Effects of Food Insecurity on Hepatic Steatosis and Fibrosis in People With HIV.

BACKGROUND & AIMS: Food insecurity (FI) is a risk factor for nonalcoholic fatty liver disease (NAFLD) and advanced fibrosis in the general population, but its impact on liver disease in people with HIV (PWH) is unknown. METHODS: We examined the association of FI with prevalence of NAFLD and fibrosis in a diverse cohort of PWH. PWH aged ≥ 18 years on antiretroviral therapy, HIV RNA <200 copies/mL, and without other known liver diseases were screened for NAFLD (controlled attenuated parameter ≥263 decibels/meter) and advanced fibrosis (liver stiffness measurement ≥11 kilopascals) by vibration controlled transient elastography at 8 U.S. CENTERS: Participants were categorized as food insecure using the Six-Item Short Form Household Food Security Survey. We used multivariable logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) of NAFLD and advanced fibrosis by FI status. RESULTS: Among 654 PWH, NAFLD was present in 348 (53%) and advanced fibrosis in 41 (6%). FI was present in 203 of participants (31%), including 97/348 with NAFLD (28%) and 18/41 with advanced fibrosis (44%). In multivariable analysis, FI was associated with lower odds of NAFLD (OR, 0.57; 95% CI, 0.37-0.88) and a greater, but nonsignificant, odds of advanced fibrosis (OR, 1.38; 95% CI, 0.65-2.90). We identified a significant interaction between FI and diabetes (P = .02) on fibrosis risk, with greater odds of fibrosis among food insecure PWH and diabetes (OR, 3.83; 95% CI, 1.15-12.73) but not among food insecure nondiabetics (OR, 1.12; 95% CI, 0.47-2.98). CONCLUSIONS: FI is highly prevalent among PWH and associated with lower odds of NAFLD, and among PWH with diabetes, there is greater odds of advanced fibrosis. FI may contribute to hepatic fibrosis through mechanisms other than steatosis in PWH.

Health disparities in chronic liver disease.

The syndemic of hazardous alcohol consumption, opioid use, and obesity has led to important changes in liver disease epidemiology that have exacerbated health disparities. Health disparities occur when plausibly avoidable health differences are experienced by socially disadvantaged populations. Highlighting health disparities, their sources, and consequences in chronic liver disease is fundamental to improving liver health outcomes. There have been large increases in alcohol use disorder in women, racial and ethnic minorities, and those experiencing poverty in the context of poor access to alcohol treatment, leading to increasing rates of alcohol-associated liver diseases. Rising rates of NAFLD and associated fibrosis have been observed in Hispanic persons, women aged > 50, and individuals experiencing food insecurity. Access to viral hepatitis screening and linkage to treatment are suboptimal for racial and ethnic minorities and individuals who are uninsured or underinsured, resulting in greater liver-related mortality and later-stage diagnoses of HCC. Data from more diverse cohorts on autoimmune and cholestatic liver diseases are lacking, supporting the need to study the contemporary epidemiology of these disorders in greater detail. Herein, we review the existing literature on racial and ethnic, gender, and socioeconomic disparities in chronic liver diseases using a social determinants of health framework to better understand how social and structural factors cause health disparities and affect chronic liver disease outcomes. We also propose potential solutions to eliminate disparities, outlining health-policy, health-system, community, and individual solutions to promote equity and improve health outcomes.

Food Insecurity is Associated With Mortality Among U.S. Adults With Nonalcoholic Fatty Liver Disease and Advanced Fibrosis.

BACKGROUND & AIMS: Food insecurity is a growing public health challenge in the United States (U.S.) and has been linked to nonalcoholic fatty liver disease (NAFLD) and advanced fibrosis. However, little is known of how food insecurity impacts mortality risk and health care utilization in chronic liver disease. METHODS: Using a population-based cohort study of U.S. adults (≥20 years) in the National Health and Nutrition Examination Survey, 1999 to 2014, with NAFLD (estimated by the U.S. Fatty Liver Index) and advanced fibrosis (estimated by the NAFLD fibrosis score, aspartate aminotransferase-to-platelet ratio index, or Fibrosis-4 Index), food security was measured using the Department of Agriculture Food Security Survey Module. The primary outcome was all-cause mortality from National Death Index data and the secondary outcome was health care utilization, defined as ≥2 inpatient and ≥4 outpatient visits, with Cox and logistic regression, respectively, estimating associations between food insecurity and outcomes. RESULTS: Of 34,134 eligible participants (mean age, 47 years; 51% women; 14% in poverty), 4816 had NAFLD and 1654 had advanced fibrosis, with food insecurity present in 28% and 21%, respectively. All-cause age-adjusted mortality was 12 per 1000 person-years among participants with NAFLD (food-secure, 11; food-insecure, 15) and 32 per 1000 person-years among advanced fibrosis participants (food-secure, 28; food-insecure, 50). In multivariable analyses, food insecurity was independently associated with higher mortality among participants with NAFLD (hazard ratio, 1.46; 95% confidence interval [CI], 1.08-1.97) and advanced fibrosis (hazard ratio, 1.37; 95% CI, 1.01-1.86) and greater outpatient health care utilization in participants with NAFLD (odds ratio, 1.32; 95% CI, 1.05-1.67). CONCLUSIONS: Food insecurity is significantly associated with greater all-cause mortality in adults with NAFLD and advanced fibrosis. Interventions that address food insecurity among adults with liver disease should be prioritized to improve health outcomes in this population.

Gender and Other Factors Associated with Endoscopy Volume Among U.S. Gastroenterology Fellows.

Gastroenterology (GI) fellows' ability to perform procedures are evaluated by the level of competency in the cognitive and technical components of procedures in Accreditation Council for Graduate Medical Education-accredited fellowship programs.1 However, competency in endoscopic procedures correlates with the number of procedures performed.2 The American Society for Gastrointestinal Endoscopy has recommended that a minimum of 130 esophagogastroduodenoscopies (EGDs) and 275 colonoscopies be performed before procedural competency can be assessed.3 Few studies have examined program or trainee-related factors, such as trainee gender, that may influence procedural volume. In other procedural subspecialties, a gender gap exists in trainee procedural volumes, with female residents performing fewer surgical cases than males.4,5 However, whether gender-related disparities exist in endoscopy volume among GI trainees is unknown. The primary aim of this study was to determine the impact of GI fellow gender on endoscopic procedural volume during training. Secondary aims were to determine if fellow career choice or other training program-related factors, such as program size, location, or setting, affect procedure volume during fellowship.

Addressing Social Determinants of Liver Disease During the COVID-19 Pandemic and Beyond: A Call to Action.

The COVID-19 pandemic has exposed healthcare inequities in the USA and highlighted the importance of social conditions in shaping the health of persons. In the field of hepatology, social determinants of health (SDOH) are closely linked to disparities in liver disease prevalence, outcomes, and access to treatment. The economic disruption and physical distancing policies brought on by the COVID-19 pandemic have further exacerbated these disparities, and may have long-lasting health consequences for marginalized patients with chronic liver disease. There are several ways that hepatology providers can bridge the gap in health equity through addressing SDOH, extending from the individual to the community and societal levels. Interventions at the individual level include implementation of systematic screening for social barriers in our hepatology practices to identify gaps in the care cascade. At the community and societal levels, interventions include creating collaborative partnerships with public health workers to expand healthcare access to the community, increasing funding for research investigating the association of SDOH, health disparities, and liver disease, engaging in advocacy to support policy reform that tackles the upstream social determinants, and addressing racism and implicit bias. As hepatology practices adapt to the "new normal," now is the time for us to address our patients' social needs within the context of healthcare delivery and reimagine ways in which to provide care to best serve our most vulnerable patients with liver disease in the COVID-19 era and beyond.

Identifying an Optimal Liver Frailty Index Cutoff to Predict Waitlist Mortality in Liver Transplant Candidates.

BACKGROUND AND AIMS: Frailty, as measured by the Liver Frailty Index (LFI), is associated with liver transplant (LT) waitlist mortality. We sought to identify an optimal LFI cutoff that predicts waitlist mortality. APPROACH AND RESULTS: Adults with cirrhosis awaiting LT without hepatocellular carcinoma at nine LT centers in the United States with LFI assessments were included. Multivariable competing risk analysis assessed the relationship between LFI and waitlist mortality. We identified a single LFI cutoff by evaluating the fit of the competing risk models, searching for the cutoff that gave the best model fit (as judged by the pseudo-log-likelihood). We ascertained the area under the curve (AUC) in an analysis of waitlist mortality to find optimal cutoffs at 3, 6, or 12 months. We used the AUC to compare the discriminative ability of LFI+Model for End Stage Liver Disease-sodium (MELDNa) versus MELDNa alone in 3-month waitlist mortality prediction. Of 1,405 patients, 37 (3%), 82 (6%), and 135 (10%) experienced waitlist mortality at 3, 6, and 12 months, respectively. LFI was predictive of waitlist mortality across a broad LFI range: 3.7-5.2. We identified an optimal LFI cutoff of 4.4 (95% confidence interval [CI], 4.0-4.8) for 3-month mortality, 4.2 (95% CI, 4.1-4.4) for 6-month mortality, and 4.2 (95% CI, 4.1-4.4) for 12-month mortality. The AUC for prediction of 3-month mortality for MELDNa was 0.73; the addition of LFI to MELDNa improved the AUC to 0.79. CONCLUSIONS: LFI is predictive of waitlist mortality across a wide spectrum of LFI values. The optimal LFI cutoff for waitlist mortality was 4.4 at 3 months and 4.2 at 6 and 12 months. The discriminative performance of LFI+MELDNa was greater than MELDNa alone. Our data suggest that incorporating LFI with MELDNa can more accurately represent waitlist mortality in LT candidates.

Liver Transplantation Outcomes in a U.S. Multicenter Cohort of 789 Patients With Hepatocellular Carcinoma Presenting Beyond Milan Criteria.

BACKGROUND AND AIMS: The Organ Procurement and Transplantation Network recently approved liver transplant (LT) prioritization for patients with hepatocellular carcinoma (HCC) beyond Milan Criteria (MC) who are down-staged (DS) with locoregional therapy (LRT). We evaluated post-LT outcomes, predictors of down-staging, and the impact of LRT in patients with beyond-MC HCC from the U.S. Multicenter HCC Transplant Consortium (20 centers, 2002-2013). APPROACH AND RESULTS: Clinicopathologic characteristics, overall survival (OS), recurrence-free survival (RFS), and HCC recurrence (HCC-R) were compared between patients within MC (n = 3,570) and beyond MC (n = 789) who were down-staged (DS, n = 465), treated with LRT and not down-staged (LRT-NoDS, n = 242), or untreated (NoLRT-NoDS, n = 82). Five-year post-LT OS and RFS was higher in MC (71.3% and 68.2%) compared with DS (64.3% and 59.5%) and was lowest in NoDS (n = 324; 60.2% and 53.8%; overall P < 0.001). DS patients had superior RFS (60% vs. 54%, P = 0.043) and lower 5-year HCC-R (18% vs. 32%, P < 0.001) compared with NoDS, with further stratification by maximum radiologic tumor diameter (5-year HCC-R of 15.5% in DS/<5 cm and 39.1% in NoDS/>5 cm, P < 0.001). Multivariate predictors of down-staging included alpha-fetoprotein response to LRT, pathologic tumor number and size, and wait time >12 months. LRT-NoDS had greater HCC-R compared with NoLRT-NoDS (34.1% vs. 26.1%, P < 0.001), even after controlling for clinicopathologic variables (hazard ratio [HR] = 2.33, P < 0.001) and inverse probability of treatment-weighted propensity matching (HR = 1.82, P < 0.001). CONCLUSIONS: In LT recipients with HCC presenting beyond MC, successful down-staging is predicted by wait time, alpha-fetoprotein response to LRT, and tumor burden and results in excellent post-LT outcomes, justifying expansion of LT criteria. In LRT-NoDS patients, higher HCC-R compared with NoLRT-NoDS cannot be explained by clinicopathologic differences, suggesting a potentially aggravating role of LRT in patients with poor tumor biology that warrants further investigation.

Weighing the risks: Morbid obesity and diabetes are associated with increased risk of death on the liver transplant waiting list.

BACKGROUND & AIMS: Obesity is a growing problem in liver transplant (LT) candidates, paralleling the US obesity epidemic and increase in LT for non-alcoholic steatohepatitis (NASH). While post-LT survival appears to be similar in obese and non-obese patients, data are scarce regarding risk of waitlist dropout in patients with morbid obesity (BMI ≥ 40 kg/m2 ). We examined the impact of obesity on waitlist mortality and evaluated predictors of dropout in LT candidates with morbid obesity or NASH. METHODS: Competing risk analyses were performed in candidates listed between 3/2002-12/2013 to evaluate predictors of waitlist removal or death. Variables with P-value <.05 in univariable models or clinically relevant were included in multivariable models. RESULTS: Eighty-four thousand two hundred and fifty-four patients (34% female, median age 55, 15% Hispanic) were included. Compared to those with BMI 25-29.9 kg/m2 , candidates with BMI ≥ 40 kg/m2 were more likely to be female (46% vs 28%), diabetic (25% vs 18%) and have NASH (35% vs 13%); all P < .001. After adjusting for well-recognized predictors of waitlist dropout, including ascites severity, morbid obesity (HR = 1.27, CI 1.20-1.36) and diabetes (HR = 1.14, CI 1.11-1.17) were independent predictors of dropout. Morbid obesity remained a predictor (HR = 1.27, CI 1.10-1.47) of dropout in patients without ascites (24%). In NASH patients, morbid obesity (HR = 1.21, CI 1.07-1.37) and diabetes (HR = 1.15, CI 1.06-1.23) were also associated with a higher dropout risk. In patients with morbid obesity, diabetes trended towards a higher dropout risk but was not significant (HR = 1.12, CI 0.995-1.26). CONCLUSIONS: Morbid obesity and diabetes are independent predictors of death in LT candidates.

Use of hepatitis C-positive grafts in hepatitis C-negative liver transplant recipients is cost effective.

BACKGROUND: The number of patients needing liver transplantation (LT) exceeds the number of available allografts. The current opioid epidemic in this country has increased the number of potential donors infected with hepatitis C (HCV). METHODS: We assessed the incremental cost-effectiveness ratio (ICER) by comparing the costs and number of liver transplants performed using HCV-positive and HCV-negative grafts into patients without HCV infection in a decision analysis model with a 1-year time horizon. RESULTS: The use of HCV-positive grafts was found to have an ICER below $50 000 across all MELD scores. Using our baseline cohort with a model for end-stage liver disease (MELD) score of 15-22, the ICER was $21 233/additional LT performed. As the MELD scores increased, the ICER decreased. Above a MELD score of 23, the use of HCV-positive grafts became cost saving (-$115 419). Our model was robust to all variables tested in the sensitivity analyses, except drug costs. CONCLUSION: The results of our decision analysis model highlight the potential pharmacoeconomic benefit of utilizing HCV-positive grafts in LT candidates who are not infected with HCV. The use of HCV-positive grafts is at least cost effective and even cost saving in patients with MELD scores above 23.

Hepatitis C virus-HIV-coinfected patients and liver transplantation.

PURPOSE OF REVIEW: To review the experience to date and unique challenges associated with liver transplantation in hepatitis C virus (HCV)/HIV-coinfected patients. RECENT FINDINGS: The prevalence of cirrhosis and hepatocellular carcinoma is rising among HIV-infected individuals. With careful patient selection and in the absence of HCV infection, HIV-infected and HIV-uninfected liver transplant recipients have comparable posttransplant outcomes. However, in the presence of HCV infection, patient and graft survival are significantly poorer in HIV-infected recipients, who have a higher risk of aggressive HCV recurrence, acute rejection, sepsis, and multiorgan failure. Outcomes may be improved with careful recipient and donor selection and with the availability of new highly potent all-oral HCV direct acting antivirals (DAAs). Although all-oral DAAs have not been evaluated in HIV/HCV-coinfected transplant patients, HIV does not adversely impact treatment success in nontransplant populations. Therefore, there is great hope that HCV can be successful eradicated in HIV/HCV-coinfected transplant patients and will result in improved outcomes. Careful attention to drug-drug interactions with HIV antiretroviral agents, DAAs, and posttransplant immunosuppressants is required. SUMMARY: Liver transplant outcomes are poorer in HIV/HCV-coinfected recipients compared with those with HCV-monoinfection. The new HCV DAAs offer tremendous potential to improve outcomes in this challenging population.

Microbial Translocation and Gut Damage Are Associated With an Elevated Fast Score in Women Living With and Without HIV.

Background: Steatohepatitis is common in persons living with HIV and may be associated with gut microbial translocation (MT). However, few studies have evaluated the gut-liver axis in persons living with HIV. In the Women's Interagency HIV Study, we examined the associations of HIV and circulating biomarkers linked to MT and gut damage using the FibroScan-aspartate aminotransferase (FAST) score, a noninvasive surrogate for steatohepatitis with advanced fibrosis. Methods: Among 883 women with HIV and 354 without HIV, we used multivariable regression to examine the associations of HIV and serum biomarkers linked to MT and gut damage (kynurenine and tryptophan ratio, intestinal fatty acid-binding protein, soluble CD14, and soluble CD163) with a log-transformed FAST score after adjusting for key covariates. We used a path analysis and mediation models to determine the mediating effect of each biomarker on the association of HIV with FAST. Results: HIV infection was associated with a 49% higher FAST score. MT biomarker levels were higher in women with HIV than women without HIV (P < .001 for each). MT biomarkers mediated 13% to 32% of the association of HIV and FAST score. Conclusions: Biomarkers linked to MT and gut damage are associated with a higher FAST score and mediate the association of HIV with a higher FAST score. Our findings suggest that MT may be an important mechanism by which HIV increases the risk of steatohepatitis with advanced fibrosis.

Pragmatic strategies to address health disparities along the continuum of care in chronic liver disease.

Racial, ethnic, and socioeconomic disparities exist in the prevalence and natural history of chronic liver disease, access to care, and clinical outcomes. Solutions to improve health equity range widely, from digital health tools to policy changes. The current review outlines the disparities along the chronic liver disease health care continuum from screening and diagnosis to the management of cirrhosis and considerations of pre-liver and post-liver transplantation. Using a health equity research and implementation science framework, we offer pragmatic strategies to address barriers to implementing high-quality equitable care for patients with chronic liver disease.

A clinical trial of lovastatin for modification of biomarkers associated with breast cancer risk.

Pre-clinical and epidemiologic studies provide rationale for evaluating lipophilic statins for breast cancer prevention. We conducted a single-arm, biomarker modulation trial of lovastatin among women with increased risk of breast cancer. Eligibility criteria included a deleterious germline mutation in BRCA1, BRCA2, CDH1, or TP53; lifetime breast cancer risk of ≥20 % as estimated by the Claus model; or personal history of estrogen receptor and progesterone receptor-negative breast cancer. Participants received 40 mg of lovastatin orally twice daily for 6 months. We evaluated the following biomarkers before and after lovastatin use: breast duct cytology (primary endpoint), serum lipids, C-reactive protein, insulin-like growth factor-1, IGF binding protein-3, lipid peroxidation, oxidative DNA damage, 3-hydroxy-3-methylglutaryl CoA reductase genotype, and mammographic density. Thirty women were enrolled, and 26 (86.7 %) completed the study. For the primary endpoint of changes in breast duct cytology sampled by random periareolar fine needle aspiration, most participants [57.7 %, 95 % confidence interval (CI) 38.9-74.5 %] showed no change after lovastatin; 19.2 % (CI 8.1-38.3 %) had a favorable change in cytology, 7.7 % (95 % CI 1.0-25.3 %) had an unfavorable change, and 15.4 % (95 % CI 5.5-34.2 %) had equivocal results due to acellular specimens, usually after lovastatin. No significant changes were observed in secondary biomarker endpoints. The study was generally well-tolerated: 4 (13.3 %) participants did not complete the study, and one (3.8 %) required a dose reduction. This trial was technically feasible, but demonstrated no significant biomarker modulation; contributing factors may include insufficient sample size, drug dose and/or duration. The results are inconclusive and do not exclude a favorable effect on breast cancer risk.

Parental leave, childcare policies, and workplace bias for hepatology professionals: A national survey.

BACKGROUND: The presence of workplace bias around child-rearing and inadequate parental leave may negatively impact childbearing decisions and sex equity in hepatology. This study aimed to understand the influence of parental leave and child-rearing on career advancement in hepatology. METHODS: A cross-sectional survey of physician members of the American Association for the Study of Liver Diseases (AASLD) was distributed through email listserv in January 2021. The 33-item survey included demographic questions, questions about bias, altering training, career plans, family planning, parental leave, and work accommodations. RESULTS: Among 199 US physician respondents, 65.3% were women, and 83.4% (n = 166) were attendings. Sex and racial differences were reported in several domains, including paid leave, perceptions of bias, and child-rearing. Most women (79.3%) took fewer than the recommended 12 paid weeks of parental leave for their first child (average paid leave 7.5 wk for women and 1.7 for men). A majority (75.2%) of women reported workplace discrimination, including 83.3% of Black and 62.5% of Hispanic women. Twenty percent of women were asked about their/their partners' pregnancy intentions or child-rearing plans during interviews for training. Women were more likely to alter career plans due to child-rearing (30.0% vs. 15.9%, p = 0.030). Women were also more likely to delay having children than men (69.5% vs.35.9%). CONCLUSIONS: Women reported sex and maternity bias in the workplace and during training interviews, which was more frequently experienced by Black and Hispanic women. As two-thirds of women had children during training, it is a particularly influential time to reevaluate programmatic support to address long-term gender disparities in career advancement.

Racial and ethnic differences in diet quality and food insecurity among adults with fatty liver and significant fibrosis: a U.S. population-based study.

BACKGROUND: Food insecurity affects diet quality (DQ) and is associated with non-alcoholic fatty liver disease (NAFLD). While racial and ethnic disparities in NAFLD exist, the relationship between food insecurity and DQ by race and ethnicity is unknown. AIM: To examine the relationship between food insecurity and DQ in adults with NAFLD and significant fibrosis by race and ethnicity in a nationally representative cohort METHODS: We performed a cross-sectional analysis of U.S. adults (≥20 years) in the National Health and Nutrition Examination Survey 2017-2018 with vibration-controlled transient elastography (VCTE), DQ, and food security (FS) measurements. NAFLD and significant fibrosis were defined using validated VCTE cut-offs. We assessed total and component DQ by the healthy eating index (HEI)-2015, with poor scores defined as <25th percentile. We used multivariable linear and logistic regression to examine associations of FS and race/ethnicity with DQ. RESULTS: Of 1351 adults with NAFLD (17% food insecure; 248 with fibrosis), mean (standard error [SE]) DQ score was 49 (1) and 47(1.2) for food secure and insecure groups, respectively. Mean (SE) DQ was lowest for White (47[1.1]), followed by Black (49[0.9]), Hispanic (50[1.2]) and Asian persons (56[2.1]). In multivariable models, there was an inverse relationship between FS and DQ, although this did not reach statistical significance (estimated difference [coef]:-1.8 mean HEI score, 95% CI: -4.3-0.7; p = 0.14). Adjusted mean DQ scores were higher for Black (coef:+3.0, 95% CI:0.5-5.5; p = 0.02), Asian (coef:+7.4, 95% CI:3.4-11.5; p = 0.001) and Hispanic (coef: +4.3, 95% CI: 0.6-7.9; p = 0.03) compared to White persons. Greatest differences in DQ components by food security status were seen in White persons. CONCLUSION: Among adults with NAFLD, White persons had poorer DQ than other races/ethnicities. The influence of food insecurity on DQ may be potentiated in this group. Exploration of the sociocultural factors influencing DQ is needed to mitigate NAFLD disparities.

Plasma metabolomic analysis indicates flavonoids and sorbic acid are associated with incident diabetes: A nested case-control study among Women's Interagency HIV Study participants.

INTRODUCTION: Lifestyle improvements are key modifiable risk factors for Type 2 diabetes mellitus (DM) however specific influences of biologically active dietary metabolites remain unclear. Our objective was to compare non-targeted plasma metabolomic profiles of women with versus without confirmed incident DM. We focused on three lipid classes (fatty acyls, prenol lipids, polyketides). MATERIALS AND METHODS: Fifty DM cases and 100 individually matched control participants (80% with human immunodeficiency virus [HIV]) were enrolled in a case-control study nested within the Women's Interagency HIV Study. Stored blood samples (1-2 years prior to DM diagnosis among cases; at the corresponding timepoint among matched controls) were assayed in triplicate for metabolomics. Time-of-flight liquid chromatography mass spectrometry with dual electrospray ionization modes was utilized. We considered 743 metabolomic features in a two-stage feature selection approach with conditional logistic regression models that accounted for matching strata. RESULTS: Seven features differed by DM case status (all false discovery rate-adjusted q<0.05). Three flavonoids (two flavanones, one isoflavone) were respectively associated with lower odds of DM (all q<0.05), and sorbic acid was associated with greater odds of DM (all q<0.05). CONCLUSION: Flavonoids were associated with lower odds of incident DM while sorbic acid was associated with greater odds of incident DM.