BARCELONA – It’s taken 15 years, but newly released data are finally clearing the last shadow of doubt over metformin’s safety in type 2 diabetes.
Patients in the groundbreaking U.K. Prospective Diabetes Study (UKPDS) who took the drug along with a sulfonylurea don’t appear to have any long-term increases in the risk of diabetes-related death, all-cause mortality, or cardiovascular events, Dr. Rury R. Holman said at the annual meeting of the European Association for the Study of Diabetes.
However, although they are strongly reassuring, these data can’t yet be considered ironclad, said Dr. Holman, a primary UKPDS investigator and director of the University of Oxford Diabetes Trials Unit. "What we have here is comforting, but it is not proof positive. It’s not a second trial, but it does suggest that [the apparent early risk] may have evened out over time."
The Sulfonylurea and Metformin Substudy was designed to investigate the combination’s effect on patients in UKPDS who had not achieved the target glucose goal of 6 mmol/L (108 mg/dL), despite being on sulfonylurea monotherapy for the entire 7-year study. This group of 537 patients was secondarily randomized to either staying with the sulfonylurea monotherapy or adding metformin. "The impact of this was actually to add glycemic rescue therapy at 6 mmol/L rather than 15 mmol/L [270 mg/dL]," the level built into UKPDS, Dr. Holman said.
The substudy patients were a fairly representative mix of those in the larger study. They were a mean of 64 years old, with a mean body mass index of about 28 kg/m2. They did differ significantly in two major categories, however, Dr. Holman pointed out. Patients in the combination therapy group had lower fasting plasma glucose (9 vs. 10 mmol/L; 162 vs. 180 mg/dL), hemoglobin A1c (8.3% vs. 7.9%) and lower LDL cholesterol (3.4 vs. 3.2 mmol/L; 61.2 mg/dL vs. 57.6 mg/dL). These baseline differences could have helped set the stage for the differences in mortality outcomes, he suggested.
After the secondary randomization, those who got the addition of metformin showed an immediate, significant, decrease in fasting plasma glucose. This was a very encouraging finding, he said, because it broke the disappointing pattern of improvement followed eventually by beta-cell deterioration, which UKPDS characterized in all of its treatment arms. But, Dr. Holman said, after reaching a nadir at about 18 months, glucose levels in the dual-therapy patients did begin to creep back up, taking a parallel – although less pronounced – track to the monotherapy group.
The glucose response was encouraging, but clouded by an unexpected finding. By 6 years, those taking the drug combination were almost twice as likely to have died from a diabetes-related complication as were those in the monotherapy group (risk ratio, 1.96). The overall risk of death from any cause was also significantly increased in those taking the combination (RR, 0.60). However, there were no differences in fatal or nonfatal stroke or myocardial infarction.
"There was a separation of the mortality curves at around 3 years" that continued to separate through the entire study period, Dr. Holman said. "This was contrary to everything else UKPDS had shown, and raised considerable concern and alarm. It looked horrific and it could potentially have been horrific."
The investigators tried to make sense of the results, without much luck. Matching the cohorts with similar groups in other studies shed no light on the matter. The disconnect between the increase in all-cause mortality and stable cardiovascular events remained a puzzle. Because only 40 patients had died at that point – a reflection of the relative good baseline health of the UKPDS group – there was limited power to fully explore the finding.
Nonetheless, "when we found this, we said that these results could be real and that the addition of metformin to patients on sulfonylureas required more study, which is what we now have done."
At the meeting, Dr. Holman reported the most recent, yet-unpublished, data on these patients, who were followed for up to 15 years. In 1997, the hazard ratio for diabetes-related deaths was 1.96, with a high level of statistical significance (P = .039). It hovered near that point until 2001, when it began to decline. By 2003, it dropped to about 1.50, and was no longer statistically significant. Since then, the risk has declined close to null; it has never regained a significant value.
"Over time, the combination of sulfonylurea and metformin does not appear to have any continued adverse effect," Dr. Holman said. "We saw a diminishing risk ratio and at this point, close to 15 years after we started UKPDS, there appears to be no evidence of harm."