George N. Ioannou contributed to the study concept and design, the acquisition of data, the analysis and interpretation of data, the drafting of the article, a critical revision of the article for important intellectual content, and the statistical analysis.
Potential conflict of interest: Nothing to report.
Elevated serum uric acid (UA) levels strongly reflect and may even cause oxidative stress, insulin resistance, and metabolic syndrome, which are risk factors for the progression of liver disease. We sought to determine whether serum UA levels are associated with the development of cirrhosis or the presence of elevated serum liver enzymes. We used cohort data from the first National Health and Nutrition Examination Survey (NHANES I) to determine whether the baseline serum UA level was associated with the incidence of hospitalization or death due to cirrhosis among 5518 participants during a mean follow-up of 12.9 years (range = 4-21 years) after the exclusion of the first 4 years of follow-up. We also used cross-sectional data from NHANES 1988-1994 (n = 10,993) and NHANES 1999-2006 (n = 6186) to determine whether the serum UA level was associated with elevated serum alanine aminotransferase (ALT) or γ-glutamyl transferase (GGT), two markers of hepatic necroinflammation. Compared to persons in the lower third of the distribution of serum UA (<4.8 mg/dL), those in the top third (>6 mg/dL) had a higher risk of cirrhosis-related hospitalization or death [adjusted hazard ratio (AHR) = 2.8, 95% confidence interval (CI) =1.3-5.7], whereas the risk was not substantially increased in persons within the middle third (serum UA level = 2.6-4.8 mg/dL, AHR = 1.3, 95% CI = 0.6-2.7). A higher serum UA level was associated with greater mean serum ALT and GGT levels and a greater probability of elevated serum ALT and GGT. Conclusion: The serum UA level is associated with the development of cirrhosis and the presence of elevated serum liver enzymes after adjustments for important causes and risk factors of chronic liver disease. (HEPATOLOGY 2010;)
In humans and higher primates, uric acid (UA) is the final oxidation product of purine metabolism and is excreted in urine. Hyperuricemia has long been recognized as a cause of gouty arthritis and kidney stones. More recently, hyperuricemia has also been implicated in the development of hypertension, kidney disease, metabolic syndrome, and cardiovascular disease (reviewed by Feig et al.1 and Edwards2).
Although hyperuricemia has traditionally been considered a result of these conditions or an epiphenomenon, mechanisms have been proposed by which hyperuricemia could actually cause them. Such mechanisms include the induction by hyperuricemia of endothelial dysfunction, insulin resistance, oxidative stress, and systemic inflammation.1, 2
Oxidative stress, insulin resistance, and systemic inflammation are now known to be important risk factors for the development or progression of the most important liver diseases. For example, these conditions are considered central in the pathogenesis of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH).3 In addition, they contribute to the progression of hepatitis C virus (HCV)–related and alcoholic liver diseases.4 Therefore, we hypothesized that hyperuricemia, which strongly reflects and may even cause oxidative stress, insulin resistance, and systemic inflammation, is a risk factor for the development of cirrhosis or the presence of hepatic necroinflammation. We performed two related studies to test this hypothesis:
1A prospective cohort study to determine whether the baseline serum UA level is associated with the subsequent development of cirrhosis.
2Cross-sectional studies to determine the association between hyperuricemia and the levels of serum alanine aminotransferase (ALT) and γ-glutamyl transferase (GGT), two markers of hepatic necroinflammation.
AHR, adjusted hazard ratio; ALT, alanine aminotransferase; BMI, body mass index; CI, confidence interval; CRP, C-reactive protein; GFR, glomerular filtration rate; GGT, γ-glutamyl transferase; HBV, hepatitis B virus; HCV, hepatitis C virus; HDL, high-density lipoprotein; HOMA-IR, homeostasis model assessment insulin resistance; MDRD, Modification of Diet in Renal Disease; N/A, not applicable; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NHANES, National Health and Nutrition Examination Survey; NHEFS, First National Health and Nutrition Examination Survey Epidemiologic Follow-Up Study; RIBA, recombinant immunoblot assay; UA, uric acid.
Materials and Methods
First National Health and Nutrition Examination Survey (NHANES I) Cohort Study of the Association Between Serum UA Levels and Cirrhosis.
Data were derived from NHANES I, a cross-sectional study of a nationwide probability sample from the civilian, noninstitutionalized population of the coterminous United States conducted between 1971 and 1975.1 The survey included 14,407 participants, 25 to 74 years old, who completed extensive dietary questionnaires and underwent physical examinations and laboratory investigations. The NHANES I Epidemiologic Follow-Up Study (NHEFS)2 sought to locate these 14,407 individuals in 1982-1984, 1986, 1987, and 1992 and collected data on specific health conditions that they developed in the intervening period through personal interviews, hospitalization records, and death certificates. We merged NHANES I and NHEFS to form a nationally representative cohort of 14,407 persons with approximately 20 years of follow-up.
NHANES Cross-Sectional Studies of the Association Between Serum UA Levels and Serum Levels of ALT and GGT.
Data were derived from NHANES III (conducted between 1988 and 1994 and henceforth called NHANES 1988-1994) and from the more recent NHANES studies conducted between 1999 and 2006 (henceforth called NHANES 1999-2006). These were cross-sectional studies designed to assess the health and nutritional status of the noninstitutionalized US population.3 Participants completed personal, structured interviews at home and then attended a mobile examination center at multiple locations throughout the United States to undergo various examinations and provide blood samples.
NHANES I Cohort Study.
Among 14,407 NHANES I participants (25-74 years old), 13,861 were successfully traced on at least one of four follow-up occasions (1982-1984, 1986, 1987, or 1992). We attempted to exclude participants who suffered from cirrhosis at the time of entry into the study by excluding participants who, at the baseline, reported ever being told by a physician that they had jaundice (n = 886) or hepatitis (n = 47), who had hepatomegaly or splenomegaly at the baseline examination (n = 237), or whose level of serum albumin was less than 3 g/dL (n = 10). Serum bilirubin levels and platelet counts, which may be abnormal in advanced cirrhosis, were available only in a small minority of participants and therefore could not be used to identify participants with possible cirrhosis. Because cirrhosis may be present for a long time before it is clinically diagnosed, we also excluded participants who were diagnosed with cirrhosis within the first 4 years of follow-up or who had less than 4 years of follow-up (n = 687). We excluded 47 participants who had a malignant tumor and 90 with missing values in potential confounding variables. Serum UA levels were measured only in a subsample of participants, so 6339 participants did not have serum UA measurements; this left 5518 participants in the current analyses.
NHANES 1988-1994 and NHANES 1999-2006 Cross-Sectional Studies.
Of 16,884 NHANES 1988-1994 participants who were 25 years old or older, we excluded 168 pregnant women and participants with missing data for viral hepatitis B or C serologies (n = 2861), educational attainment (n = 186), alcohol consumption (n = 560), body mass index (BMI; n = 24), waist circumference (n = 474), diabetes (n = 9), coffee consumption (n = 18), and serum UA (n = 104). We excluded persons who fasted for ≤6 hours or lacked measurements for fasting serum insulin and plasma glucose (n = 1487); this left 10,993 persons for serum ALT analyses. Serum GGT testing was added to the NHANES 1988-1994 protocol after the study began, so serum GGT levels were not available in an additional 2359 participants; this left 8634 participants for serum GGT analyses. Identical inclusion and exclusion criteria resulted in 6186 participants for both serum GGT and ALT analyses in NHANES 1999-2006.
Measurement of Serum UA Levels
In NHANES I, serum UA was measured with an automated colorimetric phosphotungstic acid procedure, which had been validated against the uricase assay, on a Technicon SMA 12-60 (Technicon Instruments, Tarrytown, NY). In NHANES 1988-1994 and NHANES 1999-2006, serum UA was measured by oxidation with the specific enzyme uricase to form allantoin and hydrogen peroxide.5 Serum UA levels measured in NHANES studies have been used in multiple research publications.6, 7
In NHANES I, we divided participants into three groups based on tertiles of serum UA levels: 0 to 4.8, >4.8 to 6.0, and >6.0 mg/dL. The number of hospitalizations or deaths due to cirrhosis during follow-up (n = 80) precluded the division of participants into more categories.
In NHANES 1988-1994 and NHANES 1999-2006, participants were divided into four groups based on quartiles of serum UA levels: 0 to 4.2, >4.2 to 5.2, >5.2 to 6.3, and >6.3 mg/dL.
Measurement of Serum ALT and GGT in NHANES 1988-1994 and NHANES 1999-2006 and Definition of Elevated Levels
In NHANES 1988-1994, serum specimens were frozen and shipped weekly to a central laboratory (White Sands Research Center, Alamogordo, NM); there, they were stored initially at −20°C and then at −70°C before they were thawed and analyzed for ALT and GGT with a Hitachi model 737 multichannel analyzer.
In NHANES 1999-2006, serum specimens were refrigerated at 4 to 8°C and then shipped weekly to a central laboratory, at which they were tested upon arrival.8, 9 Although the central laboratory changed between 1999-2001 (Coulston Foundation, Alamogordo, NM, which used a Hitachi Model 704 multichannel analyzer) and 2002-2006 (Collaborative Laboratory Services, Ottumwa, IA, which used a Beckman Synchron LX20 analyzer), there was no difference in the ALT means of samples measured at the Coulston Foundation Laboratory in 2001 and Collaborative Laboratory Services in 2002.8, 9
We previously suggested that the method of specimen processing in NHANES 1988-1994 might have led to some degradation of ALT activity.10 Although absolute serum ALT levels are lower in NHANES 1988-1994, multiple studies by us10-12 and other investigators13, 14 have demonstrated all the expected associations with serum ALT activity, and this suggests a uniform reduction in ALT activity across all specimens.
Elevated levels were defined on the basis of recommended cutoffs as a serum ALT level > 30 U/L for men and > 19 U/L for women and a serum GGT level > 51 U/L for men and > 33 U/L for women.14
Hospitalizations and Deaths Due to Liver Cirrhosis During Follow-Up in NHANES I
Deaths and hospitalizations due to liver cirrhosis that occurred during follow-up were ascertained from hospitalization records and death certificates abstracted by specially trained NHEFS personnel. We used the following ICD-9 codes for cirrhosis: alcoholic cirrhosis, 571.2; cirrhosis without mention of alcohol, 571.5; pigmentary cirrhosis, 275.0; esophageal varices, 456.0-456.2; hepatic coma, 572.2; portal hypertension, 572.3; and hepatorenal syndrome, 572.4. Esophageal varices, hepatic coma, portal hypertension, and hepatorenal syndrome were included in the diagnosis of liver cirrhosis because the overwhelming majority of cases of these conditions in the United States are the result of liver cirrhosis. If acute necrosis of the liver (ICD-9 code 570.0) was diagnosed together with hepatic coma or hepatorenal syndrome, then the person was considered not to have cirrhosis. Other complications of cirrhosis such as ascites or peritonitis were not included as evidence of cirrhosis because they are commonly caused by other conditions. Similar diagnostic codes from NHANES I have been previously used by us and other investigators to determine the incidence of death or hospitalization related to cirrhosis.15-17
The date of the first hospital admission for each condition was used as the date of incidence. For subjects who had a death certificate recording one of these conditions but did not have a hospitalization for any of them, the date of death was used as the date of incidence.
Ascertainment of Potential Confounders and Other Baseline Characteristics
NHANES I Cohort Study.
The following were studied: age; gender and menopausal status; race, which was categorized as white (n = 4812) and nonwhite (n = 706; 653 of these were black, and only 53 were of other race, too small a number for additional racial categories); alcohol consumption over the previous 12 months, which was ascertained with a specifically designed validated questionnaire; BMI, which was calculated as the measured weight in kilograms divided by the square of the height in meters; subscapular-to-triceps skinfold ratio, which is a measure of central subcutaneous fat versus peripheral subcutaneous fat (the waist circumference was not measured in NHANES I); self-reported diabetes mellitus; coffee or tea consumption; consumption of dietary calories, proteins, carbohydrates, and fat, which was ascertained by 24-hour dietary recall; educational attainment; smoking; serum creatinine level; use of antihypertensive or diuretic medications; and geographical area of residence in the United States (Northeast, Midwest, South, and West).
Viral hepatitis B and C testing was not available in 1971-1975 when the NHANES I participants were recruited. We wanted to ensure that viral hepatitis, an important cause of cirrhosis in the United States, was not associated with serum UA levels in order to exclude the possibility that viral hepatitis was an important source of unmeasured confounding in our NHANES I cohort. We did this with data from NHANES 1988-1994 and NHANES 1999-2006 [Table 3 (shown later) shows little association between the serum UA level and the presence of viral hepatitis B or C].
NHANES I participants were not instructed to fast; hence, fasting plasma glucose, lipid, and serum insulin levels were not available.
NHANES 1988-1994 and NHANES 1999-2006 Cross-Sectional Studies.
The following were studied: age; gender and menopausal status; race/ethnicity, which was categorized as non-Hispanic white, non-Hispanic black, Mexican American, and other; alcohol consumption over the preceding 12 months; BMI; waist circumference; self-reported diabetes mellitus; fasting plasma glucose level; homeostasis model assessment insulin resistance (HOMA-IR), which was calculated as [fasting serum insulin (μU/mL) × fasting serum glucose (mmol/L)]/22.5; hepatitis B virus (HBV) infection (positive serum hepatitis B surface antigen) and HCV infection [positive serum HCV RNA in NHANES 1988-1994; positive HCV antibody confirmed by the recombinant immunoblot assay (RIBA) in NHANES 1999-2006]; educational attainment; smoking; physical activity, which was calculated by the multiplication of each recreational and nonrecreational activity by its intensity value and the addition of the products for all activity types (data presented for NHANES 1988-1994 only); average measured systolic and diastolic blood pressure; use of diuretics and antihypertensives; coffee intake (cups/day) in NHANES 1988-1994 or caffeine intake (mg/day) in NHANES 1999-2006; dietary intake of meat, seafood, dairy foods, and sugar-sweetened soft drinks, which was ascertained with a food frequency questionnaire (data presented for NHANES 1988-1994 only), and dietary intake of calories, protein, fat, and carbohydrates, which was ascertained by 24-hour dietary recall; C-reactive protein (CRP); total cholesterol and high-density lipoprotein (HDL) cholesterol; triglycerides; and glomerular filtration rate (GFR), which was estimated with the abbreviated Modification of Diet in Renal Disease (MDRD) study equation and standardized creatinine values.18
NHANES I Cohort Study.
The Cox proportional hazards model was used to determine the hazard ratio: persons with different levels of serum UA were compared with respect to the risk for cirrhosis with or without adjustments for baseline potential confounders. The date 4 years after the ascertainment of serum UA levels was used as time zero because any cases occurring within the first 4 years were excluded. We performed sensitivity analyses in which we varied the number of years after entry into the cohort that we excluded from analysis (0, 1, 2, 3, 4, 5, or 6 years).
We did not simultaneously adjust for all 18 ascertained baseline variables because some of them were ascertained only in subsets of our study population, because the number of events (n = 80) did not allow simultaneous modeling of so many variables, and because some are suspected but not proven to be associated with both UA and cirrhosis and therefore we were not sure a priori if they were significant confounders. Instead, we included all variables that are known to be associated with both UA and cirrhosis [alcohol consumption, gender/menstruation, race, age, BMI, and diabetes] and then added one by one the other variables; we retained those that resulted in a substantial change (>10%) in the adjusted hazard ratio (AHR) of the association between UA and cirrhosis. We then removed individual variables and eliminated them from the final model if their removal resulted in a <10% change in the AHR between UA and cirrhosis.
We performed additional analyses limited to persons (n = 3951) who were hospitalized at least once or who died during follow-up, so all had hospitalization records or death certificates available in which the diagnosis of cirrhosis could be sought.
Because of the importance of gender, obesity, and alcohol consumption in liver disease, we performed analyses stratifying persons into subgroups defined by these variables.
NHANES I employed a complex sampling design that incorporated stratification, clustering, and weighting processes. However, in NHANES I, the stratification and clustering were shown to have little effect on estimates, whereas sample weights can be highly variable and skewed. This can produce highly unstable results, especially when relatively small subsamples of the data are used, as in our study. Therefore, we decided to present unweighted results; we treated the observations as a simple sample, as recommended by the NCHS investigators19 and Korn and Graubard.20
NHANES 1988-1994 and NHANES 1999-2006 Cross-Sectional Studies.
Multivariate linear and logistic regression was used to determine whether serum UA levels were associated with the levels of serum ALT or GGT (linear regression) or with the likelihood of having elevated serum ALT or GGT levels (logistic regression). As in our NHANES I analyses, we included all variables known to be associated with both serum UA levels and serum liver enzyme levels in initial multivariate models and then used forward selection and backward elimination techniques to determine whether additional variables were important confounders.
In contrast to NHANES I, there is a general consensus that analyses employing NHANES 1988-1994 and NHANES 1999-2006 data should account for the complex sampling design of the studies. We did so with the survey commands of STATA 10 statistical software and with the appropriate weight, strata, and primary sampling unit variables.
NHANES I Cohort Study
Increased levels of serum UA were associated with increased age, BMI, subscapular-to-triceps skinfold ratio, serum creatinine, alcohol consumption, use of antihypertensive medications, dietary consumption of total calories, proteins, carbohydrates, and fat, nonwhite race, male gender, smoking, and lower educational attainment (Table 1). The prevalence of diabetes was slightly greater (4%) in persons in the top serum UA quartile versus persons in the lower two quartiles (3%). Serum UA levels did not appear to be associated with US geographical location or coffee or tea consumption.
Table 1. Baseline Characteristics of NHANES I Participants Stratified by Serum Uric Acid Levels
Serum UA (mg/dL)
0-4.8 (n = 1834)
>4.8-6.0 (n = 1850)
>6.0 (n = 1834)
The values are means or percentages and standard errors.
Data were available for a subset of the study population.
There were 80 incident cases of death or hospitalization due to cirrhosis, including 25 cases diagnosed only from death certificates, during a mean follow-up of 12.9 years after the exclusion of the first 4 years of follow-up. With the forward selection and backward elimination techniques described in the Materials and Methods section, the following variables remained in the final multivariate models: alcohol consumption, gender/menstruation, race, age, educational attainment, BMI, and subscapular-to-triceps skinfold ratio. The incidence of death or hospitalization due to cirrhosis increased from persons in the lowest serum UA tertile (53 per 100,000 person-years) to persons in the middle tertile [83 per 100,000 person-years, AHR = 1.3, 95% confidence interval (CI) = 0.6-2.7] to persons in the top tertile (210 per 100,000 person-years, AHR = 2.8, 95% CI = 1.3-5.7; Table 2). For every unit increase in serum UA, the AHR for cirrhosis was 1.40 (95% CI = 1.2-1.7); this association did not vary substantially among subgroups defined by gender, BMI, or alcohol consumption (Table 2). Interaction terms for these variables and serum UA were not statistically significant.
Table 2. Association Between Serum Uric Acid Levels and Hospitalization or Death Due to Cirrhosis
Serum UA (mg/dL)
Number of Subjects (n = 5518)
Deaths or Hospitalizations Related to Cirrhosis (n = 80)
Deaths or Hospitalizations Related to Cirrhosis per 100,000 Person-Years
Adjusted for daily alcohol consumption, gender/menstruation, race, age, educational attainment, BMI, and subscapular-to-triceps skinfold ratio.
Hazard Ratio per Unit Increase in Serum UA
Obese or overweight persons
Persons consuming < 1 alcoholic drink/day
Persons consuming ≥1 alcoholic drink/day
Among a subgroup of 3951 participants who either were hospitalized or died during follow-up, such that they all had hospitalization records or death certificates, the association between serum UA levels and the development of cirrhosis was similar to the association in the entire study population (AHR = 1.49 and 95% CI = 0.7-3.1 for persons with a UA level of 4.8-6.0 mg/dL and AHR = 2.95 and 95% CI = 1.4-6.2 for persons with a UA level > 6 mg/dL versus persons with a serum UA level < 4.8 g/dL). Not excluding persons who reported at the baseline ever being told by a physician that they had jaundice or hepatitis had almost no influence on the results.
Increasing the number of years following entry into the study that were excluded from analysis (in order to exclude prevalent cases of cirrhosis) from 0 to 6 years led to an increasing hazard ratio in the association between serum UA and cirrhosis (see the supporting information). The Kaplan-Meier curves (Fig. 1) show little difference between persons in different UA categories in the incidence of cirrhosis-related hospitalization or death due to cirrhosis in the first 7 years after enrollment into NHANES I, but there is a marked separation of the cumulative incidence curves after approximately 7 years. These findings suggest that the associations that we describe between serum UA levels and cirrhosis are truly due to the development of incident cases during follow-up rather than the presence of undiagnosed cases of cirrhosis at the baseline.
NHANES 1988-1994 and NHANES 1999-2006 Cross-Sectional Studies
In both NHANES studies, persons in increasing quartiles of serum UA had higher age, BMI, waist circumference, HOMA-IR, plasma triglycerides, plasma cholesterol, CRP, alcohol consumption, and consumption of dietary calories, protein, fat, and carbohydrates and lower HDL cholesterol, and they were more likely to be male and diabetic (Table 3). There was little difference between persons in different serum UA quartiles with respect to race/ethnicity or the prevalence of viral hepatitis B or C.
Table 3. Characteristics of Participants from NHANES 1999-2006 (n = 6186, Shown in Bold) and NHANES 1988-1994 (n = 10,993, Shown in Normal Font) Stratified According to Serum UA Levels
Serum UA Level (mg/dL)
This table should not be used to evaluate temporal trends in listed characteristics between 1988-1994 and 1999-2006 because of possible differences in the ascertainment of each characteristic between the two NHANES studies (which are beyond the scope of this work). The values are means or percentages and standard errors.
Diabetes was defined as self-reported diabetes (diabetics on insulin were instructed not to fast; hence, they were not included in this sample of fasting NHANES participants).
HOMA-IR was calculated as follows: [Fasting serum insulin (μ U/mL) × Fasting serum glucose (mmol/L)]/22.5
Hypertension was defined as a systolic blood pressure > 130 mm Hg, a diastolic blood pressure > 85 mm Hg, or the use of antihypertensive medications.
GFR was estimated with the abbreviated MDRD study equation27: GFR (mL/minute/1.73m2) = 175 × [Standardized serum creatinine level (mg/dL)]−1.154 × (Age)−0.203 × (0.742 if female) × (1.212 if black)
Standardized serum creatinine was derived from the measured serum creatinine. For NHANES 1988-1994
Standard creatinine = −0.184 + 0.960 × Uncalibrated serum creatinine.
For NHANES 1999-2000
Standard creatinine = 0.147 + 1.013 × Uncalibrated serum creatinine
With the forward selection and backward elimination techniques described in the Materials and Methods section, the following variables remained in our fully adjusted models predicting serum ALT or GGT levels: age, gender/menstruation, race/ethnicity, alcohol consumption, HBV infection, HCV infection, BMI, waist circumference, HOMA-IR, self-reported diabetes, fasting plasma glucose, serum CRP, diuretic medication use, hypertension, GFR, plasma triglycerides, plasma HDL cholesterol, and coffee consumption (or caffeine consumption in NHANES 1999-2006). In both NHANES studies, mean serum ALT and GGT levels both increased with increasing levels of serum UA (Table 4). The prevalence of elevated serum ALT or GGT also increased with increasing serum UA levels (Table 5).
Table 4. Association Between Serum Uric Acid Levels and Levels of Serum ALT or GGT, Data from NHANES 1999-2006 are Shown in Bold; Data From NHANES 1988-1994 are Shown in Normal Font
Serum UA Level (mg/dL)
Mean ALT (U/L)
Unadjusted Mean Difference in ALT Versus the 0-4.2 Category (U/L)
Adjusted Mean Difference in ALT Versus the 0-4.2 Category (U/L)*
Mean GGT (U/L)
Unadjusted Mean Difference in GGT Versus the 0-4.2 Category (U/L)
Adjusted Mean Difference in GGT Versus the 0-4.2 Category (U/L)*
An elevated ALT was a level > 30 U/L for men and > 19 U/L for women; an elevated GGT level was a level > 51 U/L for men and > 33 U/L for women.
Odds Ratio per Unit Increase in Serum UA
Obese or overweight persons
Persons consuming < 1 alcoholic drink/day
Persons consuming ≥ 1 alcoholic drink/day
Despite the differences in mean serum ALT and the prevalence of elevated ALT between NHANES 1988-1994 and NHANES 1999-2006, which were attributed largely to differences in specimen processing (see the Materials and Methods section), the associations between serum UA and serum ALT and GGT levels were consistent in the two studies.
The associations between serum UA and ALT or GGT were not substantially different among subgroups defined by gender, obesity, and alcohol consumption (Table 5); interaction terms for these variables and serum UA were not statistically significant.
Increased serum UA levels were associated with a greater risk of cirrhosis-related hospitalization or death and with elevated levels of serum markers of hepatic necroinflammation (ALT and GGT) even after adjustments for important causes and risk factors for cirrhosis.
A study from China reported that among 8925 employees of a chemical company, the serum UA level was associated with ultrasonographic NAFLD after adjustments for 10 anthropometric and metabolic potential confounders (although insulin resistance was not estimated).21 In an Italian study, 60 patients with ultrasonographic NAFLD had higher serum UA levels than 60 historical controls without NAFLD after adjustments for serum insulin; the investigators did not simultaneously adjust for all potential confounders.22 These studies suggest that hyperuricemia is associated with NAFLD; this would be expected because hyperuricemia is associated with many risk factors for NAFLD, such as obesity, insulin resistance, and metabolic syndrome, including each of metabolic syndrome's five components.6 We are not aware of studies other than ours investigating the associations between hyperuricemia and the development of cirrhosis.
A crucial question raised by our findings is whether hyperuricemia plays any role in directly causing hepatic necroinflammation and cirrhosis or whether it is just a marker for an adverse metabolic profile that leads to NAFLD/NASH or promotes progression of viral or alcoholic hepatitis. Observational studies such as ours cannot definitively distinguish between these two possibilities, but it is tempting and potentially useful to speculate whether hyperuricemia is a cause or a marker. Although hyperuricemia has been clearly associated with alcohol consumption, obesity, insulin resistance, systemic inflammation, and metabolic syndrome,1, 6 the associations that we describe with elevated liver enzymes persisted after adjustments for these conditions. However, we cannot exclude the possibility of residual confounding by conditions such as insulin resistance and systemic inflammation, which are likely not captured completely even after adjustments for HOMA-IR, CRP, and a host of related metabolic parameters that we included in our regression models. Nonetheless, even if hyperuricemia proves in the future to be only a marker for the presence of hepatic necroinflammation or the development of cirrhosis and not a cause, it is likely to be a useful marker because it can predict these outcomes independently of other currently available predictors. Furthermore, the associations that we describe were robust and occurred in all three NHANES studies for different outcomes (cirrhosis or elevated liver enzymes) and among different subgroups (by gender, obesity, and alcohol consumption) as well as the entire population. It has been proposed recently that hyperuricemia, rather than being simply a marker, might contribute to the cause of insulin resistance, oxidative stress, systemic inflammation, and metabolic syndrome.1, 2 Because these conditions can cause NAFLD, promote its progression to steatohepatitis, or even promote the progression of viral and alcoholic hepatitis, they represent mechanisms by which hyperuricemia can directly cause cirrhosis (Fig. 2). Hyperuricemia can induce endothelial dysfunction and reduced bioavailability of endothelial nitric oxide in rats,23 whereas treatment with allopurinol can improve endothelial function in patients with hyperuricemia.24 Glucose uptake in skeletal muscle depends in part on increases in blood flow mediated by the insulin-stimulated release of nitric oxide from endothelial cells. Therefore, hyperuricemia-induced endothelial dysfunction can potentially promote insulin resistance by impairing insulin-stimulated release of nitric oxide. Furthermore, hyperuricemia induces inflammatory and oxidative changes in adipocytes, and this process is crucial in causing metabolic syndrome in obese mice.25 Whether hyperuricemia is a cause or a result of conditions that promote the progression of liver disease is of considerable significance because pharmacological reduction of serum UA levels is possible but will be useful only if hyperuricemia is a cause rather than a result of these conditions. Similar arguments about the role of hyperuricemia as a cause or effect of cardiovascular diseases are currently ongoing.1, 2
Our NHANES I cohort study is limited by the fact that the diagnosis of cirrhosis is based on hospitalization records and death certificates. These diagnoses are likely to be accurate because cirrhosis that is advanced enough to lead to hospitalization or death presents with very typical symptoms, signs, and laboratory findings. A large review of autopsy studies found that a clinical diagnosis of cirrhosis made during life has nearly 100% specificity in comparison with autopsy data.26 Furthermore, the fact that 96.2% of the study participants were successfully traced suggests that the ascertainment of deaths or hospitalizations due to cirrhosis was nearly complete. However, cases of undiagnosed cirrhosis or diagnosed cirrhosis that did not lead to hospitalization or death were not captured. This misclassification tends to drive hazard ratios toward the null, so the hazard ratios that we report might be underestimates of the true hazard ratios; because cirrhosis is a rare outcome, such misclassification is expected to have little effect. The absence of HCV serologies is another potential limitation of our NHANES I study. However, the lack of a substantial association between HCV infection and serum UA levels (in NHANES 1988-1994 and NHANES 1999-2006) suggests that HCV infection is unlikely to be an important source of unmeasured confounding. Finally, the presence of diabetes might have been underreported in the 1970s, and fasting plasma glucose levels were not available.
Our cross-sectional NHANES 1988-1994 and 1999-2006 analyses are limited by reliance on a single measurement of serum ALT and GGT. Because serum ALT and GGT levels may fluctuate with time, this can result in nondifferential misclassification in comparison with multiple measurements over time. Such misclassification would most likely bias our results toward the null, so if multiple measurements of ALT or GGT were available, the associations between hyperuricemia and serum ALT or GGT would most likely be even greater than what we report.
We have reported novel associations between serum UA levels and the incidence of cirrhosis-related hospitalization or death or the presence of elevated serum ALT or GGT. These associations are largely independent of other known liver disease risk factors. The serum UA level might be a risk factor for the incidence of chronic liver disease. Future studies should investigate whether this association is causal or has clinical utility in the prediction of the presence or incidence of liver disease. If this is confirmed, further consideration should be given to measures that reduce the serum UA levels as a means of preventing cirrhosis in persons with elevated levels.