Diabetes for Researchers

Interpretable Machine Learning Framework Reveals Robust Gut Microbiome Features Associated With Type 2 Diabetes

OBJECTIVE

To identify the core gut microbial features associated with type 2 diabetes risk and potential demographic, adiposity, and dietary factors associated with these features.

RESEARCH DESIGN AND METHODS

We used an interpretable machine learning framework to identify the type 2 diabetes–related gut microbiome features in the cross-sectional analyses of three Chinese cohorts: one discovery cohort (n = 1,832, 270 cases of type 2 diabetes) and two validation cohorts (cohort 1: n = 203, 48 cases; cohort 2: n = 7,009, 608 cases). We constructed a microbiome risk score (MRS) with the identified features. We examined the prospective association of the MRS with glucose increment in 249 participants without type 2 diabetes and assessed the correlation between the MRS and host blood metabolites (n = 1,016). We transferred human fecal samples with different MRS levels to germ-free mice to confirm the MRS–type 2 diabetes relationship. We then examined the prospective association of demographic, adiposity, and dietary factors with the MRS (n = 1,832).

RESULTS

The MRS (including 14 microbial features) consistently associated with type 2 diabetes, with risk ratio for per 1-unit change in MRS 1.28 (95% CI 1.23–1.33), 1.23 (1.13–1.34), and 1.12 (1.06–1.18) across three cohorts. The MRS was positively associated with future glucose increment (P < 0.05) and was correlated with a variety of gut microbiota–derived blood metabolites. Animal study further confirmed the MRS–type 2 diabetes relationship. Body fat distribution was found to be a key factor modulating the gut microbiome–type 2 diabetes relationship.

CONCLUSIONS

Our results reveal a core set of gut microbiome features associated with type 2 diabetes risk and future glucose increment.

Read More

Glycemic Control and Risk of Cellulitis

OBJECTIVE

We know that diabetes predisposes to common infections, such as cellulitis and pneumonia. However, the correlation between the level of glycemic control and the rate of infection is unknown.

RESEARCH DESIGN AND METHODS

We examined the association between glycemic control in patients with diabetes and the incidence of infection in the entire population of patients with diabetes in a large HMO. During the study period, we first selected an HbA1c test for each patient and then searched for an infection diagnosis in the 60 days that followed the test. A multivariate logistic regression analysis was performed to determine the independent effect of HbA1c on the likelihood of being diagnosed with an infection. We were able to control for many confounders, such as other chronic illness, time since the diagnosis of diabetes, and use of steroids before the infection.

RESULTS

We identified 407 cases of cellulitis. Multivariate logistic regressions for cellulitis showed a 1.4-fold increased risk among patients with HbA1c >7.5% (58 mmol/mol). Factors such as obesity, Parkinson’s disease, peripheral vascular disease, and prior treatment with prednisone predisposed to cellulitis. There was an increase of 12% in the odds of cellulitis for every 1% (11 mmol/mol) elevation in HbA1c (odds ratio [OR] 1.12; CI 1.05–1.19). A similar analysis showed a trend toward an increased risk of pneumonia in patients with HbA1c >7.5% (58 mmol/mol) (OR 1.1; CI 0.9–1.4).

CONCLUSIONS

Poor glycemic control was associated in this study with the development of cellulitis. The study also suggests that exposure to oral prednisolone increases the risk of cellulitis, pneumonia, and upper respiratory infection.

Read More

Incremental Risk of Developing Severe COVID-19 Among Mexican Patients With Diabetes Attributed to Social and Health Care Access Disadvantages

OBJECTIVE

Diabetes is an important risk factor for severe coronavirus disease 2019 (COVID-19), but little is known about the marginal effect of additional risk factors for severe COVID-19 among individuals with diabetes. We tested the hypothesis that sociodemographic, access to health care, and presentation to care characteristics among individuals with diabetes in Mexico confer an additional risk of hospitalization with COVID-19.

RESEARCH DESIGN AND METHODS

We conducted a cross-sectional study using public data from the General Directorate of Epidemiology of the Mexican Ministry of Health. We included individuals with laboratory-confirmed severe acute respiratory syndrome coronavirus 2 between 1 March and 31 July 2020. The primary outcome was the predicted probability of hospitalization, inclusive of 8.5% of patients who required intensive care unit admission.

RESULTS

Among 373,963 adults with COVID-19, 16.1% (95% CI 16.0–16.3) self-reported diabetes. The predicted probability of hospitalization was 38.4% (37.6–39.2) for patients with diabetes only and 42.9% (42.2–43.7) for patients with diabetes and one or more comorbidities (obesity, hypertension, cardiovascular disease, and chronic kidney disease). High municipality-level of social deprivation and low state-level health care resources were associated with a 9.5% (6.3–12.7) and 17.5% (14.5–20.4) increased probability of hospitalization among patients with diabetes, respectively. In age-, sex-, and comorbidity-adjusted models, living in a context of high social vulnerability and low health care resources was associated with the highest predicted probability of hospitalization.

CONCLUSIONS

Social vulnerability contributes considerably to the probability of hospitalization among individuals with COVID-19 and diabetes with associated comorbidities. These findings can inform mitigation strategies for populations at the highest risk of severe COVID-19.

Read More

Patient Health Utility Equations for a Type 2 Diabetes Model

OBJECTIVE

To estimate the health utility impact of diabetes-related complications in a large, longitudinal U.S. sample of people with type 2 diabetes.

RESEARCH DESIGN AND METHODS

We combined Health Utilities Index Mark 3 data on patients with type 2 diabetes from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) and Look AHEAD (Action for Health in Diabetes) trials and their follow-on studies. Complications were classified as events if they occurred in the year preceding the utility measurement; otherwise, they were classified as a history of the complication. We estimated utility decrements associated with complications using a fixed-effects regression model.

RESULTS

Our sample included 15,252 persons with an average follow-up of 8.2 years and a total of 128,873 person-visit observations. The largest, statistically significant (P < 0.05) health utility decrements were for stroke (event, –0.109; history, –0.051), amputation (event, –0.092; history, –0.150), congestive heart failure (event, –0.051; history, –0.041), dialysis (event, –0.039), estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m2 (event, –0.043; history, –0.025), angina (history, –0.028), and myocardial infarction (MI) (event, –0.028). There were smaller effects for laser photocoagulation and eGFR <60 mL/min/1.73 m2. Decrements for dialysis history, angina event, MI history, revascularization event, revascularization history, laser photocoagulation event, and hypoglycemia were not significant (P ≥ 0.05).

CONCLUSIONS

With use of a large study sample and a longitudinal design, our estimated health utility scores are expected to be largely unbiased. Estimates can be used to describe the health utility impact of diabetes complications, improve cost-effectiveness models, and inform diabetes policies.

Read More

Clinical Impact of Residual C-Peptide Secretion in Type 1 Diabetes on Glycemia and Microvascular Complications

OBJECTIVE

To quantify the relationship of residual C-peptide secretion to glycemic outcomes and microvascular complications in type 1 diabetes.

RESEARCH DESIGN AND METHODS

C-peptide was measured in an untimed blood sample in the Scottish Diabetes Research Network Type 1 Bioresource (SDRNT1BIO) cohort of 6,076 people with type 1 diabetes monitored for an average of 5.2 years.

RESULTS

In regression models adjusted for age at onset and duration, effect sizes for C-peptide ≥200 vs. <5 pmol/L were as follows: insulin dose at baseline, 27% lower (P = 2 x 10–39); HbA1c during follow-up, 4.9 mmol/mol lower (P = 3 x 10–13); hazard ratio for hospital admission for diabetic ketoacidosis during follow-up, 0.44 (P = 0.0001); odds ratio for incident retinopathy, 0.51 (P = 0.0003). Effects on the risk of serious hypoglycemic episodes were detectable at lower levels of C-peptide, and the form of the relationship was continuous down to the limit of detection (3 pmol/L). In regression models contrasting C-peptide 30 to <200 pmol/L with <5 pmol/L, the odds ratio for self-report of at least one serious hypoglycemic episode in the last year was 0.56 (P = 6 x 10–8), and the hazard ratio for hospital admission for hypoglycemia during follow-up was 0.52 (P = 0.03).

CONCLUSIONS

These results in a large representative cohort suggest that even minimal residual C-peptide secretion could have clinical benefit in type 1 diabetes, in contrast to a follow-up study of the Diabetes Control and Complications Trial (DCCT) intensively treated cohort where an effect on hypoglycemia was seen only at C-peptide levels ≥130 pmol/L. This has obvious implications for the design and evaluation of trials of interventions to preserve or restore pancreatic islet function in type 1 diabetes.

Read More