Treating to lower glucose levels not linked to deaths in ACCORD

June 10, 2009

Intensive lowering of blood glucose to a hemoglobin A1c (HbA1c) level under 6% is not associated with the increased mortality that caused an early halt to the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial last year. The increased mortality also does not appear to be related to increased hypoglycemic episodes among those receiving intensive treatment.

Intensive lowering of blood glucose to a hemoglobin A1c(HbA1c) level under 6% is not associated with the increasedmortality that caused an early halt to the Action to Control CardiovascularRisk in Diabetes (ACCORD) trial last year. The increased mortality also doesnot appear to be related to increased hypoglycemic episodes among thosereceiving intensive treatment.

New analyses of the datashow that a greater decline in HbA1c over a one-year period wasassociated with a lower risk of death, says Matthew C. Riddle, MD, professorof medicine, Oregon Health Science University, Portland (pictured left). Dr.Riddle is also a member of the ACCORD Glycemia Management Group. "The lowerthe HbA1c, the lower the risk of death in the trial, so there is aparadoxical relationship between the increased mortality and intensiveglucose lowering. The excess mortality risk occurred in patients who failedto reduce their HbA1c," he said.

"There was a 20% increased risk of death for every 1% increase inHbA1c level above 6%, suggesting that lower blood glucose levelsmay be a worthy target in some patients," Dr. Riddle continues. The increasedrisk of death associated with higher HbA1c remained after ACCORDresults were adjusted for age, diabetes duration, and history ofcardiovascular events.

What may have caused the increased mortality in ACCORD? "We don’t know;however, people who easily achieved an HbA1c level below 7%, eventhose with cardiovascular risk factors, were less likely to have increasedrisk of death than those who struggled to reduce HbA1c below 7%,"says Dr. Riddle. Very rapid changes in weight or blood glucose levels mayalso play a role, as may some of the agents or drug combinations used in thetrial.

ACCORD researchers plan to analyze the data according to the individualantidiabetic drugs used, but since investigators had a large variety ofagents they were permitted to prescribe in order to reduce HbA1c,such analyses will be difficult to interpret, according to Dr. Riddle.

Hypoglycemia not linked to mortality

"We do not believe thatsevere hypoglycemia as defined by ACCORD was associated with the increasedrisk of death," says Denise Bonds, MD (pictured right), ACCORD medicalofficer, National Heart, Lung, and Blood Institute (NHLBI).

Severe hypoglycemia in ACCORD was defined as blood glucose levels of lessthan 50 mg/dL or symptoms consistent with hypoglycemia and recovery withglucagon or an oral carbohydrate. Hypoglycemic episodes requiring assistanceof any kind and those requiring emergency medical assistance at home or in ahospital were analyzed separately.

There were 451 total deaths across both the standard- andintensive-treatment groups, and overall, 7% of those occurred in patients whohad at least one severe hypoglycemia episode requiring emergency medicalassistance.

"There were few deaths reported within 90 days of a hypoglycemic episode,"Dr. Bonds points out. "Severe hypoglycemia overall was associated with ahigher risk of death, but in the intensive treatment arm, there was abouthalf the risk of death among those who had a hyperglycemic episode than amongpatients in the standard-treatment arm with a hyperglycemic episode."

Data analyses revealed that hypoglycemic episodes were more likely tooccur in patients with higher baseline HbA1c levels and those withpoor glycemic control during the trial. Of patients having a severehypoglycemic episode, half reported missing a meal, eating less than usual,or not counting carbohydrates for a meal preceding the episode.

"This suggests that we should institute additional educational measuresfor type 2 diabetes patients," Bonds says.

The ACCORD trial

The ACCORD trial included 10,251 adult patients (mean age, 62 years). Ofthese, 35% had a history of MI or stroke and comprised the secondaryprevention group. The remaining patients were at high risk for cardiovascularevents and had left ventricular hypertrophy, microalbuminuria, or at leasttwo of the following risk factors: high low-density lipoprotein (LDL) levels,low high-density lipoprotein (HDL) levels, hypertension, body mass indexindicating overweight or obesity, or current smoking.

Patients were randomized to receive intensive treatment aiming to reduceHbA1c to below 6% or to standard treatment with the intention ofreducing HbA1c to below 7%. A broad range of medications wereused, but 52% of patients in the intensive-treatment group versus 16% in thestandard-treatment group were likely to be taking insulin plus three oralantidiabetic agents.

The trial was discontinued early at 60% completion and 3.4 years offollow-up because of the excess mortality in the intensive-treatment arm,giving birth to the theory that HbA1c levels below 7% were linkedto increased risk of death. All current and previous ACCORD analyses havebeen based on data from the point of discontinuation. However, ACCORD hasbeen resumed and will end June 30, 2009, with complete analyses expectedwithin a year.

ACCORD is sponsored primarily by the NHLBI, with support from otherfederal health organizations.