I agree with the fundamental point made by Dr. Scheen (1) in this issue of Diabetes Care. It is unclear whether the effect of metformin to reduce the incidence of diabetes during the Diabetes Prevention Program (DPP) (2) was true prevention or simply a masking of diabetes. The posttrial washout period was very short for testing that distinction. The data cited by Dr. Scheen regarding a very rapid effect of metformin to increase insulin sensitivity suggests that dissipation of any such effect in the DPP could have been rapid as well. Thus, acute effects of metformin on insulin sensitivity could have been gone 2 weeks after the drug was stopped. It seems unlikely that we will ever know for sure because direct measures of insulin sensitivity were not made in the DPP. Additionally, we will never know whether glucose levels had stabilized or were still rising 2 weeks after metformin was stopped. The Study to Prevent (STOP)-NIDDM trial (3) included a more convincing washout period of 3 months before retesting for diabetes. During the washout, the jump in new cases of diabetes in the former acarbose group (twice as many as in the former placebo group) was even more striking than the analogous jump in the former metformin group in the DPP (49% more cases than in the placebo group). Whether this greater jump in the STOP-NIDDM trial reflects the more meaningful washout period used in that trial or a greater component of masking rather than prevention of diabetes compared with metformin can’t be determined from the published data.
We have provided strong evidence from the Troglitazone in the Prevention of Diabetes (TRIPOD) study (4), conducted in Hispanic-American women, that lowering endogenous insulin requirements can slow or stop the loss of β-cell function, which leads to type 2 diabetes. If this effect occurs in other ethnic groups (still a big “if”), then any effect of metformin or acarbose to lower glucose levels or enhance insulin sensitivity could have had an analogous, albeit probably weaker, effect in the DPP and STOP-NIDDM trials. Given the known effects of metformin and acarbose on carbohydrate metabolism, I suspect that portions of the reductions in diabetes rates observed in the metformin arm of the DPP and in the STOP-NIDDM trial were due to β-cell protection. The posttrial washouts in the two studies demonstrate that some of the apparent protection from diabetes was simply masking caused by acute glucose-lowering effects of the drugs.
We will never know precisely how much prevention and how much masking of diabetes occurred in the DPP and the STOP-NIDDM trials. Nonetheless, discourse on this issue can be helpful. It can focus diabetes prevention research on the real disease of type 2 diabetes—progressive β-cell failure—rather than on the glucose thresholds that currently define diabetes. Those thresholds are clearly important for assessing the risk of microvascular and neuropathic complications of diabetes. However, they are not very useful for tracking the underlying β-cell disease. That disease progresses for a long time before patients cross the glucose threshold to diabetes, and it continues after they have crossed that threshold. When the discussion turns to prevention (or even to initial treatment) of type 2 diabetes, the goal should become stability of glucose levels and β-cell function.
T.A.B. has served on advisory panels for Takeda N.A., RW Johnson, and Sankyo; has been on the speakers’ bureau for Takeda N.A.; has been a consultant for Takeda N.A., Novo Nordisk, RW Johnson, Metabolex, Sankyo, and Amylin; and has received grant support from Takeda N.A. and Bristol Meyers Squibb.