Deaths Higher When Insulin Is Second-Line Treatment for Type 2 Diabetes

The addition of insulin as a second-line agent after metformin in patients with type 2 diabetes is associated with increased cardiovascular risk compared with adding a sulfonylurea, a large retrospective observational study finds.

It may therefore be better to use 2 oral medications before progressing to insulin use in type 2 diabetes, says the lead author of the research, Christianne L. Roumie, MD, associate professor of internal medicine and pediatrics at Vanderbilt University, Nashville, Tennessee. The analysis of data from the US Veterans Health Administration (VA), Medicare, and the National Death Index databases was published in the June 11 issue of the Journal of the American Medical Association, a special diabetes-themed issue timed to coincide with theAmerican Diabetes Association (ADA) 2014 Scientific Sessions, which start in San Francisco on Friday.

Over a median follow-up of 14 months, patients who received insulin as second-line treatment after metformin had statistically significant increased risks of 30% for nonfatal cardiovascular outcomes and a 44% increase in all-cause mortality.

"As clinicians we are expected to evaluate the best available evidence to make informed decisions for our patients. For many patients, insulin is eventually needed to control their blood sugar because oral medications alone cannot do this. Our study suggests that it is reasonable to begin 2 oral medicines before advancing to insulin and that this combination may be safer for patients than adding insulin early to achieve blood glucose control," Dr. Roumie told Medscape Medical News.

In an accompanying editorial, Monika M. Safford, MD, a professor in the department of medicine at the University of Alabama at Birmingham, says: "The signal for potential harm is certainly concerning, suggesting that confirmatory studies addressing this issue should be conducted relatively quickly using other [US] databases (such as Medicare, Kaiser, or GroupHealth Cooperative of Puget Sound)."

However, she also cautioned that findings from work such as this can create uncertainty. The complex statistical methods needed to overcome the various sources of bias and confounding that are inherent in observational research and the fact that this is a comparative-effectiveness study make it difficult for clinicians to interpret the data, she noted. "Given these caveats, many clinicians will probably refrain from making practice changes based on this study."

Deaths Higher in Metformin/Insulin Patients, Mostly from Cancer

In the study, of 43,345 VA patients who initiated metformin therapy from 2001 to 2008 and for whom a second glucose-lowering agent was subsequently initiated at a median of 14 months, 92% (39,990) received sulfonylurea and 7% (2948) received insulin as second-line treatment. (The other 1% were excluded for either data errors or medical reasons).

Propensity-score matching was used to address the facts that the patients who received insulin intensified earlier (14 vs 18 months); had higher mean HbA1c levels (8.5% vs 7.5%); and had more comorbidities. After propensity matching, the investigators included 2436 metformin/insulin and 12,180 metformin/sulfonylurea patients.

For the primary composite outcome of acute myocardial infarction (MI), stroke hospitalization, or all-cause death, the rates were 42.7 per 1000 person-years for metformin/insulin vs 32.8/1000 for metformin/sulfonylurea, giving an adjusted hazard ratio of 1.30 (P = .009).

Combined rates of acute MI and stroke were similar between the 2 groups, 10.2 and 11.9 per 1000 person-years for metformin/insulin and metformin/sulfonylurea, respectively (HR 0.88; P = 0.52).

The main difference was in all-cause deaths, at 33.7/1000 for metformin/insulin compared with 22.7/1000 for metformin/sulfonylurea, with a hazard ratio of 1.44 (P = 0.001).

For the secondary composite outcome of fatal and nonfatal CV events, the hazard ratios were 22.8/1000 for metformin/insulin vs 22.5/1000 for metformin/sulfonylurea, a nonsignificant difference (HR 0.98; P = 0.87).

Sensitivity and subgroup analyses did not change the primary outcome significantly, although the confidence intervals were wide.

In a separate analysis looking at cause of death, hazard ratios for death from all causes were higher with insulin vs sulfonylurea, but only statistically significant for cancer death.

Dr. Safford notes that although insulin resistance has been linked to cancer death, "it is not clear whether exogenous insulin or unmeasured confounders such as the degree of insulin resistance could be responsible for the findings."

Search for Clues as to the Cause Continues

Dr. Roumie told Medscape Medical News that the underlying mechanisms for the increased risk for death found with insulin are unknown, but her group is pursuing the answers. "We have a number of studies planned to examine possible mechanisms. We are investigating type 2 diabetes outcomes associated with blood glucose swings and with episodes of hypoglycemia tied with both insulin and sulfonylurea use."

In the meantime, she advised, "Although insulin remains a reasonable option for patients who have very high glucose or who desire flexible and fast blood sugar control, most patients prefer to delay insulin initiation. The current study suggests that adding a sulfonylurea to metformin should be preferred to adding insulin for most patients who need a second diabetes drug."

In a related article also published in the same issue of JAMA, researchers show that, at least for private health insurance claims, there has been a near-universal transition to insulin analogs from human insulin over the past decade in the US — a move that has come at increased cost but without clear evidence of benefit for patients with type 2 diabetes, the authors conclude.

This project was funded by the Agency for Healthcare Research and Quality, US Department of Health and Human Services, with support from the VA. Dr. Roumie and coauthors and Dr. Safford have reported no relevant financial relationships.

JAMA. 2014;311:2288-2296; 2275-2276.

http://www.medscape.com/

Graham