Recently, the incidence of thyroid cancer (TC) has increased worldwide; in 2020, TC was one of the ten most common types of cancer globally based on GLOBOCAN data (1). In South Korea, TC ranked as the most common cancer type in terms of the age-standardized incidence rate for both sexes in 2019 (42.9 per 100,000), although the rank of TC dropped due to controversy surrounding overdiagnosis in 2014 (2). Moreover, South Korea has the highest incidence rate of TC worldwide; in this country, TC cases are predominantly cases of papillary thyroid carcinoma (3, 4).

To date, the underlying causes of TC have remained unclear. Previous findings proposed that there were some potential factors, such as the environment and lifestyle, that may play a role in TC development (5, 6). Specifically, positive associations between TC risk and several other factors, including ionizing radiation, history of benign thyroid nodules/adenoma or goiter, and sex (female), have been reported (7).

Among the modifiable risk factors for TC, diet seems key to TC development (7, 8). Of them, there are both carcinogenic substances and antineoplastic agents, which are considered to impact TC risk (9). For instance, several studies have found that iodine deficiency or extreme iodine excess and high consumption of cruciferous vegetables containing goitrogens seem to be related to a high risk of TC (10–12). Conversely, dietary patterns rich in fresh fruits and vegetables were found to prevent TC risk (13). Nevertheless, the relationship between TC and dietary habits remains ambiguous. For example, while previous findings suggested that using multivitamin supplementation during a long-term period (> 10 years) decreased the risk of papillary TC (14), another study indicated that the risk of this type of TC is increased in women who consumed multivitamins for more than 10 years (15). Moreover, there is limited evidence of a relationship between dietary habits and TC risk specifically in Asian populations, including those in South Korea, due to inconclusive findings.

Previous studies have reported that papillary TC is the primary type of TC among Koreans (4), with the majority of patients younger than 50 years at diagnosis (16), and it peaked at the age of 50 years, especially among women (17), while women were predominant (64.8%) in our study. Additionally, Haymart indicated that 50 years is the age at which age-related worsening of TC prognosis occurs in women because of menopause, which alters the impact of estrogen or follicle-stimulating hormone/luteinizing hormone on TC development (18). Moreover, the median age of participants was close to 50 years in this study. Thus, it would be important to observe the association between dietary habits and the risk of TC stratified by age (< 50 or ≥ 50 years) and sex.

Additionally, several studies have implied that smoking status was a factor that reduced TC risk (19, 20). Haymart et al. suggested that higher serum thyroid-stimulating hormone (TSH) levels could be responsible for the development of TC (21). Smokers may have lower TSH concentrations than non-smokers (22). Another possible biological explanation may be related to the tobacco components. The stimulated nicotine receptors on immune cells may be involved in the decreased risk of Hashimoto’s thyroiditis (23). Another tobacco alkaloid, anatabine, also helped to reduce not only the incidence but also the severity of thyroiditis related to thyroglobulin (24), while a positive relationship between TC and chronic thyroiditis, such as Hashimoto’s thyroiditis, was found in some studies (25, 26). However, the relationship between eating habits and TC development stratified by smoking status is still unclear and needs to be evaluated.

To address this research gap, we carried out a prospective cohort study in South Korea, which has a high incidence of TC. We aimed to identify the association between some dietary habits and TC development and evaluate this relationship after stratification by age, sex, and smoking status (ever smokers, non-smokers) in a Korean population.

This study assessed the relationship between some dietary habits and TC development in a relatively large prospective cohort in South Korea. Among the specific dietary habits, after adjustment for confounders, dairy consumption five or more days a week and meal durations longer than 10 min were found to be two protective factors against TC, especially among participants older than 50 years, women and those who never smoked.

Diet may affect TC development. This relationship can be assessed in several dietary aspects, such as nutrient intake and food consumption frequency and etc. In addition, previous findings suggest that the risk of TC is also modified by dietary habits (9). It is evident that a single nutrient or food intake may influence the cancer risk while it limits to assess the overall eating habits. However, the evaluation of the association between dietary behaviors and cancer risk could suggest a wider and more practical estimation of the relationship between eating habits and cancer risk (31). Therefore, dietary behaviors might contribute greatly to the prevention of cancer (32), including TC. In addition, since this study was carried out with specific dietary habit questionnaires in the Korean population, our findings provide evidence of the impact of eating habits on TC risk for these inhabitants. However, the results of our study may differ from those attained in other populations, such as Caucasian and Hispanic populations, due to differences in dietary patterns and lifestyle factors.

The evidence of an association between TC and dairy consumption in the Asian population is still limited. To the best of our knowledge, only one Japanese cohort study has investigated the relationship between TC risk and dairy consumption; they found reduced risk of TC in those who consumed higher amounts of dairy products, which is consistent with our findings (33). However, the impact of dairy consumption on the development of TC had inconsistent results in other geographical regions. A case-control study in Italy indicated that subjects with a higher dairy intake had a lower risk of TC (9). Conversely, another case-control study in Sweden and Norway reported that high intakes of butter and cheese were risk factors for TC development (34). Moreover, no significant relationship between dairy consumption and TC risk was found in other case-control studies (35, 36).

Biological mechanisms that could explain the protective role of dairy products in reducing TC risk have been proposed. Milk and dairy products are known as an essential source of many important nutrients, including calcium, liposoluble vitamins (A, D, E, and K), and essential fatty acids (37). Dairy fatty acids, particularly butyric acid and conjugated linolenic acid (CLnA), might be anticancer molecules with bioactivity that is mediated by the downregulation of acetyl-CoA carboxylase (ACC), fatty acid synthase (FASN), and 3-hydroxy-3-methylglutaryl coenzyme-A (HMG-CoA) or specific genes related to cell proliferation and apoptosis (38). CLnA was shown to be metabolized to rumenic acid (39), which has a possible beneficial effect on cancer and immunology (38). In addition, linolenic acid may be synthesized endogenously to docosahexaenoic acid (DHA), a long-chain fatty acid found to be inversely associated with TC risk (40). In addition, milk products contain vitamin D (41), and the type of vitamin might have a protective function against tumors (42). After absorption, this vitamin undergoes metabolism in the liver and kidney to become a biologically active metabolite (43). This metabolized compound can strongly bind to the vitamin D receptor on the surface of TC cells and consequently block the TC development stage via inhibition of cell growth (44).

In addition, a meal duration of more than 10 min was found to reduce the TC risk in our study population. To our knowledge, there have been no studies on the direct relationship between meal duration and TC development to date. The speed of eating might be considered a factor affecting the length of meal duration. Several previous studies found that eating fast could increase the risk of obesity and metabolic syndrome (45, 46) due to a reduction in insulin sensitivity (47). Otsuka et al. indicated that rapid speed of eating could trigger overeating before the stomach sensation of fullness because of a lack of satiety (46). Although there was no significant difference in BMI between individuals with and without TC in our study due to the relatively small number of individuals with incident TC and the lack of follow-up regarding anthropometric changes, previous studies reported positive associations among obesity, metabolic syndrome and TC (48, 49). Hyperinsulinemia due to obesity causes excessive insulin binding to IGF-1 receptors in the thyroid, increasing the proliferation of TC cells (50). Moreover, the mitogen-activated protein kinase (MAPK) pathway is triggered by the presence of insulin, which induces the dedifferentiation and proliferation of thyroid follicular cells (49).

TC development is believed to have a long latency period (51), and food habits from the initiation of latent disease to the diagnosis of TC could be important. Even though eating habits are relatively stable, alterations in dietary habits could occur and vary among participants over a long follow-up period (52). Dietary habits may be altered due to changes in food availability or active decisions leading to changes in dietary and lifestyle behaviors (53). Therefore, investigation of dietary habit changes during the follow-up period is necessary. However, our study examined eating habits only at baseline. In addition, there were clearly strong risk factors that might have impacted TC risk throughout the follow-up period. Previous findings indicated that ionizing radiation and xenobiotic exposure have clear positive associations with TC risk (11); these factors, combined with dietary habits, could influence the development of TC. However, we did not examine them comprehensively in this study.

In addition, during the follow-up period, a much higher proportion of women had incident TC compared to men in our study (Table 1), which is consistent with many previous findings (54, 55). Women have a higher risk for TC development due to the influence of sex hormones (56). During the reproductive period, women are influenced by estrogen, which is a strong growth factor for TC cells. High levels of estrogen in women could activate the estrogen receptor linked to the tyrosine kinase signaling pathways MAPK and PI3K for TC cell survival and invasion and TC progression (57). Herein, we additionally evaluated the association between sex and TC risk (Supplementary Table 6), and a significantly increased TC risk was observed among women (aHR, 2.95; 95% CI, 1.49–5.84). In contrast, we found no significant relationship between smoking status and TC development in the study population. Moreover, there was a low percentage of women who smoked in our study (7.40%). Thus, the association of dietary habits with TC could be impacted by women more than by smoking status. However, several previous studies indicated that smoking may decrease the risk of TC (20). Therefore, further studies are needed to confirm this association.

Healthy dietary habits play critical roles in the prevention of cancer, including TC. Thus, it is necessary to study and identify healthy eating habit recommendations. Our study indicates that dairy consumption five or more days a week and meal durations longer than 10 min could reduce TC development, which may contribute to the improvement of diet behaviors for TC prevention in the Korean population. Moreover, dairy foods are known as healthy food and their consumption has been suggested to be inversely associated with several cancer types and metabolic syndrome in previous Korean studies (58, 59). Therefore, high dairy consumption is a potential healthy lifestyle factor that decreases the risk of TC. To our knowledge, this study also presents the first investigation of the relationship between meal duration and TC development. In addition to eating speed, meal duration may also be related to food types, work traits, and other lifestyle-related factors. The findings in our study support the opinion that meal duration may be the dietary behavior marker, which is a helpful predictor of a healthy lifestyle for TC prevention.

There are several strengths of our study. To the best of our knowledge, this is the first prospective cohort study in South Korea to evaluate the relationship between some aspects of dietary habits and TC incidence. This prospective cohort study had a large sample size with a relatively long median follow-up period of 7.6 years. Moreover, we collected dietary information from study participants at enrollment; thus, selection or recall bias was prevented. In addition, analyses were adjusted for demographic information to control the impact of confounding factors. Additionally, assuming that bias may be associated with the large difference between incident TC cases and non-incident cases, we carried out sensitivity analysis using a population matched by a 1:2 ratio of TC cases and non-cases based on the propensity score matching method. However, we found that the results were consistent in both main and the matched population analyses.

Our study still has several limitations. First, we evaluated the diet information only at baseline and did not investigate whether dietary habits changed over the follow-up period. Second, our data had many missing values, as reported in Table 1. However, we excluded the missing values for the association analysis to minimize their impact on the results. Third, the dietary habit questions were only related to consumption frequency and did not include questions on dietary intake in which an objective indication of diet evaluation might be limited. Accordingly, we could not calculate the intake of each food and separate the study populations into subgroups by food intake. Fourth, the dietary habit questionnaire consisted of only 12 items assessing dietary factors, which might not comprehensively reflect the effect of all food items on the development of TC. However, we tried to formulate questions that represented the foods, dietary patterns, and eating habits specific to the Korean population. Fifth, there were several environmental exposures and nutritional factors (e.g., history of benign thyroid nodules/adenoma, ionizing radiation, and iodine intake) that were not measured completely in the subjects in the present study. These factors may result in bias in investigations of the relationship. Moreover, specific types of TC based on histology, e.g., papillary TC, should be examined in further studies.

In conclusion, our results imply that consuming milk and/or dairy products 5 or more days a week and meal durations greater than 10 min could be protective factors against TC in the population, exerting a stronger protective effect in individuals above age of 50 years, women, and those who never smoked. Further prospective studies with longer follow-up periods are needed to investigate dietary intake with more food items, including dietary patterns with respect to specific types of TC, while accounting for additional nutritional and environmental factors that could affect the results.

The original contributions presented in this study are included in this article/Supplementary material, further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Institution Review Board of NCC, Korea (No. NCCNCS-07-077). The patients/participants provided their written informed consent to participate in this study.

LN designed and conducted the research, analyzed the data, and wrote the manuscript draft. MG revised the manuscript. JL collected and analyzed the data. JK designed and conducted the research, collected the data, and revised the manuscript. All authors read and approved the final manuscript.