Does hyperuricemia affect mortality? A prospective cohort study of Japanese male workers.

A positive association between hyperuricemia and cardiovascular disease has been reported, but no study has evidenced yet the precise role of serum uric acid in the development of cardiovascular disease. In addition, no epidemiological studies have so far documented a decreased risk of cancer among people with hyperuricemia, even though the antioxidant action of uric acid has recently been stressed to inhibit DNA damage. The present prospective cohort study investigates the relationship between hyperuricemia and health hazards in a Japanese working population. The subjects were 49,413 Japanese male railroad workers, aged 25-60 years at enrollment. Serum uric acid and other baseline data were provided by annual health-survey records from 1975 to 1982. The vital status of the subjects was traced until the end of 1985 for those who remained alive. During an average 5.4-year study period, 984 deaths were recorded. Those with serum uric acid over 8.5 mg/dl showed elevated relative risks (RRs) of death in all causes (RR 1.62, p<0.01), coronary heart disease ( RR 1.52), stroke (RR 2.33, p<0.01), hepatic disease (RR 3.58, p<0.01), and renal failure ( RR 8.52, p<0.01), as compared with those with serum uric acid levels of 5.0-6.4mg/dl. The RR of death in all causes still remains statistically significant when adjusted by age and serum total cholesterol (2.00, p<0.01), age and alcohol intake (1.85, p<0.001), age and smoking (1.69, p<0.001), age and gout treatment (1.61, p<0.05), and also age and BMI (1.50, p< 0.05). On the other hand, the RR of all causes decreased but was still above 1.0 when adjusted by age and blood glucose (1.62), age and systolic blood pressure (1.32), age and GOT (1.23), and also age and history of cardiovascular disease (1.17). These results showed that hyperuricemia has a strong association with the RRs of death in all causes, coronary heart disease, stroke, hepatic disease and renal failure, and indicated that serum uric acid seems to be a considerable risk factor for reduced life expectancy.

for a few reports in which the relationship between SUA and stroke [11][12][13] was investigated. The author's previous study 14) on 7,200 Japan Railways workers revealed, as a result of multivariate analyses, that the history of gout is a significant independent risk factor for CHD. As for cancer, an epidemiological paper 15) has reported that a raised SUA level is a risk factor for cancer, while a biological study 16) has stressed the antioxidant role of uric acid inhibiting DNA damage.
The present study aims to investigate the relationship between hyperuricemia and total health hazards, based on a 10year follow up, averaging 5.4 years, of a cohort of a Japanese male working population.
The characteristics of this study population was, large in scale and relatively young in age.

SUBJECTS AND METHODS
The study population consisted of 49,413 Japanese male railway workers ranging in age from 25 to 60 years at the time of enrollment, which covered the period from 1975 through 1982. The health information on the subjects was obtained from yearly medical examinations, which included body height, body weight, blood pressure, blood count, serum chemistry, ECG, habits, history of diseases, and other factors. Over 99% of railway workers annually underwent a health survey. In case a worker had two or more records, we chose the latest record. The vital status was obtained through the company's health records for employees and the company's pension records for the retired, with underlying causes of death according to the 9th version of International Code of Disease(ICD).
SUA was determined through a multichannel Hitachi 400 analyzer using the phospho-tungstic acid procedure from 1975 through 1977 and then a Hitachi 716 analyzer using the uricase peroxidase assay 17-18) from 1978 through 1982. The precision of the serum chemistry measurement was examined using precision management serum. When the test method was changed from one to another, we checked the reappearance of the data, the linearity of the data by the new method, the correlation of the data by new and old procedure, the effect of the coexisting substance, the daily difference, and the new and old procedures were carefully compared for the sake of consistency.
Person-years were calculated for each subject from the start of the follow up to the termination of the follow up on December 31, 1985, or the time of death for the diseased. Since 1985, Japanese National Railways was made up of six regional companies for passenger service and five national companies for freight, telecommunications for JR group members. So we ended the follow up of the subjects.
The person-years and the follow-up number of deaths were aggregated and stratified by sex, attained age, follow-up intervals and factors for analyses, using DATAB computer program 19). AMFIT regression program was used for survival analysis with Poisson regression analyses 20).

The distribution of SUA level
The SUA levels at the baseline formed a nearly normal distribution, slightly skewed to higher values, with a mean of 5.77 1.33 mg/dl (Fig.1). For further analysis, SUA levels at the baseline were categorized into four groups; 1) low (0.3-4.9mg/dl), 2) moderate (5.0-6.4mg/dl), 3) slightly high (6.5-8.4mg/dl), and 4) high (over 8.5mg/dl). The range of group 2, which was often clinically regarded as a reference level, corresponds to the mean 1/2SD in this study. The number of the subjects and person-years by the SUA level category are shown in Table 1.

The characteristics of subjects by SUA category
As SUA is correlated with several other factors closely, the characteristics of subjects by SUA category are shown in Tablet. These items are selected by Trend Test at level significance p<0.001.  (1.9%) were recorded. The underlying causes of death and the number of deaths are shown in Table 3 according to the SUA category at enrollment. In this study, CHD was defined as acute coronary myocardial infarction (ICD 410-414) and "other heart disease" including chronic heart failure and valvular diseases ( Age-adjusted relative risk by SUA category Age-adjusted relative risk (RR) and 95% confidence interval were calculated against the moderate SUA level group by causes of death ( Table 4). The age-adjusted RR for all causes was 1.17 in the slightly high-level group and 1.62 (p<0.01) in the high-level group (Fig 2). The high-level group had the following relative risks for specific diseases: 1.52 in CHD, 2.97 (p<0.01) in other heart disease, 2.33 (p<0.01) in stroke, 8.52 (p<0.01) in renal failure, and 3.58 (p<0.01) in hepatic disease. To the contrary, the RR for all cancer was not elevated, being 0.6 in the high-level group.
Relative risk of all causes of death, when adjusted by age and specific factors Relative risks and 95% confidence interval of all causes of death by SUA category, when adjusted by various factors, are shown in Table 5. The RR of the following indicies still remained statistically significant, when adjusted by age and serum total cholesterol (2.00, p<0.01), age and alcohol intake (1.85, p<0.001), age and smoking (1.69, p<0.001), age and history of gout (1.61, p<0.05), and also age and BMI (1.50, p< 0.05). On the other hand, RRs decreased in the following indices; when adjusted by age and blood glucose(1.62), age and systolic blood pressure (1.32), age and GOT(1.23), and also age and history of cardiovascular disease (CVD) (1.17).

DISCUSSION
The present study aimed to investigate the relationship between the SUA baseline data and specific causes of death through a prospective cohort study of 49,413 Japanese male workers. The follow-up period ranged from 3 to 10 years after enrollment.
The observation revealed that the RR of all causes of death is lower in SUA groups below 6.4 mg/dl than in groups over 6.5 mg/dl. This result was consistent with the previous report by Freedman et al. 6), who traced 5,421 people for 12 to 16 years, and reported that RR in all causes of death was the lowest in those with SUA 5.0 to 5.9 mg/dl. The authors tried to adjust the RR of all causes of death by several specific factors (Tables). As an example, RR remained significant after the adjustment by BMI or total cholesterol. Moreover, when adjusted by the history of CVD, blood pressure, blood glucose, and GOT, RR showed a decrease but still remained above 1.0. Our results indicate a possible role of SUA in life expectancy. If other risk factors such as hypertension or poor medication of circulatory disease are improved in addition to SUA, the RR of total death shoud be expected to decrease. For the clarification of if the increase of RR in the SUA groups over 6.5ng/dl is the direct effect of SUA or not, the intervention study may be needed to control the SUA level and clarify the decrease of death.
The RR of CHD increased linearly with the elevation of SUA, though no significant difference was noted between baseline SUA groups. This finding coincides with that of Framingham study [1][2][3][4], which reported the RR of CHD being about 1.3 in those with hyperuricemia. It is noted that the Framingham study involved a general population while the present study dealt with a relatively young working population.
The biological mechanism of SUA and increased CHD risk is not yet certain, but there are several possibilities about the relationship of SUA and arteriosclerosis. An in vitro study has reported that monosodium urate crystals cause abnormal platelet functions, i.e. , the shortened platelet life span, and the increased platelet aggulutination 21). Another study by Kellermeyer observed that uric acid crystals absorb and activate the Hageman factor 22) . This causes shortness of the hemoagglutination time. Rao et al. 23) suggested that hyperuricemia causes the proliferation of smooth muscle cells in blood vessels, followed by the disturbance of blood flow.
In relation to arteriosclerosis, Hiyamuta et al.24) w made a case control study of 1,029 CHD cases to examine the relationship between arteriographic features of coronary atherosclerosis and risk factors for coronary diseases. They concluded that SUA is a significant risk factor for coronary atherosclerosis in all types of coronary stenosis.
One of the reasons that the mechanism or the role of SUA in the development of CHD in clinical studies remains uncertain is the strong correlation between SUA and other risk factors such as hypertension, obesity, hyperlipidemia, and use of thiazide diuretics, which are considered to be attributable to CHD. As shown in Table2, SUA level is correlated with other CHD risk factors closely (p<0.001). Reaven 25) suggested the possibility that resistance to insulin-stimulated glucose uptake and hyper insulinemia are involved in the etiology and clinical course of three major diseases: noninsulin-dependent diabetes mellitus(NIDDM), hypertension, and coronary artery disease(CAD). He suggested that there is a series of related variables(Syndrome X ) that tends to occur in the same individual and may be of importance in the genesis of CAD. After that, several similar conceptions are proposed, and hyperuricemia is included as one of the risk factors of atherosclerosis in Syndrome X plus 26). Wannamethee et al 10) suggested that serum urate is not a truly independent risk factor for coronary heart disease. Raised serum urate appears to be an integral part of the cluster of risk factors associated with the insulin-resistance syndrome. From our study, SUA may be a risk factor for CHD, but contribution of SUA may not be large. CHD may be prevented by completely controlling not only SUA but also these other risk factors.
When the relationship between stroke and hyperuricemia was observed, the mortality from stroke increased as SUA level increased (Table 4). It is noted that RR was significantly low in the low SUA level group, showing a linear increase of RR as SUA level rises. A prospective cohort study on 2,710 Japanese-Americans by the Honolulu Heart Program revealed a significant positive association between SUA and atherosclerotic occurence 27). Vigna 13) reported a positive correlation of SUA to the extracoronary atherosclerosis only in females. It is certain from the present study that SUA is a risk factor for stroke as is hypertension.
In regard to cancer death and SUA, there have been in vitro studies and hypotheses which reported the antioxidant activity of uric acid 16,[28][29][30]. Numerous clinical reports have pointed out the association of hypouricemia with malignant neoplasms, one report indicated a reduction in the pre-and post-secretory reabsorption of urates 31). However, some studies formed no evidence for protective associations between the levels of SUA and the risk of cancer 15,32-33).
The present study subjects are a healthy working population below 60 years of age at the time of enrollment, and the RR shows interestingly a linear decrease in cancer as SUA level rises, although the trend was not statistically significant (Table4). Further analyses may be needed to clarify the roll of hypouricemia as a cancer suppressor.
In the present study (Table 4), the cases with hepatic disorders had the lowest RR of 0.78 (not significant) in the slightly high SUA level group. The RR increased in the low-and highlevel groups, showing RRs of 1.82 and 3.58, respectively. Hypouricemia has been noted in patients with hepatocellular disease [34][35][36]. In severe liver dysfunction, poor blood flow in the liver 37) or raised uric acid clearance 38,39) may cause hypouricemia. However, liver disease may exert bi-phasic effects on SUA. A close correlation was observed between SUA and liver-function tests, such as alkaline-phosphatase, GOT, GPT, and -GTP as well as alcohol intake. It is conceivable that the background life style in drinking alcohol induces hyperuricemia, increases liver damages, and leads to a high mortality rate from liver disease.
In conclusion, a prospective cohort study in 49,413 Japanese male workers demonstrated that those with high SUA have a high RR for all causes of death, heart disease, stroke, renal fail-ure, and hepatic diseases. Thus, SUA may be a risk factor for reduced life expectancy.

ACKNOWLEDGMENTS
This study was partly made by the 1994 grant-in-aid of gout investigation study from the Japanese Foundation of Gout. A summary of this study was presented at the XIV International Scientific Meeting of the International Epidemiological Association(1996, Nagoya, Japan). The authors are grateful to the staff of the Central Health Institute of East Japan Railway Company for their cooperation of this study.
We express our appreciation to Takashi Watanabe, Ph.D., Railway Technical Research Institute, for his help on computer analyses.