Association between Dietary Patterns and Chronic Obstructive Pulmonary Disease in Korean Adults: The Korean Genome and Epidemiology Study

In addition to smoking, dietary habits may contribute to the development of chronic obstructive pulmonary disease (COPD). This study aimed to examine the association between dietary patterns and lung function in a Korean community cohort. A total of 5436 participants were included from the Ansan–Ansung cohort study. To identify the dietary patterns, we performed principal component factor analysis using the results of a semi-quantitative food frequency questionnaire. The forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio were measured by spirometry. Multiple logistic regression models were used to evaluate the association between dietary patterns and lung function after adjusting for confounders. We identified four major dietary patterns; ‘prudent’, ‘coffee, fat, and sweet’, ‘westernized’, and ‘white rice’. After adjusting for potential confounders, the ‘coffee, fat, and sweet’ dietary pattern was negatively associated with lung function, particularly the FEV1/FVC ratio. Participants with high scores for the ‘coffee, fat and sweet’ pattern had a higher risk of COPD among men but not women. Therefore, these results indicate that the ‘coffee, fat and sweet’ dietary pattern is inversely related to lung function in Korean adults. Our results indicate that dietary habits may be modifiable risk factors for COPD.


Introduction
Chronic obstructive pulmonary disease (COPD) is characterized by progressive airflow obstruction [1][2][3] and is currently the third leading cause of chronic morbidity and mortality worldwide, causing 3.23 million deaths in 2019 [4,5].
Although tobacco smoking is known to be the most common risk factor for COPD, from 25% to 45% of patients with airway obstruction are nonsmokers [6,7]. As nonsmokers account for a significant proportion of COPD patients, a deeper understanding of the other risk factors for COPD such as environmental exposure, occupational hazards, and dietary habits, becomes more important. While the relationship between lifestyle factors and the risk of developing other chronic diseases with similar burdens on quality of life, such as cancer, cardiovascular disease, or diabetes, has been reported in several previous studies [8][9][10][11], little is known about the relationship between lifestyle factors and the risk of COPD.
Dietary intake may be an important risk factor for impaired lung function, and healthy dietary habits may be protective against COPD [12]. In the past, many studies have focused on the effects of individual foods or nutrients [13]. An approach that analyzes dietary patterns rather than the individual constituents of food has recently become more appealing [14]. Such a pattern analysis may reflect the dietary habits more accurately as foods are consumed together, with interactions between individual nutrients. People eat meals with complex combinations of nutrients that are likely to be interactive or synergistic [15]. A single nutrient analysis might be confounded by the effect of dietary patterns [15,16]. In addition, considering how foods and nutrients are consumed in combination would more closely reflect the real world. Thus, a dietary pattern analysis in which multiple dietary components are operationalized as a single exposure could provide new insights into the association of diet and COPD [17,18].
However, there is still a lack of epidemiological studies on dietary patterns, and the impact of dietary intake on the risk of developing COPD. In this study, we aimed to determine the dietary patterns by using principal components analysis, and to investigate the association between dietary patterns and COPD. Additionally, we investigated the association of dietary patterns and their spirometric indices.

Study Population
The current study was based on the Korean Genome and Epidemiology Study (Ko-GES) [19], a large consortium project consisting of six population-based prospective cohort studies. The datasets were obtained from the Ansan and Ansung study among the six cohort studies. The Ansan and Ansung study recruited Korean adults who were aged from 40 to 69 years at the baseline from 2001 to 2002 [20]. The total number of participants in the base survey was 10,030, and the 7th follow-up survey and examinations were completed by 2016. All participants provided informed consent for the baseline data and biospecimens and underwent an interview and physical examination. The participants were questioned by trained interviewers regarding their socio-demographic status, lifestyle along with anthropometric measurements, and diet [19].
Participants for the current study were recruited from the 3rd follow-up assessment between 2005 and 2006. Among the 6231 eligible individuals, participants without more than two valid lung function measurements (n = 368), more than two bio-electrical impedance analyses (InBody 3.0, Biospace, Seoul, Korea) measurements [21] (n = 395), smoking history (n = 4), and implausible energy intake (<500 or >5000 kcal/day, n = 28) were excluded. As a result, a total of 5436 participants were included in the final analysis ( Figure 1).

Dietary Assessment
Certified clinical dietitians used a semi-quantitative food frequency questionnaire (SQFFQ) to assess the dietary data. For each food item, the participants accurately reported 103 responses pertaining to the frequency of consumption over the last year by checking one of nine frequency categories, which ranged from 'almost never' to '3 times/day', followed by a question on the portion size of each food, which was divided into three categories: 'less than one serving size', 'one serving size' and 'more than one

Dietary Assessment
Certified clinical dietitians used a semi-quantitative food frequency questionnaire (SQFFQ) to assess the dietary data. For each food item, the participants accurately reported 103 responses pertaining to the frequency of consumption over the last year by checking one of nine frequency categories, which ranged from 'almost never' to '3 times/day', followed by a question on the portion size of each food, which was divided into three categories: 'less than one serving size', 'one serving size' and 'more than one serving size'. The food frequency was validated by two semi-quantitative food frequency questionnaires (SQFFQ1 and SQFFQ2) repeated at 1-year intervals using 3-day diet records during each of the four seasons from December 2002 to May 2004. The median spearman correlation coefficients of most food intakes, assessed repeatedly by the SQFFQ, were 0.45 for all nutrient intakes and 0.39 for nutrient densities [1]. The daily intake of each food item was converted to grams per day by multiplying the frequency per day by the portion size score [19]. The total energy, macronutrients, vitamins, and minerals were calculated from the daily consumption.

Identification of Dietary Patterns
The dietary pattern was measured by extracting the factors using 27 common food groups, including the food items of daily consumption, based on references [22,23], and these patterns are presented in Table 1. The factors were identified by principal component analysis rotating with an orthogonal transformation to the varimax method. The rotated factors that explained the sum of the total variance by food groups were the interpreted eigenvalues. The four most meaningful factors were determined with an eigenvalues by men (>1.5) and women (>1.6). Factor analysis was conducted using the four factors, and the results were analyzed from the derived dietary patterns as the correlation of the factors and food groups with loadings ≥ ±0.30 [24].

Assessment of Lung Function
Spirometry was performed by three well-trained pulmonary technologists according to the manual of the American Thoracic Society/European Respiratory Society Task Force, using the same spirometer (Vmax-229, Sensor-Medics, Yorba Linda, CA, USA) for all participants [25]. The forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio were obtained. In addition, COPD was defined as an FEV1/FVC ratio below 0.7 according to the 2018 Global Initiative for Chronic Obstructive Lung Disease guidelines [26].

Statistical Analysis
Principal component factor analysis was performed to identify the dietary patterns. To analyze the relationship between dietary pattern and lung function, multiple logistic regression models were adjusted for variables: Model 1: adjusted for age; Model 2: adjusted for the variable from model 1 plus smoking status (No/Yes); Model 3: adjusted for the variables from model 2 plus total energy intake (kcal/day) and body-mass index (BMI) (kg/m 2 ); Model 4: adjusted for the variables from model 3 plus C-reactive protein (CRP). Quintiles of dietary intake were modeled as an ordinal variable in regression models, and the associated p value for the trend was obtained. Data were presented as means ± standard errors (SE) or the number of cases (percentage). Categorical data were compared using the Pearson's chi-squared test. Continuous variables were compared using the t-test. All p values < 0.05 were considered statistically significant.

Baseline Characteristics
The baseline characteristics of the study population are shown in Table 2. The mean age was 54.6 years, and 47.8% were men. More than half of the participants were nonsmokers (68.9%). Concerning the lung function parameters, the study population had a mean FEV1 of 2.8 L (±0.7), mean FVC of 3.6 L (±0.8), and mean FEV1/FVC of 79.1% (±6.4). The mean energy and micronutrient intakes were higher in men, except for the percent of energy from carbohydrate, calcium, vitamin C, and fiber. When we analyzed the baseline characteristics according to the FEV1, the lower quartile group of FEV1 tended to be older and thinner. In addition, the lowest quartile group of FEV1 had a lower intake of total energy, as well as lower intakes of vitamin C and vitamin E, among both men and women (Supplementary Table S1).  Table 3 shows the four major dietary patterns identified using factor analysis among men and women: Prudent pattern, involving a higher intake of potatoes, starch jelly, legumes, vegetables, kimchi, mushrooms, fruits, meat, fish and shellfish, seaweeds, soup, and other beverages; 'Coffee, fat, and sweet' pattern, characterized by a higher intake of sweets, oils and fats, and coffee; Westernized pattern with a higher intake of noodles and dumplings, cereals and snacks, bread, pizza, hamburger, starch jelly, nuts, fruits, meat, eggs, fish and shellfish, dairy products, and seasoning; White rice pattern with high factor loading with white rice and negative loading with whole grains.

Dietary Patterns and Lung Function Measurements
Tables 4 and 5 reveal multivariable regression analyses for the association between dietary pattern and lung function adjusted for models 1-4 among men and women, respectively. Regarding the relationship with the lung function parameter of FEV1/FVC ratio, pattern 2 (coffee, fat, and sweet) showed a significantly negative association after adjusting for variables in all models (Table 4). We found no associations between the dietary patterns 1, 3, and 4 and any of the lung function parameters among men. In addition, no association was found between dietary patterns and lung function parameters among women (Table 5).   Tables 6 and 7 show the results of the multiple logistic regression models estimating the association between dietary patterns and COPD among men and women, respectively. Participants with high scores for the 'Coffee, fat, and sweet' pattern had a higher risk of COPD among men. This trend remained even after adjustment for the confounders of age, smoking status, total energy intake, BMI, and C-reactive protein ( Table 6). In addition, when we analyzed the differences in intake based on the 27 common food groups, the COPD group showed a higher intake of sweets, kimchi, and soup, along with a relatively lower intake of starch jelly, mushrooms, fruits, and dairy products among men (Supplementary  Table S2). However, there was no significant association between dietary pattern and COPD among women (Table 7).

Discussion
In this study, we evaluated the relationship between dietary patterns and lung function outcomes. We have found that a 'coffee, fat, and sweet' pattern, characterized by higher intake of sweets, oils, fat, and coffee, was negatively associated with lung function, particularly the FEV1/FVC ratio, and was associated with an increased prevalence of COPD in men. In addition, although the association of the 'coffee, fat, and sweet' pattern with FEV1 was not statistically significant, it was consistent with the FEV1/FVC ratio results. This relationship was not modified after adjustment for potential confounders, such as energy intake, age, BMI, and smoking status.
Whereas the relationship between dietary patterns and the risk of other chronic diseases such as cancer, cardiovascular disease, or diabetes, has been reported in several previous studies [8][9][10][11], relatively few epidemiological studies have reported the relationship between dietary patterns and the risk of COPD. Two prospective studies [27,28] found that the 'prudent' pattern, characterized by a high intake of fruits and vegetables, oily fish, and low-fat products, was negatively associated with the diagnosis of COPD. One Chinese meta-analysis involving 550,614 participants also revealed that the healthy/prudent dietary pattern was associated with a decreased risk of COPD [29]. Our findings are consistent with these studies [27][28][29], as one of the hallmarks of the 'coffee, fat, and sweet' dietary pattern is a low vegetable intake. Our results are also in agreement with a cross-sectional analysis from the Netherlands, which found that the 'traditional' pattern, represented by higher intakes of red meat, fat, coffee, and beer, was associated with a lower lung function and higher prevalence of COPD [30]. Although the pathophysiological mechanisms are unclear, the potential beneficial effects of antioxidant and anti-inflammatory nutrients in fruits and vegetables are thought to have a positive effect on lung function [31,32]. Similar to the results of previous studies [33][34][35][36][37], the group with a lower FEV1/FVC had a lower intake of vitamin C and vitamin E in our study. They also tended to eat smaller amounts of fruits, dairy products, mushrooms, and starch jelly.
Recent studies [11,[38][39][40] have also demonstrated that a high intake of sweets and fat is associated with reduced lung function. Obesity has been proposed as a possible explanation for the association between a diet rich in sweets and fats and impaired lung function [41,42]. Reductions in FEV1 and FVC have been documented in extremely obese subjects. Fat accumulation in the rib cage and visceral cavity may reduce respiratory function [43]. Additionally, several studies [44][45][46][47] have reported that decreased lung function is associated with diabetes and high levels of fasting blood glucose. Diabetes and hyperglycemia may trigger oxidative stress-related inflammatory responses and alter the regulation of the inflammatory pathway [48], consequently leading to chronic lung function impairment [47]. These effects may be associated with alveolar morphological changes caused by increased oxidative stress [49]. Inflammatory changes in the pulmonary vascular tissue, along with thickening of the pulmonary capillary basement membrane, have been reported in diabetic animal models [49].
Studies on the effects of caffeine consumption on the risk of developing COPD have demonstrated inconsistent results. Hirayama et al. conducted a case control study that found a positive association between caffeine intake and the development of COPD in Japanese adults [50]. Conversely, a retrospective study has suggested that caffeine has no significant effect on the frequency of COPD exacerbations [51]. More research is needed to understand the relationship between caffeine intake and COPD.
In addition, we noted a sex-associated difference in the association of dietary pattern and lung function in this study. Dietary patterns and associated behavioral patterns may differ in men and women, and their impact on lung function could be different. There are several studies of gender differences in the effects of smoking on lung function [21,52]. Thresholds for the detrimental effects of pulmonary irritants are expected to be different in men and women [53]. Moreover, dietary macronutrients could have different effects on lung function in men and women [54]. Even we could not fully understand these various mechanisms, different approaches may be needed to analyze the associations between dietary patterns and lung function by gender.
The strengths of the current study are that it provides sufficient power analysis due to the large sample size (n = 5436), along with a validated and reliable food frequency questionnaire. Additionally, focusing on the overall dietary patterns rather than the individual nutrients in foods allows for a more predictive analysis of the disease risk [15,55]. Since individuals do not get nutrients from foods in isolation, we believe that our pattern analysis can provide practical guidelines for public health. Recently, dietary pattern analysis has been increasingly recognized as a complementary approach to inform public health recommendations [15,56]. The 2015 Dietary Guidelines Advisory Committee noted that the dietary patterns approach may have synergistic and cumulative effects on health and disease in its scientific report [57]. Another strength of our study is that we used spirometric measurements, which are the gold standard markers for the diagnosis of COPD.
This study has several limitations. First, our study population included only middleaged and older adults, and the findings might not be generalizable to other age groups. A second, further limitation relates to unmeasured confounding: although variable covariates including age, smoking, total energy intake (kcal/day), BMI (kg/m 2 ), and C-reactive protein were included, some potential covariates including lifestyle and sociodemographic factors may have roles as unadjusted confounders. Additionally, we could not measure the interaction [58] between dietary patterns and other covariates, such as the environment. Furthermore, although our results provide cross-sectional evidence that certain dietary patterns were associated with lung function, we did not show an association between dietary patterns and decline in lung function over a period of time. Additional longitudinal studies over an extended period of time are needed for an in-depth analysis.

Conclusions
In the current study, we observed an inverse relationship between a 'coffee, fat, and sweet' dietary pattern and lung function, as measured by the FEV1/FVC. Our results suggest that dietary habits may be modifiable risk factors for COPD.