Introduction
Metabolic dysfunction-associated fatty liver disease (MAFLD) refers to steatosis occurring in the setting of a metabolic risk condition, such as type 2 diabetes mellitus (T2DM),1,2 and is now recognized as the most prevalent chronic liver disease worldwide.3 T2DM is an important risk factor for MAFLD and vice-versa, and seems to accelerate the progression of liver disease.4–8 Despite the high prevalence and serious clinical implications of MAFLD in T2DM patients, it is often overlooked in clinical practice.4 As MAFLD entails considerable (extra-)hepatic morbidity and mortality, there is a need for increased awareness among all stakeholders (primary care physicians, specialists, and health policy-makers) for adding MAFLD as a frequent end-organ comorbidity of T2DM, along with micro- and macrovascular complications.4,9
EASL-EASD-EASO guidelines recommend screening high-risk patients for the presence of MAFLD and assessing the presence of advanced fibrosis using systematic calculation of noninvasive tests (NITs) for steatosis and fibrosis detection.4,9,10 These tests use readily available bioclinical parameters and can therefore be routinely used by general practitioners and diabetologists to detect patients who might benefit from further investigation. In the case of indeterminate results for fibrosis or suspected advanced fibrosis, patients should undergo one-dimensional ultrasound vibration-controlled transient elastography (VCTE) (Fibroscan; Echosens, Paris, France) for confirmation, and from abdominal Doppler ultrasound to assess liver surface, parenchyma, and vasculature. In the setting of elevated liver enzymes, other causes of liver disease should be ruled out. Few patients would eventually undergo liver biopsy, which remains the gold standard for MAFLD staging.
To demonstrate the feasibility and accuracy of such a systematic approach, there is a crucial need for prospective studies of MAFLD screening in high-risk patients such as those with T2DM. To date, few studies (mainly retrospective) are available,11–16 and most were focused on nonalcoholic fatty liver disease (NAFLD), thus excluding patients with significant alcohol consumption. That may not represent real-life practice. We therefore carried out a cross-sectional study targeting consecutive patients attending diabetes clinic for MAFLD and advanced fibrosis screening on the basis of NITs. Depending on individual results, this initial assessment was followed by abdominal Doppler ultrasound and/or VCTE in high-risk patients.
Methods
Type of study
We conducted a monocentric (tertiary center), cross-sectional study including T2DM outpatients attending the diabetes consultation of the Cliniques universitaires Saint-Luc (Brussels) between June 2021 and May 2022. The study protocol (No. B4032021000065) was in line with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institutional review board and ethics committee. All included patients provided written informed consent.
Inclusion and exclusion criteria
In the setting of diabetology consultations, ambulatory patients with T2DM, over the age of 18 years, able to read and understand the information letter, and benefiting from social security in health care were eligible. Patients under supervision or curatorship, deprived of liberty, and those with intercurrent disease with an estimated life expectancy of <6 months and/or inability to present to follow-up consultations or tests were excluded. Of note, to provide reliable results on the prevalence of the condition, patients who had previously been diagnosed with MAFLD were not excluded from the study. Patients with excessive alcohol consumption or other causes of secondary steatosis were also not excluded.
Study design
Informed consent forms were systematically collected during outpatient visits at the diabetes clinic over an enrollment period of 50 weeks. All included patients benefited from an initial bioclinical screening to calculate validated noninvasive scores for steatosis [fatty liver index (FLI),17 hepatic steatosis index,18 and NAFLD ridge score],19 and fibrosis [NAFLD fibrosis score (NFS),20 fibrosis-4 index (FIB-4),21,22 and Hepamet fibrosis score],23 aspartate aminotransferase (AST)-to-platelet ratio index (APRI), and isolated AST.24 Based on national and international guidelines,9,10 in those with positive screening for MAFLD using FLI or advanced liver fibrosis by NFS and FIB-4, the two most validated scores in the literature, patients were invited to undergo subsequent diagnostic tests such as abdominal Doppler ultrasound and/or VCTE. Patients with elevated transaminase levels were further assessed to at least exclude viral hepatitis B and C, alpha-1-antitrypsin deficiency, and biological markers of autoimmune hepatitis, i.e. elevated immunoglobulin G (Fig. 1). Further management was clinically guided according to current standards of care.
Analytical assessment
An interview on disease status, personal history of previous diseases, smoking status, alcohol consumption (units/week), and pharmacological therapy was collected on a routine basis. MAFLD was defined as evidence of liver steatosis by the positivity of one score based on laboratory and anthropometric parameters (i.e., FLI) and/or imaging (i.e., liver ultrasound), in the setting of a metabolic risk condition (i.e., T2DM in this study). MAFLD diagnosis is therefore based on positive criteria, as opposed to former NAFLD which, as the name suggests, requires the exclusion of excessive alcohol consumption and other causes of secondary steatosis.2
Diabetes was diagnosed by American Diabetes Association criteria.25 Macroangiopathy or established cardiovascular disease was defined as a composite of coronary artery disease (acute coronary syndromes, revascularization procedures, or stable angina), cerebrovascular disease (previous stroke, transient ischemic attack, or >35% carotid artery stenosis) and peripheral arterial disease requiring revascularization. Microangiopathy was defined as the presence of diabetic kidney disease, diabetic retinopathy, or diabetic neuropathy. Diabetic kidney disease was defined as chronic kidney disease (CKD) stage ≥3 (defined as a CKD-EPI estimated glomerular filtration rate of <60 mL/min/1.73 m2) and/or the presence of an altered albumin excretion rate defined as a urine albumin / creatinine ratio of >30 mg/g. Diabetic retinopathy was diagnosed by a dilated eye exam performed by an ophthalmologist.
Height (cm), body weight (kg), waist circumference (cm), systolic and diastolic blood pressure (mmHg) at rest were obtained during the diabetes clinic consultation. Body mass index (BMI) was reported as kg/m2. Hypertension was defined as taking antihypertensive treatment and/or by a repeated resting systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg. Plasma/serum creatinine, AST, alanine aminotransferase (ALT), gamma-glutamyltransferase (GGT), total cholesterol, high-density lipoprotein cholesterol (HDL-C), triglycerides, platelets, white blood cells, and albumin were measured at local laboratories. Low-density lipoprotein cholesterol (LDL-C) was calculated by Friedewald’s formula. Dyslipidemia was defined as taking lipid-lowering drugs and/or as an LDL-C≥70 mg/dL and/or a non-LDL-C≥100 mg/dL. VCTE quality criteria included a minimum of 10 measurements to obtain the median valid liver stiffness measurement (LSM) and interquartile range (IQR), an IQR / median LSM ratio of ≤0.3 × LSM for values >7 kPa, and a success rate of ≥60% in obtaining the 10 measurements.26,27 Liver stiffness cutoffs using medium (and extra-large) probes were: F0–F1<7.8 (6.4) kPa; F2 [7.8 (6.4)–12.5 (9.3)] kPa; F3 [12.5 (9.3)–22.3 (16.0)] kPa; and F4≥22.3 (16.0) kPa.28
Study endpoints
The primary endpoint was to assess the feasibility of outpatient systematic screening for MAFLD in T2DM patients. To do so, we determined the respective prevalence of steatosis and severe fibrosis using simple noninvasive tools. We estimated patient adherence to more accurate fibrosis screening by VCTE if indicated by bioclinical testing. Secondary endpoints were risk factors and clinical and/or biological criteria associated with steatosis or advanced fibrosis in this T2DM population.
Statistical analysis
The statistical analysis was performed with SPSS v29.0 (IBM Corp., Armonk, NY, USA). Continuous variables were reported as means±standard deviation (SD) and compared using the student unpaired t-test; or as medians and IQR and compared using the Wilcoxon rank-sum test. Categorical variables were reported as numbers and percentages and were compared using the chi-square test or Fisher’s exact test as appropriate. Logistic regression models were used to identify factors independently associated with MAFLD or advanced fibrosis. p-values ≤0.05 were considered significant.
Results
Study population
The baseline characteristics of the 213 T2DM patients who were included are shown in Table 1. Their mean age was 62 years and 67.1% were men. The vast majority of patients were overweight or obese (mean BMI 31.3 kg/m2), three men (1.4%) reported an alcohol consumption >30 g/day, and no women reported a consumption of >20 g/day. Only 33 patients (15.5%) had been previously evaluated for MAFLD in a hepatology consultation.
Table 1Patients’ characteristics for the total population, by MAFLD status and by advanced fibrosis status
| Population variable | Total (n=213) | No MAFLD (n=28) | MAFLD (n=185) | p-value | No advanced fibrosis (n=189) | Advanced fibrosis (n=24) | p-value |
|---|
| Demographic |
| Age in years, mean±SD | 62±12 | 67±12 | 61±12 | 0.035 | 62±13 | 62±11 | 0.982 |
| Female sex, n (%) | 70 (32.9) | 7 (25) | 63 (34.1) | 0.342 | 65 (34.4) | 5 (20.8) | 0.183 |
| Weight in kg, mean±SD | 91±18 | 77±12 | 93±18 | <0.001 | 90±17 | 99±20 | 0.027 |
| Height in cm, mean±SD | 170±10 | 169±10 | 171±9 | 0.555 | 170±10 | 170±8 | 0.774 |
| BMI in kg/m2, mean±SD | 31±5 | 27±3 | 32±5 | <0.001 | 31±5 | 34±7 | 0.033 |
| Waist circumference in cm, mean±SD | 109±13 | 97±8 | 110±12 | <0.001 | 108±12 | 116±14 | 0.004 |
| Lifestyle, n (%) |
| Alcohol consumption >30 g/d in M | 3 (1.4) | 0 | 3 (1.6) | 0.497 | 1 (0.5) | 2 (8.3) | 0.002 |
| Alcohol consumption >20 g/d in F | 0 | 0 | 0 | | 0 | 0 | |
| Smoking history | 81 (38.0) | 11 (39.3) | 73 (39.5) | 0.700 | 70 (37.0) | 14 (58.3) | 0.079 |
| History / comorbidities, n (%) |
| Hypertension | 146 (68.5) | 15 (57.1) | 130 (70.3) | 0.163 | 128 (67.7) | 18 (75.0) | 0.470 |
| Dyslipidemia | 177 (83.1) | 23 (82.1) | 154 (83.2) | 0.885 | 158 (83.6) | 19 (79.2) | 0.585 |
| Coronary artery disease | 41 (19.2) | 7 (25.0) | 34 (18.4) | 0.408 | 37 (19.6) | 4 (16.7) | 0.733 |
| Cerebrovascular disease | 14 (6.6) | 3 (10.7) | 11 (5.9) | 0.343 | 14 (7.4) | 0 | 0.168 |
| Peripheral arterial disease | 17 (8.0) | 5 (17.9) | 12 (6.5) | 0.039 | 15 (7.9) | 2 (8.3) | 0.946 |
| Diabetic kidney disease | 63 (29.6) | 16 (57.1) | 57 (70.3) | 0.311 | 56 (29.6) | 7 (29.2) | 0.963 |
| Chronic kidney disease, stage ≥3 | 35 (16.4) | 16 (57.1) | 29 (70.3) | 0.444 | 33 (17.5) | 2 (8.3) | 0.256 |
| Microalbuminuria, AER>30 mg/g creatinine | 46 (21.6) | 4 (14.3) | 42 (22.7) | 0.313 | 40 (21.2) | 6 (25) | 0.667 |
| Diabetic retinopathy | 23 (10.8) | 4 (14.3) | 19 (10.3) | 0.478 | 19 (10.1) | 4 (16.7) | 0.330 |
| Diabetic neuropathy | 19 (8.9) | 0 | 19 (10.3) | 0.076 | 16 (8.5) | 3 (12.5) | 0.514 |
| Sleep apnea syndrome | 31 (14.6) | 1 (3.6) | 30 (16.2) | 0.077 | 26 (13.8) | 5 (20.8) | 0.354 |
| Bariatric surgery | 8 (3.8) | 1 (3.6) | 7 (3.8) | 0.952 | 6 (3.2) | 2 (8.3) | 0.213 |
| Diabetes medications, n (%) |
| Metformin | 180 (84.5) | 22 (78.6) | 158 (85.4) | 0.352 | 159 (84.1) | 21 (87.5) | 0.667 |
| SGLT2-i | 50 (23.5) | 8 (28.6) | 42 (22.7) | 0.495 | 45 (23.8) | 5 (20.8) | 0.746 |
| DPP4-i | 26 (12.2) | 4 (14.3) | 130 (70.3) | 0.718 | 26 (13.8) | 0 | 0.052 |
| GLP1-RA | 43 (20.2) | 2 (7.1) | 41 (22.2) | 0.065 | 35 (18.5) | 8 (33.3) | 0.089 |
| Sulfonylurea | 48 (22.5) | 5 (17.9) | 43 (23.2) | 0.525 | 40 (21.2) | 8 (33.3) | 0.179 |
| Glitazone | 3 (1.4) | 0 | 3 (1.6) | 0.497 | 3 (1.6) | 0 | 0.534 |
| Glinide | 12 (5.6) | 1 (3.6) | 11 (5.9) | 0.612 | 10 (5.3) | 2 (8.3) | 0.543 |
| Insulin | 90 (42.3) | 15 (53.6) | 75 (40.5) | 0.193 | 80 (42.3) | 10 (41.7) | 0.951 |
| Cardiovascular medications, n (%) |
| ACE inhibitor / sartan | 120 (56.3) | 16 (57.1) | 104 (56.2) | 0.927 | 106 (56.1) | 14 (58.3) | 0.834 |
| Thiazide / indapamide | 53 (24.9) | 4 (14.3) | 47 (25.4) | 0.650 | 48 (25.4) | 5 (20.8) | 0.626 |
| Loop diuretic | 13 (6.1) | 2 (7.1) | 11 (5.9) | 0.811 | 11 (5.8) | 2 (8.3) | 0.633 |
| Βetablocker | 60 (28.2) | 4 (14.3) | 54 (29.2) | 0.395 | 54 (28.6) | 6 (25.0) | 0.714 |
| Calcium channel blocker | 41 (19.2) | 2 (7.1) | 39 (21.1) | 0.081 | 38 (20.1) | 3 (12.5) | 0.373 |
| Statin | 153 (71.8) | 20 (71.4) | 133 (71.9) | 0.959 | 136 (72.0) | 17 (70.8) | 0.908 |
| Ezetimibe | 44 (20.7) | 3 (10.7) | 41 (22.2) | 0.163 | 39 (20.6) | 5 (20.8) | 0.982 |
| Fibrate | 34 (16.0) | 1 (3.6) | 33 (17.8) | 0.055 | 28 (14.8) | 6 (25.0) | 0.199 |
| Acetylsalicylic acid | 82 (38.5) | 14 (50.0) | 68 (36.8) | 0.180 | 76 (40.2) | 6 (25.0) | 0.149 |
| Amiodarone | 3 (1.4) | 0 | 3 (1.6) | 0.497 | 3 (1.6) | 0 | 0.534 |
| Biological, median (IQR) |
| HbA1c as % | 7.0 (1.4) | 6.8 (0.9) | 7.1 (1.5) | 0.045 | 7.0 (1.3) | 6.7 (1.8) | 0.494 |
| White blood cells/mm3 | 7.20 (2.65) | 6.54 (2.63) | 7.23 (2.64) | 0.137 | 7.22 (2.55) | 6.38 (2.74) | 0.218 |
| Platelet count as × 103/mm3 | 239 (91) | 231 (54) | 242 (93) | 0.200 | 243 (88) | 175 (69) | <0.001 |
| Albumin in g/L | 45 (4) | 43 (5) | 46 (4) | 0.088 | 45 (4) | 45 (7) | 0.603 |
| AST in U/L | 22 (13) | 19 (5) | 23 (13) | 0.001 | 21 (9) | 43 (29) | <0.001 |
| ALT in U/L | 24 (18) | 16 (8) | 26 (20) | <0.001 | 22 (17) | 35 (32) | <0.001 |
| GGT in U/L | 26 (33) | 15 (7) | 28 (33) | <0.001 | 24 (27) | 70 (83) | <0.001 |
| Total cholesterol in mg/dL | 144 (59) | 139 (49) | 145 (60) | 0.117 | 144 (58) | 146 (56) | 0.715 |
| HDL-cholesterol in mg/dL | 42 (16) | 50 (20) | 41 (15) | 0.003 | 43 (16) | 41 (18) | 0.266 |
| Triglycerides in mg/dL | 136 (101) | 91 (74) | 138 (104) | <0.001 | 135 (95) | 138 (147) | 0.228 |
NITs for steatosis
A total of 77.0 % of the patients were classified in the high-risk category for steatosis (FLI>60), 18.2% in the indeterminate risk category (30≤FLI≤60), while only 4.8 % were classified as low-risk (FLI<30). When using hepatic steatosis index and NAFLD ridge score, an even lower proportion of patients were ascribed to the low-risk group (Fig. 2).
Abdominal Doppler ultrasound
Doppler ultrasound was offered to patients with an intermediate or high FLI. Eighteen patients declined and one patient with a high FLI died from cardiac arrest following food aspiration before further liver assessment. When evaluated by abdominal Doppler ultrasound, the hepato-renal echogenicity gradient was increased in 80.6% of patients at high or indeterminate risk (an FLI of >30) and in 84.0% of high-risk patients (an FLI of >60). Using a sequential combination of FLI and ultrasound (if the FLI was >30), a total of 185 patients (86.9%) were diagnosed with MAFLD. Twenty-two patients had dysmorphic liver and/or signs of portal hypertension and were offered a further assessment by VCTE. Two patients were suspected to have a hemangioma, which was confirmed by magnetic resonance imaging (MRI). None of the patients presented with liver lesions compatible with hepatocellular carcinoma.
NITs for fibrosis
In contrast to steatosis, the prevalence of scores consistent with advanced fibrosis varied greatly, ranging from 3.8% (FIB-4), 4.6% (Hepamet fibrosis score), and 19.0% (NFS). Whereas NFS seemingly classified a majority of T2DM patients in the intermediate risk group (59.0 %), the FIB-4 with age-adjusted cutoffs classified most patients in the low-risk group (75.1%; Fig. 2). Using sequential FLI and the combination of NFS and FIB-4 with age-adjusted cutoffs as recommended by the guidelines led to referring 29 patients (13.6%) who were offered further evaluation by VCTE.
VCTE
VCTE was offered to patients with dysmorphic liver features or signs of portal hypertension on abdominal Doppler ultrasound, as well as to patients with positive fibrosis screening by both NFS and FIB-4. Of 43 patients, three declined and one died from cardiac arrest after food aspiration and before VCTE evaluation. The acceptance rate for this additional testing was therefore very high (92.3%). Of 39 VCTE measurements, 37 met the quality criteria. In that select population, the mean controlled attenuation parameter was 325±47 dB/m and the LSM was 17.0±12.8 kPa. The spread of the elasticity module results is shown in Figure 3. Overall, 24 patients were classified as having advanced fibrosis, nine of whom were diagnosed with cirrhosis, i.e. F4 and dysmorphic liver or signs of portal hypertension. Of the 24 patients with VCTE-confirmed advanced fibrosis, 13 (54.2%) had never previously been evaluated by a hepatologist and therefore had newly diagnosed MAFLD that was already at an advanced stage.
Assessment for other causes of elevated transaminases
Sixty-one patients (28.6%) had elevated transaminases (local reference range, AST>36 and/or ALT>35 UI/L). Besides two patients whose excessive alcohol consumption might underlie elevated liver tests, none of the patients were found to have liver conditions other than MAFLD. Two patients had slightly decreased serum alpha-1-antitrypsin levels, and were subsequently asked to undergo phenotype determination of the protein.
Association of MAFLD and/or advanced fibrosis with demographic, clinical, and biochemical characteristics
Besides waist circumference, BMI, triglycerides, and GGT, all used for FLI calculation and thus collinear variables, there was a significant association between MAFLD status and lower age, absence of peripheral arterial disease, glycemic control (HbA1c), transaminase levels, and low HDL-C (Table 1). In multivariable models adjusted for age and sex, high HbA1c, high transaminases, and low HDL-c remained linked to MAFLD. In multivariable models adjusted for age, sex, BMI, smoking status, hypertension, and dyslipidemia (factors known to be associated with vascular disease), peripheral arterial disease was associated with decreased odds of MAFLD [OR 0.197 (95% CI: 0.040–0.963), p=0.045; Table 2].
Table 2Logistic regression model comparing the prevalence of population variables with or without MAFLD
| Population variable | MAFLD Odds ratio (95% CI) | p-value |
|---|
| Demographic and biochemical |
| Age in years | 0.961 (0.927–0.998) | 0.037 |
| Female sex, yes/no | 1.549 (0.625–3.840) | 0.345 |
| Waist circumference in cm | 1.134 (1.076–1.196) | <0.001 |
| BMI in kg/m2 | 1.406 (1.223–1.618) | <0.001 |
| Obesity, BMI≥30 kg/m2, yes/no | 7.557 (2.748–20.785) | <0.001 |
| Peripheral arterial disease, yes/no | 0.319 (0.103–0.988) | 0.048 |
| HbA1c as % | 1.534 (1.018–2.310) | 0.041 |
| AST in U/L | 1.088 (1.020–1.161) | 0.010 |
| ALT in U/L | 1.098 (1.040–1.159) | <0.001 |
| GGT in U/L | 1.062 (1.021–1.104) | 0.003 |
| HDL-C in mg/dL | 0.955 (0.926–0.984) | 0.003 |
| Multivariable adjustment |
| Age, sex-adjusted | 0.963 (0.928–0.999) | 0.042 |
| Waist circumference, age- and sex-adjusted | 1.164 (1.095–1.236) | <0.001 |
| BMI, age- and sex-adjusted | 1.384 (1.202–1.592) | <0.001 |
| Obesity, age- and sex-adjusted | 7.280 (2.628–20.164) | <0.001 |
| Peripheral arterial disease, age-, sex-, BMI-, smoking history-, hypertension-, and dyslipidemia-adjusted | 0.197 (0.040–0.963) | 0.045 |
| HbA1c, age- and sex-adjusted | 1.557 (1.010–2.400) | 0.045 |
| AST, age- and sex-adjusted | 1.085 (1.016–1.159) | 0.015 |
| ALT, age- and sex-adjusted | 1.097 (1.037–1.161) | 0.001 |
| GGT, age- and sex-adjusted | 1.059 (1.018–1.101) | 0.005 |
| HDL-C, age- and sex-adjusted | 0.951 (0.922–0.981) | 0.002 |
In addition to the parameters used for FIB-4 and NFS calculation (age, BMI, AST, ALT, platelet count, and albumin), there was a significant association between advanced fibrosis and excessive alcohol consumption (chi-square test, Table 1; univariate logistic regression, and multivariable regression adjusted for age and sex, Table 3). In multivariable models adjusted for age and sex, high waist circumference was also associated with increased odds of advanced fibrosis [OR 1.047 (95% CI: 1.011–1.084), p=0.010; Table 3].
Table 3Logistic regression model comparing the prevalence of population variables with or without advanced fibrosis
| Population variable | Advanced Fibrosis Odds ratio (95 % CI) | p-value |
|---|
| Demographic and biochemical |
| Age in years | 1.000 (0.966–1.034) | 0.982 |
| Female sex, yes/no | 0.502 (0.179–1.406) | 0.190 |
| Waist circumference in cm | 1.049 (1.014–1.086) | 0.005 |
| BMI in kg/m2 | 1.118 (1.036–1.207) | 0.004 |
| Obesity, BMI≥30 kg/m2, yes/no | 2.559 (0.973–6.731) | 0.057 |
| Alcohol consumption in units/week | 1.055 (0.995–1.119) | 0.073 |
| Excessive alcohol consumption, yes/no | 17.091 (1.489–196.231) | 0.023 |
| Insulin, yes/no | 0.973 (0.411–2.303) | 0.951 |
| Platelet count as ×103/mm3 | 0.978 (0.969–0.987) | <0.001 |
| AST in U/L | 1.057 (1.031–1.083) | <0.001 |
| ALT in U/L | 1.020 (1.005–1.035) | 0.009 |
| GGT in U/L | 1.015 (1.008–1.022) | <0.001 |
| Multivariable adjustment |
| Waist circumference, age- and sex-adjusted | 1.047 (1.011–1.084) | 0.010 |
| BMI, age- and sex-adjusted | 1.118 (1.035–1.208) | 0.004 |
| Obesity, age- and sex-adjusted | 2.637 (0.994–6.992) | 0.051 |
| Alcohol consumption, age- and sex-adjusted | 1.051 (0.992–1.114) | 0.094 |
| Excessive alcohol consumption, age- and sex-adjusted | 16.646 (1.367–202.640) | 0.027 |
| Platelet count, age- and sex-adjusted | 0.977 (0.968–0.987) | <0.001 |
| AST, age- and sex-adjusted | 1.061 (1.034–1.089) | <0.001 |
| ALT, age- and sex-adjusted | 1.024 (1.006–1.043) | 0.009 |
| GGT, age- and sex-adjusted | 1.015 (1.008–1.023) | <0.001 |
Discussion
Given the increasing prevalence of MAFLD, routine referral of all patients to specialized hepatologists is neither feasible nor sustainable. Clinicians are therefore challenged to identify a select target population at high risk of advanced MAFLD. Systematic screening with a simple algorithm of NITs including routine bioclinical parameters followed by readily available confirmatory imaging techniques such as Doppler ultrasound and/or VCTE, prospectively identified 86.9% of MAFLD, 11.3% of advanced fibrosis, and 4.2% cirrhosis patients in a population of 213 T2DM outpatients attending a diabetes clinic in a tertiary care center.
Our study is, to the best of our knowledge, the first to prospectively include consecutive outpatients with real-life use of NITs for systematic steatosis and fibrosis screening in a regular T2DM population. Few retrospective studies using NITs for steatosis and fibrosis assessment from T2DM population databases are available.11–15 Limitations of those studies mainly include their retrospective design. Consequently noninvasive scores may not have been appropriately calculated in all patients owing to lack of crucial information (e.g., abdominal circumference for FLI) or nonconcomitant collection of anthropometric and biological parameters. Secondly, those studies did not systematically assess the presence of other etiologies of steatosis or elevated liver tests such as excessive alcohol consumption or other causes of liver disease (e.g., viral infection). Finally, intermediate or high-risk noninvasive scores did not lead to medical counseling or further testing (e.g., imaging studies or liver biopsy) as per the guidelines. Therefore, the acceptance and applicability of these measures were not recorded.
A recent prospective study by Ajmera et al.16 assessed the prevalence of NAFLD, advanced fibrosis, and hepatocellular carcinoma in a population of 524 patients recruited in primary care or endocrinology clinics. The value of their study lies in the carrying out of MRI (proton density fat fraction, and elastography) in the vast majority of patients, which allowed for optimal noninvasive assessment of the degree of steatosis and fibrosis and for highlighting advanced fibrosis and even hepatocellular carcinoma in patients with low FIB-4 scores.29 However, their study has some limitations including possible recruitment bias, limited real-life availability of MRI, questionable usefulness of hepatic fat quantification, and possible bias resulting from exclusion of patients with excessive alcohol consumption.30
Large retrospective studies reported similar positivity rates of various NITs for steatosis and therefore a prevalence of MAFLD similar to that in our population.11–15 We confirmed the presence of hepatic steatosis in the majority of regular T2DM patients, which seems reasonable given the pathophysiology of these two interrelated comorbidities.4 Overall, the prevalence of MAFLD is higher than that of T2DM in the general population, and insulin resistance measured with the gold standard euglycemic-hyperinsulinemic clamp, is already present in patients with steatosis prior to T2DM onset.31 Mediators produced by the steatotic liver (hepatokines such as fetuin-A) may contribute to peripheral insulin resistance.32,33 The role of insulin resistance in the worsening of steatosis following the onset of T2DM is also well described.34
As NITs would classify nearly all T2DM patients at high or intermediate risk for steatosis, with the caveat of missing a subgroup of patients with low BMI, abdominal circumference and/or triglycerides due to more severe liver disease (cirrhosis or hepatocellular carcinoma), one might rather skip the NITs for steatosis and go directly to performing Doppler ultrasonography. Despite its lower sensitivity versus NITs for detecting hepatic steatosis (defined as >5% macrovesicular steatosis) because the sonographic hepatorenal index does not always significantly increase in mild steatosis (5–10% liver fat),9,35 liver ultrasound can yield invaluable additional information regarding the liver surface, parenchyma, and vasculature, alongside other abdominal organs of interest, in particular the spleen and splenic vein.
On one hand, we found a statistically-significant association between MAFLD and poor glycemic control, elevated transaminases and low HDL-C, which is consistent with available literature.36–40 We also found an association between MAFLD and absence of peripheral arterial disease. The latter is compatible with the paradoxical ocular protection of steatosis in T2DM patients,41,42 as there is an established association between lower-extremity arterial disease and diabetic retinopathy.43 Moreover, recent data from large cohorts indicate that patients with high visceral fat and low liver fat suffer from higher cardiovascular risk.44 However, that is contrary to other previous reports.45 A confounding factor for this association was the lower age of the MAFLD group, as shown by multivariable adjustments. Another confounding hypothesis was a more severe fibrosis level in patients without MAFLD, because the degree of steatosis may decrease with advanced fibrosis or cirrhosis,46 but has been invalidated as neither FIB-4, NFS, nor Hepamet fibrosis score were significantly higher than in the MAFLD group. On the other hand, we did not confirm the association that previous retrospective studies reported between liver steatosis and cardiovascular disease,47,48 (micro)albuminuria,12 or CKD.49,50 Moreover, we could not settle the puzzling results of previous retrospective studies about the association of steatosis and diabetic retinopathy.41,49 Neither could we show an association between certain glucose-lowering drugs and the absence of MAFLD. However, it is important to note that our study did not include long-term prospective follow-up. In the future, our database will allow us to study the evolution of MAFLD in patients particularly according to their treatments.
The prevalence of advanced fibrosis using NITs varied from 3.8% (FIB-4) and 4.2% (APRI) to 19.0% (NFS), which is similar to previous retrospective studies.12 When assessing LSM using available cutoffs28 in patients with a high risk combination of FIB-4 and NFS, or abnormal abdominal Doppler ultrasound (liver dysmorphism, signs of portal hypertension), we found 11.3 % advanced fibrosis and 4.2 % cirrhosis in T2DM patients. As VCTE was only performed in patients with a combination of high NITs results, we were not able to evaluate and compare the diagnostic performance of the scores. We confirmed the known association between a higher BMI and more severe MAFLD with advanced fibrosis.16 In our population, alcohol consumption in units per week tended to be higher in patients with liver fibrosis, and excessive alcohol consumption was a risk factor for advanced fibrosis, confirming previous data suggesting that alcohol use was a significant risk factor for the progression of liver disease in MAFLD.51,52 Although because concerning a small number of patients, this should be interpreted with much caution. Finally, we did not confirm the previously described association between liver fibrosis and CKD, cardiovascular disease, or insulin use in T2DM patients.12,16
The discrepancy of positivity rates between NITs shows that there are serious limitations of the sole use of a NIT for detection of liver fibrosis. First of all, NITs seem to perform best for detecting fibrosis in older, nonobese and nondiabetic patients.53 There are several potential reasons why NITs would underperform in T2DM.24,53,54 Indeed, patients with T2DM may only represent a relatively small part of the whole spectrum of MAFLD severity, resulting in potential spread effect. Moreover, certain glucose-lowering agents (e.g., glucagon-like peptide 1 receptor agonists, sodium-glucose co-transporter-2 inhibitors, glitazones) or lipid-lowering drugs (e.g., fibrates) could be potential confounders by modulating liver fat accumulation and/or measurements used to calculate the scores (e.g., ALT, AST, triglycerides, or BMI). In addition, many of the diagnostic tools were developed and validated in Caucasian populations and might not necessarily apply to other ethnicities. Finally, observational studies are always prone to selection bias that can affect generalizability of the study results. Specifically, there were concerns regarding use of NFS in T2DM because the algorithm included the presence of diabetes, leading to spectrum bias with an increase in the score for all patients and a decreased positive predictive value.54 In addition, it requires assessment of serum albumin, which is not a routine test at diabetes clinics. Conversely, an FIB-4 of <1.3 has been shown to have modest negative predictive value and might not accurately classify T2DM people as at low risk of advanced fibrosis.54 An option to reduce indeterminate results rate is therefore to combine different biomarkers of liver fibrosis.24,55 Second, an approach that solely bases referrals on fibrosis stage is insufficient. Indeed, it is likely to miss out on a subset of patients with nonalcoholic steatohepatitis (NASH) and early fibrosis who still need referral to a specialist because they are at high risk of developing advanced fibrosis or cirrhosis in the short term. That underlines the need for reliable noninvasive biomarkers of NASH that are currently not available for clinical routine use.24
Elevated transaminases were found in 28.6% of patients. That subgroup was further assessed for other liver diseases but eventually were confirmed as having only MAFLD or MAFLD and alcohol-related liver disease (two patients). However, we noticed that the majority of those patients had never been tested to rule out viral hepatitis B and C, alpha-1-antitrypsin deficiency, or signs of auto-immune hepatitis. We consider that elevated transaminases should not be trivialized in T2DM patients, and should at least once be subject to further assessment.
In conclusion, using simple bioclinical noninvasive tools to routinely triage T2DM patients with potentially severe liver disease is feasible. There is wide adherence of high-risk patients to noninvasive complementary testing with abdominal Doppler ultrasound and VCTE. The usefulness of noninvasive scores for steatosis detection in T2DM is questionable, as the vast majority of patients were classified as at high or indeterminate risk of steatosis, meaning that nearly all T2DM patients presented with MAFLD. A baseline abdominal Doppler ultrasound seems therefore appropriate in all T2DM patients to assess the liver surface, parenchyma, and vascularization. Systematic combined use of NFS and FIB-4, although having multiple shortcomings, allowed for detecting a significant number of patients with previously undiagnosed cirrhosis and/or advanced fibrosis. In particular, the study showed that the majority of patients could be managed by their treating physicians.