Abstract:

We aimed to compare nonalcoholic fatty liver disease (NAFLD) and metabolic dysfunction-associated steatotic liver disease (MASLD) definitions concerning diabetes prediction in a large sample of Brazilian adults. As a secondary objective, we compared associations between NAFLD/MASLD and diabetes across self-declared race/skin color groups. The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) is a prospective cohort study of Brazilian civil servants (35-74 years) enrolled from 2008 to 2010 and followed up from 2012-2014 and 2017-2019. We ascertained type 2 diabetes mellitus at baseline as well as follow-up visits based on self-reported diagnosis, medication use, and glycemic tests (fasting and 2h post-OGTT glucose and HbA1c). We excluded individuals with heavy alcohol consumption or self-reported cirrhosis/hepatitis. We analyzed 7,073 subjects. NAFLD was defined by ultrasound-based steatosis. Participants with steatosis and at least one cardiometabolic factor were considered as having MASLD. Cox proportional hazards models were performed to evaluate the association between NAFLD/MASLD and the incidence of type 2 diabetes mellitus. At baseline, 33.9% of individuals presented NAFLD and 32.5% presented MASLD. Over 9.4 years of follow-up, the relative increase in the incidence of diabetes was 78% for NAFLD (HR = 1.78; 95%CI: 1.58-2.01) and 88% for MASLD (HR = 1.88; 95%CI: 1.67-2.12). Associations did not differ significantly among race/skin color groups (p for interaction = 0.10 for MASLD and 0.08 for NAFLD). In this large cohort of middle-aged and older Brazilian adults, the relative incidence of diabetes was similar for NAFLD and MASLD definitions, with similar associations in all ethnic groups.

Keywords:
Non-Alcoholic Fatty Liver Disease; Type 2 Diabetes Mellitus; Ethnicity

Resumo:

Este estudo comparou as definições de doença hepática gordurosa não alcoólica (DHGNA) e doença hepática esteatótica associada à disfunção metabólica (DHEADM) em relação à predição de diabetes em uma grande população adulta de brasileiros. Como objetivo secundário, comparamos as associações entre DHGNA/DHEADM e diabetes em grupos de diferentes etnias/cor da pele autodeclaradas. O Estudo Longitudinal de Saúde do Adulto (ELSA-Brasil) é um estudo de coorte prospectivo de servidores públicos brasileiros (35-74 anos) recrutados de 2008 a 2010 e acompanhados nos períodos de 2012-2014 e 2017-2019. Determinamos o diabetes mellitus tipo 2 nas visitas iniciais e de acompanhamento com base em diagnóstico autorreferido, uso de medicação e testes glicêmicos (glicemia em jejum e 2 horas após o teste de tolerância à glicose oral e HbA1c). Foram excluídos indivíduos com consumo pesado de álcool ou cirrose/hepatite autorreferida. Foram analisados 7.073 indivíduos. A DHGNA foi definida por esteatose através de ultrassom. Participantes com esteatose e pelo menos um fator cardiometabólico foram considerados como portadores de DHEADM. Modelos de riscos proporcionais de Cox foram aplicados para avaliar a associação entre DHGNA/DHEADM e a incidência de diabetes mellitus tipo 2. No início do estudo, 33,9% apresentavam DHGNA e 32,5% DHEADM. Ao longo de 9,4 anos de acompanhamento, o aumento relativo na incidência de diabetes foi de 78% para indivíduos com DHGNA (HR = 1,78; IC95%: 1,58-2,01) e 88% para indivíduos com DHEADM (HR = 1,88; IC95%: 1,67-2,12). As associações não diferiram significativamente entre os grupos de etnia/cor de pele (p para interação = 0,10 para DHEADM e 0,08 para DHGNA). Nesta grande coorte de brasileiros de meia-idade e idosos, a incidência relativa de diabetes foi semelhante nas definições de DHGNA e DHEADM, com associações semelhantes entre grupos étnicos.

Palavras-chave:
Hepatopatia Gordurosa não Alcoólica; Diabetes Mellitus Tipo 2; Etnicidade

Resumen:

Este estudio comparó las definiciones de enfermedad del hígado graso no alcohólico (EHGNA) y enfermedad hepática esteatótica asociada con disfunción metabólica (EHDM) con relación a la predicción de diabetes en una gran población adulta de brasileños. Como objetivo secundario, comparamos las asociaciones entre EHGNA/MASLD y diabetes en grupos de diferentes etnias/color de piel autodeclarados. El Estudio Longitudinal de Salud del Adulto (ELSA-Brasil) es un estudio de cohorte prospectivo de servidores públicos brasileños (35-74 años) reclutados entre el 2008 y el 2010 y seguidos en los períodos 2012-2014 y 2017-2019. Determinamos la diabetes mellitus tipo 2 en las visitas iniciales y de seguimiento con base en el diagnóstico autoinformado, el uso de medicación y las pruebas de glucemia (glucemia en ayunas y prueba de tolerancia a la glucosa oral 2 horas después y HbA1c). Se excluyeron las personas con consumo excesivo de alcohol o cirrosis/hepatitis autoinformadas. Se analizaron 7.073 individuos. La EHGNA se definió por esteatosis mediante ecografía. Se consideró que los participantes con esteatosis y al menos un factor cardiometabólico tenían EHDM. Se aplicaron modelos de riesgos proporcionales de Cox para evaluar la asociación entre EHGNA/EHDM y la incidencia de diabetes mellitus tipo 2. Al inicio del estudio, el 33,9% tenía EHGNA y el 32,5% EHDM. Durante 9,4 años de seguimiento, el aumento relativo en la incidencia de diabetes fue del 78% para las personas con EHGNA (HR = 1,78; IC95%: 1,58-2,01) y del 88% para las personas con EHDM (HR = 1,88; IC95%: 1,67-2,12). Las asociaciones no difirieron significativamente entre los grupos étnicos/color de piel (p para interacción = 0,10 para EHDM y 0,08 para EHGNA). En esta gran cohorte de brasileños de mediana edad y ancianos, la incidencia relativa de diabetes fue similar en las definiciones de EHGNA y EHDM, con asociaciones similares entre grupos étnicos.

Palabras-clave:
Enfermedad del Hígado Graso no Alcohólico; Diabetes Mellitus Tipo 2; Etnicidad

Introduction

The burden of type 2 diabetes mellitus and nonalcoholic fatty liver disease (NAFLD) has increased in the last decades. From 1990 to 2019, the global incidence of NAFLD increased by 95%, and deaths and disability-adjusted life years (DALYs) attributable to NAFLD increased by 80% and 63%, respectively ¹1. Tian H, Zhang K, Hui Z, Ren F, Ma Y, Han F, et al. Global burden of non-alcoholic fatty liver disease in 204 countries and territories from 1990 to 2019. Clin Res Hepatol Gastroenterol 2023; 47:102068.. In the same period, new cases of type 2 diabetes mellitus increased by 78%, type 2 diabetes mellitus-related deaths by 68%, and DALYs by 80% worldwide ²2. Institute for Health Metrics and Evaluation. GBD compare data visualization. https://vizhub.healthdata.org/gbd-compare/ (accessed on 15/Jan/2024).
https://vizhub.healthdata.org/gbd-compar... .

Positive associations between NAFLD and incident type 2 diabetes have been well documented ³3. Mantovani A, Petracca G, Beatrice G, Tilg H, Byrne CD, Targher G. Non-alcoholic fatty liver disease and risk of incident diabetes mellitus: an updated meta-analysis of 501 022 adult individuals. Gut 2021; 70:962-9., and their causal nature was demonstrated by Mendelian randomization ⁴4. De Silva NMG, Borges MC, Hingorani AD, Engmann J, Shah T, Zhang X, et al. Liver function and risk of type 2 diabetes: bidirectional Mendelian randomization study. Diabetes 2019; 68:1681-91.. Consistent with these findings, we also found a higher risk of diabetes associated with NAFLD ⁵5. Faria LC, Diniz MFHS, Giatti L, Schmidt MI, Goulart AC, Duncan BB, et al. Liver steatosis as a predictor of incident diabetes in adults: a prospective evaluation in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Cad Saúde Pública 2023; 39:e00090522. in Brazilian adults based on an early follow-up examination of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil).

In 2023, a new definition and criteria for steatotic liver disease emerged to replace the nomenclature used in previous definitions. Proposed by a consensus of specialists, “metabolic dysfunction-associated steatotic liver disease” (MASLD) ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73. excludes other causes of steatosis but requires the presence of metabolic dysfunction. To our knowledge, the only study contrasting MASLD with the previous NAFLD definition regarding diabetes prediction was done in a Chinese sample ⁷7. He L, Zheng W, Qiu K, Kong W, Zeng T. Changing from NAFLD to MASLD: the new definition can more accurately identify individuals at higher risk for diabetes. J Hepatol 2023; 80:e85-7..

Ethnic disparities related to NAFLD occurrence and complications have been reported, with a lower prevalence of NAFLD being found in black individuals ⁸8. Riazi K, Swain MG, Congly SE, Kaplan GG, Shaheen AA. Race and ethnicity in non-alcoholic fatty liver disease (NAFLD): a narrative review. Nutrients 2022; 14:4556.^,99. Rich NE, Oji S, Mufti AR, Browning JD, Parikh ND, Odewole M, et al. Racial and ethnic disparities in nonalcoholic fatty liver disease prevalence, severity, and outcomes in the United States: a systematic review and meta-analysis. Clin Gastroenterol Hepatol 2018; 16:198-210.e2.. A North American cohort ¹⁰10. Hatano Y, VanWagner LB, Carnethon MR, Bancks MP, Carson AP, Lloyd-Jones DM, et al. Racial difference in the association between non-alcoholic fatty liver disease and incident type 2 diabetes: findings from the CARDIA study. Diabetologia 2023; 66:1235-46. showed an increased absolute and relative risk of diabetes related to NAFLD in white but not in black participants, which is surprising considering the higher burden of diabetes experienced by black individuals ¹¹11. Haw JS, Shah M, Turbow S, Egeolu M, Umpierrez G. Diabetes complications in racial and ethnic minority populations in the USA. Curr Diab Rep 2021; 21:2.^,1212. Cheng YJ, Kanaya AM, Araneta MRG, Saydah SH, Kahn HS, Gregg EW, et al. Prevalence of diabetes by race and ethnicity in the United States, 2011-2016. JAMA 2019; 322:2389-98..

To provide additional information regarding these issues, our objectives were to: (1) reassess the association of NAFLD with the incidence of type 2 diabetes mellitus and contrast it with the association found with the new MASLD definition; and (2) compare associations between NAFLD/MASLD and diabetes among self-declared Brazilian ethnic groups.

Methods

Study population and design

The ELSA-Brasil is an ongoing prospective multicenter cohort study of civil servants aimed at investigating risk factors for the development of chronic diseases such as diabetes ¹³13. Aquino EML, Barreto SM, Bensenor IM, Carvalho MS, Chor D, Duncan BB, et al. Brazilian Longitudinal Study of Adult Health (ELSA-Brasil): objectives and design. Am J Epidemiol 2012; 175:315-24.. During the baseline visit, the study enrolled 15,105 participants aged 35-74 years at public institutions in six capitals of Brazilian states (Bahia, Espírito Santo, Minas Gerais, Rio de Janeiro, Rio Grande do Sul, and São Paulo) from 2008 to 2010 ¹⁴14. Schmidt MI, Duncan BB, Mill JG, Lotufo PA, Chor D, Barreto SM, et al. Cohort profile: Longitudinal Study of Adult Health (ELSA-Brasil). Int J Epidemiol 2015; 44:68-75.. Two follow-up visits occurred between 2012-2014 and 2017-2019. Ethics committees of each institution involved in the study approved the research protocol, and participants signed informed consent forms agreeing to participate in each visit ¹⁵15. Aquino EML, Vasconcellos-Silva PR, Coeli CM, Araujo MJ, Santos SM, Figueiredo RCD, et al. Aspectos éticos em estudos longitudinais: o caso do ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:19-26..

Participants underwent standardized questionnaires, abdominal ultrasound, and blood sample collection after an overnight fasting ¹⁶16. Mill JG, Pinto K, Griep RH, Goulart A, Foppa M, Lotufo PA, et al. Aferições e exames clínicos realizados nos participantes do ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:54-62. with strict quality control ¹⁷17. Schmidt MI, Griep RH, Passos VM, Luft VC, Goulart AC, Menezes GMDS, et al. Estratégias e desenvolvimento de garantia e controle de qualidade no ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:105-12.. Blood samples were analyzed in a centralized laboratory ¹⁸18. Fedeli LG, Vidigal PG, Leite CM, Castilhos CD, Pimentel RA, Maniero VC, et al. Logística de coleta e transporte de material biológico e organização do laboratório central no ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:63-71.. In addition to clinical visits, participants responded to annual telephone questions, including whether and when a new medical diagnosis of diabetes was made since the last visit ¹⁹19. Barreto SM, Ladeira RM, Bastos MSCBO, Diniz MFHS, Jesus EA, Kelles SMB, et al. Estratégias de identificação, investigação e classificação de desfechos incidentes no ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:79-86..

In this analysis, 2,429 (16%) participants with prevalent diabetes were excluded, as well as 599 individuals without data on baseline covariates, 2,295 subjects who did not undergo baseline ultrasonography, and 1,292 participants who did not have information on ultrasonography quality or had a ultrasonography quality classified as “poor” or “unacceptable” according to liver ultrasound parameters used for the diagnosis of steatosis (see ahead). To follow NAFLD ²⁰20. European Association for the Study of the Liver (EASL), European Association for the Study of Diabetes (EASD), European Association for the Study of Obesity (EASO). EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease. J Hepatol 2016; 64:1388-402. and MASLD ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73. definitions, 464 individuals with heavy alcohol intake at baseline (> 140g of alcohol per week for females and > 210g of alcohol per week for males) were excluded in addition to other 700 who reported a medical diagnosis of cirrhosis or hepatitis at baseline. We also excluded 253 participants without information on diabetes incidence due to loss on follow-up. Finally, data from 7,073 participants were analyzed (Figure 1).

Assessment of diabetes

Prevalent diabetes was defined by (1) self-report (answering “Yes” to the question “Have you been previously told by a physician that you had/have diabetes?”; (2) medication use (answering “Yes” to the question “Have you used medication for diabetes or high blood sugar in the past two weeks?”; or (3) abnormal glycemic tests: fasting plasma glucose (FPG; ≥ 126mg/dL, ≥ 7mmol/L), 2-hour oral glucose tolerance test (2h-75g OGTT; 200mg/dL, ≥ 11.1mmol/L), and glycated hemoglobin (HbA1c; ≥ 6.5%, ≥ 48mmol/mol) ²¹21. American Diabetes Association Professional Practice Committee. 2. Diagnosis and classification of diabetes: standards of care in diabetes - 2024. Diabetes Care 2024; 47 Suppl 1:S20-42.^,2222. World Health Organization. Global report on diabetes. https://apps.who.int/iris/handle/10665/204871 (accessed on 22/Jan/2023).
https://apps.who.int/iris/handle/10665/2... .

Also new cases of diabetes developed during follow-up were identified among those free of the disease at baseline, using information gathered during follow-up visits (questionnaires and laboratory tests) or the annual telephone surveillance. The time until the development of diabetes or censure was estimated as previously reported ²³23. Schmidt MI, Bracco PA, Yudkin JS, Bensenor IM, Griep RH, Barreto SM, et al. Intermediate hyperglycaemia to predict progression to type 2 diabetes (ELSA-Brasil): an occupational cohort study in Brazil. Lancet Diabetes Endocrinol 2019; 7:267-77..

Assessment of NAFLD and MASLD

Steatosis by liver ultrasonography was assessed only at baseline ¹⁶16. Mill JG, Pinto K, Griep RH, Goulart A, Foppa M, Lotufo PA, et al. Aferições e exames clínicos realizados nos participantes do ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:54-62.^,2424. Goulart AC, Oliveira IRS, Alencar AP, Santos MSC, Santos IS, Martines BMR, et al. Diagnostic accuracy of a noninvasive hepatic ultrasound score for non-alcoholic fatty liver disease (NAFLD) in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Sao Paulo Med J 2015; 133:115-24. using the same equipment model in all research centers: a high-resolution B-mode scanner (SSA-790A, Aplio XG, Toshiba Medical Systems; https://www.global.toshiba) with a convex array transducer (model PVT-375BT) adjusted with 3.5 MHz central frequency and fundamental frequency ranging 1.9-5.0 MHz. The exam was carried out by board-certified radiologists or trained technicians following a standardized protocol. The parameters evaluated were hepatic beam attenuation, anteroposterior diameter of the right hepatic lobe, and hepatorenal index. A centralized reading center in São Paulo interpreted the exams, and a senior radiologist evaluated their quality.

Steatosis was defined based on hepatic beam attenuation, our study’s most accurate measure for hepatic steatosis when compared with high-resolution computerized tomography in participants with NAFLD ²⁴24. Goulart AC, Oliveira IRS, Alencar AP, Santos MSC, Santos IS, Martines BMR, et al. Diagnostic accuracy of a noninvasive hepatic ultrasound score for non-alcoholic fatty liver disease (NAFLD) in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Sao Paulo Med J 2015; 133:115-24.. Hepatic beam attenuation was based on the visibility of intra-hepatic vessels and the diaphragm posterior to the right hepatic lobe. The absence of steatosis was defined by a normal hepatic attenuation with complete diaphragm visualization. Abnormal attenuation was classified into mild (diaphragm > 50% visible), moderate (diaphragm < 50% visible), or severe (no visible diaphragm) ²⁴24. Goulart AC, Oliveira IRS, Alencar AP, Santos MSC, Santos IS, Martines BMR, et al. Diagnostic accuracy of a noninvasive hepatic ultrasound score for non-alcoholic fatty liver disease (NAFLD) in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Sao Paulo Med J 2015; 133:115-24.. The ultrasonography quality was based on identifying four anatomic marks (anterior hepatic surface, posterior hepatic surface, gallbladder, and portal vein) in images of the right hepatic lobe in the axial plane. Exams in which fewer than two anatomical landmarks were identified were considered “poor”.

NAFLD was defined by ultrasound-based steatosis at baseline. MASLD was defined by ultrasound-based steatosis and the presence of at least one of the following cardiometabolic risk factors, according to the MASLD definition: overweight/obesity defined by elevated body mass index (BMI) or elevated waist circumference (WC), prediabetes, elevated blood pressure, hypertriglyceridemia, and low HDL-cholesterol ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73..

Height was measured with a stadiometer and weight with an electronic scale. WC was measured using an inelastic tape at the midpoint between the lower rib margin and the anterior superior iliac crest in the mid-axillary line. BMI was calculated as weight (kg) divided by height squared (m²). Overweight/obesity as a cardiometabolic criterion for MASLD was defined as BMI ≥ 23kg/m² for self-declared Asian and a BMI ≥ 25kg/m² for other self-declared ethnicities groups, or WC > 94cm for males and > 80cm for females ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73..

Prediabetes was defined by blood tests performed in each visit: FPG ≥ 100mg/dL (5.6mmol/L), 2h-75g OGTT ≥ 140mg/dL (7.8mmol/L), or HbA1c ≥ 5.7% (39mmol/mol) ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73..

Blood pressure was measured trice during the clinic visits, and the mean of the last two measurements was calculated. Use of anti-hypertensive medication was defined as the presence of both: (1) the self-reported use of medication to treat arterial hypertension in the previous two weeks and (2) confirmation of an anti-hypertensive drug by cross-checking prescriptions and boxes brought by the participants. Elevated blood pressure as a MASLD criterion ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73. was defined as a mean systolic pressure ≥ 130mmHg and/or a mean diastolic pressure ≥ 85mmHg, or by use of anti-hypertensive medication.

Hypertriglyceridemia as a MASLD criterion ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73. was defined as plasma triglycerides ≥ 150mg/dL (1.7mmol/L) in fasting blood collection or using a triglyceride-lowering medication (fibrates or nicotinic acid) checked in prescriptions and boxes brought by the participant ¹⁸18. Fedeli LG, Vidigal PG, Leite CM, Castilhos CD, Pimentel RA, Maniero VC, et al. Logística de coleta e transporte de material biológico e organização do laboratório central no ELSA-Brasil. Rev Saúde Pública 2013; 47 Suppl 2:63-71..

Low HDL-cholesterol as a MASLD criterion ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73. was defined as plasma HDL-cholesterol ≤ 40mg/dL (1.0mmol/L) in males or ≤ 50mg/dL (1.3mmol/L) in females in fasting blood collection, or by use of a lipid-lowering medication (same as in hypertriglyceridemia).

Baseline covariates

During interviews, data were collected on: age (in years), sex (female/male), self-reported skin color/race (Brazilian census categories of white, black, brown [mixed-race], Asian, and Indigenous), education (less than high school, high school, university degree), per capita household income (in Brazilian Real - BRL), history of diabetes in first-degree relatives (yes/no), and smoking status (current, former, or never).

Information about alcohol consumption (g/week) was obtained by a questionnaire considering the frequency and quantity of beer (considered as having 5% alcohol), wine, and distillates consumption. Participants were categorized into abstainers (0g/week) and light to moderate (1-140g/week for females and 1-210g/week for males) drinkers. The long form of the International Physical Activity Questionnaire (IPAQ) for leisure and transportation domains ²⁵25. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML, Ainsworth BE, et al. International Physical Activity Questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003; 35:1381-95., validated for the Brazilian population ²⁶26. Hallal PC, Simoes E, Reichert FF, Azevedo MR, Ramos LR, Pratt M, et al. Validity and reliability of the telephone-administered International Physical Activity Questionnaire in Brazil. J Phys Act Health 2010; 7:402-9.^,2727. Hallal PC, Victora CG. Reliability and validity of the International Physical Activity Questionnaire (IPAQ). Med Sci Sports Exerc 2004; 36:556., was used to estimate the weekly volume of physical activity in metabolic equivalents (METs) according to instrument guidelines.

Statistical analysis

Continuous normal variables were presented as means and standard deviation, and continuous non-normal variables as medians and interquartile intervals. Normality was verified with the Shapiro-Wilk test. Categorical variables were described as frequencies and percentages. The Cox proportional hazards model was used to estimate hazard ratios (HR) and 95% confidence intervals (95%CI) for diabetes. The proportional hazards assumption was assured with the analysis of Schoenfeld residuals.

The prevalence of hypertension was described at baseline only for description purposes, since our exposure variable (MASLD) also contains a measure of high blood pressure. In Table 1, the variable “Hypertension” was defined as confirmed use of anti-hypertensive medications in the previous two weeks or mean high blood pressure (≥ 140 and/or ≥ 90 mmHg) measured at the clinic ²⁸28. Barroso WKS, Rodrigues CIS, Bortolotto LA, Mota-Gomes MA, Brandão AA, Feitosa ADDM, et al. Diretrizes brasileiras de hipertensão arterial - 2020. Arq Bras Cardiol 2021; 116:516-658..

For each predictor (NAFLD and MASLD) two models were performed. Model 1 was unadjusted and Model 2 included age (years), sex (male/female), study center (Bahia, Espírito Santo, Minas Gerais, Rio de Janeiro, Rio Grande do Sul, São Paulo), race/skin color (white, mixed-race, black, others), education (less than high school, high school, university degree), per capita household income (BRL), smoking status (current, former, never), alcohol intake categories (abstinence, light-moderate drinking), physical activity (METs/week), and family history of diabetes mellitus (yes/no). The variables for adjustment were selected based on previous literature evaluating diabetes risk ²⁹29. Bellou V, Belbasis L, Tzoulaki I, Evangelou E. Risk factors for type 2 diabetes mellitus: an exposure-wide umbrella review of meta-analyses. PLoS One 2018; 13:e0194127.. These variables are generally also related to MASLD. Variables that were part of the MASLD definition were not included in the model adjustment. Variance inflation factors were used to assess important collinearity (> 5).

Additional models were performed, including interaction terms for NAFLD and MASLD with race/skin color groups. For interaction analyses, subjects who self-declared as Asian or Indigenous were grouped in the category “Other”, due to small numbers. However, black and brown/mixed-race individuals were not grouped to permit assessment of a graded increased risk of diabetes among all groups. Black and mixed-race subjects are at increased risk of diabetes, and they are unfavorably exposed to the social determinants of health compared to white subjects ³⁰30. Hill-Briggs F, Ephraim PL, Vrany EA, Davidson KW, Pekmezaris R, Salas-Lopez D, et al. Social determinants of health, race, and diabetes population health improvement: Black/African Americans as a population exemplar. Curr Diab Rep 2022; 22:117-28.. Thus, tests were performed for effect modification by race/skin color categories using the likelihood ratio test, comparing the goodness of fit between the models with and without the interaction terms.

All the analyses were performed using software R, version 4.3.0 (http://www.r-project.org). A 2-sided p-value < 0.05 was considered statistically significant for all associations.

Results

Of the 7,073 individuals in the analytic sample free of diabetes at baseline, 2,395 participants (33.9%) were classified as having NAFLD and 2,298 (32.5%) had MASLD. Most subjects with NAFLD (96%) were also classified as having MASLD.

Our sample comprised males and females with an average age of 51 years and of various self-declared multiethnic groups, including 1,036 (14.6%) individuals self-declared as black and 1,778 (25.1%) as mixed-race. Less than 4% individuals self-declared as Asian or Indigenous. Almost half did not have a university degree. Family history of diabetes and diagnosis of hypertension were frequent, and current smoking was uncommon (Table 1).

Those with NAFLD or MASLD at baseline were similar regarding these characteristics in all ethnic groups. The most common cardiometabolic risk factor among those with MASLD was overweight/obesity (BMI ≥ 25kg/m² or BMI ≥ 23kg/m² in Asians) in all ethnic groups. Compared to white and mixed-race subjects, black individuals with MASLD presented a higher prevalence of the following metabolic abnormalities: elevated BMI, elevated WC, elevated blood pressure or medication, prediabetes, and hypertriglyceridemia, but a smaller prevalence of low HDL (Tables 2, 3, and 4).

Prediction of incident diabetes from NAFLD and MASLD

A total of 1,157 (16%) new cases of type 2 diabetes mellitus were identified during a median (interquartile range - IQR) follow-up of 9.43 (8.86; 9.87) years. In unadjusted models, people with NAFLD (HR = 1.87; 95%CI: 1.67-2.10) or MASLD (HR = 1.99; 95%CI: 1.77-2.23) were at higher risk of developing type 2 diabetes mellitus. After adjustment for covariates, the association between liver disease and incidence of type 2 diabetes mellitus remained significant (NAFLD HR = 1.78; 95%CI: 1.58-2.01; MASLD HR = 1.88; 95%CI: 1.67-2.12) (Table 5).

Heterogeneity among race/skin color groups

We did not find statistically significant heterogeneity in the associations among race/skin color groups for both definitions of steatotic liver disease (p for heterogeneity for NAFLD = 0.08; MASLD = 0.10), although p-values were close to the 5% criterion. In addition, as seen in Figure 2, white, black, and mixed-race individuals had increased diabetes risk related to steatotic liver disease, regardless of the definition used.

Discussion

To the best of our knowledge, this is the first study investigating the association between MASLD and diabetes in a non-Asian sample. Our findings show that within an average of 9 years of follow-up, having MASLD at baseline predicted the development of type 2 diabetes in a similar manner as having NAFLD. Although primarily based on white participants, we had a considerable number of black and mixed-race subjects, finding a similar increased risk in all ethnic groups.

We found a major overlap between NAFLD and MASLD, consistent with a previous report ³¹31. Perazzo H, Pacheco AG, Griep RH, Gracindo R, Goulart AC, Fonseca MJM. Changing from NAFLD through MAFLD to MASLD: similar prevalence and risk factors in a large Brazilian cohort. J Hepatol 2023; 80:e72-4. in the same sample, highlighting the fact that metabolic dysfunction plays a central role in this type of steatotic liver disease. A major overlap was also seen in a Chinese sample of individuals aged 18 years or older ⁷7. He L, Zheng W, Qiu K, Kong W, Zeng T. Changing from NAFLD to MASLD: the new definition can more accurately identify individuals at higher risk for diabetes. J Hepatol 2023; 80:e85-7.. However, our findings differ from those of the Chinese sample in two ways. First, our prevalence of steatosis was much higher, 34% vs. 18%. Second, the associations with diabetes in the Chinese sample were much larger for both NAFLD and MASLD. In fact, the associations between NAFLD and the incidence of diabetes found in that study are among the largest published so far ³3. Mantovani A, Petracca G, Beatrice G, Tilg H, Byrne CD, Targher G. Non-alcoholic fatty liver disease and risk of incident diabetes mellitus: an updated meta-analysis of 501 022 adult individuals. Gut 2021; 70:962-9.^,3232. Mantovani A, Byrne CD, Bonora E, Targher G. Nonalcoholic fatty liver disease and risk of incident type 2 diabetes: a meta-analysis. Diabetes Care 2018; 41:372-82..

Of note, obesity, which is occurring in epidemic proportions globally ³³33. Chong B, Jayabaskaran J, Kong G, Chan YH, Chin YH, Goh R, et al. Trends and predictions of malnutrition and obesity in 204 countries and territories: an analysis of the Global Burden of Disease Study 2019. EClinicalMedicine 2023; 57:101850., was the most important component of steatotic liver disease regardless of the definition used. This highlights the importance of interventions to counter obesity in the population.

Concerning race/color, we provide new evidence complementing previous findings based on a U.S. sample that reported an increased risk of diabetes associated with NAFLD in whites but not in black individuals ¹⁰10. Hatano Y, VanWagner LB, Carnethon MR, Bancks MP, Carson AP, Lloyd-Jones DM, et al. Racial difference in the association between non-alcoholic fatty liver disease and incident type 2 diabetes: findings from the CARDIA study. Diabetologia 2023; 66:1235-46.. Rather than the absence of an increased risk of NAFLD-related diabetes, we demonstrated that black and mixed-race subjects exposed to NAFLD or MASLD have an increased risk of diabetes compared to their non-exposed counterparts.

Explanations previously proposed for racial differences related to NAFLD were potential ethnic differences in body fat distribution ³⁴34. Agbim U, Carr RM, Pickett-Blakely O, Dagogo-Jack S. Ethnic disparities in adiposity: focus on non-alcoholic fatty liver disease, visceral, and generalized obesity. Curr Obes Rep 2019; 8:243-54., lipid profile ³⁴34. Agbim U, Carr RM, Pickett-Blakely O, Dagogo-Jack S. Ethnic disparities in adiposity: focus on non-alcoholic fatty liver disease, visceral, and generalized obesity. Curr Obes Rep 2019; 8:243-54.^,3535. Yu SSK, Castillo DC, Courville AB, Sumner AE. The triglyceride paradox in people of African descent. Metab Syndr Relat Disord 2012; 10:77-82., and health social determinants ³⁶36. Ochoa-Allemant P, Marrero JA, Serper M. Racial and ethnic differences and the role of unfavorable social determinants of health across steatotic liver disease subtypes in the United States. Hepatol Commun 2023; 7:e0324.. The Brazilian self-declared classification of Afro-descendants into black and brown [mixed-race] individuals is a way to capture differential exposure to health social determinants related to the construct of race/skin color. Previous studies from ELSA-Brasil have shown that both black and mixed-race subjects face disproportionate exposure to adverse life conditions compared to whites, such as more residential segregation ³⁷37. Barber S, Diez Roux AV, Cardoso L, Santos S, Toste V, James S, et al. At the intersection of place, race, and health in Brazil: residential segregation and cardio-metabolic risk factors in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Soc Sci Med 2018; 199:67-76. and lower socioeconomic position ³⁸38. Mendes PM, Nobre AA, Griep RH, Juvanhol LL, Barreto SM, Fonseca MJM, et al. Association between race/color and incidence of hypertension in the ELSA-Brasil population: investigating the mediation of racial discrimination and socioeconomic position. Ethn Health 2022; 27:1047-57.. However, compared to mixed-race, black individuals experienced higher levels of racial discrimination ³⁹39. Camelo LV, Machado AV, Chor D, Griep RH, Mill JG, Brant LCC, et al. Racial discrimination is associated with greater arterial stiffness and carotid intima-media thickness: the ELSA-Brasil study. Ann Epidemiol 2022; 72:40-7., were more likely to live in segregated neighborhoods, and had higher adjusted prevalences of hypertension and diabetes ³⁷37. Barber S, Diez Roux AV, Cardoso L, Santos S, Toste V, James S, et al. At the intersection of place, race, and health in Brazil: residential segregation and cardio-metabolic risk factors in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Soc Sci Med 2018; 199:67-76.. Despite these disparities, our findings did not show a gradient in the risk of diabetes among race groups.

Potential limitations of our findings must be considered. First, residual confounding is always possible in observational studies, although we included important variables related to diabetes and MASLD. Second, although our diabetes ascertainment was very comprehensive, based on fasting and 2h glycemia, as well as HbA1c levels, it may have high sensitivity but low specificity, perhaps introducing more false positive incident cases and thus decreasing the true magnitude of the association. Third, we did not exclude other less frequent secondary causes of steatosis and did not evaluate indicators of liver fibrosis and other advanced liver diseases due to the unavailability of this information. Fourth, our models were adjusted only by confounding variables measured at baseline, overlooking potential changes in the effect of confounders across time, which seems especially relevant for lifestyle habits such as alcohol consumption and physical activity.

Our findings have some strengths to be highlighted. Being a large prospective cohort study with highly standardized measurements and numerous relevant covariates, we were able to adjust for multiple confounders addressed in the literature ⁴⁰40. Huh Y, Cho YJ, Nam GE. Recent epidemiology and risk factors of nonalcoholic fatty liver disease. J Obes Metab Syndr 2022; 31:17-27. and to evaluate possible interactions among race/skin color groups. In this regard, we provided information on self-reported race/skin color in the association of incident diabetes with MASLD, supporting the use of the new MASLD definition to predict diabetes in various ethnic groups.

Finally, although the new MASLD definition provides similar information compared to the previous NAFLD in terms of diabetes prediction, it may offer additional advantages at the nomenclature level. However, additional investigation is needed. The inclusion of individuals with heavy alcohol intake and secondary causes of liver disease, now defined as metabolic dysfunction and alcohol-associated liver disease (MetALD) and MASLD with combined etiology ⁶6. Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multi-society Delphi consensus statement on new fatty liver disease nomenclature. Hepatology 2023; 313:2263-73., have not been evaluated regarding diabetes and other cardiometabolic conditions. In addition, the impact of applying ethnic-specific waist circumference cutoffs in the MASLD definition merits further exploration. Finally, further evaluation of health social determinants regarding the MASLD-diabetes association in different ethnic groups is warranted.

Conclusion

In this Brazilian sample of self-declaring white, mixed-race, black, and in lower proportions also Asian and Indigenous subjects, MASLD predicted similar relative risks of diabetes compared to NAFLD, with similar magnitudes in all ethnic groups.

Acknowledgments

The authors thank the staff and the participants of ELSA-Brasil for their essential contributions. To Brazilian Ministry of Health, Brazilian Ministry of Science, Technology, and Innovation, and Brazilian Coordination for the Improvement of Higher Education Personnel (CAPES) for the financial support. The ELSA-Brasil study was supported by the Brazilian Ministry of Health, Brazilian Ministry of Science, Technology and Innovation, and the Brazilian National Research Council (CNPq).

References

Publication Dates

History