Serum Uric Acid : Correlation with Biological , Clinical and Behavioral Factors in Japanese Men

Cross-sectional associations between biological, clinical and behavioral factors and serum uric acid (SUA) levels were examined in 2,438 Japanese male office workers aged 20 to 59 years in Osaka, Japan. Stepwise regression analysis for SUA was carried out for all persons and repeated excluding those under medication for hypertension, hyperuricemia or diabetes mellitus. The results were essentially the same without change in the sequence of the seven most important variables. When 150 men under medication were excluded, independent correlates with SUA levels were, in order of relative importance, history of gout, log triglyceride, creatinine, hemoglobin A,. (negative association), body mass index, total protein, alcohol intake , age (negative association), and total cholesterol. 32.7 percent of total variation in SUA was accounted for by these variables combined. Our data suggest that weight and serum lipids control and avoiding excessive drinking may be beneficial in the prevention of hyperuricemia. J Epidemiol, 1999 ; 9 : 99-106

Japanese dietary habits are characterized by a higher intake of carbohydrates and a lower intake of fat and protein than those in western countries 15,16,17) However, Japan has recently experienced drastic changes in dietary habits with the rapid westernization of lifestyle, and the intake of animal fat and protein and alcohol consumption have increased remarkably.According to the National Survey of Circulatory Disorders in 1990 18), the mean SUA concentration among male adults was estimated to be 5.6 mg/dl (standard deviation: 1.3 mg/dl) and only 4.2 percent of all observed SUA levels were 8.0 mg/dl or more.However, recent changes in lifestyle warrant concern .
Although secondary prevention with the focus on diagnosis and treatment of hyperuricemia and gout is still the goal of current clinical efforts, long-term antihyperuricemic therapy carries both costs and risks 3,19).Therefore, the eventual public health goal should be the primary prevention of hyperuricemia .
In this report on a cross-sectional population study based on annual health examinations at the workplace , we have made an attempt to identify the relative importance of biological , clinical and behavioral factors in the association with SUA among Japanese men.

MATERIALS AND METHODS
A survey to evaluate the factors related to SUA levels was conducted in 1996 among employees of T Corporation, which is one of the biggest building contractors in Osaka, Japan.The participants in the 1996 survey consisted of 2,438 Japanese male office workers aged 20-59 years, and the participation rate was 99.8%.
The survey included a questionnaire, physical examinations, blood pressure measurement, and collection of blood samples for laboratory analysis.Data on alcohol intake and smoking habits were obtained by interview.The subjects were asked to quantify their consumption of alcohol in terms of what kind of alcoholic beverage they drank, how much they drank per day and how often they drank per week.Alcohol intake was then converted into the equivalent number of go, a traditional Japanese unit of volume for sake (Japanese rice wine) (1 go = 180 ml and contains 23 g of ethanol).The number of cigarettes currently smoked per day was noted for smokers.Medical history and history of use of prescribed drugs of each subject was also assessed by examining physicians.Weight and height were measured with the subjects in typical indoor clothing but without shoes.Body mass index (BMI) was calculated as weight/height 2 (kg/m2).Glycosuria and proteinuria were measured with the urinalysis test.Blood pressure was measured to the nearest 2 mmHg with a standard sphygmomanometer on the right arm of subjects sitting after 5 min rest, and Korotkoff phases I and V were taken as representing systolic and diastolic blood pressure, respectively.After the measurement of the blood pressure and while the participant was fasting, blood samples were drawn from an antecubital vein.SUA concentrations were determined according to a uricase method 20 ) with an auto analyzer Olympus AU-5200 (Olympus Japan Co, Ltd., Tokyo, Japan) by the Nihon Clinical Laboratories Inc. (Tokyo, Japan).Biochemical measurements made by the Olympus AU-5200 also included determinations of total cholesterol, high density lipoprotein (HDL) cholesterol, triglyceride, total protein, creatinine and blood urea nitrogen (BUN).Hemoglobin A, (HbA,,) and hematocrit were determined with a JOOKOO's automated glycohemoglobin analyzer system HS-8 (Jookoo Co, Ltd., Tokyo, Japan) and an auto analyzer Sysmex E-4000 (Toa Medical Electronics Co, Ltd., Tokyo, Japan), respectively.Continuous variables were classified into five sub-categories approximating to quintile and discrete variables into three sub-categories according to qualitative differences.Regarding creatinine level, cigarette smoking and alcohol intake, subjects were divided into five classes but in unequal numbers.
For statistical assessment, comparisons across age group means of SUA were made using analysis of variance.Analysis of covariance was used to examine SUA levels according to categories of selected factors adjusting for age.Multiple regression analysis was performed to examine an independent association of factors and their relative importance as determinants of SUA levels.In the statistical analyses, the logarithm for the triglyceride (because of the non-gaussian distribution of the frequency for this variable) was used.In multiple linear regression analyses, exact values were used for continuous variables.For discrete variables, estimates of relative importance compared with the reference level of each subclass were calculated by creating two dummy variables for each variable as follows: x1=0, x2=0 for the reference level (negative for history or urinalysis); xl=1, x2=0 for the second level (positive without medication for history, trace for urine protein, and 1+ or more for urine glucose); and xl=0, x2=1 for the third level (positive with medication for history, 1+ for urine protein, and 2+ or more for urine glucose).
Data analysis was performed with the SPSS/PC statistical package (Marija J. Norusis, SPSS Inc., Chicago, IL, USA).All reported P-values are two-tailed and a P-value of less than 0.05 was considered significant.

RESULTS
Table I shows the biological, clinical and behavioral features of 2,438 Japanese men aged 20-59 years.The mean SUA level was 5.83 mg/dl with the standard deviation of 1.29 mg/dl, and the frequency distribution of SUA levels showed a unimodal curve with a median of 5. Table 3 shows interclass differences of age-adjusted mean SUA levels among sub-categories of continuous variables.Creatinine had the strongest positive association with SUA.Triglyceride and BMI were the second and third most significant variables showing a clear trend of positive association, respectively.Other continuous variables showing a significant positive association with SUA include total protein, alcohol intake, systolic and diastolic blood pressures, total cholesterol, BUN, and hematocrit.On the other hand, HDL-cholesterol showed a negative association with SUA.Cigarette smoking and HbA1c did not show a consistent association with SUA.
Table 4 shows interclass differences of age-adjusted mean SUA levels among sub-categories of discrete variables.The most significant differences of SUA was noted for history of gout, and those with history but without medication showed the highest values.History of renal disease or hypertension had a positive association with SUA, and those with history and medication showed the highest values.On the other hand,  urine glucose and history of diabetes mellitus showed a clear trend of negative association with SUA.Urine protein did not show a significant association with SUA.
Table 5 shows the results of stepwise regression analyses for SUA.The analysis was carried out separately once for all persons and repeated excluding those under medication for hypertension, hypenricemia or diabetes mellitus.The results were essentially the same without change in the sequence of the seven most important variables, and the cumulative percentage of variation was quite similar (R2=0.328for all persons and R2=0.327 after excluding 150 men under medication for hypertension, hyperuricemia or diabetes mellitus).When the sub-jects under medication for hypertension, hyperuricemia or diabetes mellitus were excluded, independent and significant correlates with the SUA levels were, in order of relative importance; history of gout, log triglyceride, creatinine, HbA,.(negative association), BMI, total protein, alcohol intake, age (negative association), and total cholesterol.On the other hand, the correlation between SUA and systolic and diastolic blood pressure, HDL-cholesterol, BUN, hematocrit, number of cigarettes per day or glycosuria turned out to be insignificant when the other variables were taken into consideration.DISCUSSION Table 5. Stepwise regression analysis of serum uric acid (mg/dl) on selected variables * * Also tested in the model but not contributing significantly at the 5% level were: systolic blood pressure , diastolic blood pressure, High-density lipoprotein cholesterol, blood urea nitrogen, hematocrit, proteinuria, number of cigarettes per day and history of hypertension, diabetes mellitus and renal disease.Percentage of variation accounted for = R2X100.R: multiple correlation.
In this study, history of gout was the strongest predictor of the levels of SUA among all persons examined before and after excluding those under medication for hypertension, hyperuricemia or diabetes mellitus, and accounted for about 23 percent of the variation in SUA.These results indicate that the level of SUA is higher among those with past history of gout than among those without and that hyperuricemia is not well controlled among those with past history of gout.
A number of clinical studies have asserted that hyperuricemia or gout is positively associated with triglyceride 21,22,23).Not many population studies have addressed the relation between triglyceride and SUA.A strong correlation was demonstrated in studies of French men 6) and of Japanese men in Hawaii 8) and in Japan 12).As for total cholesterol, epidemiological studies are inconsistent as to the relation between total cholesterol and SUA [4][5][6][7][8][9][10][11][12].In a study of French men, SUA was positively related to total cholesterol after adjusting for body fat, but not after controlling for triglyceride 6).A study of Japanese men in Hawaii also found a positive relation between total cholesterol and SUA in the univariate analysis, but not in the multivariate analysis controlling for BMI, triglyceride and other variables 8).Kono et al 12) have recently reported that total cholesterol was an independent cor-relate of SUA levels after controlling for BMI, triglyceride and other potential factors.On the other hand, evidence is sparse as to the relation between HDL-cholesterol and SUA.In a study of Israeli men 9), a significant negative association with HDLcholesterol was found in a multiple regression analysis including total cholesterol and other covariates but not triglyceride.Kono et al 12) found a weak but independent negative relation between HDL-cholesterol and SUA.In this study, log triglyceride was the strongest correlate of SUA among biological and behavioral variables under this study, and total cholesterol also had a weak but independent association with SUA .HDL-cholesterol showed a negative correlation with age-adjusted SUA levels, but did not remain as statistical significance in the multivariate analysis.These results of our study indicate that the level of triglyceride was the most significant and independent predictor of SUA levels among biological variables, whereas an independent contribution of total cholesterol and HDL-cholesterol were much smaller compared with that of triglyceride.
Creatinine, BUN and history of renal disease showed a positive association with age-adjusted SUA levels .Creatinine remained as statistical significance in the multivariate analysis.The strong association between creatinine and SUA is in agreement with another observation based on the popula-tion 10,11,12).Cross-sectional data can not address the causal relationship, but prospective studies have found no renal deterioration resulting from aymptomatic hyperuricemia 3,24).Asymptomatic hyperuricemia may simply reflect insidious renal deterioration; alternatively both hyperuricemia and raised creatinine levels may be linked with unknown common factors.
As for the relationship between obesity and SUA, both absolute and relative weight have been shown to be positively associated with SUA levels in population studies 8,9).Our study confirmed that BMI had a significant and independent impact on SUA levels.It may be important enough to promote the health education program for appropriate body weight at the workplace, as shown by weight control in therapeutic trials 25).
Higher levels of SUA have been observed among those with prediabetic status or impaired glucose tolerance 11-14),while lower levels of SUA have been observed in diabetics 6-13).In the range of glucose concentrations where glycosuria rarely occurs, the British Regional Heart Study showed a positive relation between glucose and SUA, independent of BMI, alcohol intake and creatinine 13).In the present study, HbA1c levels, glycosuria and history of diabetes mellitus had a significant and negative correlation with age-adjusted SUA levels.HbA1c remained as an independent negative factor in the multivariate analysis even when the subjects under medication for diabetes mellitus were excluded.These results indicate that high HbA1c levels may lower SUA levels by enhancing renal excretion of uric acid possibly by inhibiting reabsorption in the proximal tubuli [11][12][13][14].
Alcohol may increase uric acid production by enhancing the turnover of adenine nucleotides 26).A decreased renal clearance of uric acid may also be linked to hyperuricemia in excessive alcohol consumption through alcohol-induced elevations in blood lactate levels 27).Additional causes are alcohol related hyperlipidemia, fat-dependent ketogenesis and starvation 28).The present study corroborated an independent association between alcohol intake and SUA.Alcohol intake in Japan, especially among men, is high and has been gradually increasing 'T.Considering the beneficial effect of a cessation in alcohol intake on uric acid 29), such a reduction might prove to be a factor in achieving the primary prevention of hyperuricemia for this population.
Total protein was also demonstrated to be an independent correlate of SUA in this study.This finding is in agreement with another population study in Japan 10).We have no clear explanation for this finding, but the association is probably due to the same causes, derived from the consumption of large quantities of high protein diets 30).
It is uncertain whether untreated hypertension is related to an increased level of SUA in the absence of overt renal disease.Several studies have found a positive association between untreated hypertension and SUA 5,8,9,12).On the other hand, other studies have suggested that a positive association between raised blood pressure and SUA can be ascribed largely to obesity or alcohol intake 6,31).An association between untreated hypertension and hyperuricemia may also be due to unrecognized early renal dysfunction 32).In the present study, systolic and diastolic blood pressure was significantly associated with elevated age-adjusted SUA levels, but did not correlate with SUA levels independently of other potential factors.As for the effects of antihypertensive drugs on SUA, it is well known that the administration of long-term diuretic therapy increases SUA levels in hypertensive patients 33,34).On the other hand, calcium antagonists, angiotensin II receptor antagonists, and adenosine-1 blockade, which have been used extensively in Japan as first-line antihypertensive agents 35), have been recently reported to lower the SUA concentration, suggesting a favorable effect on uric acid metabolism in patients with hypertension 36-39).In the present study, hypertensive persons with antihypertensive medication showed the higher ageadjusted SUA levels than those without, but use of hypotensitve drugs did not remain as statistically significant, controlling for other potential factors.This may in part be due to the fact that most of the subjects on hypotensive medication in this study were taking not diuretics but other hypotensive drugs, including calcium antagonists or angiotensin II receptor antagonists.
Some studies have reported a reduced level of SUA among cigarette smokers without showing a clear dose-response relation 6, 9,12) while others failed to find such an association 8,10).In the present study, cigarette smoking did not show a consistent association with age-adjusted SUA levels, and did not remain as statistical significance in the multivariate analysis.It remains to be answered if and how usage of tobacco affects SUA levels, but associations with cardiovascular diseases, irrespective of hyperuricemia, warrant the promotion of cessation of smoking 40).
Our findings indicate the associations of the levels of SUA with biological and behavioral factors and highlight the clinical importance of determining hyperuricemia in persons with other metabolic disorders and vice versa.Underlying mechanisms remain to be elucidated as to the strong association of SUA with triglyceride, creatinine and total protein.

Table 1 .
Characteristics of 2,438 men aged 20 to 59 for selected variables go= 180ml Japanese sake

Table 2 .
Mean levels and standard deviations of serum uric acid (mg/dl) by age group

Table 3 .
Interclass differences of age-adjusted mean uric acid levels and significance test for continuous variables * Q1-Q5 indicate the first (lowest) to the fifth (highest) quintile values of each continuous variable , respectively.In alcohol intake, Q1-Q4 indicate the first to the fourth quartiles, respectively, excluding non-drinkers (none).Parentheses show age-adjusted mean uric acid levels.

Table 4 .
Interclass differences of age-adjusted mean uric acid levels and significance test for discrete variables * Parentheses show age -adjusted mean uric acid levels .