Natural course of fatty liver in 36,195 South Korean adults

Nonalcoholic fatty liver disease (NAFLD) is the most common cause of liver disease, and yet the natural course remains unclear. Study population included 36,195 individuals who participated in a health-screening program and diagnosed with fatty liver by abdominal ultrasound. Participants were provided written information regarding fatty liver and advised to make lifestyle changes. Ultrasound was repeated after at least 6 months. After a mean follow up of 4.9 years (±3.4), 19.6% resolved their fatty liver. Individuals who resolved were more likely female (22.9% vs. 12.3%), thinner (body mass index [BMI], 25.2 ± 2.7 vs. 26 ± 2.7), and with lower HOMA-IR (1.4 vs. 1.7) (P .70.001). Decrease in BMI predicted resolution of fatty liver with 42% of those in the top quartile of BMI decline resolving compared with 5.7% in the lowest quartile (odds ratio [OR] (95% confidence interval [CI]) 15.65 (14.13–17.34), P < 0.001)). Baseline HOMA-IR also predicted resolution with those in the top quartile (most insulin resistant) being least likely to resolve (12%) vs. those in the lowest quartile (25%) (OR 0.36 [0.31–0.42], P < 0.001). Fatty liver disease is persistent. Individuals with higher degree of insulin resistance are also the most likely to have persistent steatosis at follow up.


Results
As can be seen in Table 1, the overall group was young with a mean age of 37, and more than half were obese (BMI ≥ 25 kg/m 2 ) at baseline. Fatty liver persisted in a majority of individuals at follow and only 19.6% resolved their fatty liver. Those with resolution were slightly older. A greater proportion of women resolved than men (31% vs. 18%). Those with resolution were also more likely to be thinner with better metabolic profile (lower glucose, insulin, HOMA-IR, low-density lipoprotein [LDL], and triglyceride, and higher high-density lipoprotein cholesterol [HDL-C]) and lower prevalence of prediabetes and hypertension at baseline. They were also slightly less likely to have education beyond high school and less likely to be current smokers. NAFLD fibrosis score was overall low but slightly higher in those who resolved their fatty liver at follow up. This difference is likely due to higher age and lower albumin in those who resolved. Incidence for resolution of fatty liver /100 person-years of follow-up was 3.97 (Male: 3.52, Female: 6.94) ( Table 2).
We evaluated difference in BMI change between those with and without resolution of fatty liver. Those who resolved had a decrease in BMI compared with an increase in those with persistent fatty liver (mean ± standard deviation [SD], −1.0 ± 1.5 vs. 0.2 ± 1.2, P < 0.001). We further evaluated the resolution of fatty liver by quartiles of BMI change. As seen in Fig. 1, 42% of those who had the greatest decline in BMI (Q4) resolved their fatty liver at follow up. BMI change remained a significant predictor of fatty liver resolution in a multiple regression  Table 1. Baseline characteristics based on resolution of fatty liver. Data are mean ± standard deviation or median (interquartile range) unless otherwise specified. a Insulin concentration and HOMA-IR were available for 13, 688 individuals. b Obesity was defined by a BMI ≥25 kg/m 2 . c NAFLD fibrosis score = −1.675 + 0.037 × age (years) + 0.094 × BMI (kg/m 2 ) + 1.13 × IFG/diabetes (yes = 1, no = 0) + 0.99 × AST/ALT ratio − 0.013 × platelet (×10 9 /l) − 0.66 × albumin (g/dl). AST aspartate transaminase, ALT alanine aminotransferase, BMI body mass index, CI confidence interval, CVD cardiovascular disease, HDL high-density lipoprotein, HOMA-IR homeostatic model assessment-Insulin resistance, hs-CRP high-sensitivity c-reactive protein, LDL low-density lipoprotein, NAFLD nonalcoholic fatty liver disease, OR odds ratio.
Insulin concentration and HOMA-IR were available for 13,688 individuals (38%). In this subgroup, BMI change remained significantly associated with fatty liver resolution when adjusted for HOMA-IR (Supplementary Table 1). We also evaluated the association between baseline HOMA-IR and resolution of fatty liver (Table 4) to evaluate the effect of insulin resistance to predict fatty liver change. The most insulin-resistant quartile (Q4) were the least likely to resolve their fatty liver. Only 12% resolved compared with 25% in the most insulin-sensitive quartile (Q1). Thus, the OR for fatty liver resolution declined as HOMA-IR increased and this was similar for both men and women, after adjusting for covariates. Table 4 also shows the baseline BMI and BMI change by quartile of HOMA-IR. Baseline BMI increased from HOMA IR Q1 to Q4. Interesting, BMI change declined from Q1 to Q4. Thus, average BMI was more likely to decrease in the HOMA-IR Q4 than Q1, although resolution of fatty liver was lowest in Q4.

Discussion
The global prevalence of NAFLD is 25% 2 , which is similar to the estimate in our population of South Korean adults. After a mean follow-up of 4.9 years, approximately 1 in 5 adults resolved their fatty liver. The individuals who resolved were more likely female, less insulin resistant, and with fewer metabolic abnormalities. In addition, those who resolved were more likely to have lost weight.
NAFLD reflects a spectrum of disease from steatosis alone to steatohepatitis. Our study population had low NAFLD Fibrosis Score, and most likely had steatosis alone. Although individuals with steatohepatitis are at highest risk to progress to fibrosis and cirrhosis 1 , individuals with steatosis alone can also develop progressive hepatic fibrosis 9 . Insulin resistance is important to the pathogenesis and progression of NAFLD 10,11 . Our study population was selected to have fatty liver disease by ultrasound and likely more insulin resistant than individuals without fatty liver disease 7 . Even in this population selected for fatty liver, the degree of insulin resistance was negatively associated with fatty liver resolution at follow up. Thus, those in the highest HOMA-IR quartile were 64% less likely to resolve their fatty liver compared with the lowest HOMA-IR quartile.  Table 2. Incidence for resolution of fatty liver/100 person-years of follow-up.   www.nature.com/scientificreports www.nature.com/scientificreports/ As expected, higher HOMA-IR was associated with higher BMI. Interestingly, individuals with higher HOMA-IR had greater decline in BMI, albeit modest, at follow-up. This paradox likely reflects the ineffectiveness of modest weight loss and not to inherent futility of weight loss in insulin-resistant individuals. Wong et al. showed in a Hong Kong population with fatty liver that less than 3% weight loss was associated with 13% resolution of NAFLD versus 97% resolution in those who lost at least 10% of baseline weight 12 . Their population had a similar mean BMI ~25 kg/m 2 as ours. Furthermore, prior studies have supported that individuals with fatty liver with more unfavorable metabolic risk factors at baseline (e.g., hyperglycemia) require greater degree of weight loss to resolve fatty liver 13 . The mean decline in BMI in the HOMA-IR Q4 was 0.09 kg/m 2 . Thus, a male weighing 79.5 kg pounds with a height of 1.7 meters would have lost only 0.3% of his baseline weight. Thus, mean weight loss was minor.
There are limitations to our study. We diagnosed fatty liver based on ultrasound. Ultrasound has been shown to be reliable and accurate for detecting moderate to severe fatty liver 14 . However, ultrasound may miss mild degree of fatty liver. Nevertheless, ultrasound provides a noninvasive and feasible means to evaluate fatty liver in a large population.
In conclusion, we show in a large cohort of individuals that fatty liver disease is persistent. Unfortunately, individuals with the greatest risk for progressive liver disease-those with higher degree of insulin resistance and metabolic abnormalities-are also the most likely to have persistent steatosis at follow up. Weight loss, as confirmed in this study, remains the major intervention to resolve fatty liver. However, usual care is likely not sufficient to motivate major weight loss.

Materials and Methods
The study population consisted of individuals who participated in a comprehensive health-screening program, at least twice, at Kangbuk Samsung Hospital, Seoul and Suwon, Korea from 2002 to 2014 (n = 259,011). Fatty liver, as determine by ultrasound, was present at baseline in 67,138 individuals (25%). Individuals with fatty liver were excluded from the study if they were less than 20 years old and consumed >30 g/day (men) or >20 g/day (women) of alcohol. We also excluded individuals who were receiving treatment for diabetes, hypertension, or hyperlipidemia (including statin treatment). Individuals were also excluded for potential secondary etiologies of liver disease: positive hepatitis C antibody status (n = 34 at baseline and n = 69 at follow up); positive hepatitis B surface antigen status (n = 840 at baseline and n = 1,753 at follow up); and evidence of cancer (n = 682 at baseline and n = 1,675 at follow up). Individuals were also excluded for missing data (n = 11,299 at baseline and n = 12,642 at follow up). After these exclusions, the final number of study participants was 36,195. Mean follow up was 4.93 years (±3.39); median was 3.94 years (maximum 12.65 years). The Institutional Review Board of Kangbuk Samsung Hospital approved the study and waived informed consent as de-identified information was retrieved retrospectively.

Measurements.
As part of the health-screening program, individuals completed self-administered questionnaires, related to their medical and social history and medication use. Individuals were asked about duration of education (years), smoking history (never, former, or current) and alcohol consumption (g/day). We also assessed frequency of moderate-or vigorous-intensity physical activity per week.
Trained staff collected anthropometric measurements and vital statistics. Body weight was measured in light clothing with no shoes to the nearest 0.1 kilogram using a digital scale. Height was measured to the nearest 0.1 centimeter. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Blood samples were collected after at least 10-hours of fasting and analyzed in the same core clinical laboratory. The core clinical laboratory has been accredited and participates annually in inspections and surveys by the Korean Association of Quality Assurance for Clinical Laboratories.
Insulin concentrations were available for 13,688 individuals (38%). Homeostatic model assessment of insulin resistance (HOMA-IR) was used as a surrogate measure of insulin resistance and calculated using the following equation:   Table 4. OR (95% CI) for fatty liver resolution by HOMA-IR quartiles. Adjusted for age, sex, baseline BMI, BMI change, education, exercise, smoking and alcohol intake (g/day). a BMI change was calculated as BMI follow-up − BMI baseline . BMI body mass index, CI confidence interval, HOMA-IR homeostatic model assessment-Insulin resistance, OR odds ratio.