NEW YORK (Reuters Health) – Tight glycemic control may increase the risks of mortality and cardiovascular disease, new research suggests. Moderate control, with glycosylated hemoglobin A1c (HbA1c) levels around 7.5, seemed to achieve optimal outcomes.

Furthermore, insulin was more hazardous than oral antihyperglycemic agents in the retrospective study. This and other findings, based on data from more than 40,000 patients, were published online January 27 in The Lancet.

The authors of the report note that many randomized trials have linked tight glycemic control with positive or equivocal outcomes. The ACCORD trial, however, which was published in 2008, showed increased mortality associated with intensive treatment.

“This is one of the rare instances where the epidemiology followed randomized controlled trials,” lead author Dr. Craig J. Currie, from Cardiff University, UK, told Reuters Health by phone. “There is certainly sufficient evidence to suggest that a case can be made with regards to safety and extremely tight glucose control.”

Dr. Currie and his colleagues used the UK’s General Practice Research Database to identify patients aged 50 years and older who were treated for type 2 diabetes between 1986 and 2008.

They divided patients into two cohorts. The first contained 27,965 patients (mean follow-up 4.5 years) whose treatment had been intensified from oral monotherapy to combination therapy with metformin and a sulfonylurea. The second cohort consisted of 20,005 patients (mean follow-up 5.2 years) who had switched from oral agents alone to insulin, with or without oral agents. The average age in both groups was 64 years.

When the researchers compared deciles of HbA1c truncated at the lower and upper quartile, results for both cohorts after treatment escalation showed a U-shaped relationship between all-cause mortality and deciles of HbA1c. The fourth decile (median HbA1c 7.5) was associated with the lowest risk.

In cohort 1 (intensified therapy with oral agents), mortality was 7% in decile 4, compared to 9% in decile 1 (median HbA1c 6.4%) and 9% in decile 10 (median 10.5%).

In a Cox proportional hazards model adjusted for age, sex, smoking, total cholesterol, body mass index, and general comorbidity, the hazard ratio for mortality was 1.30 for decile 1 (p = 0.0072) and 1.93 for decile 10 (p < 0.0001), compared with decile 4.

In cohort 2 (intensified therapy with insulin), mortality was 12% in decile 4, 18% in decile 1 and 17% in decile 10. In this cohort, the adjusted hazard ratios for mortality were 1.79 for decile 1 (p < 0.0001) and 1.80 for decile 10 (p < 0.0001). The authors note that significant differences were also evident for deciles 2, 3 and 9.

When patients with high cardiovascular risk or renal impairment were excluded and the two cohorts were compared, the hazard ratio for mortality was 1.46 with insulin-based therapy compared to oral medications.

The adjusted risk for progression to a large-vessel disease event was also higher with insulin treatment (HR 1.31). For both cohorts combined, the adjusted risk of progression to a large-vessel disease event was 1.54 in decile 1 and 1.36 in decile 10. .

“This study will raise a few eyebrows,” Dr. Currie said. “Conventionally, doctors have always been told to drive down (blood sugar levels) as low as possible. It will come as a major surprise to many doctors that taking people down too far appears to be quite risky.

He added that the findings on insulin should not prompt urgent action, but patients should “arrange to see their doctor sometime over the next few weeks to discuss it with them.”

The research team suggests that “diabetes guidelines might need revision to include a definition of an HbA1c minimum value.”

However, Dr. Beverley Balkau from INSERM in Villejuif, France and Dr. Dominique Simon from Groupe Hospitalier Pitie Salpetriere in Paris point out in a commentary that epidemiological studies “cannot show a causal relation, and such an observational database does not provide the detailed information that is available in a randomized clinical trial.”

In particular, they note that Dr. Currie’s team provided no information on specific drugs used for treatment, medication dosages, disease severity, causes of death, or frequency of hypoglycemia.

The wisest course of action for now, they suggest, is to prescribe insulin sensitizers for as long as possible, because these agents allow patients to maintain a low HbA1c level without any risk of hypoglycemia.

The research was sponsored by Eli Lilly and Company, who participated in all phases of the study.

Additional reporting by Kate Kelland in London and Julie Steenhuysen in Chicago.

Reference:
Lancet 2010.