Diabetes Risk Factors in People With HIV Receiving Pitavastatin Versus Placebo for Cardiovascular Disease Prevention : A Randomized Trial

Abstract

Background: REPRIEVE (Randomized Trial to Prevent Vascular Events in HIV) led to new guidelines for statin use among people with HIV (PWH) with low to moderate risk for atherosclerotic cardiovascular disease (ASCVD). Little is known about the natural history of diabetes mellitus (DM) or mechanisms contributing to statin effects on DM among this population.

Objective: To determine the contribution of known DM risk factors to excess risk for DM with pitavastatin in REPRIEVE.

Design: Phase 3, primary ASCVD prevention trial over a median of 5.6 years of follow-up. (ClinicalTrials.gov: NCT02344290).

Setting: Global, multicenter trial.

Participants: 7731 PWH aged 40 to 75 years with low to moderate ASCVD risk (by the pooled cohort equations from the American College of Cardiology and American Heart Association) without DM at study entry.

Intervention: Random 1:1 assignment to pitavastatin, 4 mg daily, or placebo.

Measurements: New-onset DM was determined at each visit by clinical diagnosis requiring initiation of medication treatment for DM. The incidence of new-onset DM was assessed in relation to predefined demographic and metabolic risk factors, stratified by treatment group. Treatment effects of pitavastatin on progression to new DM in key subgroups were determined.

Results: Participants with at least 3 DM risk factors (vs. no risk factors) had increased risk for DM in each treatment group (incidence rate, 3.24 per 100 person-years [PY] vs. 0.34 per 100 PY [pitavastatin] and 2.66 per 100 PY vs. 0.27 per 100 PY [placebo]). The incidence of DM was highest in South Asia. In adjusted analyses, high body mass index, prediabetes, and metabolic syndrome components were strongly associated with new-onset DM (all P < 0.005).

Limitation: Pitavastatin was the only statin assessed; DM was assessed clinically.

Conclusion: Metabolic risk factors, including prediabetes and obesity, contributed to new-onset DM in statin- and placebo-treated participants. A clinically significant effect of pitavastatin on DM was seen primarily among those with multiple risk factors for DM at entry. Strategies targeting key metabolic risk factors, like obesity and prediabetes, may help protect against DM among PWH.

Primary funding source: National Heart, Lung, and Blood Institute of the National Institutes of Health.

Figure 1:. Cumulative Probability of New-Onset Diabetes Over Time by Number of Diabetes Risk Factors, by Treatment Group

Participant follow-up was calculated as the number of days from date of randomization to the date of first event or date of last contact, whichever was earlier; participants with no contact after entry were included with 1 day imputed as censoring time. Months on study is defined in terms of calendar months (30.44 days). Deaths are treated as competing events. The following are considered diabetes risk factors: taking anti-diabetic medication, fasting glucose >100 mg/dL, BMI >30 kg/m², elevated waist circumference, reduced HDL, and elevated triglycerides (or taking non-statin lipid-lowering therapy). Participants are classified occurring to available diabetes risk factors.

Figure 2:. Modification of Treatment Effect on Confirmed Diabetes

Each risk factor shown was evaluated in a separate model limited to the risk factor, treatment group, and the treatment group, risk factor interaction. P-value is a type 3 p-value of an overall interaction from log linear (Poisson) regression model. Participants missing a covariate value are excluded from the modeling for that risk factor (see Table 1 and 2 for details of missing data). The following are considered diabetes risk factors: taking anti-diabetic medication, fasting glucose >100, BMI >30kg/m², elevated waist circumference, reduced HDL, and elevated triglycerides (or taking non-statin lipid-lowering therapy). Prediabetes is defined as fasting glucose >100 or currently taking anti-diabetic medication. Race-specific cut-offs for BMI to define overweight and obese use 23-<27.5 kg/m² and >27.5 kg/m² for participants of Asian race and the standard WHO cut-offs of 25-<30 kg/m² and >30 kg/m² otherwise.

Figure 3:. Predicted Diabetes Incidence by Risk Factor Profile

Predicted values with 95% confidence intervals grouped by risk factor profile and standardized by the exponent of participant follow-up time are derived from the log linear (Poisson) model and presented per 100 person years. Model includes GBD region and race (within the High Income setting), treatment group, pre-diabetes, BMI, hypertension, and other metabolic syndrome components (elevated waist circumference, elevated triglycerides, and reduced HDL, respectively); model also includes an interaction between BMI and pre-diabetes and between GBD region/race and treatment group (see Supplementary Figure 5). Race-specific cut-offs for BMI to define overweight and obese use 23-<27.5 kg/m² and >27.5 kg/m² for participants of Asian race and the standard WHO cut-offs of 25-<30 kg/m² and >30 kg/m² otherwise. Participants with missing covariate information (n=557) are excluded from modeling. Risk factor combinations shown are those observed in both treatment groups within the subgroup, limited to the 35 most frequently observed combinations plus any with a predicted diabetes incidence >6.5 /100 PY. In panel b, the blue shaded region shows IR difference of ±1 /100PY to reflect a clinically meaningful difference of one additional diabetes case per 100PY between treatment groups; the marker size reflects the number of participants with the risk factor combination with the REPRIEVE cohort.

Figure 3:. Predicted Diabetes Incidence by Risk Factor Profile

Predicted values with 95% confidence intervals grouped by risk factor profile and standardized by the exponent of participant follow-up time are derived from the log linear (Poisson) model and presented per 100 person years. Model includes GBD region and race (within the High Income setting), treatment group, pre-diabetes, BMI, hypertension, and other metabolic syndrome components (elevated waist circumference, elevated triglycerides, and reduced HDL, respectively); model also includes an interaction between BMI and pre-diabetes and between GBD region/race and treatment group (see Supplementary Figure 5). Race-specific cut-offs for BMI to define overweight and obese use 23-<27.5 kg/m² and >27.5 kg/m² for participants of Asian race and the standard WHO cut-offs of 25-<30 kg/m² and >30 kg/m² otherwise. Participants with missing covariate information (n=557) are excluded from modeling. Risk factor combinations shown are those observed in both treatment groups within the subgroup, limited to the 35 most frequently observed combinations plus any with a predicted diabetes incidence >6.5 /100 PY. In panel b, the blue shaded region shows IR difference of ±1 /100PY to reflect a clinically meaningful difference of one additional diabetes case per 100PY between treatment groups; the marker size reflects the number of participants with the risk factor combination with the REPRIEVE cohort.

Figure 3:. Predicted Diabetes Incidence by Risk Factor Profile

Predicted values with 95% confidence intervals grouped by risk factor profile and standardized by the exponent of participant follow-up time are derived from the log linear (Poisson) model and presented per 100 person years. Model includes GBD region and race (within the High Income setting), treatment group, pre-diabetes, BMI, hypertension, and other metabolic syndrome components (elevated waist circumference, elevated triglycerides, and reduced HDL, respectively); model also includes an interaction between BMI and pre-diabetes and between GBD region/race and treatment group (see Supplementary Figure 5). Race-specific cut-offs for BMI to define overweight and obese use 23-<27.5 kg/m² and >27.5 kg/m² for participants of Asian race and the standard WHO cut-offs of 25-<30 kg/m² and >30 kg/m² otherwise. Participants with missing covariate information (n=557) are excluded from modeling. Risk factor combinations shown are those observed in both treatment groups within the subgroup, limited to the 35 most frequently observed combinations plus any with a predicted diabetes incidence >6.5 /100 PY. In panel b, the blue shaded region shows IR difference of ±1 /100PY to reflect a clinically meaningful difference of one additional diabetes case per 100PY between treatment groups; the marker size reflects the number of participants with the risk factor combination with the REPRIEVE cohort.