2 new articles in NEJM from the european society of cardiology congress looked at cardiac outcomes after improved A1c levels with 2 different DPP-4 inhibitors
1. saxagliptin, a dipeptidyl peptidase 4 (DPP-4) inhibitor vs placebo given to 16.5K patients with DM2 with hx or at risk for CAD events (high risk: men >54 yo or women >59yo, with at least one of: dyslipidemia, htn, active smoking) followed 2.1 yrs. other dm meds adjusted per physicians. primary endpoint of composite of cardiovascular death, MI or ischemic stroke. secondary endpt of above plus hosp for heart failure, coronary revasc, or unstable angina. (see DOI: 10.1056/NEJMoa1307684 ). drug company sponsored study
–fasting glucose levels lower in saxagliptin group; A1c also lower (7.5% vs 7.8% at year 2, with 36.2% vs 27.9% having A1c<7% by the end of the trial) — all highly significant. associated decrease in albumin/creat ratio.
–primary endpt in 7.3% in saxagliptin vs 7.2% in placebo group — nonsignificant
–secondary endpt in 12.8% vs 12.4% — also nonsignificant, though statistically more patients were hospitalized for heart failure (3.5% vs 2.8%) with saxagliptin
–more hypoglycemia with med. no diff in pancreatitis, though infrequent event. (important, since exanitide assoc with pancreatitis in recent study: see dm exenatide and pancreatitis jamaintmed 2013 in dropbox or doi:10.1001/jamainternmed.2013.2720) . the DPP-4 inhibitors inhibit breakdown of GLP-1 (exenatide is a GLP-1 agonist)
2. alogliptin (another DPP-4 inhibitor) vs placebo given to 5400 pts with acute coronary syndrome (acute MI or hosp for unstable angina within past 15-90 days), followed median 18 months. primary endpt of composite of cardiovasc death, nonfatal MI or nonfatal stroke. secondary endpt same, with addition of urgent revasc due to unstable angina (see DOI: 10.1056/NEJMoa1305889). drug company sponsored study
–A1c lower with alogliptin vs placebo (initial median of 8%, decreased to 7.7% with med vs no change with placebo) — highly significant
–primary outcome in 11.3% if pts on alogliptin vs 11.8% on placebo — nonsignificant
–no diff in adverse events, including pancreatitis
it seems to be pretty consistent using different types of hypoglycemic agents that lowering A1C is associated with decreased development or progression of certain microvascular outcomes (though for nephropathy, most of the studies are looking at the surrogate marker of microalbumenuria and not renal failure). however, most diabetics die from macrovascular causes. metformin pretty clearly decreases cardiac deaths. the data are less clear with sulfonylureas/insulin (eg, huge retrospective VA study showing cardiac mortality is about 1/2 with metformin compared to sulfonylureas, see Ann Intern Med. 2012;157:601-610 and below). so, this makes it problematic in using A1C as a surrogate marker for macrovascular clinical diabetes outcomes. and some of the agents may lower A1C but increase cardiac mortality (rosiglitazone), or heart failure (thiazolidinediones) and perhaps DPP-4 inhibitors (above). i bring these articles up mostly to reinforce that it is problematic using surrogate markers (eg A1C) instead of real clinical outcomes. these were pretty short-term studies, and the difference in A1c was not dramatic though highly significant, but these are high risk patients. in the VA study noted below, the outcome curve diverged within 2 years, showing metformin benefit.
Ann Intern Med. 2012;157:601-610: Retrospective VA study of 250K patients with diabetes started on metformin (155K) or sulfonylurea (98K) — see dm rx metf vs sulfon VA annals 2012 in dropbox. 5 yrs of followup data, with adjusted increased risk of CVD outcomes (acute MI or stroke) 21% higher in those on sulfonylurea than metformin. Divergence of curves after 2 years. Since VA study, 95% men and 75% white. Glyburide with 25% increase and glipizide with 15% increase. These results are consistent with very old University Group Diabetes study, UKPDS and others (which have led to suggestion that metformin be first-line therapy).
Reasons for inc morality in sulfonylureas — they note increased weight, increased lipids, and increased incidence of hypoglycemia. I would add increased hyperinsulinemia. Several studies, best of which is the Quebec study about 10-15 years ago, which found that hyperinsulinemia is epidemiologically linked to CAD to at least the same extent as hyperglycemia (and indepedently). Likely attributable to the effect of insulin on platelet adherence, trophic increases in smooth muscle (part of atherogenic process), local increases in HMG-CoA reductase activity, and, as i remember, increases in endothelial dysfunction. In this light, not surprising that glipizide, which is short-acting and causes less hyperinsulinemia, seems to be better than glyburide (and probably should be the perferred sulfonylurea used, as suggested in some recent reviews). Also, this model would suggest that other medications which decrease the hyperinsulinemia may be better (eg, i would guess that exanitide would be physiologically better than insulin injections … Though we need to look at long-term data, and, i have been wrong more than once in guessing like this).
geoff