We appreciate the comments of Drs. Scheen (1) and Buchanan (2) regarding our article (3), which described the effects of withdrawal of metformin on diabetes status in the Diabetes Prevention Program (DPP). As Dr. Scheen comments, and as we pointed out in our article, clearly part of the effect of metformin was due to an acute pharmacological effect of metformin, but the great majority of the effect persisted beyond this period of washout. As he notes, the washout period of 1–2 weeks was indeed beyond the usual required period of five times the plasma half-life of the drug. The duration and nature of metformin’s effects beyond that time frame are not well established. However, there are no data to support Dr. Scheen’s conjecture that waiting for an additional time would “allow a sufficiently high number of new cases of diabetes,” which he feels limited the statistical power of the analyses. The data are what they are. Within the DPP, the investigators felt that it would not be ethical to withhold such an efficacious treatment for a prolonged period of time to answer these questions. Furthermore, pharmacological treatment of many other chronic diseases, such as hypertension and dyslipidemia, is generally given continuously. We don’t expect drugs treating these conditions to work after they are discontinued. Similarly, we shouldn’t expect drug treatment of impaired glucose regulation and other risk factors for type 2 diabetes to be effective after the drugs are discontinued.
We agree that whether drug therapy with metformin, acarbose, troglitazone, estrogens, ACE inhibitors, or angiotensin II receptor blockers truly prevent diabetes or delay its onset cannot be determined completely by the existing studies. However, delaying the onset of diabetes is clearly an important goal with major individual and public health implications. From a practical standpoint, there is little difference between delay and prevention of a disease such as type 2 diabetes with variable and late onset. If an intervention could delay the onset of diabetes long enough such that a person ultimately dies without developing the disease, delay would be equivalent to prevention. Unfortunately, neither of the DPP interventions (4) was totally effective, i.e., neither reduced the diabetes incidence rate to zero. Thus they produced only partial delay or prevention. Even partial effectiveness in this regard, however, is beneficial and is cause for optimism that more effective interventions can be developed.
