Circulating Biomarkers of Tryptophan and the Kynurenine Pathway and Lung Cancer Risk

Background: Imbalances in tryptophanmetabolism have been linked to cancer-related immune escape and implicated in several cancers, including lung cancer. Methods: We conducted a nested case–control study within the European Prospective Investigation into Cancer andNutrition (EPIC) that included 893 incident lung cancer cases and 1,748matched controls.Circulating levels of tryptophan and six of its metabolites were measured and evaluated in relation to lung cancer risk. Results: Tryptophan (Ptrend 1⁄4 2 10 ) and the kynurenine/tryptophan ratio (KTR; Ptrend 1⁄4 4 10 ) were associated with lung cancer risk overall after adjusting for established risk factors. The ORs comparing the fifth and first quintiles (OR5th vs. 1st) were 0.52 [95% confidence interval (CI), 0.37–0.74] for tryptophan and 1.74 (95% CI, 1.24–2.45) for KTR. After adjusting for plasmamethionine (available from previous work, which was strongly correlatedwith tryptophan), the associations of tryptophan (adjusted Ptrend1⁄4 0.13) and KTR (Ptrend1⁄4 0.009) were substantially attenuated. KTRwas positively associated with squamous cell carcinoma, the OR5th vs. 1st being 2.83 (95% CI, 1.62–4.94, Ptrend 1⁄4 3 10 ) that was only marginally affected by adjusting for methionine. Conclusions: This study indicates that biomarkers of tryptophan metabolism are associated with subsequent lung cancer risk. Although this result would seem consistent with the immune system having a role in lung cancer development, the overall associations were dependent on methionine, and further studies are warranted to further elucidate the importance of these metabolites in lung cancer etiology. Impact: This is the first prospective study investigating the tryptophan pathway in relation to lung cancer risk. Cancer Epidemiol Biomarkers Prev; 23(3); 461–8. 2013 AACR. Authors' Affiliations: School of Public Health; Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, London; Institute for Ageing and Health, Newcastle University, Newcastle; Cancer Epidemiology Unit, Nuffield Department of Clinical Medicine, University of Oxford, Oxford; Department of Public Health and Primary Care, University of Cambridge, Cambridge, United Kingdom; International Agency for Research on Cancer, Lyon; Inserm U1018, Centre for Research in Epidemiology and Population Health (CESP); University Paris Sud, UMRS 1018; IGR, Villejuif, France; Section of Pharmacology, Institute of Medicine; Department of Public Health and Primary Health Care, University of Bergen; Laboratory of Clinical Biochemistry, Haukeland University Hospital; Bevital AS; Division of Epidemiology,Norwegian Institute of Public Health, Bergen; Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, Tromsø, Norway; WHO Collaborating Center for Food and Nutrition Policies, Department of Hygiene, Epidemiology and Medical Statistics, University of Athens Medical School; BureauofEpidemiologicResearch,AcademyofAthens; HellenicHealth Foundation, Athens, Greece; Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts; Department of Epidemiology, Julius Center for Health Sciences and Primary Care, University Medical Center; Department of Gastroenterology and Hepatology, University Medical Centre, Utrecht; National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands; Department of Epidemiology, German Institute of HumanNutrition Potsdam-Rehbruecke, Nuthetal; Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany; Cancer Registry and Histopathology Unit, "Civile M.P.Arezzo" Hospital, ASP Ragusa; Nutritional Epidemiology Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan; Molecular and Nutritional Epidemiology Unit, Cancer Research and Prevention Institute–ISPO, Florence; Human Genetics Foundation (HuGeF), Torino, Italy; Navarre Public Health Institute, Pamplona; Andalusian School of Public Health, Granada; CIBER Epidemiología y Salud P ublica (CIBERESP), Madrid; Public Health Directorate Asturias, Oviedo; Department of Epidemiology, Murcia Regional Health Authority, Murcia; Unit of Nutrition, Environment, and Cancer, Catalan Institute of Oncology, Barcelona, Spain Department of Radiation Sciences & Oncology, University Hospital; and Department of Medical Biosciences, Clinical Chemistry, Umea University, Umea , Sweden Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/). S.-C. Chuang and A. Fanidi contributed equally to this work. Current affiliation for Shu-Chun Chuang: Division of Health Policy Translation, Institutes of Population Health Sciences, National Health Research Institutes, Miaoli, Taiwan. Corresponding Author: Mattias Johansson, International Agency for ResearchonCancer (IARC), 150coursAlbert Thomas, 69008Lyon, France. Phone: 33-4-7273-8023; Fax: 33-4-7273-8342; E-mail: johanssonm@iarc.fr doi: 10.1158/1055-9965.EPI-13-077


Introduction
Lung cancer is annually responsible for 1.4 million deaths and remains the most lethal cancer worldwide (1). In Europe, lung cancer was the third most common cancer (12% of all new cancers) and the most common cause of death from cancer (20% of total cancer deaths) in 2008 (2), the age-and area-adjusted 5-year survival rate being 12% (3). There are four major histologic types of lung cancer, which comprise more than 90% of all cases: squamous cell carcinoma (SCC), large cell carcinoma (LCC), adenocarcinoma, and small-cell lung carcinoma (SCLC; ref. 4). Cigarette smoking is the single most important risk factor for all types of lung cancer; however, adenocarcinoma typically shows a weaker association with smoking than do the other histologic types (5,6).
Another potential mechanism for the role of tryptophan metabolism is related to cigarette smoking. Tobacco smoke contains a wide array of known carcinogens, including polycyclic aromatic hydrocarbon (PAHs), nitrosamines, and aromatic amines (15). In particular benzo[a] pyrene, a strong PAH carcinogen in the lung, is mediated by the aryl hydrocarbon receptor (AHR), and recent data from Opitz and colleagues linked TDO expression to AHR activation (9,16). This provides an additional mechanism by which enzymes related to metabolism of tryptophan through the kynurenine pathway may play a role in lung cancer etiology. It is methodologically complex to measure cellular IDO and TDO activity directly in an epidemiologic setting, but the ratio between circulating  . The tryptophan degradation pathway and potential implications of IDO induction. Tryptophan is an essential amino acid and is obtained by external sources. IDO transforms tryptophan into kynurenine, which is then metabolized into other catabolites within the kynurenine pathway. IDO induction decreases the local concentration of tryptophan, while increasing the concentration of downstream metabolites. IDO, indolamine 2,3-dioxygenase; TDO, tryptophan 2,3-dioxygenase; NAD, nicotinamide adenine dinucleotide; AHR, aryl hydrocarbon receptor; B2, riboflamin; B6, pyridoxal 5 0 -phosphate. kynurenine and tryptophan (KTR) is widely accepted as an indirect measure of IDO activity (11,13). Circulating tryptophan and kynurenine concentrations have been associated with poor outcome among patients with lung cancer (17,18), but have not yet been evaluated in relation to lung cancer incidence in a prospective setting.
The aim of the current study was to prospectively investigate the associations between circulating tryptophan, kynurenine pathway metabolites, and lung cancer risk in a large cohort.

Materials and Methods
The European Prospective Investigation into Cancer and Nutrition study European Prospective Investigation into Cancer and Nutrition (EPIC) is a multicenter prospective cohort study, which recruited subjects from 23 centers in 10 countries between 1992 and 2000. EPIC recruitment procedures and collection of questionnaire data and blood samples have been described in detail elsewhere (19). In brief, standardized questionnaire data were collected from 521,330 individuals across Europe, of whom 385,747 provided a blood sample. Blood samples were fractionated (serum, plasma, red cells, and buffy coat), aliquoted into 0.5 mL straws, and the majority were stored in liquid nitrogen tanks at the International Agency for Research on Cancer (IARC) in Lyon, France at À196 C, and four centers (Umea , Malm€ o, Copenhagen, and Aarhus) stored their samples in local biobanks. Informed consent forms were completed at each local center and the study was approved by the Institutional Review Board at the IARC and by the local ethical committees.

Ascertainment of cases and control selection
Incident cancer cases were identified through population cancer registries (Denmark, Italy except Naples, the Netherlands, Norway, Spain, Sweden, and the United Kingdom) or by active follow-up (France, Germany, Greece, and Naples). Active follow-up involved a combination of methods, including review of health insurance records, cancer and pathology registries, as well as direct contact with participants and their next of kin. Subjects were followed up from study entry until cancer diagnosis (except nonmelanoma skin cancer), death, emigration, or the end of the follow-up period for the relevant study center. End of follow-up was defined as the latest date of complete follow-up for both cancer incidence and vital status and varied between study centers from 2002 to 2006.
The selection criteria of the EPIC lung cancer study have been described previously (20). In brief, the study included eight of the 10 participating countries: the Netherlands, the United Kingdom, France, Germany, Spain, Italy, Greece, and Sweden (except for Malm€ o center). For each incident case, two controls were selected by incidence density sampling and matched by center, gender, date of blood collection (AE1 month, relaxed to AE5 months for sets without available controls), and date of birth (AE1 year, relaxed to AE5 years for sets without available controls).
Overall, 904 cases and 1,825 controls were included in the study. We further excluded subjects without kynurenine and tryptophan measurements (nine cases and 23 controls), two additional cases without matched controls, and 54 controls without matched cases. The final dataset for analysis was composed of 893 cases and 1,748 controls.

Biologic samples and laboratory analyses
Concentrations of tryptophan and kynurenine were determined simultaneously by gas chromatography/tandem mass spectrometry (21). Concentrations of anthranilic acid, xanthurenic acid, 3-hydroxykynurenine, kynurenic acid and HAA, neopterin, and cotinine were determined by liquid chromatography/tandem mass spectrometry (22). Measurements of methionine and vitamin B6 (pyridoxal 5'-phosphate; PLP) were available from a previous investigation of the same study population (20). All laboratory analyses were performed at Bevital AS (23).
Samples were analyzed in batches of 86 and quality control included six calibration samples, two control samples, and one blank sample in each batch. Samples from case and control participants were kept at À80 C and analyzed in random order. The within and between day coefficients of variation for the assays were 3% to 16% (22). The staffs at the Bevital laboratory were blinded to the casecontrol and the demographic status of the blood samples.

Statistical analysis
Quintile cutoff values for each biomarker were defined on the basis of the distributions among controls. The correlation between biomarkers was assessed by Spearman ranked partial correlation coefficients (r) adjusted for age, sex, country, cotinine concentrations (nmol/L), and body mass index (BMI; kg/m 2 ).
Conditional logistic regression was used to estimate OR and 95% confidence intervals (CI) for lung cancer, with the lowest quintile serving as the reference category. All models included adjustments by tobacco smoking (never, former, current, unknown, and cotinine concentrations). Adjusting for additional smoking variables (duration of smoking, average cigarettes smoked per day) did not alter the results notably. Further adjustments were conducted for education (no degree/primary school, technical or professional school, secondary school, university degree, and not specified/missing), BMI (kg/m 2 ), as well as quintiles of PLP and methionine that were strongly associated with lung cancer risk in a previous EPIC study (20).
Stratified risk analyses were conducted by histology, smoking status, and time from blood draw to diagnosis by unconditional logistic regression adjusted for the matching variables (age, in 5-year categories, sex, and country). The overall trend for each biomarker (P trend ) and stratified analyses were calculated by including the base 2 logarithm (log 2 ) of the biomarker concentrations as a continuous variable in separate logistic regression models. The OR trend estimate from this model may be interpreted as the relative risk associated with a doubling in circulating concentrations of the biomarker of interest.
The association between lifestyle factors with biomarker levels was investigated using linear regression models, adjusting for age, sex, and country, and further adjusting for cotinine (in quintiles) when appropriate.
All statistical tests were two sided and performed using SAS 9.1.

Results
Detailed baseline characteristics of the study population were presented previously (20). Briefly (Table 1), 893 lung cancer cases and 1,748 controls in total were included in the current study, out of whom 62% were men and 38% were women. The proportion of current smokers was 59% among cases and 23% among controls, whereas 29% of cases and 37% of controls were former smokers. The overall concentrations of tryptophan and kynurenines in cases and controls are shown in Table 1. The Spearman ranked correlation coefficients between biomarkers are presented in Supplementary Table S1. Table 2 shows the associations between tryptophan, kynurenine, KTR, and lung cancer risks overall. After controlling for tobacco smoking and cotinine concentration, tryptophan displayed an inverse association with lung cancer risk (P trend ¼ 1 Â 10 À5 ) and KTR displayed a corresponding positive association with risk (P trend ¼ 3 Â 10 À5 ). The OR comparing the fifth versus first quintiles (OR 5th vs. 1st ) were 0.52 (95% CI, 0.37-0.72) for tryptophan and 1.72 (95% CI, 1.23-2.40) for KTR. Further adjustment for educational attainment and BMI did not notably affect the associations of tryptophan and KTR with lung cancer risk, but controlling for PLP and methionine levels considerably attenuated both the association of tryptophan (adjusted P trend ¼ 0.13) and KTR (adjusted P trend ¼ 0.009), this being primarily driven by a strong correlation between methionine and tryptophan (r ¼ 0.51; Supplementary Tables S1 and S2). Further adjusting for combined intake of dairy and total fruit, nut and seed did not materially affect the overall association of circulating tryptophan or KTR with lung cancer risk (adjusted P trend ¼ 0.17 and P trend ¼ 0.009, respectively). As a secondary analysis, we also investigated other catabolites in the kynurenine pathway (Supplementary Table S3). In particular, plasma levels of kynurenic acid and xanthurenic acid were inversely associated with lung cancer risk (P trend ¼ 0.009 and 1 Â 10 À4 , respectively), but the associations were attenuated and no longer statistically significant after further adjustment by methionine and PLP (P trend ¼ 0.39 and 0.07, respectively).
Stratified risk analysis for KTR by smoking status did not reveal large differences in its association with lung cancer risk [P heterogeneity (P het ) ¼ 0.31], but the association differed substantially by histology (P het ¼ 0.01), being more prominent for cancers classified as SCCs and others ( Fig. 2 and Supplementary Table S4). The OR 5th vs. 1st for SCC was 2.83 (95% CI, 1.62-4.94, P trend ¼ 3 Â 10 À5 ), and adjusting for PLP and methionine only moderately attenuated this association (OR 5th vs. 1st ¼ 2.32, 95% CI, 1.30-4.15, P trend ¼ 4 Â 10 À4 ). The association between KTR and lung cancer risk was only apparent in former smokers (P trend ¼ 0.002 and Supplementary Table S5), but testing for heterogeneity did not indicate notable differences in OR when compared with never-and current smokers (P het ¼ 0.33; Fig. 2

Discussion
This is the first study exploring the association between tryptophan metabolites and lung cancer risk in a prospective study design. We found that the KTR was positively associated with overall risk, an association that was largely explained by a strong correlation between methionine and tryptophan. The association of KTR with risk was particularly prominent for SCC.
The key enzymes in degradation of tryptophan through the kynurenine pathway, IDO and TDO, as well as downstream catabolites may influence the immune system (11,24), and thus affect cancer development and progression (7,8,10,12,17,18,25). KTR is considered an indirect marker of IDO activity and was in the current study associated with increased risk of lung cancer. Although no previous prospective studies have investigated KTR in relation to lung cancer, our data would seem consistent with studies conducted in case series in which tryptophan depletion was associated with worse clinical outcome (17,18). However, the association of KTR with lung cancer risk was primarily driven by an inverse association of tryptophan, to a large extent explained by its strong correlation with methionine, which was previously found associated with lung cancer risk in the same study population (20). The association of KTR with lung cancer risk was particularly prominent for cancers classified as SCC, and controlling for methionine in this subgroup had little impact on the OR estimates. This observation is intriguing because kynurenine is an endogenous ligand of AHR (16), which in turn mediates the metabolism of PAHs, including potent carcinogens that are specifically relevant in SCC development (26,27). The association of KTR with SCC cancers may therefore indicate activation of a pathway related to detoxification of environmental chemicals.
There is limited evidence for the other kynurenines, anthranilic acid, kynurenic acid, xanthurenic acid, and HAA, in relation to cancer development or progression (Supplementary Table S5). For xanthurenic acid, we observed an inverse association with overall risk that was independent of KTR. Immunologic studies suggest that the downstream kynurenines may serve as a feedback of TH1 cell suppression and enhance TH2 response (24). Indeed, inflammation biomarkers have previously been strongly associated with lung cancer risk, indicating that inflammatory processes may be important in early lung cancer pathogenesis (28,29). It would be useful to obtain measurement of inflammation biomarkers, such as C-reactive protein, to further evaluate whether these associations are linked to inflammation. Alternatively, the associations may reflect the PLP-dependent catabolism of kynurenine ( Fig. 1; ref. 14) and is consistent with the positive correlations observed between PLP and kynurenic acid, xanthurenic acid and HAA (Supplementary Table S1). As previously reported (20), PLP is inversely associated with lung cancer risk in this population, but adjusting for PLP did not affect the association with xanthurenic acid (data not shown). The concentrations of tryptophan and kynurenines measured in the current study are similar to those published for other cohorts (30,31). This suggests adequate preanalytical samples handling because hydroxylated kynurenines are degraded in serum and plasma at room temperature or during prolonged storage at À25 C (30,32). KTR showed positive associations with age and BMI as previously shown (33)(34)(35)(36)(37), and with neopterin, which can be explained by the fact that both neopterin and IDO are stimulated by INF-g (38)(39)(40). Both KTR and neopterin were negatively associated with PLP, which is in accordance with multiple observations of inflammation or immune activation in subject with low PLP status (41). We also observed an inverse association of KTR, and several kynurenines with cotinine, in line with published results (42), which may reflect the immunosuppressive effect from smoking (43).
An unexpected observation of this study was the strong correlation between tryptophan and methionine (Supplementary Table S1 and S2). To our knowledge, there is currently little evidence implicating methionine in metabolism of tryptophan, but it is possible that intake, uptake, and metabolism of methionine and tryptophan are coregulated (44). The fact that methionine accounted for the overall association with risk of tryptophan, and consequently, to a large extent, the association of KTR, suggests that these associations with overall lung cancer risk are dependent of methionine. Whether they reflect common and correlated sources of these amino acids, a common pathway in which they act in relation to risk, or whether tryptophan is confounded by methionine, are unresolved questions that will require further research.
In conclusion, this study indicates that biomarkers of tryptophan metabolism are associated with subsequent lung cancer risk. Although this result would seem consistent with the immune system being important in lung cancer development, the associations were dependent on