麻豆果冻传媒影音

CHICAGO -- Higher insulin use doesn't account for the higher mortality risk with intensive glucose management in the ACCORD trial, a subanalysis determined, although whether those results were reassuring for the clinic was subject to debate.

Each unit of insulin per kilogram body weight was associated with an 83% higher unadjusted risk of cardiovascular mortality in the trial, Elias Siraj, MD, of Temple University in Philadelphia, and colleagues reported here at the American Diabetes Association meeting.

But that association was attenuated down to a nonsignificant 12% difference after adjustment for other medication differences and clinical characteristics, and disappeared entirely after additional control for treatment arm in the trial, suggesting no independent impact.

The overall ACCORD trial showed a 22% elevated all-cause mortality and 35% elevated cardiovascular mortality with the intensive management arm targeting a hemoglobin A1c (HbA1c) of 6.5%, versus a standard strategy aiming at 7.0% HbA1c.

Prior post hoc analyses have ruled out severe hypoglycemia, weight gain, rapid reduction in HbA1c, low HbA1c itself, and other specific medications as potential mechanisms, Siraj explained.

However, the audience at the President's Session where the results were presented was sharply divided in its interpretation of the new post-hoc analysis.

Geremia B. Bolli, MD, of the University of Perugia, Italy, called the results reassuring and suggested reverse causality was at play.

"We should not think that when a high insulin dose is given to people, this could be dangerous itself," he said. "These people require huge doses of insulin because they are highly insulin resistant. Maybe they die because they are very sick and not because we give insulin."

However, "from the data, I'm more convinced that there actually is an association between insulin and cardiovascular death," said another commenter, citing a statistical analysis that was "completely overparameterized."

High doses of insulin still could be a problem for subgroups of highly insulin resistant patients hidden in the covariates, suggested Robert Fredericks, MD, an endocrinologist in private practice in Reno, Nev.

"If the real problem is an interaction between insulin resistance and higher doses of insulin, not just insulin per se but pushing insulin as a way to control sugar, we may actually be aggravating the underlying pathology," he told MedPage Today.

Another possibility is that the insulin dose escalation mandated by the aggressive management protocol really was killing patients poorly adherent to therapy, suggested Robert E. Ratner, MD, chief scientific and medical officer for the ADA.

"You're driving a system in which there's a disconnect between what the physician is telling the patient to do and what the patient is doing, and every now and then they actually take the dose of insulin you've actually told them to take, which is now considerably more than anything they really needed," he said at the session.

The issue is almost impossible to sort out without accurate continuous glucose monitoring for the entire trial, and may never be settled, he told MedPage Today.

"No one has found the silver bullet yet. It may be some combination of factors that are difficult to sort out," agreed Cora Lewis, MD, MSPH, director of the Preventive Medicine Clinic at the University of Alabama at Birmingham. "I wouldn't hold your breath."

The one ACCORD subanalysis that did show a possible explanation for the mortality impact suggested that unintended high or low glucose levels and dramatic swings in glucose were predictive.

Siraj's analysis included 10,163 individuals with type 2 diabetes who had both cardiovascular mortality and insulin dose data in the trial, with a mean follow-up of 5 years.

As expected, mean total, basal, and bolus insulin doses were significantly higher in the intensive management arm of the trial than in the standard goal arm.

Both treatment arms showed a progressively higher insulin dose with higher average HbA1c, though with a higher insulin dose at every HbA1c level in the intensive group.

Hazard ratios for cardiovascular mortality per 1 unit/kg insulin across the models were:

• In the unadjusted model, 1.83 for total insulin: 2.29 for basal, and 3.36 for bolus (all P<0.0001)
• In the model adjusted for the blood pressure and lipid treatment proportion of the trial, severe hypoglycemia, and weight change: 1.21 for total insulin, 1.30 for basal, and 1.65 for bolus (none P<0.05)
• In the model also adjusted for updated average HbA1c levels: 1.12 for total, 1.13 for basal, and 1.48 for bolus (none P<0.05)
• In the model additionally adjusted for randomization in the glucose management of the trial, 0.99 for total, 1.94 for basal, and 1.23 for bolus (none P<0.05)

The abstract also mentioned analysis by all-cause mortality, but those results were not presented.

Comment