Elevated Liver Function Enzymes Are Related to the Development of Prediabetes and Type 2 Diabetes in Younger Adults: The Bogalusa Heart Study

The Bogalusa Heart Study is being conducted in the semirural, biracial (65% white and 35% black) community of Bogalusa, Louisiana. Between 1985 and 1996, three cross-sectional surveys of younger adults were conducted (baseline surveys). In addition, four cross-sectional surveys of subjects who had been previously examined were conducted between 1997 and 2010 (follow-up surveys). A total of 1,055 fasting subjects (72% white; 40% male) were selected from the last four surveys (1997–2010) of adults to form a retrospective cohort for this study. At baseline examination, individuals with a history of treatment for diabetes or those with a fasting glucose level ≥100 mg/dL were excluded. At the initial screening, the mean (SD) age was 25.1 (4.0) years (range, 18–38 years). At the most recent screening, the mean age was 41.3 (5.1) years (range, 26–50 years). The mean follow-up interval was 16.2 (4.9) years. The number of screenings and follow-up visits since young adulthood ranged from two to seven times. In all, 84% of subjects were screened three or more times and 74% were screened three to five times, with a total of 4,135 observations.

On the basis of data from the follow-up surveys, adult subjects were classified as normoglycemic, prediabetic, or diabetic according to American Diabetes Association criteria (13). Individuals with a fasting glucose level of 99 mg/dL or lower were considered normoglycemic (n = 874), those with a fasting glucose level between 100 and 125 mg/dL were considered prediabetic (n = 101), and subjects with a fasting glucose level of 126 mg/dL or higher or who had a history of treatment for the condition were considered diabetic (n = 80). All participants provided informed consent, and the institutional review board of the Tulane University Health Sciences Center approved the study.

Identical protocols were used by trained examiners across all surveys (14). Briefly, subjects were instructed to fast overnight (at least 8 h) before the screening, with compliance ascertained by interview on the day of examination. Information on personal health and medication history were obtained by questionnaires. Anthropometric and blood pressure measurements were collected in replicate and mean values were used. BMI (calculated as weight in kilograms divided by height in meters squared) was used as a measure of overall adiposity. Blood pressure measurements were obtained using the right arm of the subjects in a relaxed, sitting position. Systolic and diastolic blood pressures were recorded at the first and fifth Korotkoff phases using a mercury sphygmomanometer. Mean arterial pressure (MAP) was calculated as diastolic blood pressure plus one-third pulse pressure. Individuals were considered smokers if they reported currently smoking or having stopped smoking only within the past year. In a similar manner, individuals were considered alcohol drinkers if they reported current consumption of alcohol or having stopped alcohol drinking only within the past year.

From 1973 to 1986, cholesterol and triglyceride levels were measured using chemical procedures on a Technicon Autoanalyzer II (Technicon Instrument Corporation, Tarrytown, NY), according to the laboratory manual of the Lipid Research Clinics Program. These variables were later determined by enzymatic procedures on the Abbott VP instrument (Abbott Laboratories, North Chicago, IL) between 1987 and 1996 and on the Hitachi 902 Automatic Analyzer (Roche Diagnostics, Indianapolis, IN) afterward. Both chemical and enzymatic procedures met the performance requirements of the Lipid Standardization Program of the Centers for Disease Control and Prevention, which has routinely monitored the precision and accuracy of cholesterol, triglyceride, and HDL cholesterol measurements since the beginning of this study. Serum lipoprotein cholesterols were analyzed using a combination of heparin-calcium precipitation and agar-agarose gel electrophoresis procedures (15). The intraclass correlation coefficients between the blind duplicate (10% random sample) values ranged from 0.86 to 0.98 for HDL cholesterol, 0.86 to 0.98 for LDL cholesterol, and 0.88 to 0.99 for triglycerides. Glucose, ALT, and GGT were measured as part of a multichemistry profile (SMA20) by enzymatic procedures using the multichannel Olympus Au-5000 Analyzer (Olympus, Lake Success, NY). Plasma immunoreactive insulin levels were measured by a commercial radioimmunoassay kit (Phadebas; Pharmacia Diagnostics, Piscataway, NJ). The intraclass correlation coefficients between blind duplicate values ranged from 0.94 to 0.98 for insulin, 0.86 to 0.98 for glucose, 0.97 for ALT, and 0.99 for GGT. In addition, an index of insulin resistance was calculated according to the homeostasis model assessment of insulin resistance (HOMA-IR) equation (16): HOMA-IR = [insulin (microunits per milliliter) × glucose (millimoles per liter)/22.5].

All statistical analyses were performed with SAS version 9.2 (SAS Institute, Cary, NC). In the analyses, since there was no interaction between race (or sex) with liver function enzyme variables (ALT and GGT), the race and sex groups were combined to increase statistical power and to simplify the presentation. Continuous variables were tested for normality using a Kolmogorov-Smirnov test. Values of triglycerides, glucose, insulin, and insulin resistance index (HOMA-IR) were log transformed to improve normality, as applicable. General linear models were used to examine the cardiometabolic risk factor variables by status of adult diabetes, adjusted for age, race, and sex. The trends of follow-up diabetes by baseline ALT and GGT quartile levels were examined using the Cochran-Armitage trend test.

Longitudinal multivariate logistic regression analyses (generalized equation estimation method) were used to determine which longitudinal changes in risk variables since baseline predicted follow-up adult diabetes status. Risk variables were standardized to z scores based on age-, race-, and sex-specific means and SDs. The model included smoking (yes/no) and alcohol drinking (yes/no) observed since baseline as longitudinal categorical variables and BMI, MAP, HDL cholesterol, LDL cholesterol, triglycerides, and HOMA-IR as longitudinal continuous variables, as well as ALT and GGT (model 1 and 2, respectively), adjusted for study year, age, age squared, race, sex, and sex by race interaction, as applicable. Odds ratios are expressed per 1-SD increment in BMI, MAP, HDL cholesterol, LDL cholesterol, triglycerides, HOMA-IR, ALT, and GGT. Nonsignificant terms (P > 0.05) were removed from the model by backward stepwise procedure. To evaluate the discriminatory capability of the models using the area under the receiver operating characteristic (ROC) curve (C statistic), the multivariate C statistic logistic regressions were performed on the association of the baseline liver function enzymes and glucose homeostasis variables (glucose, insulin, and HOMA-IR) with prediabetes or diabetes status at the follow-up in adulthood adjusted for age, race, sex, smoking, drinking, BMI, MAP, HDL cholesterol, LDL cholesterol, and triglycerides. ROCs (C values) were tested for equality by diabetes status and by pairwise comparison of each model with the rest.

Mean levels of anthropometric, hemodynamic, metabolic, and liver function enzyme variables at baseline are presented in Table 1 by follow-up diabetes status. The prediabetic group, versus the normoglycemic group, displayed significantly higher systolic blood pressure, MAP, LDL cholesterol, glucose, insulin, HOMA-IR, and ALT. The diabetic group, versus the normoglycemic group, displayed higher BMI, systolic blood pressure, diastolic blood pressure, MAP, LDL cholesterol, triglycerides, glucose, insulin, HOMA-IR, ALT, and GGT and lower HDL cholesterol.

The follow-up prevalence rate of diabetes status after a 16-year interval by quartiles of baseline ALT and GGT is shown in Fig. 1. Significant adverse trends of ALT and GGT were noted for both prediabetic (P < 0.01) and diabetic (P < 0.05) groups.

Table 2 shows the results of a multivariable adjusted longitudinal logistic regression model that included BMI, MAP, HDL cholesterol, LDL cholesterol, triglycerides, and HOMA-IR along with ALT (model 1) and GGT (model 2) as a multiple repeated-measurement standardized variable (z scores) and smoking (yes/no) and alcohol drinking (yes/no) as a longitudinal repeated-measurement categorical variable observed since baseline. After adjusting for study year, age, race, sex, and race by sex interaction (as applicable), expressed per 1-SD increase, HOMA-IR showed a significant odds ratio of 1.70 (P < 0.0001) for developing follow-up prediabetes after an average of 16 years of follow-up. Neither ALT nor GGT showed significant odds ratios in this regard. With respect to developing follow-up diabetes, the odds ratios per 1-SD increase were 1.16 (P = 0.05) for ALT and 1.20 (P < 0.01) for GGT. BMI, triglycerides, and HOMA-IR displayed odds ratios of 1.77, 1.26, and 1.57, respectively (P < 0.05); BMI and triglycerides showed odds ratios of 2.23 and 1.36, respectively, in model 2 (P < 0.001). No significant race difference in the predictive value of liver function enzymes was observed for either prediabetes or diabetes (data not shown).

In terms of discriminative values of different baseline liver function enzymes and glucose homeostasis variables (glucose, insulin, and HOMA-IR) (Table 3), the predictive models produced C values ranging from 0.701 to 0.751 and from 0.814 to 0.830 for prediabetes and diabetes for these measures, respectively, which were relatively similar to each other in magnitude. Compared with the prediabetic group, the diabetic model had significantly higher C values of insulin, HOMA-IR, ALT, and GGT, regardless of race, sex, adiposity, and other conventional cardiometabolic risk factor variables (P < 0.05).

The current data links development of prediabetes and type 2 diabetes in apparently healthy adults with concurrent longitudinal changes in the liver function enzyme, along with some cardiometabolic risk variables, since young adulthood. Significant adverse trends of baseline ALT and GGT were noted for both follow-up prediabetic and diabetic groups. Furthermore, there was a significant increased risk (1.2-fold) of developing type 2 diabetes for elevated ALT and GGT levels (per 1-SD increment) in young adults over a 16-year period, after controlling for alcohol consumption, obesity, and other traditional cardiometabolic risk factors. In terms of predictive value as depicted by the area under the ROC curve analysis, adjusted for relevant covariates, none of the glucose homeostasis variables were significantly better than baseline ALT and GGT in identifying increased risk of follow-up prediabetes and diabetes over an average of a 16-year period.

In the current study, the prevalence of both prediabetes and diabetes was lower than that reported previously (3,17), which could be explained by the lower average age of our cohort. Of note, adiposity, as depicted by BMI, was the strongest predictor of prediabetes and diabetes in this study and in our earlier study (18). As is well known, obesity is pathologically linked to insulin resistance/hyperinsulinemia and related development of dysglycemia. This is consistent with our earlier studies in that obesity precedes, not follows, hyperinsulinemia/insulin resistance, or metabolic syndrome (19,20). In the current study, the residual effect of these enzymes on the onset of diabetes remained even after controlling for BMI.

The current study, consistent with earlier reports (4,6–9,11), demonstrates an association of ALT and GGT with prediabetes and diabetes. These liver enzymes likely represent the impaired glucose regulation state. The liver is an important site for insulin clearance (8) and production of inflammatory cytokines (8). The adverse association of liver function enzymes to type 2 diabetes may result from the link between excess central (visceral) adiposity, NAFLD, and hepatic insulin resistance mediated by elevated hepatic free fatty acid flux from visceral fat that induces increased hepatic lipogenesis and triglyceride-rich lipoprotein secretion (4,5,21,22). Indeed, an overaccumulation of unoxidized long-chain fatty acids can saturate the storage capacity of adipose tissue, which induces a lipid “spill over” to nonadipose tissue (liver, muscle, heart, and pancreatic β-cells) (23). In such ectopic deposition, excess lipids (hepatic free fatty acids) are driven into an alternate nonoxidative pathway, causing the formation of reactive lipid moieties that evolve into relevant cellular dysfunction and programmed cell death (23). This metabolic disturbance caused by fat infiltration could induce abnormal liver enzyme. Controlling for obesity may not exclude this cellular dysfunction of central deposition of fat.

Moreover, excess central adiposity (and by inference, NAFLD) augments the expression of proinflammatory adipocytokines, including tumor necrosis factor-α, and reduces the expression of insulin-sensitizing and anti-inflammatory adiponectin, which causes an increase in insulin resistance (4,5,24). Insulin resistance, in turn, increases reactive oxygen species and oxidative stress by attenuating the inhibitory effect of insulin on lipid oxidation and by activating CYP2E1, a component of the cytochrome P-450 system (5,25).

Balkau et al. (10) have demonstrated over a 9-year period that fatty liver indices, including liver function enzymes, are simpler clinical tools for predicting diabetes in relatively older adults. In the current long-term longitudinal study, baseline young adult ALT and GGT levels predicted a 16-year incident type 2 diabetes, even after controlling for alcohol intake, obesity, and other cardiometabolic risk factors.

The current study has certain limitations in that it lacks direct assessments of postchallenge glucose, in vivo insulin action and secretion, body fat mass and distribution, and liver fat content. Instead, we used the simple surrogate measure of glucose homeostasis that is applicable to population studies. Furthermore, information on baseline adipocytokines might have provided additional insight into the development of diabetes.

In summary, elevations in levels of liver function enzymes ALT and GGT relate to incident prediabetes and type 2 diabetes in apparently healthy younger adults. These results underscore the potential utility of ALT and GGT as biomarkers, along with obesity status, in the evaluation of diabetes risk in this age-group.

This study was supported by grants HD-061437 and HD-062783 from the National Institute of Child Health and Human Development and grant AG-16592 from the National Institute on Aging.

No potential conflicts of interest relevant to this article were reported.

Q.M.N. researched data, contributed to discussion, and wrote, reviewed, and edited the manuscript. S.R.S. researched data, contributed to discussion, and reviewed and edited the manuscript. J.-H.X. researched data. W.C. researched data and reviewed and edited the manuscript. S.H. and J.R. contributed to discussion. G.S.B. researched data and reviewed and edited the manuscript.

The Bogalusa Heart Study is a joint effort of many investigators and staff members, whose contributions are gratefully acknowledged. The authors thank the study participants.