Cysteinyl Leukotriene Receptor Antagonists Decrease Cancer Risk in Asthma Patients

Previous in vitro and in vivo studies have demonstrated the potential of using cysteinyl leukotriene receptor antagonists (LTRAs) for chemoprevention, but this has not been investigated in any clinical setting. We therefore investigated the chemopreventive effect of LTRAs in a nationwide population-based study. From the Taiwan National Health Insurance Research Database, we enrolled adults with newly-diagnosed asthma between 2001 and 2011. Among these patients, each LTRA user was matched with five randomly-selected LTRA non-users by sex, age, asthma diagnostic year and modified Charlson Comorbidity Index score. We considered the development of cancer as the outcome. Totally, 4185 LTRA users and 20925 LTRA non-users were identified. LTRA users had a significantly lower cancer incidence rate than LTRA non-users did. Multivariable Cox regression analyses adjusting for baseline characteristics and comorbidities showed LTRA use was an independent protecting factor (hazard ratio = 0.31 [95% CI: 0.24–0.39]), and cancer risk decreased progressively with higher cumulative dose of LTRAs. In conclusion, this study revealed that the LTRA use decreased cancer risk in a dose-dependent manner in asthma patients. The chemopreventive effect of LTRAs deserves further study.

Scientific RepoRts | 6:23979 | DOI: 10.1038/srep23979 LTRAs. Using a retrospective cohort study design, we found that LTRA use was associated with a decreased cancer risk in a dose-dependent manner.

Methods
Data Source. The Taiwan National Health Insurance (NHI) has covered ambulatory care, inpatient care and prescription drugs in Taiwan since 1996. The NHI coverage rate was 96.2% of whole population in 2000 and increased to > 99% by 2005 2,17-21 . The NHI Research Database therefore comprises comprehensive health care information from nearly the entire population of 23.72 million in Taiwan, becoming one of the largest insurance databases in the world 17,[19][20][21][22][23] . The database used for this study is a cohort of two million subjects randomly sampled from NHI beneficiaries in 2000, and has been verified to be representative of the overall population of beneficiaries in terms of age, sex, geographic distribution and healthcare costs. The database includes information on medical reimbursement claims (such as ambulatory care claims, inpatient care claims, prescriptions, and registration entries) as well as information from Catastrophic Illness Registry, National Cancer Registry and National Register of Deaths. The database is managed by the Collaboration Center of Health Information Application (CCHIA), Ministry of Health and Welfare. For protection of confidentiality, patient identification has been already encrypted, and the authorized researchers are only permitted to perform data linkage, processing and statistical analyses with a specified computer in a closely monitored room. Using the scrambled personal identifier for each subject, the researchers are able to link the files to obtain socio-demographic information, longitudinal medical history and other information. Only statistical results were allowed to be brought out.
Study population. From the dataset, patients with newly diagnosed asthma were identified by the algorithm showed in Fig. 1. Patients with asthma diagnosis (International Classification of Diseases, Ninth Revision, Clinical Modification code [ICD-9-CM] of 493) in the ambulatory or inpatient claim database were identified, and only those with asthma diagnosis in at least three ambulatory claims or one inpatient claim were enrolled 18 . To ensure newly diagnosed adult asthma, those having asthma diagnosis before 2001 or those younger than 18 years old on their first asthma diagnosis were excluded.
The subjects who had ever used either montelukast or zafirlukast, the LTRAs available in Taiwan, after their asthma diagnoses were identified. After excluding those with neoplasm diagnosis (ICD-9-CM of 140-239 in any claims) before the end of the first year of LTRA use and those with the interval between first LTRA prescription and end of follow-up ≤ 1 year, subjects using LTRA for ≥ 30 days before the end of follow-up were identified as candidates for LTRA user cohort. The subjects who had never used LTRA were identified as candidates for LTRA non-user cohort.
Definitions of variables. The endpoint of this study was the development of cancer, defined by the appearance of cancer diagnosis in Catastrophic Illness Registry or National Cancer Registry. Pathological confirmation is generally required for reporting a cancer diagnosis to these registries. The date of death was obtained from the National Register of Deaths.
The presence of comorbidity was identified by the presence of any corresponding diagnostic codes before the index date in the claim databases and confirmed by the presence of the codes for at least three times in the ambulatory claim database or at least once in the inpatient claim database. Based on the comorbidities, modified Charlson Comorbidity Index (mCCI) score was calculated by subtracting chronic pulmonary disease from the original Charlson Comorbidity Index score 24 . Study cohorts. Each LTRA user was matched with five randomly-selected LTRA non-users by sex, age (± 2), asthma diagnostic year (± 2) and mCCI score. The index date was defined as the date of first LTRA prescription for LTRA users; the LTRA non-users were given the index date with the same interval from their first asthma diagnosis as their corresponding LTRA users. During the matching process, the same exclusion criteria for the LTRA users were also applied while selecting LTRA non-users to ensure enough follow-up time and absence of any cancer diagnosis before the end of the first year after index date.
To minimize immortal time bias, the follow-up period was calculated from a year after the index date. The subjects were followed from a year after the index date to either development of cancer, death or the end of 2011, whichever came first. The defined daily doses (DDD) were 10 mg and 40 mg for montelukast and zafirlukast, respectively. To quantify individual's exposure to LTRA, the cumulative defined daily doses of LTRA from the index date to the end of follow-up (cDDD) and to a year after the index date [cDDD(1y)] were calculated. Statistical analysis. The demographic data and comorbidities were compared between LTRA users and non-users using Pearson's χ 2 test for categorical variables or Student's t test for continuous variables, as appropriate. The cancer incidence rate (IR) was calculated as the number of cancer developed during the follow-up period divided by the total person-year. The cancer IRs in LTRA users and non-users were compared by estimating the incidence rate ratio (IRR) using Poisson regression and adjusted IRR (aIRR) using multivariable Poisson regression after adjusting for age, residency, income level, marriage status, education level and the presence of various comorbidities. Cumulative incidence of cancer was calculated and compared with Kaplan-Meier method and log-rank test. To further assess the effect of LTRA, multivariable Cox proportional hazards regression analyses were performed with adjustment of the same covariates as in Poisson regression. In addition, stratified analyses were also performed for Poisson and Cox regression in subgroups of covariates. To determine the effect of LTRA on the risk of different cancers, we also calculated the hazard ratios of LTRA use for several major cancers in Taiwan.
Extraction and computation of data, data linkage, processing and sampling and statistical analyses were performed using SAS system (version 9.3 for Windows, SAS Institute Inc., Cary, NC). The statistical significance level was set at a two-sided p value of < 0.05.  design. From the dataset, adult patients with newly diagnosed asthma were identified. Through the algorism, subjects using LTRA for more than a month (30 days) before the end of follow-up were identified as candidates for LTRA user cohort. The subjects who had never used LTRA were identified as candidates for LTRA non-user cohort. Each LTRA user was matched with five randomlyselected LTRA non-users by sex, age (± 2), asthma diagnostic year (± 2) and mCCI score. The index date was defined as the date of first LTRA prescription for LTRA users; the LTRA non-users were given the index date with the same interval from their first asthma diagnosis as their corresponding LTRA users. During the matching process, the same exclusion criteria for the LTRA users were also applied while selecting LTRA non-users to ensure enough follow-up time and absence of any cancer diagnosis before the end of the first year after index date. The subjects were followed from a year after the index date to either development of cancer, death or the end of 2011, whichever came first. The cumulative defined daily doses of LTRA were calculated from the index date to the end of follow-up (cDDD) and
LTRA users had a significantly lower cancer IR than LTRA non-users did (5.8 vs. 13.1 per 1000 patient-years; aIRR = 0.41 [95% CI: 0.36-0.47], p < 0.0001) ( Table 2), and all stratified analyses showed consistent findings. The cumulative cancer incidence was significantly lower in LTRA users than in LTRA non-users (p < 0.0001) (Fig. 2a). On stratified analyses, the LTRA users had a significantly lower cumulative cancer incidence as compared with LTRA non-users in strata of female, male, younger and elder subjects (all p < 0.0001) (Fig. 2b-e).  On multivariable Cox proportional hazards regression analyses adjusting for age, residency, income level, marriage status, education level and comorbidities, LTRA use was associated with a decreased cancer risk (hazard ratio = 0.31 [95% CI: 0.24-0.39], p < 0.0001) ( Table 3, model 1). The cancer risk decreased progressively with higher cumulative dose of LTRA use as compared with LTRA non-users. LTRA users with lower and higher cDDD of LTRA had 60% and 78% cancer risk reduction, respectively (Table 3, model 2). Similarly, LTRA users with lower and higher cDDD(1y) of LTRA had a 66% and 72% cancer risk reduction, respectively (Table 3, model  3). On stratified analyses, LTRA use was associated with a significantly lower cancer risk in all strata (Fig. 3a). LTRA users with higher cDDD or cDDD(1y) use had lower cancer risk than those with lower cDDD or cDDD(1y) did in nearly all strata (Fig. 3b,c). The significant effect of LTRA on cancer risk reduction was observed mainly in lung, colorectal, liver and breast cancer (Table 4).

Discussion
This large population-based study revealed that LTRA use was associated with a decreased cancer risk in asthma patients. Particularly, the chemopreventive effect appeared larger with a higher cumulative dose, indicating a dose-dependent manner of LTRA in this issue. The strengths of this study are its population-based sampling, avoidance of selection bias, adjustment for confounders, and, most importantly, the demonstration of dose-dependent protection effect. To the best of our knowledge, we are not only the first to report the chemopreventive effect of LTRAs in the clinical setting but also the first to demonstrate a dose-response relationship between the use of LTRAs and reduced risk of cancer. Further clinical studies are required to confirm our findings, and further in vivo and in vitro studies should be taken to investigate the chemopreventive mechanisms of LTRAs.
As inflammation is a major contributor for carcinogenesis and cancer progression, immune responses are the most important mechanisms running in tumour microenvironment. Indeed, the interaction between cancer cells and the surrounding immune cells have been noted to form a milieu which is suitable for carcinogenesis,  Table 2. Incidence rates of cancer in LTRA users and non-users. The subjects were followed from a year after the index date to either development of cancer, death or the end of 2011, whichever came first. The incidence rate (IR) is expressed as incident cancer per 1000 patient-years. The IRs of cancer in LTRA users and non-users were compared by estimating the incidence rate ratio (IRR) using Poisson regression and adjusted IRR (aIRR) using multivariable Poisson regression after adjusting for age, residency, income level, marriage status, education level and the presence of various comorbidities (except for the variable used for stratification).  The cumulative cancer incidence of (a) the whole study population, (b) female patients, (c) male patients, (d) subjects ≤ 65 years old, and (e) subjects > 65 years old. The red dashed lines and blue continuous lines show the cumulative cancer incidence of LTRA non-users and LTRA users, respectively. LTRA users had a significantly lower cumulative cancer incidence than LTRA non-users did (p < 0.0001). microenvironment 7 . Using selective cyclooxygenase-2 (COX-2) inhibitors for chemoprevention is therefore widely discussed. Our previous population-based study indicated that selective COX-2 inhibitors reduced development of colorectal cancer by at least 10% 3 . In recent years, the role of leukotriene pathway in carcinogenesis and tumour-mediated immunosuppression has been increasingly recognized 7,26 . While much effort has been made in identifying the role of LTB 4 pathway in cancer, the tumour-promoting role of cysteinyl leukotrienes, including LTC 4 , LTD 4 and LTE 4 , is less studied. Cysteinyl leukotrienes are originally recognized for their effect to promote bronchoconstriction, inflammation, microvascular permeability and mucus secretion 5 . Since more than a decade ago, LTD 4 has been shown to reduce apoptosis, enhance proliferation, induce transcriptional activity of potentially oncogenic genes and induce migration of intestinal epithelial cells 27 . Clinically, increased expression of CysLT 1 R was noted in specimens from colorectal, gastric and breast cancers, and the elevated CysLT 1 R expression correlated to poorer survival 13,28-30 . The circulating LTD 4 level was significantly higher in patients with hepatocellular carcinoma than in healthy subjects 31 . Over-expression of CysLT 1 R has also been shown in prostate cancer, renal cell carcinoma, transitional cell carcinoma and testicular cancer, and montelukast induces early apoptosis of these cancer cells [8][9][10][11][12]14 .
In addition to the pro-apoptotic effect of montelukast on few cancer cell lines, however, only few in vivo studies have reported chemoprevention effect of leukotriene pathway inhibitors in the literature while no clinical study is available currently. An early study demonstrated chemopreventive effect of leukotriene pathway inhibitors, accolate, zileuton and MK-866, in vinyl carbamate-induced lung tumours in mice 15 . In an in vivo LLC cells metastasis model, pranlukast and montelukast prevented tumour metastasis through peripheral capillaries 16 . A recent study using nude mice demonstrated that an LTRA, ZM198,615 or montelukast, inhibited the growth of colon cancer xenografts 14 .
In contrast to our previous study showing about 50% cancer risk reduction in users of selective COX-2 inhibitor, the present study showed an impressive 60-78% cancer risk reduction with using LTRAs 3 . Many adverse    effects of selective COX-2 inhibitors, especially renal failure and cardiovascular complications, prevent their wide application. LTRAs used in current clinical practice are generally so safe that can be used in paediatric asthma patients 6 . After our findings are further validated in other clinical studies, using LTRAs for chemoprevention might be much easier due to their satisfying safety profiles. There are several limitations in our studies. First, some well-known potentially important clinical covariates, such as smoking history and environmental exposure, are not available in the database. Interpreting of our results must be careful to account for possible impacts of these risk factors. Nevertheless, LTRA users and LTRA non-users are matched by sex, age, asthma diagnostic year and mCCI score, and Cox regression analyses were adjusted for age, residency, income level, marriage status, education level and comorbidities. To address the potential issues caused by the administrative database, we also conducted various stratified analyses and found consistent results. Because smoking rate was much lower in female (4.2%) than in male (46.8%) in Taiwan 2,32 , the results of stratified analyses in female subjects might be taken as a proxy for the effect of LTRAs in non-smokers. In this study, the chemopreventive effect of LTRAs seemed more pronounced in the female and younger subjects, as compared to male subjects and elder subjects, respectively. These findings suggested that the chemopreventive effect of LTRAs might be more pronounced in non-smokers than in smokers, and the detailed mechanisms deserves further study. Second, because our studies enrolled only asthma patients, whether the results can be applied to patients without asthma needs further study. However, because LTRAs are mainly used for allergic asthma, choosing subjects from asthma patients are therefore required to homogenize the case and control cohorts. Third, time-related biases were always a concern as in many observational studies 33 . Our study design inherently avoided time-lagging bias by unifying the interval between the initial asthma diagnosis and the index date of an LTRA non-user with that of the corresponding LTRA user. Although immortal time bias and time-window bias were not totally avoided in this study, our study design substantially minimized the impact of these biases. Besides, asthma was not associated with significantly increased cancer incidence 17 . Furthermore, the dose-dependent effect shown in multivariable Cox regression analyses increased the reliability of our results. Finally, this study was conducted in patients of Han Chinese ethnicity. Whether the findings are also applicable to other ethnic population require further evaluation.
In summary, our study reveals that the use of LTRA in asthma patients is associated with a decreased risk of cancer in a dose-dependent manner. The utility of LTRAs as chemopreventive agents deserves further study in depth.  This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/