Vitamin D Flops for Diabetes Prevention

SAN FRANCISCO — Vitamin D supplementation for a median 2.5 years did not significantly affect development of type 2 diabetes in a high-risk population, researchers said here.

In the so-called D2d study, involving more than 2,400 non-vitamin D deficient adults, no significant difference was observed between participants given 4,000 IU vitamin D₃ per day or placebo (hazard ratio 0.88, 95% CI 0.75-1.04, P=0.12), reported Anastassios Pittas, MD, of Tufts Medical Center in Boston, Massachusetts.

No significant differences were found between groups, either, in terms of adverse events such as hypercalcemia, fasting urine calcium, glomerular filtration rates, or nephrolithiasis, Pittas said here at the annual American Diabetes Association meeting. Findings were also published simultaneously in the New England Journal of Medicine.

“We powered this study to detect an effect of 25% or higher and we were unable to show this,” Pittas said during his oral presentation of the findings.

Roughly 80% of the participants in the study had sufficient vitamin D levels, “which could have potentially limited the study’s ability to detect a statistically significant effect,” he added.

Prior studies have linked vitamin D levels to impaired pancreatic beta-cell function and insulin resistance, suggesting that supplementation may delay the onset of type 2 diabetes in a population with prediabetes, wrote Deborah Wexler, MD, of Massachusetts General Hospital, in an accompanying editorial.

Notably, the hazard ratio presented here “does not rule out a modest benefit of vitamin D,” although she noted that the landmark Diabetes Prevention Program trial in 2002, with similar group sizes and follow-up, showed much larger impacts on diabetes incidence with a lifestyle intervention and metformin (reductions of 58% and 31%, respectively).

Wexler also allowed that patients with vitamin D deficiency might be expected to see the most benefit from supplementation.

Indeed, a post hoc analysis of 103 participants in the current study showed those with serum 25-hydroxyvitamin D levels under 12 ng/mL did get significant benefit from supplementation compared to placebo (HR 0.38, 95% CI 0.18-0.80). Among those with baseline levels of at least 12 ng/mL, however, the hazard ratio was 0.92 (95% CI 0.78-1.08) versus placebo.

“Although subgroup analysis are underpowered, it may be that targeting the intervention to those at risk by virtue of vitamin D deficiency would have yielded greater benefit,” Wexler wrote.

The authors intentionally did not target patients with vitamin D deficiencies because in prior studies the intervention worked regardless of baseline serum 25-hydroxyvitamin D levels, said co-author Myrlene Staten, MD, of the National Institute of Diabetes and Digestive and Kidney Diseases.

“If in clinical practice, you wouldn’t give vitamin D until you’d actually measured [a patient’s] levels, it makes it less generalizable than if you could treat without having to measure,” she said.

D2d enrolled people at 22 U.S. academic medical centers who met two of three 2010 ADA glycemic criteria for prediabetes and not taking more than 1,000 IU vitamin D or 600 mg calcium per day. Potential participants were identified through electronic health record data, and those consenting were screened in person; blood glucose levels were confirmed with three diagnostic tests.

Participants were then assigned 1:1 to receive either 4,000 IU of once-daily, soft-gel vitamin D₃ pills or placebo and stratified by trial site, body mass index, and race. All individuals in the study received handouts addressing lifestyle activities including diet and activity levels.

Study participants returned bottles with unused pills to measure adherence (85.8% of pills were taken) and were examined again at 3 and 6 months, and every 6 months thereafter. Half of patients had at least 2.5 years of follow-up (IQR 1.9-3.5 years for active supplementation, 1.7-3.5 years for placebo).

Of 2,423 participants (44.8% female, mean age 60), about one-third were non-white and 9.3% were Hispanic. Mean BMI among participants was 32.1 and mean HbA1c was 5.9%. Mean baseline levels of serum 25-hydroxy-vitamin D were about 28 ng/mL in each group; mean at 1 and 2 years was higher, however, in the vitamin D group versus the placebo group.

At the trial’s end at 2.5 years, diabetes developed in 293 participants in the treatment arm (24.2%) versus 323 in the placebo group (26.7%).

No meaningful difference was observed across subgroup analysis for race, BMI, or glycemic criteria for prediabetes.

In terms of safety, 3.9% of participants in the vitamin D arm and 3.1% in the placebo ceased taking the pills due to an adverse event (difference 0.8 percentage points, 95% CI -0.7 to 2.3).

Additionally, 13.9% of participants in the vitamin D arm discontinued the pills, began diabetes medication, or took additional vitamin D (above 1,000 IU) outside of the study, while 14.1% of the placebo group did. Still, in an analysis incorporating these three factors, no difference was observed between the groups in terms of diabetes development (HR 0.84, 95% CI 0.71-1.00).

Staten said the team still has other analyses to complete.

“We want to do the individual patient level data analysis and need to finish fully evaluating all of our data in terms of the continuous variables and the effect on insulin sensitivity and secretion,” Staten said. “In terms of a major risk reduction for type 2 diabetes, D2d was a well-done trial that has answered that question.”

last updated 06.07.2019