Non-invasive screening for NAFLD and advanced fibrosis in diabetes in primary care setting by MRI and MRE

SUMMARY

Background

Current guidelines do not recommend screening for nonalcoholic fatty liver disease (NAFLD) or advanced fibrosis. Patients with type 2 diabetes mellitus (T2DM) are known to be at increased risk for NAFLD and advanced fibrosis.

Aim

We aimed to assess the feasibility in diabetics in a primary care setting of screening for NAFLD and advanced fibrosis by using non-invasive magnetic resonance imaging (MRI) to estimate the hepatic proton density fat fraction (MRI-PDFF) and magnetic resonance elastography (MRE) to estimate hepatic stiffness.

Methods

We performed a cross-sectional analysis of a prospective study that included 100 (53% men) consecutively enrolled diabetics who did not have any other etiology of liver disease. All patients underwent a standardized research visit, laboratory tests, MRI-PDFF, and MRE.

Results

Mean (± SD) age and BMI was 59.7 (±11.2) years and 30.8 (±6.5) kg/m², respectively. The prevalence of NAFLD (defined as MRI-PDFF ≥ 5%) and advanced fibrosis (defined as MRE ≥ 3.6 kPa) was 65% and 7.1%, respectively. One patient with advanced fibrosis had definite hepatocellular carcinoma. When compared to those without NAFLD, patients with NAFLD were younger (p=0.028) and had higher mean BMI (p=0.0008), waist circumference (p<0.0001) and prevalence of metabolic syndrome (84.6% vs. 40.0%, p<0.0001). Only 26% of those with NAFLD had elevated ALT.

Conclusions

This proof-of-concept study demonstrates that T2DM has significant rates of both NAFLD and advanced fibrosis. Concomitant screening for NAFLD and advanced fibrosis by using MRI-PDFF and MRE in T2DM is feasible and may be considered after validation in a larger cohort.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease in the Western world with an estimated prevalence in the United States ranging between 19–46% depending on the study population characteristics and diagnostic modality.^{1, 2} NAFLD is a frequent finding in patients with type 2 diabetes mellitus (T2DM) due to their common underlying pathogenic mechanism of insulin resistance.^{3, 4} Previous studies on the prevalence of NAFLD in T2DM were based on abdominal ultrasound evaluation,^{5, 6} magnetic resonance spectroscopy (MRS)^{7, 8} or small samples of patients with liver biopsies.^{2, 9} Due to the differences in the study methodologies and diagnostic techniques, the reported prevalence of NAFLD in T2DM patients ranges broadly between 43% and 94%.^{2, 5–9} It is also known that the presence of T2DM is an independent predictor of advanced fibrosis in NAFLD,¹⁰ with a greater prevalence of cirrhosis in diabetic compared to non-diabetic patients.¹¹ However, the exact prevalence of NAFLD in T2DM as well as the utility of screening for NAFLD in T2DM remain uncertain.

With the steadily increasing prevalence of obesity and T2DM, the health burden of NAFLD in diabetes raises an important concern and imposes the need for early identification of patients who are at an increased risk of progressive liver disease and cirrhosis. Currently, routine screening of NAFLD in either diabetics (as a high-risk group for progression of liver disease) or in the general population is not recommended due to controversies regarding the lack of accurate non-invasive diagnosis of NAFLD and advanced fibrosis and limited therapeutic options.¹²

Serum aminotransferase elevation is an inaccurate marker of NAFLD as normal alanine aminotransferase (ALT) has been noted in up to 86% of patients with NAFLD.⁵ Abdominal ultrasound lacks sensitivity for mild steatosis (< 30%) and has high inter- and intra-observer variability.¹³ Computed tomography (CT) has limited sensitivity for mild disease and is associated with radiation exposure. Therefore, liver biopsy remains the gold standard for the diagnosis and staging of NAFLD and assessment of fibrosis.¹² However, liver biopsy is an invasive procedure, which is limited due to its associated cost and risk of complications, including pain, infection, bleeding, and very rarely even death. Furthermore, liver biopsy has significant sampling variability, and inter-observer variability, and it is impractical tool for screening of patients who are at increased risk for presence of NAFLD and advanced fibrosis.¹⁴

Therefore, we aimed to investigate the feasibility of screening for NAFLD in T2DM using magnetic resonance imaging-estimated proton density fat fraction (MRI-PDFF), a novel, recently validated imaging-based biomarker which allows for the quantification of liver fat content in the entire liver that correlates strongly with MR spectroscopy – measured liver fat^15–17 and liver histology-determined steatosis grade.^18–20 We also aimed to assess the usefulness of magnetic resonance elastography (MRE), a noninvasive imaging modality shown to differentiate advanced from non-advanced fibrosis,^{21, 22} to screen our cohort for advanced fibrosis. Our hypotheses were that the feasibility of using MRI-PDFF and MRE to screen for NAFLD and advanced fibrosis would be excellent with no or low failure rates, and that asymptomatic diabetics in primary care setting would have high prevalence of NAFLD and advanced fibrosis. We followed STROBE guidelines for the reporting of prospectively conducted cross-sectional studies.²³

METHODS

Study design and setting

This is a cross-sectional analysis of a prospective study involving 100 consecutive eligible adult patients with T2DM recruited via newspaper advertisement and from primary care practices in the greater San Diego area. The patients underwent a standardized research visit including history, physical exam, anthropometric measurements, and biochemical and serological laboratory tests at the Clinical and Translational Research Institute, University of California, San Diego (UCSD). Within 30 days of the research visit, patients were scanned at the UCSD MR3T Research Laboratory and underwent same day MRI and MRE to estimate hepatic PDFF and hepatic stiffness, respectively. The study was approved by the UCSD Institutional Review Board. All enrolled patients provided written informed consent before data collection.

Derivation of cohort

Between March 2013 and September 2014, one hundred ninety-five subjects were screened. Ninety-five were excluded and 100 subjects were enrolled in the study. See Supplementary Figure 1 for details of cohort derivation.

Patient Population

We included patients with T2DM, age 21 years or older, diagnosed with T2DM according to the American Diabetes Association (ADA) clinical practice recommendations-2012²⁴: hemoglobin A1c ≥ 6.5% or fasting plasma glucose ≥ 126 mg/dL or 2-hour plasma glucose ≥ 200mg/dL during an oral glucose tolerance test (OGTT) or patient with classic symptoms of hyperglycemia or hyperglycemic crisis, a random plasma glucose ≥ 200mg/dL. All enrolled patients were asymptomatic with respect to liver disease and did not have any history of known or diagnosed liver disease.

Exclusion criteria: Subjects were excluded if they had evidence of any other chronic liver disease shown by the presence of hepatitis B surface antigen (HBsAg) or hepatitis C antibody (anti-HCV Ab), documented history of any other liver disease (such as alpha-1 antitrypsin deficiency, autoimmune hepatitis, drug-induced liver injury, primary biliary cirrhosis, primary sclerosing cholangitis or any other liver disease), or cirrhosis, bile duct obstruction on prior radiology tests, use of medications associated with secondary NAFLD (corticosteroids, tamoxifen, amiodarone, methotrexate), major diabetic end-organ damage (advanced cardiovascular disease requiring coronary artery bypass graft surgery or congestive heart failure; chronic kidney disease, stage ≥ 3), alcohol consumption of more than 30 grams per day in the previous 10 years or greater than 10 grams per day in the previous year assessed with the Alcohol Use Disorders Identification Test (AUDIT) questionnaire²⁵, positive HIV test, pregnancy or contraindications to MRI.

Clinical assessment and formulas

All patients were evaluated at the UCSD NAFLD Clinical and Translational Research Institute. A detailed history was obtained from all patients. A physical exam including vital signs, height, weight, and anthropometric measurements was performed by a trained investigator. Alcohol consumption was assessed by using the AUDIT questionnaire, a validated tool used to screen for heavy drinking and/or active alcohol abuse or dependence.²⁵ All patients had blood drawn in a fasting state for laboratory analysis.

Body mass index (BMI) was calculated by dividing body weight (in kilograms) by the square of the height (in meters). Homeostatic model assessment of insulin resistance (HOMA-IR) was calculated in a fasting state based on the formula: HOMA-IR = [glucose (mg/dL) x insulin (mIU/mL)]/405. Adipose tissue IR (Adipo-IR) was calculated in a fasting state using the formula: Adipo-IR = free fatty acids (mmol/L) x insulin (mIU/mL).

Covariates

The following demographic and anthropometric characteristics were collected: age, gender, race, BMI and waist circumference. Clinical characteristics included: presence of hypertension (defined as blood pressure ≥ 130/85 mmHg or use of antihypertensive medication), metabolic syndrome based on the American Heart Association and the National Heart, Lung, and Blood Institute criteria²⁶ (3 out of 5 conditions present: hypertension as defined above; fasting glucose ≥ 100 mg/dL (5.6 mmol/L) or receiving therapy for hyperglycemia; hypertriglyceridemia ≥ 150 mg/dL (1.7 mmol/L); high-density lipoprotein cholesterol (HDL-C) < 50 mg/dL (1.29 mmol/L) in women and < 40 mg/dL (1.03 mmol/L) in men; waist circumference ≥ 102 cm in men or ≥ 88 cm in women, or ≥ 90 cm in Asian men and ≥ 80 cm in Asian women), duration of diabetes, diabetic complications (diabetic retinopathy, neuropathy or nephropathy), current treatment of diabetes, antilipemic therapy, and current therapy with vitamin E.

Primary and secondary outcomes

The primary outcome of this study was the assessment of the feasibility of screening for NAFLD in T2DM in primary care setting by using MRI to estimate PDFF. Our secondary outcome was the evaluation of the feasibility of screening for advanced fibrosis in T2DM in primary care setting by using MRE.

MR Examinations

MR examinations were performed using a 3T research scanner (GE Signa EXCITE HDxt, GE Healthcare, Waukesha, WI) at the UCSD MR3T Research Laboratory. The MR examinations were performed by experienced MR technologists and analyzed by a single image analyst who was blinded to clinical and biochemical data. Patients were scanned in the supine position with a torso phased-array coil placed over the abdomen. To reduce potential physiologic confounding factors, patients were instructed to fast for a minimum of four hours before the exam. Two MR techniques were performed: MRI to estimate hepatic PDFF for NAFLD screening and MRE to estimate hepatic stiffness for advanced fibrosis screening.

For screening of NAFLD: MRI-PDFF protocol for assessment of liver fat content

The MRI-PDFF protocol has been previously described in details.^{19, 27} For full protocol, see supplementary materials.

Rationale for using MRI-PDFF for NAFLD screening

MRI-PDFF was chosen to screen for NAFLD, as this is an accurate and robust quantitative imaging-based biomarker that allows for precise fat quantification in all nine segments of the liver.^{20, 28, 29} It has been proven to correlate well with MRS^{15–17, 20} and histology-proven steatosis grade.^{18–20, 30} This technique is accurate, reproducible,³¹ has high inter- and intra-examination repeatability,^{15, 28} and can serve as a marker for longitudinal assessment.

For screening of advanced fibrosis: MRE protocol for assessment of advanced fibrosis

The MRE protocol has been previously described in detail.²¹ For full protocol, see supplementary materials.

Rationale for using MRE for screening of advanced fibrosis

MRE is an accurate, safe, non-invasive imaging with excellent diagnostic accuracy in differentiating between normal liver and mild fibrosis (stage 0–2) and between non-advanced fibrosis and advanced fibrosis (stage 3–4).^{22, 32} A recent prospective study has confirmed this finding with the diagnostic accuracy of AUROC 0.92 (p-value < 0.0001). The most optimal threshold for the diagnosis of advanced fibrosis on MRE was 3.6 kPa with a negative predictive value of 0.97 (95% CI, 0.91–0.99).²¹

Statistical analysis

Patients’ demographic, anthropometric, clinical and biochemical characteristics were summarized. Categorical variables were shown as counts and percentages and associations were tested using a Chi-square test (χ²) or Fisher’s exact test. Normally distributed continuous variables were shown as mean (+/− standard deviation) and differences between groups were analyzed using a two-independent- samples t-test. Values that were not normally distributed were reported as medians and interquartile ranges (IQR) with differences assessed using the Wilcoxon-Mann-Whitney test. A two-tailed p-value of <0.05 was considered statistically significant for all analyses. Statistical analyses were performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

Baseline characteristics

Mean (±SD) age and BMI was 59.7 (±11.2) years and 30.8 (±6.5) kg/m², respectively. 53% were men and the mean duration of T2DM was 8.5 years. 69% had metabolic syndrome and 66% had hypertension. Detailed description of anti-diabetic and antilipemic therapies is listed in Supplementary Table 1. Elevated ALT (>40 U/L) was noted only in 18% with median ALT of 24 U/L.

Primary Outcome

All 100 patients successfully underwent MRI to estimate PDFF. NAFLD as assessed by MRI-PDFF ≥ 5% was present in 65% of the patients (Figure 1). There were no technical failures.

Figure 1. Prevalence of NAFLD and advanced fibrosis among patients with type 2 diabetes in primary care.

Patients with type 2 diabetes in the primary care setting were screened for NAFLD with magnetic resonance imaging-estimated proton density fat fraction (MRI-PDFF). NAFLD was defined by the presence of hepatic steatosis ≥ 5% on MRI-PDFF. Screening for advanced fibrosis was performed using magnetic resonance elastography (MRE) with a threshold of 3.6 kPa to identify those with advanced fibrosis.

Demographic, clinical, biochemical and imaging characteristics in diabetics with NAFLD compared to those without NAFLD

Table 1 presents a detailed description of patients with and without NAFLD. In comparison to those without NAFLD, patients with NAFLD were younger (p = 0.028), and had higher mean BMI (32.5 vs. 27.6, p = 0.0008), mean waist circumference (106.6 cm vs. 95.1 cm, p <0.0001), and prevalence of metabolic syndrome (84.6% vs. 40%, p <0.0001). Patients with NAFLD were more likely to have elevated aspartate aminotransferase (AST) and ALT although only 26.2% had elevated ALT. Average ALT and gamma-glutamyl transpeptidase (GGT) were higher in those with NAFLD but these were within normal reference limits. Lower HDL-C and higher triglyceride levels characterized lipid profile of NAFLD patients when compared to those without NAFLD. The duration of diabetes and control of diabetes as assessed by hemoglobin A1c did not differ between the NAFLD versus the non-NAFLD group. However, as expected, compared to non-NAFLD group, participants in the NAFLD group had higher fasting glucose (p = 0.0366), fasting insulin (p = 0.0151), HOMA-IR (p = 0.0007), and adipo-IR (p = 0.0033).

Table 1.

Baseline population characteristics and comparison between diabetics with and without NAFLD

	N	NAFLD (MRI-PDFF ≥ 5%) (N=65)	No NAFLD (MRI-PDFF <5%) (N=35)	p-value
Demographics
Male, n (%)	100	33 (50.8%)	20 (57.1%)	0.5425
Age (yrs), mean ± SD	100	57.9 ± 11.3	63.1 ± 10.5	0.0280
Race	100			0.1125
Am Indian/Alaskan Native, n (%)		1 (1.5%)	0
Asian, n (%)		9 (13.8%)	2 (5.7%)
Black, n (%)		5 (7.7%)	12 (12%)
White, n (%)		34 (52.3%)	22 (62.9%)
Hispanic, n (%)		16 (24.6%)	4 (11.4%)
Anthropometric
BMI (kg/m2), mean ± SD	100	32.5 ± 5.4	27.6 ± 7.2	0.0008
Waist (cm), mean ± SD	100	106.6 ± 12.6	95.1 ± 12.9	<0.0001
Clinical
Hypertension, n (%)	100	42 (64.6%)	24 (68.6%)	0.6904
Metabolic syndrome, n (%)	100	55 (84.6%)	14 (40.0%)	<0.0001
Duration of DM (years)	84	7.0 (8.0)	11.0 (11.0)	0.1869
Oral hypoglycemic therapy, n (%)	100	55 (84.6%)	26 (74.3%)	0.2091
Metformin, n (%)	100	49 (75.4%)	25 (71.4%)	0.6671
Thazolidinediones, n (%)	100	2 (3.1%)	0	0.5404
Antilipemic therapy, n (%)	100	37 (56.9%)	19 (54.3%)	0.7999
Insulin therapy, n (%)	100	13 (20.0%)	10 (28.6%)	0.3313
Vitamin E, n (%)	100	1 (1.5%)	1 (2.9%)	1.0000
Biochemical
AST (U/L), median (IQR)	100	22.0 (12.0)	20.0 (10.0)	0.1965
Elevated AST (>40 U/L), n (%)	100	8 (12.3%)	0	0.0477
ALT (IU/L), median (IQR)	100	29.0 (23.0)	19.0 (12.0)	0.0003
Elevated ALT (>40 U/L), n (%)	100	17 (26.2%)	1 (2.9%)	0.0027
GGT (U/L), median (IQR)	99	27.5 (19.0)	20.0 (11.0)	0.0119
Alkaline phosphatase (U/L), median (IQR)	100	72.0 (20.0)	77.0 (44.0)	0.6078
HDL-C (mmol/L), median (IQR)	100	1.27 (0.49)	1.58 (0.85)	0.0360
LDL-C (mmol/L), median (IQR)	100	2.28 (1.14)	2.28 (1.27)	0.7124
Triglycerides (mmol/L), median (IQR)	100	1.83 (0.85)	1.04 (0.65)	<0.0001
Platelet (109/L), mean ± SD	100	261.8 ± 73.0	247.7±66.4	0.3447
HbA1C (%), median (IQR)	100	7.1 (1.6)	7.0 (1.3)	0.2592
Ferritin (pmol/L), median (IQR)	98	200.0 (382.0)	155.0 (283.1)	0.2062
Fasting glucose (mmol/L), median (IQR)	98	7.33 (3.44)	5.89 (3.00)	0.0366
Fasting insulin (mIU/mL/ pmol/L), median (IQR)	94	17.5/121.5 (20.0/138.9)	11.0/76.4 (14.0/97.2)	0.0151
HOMA-IR, median (IQR)	92	5.8 (5.3)	3.0 (3.7)	0.0007
FFA, (mmol/L), mean ± SD	94	0.55 ± 0.21	0.52 ± 0.27	0.5181
Adipo –IR, median (IQR)	93	9.1 (10.9)	5.1 (8.4)	0.0033
Imaging
MRI-PDFF (%), median (IQR)	100	12.3 (9.2)	2.7 (1.9)	<0.0001
NAFLD (MRI-PDFF ≥ 5%), n (%)	100
MRE (kPa), median (IQR)	98	2.6 (0.7)	2.4 (0.5)	0.1127
Advanced Fibrosis (MRE ≥ 3.6 kPa), n (%)	98	5 (7.8%)	2 (5.9%)	1.0000

p-values: n (%) → chi-square test or Fisher’s exact test, mean ± SD → two independent samples t-test, median (IQR) → Wilcoxon-Mann-Whitney test

Abbreviations: Adipo-IR: adipose tissue insulin resistance, ALT: alanine aminotransferase, AST: aspartate aminotransferase, BMI: body mass index, DM: diabetes mellitus, FFA, free fatty acids, GGT: gamma-glutamyl transpeptidase, HbA1C: hemoglobin A1c, HDL-C: high-density lipoprotein cholesterol, HOMA-IR: Homeostatic Model Assessment of Insulin Resistance, NAFLD: nonalcoholic fatty liver disease; MRE: magnetic resonance elastography, MRI-PDFF: magnetic resonance imaging-estimated proton density fat fraction.

Secondary outcome

Ninety-eight patients (98%) underwent MRE. Two patients did not have MRE due to scheduling conflict. There were no MRE technical failures. The prevalence of advanced fibrosis (defined as MRE ≥ 3.6 kPa) was 7.1% (7 out of 98 patients who underwent MRE) (Figure 2A).

Figure 2.

Figure 2A. Hepatic steatosis by MRI-PDFF (upper row) and liver stiffness by MRE (lower row) in 4 different patients. (A) Example of a patient with no NAFLD and no advanced fibrosis, (B) patient with NAFLD, but no advanced fibrosis, (C) patient without NAFLD, but with advanced fibrosis, and (D) patient with both NAFLD and advanced fibrosis. Fat fraction on MRI-PDFF was expressed in percentages and liver stiffness in kPa, shown underneath each image.

Figure 2B. MRI-PDFF, MRE, and Gadoxetic acid-enhanced 3T MRI of the patient with hepatocellular carcinoma. (A) MRI-PDFF shows 5.8 % hepatic steatosis, (B) MRE shows significant liver stiffness of 6.8 kPa, (C1) Pre-contrast 3D MRI image, (C2) Late arterial phase image shows 3-cm hyperenhancing lesion (arrow) in segment VI of the liver, (C3) Portal venous phase with definite “washout” (arrow) and “capsule” (arrowhead) of the lesion, typical features confirming the diagnosis of HCC; (C4) Hepatobiliary phase reveals markedly hypointense lesion (arrow)

Other outcomes

One patient was noted to have definite hepatocellular carcinoma (HCC) (1% of the entire population) (Figure 2B).

Demographic, clinical, biochemical and imaging characteristics in diabetics with and without advanced fibrosis

Table 2 provides detailed comparison between groups stratified by presence or absence of advanced fibrosis. Two patients with advanced fibrosis had MRI-PDFF < 5%, likely reflecting more advanced liver disease. There was no difference between the two groups in age, gender, ethnicity, metabolic syndrome or BMI. As expected, the advanced fibrosis group had higher AST and lower low-density lipoprotein cholesterol (LDL-C), both suggestive of advanced liver disease. Similar to the NAFLD group, advanced fibrosis group was also characterized by significantly elevated fasting serum insulin (p = 0.0196), higher IR as assessed by HOMA (p = 0.0452) and adipo-IR (p = 0.0062).

Table 2.

Comparison between diabetics with and without advanced fibrosis

	Advanced Fibrosis (MRE ≥ 3.6kPa) (N=7)	No Advanced Fibrosis (MRE < 3.6kPa) (N=91)	p-value
Demographics
Male, n (%)	5 (71.4%)	47 (51.7%)	0.4423
Age (yrs), mean ± SD	67.0 ± 8.9	59.2 ± 11.3	0.0779
Race			0.9397
Am Indian/Alaskan Native, n (%)	0	1 (1.1%)
Asian, n (%)	0	11 (12.1%)
Black, n (%)	1 (14.3%)	11 (12.1%)
White, n (%)	5 (71.4%)	49 (53.9%)
Hispanic, n (%)	1 (14.3%)	19 (20.9%)
Anthropometric (mean ± SD)
BMI (kg/m2), mean ± SD	31.0 ± 8.3	30.8 ± 6.4	0.9396
Waist (cm), mean ± SD	104.0 ± 22.9	102.3 ± 13.1	0.8522
Clinical
Hypertension, n (%)	7 (100.0%)	57 (62.6%)	0.0921
Metabolic syndrome, n (%)	5 (71.4%)	63 (69.2%)	1.0000
Duration of DM (years)	10.0 (3.0)	8.0 (8.0)	0.8459
Oral hypoglycemic therapy, n (%)	6 (85.7%)	74 (81.3%)	1.0000
- Metformin, n (%)	6 (85.7%)	67 (73.6%)	0.6743
- Thiazolidinediones, n (%)	0	2 (2.2%)	1.0000
Antilipemic therapy, n (%)	6 (85.7%)	48 (52.8%)	0.1253
Insulin therapy, n (%)	2 (28.6%)	19 (20.9%)	0.6401
Vitamin E, n (%)	0	2 (2.2%)	1.0000
Biochemical
AST (U/L), median (IQR)	28.0 (19.0)	20.0 (10.0)	0.0154
Elevated AST (>40 U/L), n (%)	3 (42.9%)	5 (5.5%)	0.0110
ALT (IU/L), median (IQR)	28.0 (22.0)	23.0 (16.0)	0.2878
Elevated ALT (>40 U/L), n (%)	2 (28.6%)	16 (17.6%)	0.6090
GGT (U/L), median (IQR)	38.0 (36.0)	25.0 (16.0)	0.2163
Alkaline phosphatase (ALP) (U/L), median (IQR)	73.0 (29.0)	73.0 (31.0)	1.0000
HDL-C (mmol/L), median (IQR)	1.45 (0.49)	1.34 (0.72)	0.7825
LDL-C (mmol/L), median (IQR)	1.78 (0.85)	2.33 (1.11)	0.0372
Triglycerides (mmol/L), median (IQR)	1.38 (0.75)	1.58 (1.00)	0.5302
Platelet (109/L), mean ± SD	217.0 ± 115.1	259.8 ± 67.0	0.3668
HbA1C (%), median (IQR)	6.7 (1.5)	7.1 (1.6)	0.8684
Ferritin (pmol/L), median (IQR)	184.3 (343.8)	186.5 (283.1)	0.6828
Fasting glucose (mmol/L), median (IQR)	6.97 (5.55)	6.89 (2.72)	0.7969
Fasting insulin (mIU/mL / pmol/L), median (IQR)	45.0/312.5 (114.0/791.7)	15.0/104.2 (16.0/111.1)	0.0196
HOMA-IR, median (IQR)	9.1 (14.7)	5.0 (5.4)	0.0452
FFA, (mmol/L), mean ± SD	0.65 ± 0.28	0.53 ± 0.23	0.2418
Adipo –IR, median (IQR)	22.6 (28.1)	7.4 (8.3)	0.0062
Imaging
MRI-PDFF (%), median (IQR)	6.4 (10.6)	8.2 (9.7)	0.7932
MRE (kPa), median (IQR)	4.2 (2.0)	2.5 (0.6)	<0.0001
NAFLD (MRI-PDFF ≥5%), n (%)	5 (71.4%)	59 (64.8%)	1.0000

p-values: n (%) → chi-square test or Fisher’s exact test, mean ± SD → two independent samples t-test, median (IQR) → Wilcoxon-Mann-Whitney test

Abbreviations: Adipo-IR: adipose tissue insulin resistance, ALT: alanine aminotransferase, AST: aspartate aminotransferase, BMI: body mass index, DM: diabetes mellitus, FFA, free fatty acids, GGT: gamma-glutamyl transpeptidase, HbA1C: hemoglobin A1c, HDL-C: high-density lipoprotein cholesterol, HOMA-IR: Homeostatic Model Assessment of Insulin Resistance, NAFLD: nonalcoholic fatty liver disease; MRE: magnetic resonance elastography, MRI-PDFF: magnetic resonance imaging-estimated proton density fat fraction.

When divided into different age groups, younger patients had higher prevalence of NAFLD and lower prevalence of advanced fibrosis with p = 0.026 for the association between age and NAFLD (Figure 3).

Figure 3.

Prevalence of NAFLD and advanced fibrosis in different age groups.

DISCUSSION

Main findings

This is the first prospective study that assessed the feasibility of screening for both NAFLD and advanced fibrosis in type 2 diabetic patients in a primary care setting by using two accurate, precise, and validated non-invasive image-based biomarkers: MRI-PDFF and MRE-stiffness. The study cohort included a diverse population of patients with T2DM that has been managed and followed by primary care providers, and was conducted in the population likely to benefit from such screening program. All patients underwent MRI to estimate hepatic PDFF (100%) and 98% had MRE to estimate hepatic stiffness. We found a 65% prevalence of NAFLD and 7.1% prevalence of advanced fibrosis. One patient (1%) was noted to have definite HCC.

In context with published literature

Previous studies that aimed to screen for NAFLD in T2DM were mainly based on ultrasound imaging.^{5, 6} Targher and colleagues⁵ conducted a cross-sectional study enrolling 2839 diabetic patients in Italy and found a 71.1% age-adjusted prevalence of NAFLD in men and 68% in women based on ultrasound evaluation. Similarly, 939 randomly selected participants in the Edinburgh Type 2 Diabetes Study (ET2DS) were screened for NAFLD using ultrasound, but the prevalence of NAFLD was only 42.6%.⁶ The cohort of ET2DS was further screened for advanced fibrosis by using elevated hyaluronic acid level,³³ a test with high negative predictive value (96%), but moderate positive predictive value (51%).³⁴ Based on this test, 5.7% had evidence of hepatic fibrosis.³³ Two patients (0.2%) in that cohort were diagnosed with HCC.

Only small population samples of diabetics have been assessed with liver biopsies.^{2, 9, 35, 36} Williams and colleagues² analyzed 54 patients with T2DM and ultrasound-based diagnosis of NAFLD who underwent liver biopsies. NAFLD was confirmed in 74% and 22% had NASH in comparison to 11% in nondiabetics (p = 0.03). Furthermore, studies from Brazil and India have shown even higher histologic prevalence of NASH in diabetics ranging between 63 and 78%.^{9, 36} However, these studies also initially screened diabetics with ultrasound, and biopsies were performed only in those patients with positive imaging findings. Two studies have used MRS to assess NAFLD prevalence:^{7, 8} Petit and colleagues⁸ found NAFLD prevalence of 60.3% in 101 outpatient diabetics based on MRS > 5.5%. The other recent study of Portillo Sanchez and colleagues⁷ enrolled 103 mainly obese male diabetics with normal aminotransferases. The NAFLD prevalence was 76% based on MRS > 5.5% while NASH was noted in 56% based on liver biopsies of 37 patients. The prevalence of NAFLD in our study was based on an accurate and reproducible image-based biomarker, MRI-PDFF, and is in accordance with the previously reported prevalences.

In our cohort of diabetic patients, we found a 7.1% prevalence of advanced fibrosis using the precise method of MRE and based on a recently determined threshold.²¹ This finding emphasizes the potential importance of initiating concomitant NAFLD and advanced fibrosis screening in diabetics early, perhaps at the time of diagnosis of diabetes, in order to prevent further damage. Previous studies have demonstrated more aggressive course of NAFLD in diabetes and higher risk of cirrhosis in patients with T2DM and NAFLD.^{37, 38} The risk of advanced liver disease was present even when newly diagnosed diabetes patients were evaluated in a large population-based study.³⁹ Moreover, patients with T2DM and NAFLD were noted to have a 2.2-fold increased risk of overall mortality compared to those without NAFLD, which was partially attributed to liver-related death.⁴⁰ Population-based studies and meta-analysis also indicate that T2DM is associated with 2–3-fold increase in the risk of HCC^{41, 42} and HCC-related death (HR 3.38, CI 2.35–4.86)⁴³, which further strengthens the importance for surveillance in this population. We also identified one patient with HCC although our study was not designed for HCC screening.

Two patients with advanced fibrosis in our study did not meet criteria for NAFLD. This finding is in accordance with recent meta-analysis that showed fibrosis progression over time in patients with NAFLD.⁴⁴ Moreover, it again confirms the previously shown observation that liver fat declines in patients with advanced fibrosis (burnt-out NASH)⁴⁵ and can be missed if diabetics are screened for NAFLD only.

Cost and limited availability of MRI-PDFF and MRE remain a concern. Furthermore, several serum biomarkers as well as ultrasound-based imaging biomarkers such as transient elastography (TE) have been studied in the non-invasive assessment of steatosis and fibrosis in patients with NAFLD.^46–49 Although the available clinical prediction rules and TE are less expensive and more widely available than MRI, these tests have several limitations, especially in patients with NAFLD.⁴⁶ TE is a reproducible, easy to perform, non-invasive method measuring both fibrosis and steatosis. However, it carries several disadvantages. Controlled Attenuation Parameter (CAP) for detection of steatosis can be used only with the conventional M probe, but not with the XL probe that has been designed to be used in overweight and obese patients. A large prospective study showed an overall CAP failure rate of 7.7%, with failure of 19.4% and 58.4% in those with BMI ≥ 30 kg/m² and >40 kg/m², respectively.⁵⁰ CAP failure was independently associated with female gender, age >55 years, BMI ≥ 30 kg/m2, and metabolic syndrome. Similarly, liver stiffness measurement by TE fails to determine hepatic fibrosis in 3–10%, while unreliable examinations range between 15 and 26% when using the M probe.^{51, 52} A large study that evaluated performance of XL probe in biopsy-proven NAFLD found an overall reliable measurement by XL probe of 75% and of only 65% in patients with BMI ≥ 30 kg/m^2.52 Moreover, when compared to histology, the results by XL probe showed discordance of at least two stages in 9% with single risk factor for discordance being BMI ≥ 35 kg/m². All these disadvantages make TE a suboptimal technique for routine screening of steatosis and advanced fibrosis in diabetics as obesity is very common in patients with NAFLD.

MRE is thought to have higher accuracy, sensitivity and specificity for the diagnosis of advanced fibrosis than TE.^{21, 22, 53} Furthermore, MRE is not limited by obesity or presence of cirrhosis while TE has a high failure rate in these patients.⁵⁴ In future, we expect that it may be feasible to stratify patients based upon their BMI and utilize MRE in those who are obese versus performing ultrasound-based tests such as TE or acoustic radiation force impulse (ARFI) in those who are not obese. Further studies are needed to address these issues.

Several clinical models for prediction of steatosis, steatohepatitis, and fibrosis have been studied and proposed in NAFLD. However, although they are fast, inexpensive, and widely available, these tests lack accuracy and are unable to appropriately classify patients into distinct categories.⁴⁷ A large study evaluated the utility of 5 clinical biomarkers for prediction of steatosis in biopsy-proven NAFLD and noted that the biomarkers were able to detect the presence of steatosis, but unable to accurately quantify hepatic steatosis and more importantly, fibrosis and inflammation were significant confounders.⁴⁸ Cui et al recently demonstrated in a prospective study that two-dimensional MRE had significantly higher accuracy for diagnosis of advanced fibrosis than 8 commonly used prediction rules.⁵⁵ These data suggest that the best approach in screening diabetics for NAFLD would be a sequential algorithm using both clinical models and imaging modalities.⁵⁶ Future studies should evaluate the best and most cost-effective modalities for screening of advanced fibrosis in diabetics with NAFLD.

Recent analyses of prevalence and incidence of diabetes in the U.S. population showed doubling of both parameters between 1990 and 2008 and plateauing during 2008–2012.⁵⁷ However, currently 29.1 million Americans - 9.3 % of the U.S. population - have diabetes.⁵⁸ Of those, 90–95% are type 2 diabetics. Extrapolating the prevalences observed in our study (59% with NAFLD only, 5 % with NAFLD and advanced fibrosis, and 2% with advanced fibrosis, but no NAFLD) in asymptomatic type 2 diabetics to the entire population of diabetics, we estimate that there are about 15.5 million diabetic persons with NAFLD in the U.S.; approximately 1.3 million may have NAFLD and advanced fibrosis, and more than half million may have advanced fibrosis alone and could be missed if only NAFLD screening were performed (Figure 4). Compounding the problem, patients with NAFLD, and especially those with metabolic disturbances, have low awareness of NAFLD and its complications and management.⁵⁹ Furthermore, a recent study assessed NAFLD management in primary care settings and noted that the magnitude and proportion of ALT elevation was the only predictor of NAFLD care and only 3% of those with high risk of fibrosis were referred to a specialist.⁶⁰ All these findings imply that prompt changes are needed in surveillance strategy, management, and education of both patients and primary care providers taking care of patients at high risk of NAFLD, especially those with T2DM. We found that concomitant use of MRI to estimate PDFF and MRE to estimate stiffness are feasible for NAFLD and advanced fibrosis screening and can be used to identify patients at risk.

Figure 4.

Estimated population with NAFLD and advanced fibrosis in type 2 diabetes based on diabetes prevalence in 2012^* and results of our study^#

Strengths and limitations

The key strengths of our study are that this was a prospective study, which included asymptomatic patients in different clinical stages of T2DM. Patients were referred for the study by their primary care providers and constitute a representative sample of diabetics. Advanced MR imaging techniques were used for concomitant screening of NAFLD (MRI-PDFF) and advanced fibrosis (MRE-stiffness). Both imaging techniques are safe, accurate and reproducible and have been validated to correlate well with histologic changes in liver fat and fibrosis.^{20, 21} However, we would like to acknowledge some limitations of our study. This is a single-center study conducted at a specialized center for both clinical and MRI research in NAFLD. Although validated, MRI-PDFF and MRE-stiffness have not been approved for screening of NAFLD and advanced fibrosis and currently are available mainly in specialized tertiary centers.

Implications for future research and patient care

Patients with T2DM have a high prevalence of NAFLD and advanced fibrosis. This frequently remains unrecognized and overlooked in the primary care setting. This is the first proof-of-concept study that used advanced MR methods to accurately screen for both NAFLD and advanced fibrosis. Future studies should include larger populations of diabetic patients and utilize different non-invasive clinical prediction models and imaging techniques, (including advanced and emerging ultrasound-based methods)^{53, 61} to better assess the most accurate and cost-effective modality, the best time to start screening, and the optimal screening algorithm in this population in order to prevent advanced liver damage and reduce liver-related complications and mortality in diabetics.

Abbreviations

Adipo-IR

adipose tissue insulin resistance

ALT

alanine aminotransferase

AST

aspartate aminotransferase

AUDIT

Alcohol Use Disorders Identification Test

BMI

body mass index

CT

computed tomography

ET2DS

Edinburgh Type 2 Diabetes Study

GGT

gamma-glutamyl transpeptidase

HDL-C

high-density lipoprotein cholesterol

HCC

hepatocellular carcinoma

HOMA-IR

Homeostatic Model Assessment of Insulin Resistance

LDL-C

low-density lipoprotein cholesterol

MRE

magnetic resonance elastography

MRI-PDFF

magnetic resonance imaging-estimated proton density fat fraction

MRS

magnetic resonance spectroscopy

NAFLD

nonalcoholic fatty liver disease

NASH

nonalcoholic steatohepatitis

TE

transient elastography

T2DM

type 2 diabetes mellitus

UCSD

University of California, San Diego