High Prevalence of Advanced Liver Fibrosis Assessed by Transient Elastography Among U.S. Adults With Type 2 Diabetes

Nonalcoholic fatty liver disease (NAFLD) has become the most common chronic liver disorder worldwide (1), is associated with a substantial economic and clinical burden (2), and is the most rapidly growing indication for liver transplant, ranking second in the United States (3). NAFLD is tightly connected with obesity, ectopic fat deposition and insulin resistance (4); thus, its prevalence is particularly high in individuals at high metabolic risk, such as patients with type 2 diabetes mellitus (T2DM) (5), who are also at higher risk of progression toward nonalcoholic steatohepatitis, advanced fibrosis, and cirrhosis.

The best predictor of future clinical outcomes related to NAFLD is the degree of fibrosis, and the gold standard for its measurement is liver biopsy (6). Nonetheless, given the prevalence of this condition, noninvasive and readily available tools to identify patients with advanced fibrosis were developed (7). These can be classified as blood-based biomarkers and imaging techniques. Although the former are easier to implement in routine clinical practice, there is low concordance between different tests (8), and their performance has been questioned, especially in patients with T2DM (9), with some authors advocating the direct use of transient elastography (TE) to screen for both steatosis and fibrosis in all patients with T2DM (10). However, this strategy has not yet been tested in unselected patients with T2DM in the United States, and available studies on the prevalence of advanced fibrosis and occult cirrhosis in this population mostly come from tertiary care centers (11).

Data show that in the 2017–2018 cycle of the National Health and Nutritional Examination Survey (NHANES), TE was performed for the first time in a U.S. nationally representative sample. We used data from the survey in the present study to obtain an estimate of the prevalence of steatosis and fibrosis in U.S. adults with T2DM and identify factors associated with these conditions.

This is an analysis of data from the 2017–2018 cycle of NHANES, which is conducted in the United States by the National Center for Health Statistics of the Centers for Disease Control and Prevention. NHANES is a cross-sectional survey program aimed at including individuals representative of the general, noninstitutionalized population of all ages. To this end, a stratified, multistage, clustered probability sampling design is applied with oversampling of non-Hispanic Black, Hispanic, and Asian persons, people with low income, and older adults. The survey consists of a structured interview conducted in the home, followed by a standardized health examination that includes a physical examination and laboratory tests. Full methodology of data collection is available elsewhere (12). The original survey was approved by the Centers for Disease Control and Prevention Research Ethics Review Board and written informed consent was obtained from all adult participants. The present analysis was deemed exempt by the Institutional Review Board at our institution because the data set used in the analysis was completely de-identified.

Body measurements including height (cm), weight (kg), and waist circumference (cm) were ascertained during the mobile examination center visit; BMI was calculated as weight in kilograms divided by height in meters squared, and obesity was defined as a BMI ≥30 kg/m². Hypertension was defined as a systolic blood pressure (SBP) value ≥140 mmHg and/or a diastolic blood pressure (DBP) value ≥90 mmHg or currently taking antihypertensive drugs (13). The remaining participants were further categorized as having optimal (SBP <120 mmHg and DBP <80 mmHg), normal (SBP 120–129 mmHg and/or DBP 80–84 mmHg) and high-normal BP (SBP 130–139 mmHg and/or DBP 85–89 mmHg). Diabetes was defined in accordance with the American Diabetes Association criteria if any of the following conditions were met: 1) A self-reported diagnosis of diabetes; 2) use of antidiabetic drugs; 3) a hemoglobin A1c (HbA_1c) level ≥6.5% (48 mmol/mol); 4) a fasting plasma glucose level ≥126 mg/dL; and 5) a random plasma glucose level ≥200 mg/dL (14).

Laboratory methods of measuring HbA_1c, lipid profile, platelet count, and levels of ALT, AST, γ-glutamyltranspeptidase (GGT), creatinine, and albumin are reported in detail elsewhere (15). LDL cholesterol level was calculated using the Friedewald et al. (16) formula. The Fibrosis-4 score, which is based on age, AST and ALT levels, and platelet count, was calculated as originally proposed (17). Hepatitis C virus infection was indicated by presence of viral RNA and/or a confirmed antibody test, and hepatitis B virus infection was indicated by a positive surface antigen test, as described (18).

Estimated glomerular filtration rate was computed according to the Chronic Kidney Disease Epidemiology Collaboration equation and chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate <60 mL/min/1.73 m². On the basis of the measured urine albumin-to-creatinine ratio (UACR), participants were defined as having normo-albuminuria (UACR <30 mg/g), microalbuminuria (UACR 30–300 mg/g), or macro-albuminuria (UACR ≥300 mg/g).

Information regarding diabetic retinopathy, smoking status, history of heart failure, coronary artery disease, and stroke were based on self-report. Cardiovascular disease (CVD) was defined as a composite of coronary artery disease and stroke or transient ischemic attacks.

In the 2017–2018 NHANES cycle, vibration-controlled TE was performed by NHANES technicians after a 2-day training program with an expert technician, using the FibroScan model 502 V2 Touch (Echosens, Paris, France) equipped with medium (M) and extra-large (XL) probes. The M probe was used initially unless the machine indicated use of the XL probe. Interrater reliability between health technicians and expert FibroScan technicians (tested on 32 participants) was 0.86 for stiffness (mean difference 0.44 ± 1.3 kPa) and 0.94 for controlled attenuation parameter (CAP) (mean difference 4.5 ± 19.8 dB/m).

Examinations were considered reliable only if ≥10 liver stiffness measurements (LSMs) were obtained after a fasting time of ≥3 h, with an interquartile range/median <30%. Median CAP values ≥274 dB/m, 290 dB/m, and 302 dB/m were considered indicative of S1, S2, and S3 steatosis, respectively, in accordance with a recent study by Eddowes et al. (19). We also applied the threshold for any steatosis proposed by Caussy et al. (11) (288 dB/m). A median LSM ≥8.2 kPa was considered indicative of significant fibrosis (≥F2), whereas values ≥9.7 kPa and 13.6 kPa were considered indicative of advanced fibrosis (F3), and cirrhosis (F4), respectively (19).

A total of 5,265 participants aged ≥20 years attended a mobile examination-center visit. We initially excluded individuals without diabetes or with probable type 1 diabetes (defined as a diagnosis at age <30 years and the use of insulin as the only antidiabetic therapy), resulting in a population of 996 patients with T2DM. Among these, 43 patients were considered ineligible for TE for different reasons (e.g., unable to lie down, currently pregnant, presence of an implanted electronic medical device, presence of lesions where measurements would be taken), and 40 additional patients were excluded because TE was not performed because of refusal or insufficient time for the examination. Of the remaining 913 patients, 88 (9.6%) had an incomplete TE because of fasting for <3 h (n = 31), inability to obtain 10 valid measures (n = 32), and an interquartile range/median LSM value ≥30% (n = 25). Thus, the final sample comprised 825 patients with T2DM and complete TE examination (Fig. 1).

All analyses were conducted using SAS, version 9.4 (SAS Institute Inc., Cary, NC), accounting for the complex survey design of NHANES. We used appropriate weighting for each analysis, as suggested by the National Center for Health Statistics. Data are expressed as numbers and weighted proportions for categorical variables and as weighted means ± SE for continuous variables.

Participants’ characteristics by liver steatosis and fibrosis status were compared using one-way ANOVA for continuous variables and the design-adjusted Rao-Scott χ² test for categorical variables. Logistic regression analysis was performed to evaluate the effect of different variables on the presence of steatosis and fibrosis. A biological plausibility approach was followed for the choice of predictors, including known risk factors for steatosis such as age, sex, race/ethnicity, diabetes duration, BMI, and liver enzymes, with the addition of platelet count and albumin levels for fibrosis. A two-tailed value of P < 0.05 was considered statistically significant.

Of the 825 patients with reliable TE examinations, 484 (53.7%) were examined using the M probe and 341 (46.2%) using the XL probe. Clinical features of the entire cohort and comparison between patients in the M and XL probe groups are shown in Supplementary Table 1.

The mean age of participants was 60.6 years and 52.9% were men. The proportion of female participants was higher among patients evaluated with the XL compared with the M probe. As expected, waist circumference and the prevalence of obesity were higher in the XL group than in the M group.

Patients evaluated in the XL group were more commonly non-Hispanic White and less commonly Asian. The two groups did not differ significantly for comorbidities such as hypertension, CVD, and CKD. Finally, results of liver function tests were similar between the two groups, with the exception of albumin level, which was significantly lower in the XL group.

Distribution of the study population according to the degree of liver steatosis is shown in Table 1. The weighted prevalence of any degree of steatosis was 73.8% (95% CI 68.5%–78.5%) and the number of patients with grade 1, 2, and 3 steatosis was 70 (7.2%, 95% CI 4.9%–10.8%), 54 (8.3%, 95% CI 4.5%–15.3%), and 433 (58.3%, 95% CI 51.5%–64.3%), respectively. The prevalence of steatosis according to the threshold proposed by Caussy et al. (11) (288 dB/m) was 67.5% (95% CI 62.0%–72.6%). A CAP interquartile range <30 dB/m was present in 46.6% of patients (95% CI 43.3%–50.0%).

Waist circumference, prevalence of obesity, and triglyceride, total cholesterol, fasting plasma glucose, and HbA_1c levels increased progressively from patients without steatosis to those with S3 steatosis, whereas a progressive decrease was evident for HDL cholesterol. Moreover, as the degree of steatosis increased, so did the proportion of non-Hispanic White and Hispanic patients, with an opposite trend for non-Hispanic Black and Asian participants. As expected, AST, ALT, and GGT values were positively associated with increasing steatosis grade (all P < 0.01). No significant differences were found in the prevalence of hypertension, CVD, CKD, and altered UACR.

As shown in Table 2, the weighted prevalence of significant fibrosis (≥F2) was 23.8% (95% CI 19.7%–28.7%); the number of patients with advanced fibrosis (≥F3) was 119 (15.4%, 95% CI 12.2%–19.0%) and 52 had cirrhosis (F4) (7.7%, 95% CI 4.8%–11.9%). For sensitivity analyses, we used LSM cutoffs with 90% specificity derived from Hsu et al. (20). Significant fibrosis, advanced fibrosis, and cirrhosis were present in 14.1% (95% CI 10.2%–17.9%), 8.1% (95% CI 5.1%–12.6%), and 3.9% (95% CI 2.1%–5.6%) of patients with thresholds of 10.1 kPa, 13.4 kPa, and 16.1 kPa, respectively. In univariate analysis, higher fibrosis stage was associated with greater waist circumference, and higher spot UACR, AST, ALT, and GGT values, whereas no significant differences were found in sex and ethnicity.

Distributions of liver steatosis and fibrosis in specific subgroups are shown in Supplementary Tables 2 and 3, respectively. No significant differences were found for sex and Hispanic ethnicity, whereas obese patients had a higher prevalence of both steatosis and advanced fibrosis.

We found a weak, although statistically significant, correlation between fibrosis 4 (FIB-4) score and LSM (r = 0.30, P < 0.001). Agreement between LSM and FIB-4 in identifying advanced fibrosis was poor both when using an FIB-4 cutoff of 1.3 (κ = 0.05, P = 0.02) and when using a cutoff of 2.67 (κ = 0.17, P < 0.01). Finally, 56.7% of patients with LSM ≥9.7 kPa had a FIB-4 score <1.3.

On multivariable logistic regression analysis, BMI and ALT values were positively associated with steatosis, whereas non-Hispanic Black race was associated with lower odds of having steatosis. No differences were found for age, sex, and diabetes duration (Table 3). Moreover, higher BMI, AST, and GGT values were positively associated with advanced fibrosis, whereas non-Hispanic Black race was associated with a decreased risk.

In this study, we report the prevalence of liver steatosis and fibrosis assessed by TE in a nationally representative sample of U.S. patients with T2DM. By applying validated cutoffs for both CAP and LSM values, we estimate that 67.5%–73.8% of patients may have evidence of steatosis, approximately 15% may have advanced (≥F3) fibrosis and 7.7% may have cirrhosis. Nonetheless, it is possible that TE overestimated advanced fibrosis and cirrhosis and that the true prevalence might be lower (21).

According to previous studies, T2DM is one of the most important risk factors for advanced fibrosis in patients with NAFLD (22), and patients with T2DM are two to three times more likely to die of liver-related causes compared with the general population (23). In 2016, the European Association for the Study of the Liver, European Association for the Study of Diabetes, and European Association for the Study of Obesity jointly published guidelines that recommend routine screening for NAFLD and advanced fibrosis in patients with T2DM (24). The guidelines suggest using abdominal ultrasound to make a diagnosis of steatosis and serum-based biomarkers (e.g., FIB-4 or NAFLD fibrosis score) to evaluate the risk of advanced fibrosis. Nonetheless, because they are based on metabolic variables, these scores underperform in patients with diabetes compared with the general NAFLD population (9,25), with different scores leading to different estimates in the number of patients with advanced disease (26).

On the other hand, because TE is easy to perform and does not depend on a calculation based on metabolic features, TE might be suitable for screening patients with T2DM. Furthermore, LSM can predict the occurrence of liver-related events, both as a baseline measure and when changes are prospectively evaluated (27).

Few studies have applied TE to patients with T2DM in a systematic way. In a recent study from Romania, Sporea et al. (28) evaluated 534 White patients attending a single center. Using the same cutoffs applied to our population, Sporea et al. found a prevalence of steatosis of 76.1%, with 19.6% of patients having advanced fibrosis and 8.2% with cirrhosis. Our results in term of steatosis also align with those of previous studies using MRI-proton density fat fraction or MRS, showing a prevalence of 65%–75% (29,30).

In contrast, our estimate of advanced fibrosis is higher than what was previously reported by Doycheva et al. (29) in 100 U.S. patients with T2DM using magnetic resonance elastography (7.1%). We believe that less stringent inclusion criteria in this study for alcohol consumption and known liver disease may explain, in part, the discrepancy. On the other hand, Bril et al. (9) recently found biopsy-proven advanced fibrosis in 19% of patients with T2DM and steatosis.

Also in line with previous studies, we found a strong positive association between BMI and increased LSM, whereas advanced fibrosis was less common in non-Hispanic Black patients. AST and GGT levels were also predictive of F3-F4, whereas ALT correlated poorly with disease severity, being only associated with increasing steatosis grade. Interestingly, GGT values, which are not considered in most noninvasive scores of fibrosis, also were identified as independent predictors of increased LSM in a French study evaluating the use of TE in T2DM (31).

In contrast with that study, however, our analysis shows that patients with advanced fibrosis and cirrhosis were not older than their counterparts, suggesting the need for a high index of suspicion even in relatively young individuals. Finally, as previously reported (32), we found only a weak correlation between LSM and FIB-4, and concordance was poor when original cutoffs were applied, suggesting that negative predictive value of FIB-4 may not be high in patients with T2DM.

Our study has the strength of a large sample of unselected patients from the general U.S. population and the use of one of the best-performing noninvasive tests for assessing liver steatosis and fibrosis. Moreover, the availability of the XL probe, which was essential given the prevalence of obesity in our population, made it possible to obtain valid LSM measurements in the vast majority of patients (93.5%). Nevertheless, we acknowledge the presence of some limitations. First, regarding the exclusion of other forms of chronic liver disease, we showed that the prevalence of viral hepatitis was low and unlikely affected the results, whereas data on alcohol intake were not available because they have not yet been published by the Centers for Disease Control and Prevention. Therefore, a formal diagnosis of NAFLD could not be obtained. Second, the absence of histologic data prevents us from reporting the exact prevalence of steatosis and advanced fibrosis according to the gold standard technique. Moreover, there is no universal cutoff guideline for CAP score, and its ability to distinguish between different degrees of liver steatosis is suboptimal (19). However, we used various cutoff points from previous studies performed in Western populations (11,19). Finally, TE has several limitations and is less accurate in the context of active liver inflammation, severe obesity, and liver congestion. As a consequence, its accuracy is lower when compared with magnetic resonance techniques for assessing both steatosis and fibrosis (20,33).

In conclusion, patients with T2DM from the general U.S. population have a high prevalence of both steatosis and advanced fibrosis, supporting routine screening for these conditions. Patients with high BMI and AST and GGT levels might be at particularly high risk and can be prioritized for liver assessment.

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

Author Contributions. S.C. and G.P. designed the study and wrote, reviewed, and edited the manuscript. S.C. participated in data analysis. T.M. researched and analyzed data and participated in writing and editing the manuscript. All authors approved the final version of the manuscript to be published. G.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.