The Association Between Different Obesity Phenotypes and Liver Fibrosis Scores in Elderly Individuals with Fatty Liver in Taiwan

Introduction

Nonalcoholic fatty liver disease (NAFLD) is one of the most common liver disorders worldwide. Its prevalence ranges from 24.0% to 46.0% in the general population across western countries and from 7.9% to 54% in Asian countries.¹ In Taiwan, its prevalence is approximately 11.5–41.0%,^2,3 but one study has disclosed a higher rate (50.1%) in the elderly population.⁴

NAFLD is characterized by excess fat in the liver independent of secondary causes of fat accumulation in the liver. Approximately 23–44% of the patients with NAFLD have nonalcoholic steatohepatitis (NASH),^5,6 a more rapidly progressive form,⁷ and 37–41% of the patients with NASH develop progressive fibrosis.^8,9 Besides, about 10–20% of the patients with NASH progress to cirrhosis within 10 years, whereas the cumulative incidence of hepatocellular carcinoma (HCC) in NASH-related cirrhosis ranges from 2.4% to 12% within 3–7 years.^10,11 The presence and severity of fibrosis predict patients’ overall and liver-related mortality.^12,13 Although liver biopsy is the gold standard procedure for assessing NAFLD severity, it is invasive, with the risk of periprocedural complications and mortality.¹⁴ In addition, performing liver biopsy for all individuals having or being at risk with NASH is unrealistic. Therefore, noninvasive methods for liver fibrosis assessment have been developed. NAFLD fibrosis score (NFS) and Fibrosis-4 score (FIB-4) are among the most common noninvasive markers of fibrosis with acceptable accuracy to detect advanced fibrosis in NAFLD.^14–17 Advance fibrosis can be diagnosed with high accuracy (positive predictive value [PPV], 90%) using the high cutoff score (>0.676) in NFS and a PPV of 80% using the high cutoff score (>2.67) for FIB-4.^14–16

Risk factors for NAFLD include obesity, type 2 diabetes mellitus (T2DM), hypertension, and dyslipidemia, which are often categorized together as the insulin resistance syndrome or metabolic syndrome.¹⁸ With the progressive epidemic of obesity and metabolic syndromes, NAFLD prevalence is expected to rise. In a cross-sectional survey, Fan et al reported that fatty liver was 11.6-fold (95% confidence interval [CI], 8.2–16.5) higher in patients with obesity and was increased by 39.3-fold (95% CI, 17.8–87.1) in patients with metabolic syndrome.^19,20 Moreover, evidence showed that metabolic healthy obesity (MHO) may significantly be associated with increased risk of NAFLD and advanced fibrosis, with risks intermediate between metabolic healthy non-obese (MHNO) and metabolic unhealthy obese (MUO).²¹ A multicenter cross-sectional study in Spain showed that NASH was more frequent in metabolic unhealthy individuals (metabolic unhealthy non-obese [MUNO], 55.8%; MUO, 57.8%) than metabolic healthy individuals (MHNO, 26.9%; MHO, 42.4%)²² and another study in Europe reported that fibrosis increased progressively with numbers of metabolic alteration.²³

In addition, accumulating evidence suggest that NAFLD may be both a consequence and a cause of various components of metabolic syndrome, particularly T2DM and to a lesser extent, hypertension.²⁴ A large Taiwanese population study comprised 132,377 nondiabetic individuals with two or more health examinations during 1996–2014, showed that ultrasound-diagnosed NAFLD was significantly associated with two-fold increase in incident T2DM.²⁵ Another retrospective cohort study of 11,448 participants with 5-year follow-up in South Korea revealed that development of NAFLD was associated with increased risk of incident hypertension (odds ratio [OR], 1.36 [1.1–1.7]) compared with resolution of NAFLD at follow-up.²⁶

Furthermore, patients with NAFLD at an advanced age are more at risk for cirrhosis and even a higher mortality.²⁷ In a Japanese retrospective study, severe fibrosis was significantly more prevalent in older patients (≥60 years old) than in younger patients (29% vs 53%, p = 0.02).²⁸ Therefore, health professionals must know how to identify NAFLD in elderly individuals. However, NAFLD is often overlooked, with few studies of NAFLD focusing on the elderly. The insight into whether obesity itself or metabolic abnormalities have more influence on NAFLD remains unconfirmed. Additionally, sex differences do exist in the prevalence of NAFLD and liver fibrosis severity, which are higher in men than women during the reproductive age but changes after menopause.²⁹ Nevertheless, the connection is not fully understood and more studies are needed to fill in the research gap. With the elderly as the major group for metabolic syndrome and obesity, we are interested in the relationship between NAFLD and fibrosis risks. Hence, this cross-sectional study aimed to investigate the association between the different phenotypes of obesity or metabolic syndrome and the NFS and FIB-4 score in a Taiwanese elderly population with fatty liver.

Materials and Methods

Study Population

A total of 17,115 participants (aged ≥65 years) underwent health examinations in Taipei Mackey Memorial Hospital between 2008 and 2018. Among them, 4105 participants with fatty liver diagnosed by abdominal sonography were included in this study.

However, we excluded those who had chronic viral hepatitis, which was defined in this study as having a history of hepatitis B virus or hepatitis C virus or having tested positive for hepatitis B antigen or hepatitis C antibodies documented in medical records; those who reported heavy drinking upon examination, defined as more than 4 drinks on any day for men or more than 3 drinks for women according to National Institute on Alcohol Abuse and Alcoholism; those with incomplete data on fasting glucose level, platelet count, triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), albumin, AST and ALT; and those with repeated visits (Figure 1).

Figure 1 Flow chart of evaluated participants with fatty liver.

The study protocol was evaluated and approved by the Human Research Ethics Committee of Mackay Memorial Hospital (project research number 18MMHIS137).

Data Collection and Measurements

To obtain medical history, all participants completed a questionnaire that included demographic characteristics, personal health history, current medication use, smoking habit, and alcohol consumption. Here, they self-reported their current smoking status as either nonsmoker, social smoker, or current smoker, and their alcohol consumption as either no drinking, social drinking, or heavy drinking. We also collected data on blood pressure (BP), body weight, height, waist circumference, and other biomarker measurements including complete blood counts, liver and renal function tests, lipid panels, glucose, and albumin. The Friedewald equation and Martin/Hopkins density lipoprotein cholesterol (LDL-C) method (if TG ≥ 400) were used in those without serum levels for LDL-C.^30,31 Furthermore, well-trained gastroenterologists performed abdominal sonography to categorize fatty liver into two groups, namely, mild and moderate-to-severe groups.³²

Assessment of Obesity and Metabolic Health Status

Obesity and metabolic syndrome were defined using the ethnic-specific criteria made by Taiwan Health Promotion Administration, Ministry of Health and Welfare. BMI was calculated as weight in kilograms divided by the square of the participants’ height in meters. Obesity was defined as BMI ≥ 27, and non-obese as BMI < 27.³³ Meanwhile, metabolic syndrome was diagnosed in a participant with three or more of the following criteria: 1) central obesity (abdominal circumference ≥90 cm for men and ≥80 cm for women); 2) elevated serum triglyceride level (≥150 mg/dL) or taking lipid-lowering drugs; 3) decreased HDL cholesterol level (<40 mg/dL for men and <50 mg/dL for women); 4) elevated BP (systolic BP ≥ 130 mmHg or diastolic BP ≥ 85 mmHg) or history of hypertension or taking antihypertensive drug; and 5) elevated fasting glucose level (≥ 100 mg/dL) or history of diabetes mellitus or taking hypoglycemic agents. Then, using the definition above, we categorized the participants into following four phenotypes: 1) metabolically healthy non-obese (MHNO), 2) metabolically unhealthy non-obese (MUNO), 3) metabolically healthy obese (MHO), and 4) metabolically unhealthy obese (MUO).^34,35

Noninvasive Markers of Fibrosis and Fibrosis Severity Categories

We used NFS and FIB-4 to calculate correlated fibrosis severity.^15,16 NFS and FIB-4 were calculated using the following equations: NFS = (−1.675) + 0.037 − age (years) + 0.094 − BMI (kg/m²) + 1.13 × impaired fasting glucose (IFG)/diabetes (yes = 1, no = 0) + 0.99 × AST/ALT ratio − 0.013 × platelet count (× 10⁹/L) − 0.66 × albumin (g/dL); and FIB-4 = age × AST (IU/L)/platelet count (× 10⁹/L) × √ALT (IU/L).

For each score, advanced fibrosis was defined using previously published cutoff values.^14–16,36 The fibrosis severity defined by NFS was categorized into two groups: no-to-mild fibrosis (≤0.676) and advanced fibrosis (>0.676). Fibrosis severity divided by the FIB-4 score used the cutoff value of ≤2.67 for no-to-mild fibrosis and >2.67 for advanced fibrosis.

Statistical Analysis

We used descriptive statistics to summarize the characteristics of the participants according to the four phenotypes (MHNO, MUNO, MHO, and MUO). These descriptive statistics were expressed as mean ± SD for continuous variables and percentages for categorical variables. Significant differences in the characteristics among the four phenotypes were determined by analysis of variance tests for the continuous variables and chi-square tests for categorical variables.

Using logistic regression and MHNO as control group, we calculated the OR and 95% CI to describe the association between obese phenotypes and advanced fibrosis. Potential confounders were adjusted via the following three models: Model 1, sex; Model 2, factors in Model 1 plus smoking status and alcohol consumption; Model 3, factors in Model 2 plus fasting glucose, total cholesterol (TC), TG and HDL. Subgroup analysis stratified by sex was performed and adjusted for potential confounders as above.

All statistical data were analyzed using the SPSS version 24.0 software (IBM Corp., Armonk, NY). A p value of <0.05 was considered to be statistically significant.

Results

A total of 1817 participants (587 males and 1230 females, aged 72.9 ± 5.9 years) were analyzed. Among them, 580, 671, 131, and 435 were classified into MHNO, MUNO, MHO, and MUO groups, respectively. Table 1 presents the characteristics of study participants stratified by obesity phenotypes and characteristics of sex-based subgroups are shown in Supplemental Table 2.1, 2.2 and Supplemental Table 3.1, 3.2. The percentage for moderate-to-severe fatty liver diagnosed by sonography was 5.0%, 13.0%, 18.3%, and 31.3% in the MHNO, MUNO, MHO, and MUO group (p < 0.05). Furthermore, the percentage for advanced fibrosis according to the NFS score (NFS > 0.676) was 6.7%, 12.8%, 13.0%, 30.3% in the MHNO, MUNO, MHO, and MUO groups (p < 0.05), with an average NFS score of −0.840 (±1.0), −0.361 (±1.0), −0.292 (±0.9), and 0.176 (±0.4) (p < 0.05), respectively. Meanwhile, the percentage for advanced fibrosis according to the FIB-4 score (FIB-4 >2.67) was 8.3%, 9.5%, 16.8%, and 9.7% in the MHNO, MUNO, MHO and MUO groups (p = 0.031), with an average FIB-4 score of 1.84 (±0.8), 1.83 (±0.7), 1.82 (±0.9), and 1.98 (±0.7) (p = 0.20), respectively. Compared with the MHNO group, the MUNO, MHO, and MUO groups were more likely to have moderate-to-severe fatty liver (MUNO: OR, 2.83 [1.83–4.37]; MHO: OR, 4.30 [2.41–7.68]; MUO: OR, 8.64 [5.65–13.22]) (Supplemental Table 1.1, 1.2).

Table 1 Characteristics of Study Participants According to Obesity Phenotypes

Table 2 shows the comparison of advanced fibrosis categorized by the NFS score (NFS > 0.676) between the MUNO, MHO, and MUO groups and the MHNO group by multiple logistic regression analysis. Compared with the MHNO group, the odds of liver fibrosis were significantly higher in MUNO, MHO, and MUO groups (2.03 [1.37–3.02], 2.09 [1.14–3.82], and 6.04 [4.12–8.87], respectively). After we adjusted for covariates such as sex, smoking habit, alcohol consumption, fasting glucose, TC, TG, and HDL, the results were fundamentally the same (MUNO: OR, 2.43 [1.50–3.93]; MHO: OR, 2.35 [1.25–4.41]; MUO: OR, 6.11 [3.90–9.59] compared with the MHNO group). In the subgroup analysis stratified by sex, the results showed that compared with the MHNO group, the odds of liver fibrosis were significantly higher in MUNO and MUO groups in both women (MUNO: OR, 2.05 [1.16–3.62]; MUO: OR, 7.01 [4.04–12.15]) and men (MUNO: OR, 2.73 [1.55–4.81]; MUO: OR, 6.17 [3.52–10.83]) with a positive trend in the MHO group (women: OR, 2.09 [0.87–5.12]; men: OR, 2.23 [0.96–5.21]) (Table 3).

Table 2 Univariate Analysis and Multivariate Analysis for Advanced Fibrosis Using NFS (>0.676) According to Obese Phenotypes

Table 3 Subgroup Analysis Stratified by Sex for Advanced Fibrosis Using NFS (>0.676) Category

Table 4 demonstrates the comparison of advanced fibrosis categorized by the FIB-4 score (FIB-4 >2.67) between the MUNO, MHO, and MUO groups and the MHNO group by multiple logistic regression analysis. Compared with that in the MHNO group, the odds of advanced fibrosis by the FIB-4 score was 1.16 (0.79–1.72), 2.26 (1.31–3.90), 1.18 (0.77–1.83) in the MUNO, MHO, and MUO groups, respectively. After adjusting potential confounding covariates, we found a positive trend for association, with significant finding in the MHO group (MUNO: OR, 1.34 [0.83–2.18]; MHO: OR, 2.37 [1.36–4.13]; MUO: OR, 1.38 [0.82–2.31] compared with the MHNO group). When dividing the study population according to sex in the subgroup analysis, the association was stronger in women, with significant finding in the MHO group (MUNO: OR, 1.45 [0.82–2.56]; MHO: OR, 3.84 [1.86–7.95]; MUO: OR, 1.54 [0.83–2.85] compared with the MHNO group), whereas no differences were found in men (Table 5).

Table 4 Univariate Analysis and Multivariate Analysis for Advanced Fibrosis Using FIB-4 (>2.67) According to Obese Phenotypes

Table 5 Subgroup Analysis Stratified by Sex for Advanced Fibrosis Using FIB-4 (>2.67) Category

Discussion

In our study, both metabolic syndrome and obesity were positively associated with more advanced fibrosis according to the NFS, which showed significant findings, but only a positive trend was revealed when using the FIB-4 score with significantly positive association in the MHO group. Regardless of the metabolic status, those with obesity had a higher percentage of moderate-to-severe fatty liver diagnosed by sonography. Obesity seemed to have more detrimental effects than metabolic abnormalities per se on the severity of more advanced fibrosis in the elderly according to the FIB-4 score. The effect was even more pronounced in elderly women with a three-fold increase in the MHO group, whereas no statistical difference was observed in men.

Our finding is consistent with those of previous studies,^37–39 which concluded that metabolically healthy individuals with obesity had higher odds of developing NAFLD (OR for MHO: 3.55 [3.37–3.74]) and worsening of noninvasive fibrosis markers compared with those with normal weights (adjusted hazard ratio [HR] for BMI ≥ 30 compared with BMI 18.5–22.9 was 3.52 [2.64–4.69]). The association was also stronger in women than in men (adjusted HR for BMI ≥ 30 compared with BMI 18.5–22.9 in women was 5.76 [5.23–6.34] and 2.87 [2.70–3.05] in men).^37,38 Similarly, fatty liver was more prevalent in the MHO and MUO groups as well (adjusted OR for MUNO, MHO, and MUO: 3.75 [3.17–4.42], 7.52 [6.13–9.22], and 20.10 [16.20–24.90], respectively).³⁹ A recent Taiwanese study also reported that both metabolic syndrome (OR, 3.19 [2.41–4.22]) and higher BMI (OR, 5.66 [3.91–8.19]) were positively associated with higher odds of fatty liver in the elderly.⁴⁰

Conversely, Chen et al found that by using FIB-4 score to evaluate the risk for advanced fibrosis, higher BMI was not a risk factor (OR, 0.72 [0.37–1.40]) and Miyaaki et al stated that although there were more obese cases in the group with severe fibrosis, the association was not statistically significant (mild:severe = 67%:78%; p = 0.23).²⁸ These findings were contrary to our study, which showed that the effect of obesity on more advanced fibrosis was significant. Furthermore, Gutiérrez-Grobe et al showed that advanced fibrosis was more prevalent in the MUO group than in the MHO group (28% vs 6.5%, respectively; p < 0.05), similar with our study results using NFS definition, but different when we used FIB-4 definition.⁴¹ In addition, Kim et al reported that in sex-stratified analysis, fibrosis progression was stronger in men than in women (HR for BMI ≥30 kg/m² compared with BMI 18.5–22.9 kg/m² in men was 4.83 [4.31–5.43] and 3.49 [2.90–4.20] in women), which was opposite from our finding.³⁸ Their study composed of mainly premenopausal women, which estrogen may play a protective role, while ours are elder women with postmenopausal status.^29,38,42

In our study, there are some differences in the results of advanced fibrosis assessed by NFS and FIB-4 scores (Tables 2 and 4). Several reasons might have accounted for this discrepancy. First, the prevalence rates of advanced fibrosis and the cutoff values affect the diagnostic performance of those tests.^14,36 The best cutoff value is still debatable in the elderly and ethnic differences should take into account. Second, NFS formula included BMI and hyperglycemia, which are related directly to metabolic syndrome and obesity, while FIB-4 scores only included data on age, platelets, and liver enzymes. This may pose bias for detecting the association. Third, some patients with advanced fibrosis may not be detected and fall into the wrong category.^43,44 Nevertheless, previous studies validated that NFS and FIB-4 scores best predict NAFLD advance fibrosis and are widely used in primary setting.^36,44

Aging and visceral adiposity promotes metabolic abnormalities and decline insulin sensitivity.^45,46 Excessive free fatty acids accelerate NAFLD pathogenesis and progression.⁴⁷ Flux of free fatty acids from the adipose tissues into the liver can possibly induce lipid peroxidation and increasing proinflammatory cytokines, which can ultimately cause hepatic insult.⁴⁸ With age-related decline in Kupffer cell deactivation of endotoxins and a slower, reduced ability to regenerate hepatocytes, fatty liver may progress to steatohepatitis and liver fibrosis.⁴⁹ Estrogen is believed to have antioxidant and antisteatotic effects in the liver through sensitization of Kupffer cells to toxic stimuli and in driving the anti-inflammatory polarization of Kupffer cells.^50,51 Perhaps loss of estrogen’s protective role in NAFLD may contribute to fibrotic progression in postmenopausal women. In addition, women’s adipose tissue distribution changes as estrogen declines in the elderly, shifting from gluteal-femoral area to abdominal region after menopause.⁵² A recent systemic review and meta-analysis concluded that women have lower risk of NAFLD than men in the general population but have higher risk of advanced fibrosis once NAFLD is established. The sex difference relationship is more evident among those aged over 50 years (RR, 1.56 [1.36–1.80]).⁵³ The exact mechanism of NAFLD in postmenopausal women is unclear, but sex hormones and fat distribution may play important roles in regulating steatohepatitis in the elderly. The present study, which demonstrated that NAFLD was associated with obesity and metabolic syndrome in the elderly, suggests that obesity phenotypes and metabolic health status are key features for further studies of NAFLD pathophysiology. In addition, obesity may be a more important contributor to fatty liver in elderly women than in men, perhaps because of the lack of estrogen protection and alterations in body composition as well as fat distribution. Further research is required to understand the sex-related difference in the elderly.

Our study indicates that obesity and metabolic syndrome have higher odds for advanced fibrosis in elderly individuals with NAFLD, and that using noninvasive fibrosis markers is more cost-effective and accessible for screening advanced fibrosis, especially in elderly women with obesity. Early management of liver fibrosis such as lifestyle modification, which includes changing dietary habits, weight reduction, and exercise can be encouraged. Referral to gastrointestinal specialist for a more intense workup including FibroScan and acoustic radiation force impulse can also be suggested early on. Thus, progression of NAFLD-related complications such as cirrhosis and HCC may therefore be avoided.

Most previous studies focused on NAFLD prevalence, and the association of such prevalence with obesity phenotypes and metabolic health status is limitedly investigated, especially in the elderly population. To the best of our knowledge, this study is the first to explore the association between different obesity phenotypes and liver fibrosis score in elderly individuals with fatty liver in Taiwan. However, our study has several potential limitations. First, this study only focused on one center; thus, its generalizability to other population is uncertain. Second, no biopsy-proven pathologic diagnosis of NAFLD has been found, and the cutoff values for NFS and FIB-4 were based on previous studies that only included a small number of elderly individuals.⁵⁴ Therefore, these values may not be appropriate for the elderly population. Furthermore, data on alcohol consumption, smoking status, past medical history, and current medication use were self-reported and may include past smokers and past drinkers; hence, these data may have some bias. Influence of exercise was not also considered. Moreover, agreement on a universally accepted definition of MHO, would improve the design of future studies and would facilitate better comparisons between studies. Larger studies with a more representative sample are required to obtain a more accurate description of the association between obesity phenotypes and NAFLD.

Conclusion

Metabolic syndrome and obesity were positively associated with a more advanced category of fibrosis according to NFS, which showed significant findings contrary to FIB-4 that only revealed a positive trend. Obesity seems to have a more detrimental effect than metabolic abnormalities per se on the severity of more advanced fibrosis in the elderly according to FIB-4 score, especially in women.