Association Between Thyroid Disorders and Colorectal Cancer Risk in Adult Patients in Taiwan

IMPORTANCE Thyroid hormones have been shown to affect several important pathways in cancer development, including colorectal cancer (CRC). Clinical studies examining the association between thyroid disorders and colorectal cancer have conflicting results and have predominantly involved white populations. OBJECTIVE To determine if a diagnosis of hyperthyroidism or hypothyroidism is associated with the risk of developing colorectal cancer in an East Asian population. DESIGN, SETTING, AND PARTICIPANTS This nationwide population-based case-control study was conductedfromApril27,2018,toNovember8,2018,usingtheTaiwaneseNationalHealthInsurance Research Database. Participants were adults (n = 139426) either with a new diagnosis (between 2008 and 2013) of primary colorectal cancer without a history of cancer, or without cancer. Cases and controls were matched 1:1 by age, sex, and index date. Diagnosis of hyperthyroidism or hypothyroidism prior to the diagnosis of colorectal cancer (or the same index date in controls) was then determined. MAIN OUTCOMES AND MEASURES Risk differences in developing colorectal cancer among patients with a medical history of hyperthyroidism or hypothyroidism. waiver of data are deidentified. Conducted from April 27, 2018, to November 8, 2018, this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. We designed a large, nationwide, population-based case-control study in which cases were definedaspatientswithanewdiagnosisofCRC(diagnosedbetweenJanuary1,2008,andDecember 31,2013)andnohistoryofcancer,asrecordedintheNHIRD.Controls,thosewithoutaCRCdiagnosis, were randomly and matched 1:1 to the cases by age (by a margin of 1 month) and sex. We subsequently established the presence or absence of thyroid dysfunction prior to the diagnosis of CRC in the case group and, using the same index date as the matched case patient, in the control group. design; and the adjustment for multiple CRC risk factors, including age; sex; aspirin use; and history of IBD, obesity, benign colorectal polyps, diabetes, iodinetreatment,menopausalhormonetherapy,andcolonoscopyintheprevious10years.Byusing


Introduction
Colorectal cancer (CRC) is an increasingly common disease with far-reaching consequences. 1 It is the third most common cancer worldwide 2 and the third leading cause of cancer-associated deaths in the United States, in which 1 in 22 men and 1 in 24 women are expected to develop the disease over the course of their lifetimes. 3 Great efforts have been made to further characterize and understand the pathogenesis of CRC to develop new, more effective treatments for the disease. [4][5][6] As is the case with many neoplastic processes, with CRC genetic and molecular alterations appear to be involved, leading to uncontrolled proliferation of these abnormal cells. 4 Hormones have been identified as key factors associated with the development and evolution of many cancers and are being pursued as potential targets for therapy. [7][8][9][10][11] This association may be most obvious within the realm of sex hormones and gynecologic cancers, but the role of trophic hormones, including the family of thyroid hormones, has come under greater scrutiny. Thyroid hormones, their nuclear and cellular surface receptors, and even antithyroid antibodies have been shown to be associated with several important pathways in cancer development. 5,6 Studies in largely iodine-replete, white populations report the prevalence of hypothyroidism to be between 1% and 2%. 12 The prevalence of hyperthyroidism ranges from 0.1% to 0.5% and is higher among females than males. 13 Although numerous epidemiologic studies have detailed the association between thyroid dysfunction and cancer, to date relatively few studies have looked specifically at the association with CRC. 5,6,[14][15][16][17][18][19] Given that CRC is the fourth leading cause of death in Taiwan 20 and that previous studies have involved predominantly non-East Asian populations, we conducted, to our knowledge, the first population-based study in Taiwan to elucidate the association between hyperthyroidism, hypothyroidism, and CRC. For this purpose, we used the Taiwanese National Health Insurance Research Database (NHIRD), what we believe is one of the most comprehensive health care databases in the world. The goal of this study was to uncover the association between thyroid dysfunction and CRC risk in an Asian population.

Methods
This study was approved by the institutional review board of Taichung Veterans General Hospital, Taichung, Taiwan, which granted a waiver of informed consent because data are deidentified.
Conducted from April 27, 2018, to November 8, 2018, this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
We designed a large, nationwide, population-based case-control study in which cases were defined as patients with a new diagnosis of CRC (diagnosed between January 1, 2008, and December 31, 2013) and no history of cancer, as recorded in the NHIRD. Controls, those without a CRC diagnosis, were randomly selected and matched 1:1 to the cases by age (by a margin of 1 month) and sex. We subsequently established the presence or absence of thyroid dysfunction prior to the diagnosis of CRC in the case group and, using the same index date as the matched case patient, in the control group.

Taiwanese NHIRD
In 1995, the Taiwanese government established the National Health Insurance Program to provide coverage for most of the country's population. 21  We used the presence of these codes, combined with a certificate of catastrophic illness, to identify the initial pool of 74 057 potential patient cases.
We excluded from the case group individuals whose age and sex were not known, those who were younger than 18 years or older than 120 years, and those who were deceased. We also excluded those with a diagnosis of malignant diseases before the CRC diagnosis, considering the possibility of metastasis to the colon or rectum. On the basis of these criteria, we identified a total of 69 713 cases.
A total of 69 713 policyholders were selected as controls in a 1:1 match with cases; they were randomly paired for age, sex, and the same index date (the month and year of CRC diagnosis in the case group) from the 2013 version (consistent with the study time frame) of the Longitudinal Health Insurance Database. Similar to our process for the case group, we excluded individuals from the control group whose age and sex were not known, who were younger than 18 years or older than 120 years, who had a diagnosis of other malignant diseases, and who were deceased before the index date.

Hyperthyroidism, Hypothyroidism, and Other Adjustments
To identify patients with a thyroid disorder diagnosis, we searched for ICD-9-CM code 242 for hyperthyroidism and ICD-9-CM codes 243 and 244 for hypothyroidism. Additional criteria for inclusion were as follows: same diagnosis in at least 3 outpatient visits or 1 inpatient diagnosis followed by either another inpatient or outpatient visit with the same diagnosis. Patients with a diagnosis strictly from a single inpatient visit were excluded. The first diagnosis of a thyroid disorder must occur before the first CRC diagnosis in the case group or before the index date in the control group. We distinguished patients with acquired hypothyroidism from those with primary hypothyroidism by identifying the policyholders who first had a hyperthyroidism diagnosis (ICD- We adjusted for a history of radioactive iodine treatment, diabetes, benign colorectal polyps, obesity, inflammatory bowel disease (IBD), and aspirin use. We also adjusted for age; sex; medication or surgical treatment for thyroid disease; colonoscopy within the 10 years prior to CRC diagnosis (or index date); and menopausal hormone therapy, including estrogen-only and estrogen-plusprogesterone treatments and labeling them as either ever used or never used.

Statistical Analysis
We used the χ 2 test to analyze categorical variables, to observe differences in clinical characteristics between the case and control groups. A conditional logistic regression analysis was applied to examine the association between thyroid disorders (hyperthyroidism, hypothyroidism, and acquired hypothyroidism) and the risk of developing CRC (CRC, colon cancer, and rectal cancer), in which we controlled for possible confounders (listed earlier). The confounders were known or were possible factors associated with CRC development or diagnosis. In the subgroup analysis, a logistic regression model was applied to examine the association between hyperthyroidism and hypothyroidism treatments and the risk of developing CRC, adjusted for sex; age; aspirin use; and history of IBD, obesity, benign polyps, diabetes, radioactive iodine treatment, hormone therapy, and colonoscopy.
Statistical tests were all 2-sided, with P = .05 indicating the statistical significance level, and reported using 95% CIs and/or P values. All analyses were run using SAS, version 9.4 (SAS Institute Inc). Data were analyzed from May 3, 2018, to May 17, 2018.

Discussion
In this case-control study of 139 426 patients that used insurance claims data of nearly the entire population of Taiwan, we found that, after adjustment for many known risk factors, both hyperthyroidism and hypothyroidism were associated with a decreased risk of being diagnosed with CRC. When stratifying by age, the inverse association between hypothyroidism and rectal cancer appeared most pronounced in the subgroup of patients older than 50 years, whereas the inverse association between hyperthyroidism and colon cancer was most profound in patients younger than 50 years.
This study adds to the body of literature linking thyroid disease and CRC, which has accumulated some interesting, and seemingly contradictory at times, clinical results. 5,6,19,[22][23][24][25][26] One Abbreviations: NA, not applicable; OR, odds ratio. a Odds ratio was adjusted for sex; age; aspirin use; and history of inflammatory bowel disease, obesity, benign polyps, diabetes, radioactive iodine treatment, menopausal hormone therapy, and colonoscopy by logistic regression analysis. study found that, in general, patients with Graves disease have a higher risk of cancer. 24 Another study seems to clarify this finding, reporting the increased cancer risk in hyperthyroid function to be associated with lung and prostate cancer but not with colon cancer. 26 Boursi et al 5 and Rennert et al 6 both found a negative association between long-term thyroid hormone replacement with levothyroxine sodium and CRC risk. However, Rennert et al 6 did not control for the preceding hypothyroidism, and Boursi et al 5 found an overall higher risk of CRC among those with thyroid hormone replacement treatment for less than a year, untreated hypothyroidism, or hyperthyroidism.

JAMA Network Open | Diabetes and Endocrinology
Shu et al 19 found that, in patients hospitalized for Graves disease, risk of colon cancer was statistically significantly decreased. We hypothesized that these conflicting findings were largely associated with the complex interactions of thyroid hormones and their associated receptors with both normal and neoplastic colorectal tissues.
The biochemical association between thyroid hormones and colorectal cell differentiation, proliferation, tumorigenesis, and apoptosis is an involved and multifaceted process. The thyroid primarily secretes 2 hormones: thyroxine (T 4 ) and triiodothyronine (T 3 ). 27 Their adverse effects are mediated by nuclear receptors TRα1 and TRβ1 and the plasma membrane integrin αVβ3, which has 2 thyroid hormone-binding sites (S1 and S2). 28 Only T 3 at physiological concentrations can bind to S1, triggering the phosphorylation and thus activation of the phosphatidylinositol-3-kinase pathway, which promotes cell proliferation and inhibits apoptosis. 29 Activated by both hormones but primarily T 4 , S2 triggers the oncogenic extracellular signal-regulated kinase 1/2, facilitating a similar adverse effect while stimulating angiogenesis and the expression of fibroblast growth factor 2, components that are essential for rapid tumorigenesis. 30,31 Alterations in thyroid hormone receptor TRβ1 have been associated with colorectal adenomas and cancer, with data suggesting that TRβ1 may play a tumor suppressor-type role in malignancy progression. 4 Conversely, overexpression of thyroid hormone receptor TRα1 appears to be associated with accelerated tumor appearance and progression. 32 Similarly, thyroid hormone receptor interactor 13 has been shown to promote CRC cell proliferation, migration, and invasion 33 ; T 4 has been shown to promote β-catenin activation and cell proliferation in CRC 34 ; and thyroid hormone binding of cell surface receptor αVβ3 has been shown to lead to increased tumor cell proliferation and angiogenesis. 5 TRβ1 can also mediate the activation of the phosphatidylinositol-3kinase pathway through interplay with T 3 to ultimately promote the expression of transcription factor hypoxia-inducible factor 1's α subunit; its genomic targets facilitate hypoxia resistance and angiogenesis found in metastasizing, rapidly growing tumors. 27,35  Lin et al 31 argued that T 4 's role in carcinogenesis must be larger than T 3 because of its greater binding affinity to αVβ3 and naturally higher free-circulating levels. In addition, Lee et al 34 provided evidence that T 4 's nongenomic actions through binding with αVβ3 also promoted nuclear β-catenin accumulation and that T 4 in a dose-dependent manner promoted cell viability in CRC cell lines.

JAMA Network Open | Diabetes and Endocrinology
Therefore, a potential mechanism may exist in which low free T 4 from primary hypothyroidism could be protective against CRC by decreased interactions with the integrin.

Strengths and Limitations
Strengths of the current study include the large sample size; the source of data representing nearly the entire population of Taiwan; the case-control design; and the adjustment for multiple CRC risk factors, including age; sex; aspirin use; and history of IBD, obesity, benign colorectal polyps, diabetes, iodine treatment, menopausal hormone therapy, and colonoscopy in the previous 10 years. By using insurance claims data from the NHIRD, we captured the desired data as completely and accurately as possible. The case-control nature of the study using stored, computerized data from the period in question nearly eliminates the potential for recall bias, and the large pool of data from which we pulled the cases and controls lends strength to the associations we found.
Limitations of the study include the homogeneous nature of the study population; lack of data on the compliance of thyroid hormone replacement; CRC type and stage; and variants of thyroid disease, such as subclinical hypothyroidism, thyroid adenoma, and toxic nodular goiter. Thyroid hormone replacement therapy seems to have protective properties against CRC, depending on the duration of treatment. 5 We were not able to completely gather this information from the data, and this information could affect the associations we found. In addition, the use of ICD-9-CM codes that exactly matched hyperthyroidism and hypothyroidism may not have captured many cases along the spectrum of thyroid disease (eg, subclinical hypothyroidism, thyroid adenomas, and toxic nodular goiter), which have been suggested to have implications for cancer risk. 22,26 We were not able to collect data on the type of CRC of the case patients. Colorectal cancer is believed to develop mostly through the adenoma-carcinoma sequence, but also either de novo or from flat adenomas. 4 The pathways and signaling in these varying CRC types may differ considerably and therefore display different associations with hyperthyroid and hypothyroid states.

Conclusions
In this case-control study, both hyperthyroidism and hypothyroidism appeared to be associated with a decreased risk of being diagnosed with CRC. To be more specific, hyperthyroidism was associated with a lower risk of colon cancer, and hypothyroidism was associated with a lower risk of rectal