We thank Chan et al. (1) for their interest in our article and agree that selection bias and a relatively short duration of follow-up (primarily due to a high rate of censoring) are potential limitations to our study. These and other potential limitations are described in detail in our discussion section (2). It should be noted that the cohort was comprised of 76,176 individuals in the primary analysis and 100,653 individuals in the secondary analysis (instead of 725,469 individuals, as referenced by Chan et al.).
We have previously systematically reviewed the evidence regarding antihypertensive drug therapy and type 2 diabetes incidence, including the ACE inhibitor drug class (3). Based on current evidence, we cannot confidently conclude that ACE inhibitors prevent diabetes. Chief among our concerns is the fact that no placebo-controlled ACE inhibitor trial has ever evaluated diabetes incidence as a blinded, predefined, primary end point, although such trials are currently underway (2). Chan et al. mention the PEACE (Prevention of Events with Angiotensin Converting Enzyme Inhibition) trial, in which diabetes incidence was assessed in a post hoc analysis (4). How many other ACE inhibitor trials have found nonsignificant results for this end point in post hoc analysis and, hence, have not published the results? Additional potential limitations of studies to date include treatment contamination, failure to control for concomitant drugs affecting glycemic control, the lack of a placebo control group, and a lack of laboratory confirmation of diabetes cases.
However, the critical question is not whether ACE inhibitors lower diabetes incidence, but whether this reduction represents a true preventative effect rather than the simple masking or delaying of latent diabetes by lowering glucose levels below an arbitrary diagnostic threshold. Do ACE inhibitors prevent diabetes-related complications independent of their blood pressure-lowering effects? Does this putative, preventative effect persist when the drug is stopped? Does therapy stabilize or even reverse β-cell dysfunction and/or insulin resistance? Because the interval between the onset of β-cell dysfunction and overt diabetes averages 10 years (5), the answers to these and other questions will require rigorously designed trials with much longer durations of follow-up. Accordingly, we feel that the role of ACE inhibition in the prevention of type 2 diabetes is far from definitively established.
