Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Nov 1.
Published in final edited form as: Cancer Prev Res (Phila). 2018 Sep 25;11(11):727–734. doi: 10.1158/1940-6207.CAPR-18-0111

Methylation-derived neutrophil-to-lymphocyte ratio and lung cancer risk in heavy smokers

Laurie Grieshober a, Stefan Graw b,c, Matt J Barnett d, Mark D Thornquist d, Gary E Goodman d, Chu Chen d,e,f, Devin C Koestler b,c,#, Carmen J Marsit g,#, Jennifer A Doherty a,d,#
PMCID: PMC6214718  NIHMSID: NIHMS1508256  PMID: 30254071

Abstract

The neutrophil-to-lymphocyte ratio (NLR) is a biomarker that indicates systemic inflammation and can be estimated using array-based DNA methylation data as methylation-derived NLR (mdNLR). We assessed the relationship between pre-diagnosis mdNLR and lung cancer risk in a nested case-control study in the β-Carotene and Retinol Efficacy Trial (CARET) of individuals at high risk for lung cancer due to heavy smoking or substantial occupational asbestos exposure. We matched 319 incident lung cancer cases to controls based on age at blood draw, smoking, sex, race, asbestos, enrollment year, and time at risk. We computed mdNLR using the ratio of predicted granulocyte and lymphocyte proportions derived from DNA methylation signatures in whole blood collected prior to diagnosis (median 4.4 years in cases). Mean mdNLR was higher in cases than controls (2.06 vs 1.86, p=0.03). Conditional logistic regression models adjusted for potential confounders revealed a 21% increased risk of lung cancer per unit increase in mdNLR (Odds Ratio (OR) 1.21, 95% Confidence Interval (CI) 1.01–1.45). A 30% increased risk of non-small cell lung cancer (NSCLC) was observed for each unit increase in mdNLR (n=240 pairs; OR 1.30, 95% CI 1.03–1.63), and there was no statistically significant association between mdNLR and small cell lung cancer risk. The mdNLR-NSCLC association was most pronounced in those with asbestos exposure (n=42 male pairs; OR 3.39, 95% CI 1.32–8.67). A better understanding of the role of mdNLR in lung cancer etiology may improve prevention and detection of lung cancer.

Introduction

Inflammation is an established characteristic and hallmark of cancer development and progression (1). White blood cell distributions, such as increased neutrophil counts and decreased lymphocyte counts, are markers of systemic inflammation that are associated with morbidity and mortality in cancer patients (2,3). The neutrophil-to-lymphocyte ratio (NLR) is computed as the ratio of the respective whole-blood proportions of neutrophils and lymphocytes, and therefore reflects the balance between the innate immune response and adaptive immune response (4). Elevated pre-treatment NLR is an emerging biomarker of cancer prognosis, as it has been associated with worse outcomes for various tumor types including lung cancer (2,3,5–13). Investigations of NLR and cancer risk are limited (14–16), and only one study to date has evaluated the association between NLR and lung cancer risk (15).

White blood cell distributions are easily determined from peripheral blood in routine clinical care; however, the assessment of these parameters becomes unstable in stored blood after as little as 24 hours (17), which makes their examination in biorepository collections challenging. Blood cells originate from hematopoietic stem cells, and the lineage of each blood cell is dependent on epigenetic differentiation signatures including methylation (18). Obtaining blood cell distribution estimates from methylation data measured in archival blood using deconvolution algorithms has been described and validated (19–22). Recent work has highlighted the utility of these methods for estimating NLR, denoted methylation-derived NLR (mdNLR), and in utilizing these estimates to examine associations with cancer risk and outcomes (16,21,23). The lifetime probability that a smoker will develop lung cancer is estimated to be <15% (24). Thus, there is a great need for biomarkers to differentiate between smokers who will and will not develop lung cancer, especially in the context of lung cancer screening, which generates a large proportion of false positives. We evaluated whether increased levels of inflammation as indicated by higher mdNLR values are associated with lung cancer risk in a case-control study nested in the β-Carotene and Retinol Efficacy Trial (CARET) of heavy smokers at high risk for lung cancer. The CARET population is primarily composed of individuals who would have been eligible for lung cancer screening at blood draw based on the current guidelines from various organizations including the U.S. Preventive Task Force (e.g., ages 55–80, 30+ pack year smoking history, current smokers or former smokers who quit less than 15 years ago) (25).

Materials and methods

Study population

Participants in this study were drawn from the multicenter β-Carotene and Retinol Efficacy Trial (CARET) (26). CARET was a randomized, double-blinded, placebo-controlled chemoprevention trial designed to assess the safety and efficacy of daily supplementation with β-carotene and retinyl palmitate among individuals at high risk of developing lung cancer (27–29). Men and women ages 50–69 years were eligible for the trial if they were current or former smokers (quit at most six years prior to enrollment) with a heavy cigarette smoking history of ≥20 pack-years (n=14,254). Additionally, men ages 45–69 years with occupational asbestos exposure who were current or former heavy smokers (quit at most fifteen years prior to enrollment) were included (n=4,060). Extensive information about smoking history and other risk factors was collected through annual questionnaires. Whole blood was collected after enrollment at participant visits between 1994 and 1997. Due to higher lung cancer incidence and overall mortality rates in the intervention versus placebo arm, the intervention was stopped in 1996.

The present study includes individuals selected for a nested case-control study based on endpoint information collected from 1985 to 2005, during active participant follow-up, as described previously (26). Briefly, all participants who provided a blood specimen and who were free of lung cancer at the time of blood collection were eligible to be matched in a 2:1 ratio to 793 lung cancer cases. Matching factors included age at enrollment (±4 years), sex, race/ethnicity, enrollment year (±2 years), smoking status at enrollment (current or former), history of occupational asbestos exposure, and length of follow-up. For our study, we selected those with known histotype to the extent possible (240 NSCLC, 68 SCLC, 11 unknown). We matched controls to the 319 cases in a 1:1 ratio using age at blood draw (±5 years), sex, race/ethnicity, enrollment year (±2 years), smoking status at blood draw (current or former), history of occupational asbestos exposure, and length of follow-up. The Institutional Review Boards for each of the participating CARET institutions approved all study protocols, which were in accordance with recognized ethical guidelines, and written informed consent was obtained from all participants.

Methylation array, Quality Control, Preprocessing and Normalization of Methylation Data

We assessed DNA methylation in whole blood using the Illumina HumanMethylation850 BeadArray platform at the University of Southern California Epigenomics Core Facility, following standardized protocols described by the manufacturer (Illumina, Inc). Quality control, preprocessing, and normalization of Illumina HumanMethylation850 BeadArray methylation data was performed using both minfi and wateRmelon Bioconductor packages (30,31). To ensure high-quality methylation data, we utilized the within-array normalization combination of Noob+β-mixture quantile (BMIQ), as this processing combination has been previously demonstrated to result in improved signal intensity compared to competing approaches (32). Within-array correction for background fluorescence and dye biases were performed using the Noob methodology via the function “preprocessNoob” in the minfi Bioconductor package (30). Following application of the Noob, we identified and removed poor quality samples and probes. Samples were excluded if >20% of their probes had detection p-values >1×10−5 or if they exhibited irregularities in the distribution of their control probes (33). In addition, probes with a median detection p-value >0.05 were removed and excluded from subsequent statistical analyses. After sample- and probe-level quality control, we corrected the type II probe bias to make the methylation distribution of type II features comparable to the distribution of type I features using the BMIQ dilation methodology (34), implemented using the function “BMIQ” in the wateRmelon Bioconductor package. For each sample, BMIQ involves fitting a three-class β-mixture model (class 1=un-methylated, class 2=hemi-methylated, and class 3=methylated probes) to the type I and type II probes separately. Probes are assigned to the class with the highest posterior predictive probability (i.e., probability of class membership conditional on the observed data). β-values for the type II features are then normalized by class to the distributions for the same class estimated in type I features. Following the Noob+BMIQ processing combination, the presence of batch-effects and/or other technical sources of variation were inspected by first performing a principal components analysis (PCA) on the normalized methylation data, followed by an examination of the association between the top K principal components (K determined using random matrix theory (35) and plate, beadchip, and chip position). We observed only mild associations between these technical features and the top K principal components, and concluded that a formal batch-effect correction was not required for these data.

Estimation of the methylation-derived neutrophil-to-lymphocyte ratio (mdNLR)

We used a recently developed method by Koestler et al. to estimate the mdNLR for each of the study samples from blood-derived DNA methylation data (16). We first applied reference-based cell-mixture deconvolution to estimate the proportion of leukocyte subtypes (i.e., CD4T, CD8T, natural killer cells, B cells, monocytes, and granulocytes) for each sample using the “estimateCellCounts” function in the minfi Bioconductor package. In total, 600 leukocyte discriminating cytosine-phosphate-guanine (CpG) sites were used as the basis for cell composition estimation (36). The mdNLR was computed for each sample by taking the ratio of its predicted granulocyte and lymphocyte fractions (the sum of the predicted fraction of lymphocyte subtypes) (16).

Of the 600 CpGs used to estimate cell composition (36), 96 CpGs overlapped with the 18,760 false discovery rate-significant CpGs identified in the 2016 Joehanes et al. meta-analysis of differential DNA methylation between current and never smokers (37). We performed permutation testing to assess whether there was significant enrichment of smoking-associated CpGs among those used to deconvolute cell composition. We also compared cell type proportion estimates, and the resultant mdNLR, obtained before (600 CpGs) and after removing the 96 smoking-associated CpGs (504 CpGs) in our population (n=638) by computing the Pearson correlations and root mean squared error (RMSE).

Statistical analysis

Given the nested case-control study design with 1:1 case-control matching, a series of conditional logistic regression models were used to examine the association between continuous mdNLR and lung cancer risk. Models were fit using the “clogit” function in the survival R package with the following matching factors: age at blood draw (±5 years), sex, race/ethnicity, enrollment year, smoking status at blood draw, history of occupational asbestos exposure, and length of follow-up, and were adjusted for potential confounding factors, including: age (continuous), pack years of smoking, cigarettes per day, and body mass index (BMI), all at blood draw. Adjustment for continuous age and smoking variables help to control for any existing residual confounding in the corresponding matching factors. We selected potential confounders based on biologic plausibility and the impact of addition to crude conditional logistic regression models, i.e. met or exceeded a 10% change in the mdNLR coefficient.

Conditional logistic regression models were fit to estimate the relationship between mdNLR and lung cancer risk overall, as well as the association between mdNLR and risk by lung cancer histotype (NSCLC, SCLC), race (white, non-white), sex (male, female), study (asbestos exposure, heavy smoking only), smoking status at blood draw (yes, no), and intervention arm (placebo, active). NSCLC cases were defined as those with a diagnosis of lung adenocarcinoma, squamous cell carcinoma, and not otherwise specified (NOS) NSCLC. Because the time between blood draw and diagnosis varied considerably across cases (0.1 years to 10.1 years, median=4.4 years), conditional logistic regression models stratified by the median length of time between blood draw and diagnosis (<4.4 years and ≥4.4 years) were fit as a means to gain insight into whether the association between mdNLR and risk varies by the proximity to the time of diagnosis.

Results

Characteristics of the cases (all, NSCLC, and SCLC) and controls are presented in Table 1. Briefly, both cases and controls were approximately 64.0 years of age at blood draw and the majority were white (97%), male (66%), and current smokers at blood draw (65%). Cases were generally diagnosed at late stage, with 73% presenting with stage III or IV lung cancer. Compared to controls, mean mdNLR was higher for cases (mean, standard deviation (SD) for controls versus cases: 1.86 (0.90) versus 2.06 (1.31), p=0.03) and was higher for NSCLC than SCLC (mean, SD: 2.07, 1.31 and 1.88, 1.21, respectively).

Table 1.

Participant characteristics

Controls
(n=319)		 Cases
Alla lung
cancers

(n=319)		 Non-small
cell lung
cancerb
(n=240)		 Small cell
lung
cancer
(n=68)
Matching variables
Age at blood draw, mean years (SD) 63.5 (5.7) 64.5 (5.6) 64.5 (5.5) 64.5 (5.8)
 45–54, n (%) 21 (7) 21 (7) 16 (7) 4 (6)
 55–59, n (%) 77 (24) 42 (13) 31 (13) 9 (13)
 60–64, n (%) 86 (27) 101 (32) 77 (32) 23 (34)
 65–74, n (%) 135 (42) 155 (49) 116 (48) 32 (47)
Sex, n (%) female 109 (34) 109 (34) 78 (33) 28 (41)
Enrollment year, n (%)
13 (4)
13 (4)
79 (25)
146 (46)
68 (21)
15 (5)
14 (4)
78 (24)
144 (45)
68 (21)
13 (5)
9 (4)
57 (24)
107 (45)
54 (22)
2 (3)
4 (6)
16 (24)
33 (49)
13 (19)
 1985–1986
 1987–1988
 1989–1990
 1991–1992
 1993–1994
Asbestos exposure, n (%) 53 (17) 52 (16) 42 (18) 9 (13)
Current smoker at blood draw, n (%) 205 (64) 206 (65) 155 (65) 41 (60)
Other characteristics at blood draw
Stage III/IVc, n (%) - 197 (73) 143 (67) 53 (95)
Cigarettes per day, lifetime mean (SD) 14.0 (13.3) 15.5 (15.0) 15.0 (14.7) 15.7 (15.7)
Pack-years, mean (SD) 53.8 (23.7) 59.3 (22.5) 58.5 (22.2) 60.1 (22.9)
Years since quit smoking, mean (SD) 2.4 (5.0) 2.3 (4.3) 2.4 (4.3) 2.5 (4.3)
BMI (kg/m2), mean (SD) 28.1 (5.6) 27.7 (4.9) 27.6 (4.8) 28.1 (5.4)
 <18.5 (%) 4 (1) 2 (1) 2 (1) --
 18.5–24.9 (%) 85 (27) 92 (29) 69 (29) 19 (28)
 25.0–29.9 (%) 130 (41) 136 (43) 101 (42) 30 (44)
 ≥30.0 (%) 100 (31) 89 (28) 68 (28) 19 (28)
Intervention arm, n (%) assigned to active 168 (53) 166 (52) 123 (51) 39 (57)
mdNLR at blood draw
Mean (SD) 1.86 (0.90) 2.06 (1.31) 2.07 (1.31) 1.88 (1.21)
Open in a new tab

Abbreviations: SD = Standard Deviation; BMI = Body Mass Index.

a

”All lung cancers” includes adenocarcinoma, squamous cell carcinoma, and small cell lung cancer as well as cases for whom histotype was not otherwise specified non-small cell lung cancer (NOS; n=9) or missing (n=11).

b

The “Non-small cell lung cancer” category includes adenocarcinoma, squamous cell carcinoma, and not otherwise specified non-small cell lung cancer (NOS; n=9).

c

Stage information was not available for 50 lung cancer cases (28 NSCLC, 12 SCLC, and 10 missing histotype).

Permutation-based testing indicated statistically significant enrichment of smoking-associated CpGs (37) among the 600 CpGs used to estimate white blood cell proportions (permutation p<0.0001) (36). However, cell type estimates for our population (n=638) obtained with and without the removal of the 96 smoking-associated CpGs were highly correlated (Pearson r≥0.99 across all cell types). The average difference between estimates obtained with and without the removal of smoking-associated CpGs was less than 2% in absolute percentage points (e.g., RMSE≤2 for all cell types) and less than 1% in 4 out of the 6 cell types. Importantly, mdNLR estimates were highly consistent (Pearson r=1.00, RMSE=0.11).

Higher mdNLR values were associated with risk of lung cancer, especially NSCLC, in our study. After controlling for age, pack years, cigarettes per day, and BMI, all at blood draw, each unit increase in mdNLR was associated with a 21% increased risk of lung cancer (95% CI: 1.01–1.44; Table 2). For NSCLC, we observed a 30% increased risk of lung cancer (95% CI: 1.03–1.63) for each unit increase in mdNLR. Associations with mdNLR were similar for adenocarcinoma and squamous cell carcinoma. We did not observe statistically significant associations between mdNLR and SCLC, which may have been due to a smaller sample size (68 pairs; Table 2).

Table 2.

Continuous methylation-derived neutrophil-to-lymphocyte ratio and lung cancer risk overall and by sex, asbestos exposure, smoking status, intervention arm, and median time at risk.

Alla lung cancers Non-small cell lung
cancer		 Small cell lung
cancer
ORb (95% CI) ORb (95% CI) ORb (95% CI)
Continuous mdNLR 1.21 (1.01, 1.44) 1.30 (1.03, 1.63) 1.06 (0.77, 1.47)
Sex
Male 1.30 (1.04, 1.63) 1.34 (1.02, 1.75) 1.24 (0.79, 1.93)
Female 1.05 (0.76, 1.45) 1.22 (0.75, 1.98) 0.94 (0.53, 1.66)
Asbestos exposure
Yes 2.13 (1.12, 4.05) 3.39 (1.32, 8.67) 1.27 (0.54, 2.95)
No 1.09 (0.90, 1.32) 1.15 (0.92, 1.45) 0.96 (0.61, 1.52)
Smoking status
Never/former 1.26 (0.98, 1.63) 1.24 (0.92, 1.68) 1.52 (0.80, 2.90)
Current 1.16 (0.91, 1.47) 1.40 (0.98, 2.00) 0.76 (0.44, 1.34)
Stage
Early (I/II) 1.30 (0.79, 2.13) 1.29 (0.79, 2.12) -- --
Late (III/IV) 1.35 (1.08, 1.70) 1.44 (1.08, 1.92) 1.19 (0.81, 1.74)
Intervention arm
Placebo 1.11 (0.78, 1.59) 1.32 (0.86, 2.02) 0.70 (0.26, 1.87)
Active 1.70 (1.12, 2.58) 1.53 (0.89, 2.62) 2.47 (0.79, 7.73)
Time at risk (median in cases = 4.4 years)
<4.4 years 1.25 (0.99, 1.58) 1.34 (0.99, 1.80) 1.20 (0.81, 1.76)
≥4.4 years 1.18 (0.89, 1.57) 1.27 (0.88, 1.84) 0.80 (0.37, 1.71)
Open in a new tab

Abbreviations: OR = odds ratio; CI = confidence interval; mdNLR = methylation-derived neutrophil-to-lymphocyte ratio; BMI = body mass index.

a

”All lung cancers” includes non-small cell lung cancer pairs (N=240) and small cell lung cancer pairs (N=68), as well as pairs for whom case histotype was missing (N=11).

b

Conditional logistic regression models adjusted for age, pack years, cigarettes per day, and BMI at blood draw.

Bold OR and CI indicate statistical significance.

Subgroup analyses revealed that lung cancer risk associations were most notable in males (OR, 95% CI: 1.30, 1.04–1.63), those with a history of asbestos exposure (OR, 95% CI: 2.13, 1.12–4.05), and those assigned to the trial’s active intervention arm (OR, 95% CI: 1.70, 1.12–2.58). As in the overall analyses, associations were larger when restricted to NSCLC pairs for males (OR, 95% CI: 1.34, 1.02–1.75) and asbestos exposure (OR, 95% CI: 3.39, 1.32–8.67), but the NSCLC association for the active intervention arm was not statistically significant (OR, 95% CI: 1.53, 0.89–2.62). We observed associations of similar magnitude for early (I/II) and late (III/IV) stage lung cancers, although the late stage odds ratios were statistically significant for overall and NSCLC (Table 2). Associations were similar though no longer statistically significant in strata of smoking status or time at risk.

Discussion

We observed that elevated pre-diagnosis mdNLR is associated with an increased risk of lung cancer in a population of heavy smokers at high risk for lung cancer. Our findings are compelling, as blood was collected prior to diagnosis (median 4.4 years) and lung cancer cases and controls share a history of heavy smoking. Smoking has been reported to be associated with blood methylation signatures (38–40) and immune profiles (41). We observed that the residual effects due to smoking on the 600 CpGs used for white blood cell deconvolution have a minimal effect on cell composition estimates and therefore on mdNLR estimation. Nevertheless, smokers tend to have higher NLR values than non-smokers (42) and both smoking and nicotine are reportedly immunosuppressive (43). Observing an association between mdNLR and lung cancer risk after rigorous control for smoking history through matching and model adjustment implies that there are other factors, intrinsic or extrinsic, that are likely contributing to the relationship between mdNLR and lung cancer risk. Identifying a biomarker for lung cancer risk among those at highest risk is a salient issue, as <25% of smokers will develop lung cancer during their lifetime assuming they survive competing causes of death, or <15% without conditioning on competing causes of death (24). Successfully distinguishing between those who ultimately will develop lung cancer among those eligible for screening could improve lung cancer screening by limiting false-positives and overdiagnosis, two of the major concerns about lung cancer screening (44).

The strongest subgroup association between mdNLR and lung cancer risk was observed in those with occupational asbestos exposure. Asbestos dysregulates the immune response both alone (45,46) and in combination with cigarette smoke (47,48). In our study, cases and controls were matched on asbestos and smoking exposure prior to lung cancer diagnosis; even after controlling for these well-recognized risk factors for lung cancer, we still observed that mdNLR was associated with lung cancer risk. We note that all asbestos exposed workers included in our study were male, and that NSCLC associations for male (OR, 95% CI: 1.15, 0.88–1.51) and female (OR, 95% CI: 1.22, 0.75–1.98) participants without asbestos exposure were similar. Therefore, we attribute the strengthened association among men to the subset of men with occupational asbestos exposure. We postulate that elevated pre-diagnosis mdNLR in asbestos-exposed lung cancer cases may be indicative of an altered inflammatory profile induced by asbestos exposure that is permissive to the development of lung cancer.

Previous work has established the relationship between systemic inflammation and cancer, which includes tumor development, angiogenesis, and inhibition of apoptosis (49). Thus it is plausible that the high inflammatory profile reflected by elevated NLR (i.e., high neutrophil counts and/or low lymphocyte counts) may be more favorable for cancer development, tumor growth, tumor initiation, or early tumor progression. Specifically, lymphocytes act as tumor defense by inducing cytotoxic cell death and inhibiting proliferation and migration (49), while neutrophils have been shown to inhibit the activity of lymphocytes such as T-cells (4). Our findings parallel research in circulating immune markers and lung cancer risk (43,46) that have helped to establish roles for inflammation in the initiation and promotion of lung cancer (50).

Research on the possible relationship between elevated NLR or mdNLR and cancer risk is limited, and the only other lung cancer study reported a null association between NLR and lung cancer risk (15). The Sanchez-Salcedo study (15) included 135 participants from an LCDT lung cancer screening cohort, 32 of whom developed lung cancer during 4 to 10 years of follow-up, with the majority (72%) of cases diagnosed at early stage (I or II). This is in contrast to our study, where the majority (73%) of participants were diagnosed with late stage (III or IV) lung cancers and were diagnosed prior to use of lung cancer screening. In addition, the enrollment criteria for the Sanchez-Salcedo study (15) was relaxed in comparison to the U.S. Preventive Task Force guidelines for lung cancer screening eligibility (25), as participants were enrolled and screened as young as 40 years with a minimum smoking history of 10 pack years. Therefore, participants screened in the Sanchez-Salcedo study would not have been eligible for lung cancer screening in the U.S. We note that the majority (76%) of CARET case-control pairs included in this study meet the eligibility criteria for screening in the U.S., but were not screened since enrollment occurred long before lung cancer screening was implemented in the U.S. (44). Regarding other cancer types, a study of NLR and breast cancer risk observed that elevated NLR was associated with increased risk (14), which was confirmed in a study of mdNLR (16). In that study, elevated mdNLR was also observed to be associated with risk of head and neck squamous cell carcinoma and ovarian cancer, but there was no association with bladder cancer (16).

Despite our careful control of confounders in our study, this work is not without limitations. Our study sample consisted solely of heavy smokers who elected to participate in the CARET clinical trial, and therefore our results may have limited generalizability. Since heavy smokers are at the highest risk for lung cancer, and because the majority of smokers will not develop lung cancer during their lifetimes (24), this is an ideal population in which to investigate features that can differentiate between smokers who go on to develop lung cancer from those who will not. Based on our results, it is possible that mdNLR may be able to differentiate lung cancer risk in individuals at high enough risk to be eligible for screening. Additional work is needed to clarify whether mdNLR in screening-eligible populations can be leveraged to identify lung cancer at earlier stages among those who are screened. Small sample sizes for subgroups may have limited our statistical power to detect associations. We did not have cytological NLR measurements for the participants in our study. Nevertheless, previous work has validated that mdNLR is highly correlated with cytological NLR (R2=0.99) (16), and our estimated mdNLR aligns with the mean cytological NLR of 2.15 reported for a nationally representative sample of over 9,000 individuals from the National Health and Nutrition Examination Survey (NHANES) (42). Finally, our results have not yet been replicated in an external validation data set.

Our results demonstrate the utility of DNA methylation profiling for NLR estimation (e.g., mdNLR) for prospective analyses using archival samples. NLR determined from complete blood count measures assessed for standard clinical care are inexpensive and reproducible (6), and mdNLR is a validated proxy for NLR in archival blood samples (16). Thus, there is broad potential for mdNLR approaches to be used to improve biomarker discovery for risk prediction in other prospective studies of cancer.

We posit that NLR may convey information about both lung cancer risk and etiology among heavy smokers, especially those with a history of asbestos exposure. A deeper understanding of the mechanisms and pathways reflected by elevated NLR in lung cancer development will inform whether this potential marker has clinical utility in prevention and/or early detection, especially in high-risk populations who are actively undergoing (or have undergone) screening for lung cancer.

Acknowledgements:

The research reported in this publication was supported by the National Cancer Institute (NCI) of the National Institutes of Health R01 CA151989 to JA Doherty, the Munck-Pfefferkorn Fund at Dartmouth College to JA Doherty and CJ Marsit, the Huntsman Cancer Foundation to JA Doherty, and the Kansas IDeA Network of Biomedical Research Excellence Bioinformatics Core to DC Koestler, and supported in part by the National Institute of General Medical Science (NIGMS) Award P20 GM103418 to DE Wright, and the NCI under Award Numbers P30 CA042014 to ME Beckerle, P30 CA168524 to RA Jensen, and R01 CA111703 to C Chen. Support for CARET is from NCI grants UM1 CA167462 and U01 CA63673 to GE Goodman.

Abbreviations:

BMI

body mass index

BMIQ

β-mixture quantile

CARET

β-Carotene and Retinol Efficacy Trial

CI

confidence interval

CpG

cytosine-phosphate-guanine

mdNLR

methylation-derived neutrophil-to-lymphocyte ratio

NLR

neutrophil to lymphocyte ratio

NOS

not otherwise specified

NSCLC

non-small cell lung cancer

OR

odds ratio

RMSE

root mean squared error

SCLC

small cell lung cancer

Footnotes

Conflict of interest statement: The authors declare no potential conflicts of interest.

References

  • 1.Hanahan D, Weinberg Robert A. Hallmarks of cancer: The next generation. Cell 2011;144(5):646–74 doi 10.1016/j.cell.2011.02.013. [DOI] [PubMed] [Google Scholar]
  • 2.Guthrie GJK, Charles KA, Roxburgh CSD, Horgan PG, McMillan DC, Clarke SJ. The systemic inflammation-based neutrophil–lymphocyte ratio: Experience in patients with cancer. Crit Rev Oncol Hematol 2013;88(1):218–30 doi 10.1016/j.critrevonc.2013.03.010. [DOI] [PubMed] [Google Scholar]
  • 3.Templeton AJ, McNamara MG, Seruga B, Vera-Badillo FE, Aneja P, Ocana A, et al. Prognostic role of neutrophil-to-lymphocyte ratio in solid tumors: A systematic review and meta-analysis. J Natl Cancer Inst 2014;106(6):dju124 doi 10.1093/jnci/dju124. [DOI] [PubMed] [Google Scholar]
  • 4.Spiering MJ. Primer on the immune system. Alcohol Res 2015;37(2):171–5. [PMC free article] [PubMed] [Google Scholar]
  • 5.Mei Z, Shi L, Wang B, Yang J, Xiao Z, Du P, et al. Prognostic role of pretreatment blood neutrophil-to-lymphocyte ratio in advanced cancer survivors: A systematic review and meta-analysis of 66 cohort studies. Cancer Treat Rev 2017;58:1–13 doi 10.1016/j.ctrv.2017.05.005. [DOI] [PubMed] [Google Scholar]
  • 6.Yu Y, Qian L, Cui J. Value of neutrophil-to-lymphocyte ratio for predicting lung cancer prognosis: A meta-analysis of 7,219 patients. Mol Clin Oncol 2017;7(3):498–506 doi 10.3892/mco.2017.1342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Liu D, Huang Y, Li L, Song J, Zhang L, Li W. High neutrophil-to-lymphocyte ratios confer poor prognoses in patients with small cell lung cancer. BMC Cancer 2017;17(1):882 doi 10.1186/s12885-017-3893-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Gu XB, Tian T, Tian XJ, Zhang XJ. Prognostic significance of neutrophil-to-lymphocyte ratio in non-small cell lung cancer: A meta-analysis. Sci Rep 2015;5:12493 doi 10.1038/srep12493. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Derman BA, Macklis JN, Azeem MS, Sayidine S, Basu S, Batus M, et al. Relationships between longitudinal neutrophil to lymphocyte ratios, body weight changes, and overall survival in patients with non-small cell lung cancer. BMC Cancer 2017;17(1):141 doi 10.1186/s12885-017-3122-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Wang J, Kalhor N, Hu J, Wang B, Chu H, Zhang B, et al. Pretreatment neutrophil to lymphocyte ratio is associated with poor survival in patients with stage I-III non-small cell lung cancer. PLoS One 2016;11(10):e0163397 doi 10.1371/journal.pone.0163397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Ozyurek BA, Ozdemirel TS, Ozden SB, Erdogan Y, Kaplan B, Kaplan T. Prognostic value of the neutrophil to lymphocyte ratio (NLR) in lung cancer cases. Asian Pac J Cancer Prev 2017;18(5):1417–21 doi 10.22034/apjcp.2017.18.5.1417. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Deng M, Ma X, Liang X, Zhu C, Wang M. Are pretreatment neutrophil-lymphocyte ratio and platelet-lymphocyte ratio useful in predicting the outcomes of patients with small-cell lung cancer? Oncotarget 2017;8(23):37200–7 doi 10.18632/oncotarget.16553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Bar-Ad V, Palmer J, Li L, Lai Y, Lu B, Myers RE, et al. Neutrophil to lymphocyte ratio associated with prognosis of lung cancer. Clin Transl Oncol 2017;19(6):711–7 doi 10.1007/s12094-016-1593-y. [DOI] [PubMed] [Google Scholar]
  • 14.Okuturlar Y, Gunaldi M, Tiken EE, Oztosun B, Inan YO, Ercan T, et al. Utility of peripheral blood parameters in predicting breast cancer risk. Asian Pac J Cancer Prev 2015;16(6):2409–12. [DOI] [PubMed] [Google Scholar]
  • 15.Sanchez-Salcedo P, de-Torres JP, Martinez-Urbistondo D, Gonzalez-Gutierrez J, Berto J, Campo A, et al. The neutrophil to lymphocyte and platelet to lymphocyte ratios as biomarkers for lung cancer development. Lung Cancer 2016;97:28–34 doi 10.1016/j.lungcan.2016.04.010. [DOI] [PubMed] [Google Scholar]
  • 16.Koestler DC, Usset J, Christensen BC, Marsit CJ, Karagas MR, Kelsey KT, et al. DNA methylation-derived neutrophil-to-lymphocyte ratio: An epigenetic tool to explore cancer inflammation and outcomes. Cancer Epidemiol Biomarkers Prev 2017;26(3):328–38 doi 10.1158/1055-9965.epi-16-0461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zini G Stability of complete blood count parameters with storage: Toward defined specifications for different diagnostic applications. Int J Lab Hematol 2014;36(2):111–3 doi 10.1111/ijlh.12181. [DOI] [PubMed] [Google Scholar]
  • 18.Farlik M, Halbritter F, Müller F, Choudry FA, Ebert P, Klughammer J, et al. DNA methylation dynamics of human hematopoietic stem cell differentiation. Cell Stem Cell;19(6):808–22 doi 10.1016/j.stem.2016.10.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics 2012;13:86 doi 10.1186/1471-2105-13-86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Koestler DC, Christensen BC, Karagas MR, Marsit CJ, Langevin SM, Kelsey KT, et al. Blood-based profiles of DNA methylation predict the underlying distribution of cell types: A validation analysis. Epigenetics 2013;8(8):816–26 doi 10.4161/epi.25430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Titus AJ, Gallimore RM, Salas LA, Christensen BC. Cell-type deconvolution from DNA methylation: A review of recent applications. Hum Mol Genet 2017;26(R2):R216–R24 doi 10.1093/hmg/ddx275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Reinius LE, Acevedo N, Joerink M, Pershagen G, Dahlén S-E, Greco D, et al. Differential DNA methylation in purified human blood cells: Implications for cell lineage and studies on disease susceptibility. PLoS One 2012;7(7):e41361 doi 10.1371/journal.pone.0041361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Wiencke JK, Koestler DC, Salas LA, Wiemels JL, Roy RP, Hansen HM, et al. Immunomethylomic approach to explore the blood neutrophil lymphocyte ratio (NLR) in glioma survival. Clin Epigenetics 2017;9:10 doi 10.1186/s13148-017-0316-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Thun MJ, Henley SJ, Calle EE. Tobacco use and cancer: An epidemiologic perspective for geneticists. Oncogene 2002;21(48):7307–25 doi 10.1038/sj.onc.1205807. [DOI] [PubMed] [Google Scholar]
  • 25.Moyer VA. Screening for lung cancer: U.S. Preventive services task force recommendation statement. Ann Intern Med 2014;160(5):330–8 doi 10.7326/m13-2771. [DOI] [PubMed] [Google Scholar]
  • 26.Sakoda L, Loomis M, Doherty J, Julianto L, Barnett M, Neuhouser M, et al. Germ line variation in nucleotide excision repair genes and lung cancer risk in smokers. Int J Mol Epidemiol Genet 2012;3(1):1–17. [PMC free article] [PubMed] [Google Scholar]
  • 27.Goodman GE, Thornquist MD, Balmes J, Cullen MR, Meyskens FL, Jr., Omenn GS, et al. The Beta-Carotene and Retinol Efficacy Trial: Incidence of lung cancer and cardiovascular disease mortality during 6-year follow-up after stopping beta-carotene and retinol supplements. J Natl Cancer Inst 2004;96(23):1743–50 doi 10.1093/jnci/djh320. [DOI] [PubMed] [Google Scholar]
  • 28.Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A, et al. Risk factors for lung cancer and for intervention effects in CARET, the Beta-Carotene and Retinol Efficacy Trial. J Natl Cancer Inst 1996;88(21):1550–9. [DOI] [PubMed] [Google Scholar]
  • 29.Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A, et al. Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med 1996;334(18):1150–5 doi 10.1056/nejm199605023341802. [DOI] [PubMed] [Google Scholar]
  • 30.Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, et al. Minfi: A flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics 2014;30(10):1363–9 doi 10.1093/bioinformatics/btu049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Pidsley R, CC YW, Volta M, Lunnon K, Mill J, Schalkwyk LC. A data-driven approach to preprocessing Illumina 450k methylation array data. BMC Genomics 2013;14:293 doi 10.1186/1471-2164-14-293. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Liu J, Siegmund KD. An evaluation of processing methods for HumanMethylation450 beadchip data. BMC Genomics 2016;17:469 doi 10.1186/s12864-016-2819-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Wilhelm-Benartzi CS, Koestler DC, Karagas MR, Flanagan JM, Christensen BC, Kelsey KT, et al. Review of processing and analysis methods for DNA methylation array data. Br J Cancer 2013;109(6):1394–402 doi 10.1038/bjc.2013.496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Teschendorff AE, Marabita F, Lechner M, Bartlett T, Tegner J, Gomez-Cabrero D, et al. A beta-mixture quantile normalization method for correcting probe design bias in Illumina Infinium 450 k DNA methylation data. Bioinformatics 2013;29(2):189–96 doi 10.1093/bioinformatics/bts680. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Teschendorff AE, Zhuang J, Widschwendter M. Independent surrogate variable analysis to deconvolve confounding factors in large-scale microarray profiling studies. Bioinformatics 2011;27(11):1496–505 doi 10.1093/bioinformatics/btr171. [DOI] [PubMed] [Google Scholar]
  • 36.Koestler DC, Jones MJ, Usset J, Christensen BC, Butler RA, Kobor MS, et al. Improving cell mixture deconvolution by identifying optimal DNA methylation libraries (idol). BMC Bioinformatics 2016;17:120 doi 10.1186/s12859-016-0943-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Joehanes R, Just AC, Marioni RE, Pilling LC, Reynolds LM, Mandaviya PR, et al. Epigenetic signatures of cigarette smoking. Circ Cardiovasc Genet 2016;9(5):436–47 doi 10.1161/circgenetics.116.001506. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Lee KWK, Pausova Z. Cigarette smoking and DNA methylation. Frontiers in Genetics 2013;4:132 doi 10.3389/fgene.2013.00132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Dogan MV, Shields B, Cutrona C, Gao L, Gibbons FX, Simons R, et al. The effect of smoking on DNA methylation of peripheral blood mononuclear cells from African American women. BMC Genomics 2014;15(1):151 doi 10.1186/1471-2164-15-151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Ambatipudi S, Cuenin C, Hernandez-Vargas H, Ghantous A, Le Calvez-Kelm F, Kaaks R, et al. Tobacco smoking-associated genome-wide DNA methylation changes in the EPIC study. Epigenomics 2016;8(5):599–618 doi 10.2217/epi-2016-0001. [DOI] [PubMed] [Google Scholar]
  • 41.Arnson Y, Shoenfeld Y, Amital H. Effects of tobacco smoke on immunity, inflammation and autoimmunity. J Autoimmun 2010;34(3):J258–65 doi 10.1016/j.jaut.2009.12.003. [DOI] [PubMed] [Google Scholar]
  • 42.Azab B, Camacho-Rivera M, Taioli E. Average values and racial differences of neutrophil lymphocyte ratio among a nationally representative sample of United States subjects. PLoS One 2014;9(11):e112361 doi 10.1371/journal.pone.0112361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Shiels MS, Katki HA, Freedman ND, Purdue MP, Wentzensen N, Trabert B, et al. Cigarette smoking and variations in systemic immune and inflammation markers. J Natl Cancer Inst 2014;106(11):dju294-dju doi 10.1093/jnci/dju294. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Smith RA, Andrews KS, Brooks D, Fedewa SA, Manassaram-Baptiste D, Saslow D, et al. Cancer screening in the United States, 2018: A review of current American cancer society guidelines and current issues in cancer screening. CA Cancer J Clin 2018. doi 10.3322/caac.21446. [DOI] [PubMed] [Google Scholar]
  • 45.Maeda M, Nishimura Y, Kumagai N, Hayashi H, Hatayama T, Katoh M, et al. Dysregulation of the immune system caused by silica and asbestos. J Immunotoxicol 2010;7(4):268–78 doi 10.3109/1547691X.2010.512579. [DOI] [PubMed] [Google Scholar]
  • 46.Dostert C, Petrilli V, Van Bruggen R, Steele C, Mossman BT, Tschopp J. Innate immune activation through NALP3 inflammasome sensing of asbestos and silica. Science 2008;320(5876):674–7 doi 10.1126/science.1156995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Morris GF, Danchuk S, Wang Y, Xu B, Rando RJ, Brody AR, et al. Cigarette smoke represses the innate immune response to asbestos. Physiological Reports 2015;3(12):e12652 doi 10.14814/phy2.12652. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Brown TA, Holian A, Pinkerton KE, Lee JW, Cho YH. Early life exposure to environmental tobacco smoke alters immune response to asbestos via a shift in inflammatory phenotype resulting in increased disease development. Inhal Toxicol 2016;28(8):349–56 doi 10.1080/08958378.2016.1175526. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Coussens LM, Werb Z. Inflammation and cancer. Nature 2002;420(6917):860–7 doi 10.1038/nature01322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Engels EA. Inflammation in the development of lung cancer: Epidemiological evidence. Expert Rev Anticancer Ther 2008;8(4):605–15 doi 10.1586/14737140.8.4.605. [DOI] [PubMed] [Google Scholar]

RESOURCES