The Inverted U-Shaped Association of Caffeine Intake with Serum Uric Acid in U.S. Adults

Abstract

Background & Aims

Caffeine is a worldwide popularly consumed constituent in foods that can exert physiological effects. However, previous researches about the relationship between caffeine intake and serum uric acid (SUA) were limited and controversial. Therefore, we sought to investigate that relationship in U.S. adults.

Methods

In this cross-sectional study, the total sample of 7888 selected participants (3838 males and 4050 females) were identified from the National Health and Nutritional Examination Surveys (NHANES) 2015–2018. All subjects were tested for serum uric acid levels (µmmol/L), and their daily caffeine intakes (mg/d) were obtained by an average of two 24-hour dietary recalls. Multivariate linear regression models were used to evaluate the association between two variables in total subjects and subgroup analyses. Generalized additive models with smooth curve fittings were also performed.

Results

Multivariate regression analyses showed caffeine intake was negatively correlated with SUA after adjustment of other confounders. The subgroup analyses stratified by gender showed the negative correlation of caffeine intake with SUA was statistically significant in males but not in females. Furthermore, we observed a nonlinear inverse association of caffeine intake with SUA (P nonlinear <0.001) in the generalized additive model, followed by an inverted U-shaped curve (inflection point: 60.5mg/d) for all participants. This inverted U-shaped relationship between them could also be found in both genders, individuals aged below 60 years old, overweight (BMI of 25 to 30), and Non-Hispanic White individuals.

Conclusions

This study indicated that caffeine intake exhibited an inverse correlation with SUA, especially in males. In addition, this inverse relationship was nonlinear, which followed an inverted U-shaped curve.

Abbreviations

NHANES

National Health and Nutrition Examination Survey

SUA

Serum uric acid

NCHS

National Center for Health Statistics

CDC

Centers for Disease Control and Prevention

IQR

Interquartile range

BMI

Body mass index

PIR

poverty to income ratio

Hs-CRP

Hypersensitive C-reactive protein

Scr

Serum creatinine

BUN

Blood urea nitrogen

GAM

Generalized additive models

CI

Confidence interval

Introduction

Caffeine is the commonly consumed micronutrient worldwide, mainly taken in dietary sources such as tea leaves, roasted coffee beans (1), and beverages (2). The average daily caffeine intake is estimated to be 135 mg per day (equal to 1.5 standard cups of coffee, eight fluid oz [235 ml] is defined as a standard cup (3).) among 85% of U.S. adults who consume caffeine (2). According to the 2001–2010 National Health and Nutritional Examination Surveys (NHANES) dataset, nearly nine-tenth U.S. adults consume caffeine, and an average caffeine intake of the 90th percentile is ~440 mg/day among all adults (4). Moderate caffeine consumption (considered as less than 400 mg / day) has been shown to have no adverse health effects in the adult population (5). CYP1A2 (represents 15% of cytochrome P450) accounts for more than 90% of caffeine metabolism. The caffeine metabolites are 84% paraxanthine, 12% theophylline, and 4% theobromine, which are all ultimately oxidized by xanthine oxidase (XO) to produce uric acid (6). Finally, two-thirds of uric acid is excreted through the urinary tract, and the remainder is through intestinal excretion (7). High serum uric acid (SUA) levels can cause hyperuricemia, which might develop into gout with various medical conditions like renal and cardiovascular diseases, insulin resistance, etc. (8, 9, 10, 11).

Previous studies have reported coffee consumption was negatively correlated with SUA levels and the risk of hyperuricemia (12, 13, 14). A meta-analysis also illustrated that both men and women might benefit from moderate coffee consumption for its effects on primary prevention of gout or hyperuricemia (15). In contrast, some other studies indicated that coffee consumption was either positively correlated with SUA levels (16, 17), or had no significant relationship with SUA (18). Coffee is a beverage containing more than 1,000 components (19). Whereas caffeine is one of the main components in coffee, it is challenging to identify caffeine's role on SUA. An in vivo test showed that caffeine metabolism is dose-dependent, resulting in nonlinear clearance (20). Of note, medicine in human systems often exists with complex dynamic nature that cannot be adequately described by linear models, and nonlinear models could be applied to assist in explaining and predicting biological phenomena (21). Since previous researches about the relationship between coffee consumption and SUA were controversial, data about the association of caffeine intake with SUA were limited, and the influencing mechanism was not determined (15). Therefore, this study aimed to investigate the association between daily caffeine intake (mg/d) and SUA (µmol/L) and explore whether there is a nonlinear relationship.

Methods

Study population

The NHANES is a national population-based survey, providing the data of nutritional status and health information, which collected by the National Center for Health Statistics (NCHS) at the Centers for Disease Control and Prevention (CDC). In our analyses, the data were collected from the two most recent 2-year cycles (2015–2018) of the cross-sectional surveys in NHANES. The NHANES data comprise demography, dietary, physical examinations, biochemical analyses, and questionnaires. Additional information and detailed datasets on NHANES are publicly available at: http://www.cdc.gov/nchs/nhanes/ (the NCHS website). The NCHS Research Ethics Review Board approved all NHANES protocols, and all respondents gave their written informed consent (22).

Of the 19,225 U.S. adults (aged ≥ 20 years) in the NHANES (2015–2018) were interviewed. In this study, the exclusion criteria were: (1) missing data on SUA; (2) no record or unreliable 24-h recall for caffeine intake; (3) incomplete data for other potential confounders. After exclusions of 7070 subjects without data of SUA, 930 subjects without recorded data of two 24-h recalls, and 3337 subjects without data of other potential confounders, a total of 7888 adults (3838 men and 4050 women) were included in our analyses.

Variables

The exposure variable in our study was caffeine intake, while the outcome variable was SUA. Daily caffeine intake(mg/d) was obtained from the Total Nutrient Intakes dataset, which provided information on the respondent's foods (and beverages) consumption. Specifically, caffeine intake was assessed by a 24-hour dietary recall (midnight to midnight on the previous day of the interview) from the total nutrient intakes file. The Food and Nutrient Database for Dietary Studies (FNDDS) of the United States Department of Agriculture (USDA) converted the coding of the interview data to calculate the total nutrient intakes (http://www.ars.usda.gov/ba/bhnrc/fsrg).

There were two 24-hour dietary recalls that were not consecutive. The first was a face-to-face interview in the Mobile Examination Center (MEC). The second was consulted by the telephone three to ten days later. If a subject completed both recalls, caffeine intake was determined by using an average of two 24-hour dietary recalls. Otherwise, a single reliable dietary recall was used. All laboratory data were obtained from standardized and accredited analytical methods, and SUA (µmol/L) was measured by Beckman Coulter UniCel DxC 660i Synchron.

Other covariates in this study included continuous and categorical variables. Continuous variables included age(year), body mass index (BMI, kg/m2), minutes of sedentary activity, poverty to income ratio (PIR), waist circumference (cm), serum albumin(mg/dL), serum creatinine (Scr, mg/dL), hypersensitive C-reactive protein (hs-CRP, mg/L), and blood urea nitrogen (BUN, mg/dL). Albumin, the binding protein, plays an important role in caffeine metabolism (23). In healthy individuals or CVD patients, SUA (24) or SUA/Scr (25), hs-CRP (26), BUN/Scr (27) are associated with poor prognosis. However, coffee consumption may be associated with a lower risk of CVD and death from CVD (28, 29). Given those associations, we hypothesized that the blood markers described above might have potential confounding effects on the caffeine-SUA relationship. Thus, we adjusted for these covariates. Categorical variables included Race/Ethnicity, gender, marital status, smoking status, education, and physical activity. Details of all variables and acquisition procedures are available on the website (https://www.cdc.gov/nchs/nhanes/).

Statistical analysis

To evaluate the population sizes and oversampling of specific subgroups, sampling weights provided by NHANES were used (30). Data analyses were based on the guidelines of the CDC official website (https://wwwn.cdc.gov/nchs/nhanes/tutorials/default.aspx). Continuous variables conforming to normal distribution were expressed as mean ± SD (standard deviation), while skewed distribution variables were expressed as median (interquartile range, IQR). Categorical variables were presented as n (%, weighted). The statistical differences in each group were calculated by Student's t-test or Mann-Whitney U test (continuous variables), and chi-square test (categorical variables). To assess the association between the caffeine intake and SUA, multivariate linear regression analysis was used. We also performed the subgroup analyses of gender (male and female), BMI (normal, overweight, and obese), age (<60 years and ≥60 years), and Race/Ethnicity (Mexican American, Non-Hispanic White, Non-Hispanic Black, and other Races).

Additionally, generalized additive models (GAM) and restricted cubic splines (smooth curve fittings) were used to explore the nonlinear association between caffeine intake and SUA. A recursive algorithm was performed to estimate the inflection point if the nonlinear association was observed from smooth curve fittings. We constructed a two-piecewise linear regression model to calculate the threshold effect. To assess whether the best fit model was linear or nonlinear, the log-likelihood ratio test was applied to calculate the P nonlinear value. All statistical analyses were performed with the statistical software EmpowerStats version 2.0 (http://www.empowerstats.com, X&Y solution, Inc.) and R version 3.6.1 (http://www.R-project.org, the R Foundation). Statistically significant difference was considered when P value < 0.05.

Results

A total of 7888 participants (3838 men and 4050 women) selected from NHANES 2015–2018 were identified in our analyses. The weighted demographic characteristics and other covariates of the participants based on gender are presented in Table 1. Mean age was 47.33 ± 16.49 years for men and 48.38 ± 16.85 years for women, and mean serum acid was 361.11 ± 75.16µmol/L for men and 283.57 ± 73.17µmol/L for women. Caffeine intake was shown as a median (interquartile range, IQR), 111.00 (34.12∓217.50) mg/d for men and 83.50 (22.00∓171.50) mg/d for women. There were no significant differences in the minutes of sedentary activity and the Race/ Ethnicity between men and women. While in the other baseline characteristics, men were more likely to be in current smoking status, with higher caffeine intake, high SUA levels, higher waist circumference, serum albumin, BUN, and Scr than women.

Table 1.

Characteristics of the participants based on gender from NHANES (2015–2018).

Characteristic	Total	Gender		P value
		Male	Female
N	7888	3838	4050
Caffeine intake (mg/d)	97.00 (27.50∓193.00)	111.00 (34.12∓217.50)	83.50 (22.00∓171.50)	<0.001
Age (years)	47.87 ± 16.69	47.33 ± 16.49	48.38 ± 16.85	0.0056
BMI (kg/m2)	29.67 ± 7.07	29.46 ± 6.18	29.86 ± 7.80	0.0123
Waist Circumference (cm)	101.04 ± 17.06	103.32 ± 16.27	98.92 ± 17.50	<0.0001
Minutes Sedentary Activity	373.25 ± 198.65	374.42 ± 202.05	372.16 ± 195.43	0.615
PIR	2.12 (1.18∓4.04)	2.16 (1.22∓4.15)	2.40 (0.93∓5.55)	0.005
Marital status				<0.0001
Married or living with partner	65.14	68.7	61.83
Singled, divorced or widowed	34.86	31.3	38.17
Race/Ethnicity				0.2496
Mexican American	8.26	8.56	7.99
Non-Hispanic White	66.59	66.52	66.65
Non-Hispanic Black	10.11	9.5	10.67
Other Races	15.04	15.42	14.7
Education				0.0013
Less than high school	11.08	12.26	9.99
High school or above	88.92	87.74	90.01
Smoking behavior				<0.0001
None	40.74	26.15	54.28
Former	31.8	32.85	30.82
Current	27.46	41	14.9
Work activity				<0.0001
Vigorous	3.52	4.39	2.72
Moderate	27.11	24.49	29.54
Other	69.37	71.12	67.74
Recreational activity				<0.0001
Vigorous	7.98	10.27	5.85
Moderate	27.09	24.53	29.46
Other	64.94	65.21	64.69
Hs-CRP (mg/L)	1.98 (0.84∓4.54)	1.62 (0.77∓3.64)	2.09 (1.12∓3.92)	<0.001
Serum Albumin (g/dl)	4.23 ± 0.36	4.33 ± 0.34	4.14 ± 0.35	<0.0001
Blood Urea Nitrogen (mg/dl)	14.72 ± 5.17	15.61 ± 5.11	13.89 ± 5.10	<0.0001
Serum Creatinine (mg/dl)	0.87 ± 0.31	0.99 ± 0.34	0.76 ± 0.22	<0.0001
Serum Uric Acid (µmol/L)	320.90 ± 83.65	361.11 ± 75.16	283.57 ± 73.17	<0.0001

Continuous variables were shown as mean ± SD (standard deviation) or median (interquartile range, IQR): P value was calculated by weighted Student's t-test or Mann-Whitney U test. Categorical variables were shown as percent (%). P value was calculated by weighted chi-square test. Abbreviations: Hs-CRP, hypersensitive C-reactive protein; PIR, poverty income ratio; BMI, body mass index.

The associations between caffeine intake and SUA by multivariate linear regression analyses and subgroup analyses are shown in Table 2. As can be seen in adjusted models, the negative association between caffeine intake and SUA was presented in model I (adjustment for age and gender) [ß = −0.016, 95%CI: (−0.026, −0.005), P = 0.0042], and in model II (fully adjusted model) [ß = −0.014, 95%CI: (−0.025, −0.004), P = 0.0071] for all participants (Fig. 1a). Similarly, in fully adjusted model, the negative correlation of caffeine intake with SUA remained in men [ß = −0.019, 95%CI: (−0.032, −0.005), P = 0.0086], aged below 60 years [ß = −0.022, 95%CI: (−0.035, −0.010), P = 0.0284], in overweight participants (25 kg/m²≤ BMI <30 kg/m2) [ß = −0.025, 95%CI: (−0.044, −0.006), P = 0.0114] and other races [ß = −0.020, 95%CI: (−0.040,−0.000), P = 0.0496].

Table 2.

Association between caffeine intake and serum uric acid by multivariate linear regression and subgroup analyses.

Subjects (N)	Crude model, ß (95% CI) P	Model I, ß (95% CI) P	Model II, ß (95% CI) P
Total (7888)	0.017 (0.005, 0.029) 0.0044	−0.016 (−0.026, −0.005) 0.0042	−0.014 (−0.025, −0.004) 0.0071
Subgroup analysis stratified by gender
Male (3838)	−0.016 (−0.030, −0.003) 0.0190	−0.016 (−0.030, −0.002) 0.0258	−0.019 (−0.032, −0.005) 0.0086
Female (4050)	0.005 (−0.013, 0.022) 0.5995	−0.011 (−0.028, 0.006) 0.2185	−0.007 (−0.023, 0.010) 0.4260
Subgroup analysis stratified by age (years)
<60 (5207)	0.020 (0.005, 0.034) 0.0084	−0.011 (−0.024, 0.002) 0.0864	−0.015 (−0.029, −0.002) 0.0284
≥60 (2681)	0.005 (−0.015, 0.024) 0.6450	−0.011 (−0.030, 0.008) 0.2553	−0.002 (−0.020, 0.017) 0.8410
Subgroup analysis stratified by BMI (kg/m2)
<25 (2066)	0.005 (−0.014, 0.025) 0.6093	−0.021 (−0.038, −0.004) 0.0153	−0.014 (−0.032, 0.004) 0.1248
≥25, <30 (2521)	0.003 (−0.018, 0.024) 0.7704	−0.027 (−0.046, −0.008) 0.0045	−0.025 (−0.044, −0.006) 0.0114
≥30 (3301)	0.026 (0.007, 0.045) 0.0066	−0.012 (−0.030, 0.005) 0.1578	−0.009 (−0.026, 0.008) 0.2851
Subgroup analysis stratified by Race/Ethnicity
Mexican American (1204)	0.064 (0.023, 0.105) 0.0023	0.016 (−0.022, 0.053) 0.4199	−0.001 (−0.035, 0.036) 0.9732
Non-Hispanic White (2897)	0.008 (−0.008, 0.024) 0.3489	−0.018 (−0.033, −0.004) 0.0145	−0.012 (−0.026, 0.001) 0.0747
Non-Hispanic Black (1645)	0.043 (−0.002, 0.088) 0.0627	−0.016 (−0.058, 0.025) 0.4358	−0.009 (−0.047, 0.029) 0.6368
Other Races (2142)	0.012 (−0.012, 0.035) 0.3201	−0.020 (−0.041, 0.001) 0.0615	−0.020 (−0.040, −0.000) 0.0496

Crude model: non-adjusted (univariate analysis); Model I: age and gender were adjusted; Model II: Model I plus BMI, waist circumference, education, PIR, minutes sedentary activity, marital status, education, Race/Ethnicity, smoking behavior, physical activity (work and recreational), serum albumin, hs-CRP, BUN, and Scr were adjusted; Subgroups stratified by gender, age, BMI, and Race/Ethnicity, were adjusted for other factors in Model II except for the stratification factor itself.

Figure 1.

The relationship between caffeine intake (mg/d) and serum uric acid (µmol/L)

(a) Each black dot represents a subject, and the red dots were connected in a line by multivariate linear regression; (b) The red dotted line represents the smooth curve fit between two variables, and the 95% CI of the red line is an area between two blue dotted lines. The outcomes were adjusted for all the covariates presented in Table 1.

GAM with smooth curve fittings were also performed to estimate the nonlinear associations between caffeine intake and SUA (Figure. 1b and Figure. 2). Interestingly, we used a two-piecewise linear regression model to discover a nonlinear inverse relationship between caffeine intake and SUA in total subjects, which displayed an inverted U-shaped curve with the inflection point of 60.5 mg/d (Table 3). For individuals with caffeine intake <60.5 mg/d, every 1 mg/d increase in caffeine intake was correlated with a 0.11 µmol/L higher SUA [95%CI: (0.03, 0.18)]. By contrast, when caffeine intake >60.5 mg/d, every 1 mg/d increase in caffeine intake was correlated with a 0.03 µmol/L lower in SUA [95%CI: (−0.04, −0.02)]. We also observed inverted U-shaped relationships in both men and women, individuals aged below 60 years, overweight participants, and non-Hispanic withes (Figure. 2). Table 3 shows the threshold effect of inflection points in different subgroups, as measured by a two-piecewise linear regression model. Comparisons with previous studies about the impacts of coffee and caffeine on SUA, hyperuricemia, and gout are shown in Table 4.

Figure 2.

The relationships between caffeine intake and serum uric acid were stratified by gender (a), age (b), BMI (c), Race/ Ethnicity (d). All covariates presented in Table 1 except for the stratification factor were adjusted respectively

Table 3.

Threshold effect analysis of caffeine intake on serum uric acid by using a two-piecewise linear regression

Inflection point (caffeine intake, mg/d)	N	Adjusted ß (95% CI), P value	P nonlinear value (P for log-likelihood ratio test)
Total subjects	7888		<0.001
<60.5	2967	0.11 (0.03, 0.18) 0.0057
>60.5	4921	−0.03 (−0.04, −0.02) <0.0001
Male	3838		<0.001
<29	887	0.43 (0.18, 0.69) 0.0009
>29	2951	−0.03 (−0.04, −0.01) 0.0002
Female	4050		0.015
<101.5	2252	0.06 (0.00, 0.12) 0.0368
>101.5	1798	−0.02 (−0.04, −0.00) 0.0258
Age below 60 years	5207		<0.001
<36	1554	0.21 (0.06, 0.36) 0.0049
>36	3653	−0.03 (−0.05, −0.02) <0.0001
Overweight (25 ≤ BMI < 30)	2521		0.015
<68	980	0.12 (−0.00, 0.24) 0.0539
>68	1541	−0.04 (−0.06, −0.02) 0.0002
Non-Hispanic White	2897		<0.001
<35.5	521	0.45 (0.20, 0.70) 0.0004
>35.5	2376	−0.02 (−0.04, −0.01) 0.0021

Age, gender, BMI, waist circumference, education, PIR, minutes sedentary activity, marital status, education, Race/Ethnicity, smoking behavior, physical activity (work and recreational), serum albumin, hs-CRP, BUN, and Scr were adjusted. For the subgroups of both male and female, aged below 60 years, overweight and non-Hispanic white, the model adjusted for other factors except for gender, age, BMI, and race/ethnicity, respectively.

Table 4.

Summary for the effect of coffee and caffeine on SUA, hyperuricemia and gout.

Studies	Study design	Country	Number	Mean age	Gender	Population	Coffee consumption range	Results or Conclusions
Pham et al (12).	Cross-sectional	Japan	11,662	62	43% male	The community population	Quintiles of coffee cups (0, 1–2, 3–4, 5–6, ≥7) /day	Negative relations of coffee-SUA and coffee-hyperuricemia in male, but not in female.
Kiyohara et al (13).	Cross-sectional	Japan	2,240	52	100% male	Self-defense officials	Quartiles of coffee cups (<1, 1–2, 3–4, ≥5) /day	Negative in the coffee-SUA relation
Choi et al (14).	Cross-sectional	American	14,758	45	48% male	The NHANES-III Study	Quintiles of coffee cups (0,<1, 1–3, 4–5, ≥6) /day	Negative in the coffee-SUA relation, no correlation in caffeine-SUA
Bae et al (38).	Cross-sectional	Korean	9,400	62	37.9% male	The community population	Quintiles of caffeine intake (0-339.2mg) /day	Positive relation of caffeine-SUA in female, no correlation in male.
Choi et al (45).	Prospective	American	45,869	54	100% male	The Health Professionals Follow-up Study	Quintiles of coffee cups (0, <1, 1–3, 4–5, ≥6) /day	Negative in the coffee-gout relation, no correlation in caffeine-gout
Choi et al (46).	Prospective	American	89,433	46	100% female	The Nurses' Health Study	Quartiles of coffee cups (0, <1, 1–4, >4) /day	Negative in the coffee-gout relation
Current study	Cross-sectional	American	7,888	48	48.7% male	The NHANES 2015–2018 Study	M (IQR) of caffeine intake 97.00 (27.50-193.00) mg/day	1- Negative relation of caffeine-SUA in male, but not in female. 2- The inverted U-shaped relations of caffeine-SUA are in both genders.

SUA, serum uric acid; M (IQR), median (interquartile range)

We also evaluated weighted baseline characteristics of the population and depicted SUA levels by quartiles of caffeine intake (Supplementary Table S1, Figures S1-S2). To detect linear trends in subgroups, multivariate linear regression analyses by quartiles of caffeine intake were performed (Supplementary Table S2). However, we did not find any statistical significance in linear regression by trend tests, which allowed us to explore the possible nonlinear relationship between caffeine intake and SUA.

Discussion

Our primary purpose in this cross-sectional study was to investigate the relationship between caffeine intake and SUA, and explore whether there is a nonlinear relationship. Given that high SUA levels have been reported to be associated with cardiovascular disease mortality (31), and coffee consumption is inversely correlated with the risk of death from cardiovascular disease and all-cause death (29), we hypothesize that caffeine intake might regulate SUA and benefit cardiovascular health in a way. The present study elucidated that caffeine intake was negatively correlated with SUA after adjusting for confounders when caffeine intake was up to the inflection point (60.5mg/d), indicating a nonlinear inverse relationship between caffeine intake and SUA.

Furthermore, subgroup analyses were performed to make better use of the data as recommended by the STROBE statement (32). In our subgroup analyses by multivariate linear regression, the significant inverse associations were mainly observed in males, individuals aged below 60 years old, overweight participants, and other races. On GAM with smooth curve fitting, we found this association presented with nonlinearity after adjusting for confounders in all participants (P nonlinear <0.001). This nonlinear relationship could also be found in some subgroups, for instance, stratified by gender, the inflection points were 29 mg per day for men and 101.5 mg per day for women. The inflection points differed in subgroups may be due to inconsistent intake of caffeine and levels of SUA at baseline.

CYP1A2, the main enzyme responsible for caffeine clearance, is more active in men than women (6). CYP1A2 also metabolizes estrogen (33), and estrogen is a competitive inhibitor of caffeine metabolism (34). The gender difference in caffeine-SUA relation may be due to the differences in CYP1A2 enzyme activity and sex hormone. Age and obesity appear to have no effect on caffeine clearance, whereas caffeine half-life increases in age- and obesity-related manner (35, 36). For different race/ethnic groups, differences in the caffeine-SUA relationship may reflect variations in caffeine metabolic rates and genetic variability on physiological responses to caffeine (37). Overall, caffeine intake, absorption, metabolism, and physiological and functional effects appear to be influenced by various exogenous and endogenous factors, such as diet, smoking, drugs, age, sex (hormonal status), obesity, and genetic background (6). These factors may result in the various effects of caffeine on SUA, showing different inflection points in stratified subgroups.

Similar to this study, Kiyohara C et al. (13) conducted a cross-sectional study with 2,240 Japanese men and found that coffee consumption was inversely related to SUA. According to another research conducted by Pham et al. (12), increased coffee consumption was associated with lower SUA levels and a lower incidence of hyperuricemia in men than in women, which is consistent with the results of our study. In contrast, a study (38) on 9,400 Korean participants (37.9% male) showed caffeine intake might affect SUA levels in women. These conflicting results suggested that caffeine might play a role in regulating SUA levels, and its effect varied with gender. In our research, multivariate linear regression in subgroup analysis identified that higher caffeine intake was independently associated with decreased SUA in males, which, however, was not observed in females.

Caffeine, chemically, is a methylxanthine (1,3,7-trimethylxanthine), which is the substrate of cytochrome P450 (CYP) enzymes (CYP1A2 in particular). Caffeine metabolites include paraxanthine (main), theophylline and theobromine (smaller amount), all of which could be eventually converted to uric acid and excreted renally (6). This seems to indicate that SUA might be elevated with increased caffeine intake, however, the results are not the same. An experimental study in rats was tested in vitro and in vivo, demonstrating that methylxanthine could be considered as a xanthine oxidase inhibitor (39), which provided evidence that the metabolites of caffeine (paraxanthine, theophylline, and theobromine) might have an inhibitory effect on the conversion of SUA and reduce the risk of hyperuricemia in humans. In another study (40), theobromine was found to inhibit uric acid nucleation and crystal growth. Besides, high caffeine intake can exert a diuretic effect when resting (41). Thus, they might decrease SUA levels by increasing excretion in the urine. Given that one-third of uric acid is excreted through the intestine (7), and coffee has been reported to have beneficial effects on the composition of the gut microbiota (42), we speculate that caffeine may also promote uric acid intestinal excretion to reduce SUA. From all the above, on the one hand, caffeine metabolites could be converted into uric acid to increase SUA levels, and on the other hand, they can inhibit the conversion of uric acid or promote its excretion, which appears to be consistent with our results of nonlinear negative correlations between caffeine intake and SUA.

Generally, a large body of evidence suggested that daily moderate coffee consumption has been correlated with a lower risk of a variety of chronic diseases, including coronary artery disease and stroke (43), type 2 diabetes (9), Parkinson's disease (44), as well as gout (45, 46). Simultaneously, consumption of caffeinated coffee in U.S. adults is unrelated to the increased prevalence of major chronic diseases, such as cancers and cardiovascular diseases (28). During the past few decades in the U.S., the prevalence of hyperuricemia has been increasing rapidly (47). Hyperuricemia is correlated with various medical conditions, including obesity, chronic kidney disease, and hypertension, and serves as the primary cause of gout (48, 49). Coffee consumption is associated with a lower risk of death from cardiovascular disease (29). SUA levels, on the other hand, are inversely related to the risk of cardiovascular mortality (31). It appears that some association between caffeine and SUA may be established.

Specifically, moderate coffee consumption, usually defined as 3∓5 standard cups per day (caffeine intake ≤400 mg/day in adults or ≤200 mg/day in pregnant women), has been associated with a reduced risk of multiple chronic diseases (28). However, side effects of caffeine intake at high doses (>200 mg per occasion or >400 mg per day) include prolonged sleep latency, insomnia, anxiety, restlessness, nervousness, excitement, rambling flow of thought, and psychomotor agitation (50). In our study, all inflection points of the inverted U-shaped trajectories were located on the left side of the moderate coffee consumption. Thus, coffee consumption in the range between the inflection point and moderate caffeine intake may be associated with reduced SUA and prevent several chronic diseases. Of note, it should be alert that caffeine intake below the inflection point may have the risk of increased SUA.

We observed that SUA increased first and then decreased (inflection point of 60.5 mg/d) with increasing caffeine intake for all participants in our study. However, most previous studies focused on coffee consumption and SUA demonstrated that a higher frequency of coffee consumption might reduce SUA levels, and few identified the nonlinear relationship between them (12, 13, 15). In addition, other sources of coffee, such as chlorogenic acid (51), may have biological effects on SUA by inhibiting xanthine oxidase (52). In the third NHANES (NHANES-III) study (14), the results indicated no statistical significance in linear relationship between caffeine intake (quintiles) and SUA levels (P-trend = 0.14). Another study conducted by Choi, H.K. et al. (45) revealed a trend of lower risk of gout with higher caffeine intake for males, but this was not statistically significant. Although epidemiological evidence suggests a relationship between caffeine intake and SUA level (14, 45, 46), the regulation mechanism has not been determined (15).

To elucidate the nonlinear caffeine-SUA relationship, we selected a two-piecewise linear regression. First, a two-piecewise linear regression was presented concisely for the trend of smooth curve fittings, which showed a sensible degree of discrimination by the inflection point. Second, if we use three or more piecewise regressions, the sample size in the segments may be reduced, resulting in poor statistical power. Third, we found a two-piecewise linear regression easier to explain for the physiological efficacy of caffeine on SUA, satisfying the trend of first increasing and then decreasing. Nevertheless, it would be helpful to compare the pros and cons of different piecewise linear regressions for the underlying mechanism of the caffeine-SUA relationship in future research.

In this study, there are several strengths and limitations. The study participants were representative and distributed nationwide, as well as the sample size was large. We used two 24-hour dietary recalls to get the average caffeine intake data that reduced the random measurement errors. In addition, we used the GAM to address nonlinearity. To our knowledge, we are the first to provide evidence for a nonlinear inverse relationship between caffeine intake and SUA in U.S. adults. However, we are supposed to acknowledge some limitations. No causality could be inferred since the nature of the cross-sectional study. Caffeine is a component of coffee or tea, and the effects on SUA of other compounds in coffee or tea cannot be determined. The NHANES does not include drug information that may interfere with SUA. Therefore, the bias caused by drug factors and other unmeasured confounding factors are not excluded. Besides, the SUA levels of the study population were mainly in the physiological range, and the nonlinear relationship might not apply to the patients with hyperuricemia or gout. Finally, men had a significantly higher SUA than women at baseline, but needed less caffeine to lower SUA. This gender difference should be identified in future studies.

Conclusions

In conclusion, our research observed a nonlinear negative association between caffeine intake and SUA in U.S. adults, represented by an inverted U-shaped curve. This relationship suggests that SUA might decline when caffeine intake reaches a certain amount. The influencing mechanisms of caffeine for SUA and gender differences require further study.

Authors' contributions

Statistical analyses and drafting of the manuscript: AB.L. and C.J. Data collection: QJ.L., H.Y. and HF.Z.

Concept and design

H.M. and QS.G. All authors read and approved the final manuscript.

Funding

This work was supported by grants (DFJH2020003, DFJH201922, and DFJH201811) from the Guangdong Provincial People's Hospital with High-level Hospital Construction Project, (No. 20201008) from Traditional Chinese Medicine Bureau of Guangdong Province, (No. 2019118152336191) from Guangdong Medical Science and Technology Research Foundation, and (No. 8160284) from National Natural Science Foundation of China.

Ethical Approval

Not applicable.

Conflicts of Interest

No disclosures were reported.

Electronic Supplementary Material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-022-1767-y and is accessible for authorized users.

Supplementary material, approximately 23.8 KB.

Supplementary material, approximately 347 KB.

Supplementary material, approximately 374 KB.