Table 1.
Observational studies published from 2016 to 2020 that assessed the association between FLD (as detected by imaging or biopsy) and the risk of incident type 2 diabetes
| Author, year | Study characteristics; follow‐up length | Diagnosis of FLD | Diagnosis of diabetes | Covariate adjustment | Main findings |
|---|---|---|---|---|---|
| Chen, 2016 22 | Prospective cohort study of 6,542 (3.2% with FLD) Chinese nondiabetic subjects without known chronic liver diseases; 6 years | Ultrasound | Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or drug treatment |
Age, obesity, hypertriglyceridaemia, IFG |
FLD was independently associated with incident diabetes (adjusted HR 2.17, 95% CI 1.6‐3.0) |
| Li, 2017 23 | Prospective cohort study of 18,111 (31.9% with FLD) Chinese nondiabetic subjects without known chronic liver diseases; 4.6 years | Ultrasound | Fasting glucose ≥ 7.0 mmol/L, clinical history or drug treatment |
Age, sex, BMI, waist circumference, alcohol intake, smoking, exercise, family history of diabetes, fasting glucose, triglycerides, total cholesterol |
The adjusted HRs for incident diabetes compared with those without FLD were as follows: 1.88 (95% CI 1.6‐2.2) in the mild FLD group and 2.34 (95% CI 1.9‐3.0) in the moderate‐severe FLD group (P‐trend < 0.001) |
| Ma, 2017 24 | Prospective cohort study of 1,051 (17.8% with FLD) United States nondiabetic subjects without known chronic liver diseases; 6.2 years | Computed tomography |
Fasting glucose ≥ 7.0 mmol/L or drug treatment |
Age, sex, BMI, smoking, exercise, alcohol intake, fasting glucose, changes in BMI and liver phantom ratio during follow‐up |
FLD was independently associated with incident diabetes (adjusted OR 2.85, 95% CI 1.4‐6.0, P = .006) |
| Chen, 2017 25 | Prospective cohort study of 132,377 (32% with FLD, 18.1% with chronic liver diseases) Taiwanese nondiabetic subjects; 5.8 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, clinical history or drug treatment |
Age, obesity, hypertension, dyslipidemia, family history of diabetes, smoking, alcohol intake, exercise, AST, ALT, GGT, ALP | FLD was independently associated with incident diabetes (adjusted HR 2.08, 95% CI 1.9‐2.2, P < .001 in men and adjusted HR 2.65, 95% CI 2.4‐2.9, P < .001 in women). Exclusion of chronic liver diseases did not attenuate the association |
| Liu, 2017 26 | Retrospective cohort study of 18,507 (18.8% with FLD) Chinese elderly nondiabetic males without known chronic liver diseases; 5 years | Ultrasound | Fasting glucose ≥ 7.0 mmol/L, 2‐h plasma glucose ≥ 11.1 mmol/L during 75‐g OGTT, clinical history or drug treatment |
Age, BMI, ALT, smoking, marriage status, alcohol intake, hypertension, dyslipidemia |
FLD was independently associated with incident diabetes (adjusted RR 1.67, 95% CI 1.4‐2.1, P < .001) |
| Björkström, 2017 27 | Retrospective cohort study of 396 (100% with FLD) Swedish nondiabetic subjects without known chronic liver diseases; 18.4 years | Biopsy | Clinical history or drug treatment | Age, sex, BMI, triglycerides | Liver fat content was independently associated with incident diabetes in the fibrosis stages 0‐2 (adjusted HR 1.36, 95% CI 1.0‐1.8; P = .03), not in the fibrosis stages 3‐4 (adjusted HR 1.24, 95% CI 0.4‐3.7, P = .71) |
| Tokita, 2017 28 | Retrospective cohort study of 2,408 (11.2% with FLD) Japanese nondiabetic subjects without known chronic liver diseases; 10 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or drug treatment |
Age, sex, HbA1c, HDL, triglycerides, systolic blood pressure | FLD was independently associated with incident diabetes (P = .0001) |
| Mitsuhashi, 2017 29 | Retrospective cohort study of 17,810 (21.6% with FLD) Japanese nondiabetic subjects without known chronic liver diseases; 5.1 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, clinical history or drug treatment |
Age, BMI, smoking, exercise, alcohol intake, family history of diabetes, log ALT, fasting glucose | The adjusted HR for incident diabetes compared with those without FLD and MetS were as follows: 2.35 (95% CI 1.9‐2.9, P < .001) in the non‐MetS with FLD group, 1.70 (95% CI 1.3‐2.2, P < .001) in the MetS without FLD group, and 2.33 (95% CI 1.9‐2.9, P < .001) in the MetS with FLD group |
| Bae, 2018 30 | Retrospective cohort study of 7,849 (46.7% with FLD) Korean nondiabetic subjects without known chronic liver diseases; 4 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, clinical history or drug treatment |
Age, sex, fasting glucose, HbA1c, BMI, LDL, HDL, triglycerides, systolic blood pressure, HOMA‐IR, smoking | Changes in FLD status were independently associated with incident diabetes. The adjusted HRs compared with those without FLD were as follows: 1.50 (95% CI 1.1‐2.0) in the persistent FLD group and 0.99 (95% CI 0.8‐1.3) in the resolved FLD group |
| Seko, 2018 31 | Retrospective cohort study of 89 (100% with FLD) Japanese nondiabetic subjects (58% with IGT) without known chronic liver diseases; 5.2 years | Biopsy | Fasting glucose ≥ 7.0 mmol/L, 2‐h plasma glucose ≥ 11.1 mmol/L during 75‐g OGTT, HbA1c ≥ 6.5% or drug treatment | Age, sex, BMI, ferritin, fibrosis stage, NAS, AST/ALT ratio, fasting glucose, 30‐min and 1‐h postload plasma glucose, HbA1c, 1‐h postload insulin, HOMA‐B, HOMA‐IR | Insulin resistance was independently associated with incident diabetes (adjusted HR 40.1, 95% CI 1.4‐119.3, P = .033) |
| Kim, 2018 32 | Retrospective cohort study of 2,920 (31.6% with FLD, 3.5% with diabetes) Korean subjects without known chronic liver diseases; 5.1 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or drug treatment |
Age, sex, waist circumference, triglycerides, HDL, LDL, uric acid, smoking | FLD was independently associated with incident diabetes. The adjusted HRs compared with the nonobese without FLD group were as follows: 2.69 (95% CI, 1.7‐4.2, P < .001) in the nonobese with FLD group, 1.92 (95% CI, 1.1‐3.4, P = .022) in the obese without FLD group, and 2.89 (95% CI, 1.7‐4.8, P < .001) in the obese with FLD group |
| Shen, 2018 33 | Prospective cohort study of 41,650 (28.4% with FLD) Chinese nondiabetic subjects without known chronic liver diseases; 3.6 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L or drug treatment |
Age, sex, smoking, exercise, education, family incomes, family history of diabetes, waist circumference, ALT, LDL, HDL, triglycerides, fasting glucose, uric acid, C‐reactive protein, hypertension, metabolic syndrome | The severity of FLD was associated with higher risk of incident diabetes. The adjusted HRs compared with those without FLD were as follows: 1.62 (95% CI 1.5‐1.8) in the whole FLD group, 1.46 (95% CI 1.3‐1.6) in the mild FLD group, 1.92 (95% CI 1.7‐2.2) in the moderate FLD group and 2.66 (95% CI 2.2‐3.3) in the severe FLD group (P‐trend < 0.001). Similar associations were observed between FLD and incident IFG |
| Brunner, 2019 34 | Retrospective cohort study of 808 (14% with FLD, 2.5% with diabetes) United States subjects; 6.2 years | Computed tomography | Fasting glucose ≥ 7.0 mmol/L or drug treatment | Age, sex, BMI, smoking, alcohol intake, liver phantom ratio, fasting glucose, changes in BMI during follow‐up | Increasing liver fat content during follow‐up was independently associated with incident diabetes (adjusted OR 1.68, 95% CI 1.2‐2.3, P = .001) |
| Okamura, 2019 35 | Retrospective cohort study of 15,464 (17.7% with FLD) Japanese nondiabetic subjects without known chronic liver diseases; 5.1 years | Ultrasound | Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or clinical history | Age, alcohol intake, smoking, exercise, fasting glucose | FLD was independently associated with incident diabetes (adjusted HR 4.74, 95% CI 1.9‐11.7, P = .006 in men and adjusted HR 14.0, 95% CI 7.2‐27.1, P < .001 in women). The clustering of obesity, visceral obesity and FLD markedly increased the risk of developing diabetes (adjusted HR 10.5, 95% CI 8.0‐13.8, P < .001 in men and adjusted HR 30.0, 95% CI 18.0‐50.0, P < .001 in women) |
| Cho, 2019 36 | Retrospective cohort study of 2,726 (30.3% with FLD) Korean nondiabetic subjects without known chronic liver diseases; 62.2 months | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or drug treatment |
Age, sex, BMI, fasting glucose, ALT | Changes in FLD status were independently associated with incident diabetes. The adjusted HRs compared with those without FLD were as follows: HRs 3.59 (95% CI 2.1‐6.3, P < .001) in the persistent FLD group, 1.94 (95% CI 1.1‐3.5, P = .026) in the incident FLD group and 1.21, 95% CI, 0.4‐3.6, P = .733) in the resolved FLD group |
| Sung, 2019 37 | Retrospective cohort study of 70,303 (13.1% with FLD) Korean nondiabetic subjects without known chronic liver diseases; 3.7 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5%, clinical history or drug treatment |
Age, education, exercise, smoking, alcohol intake, centre, year, family history of diabetes, waist circumference, BMI, triglycerides, LDL, drugs for hypertension and hyperlipidaemia | FLD was independently associated with incident diabetes (adjusted HR 2.17, 95% CI 1.6‐3.0 in men and adjusted HR 2.86, 95% CI 1.5‐5.5 for women) |
| Wang, 2019 38 | Retrospective cohort study of 10,064 (5.4% with FLD) Japanese nondiabetic subjects without known chronic liver diseases; 6 years | Ultrasound |
Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or clinical history |
Age, sex, BMI, alcohol intake, smoking, HbA1c | FLD was independently associated with incident diabetes (adjusted HR 2.52, 95% CI 1.6‐4.0, P < .001) |
| Lee, 2019 39 | Retrospective cohort study of 6,240 (45.4% with FLD) Korean prediabetic subjects without known chronic liver disease; 4.3 years | Ultrasound | Fasting glucose ≥ 7.0 mmol/L, HbA1c ≥ 6.5% or drug treatment | Age, sex, BMI, smoking, alcohol intake, ALT, triglycerides, HDL, systolic blood pressure, HbA1c | FLD was independently associated with incident diabetes (adjusted RR 1.81, 95% CI, 1.5‐2.2, P < .001) |
| Nasr, 2020 40 | Prospective cohort study of 106 (100% with FLD) Swedish nondiabetic subjects without known chronic liver diseases; 23.2 years | Biopsy | Fasting glucose ≥ 7.0 mmol/L with drug treatment or 2‐h plasma glucose ≥ 11.1 mmol/L during 75‐g OGTT | Age, BMI, histologic fibrosis stage | Liver fat content was independently associated with incident diabetes (adjusted HR 1.03 per 1% increase, 95% CI 1.0‐1.1, P = .02) |
Abbreviations: ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CI, confidence interval; FLD, fatty liver disease; GGT, gamma‐glutamyl transferase; HbA1c, haemoglobin A1c; HDL, high‐density lipoprotein; HOMA‐B, homeostasis model assessment of beta‐cell function; HOMA‐IR, homeostasis model assessment of insulin resistance; HR, hazard ratio; IFG, impaired fasting glycaemia; IGT, impaired glucose tolerance; LDL, low‐density lipoprotein; MetS, metabolic syndrome; NAS, NAFLD activity score; OGTT, 75‐g oral glucose tolerance test; T2D, type 2 diabetes.