. The sample quality, shipping process, long-term storage and sample thawed might result in decrease of measured IL-6 concentration.42, 43 However, we have carried out the whole procedures carefully, all the samples were measured using same batch of ELISA kits within 1 month, and there was no sample been thawed before testing. Therefore, the data presented here are reliable.
In summary, the results of this study indicate that a high circulating IL-6 level might predict inferior response to chemotherapy and is an independent prognostic marker for lung cancer-specific survival, especially for those who received chemotherapy. A prospective study with serial blood sample collection in a large sample size of patients is needed to confirm the results.
The authors acknowledge statistical assistance provided by the National Translational Medicine and Clinical Trial Resource Center.