The best magazine
Results
Characteristics of This Study Population and Prediabetes or Type 2 Diabetes Incidence Rate
Mean follow-up was 6.8 ± 0.7 years. As outlined in Figure 1, of the 349 participants in the original survey, 213 subjects were included in the follow-up cohort. Seventy-two subjects with pre-DM/DM at baseline were excluded leaving a sample size of 141 people.
(Enlarge Image)
Figure 1.
Flow chart of the study population.
Pre-DM or DM overall incidence rate at a 7-year follow-up was 54.6% (77/141); 74.3% (26/35), in those with baseline NAFLD as compared to 48.1% (51/106) in those without NAFLD. The overall incidence of DM was 5.0% (7/141); 8.6% (3/35), in those with baseline NAFLD as compared to 3.8% (4/106) in those without NAFLD. Because of the small number of subjects developing diabetes, we combined this group with the pre-DM group for all further analyses referred to as pre-DM.
Comparison Between Subjects With and Without Baseline Fatty Liver (Table 1)
Subjects with baseline NAFLD were more likely to be men and, as expected, had higher body mass index (BMI), fasting serum levels of alanine aminotransferase (ALT) and insulin and lower adiponectin, but not significantly higher leptin levels. Subjects with baseline NAFLD had almost four-fold prevalence rate of the metabolic syndrome. Fasting glucose levels were not significantly higher, but HbA1c was significantly higher in the NAFLD group. The prevalence rate of family history of DM was not significantly different between groups; however, it tended to be higher in the NAFLD group and because of its strong association with DM, it was entered as a potential confounder to the multivariate analysis. There was no significant difference between groups in different parameters of nutritional intake: calories, total fat, saturated fat, mono-unsaturated fat, poly-unsaturated fat, carbohydrates and fibre (P ≥ 0.12 for all). Consumption of sugar-sweetened soft drinks was twice the amount and performance of physical activity was half the amount among subjects with NAFLD.
Baseline Predictors for the Development of Prediabetes in Multivariate Analysis (Table 2)
Two multivariate models were built, both with similar covariates, but one with NAFLD on regular US as a qualitative variable and the other with the HRI as a continuous quantitative variable. Both NAFLD on regular US (OR = 2.95, 1.03–8.44 95 % CI) and HRI (OR = 7.77, 1.82–33.26) were independent predictors for the development of pre-DM, adjusting for age, gender, BMI, family history of DM, baseline insulin, adiponectin, glucose and physical activity.
Adjustment for baseline abdominal obesity instead of BMI (with all variables in Table 2 except for insulin that is highly correlated with waist circumference) did not eliminate the association with either NAFLD on regular US (OR = 3.48, 1.28–9.44) or HRI (OR = 6.06, 1.50–24.53). Similar results were observed with adjustment for waist circumference as a continuous variable.
Adjustment for baseline HbA1c eliminated the statistical significance of the association between NAFLD and pre-DM (OR = 2.68, 0.86–8.42, P = 0.09). However, the association with HRI remained significant also with adjustment for HbA1c (OR = 6.01, 1.21–29.86, P = 0.03).
Adjustment for Lifestyle Parameters
Further adjustment for baseline smoking, physical activity or dietary intake of calories, total fat, saturated fat, mono-unsaturated fat, poly-unsaturated fat, carbohydrates, sugar-sweetened soft drinks, fibre and weight change during follow-up did not weaken the association between NAFLD (OR = 3.52, 1.47–8.44, P = 0.005) or HRI (OR = 8.97, 2.56–31.49, P = 0.001) and pre-DM.
Baseline Metabolic Syndrome vs. NAFLD or HRI as Predictors for the Development of Prediabetes in Multivariate Analysis (Table 3)
Both NAFLD on regular US (OR = 3.51, 1.29–9.54) and HRI (OR = 10.03, 2.42–41.62) were still independent and even stronger predictors for the development of pre-DM, adjusting for the metabolic syndrome, age, gender, BMI, and family history of DM. The metabolic syndrome was not a significant predictor for pre-DM in this model adjusting for NAFLD or HRI ( Table 3 ).
Risk for Prediabetes in Different Combinations of NAFLD and High Glucose or HbA1c Levels Within the Normal Range (Fig. 2)
(Enlarge Image)
Figure 2.
Risk for prediabetes in different combinations of Non-alcoholic fatty liver disease NAFLD and: (A) High glucose levels within the normal range (upper quartile ≥ 89) n = 79, 47, 15 respectively. (B) High HbA1c levels within the normal range (upper quartile ≥5.3) n = 79, 46, 16 respectively.
To better classify the population at risk for the development of pre-DM in clinical practice, subjects were classified by their combination of NAFLD and glucose or HbA1c that were categorized by values above and below the upper quartile of this study population (≥89 for glucose, ≥5.3 for HbA1c). Subjects without NAFLD and low fasting glucose had a 44.3% incidence rate of pre-DM vs. 59.6% in those with one risk factor and as high as 93.3% in those with both risk factors (P for trend = 0.001). A similar trend was observed with HbA1c (P for trend<0.001). In both combinations of NAFLD with glucose (P = 0.01) or with HbA1c (P = 0.001), the association with the development of pre-DM remained significant adjusting for age, gender and baseline BMI.
Source: ...