A prospective study of immune and inflammation markers and risk of lung cancer among female never smokers in Shanghai

Abstract

There is a paucity of data on risk factors for lung cancer among never smokers. Here, we have carried out the first large study of circulating inflammation markers and lung cancer risk among female never smokers in Shanghai. A study of 248 lung cancer cases in female never smokers and 263 controls was nested within the Shanghai Women’s Health Study (n = 75221), matched by dates of birth and blood collection (mean follow-up time = 7.5 years). Prediagnostic plasma levels of 65 inflammation markers were measured using a Luminex bead-based assay. Odds ratios (ORs) were estimated with multivariable logistic regression. Nine of 61 evaluable markers were statistically significantly associated with lung cancer risk among never smoking Chinese women (P-trend across categories <0.05). Soluble interleukin-6 receptor [sIL-6R; highest versus lowest category OR = 2.37; 95% confidence interval (CI) 1.40–4.02) and chemokine (C–C motif) ligand 2/monocyte chemotactic protein 1; (OR = 1.62; 95% CI 0.94–2.80) were associated with an increased risk of lung cancer, whereas interleukin (IL)-21 (OR = 0.53; 95%CI 0.31–0.93), chemokine (C–X3–C motif) ligand 1/fractalkine (OR = 0.54; 95% CI 0.30–0.96), soluble vascular endothelial growth factor receptor 2 (sVEGFR2, OR = 0.45; 95% CI 0.26–0.76), sVEGFR3 (OR = 0.53; 95% CI 0.32–0.90), soluble tumor necrosis factor receptor I (OR = 0.49; 95% CI 0.29–0.83), IL-10 (OR = 0.60; 95% CI 0.34–1.05) and C-reactive protein (OR = 0.63; 95% CI 0.37–1.06) were associated with a decreased risk. sIL-6R remained significantly associated with lung cancer risk >7.5 years prior to diagnosis. Markers involved in various aspects of the immune response were associated with subsequent lung cancer risk, implicating inflammation in the etiology of lung cancer among female never smokers.

Introduction

Although cigarette smoking causes the vast majority of lung cancer diagnosed worldwide, it has been estimated that among women, 75% of lung cancer deaths in developing regions and 29% of lung cancer deaths in industrialized regions are not attributed to smoking (1). The smoking prevalence among Chinese women is low (~4%); however, the lung cancer rates exceed those of countries with much higher smoking prevalence (2), emphasizing the need for investigation into the etiology of lung cancer among female never smokers in China.

There is evidence that pulmonary inflammation may play a role in the development of lung cancer among never smokers. Chronic inflammatory lung conditions (e.g. emphysema and tuberculosis) have been shown to be associated with increased lung cancer risk among never smokers (3). However, limited data exist on the precise inflammatory markers and pathways that may be involved in the development of lung cancer among female never smokers. The majority of studies examining circulating levels of inflammation markers has been carried out in western populations and has largely included smokers (4–7), without sufficient power to examine associations among never smokers. Therefore, studies designed specifically to address inflammation and lung cancer risk among female never smokers are needed to further understand the lung cancer etiology in this population.

We have carried out the first large study of circulating inflammation markers and lung cancer risk among female never smokers, investigating the association between 65 acute-phase proteins, pro- and anti-inflammatory cytokines, chemokines, growth factors and angiogenesis factors with prospective lung cancer risk in a case–control study nested within the Shanghai Women’s Health Study.

Materials and Methods

Study Population

Our study was nested within the Shanghai Women’s Health Study (SWHS), which has been described in detail elsewhere (8). Briefly, the SWHS recruited 75221 women (aged 40–70 years old) from seven communities in urban Shanghai during 1996–2000. At baseline, participants completed an in-person interview and a self-administered questionnaire, including information on demographics, disease history, personal habits, residential history, occupational history, diet, reproductive history, physical activity, water drinking and anthropometrics. Blood and urine samples were also obtained. Lung cancers [International Classification of Diseases (ICD) codes 162.1–162.9] were ascertained through self-report during biennial study visits, and through linkage to the Shanghai Cancer Registry and vital statistics, and verified with medical charts from the diagnosing hospitals. All women included in the study gave informed consent and the Institutional Review Boards of all participating institutions gave approval.

Our nested case–control study included 290 lung cancers that occurred in the SWHS through 31 December 2009. Two hundred and ninety controls who were alive and cancer free on the case selection date were matched to cases on date of birth (±2 years) and date of blood sample collection (±3 months). Due to a laboratory aliquoting error, 17 cases and 17 controls from one of the batches were excluded. Furthermore, ever smoking women were excluded, for a final total of 248 cases and 263 controls.

Laboratory methods

Circulating levels of 65 markers were measured in plasma specimens collected at the baseline visit (10 ml of sample collected in EDTA tubes, processed and frozen within 6 hours of collection, stored at −70°C) that had not previously been thawed. Markers were chosen based on a prior methodologic study that assessed the sensitivity and reproducibility of markers measured using multiplex immune panel technology, and to be consistent with previous studies assessing circulating markers of inflammation and lung cancer risk (5,6). Markers were measured using Luminex bead-based assays (Millipore Inc., Billerica, MA), which were tested for in an earlier study (9) and included cytokines, chemokines, growth factors, soluble receptors and acute-phase proteins (see Supplementary Table 1, available at Carcinogenesis Online, for a complete list, grouped by category of marker). Fifteen of the markers were measured with a high-sensitivity assay in the subset of cases and controls with sufficient volume (n = 450). Concentrations were calculated using either a four- or five-parameter standard curve. Plasma samples were assayed in duplicate, and averaged to calculate concentrations. Cases and matched controls were included on the same analytical batch, plated into adjacent wells. Laboratory technicians were blinded to case–control status. We included one pair of blinded duplicates within each batch and a pooled plasma sample across batches to evaluate assay reproducibility and drift across batches. For the high-sensitivity panel only, a single-pooled QC was measured twice on each batch to estimate reproducibility. We calculated coefficients-of-variation of the markers and intraclass correlation coefficients (Supplementary Table 1, available at Carcinogenesis Online). One marker [i.e. [i.e. thymic stromal lymphopoietin (TSLP)] with >90% of values below the lowest limit of quantification (LLOQ) and three markers with intraclass correlation coefficients <0.7 [i.e. granulocyte colony-stimulating factor (G-CSF), interleukin (IL)-3 and tumor necrosis factor-B] were excluded from this analysis, resulting in 61 evaluable markers (Supplementary Table 1, available at Carcinogenesis Online). In the remaining markers, 87% had intraclass correlation coefficients ≥0.90. Measurements below the LLOQ were assigned a value of half the LLOQ. In addition, 16 subjects had missing values for C-reactive protein (CRP), serum amyloid A and serum amyloid P, 32 subjects had missing values for IL-10 and 30 subjects had missing values for chemokine (C–X3–C motif) ligand 1 (CX3CL1)/fractalkine.

Statistical analyses

Marker levels were categorized into groups based on the proportion of individuals with measurements < LLOQ: markers with ≥75% of samples above the LLOQ (n = 55 markers) were categorized into quartiles (based on the distribution among controls); markers with 50–75% of samples above the LLOQ (n = 3 markers) were categorized as < LLOQ, and tertiles of detectable measurements; and markers with 25–50% of the samples above the LLOQ (n = 3 markers) were categorized as < LLOQ, and below and above the median. None of the markers had only 10–25% of measurements above the LLOQ. Pearson correlations between markers were assessed on detectable measurements.

Multivariable logistic regression models were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the association of each marker with lung cancer risk, adjusted for birthdate (continuous), sample collection date (continuous), passive smoke exposure (yes, no and missing), poor indoor ventilation (ever and never) and personal history of chronic bronchitis (yes and no). Further adjustment for education, occupation, history of tuberculosis, antibiotic use in the week prior to sample collection, family history of lung cancer in first-degree female relatives and coal use as cooking fuel was also considered, but these variables were not observed to be associated with lung cancer (P > 0.2). We assessed self-reported history of several medical conditions by case–control status (Supplementary Table 2, available at Carcinogenesis Online). As lobular proliferation of the mammary gland was significantly more common among lung cancer cases, we carried out a sensitivity analysis excluding women with this condition. Adjustment for regular aspirin use was considered, but was only reported among 17 subjects. P-values for trend were calculated by treating marker categories as ordinal variables in the models. A 5% false discovery rate criterion was applied to correct for multiple comparisons (P = 0.0008).

Among those markers with P-trends <0.05, further analyses were carried out: (i) restricting to adenocarcinomas and (ii) stratifying by time from blood collection to cancer diagnosis/date of selection based on the median value (<7.5 and 7.5+ years). Interactions between marker levels and latency were assessed by including a cross-product term in the model. In sensitivity analyses, we restricted to those cancers diagnosed ≥2 years after blood collection.

Among controls, we assessed the association between age, secondhand smoke exposure, indoor ventilation and personal history of chronic bronchitis with marker levels using a linear regression model, with continuous marker values as the outcome. Models were mutually adjusted for all of the predictors.

Results

Characteristics of 248 lung cancer cases and 263 controls are presented in Table 1. No significant differences in age, passive smoke exposure, education, occupation, history of chronic bronchitis, family history of lung cancer, use of coal as cooking fuel, poor home ventilation or antibiotic use in the week prior to sample collection were observed between cases and controls.

Table 1.

Characteristics of lung cancer cases and controls

	Cases, n = 248	Controls, n = 263	P-valuea
Age at baseline, years
40–49	61 (24.6)	57 (21.7)	0.73
50–59	50 (20.2)	56 (21.3)
60–70	137 (55.2)	150 (57.0)
Ever exposed to passive smoke
No	57 (25.2)	72 (28.7)	0.40
Yes	169 (74.8)	179 (71.3)
Missing	22	12
Education
Elementary/less	101 (40.7)	116 (44.1)	0.75
Middle school	68 (27.4)	62 (23.6)
High school	53 (21.4)	55 (20.9)
College+	26 (10.5)	30 (11.4)
Occupation
Never worked	2 (0.8)	0 (0)	0.54
Technicians and professionals	56 (22.6)	54 (20.5)
Government, political and legal workers	8 (3.2)	10 (3.8)
Administrative and service workers	40 (16.1)	50 (19.0)
Production and manufacturing workers	142 (57.3)	149 (56.7)
History of chronic bronchitis
No	220 (88.7)	243 (92.4)	0.15
Yes	28 (11.3)	20 (7.6)
Family history of lung cancer in first-degree female relatives
No	246 (99.2)	262 (99.6)	0.53
Yes	2 (0.8)	1 (0.4)
Ever used coal as cooking fuel
No	84 (33.9)	92 (35.0)	0.79
Yes	164 (66.1)	171 (65.0)
Ever poor ventilation in the home
No	202 (81.5)	204 (77.6)	0.28
Yes	46 (18.6)	59 (22.4)
Antibiotic use in the previous week
No	229 (92.3)	242 (92.0)	0.89
Yes	19 (7.7)	21 (8.0)
Lung cancer histology
Adenocarcinoma	91 (36.7)
Squamous cell carcinoma	7 (2.8)
Adenosquomous	15 (6.0)
Other	3 (0.8)
Unspecified/missing	132 (53.2)

^aEstimated with the χ² test.

Of the 61 evaluable markers, 9 were significantly associated with lung cancer risk (P-trend < 0.05; Table 2; all results Supplementary Table 2, available at Carcinogenesis Online). Soluble interleukin-6 receptor (sIL-6R; highest versus lowest category OR = 2.37; 95% CI 1.40–4.02) and chemokine (C–C motif) ligand 2/monocyte chemotactic protein-1 (CCL2/MCP-1; OR = 1.62; 95% CI 0.94–2.80) were associated with an increased risk of lung cancer, whereas interleukin-21 (IL-21; OR = 0.53; 95% CI 0.31–0.93), CX3CL1/fractalkine (OR = 0.54; 95% CI 0.30–0.96), soluble vascular endothelial growth factor receptor 2 (sVEGFR2, OR = 0.45; 95% CI 0.26–0.76), sVEGFR3 (OR = 0.53; 95% CI 0.32–0.90), soluble tumor necrosis factor receptor 1 (sTNFR1; OR = 0.49; 95% CI 0.29–0.83), IL-10 (OR = 0.60; 95% CI 0.34–1.05) and CRP (OR = 0.63; 95% CI 0.37–1.06) were associated with a decreased risk. After correcting for multiple comparisons, sIL-6R remained statistically significant. Of these 9 markers, sTNFRI, sVEGFR2 and sVEGFR3 were strongly correlated with each other (r > 0.80), whereas IL-21, IL-10 and CX3CL1/fractalkine were moderately correlated (0.60 < r < 0.70; Supplementary Table 3, available at Carcinogenesis Online). When all nine markers were included in one model, sIL-6R, CCL2/MCP-1 and CRP remained statistically significant (P-trend < 0.05). After excluding women with lobular proliferation of the mammary gland, results remained generally the same, although CRP (P-trend = 0.15), IL-10 (P-trend = 0.08) and MCP-1 (P-trend = 0.05) did not meet the threshold for nominal significance.

Table 2.

Association between circulating markers of inflammation and lung cancer risk among women in the Shanghai Women’s Health Study

				Multivariable-adjusted
Marker	Category	Cases	Controls	ORs (95% CIs)
Soluble receptors
sIL-6R	1	40	65	1.0
	2	48	66	1.20 (0.69–2.10)
	3	74	66	1.91 (1.12–3.23)
	4	86	66	2.37 (1.40–4.02)
	P-trend†			0.0003
sVEGFR2	1	83	65	1.0
	2	54	66	0.63 (0.39–1.04)
	3	72	66	0.88 (0.54–1.42)
	4	39	66	0.45 (0.26–0.76)
	P-trend			0.02
sVEGFR3	1	79	65	1.0
	2	63	66	0.78 (0.48–1.26)
	3	62	66	0.80 (0.49–1.30)
	4	44	66	0.53 (0.32–0.90)
	P-trend†			0.03
sTNFRI	1	81	65	1.0
	2	55	66	0.68 (0.42–1.11)
	3	71	66	0.88 (0.54–1.41)
	4	41	66	0.49 (0.29–0.83)
	P-trend			0.03
Cytokines
IL-21	1	67	58	1.0
	2	70	59	0.99 (0.59–1.64)
	3	37	59	0.51 (0.29–0.89)
	4	41	59	0.53 (0.31–0.93)
	P-trend			0.004
IL-10	1	65	54	1.0
	2	53	55	0.85 (0.50–1.45)
	3	44	56	0.64 (0.37–1.11)
	4	38	53	0.60 (0.34–1.05)
	P-trend			0.04
Chemokines
CX3CL1/Fractalkine	1	61	55	1.0
	2	68	55	1.10 (0.65–1.84)
	3	36	56	0.57 (0.33–1.01)
	4	34	55	0.54 (0.30–0.96)
	P-trend†			0.008
CCL2/MCP-1	1	52	65	1.0
	2	52	66	1.05 (0.61–1.78)
	3	74	66	1.65 (0.97–2.81)
	4	70	66	1.62 (0.94–2.80)
	P-trend†			0.03
Acute-phase protein
CRP	1	77	63	1.0
	2	63	64	0.76 (0.47–1.25)
	3	49	64	0.57 (0.34–0.97)
	4	51	64	0.63 (0.37–1.06)
	P-trend			0.05

Multivariable models adjusted for date of birth, date of blood collection, exposure to passive smoke, poor indoor ventilation and personal history of chronic bronchitis.

When restricted to adenocarcinomas (n = 91), associations were generally in the same direction as our overall findings, but non-significant due to low power, with the exception of sIL-6R (P-trend = 0.008) and CRP (P-trend = 0.0005), where statistically significant associations remained (Supplementary Table 4, available at Carcinogenesis Online). No statistically significant interactions between marker levels and latency were observed; however, nearly all of the markers were only significantly associated with lung cancer within 7.5 years of diagnosis, with the exception of sIL-6R, which remained strongly associated with lung cancer risk >7.5 years before diagnosis (highest versus lowest category: OR = 3.52; 95% CI 1.92–4.21) and CRP and IL-21, which were moderately inversely associated with lung cancer >7.5 years before diagnosis (OR = 0.53; 95% CI 0.29–0.64; P-trend = 0.07 and OR = 0.55; 95% CI 0.31–0.70; P-trend = 0.08; Figure 1; Supplementary Table 5, available at Carcinogenesis Online). In addition, CCL15/MIP-1D, CCL22/MDC and CCL19/MIP-3B were positively associated and granulocyte-macrophage colony-stimulating factor (GM-CSF) and CXCL5/ENA-78 were inversely associated with lung cancer risk ≤7.5 years before diagnosis (all P-trend < 0.05). Furthermore, CCL4/MIP-1B was positively associated with lung cancer risk >7.5 years before diagnosis (P-trend = 0.02). Associations were similar, although attenuated for most markers, in a sensitivity analysis where we restricted to cases occurring >2 years after blood collection (n = 233; Supplementary Table 6, available at Carcinogenesis Online).

Figure 1.

Association between circulating markers of inflammation and lung cancer risk stratified by time since blood collection among women in the Shanghai Women’s Health Study. (A) <7.5 years, (B) 7.5+ years. Points indicate ORs and lines indicate 95% CIs.

Among controls, levels of sIL-6R, sVEGFR2, sVEGFR3, CCL2/MCP-1, sTNFRI and CRP were each positively associated with age at blood collection (all P < 0.05; Supplementary Table 7, available at Carcinogenesis Online). Indoor ventilation and personal history of chronic bronchitis were not significantly associated with these markers, with the exception of positive associations observed between poor indoor ventilation and sVEGFR3. Secondhand smoke exposure was associated and IL-21 and CX3CL1/fractalkine. About 159 controls had information about hour-years of passive smoke exposure, which was moderately associated with the levels of IL-10 (P = 0.11), IL-21 (P = 0.12) and CX3CL1/fractalkine (P = 0.08).

Discussion

In the first study of circulating immune markers and lung cancer risk among female never smokers, the authors identified nine markers that represent multiple aspects of the inflammatory process to be associated with lung cancer. These markers include a receptor for a key regulator of the inflammatory response (sIL-6R), two anti-inflammatory cytokines (IL-21 and IL-10), a marker of apoptosis (sTNFRI), two soluble receptors related to angiogenesis (sVEGFR2 and sVEGFR3), two chemokines (CX3CL1/fractalkine and CCL2/MCP-1) and an acute-phase protein (CRP). This study provides evidence for a role of immune response in the development of lung cancer in the absence of cigarette smoking.

Increasing levels of sIL-6R and CCL2/MCP-1 were each associated with increased risk of lung cancer among female never smokers. IL-6 is a key mediator of the inflammatory response, and many of the functions of IL-6 are either stimulated or enhanced when it is bound to its soluble receptor (sIL-6R) (10). Furthermore, the binding of IL-6 to sIL-6R is thought to promote the transition from acute to chronic inflammation (11). Importantly, the association between sIL-6R and lung cancer preceded lung cancer by more than 7.5 years, supporting a role for chronic inflammation in the development of lung cancer in the absence of smoking. CCL2/MCP-1 is a chemokine expressed by many cell types when stimulated by a number of inflammatory stimuli, and it is chemotactic for macrophages, lymphocytes and basophils (12). CCL2/MCP-1 appears to be associated with a number of inflammatory lung conditions, including allergic asthma, acute respiratory distress syndrome and idiopathic pulmonary fibrosis (12). In addition to one small study showing an association between the variants in the SNP MCP-1-2518 and non-small cell lung cancer (13), there is a body of evidence showing that CCL2/MCP-1 may play an important role in the progression and metastasis of lung cancer (14,15). Higher levels of CCL2/MCP-1 have been reported in Chinese never smokers exposed to diesel, a known lung carcinogen (16), perhaps suggesting that immune alterations are on the pathway between diesel exposure and increased lung cancer risk.

IL-10, IL-21, CX3CL1/fractalkine, sTNFRI, sVEGFR2, sVEGFR3 and CRP were each inversely associated with lung cancer in our study. IL-10, IL-21 and CX3C1/fractalkine were moderately correlated and had similar associations with lung cancer. IL-21 plays a central role in the proliferation, survival and differentiation of B cells (17). Higher levels of IL-21 may be inversely associated with lung cancer as it is proapoptotic and angiostatic (18). IL-10 inhibits the production of a number of proinflammatory cytokines (19), and variants in the IL-10 gene have been found to be associated with lung cancer risk (20). Although we found higher plasma levels of IL-10 to be inversely associated with lung cancer risk, IL-10 levels in the tumor have been shown to be associated with metastatic potential (5). The inverse associations between IL-21 and IL-10 with lung cancer are consistent with associations observed among never smokers in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (P-trends: IL-21, 0.009 and IL-10, 0.06) (Shiels et al., unpublished results). However, no association was reported between IL-10 and lung cancer among never smokers in a pooled study of prospective cohorts (4). CX3CL1/fractalkine is a chemokine that uniquely binds to CX3C receptor-1, and primarily attracts T and B lymphocytes, natural killer cells and monocytes (21). Levels of CX3CL1/fractalkine are increased in allergic asthma patients, and this chemokine has been proposed as a potential cancer gene therapy strategy for lung cancer (22).

Tumor necrosis factor plays a key role in the inflammatory response and binds to both TNFRI and TNFRII. TNFRI is expressed nearly in all tissue types and induces cell death through the activation of apoptotic signaling pathways (23). Although VEGFR2 and VEGFR3 are proangiogenic and prolymphangiogenic factors (24), respectively, the soluble forms of these growth factors appear to inhibit lymphangiogenesis (19,25). Therefore, lower levels of sTNFRI, sVEGFR2 and sVEGFR3, which are strongly correlated in plasma, may be indicators of antiapoptosis and lymphangiogenesis, potentially resulting in tumor survival and metastases. Of note, although a earlier study in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial with a limited number of never smokers (n = 83) did not find significant associations between any of these markers and lung cancer risk in never smokers, these studies were underpowered, and the effect estimates for sTNFRI, sVEGFR2 and sVEGFR3 were in the protective direction, consistent with our current results (5). CRP is an acute-phase protein, which is primarily stimulated by IL-6 and activates the complement pathway (26–28). CRP is a commonly measured marker of inflammation, and it has been shown to be positively associated with lung cancer, primarily among ever smokers, in a number of prospective studies (5,29–31). It is surprising that CRP is inversely associated with lung cancer among female never smokers in China, perhaps indicating differences in the role of inflammation and immunity in the development of smoking-related and smoking-unrelated lung cancers or a false positive result. Of note, diesel exposure, a known risk factor for lung cancer, has been shown to be associated with decreased levels of CRP in China (16).

Although only sIL-6R was associated with lung cancer risk 7.5+ years prior to diagnosis, marker associations did not differ significantly by latency. This, along with the lack of a statistically significant association with tumor size (data not shown), provides some evidence against reverse causation. However, it is possible that some of the associations observed in this study were induced by lung cancers that were present, but not clinically detectable. sVEGFR2 and sVEGFR3 appear to play a role in the spread of malignancies, and a number of studies have shown that CCL2/MCP-1 and CX3C1/fractalkine play important roles in tumor growth and metastasis (14,15,32), whereas the association between IL-10 and lung cancer prognosis is unclear (33,34). Thus, these markers may additionally play a role in the progression of lung cancers.

Our results suggest that inter-individual variability in the immune response may be associated with the development of lung cancer among female never smokers. However, it is possible that our associations have been influenced by residual and unmeasured confounding. Fortunately, our analyses are not confounded by smoking behaviors, as the women included in our analysis were all never smokers. However, although we adjusted for passive smoking and indoor ventilation, residual confounding may be present. For example, we were unable to adjust for intensity and duration of passive smoke exposure, as this information was unavailable in nearly half of participants. Furthermore, we did not have information on other exposures that may induce pulmonary inflammation, such as outdoor air pollution, other pulmonary or immune system-related medical conditions and pharmacotherapy use.

The SWHS is one of the few large cohort studies that can be used to examine risk factors for lung cancer among female never smokers. The main strengths of our study were the use of a well-designed cohort study with pre-diagnostic samples that were uniformly collected. In addition, this is the first large study of lung cancer in female never smokers to measure a broad number of immune and inflammatory markers in circulation using a novel multiplexed technology with high laboratory reproducibility. We also note several limitations. Although we have previously examined associations between these markers and lung cancers among never smokers, our prior efforts were underpowered, precluding us from making conclusions about female never smokers (5,31). The nine markers identified in this study show promise, particularly sIL-6R, which retained statistical significance after correction for multiple comparisons, but require replication given the large number of markers evaluated. Furthermore, future studies should assess marker levels at multiple time points for a better assessment of chronic inflammation and an improved understanding of how these markers change over time. It is unclear how well immune markers in circulation represent inflammation occurring locally in the lung. Finally, although these markers were associated with lung adenocarcinomas, we were not able to assess associations with other histologic types. Future studies are needed to both replicate the associations found here and to carry out targeted studies aimed at understanding the inflammatory pathways implicated by these markers and their role in lung cancer etiology.

In conclusion, this is the largest study to date to assess circulating markers of inflammation and lung cancer risk among female never smokers in China. We showed that markers involved in various aspects of the immune response, including receptors related to the proinflammatory response, apoptosis and angiogenesis and two chemokines, were associated with subsequent lung cancer risk. Although the novelty of these results requires replication in additional studies, they show that inflammation and immune response may play a role in the etiology of lung cancer among female never smokers.

Supplementary material

Supplementary data are available at Carcinogenesis online.

Funding

The Shanghai Women’s Health Study was supported by National Institutes of Health research grant R37 CA70867 and the Intramural Research Program of the National Institutes of Health, National Cancer Institute.

Abbreviations

 
• CI

  confidence intervals

 
• CRP

  C-reactive protein

 
• LLOQ

  lowest limit of quantification

 
• SWHS

  Shanghai Women’s Health Study

Acknowledgements

The authors thank Mr. Michael Curry and Ms. Anne Taylor (Information Management Services, Inc.) for statistical support.

Conflict of Interest Statement: None declared.

Author notes

†

These authors co-supervised this work.