Jump to comment:

• Do the Conclusions Fit the Results?
  John A Kao
  Published on: 23 November 2015

• Published on: 23 November 2015

  Do the Conclusions Fit the Results?

  • John A Kao, Interventional Cardiology

  Dear Editor,

  In their article published last month, Stakos et al. (Heart 2005) found that both glipizide and metformin, when administered to non-diabetic African Americans with insulin resistance, lead to a significant increase in pulse wave velocity (PWV) when compared to placebo, and conclude that, "…long term administration of oral antidiabetic glipizide or metformin to normoglycaemic African Americans with insulin re...

  Show More

  Dear Editor,

  In their article published last month, Stakos et al. (Heart 2005) found that both glipizide and metformin, when administered to non-diabetic African Americans with insulin resistance, lead to a significant increase in pulse wave velocity (PWV) when compared to placebo, and conclude that, "…long term administration of oral antidiabetic glipizide or metformin to normoglycaemic African Americans with insulin resistance may decrease the elastic properties of the aorta." Stakos et al. go on to state that, "On the basis of these observations, the use of glipizide, metformin, or both for the prevention of cardiovascular complication in normoglycaemic insulin resistant African Americans needs to be critically evaluated." On the surface, these observations and claims appear to be quite concerning. Indeed, PWV has been correlated with aortic distensibility, and an increase in PWV is associated with increased cardiovascular mortality and adverse events. However, does the article by Stakos et al. really demonstrate this to be true? There are several points which I think need to be addressed before the validity of the authors conclusions can be accepted.

  First, the implication from this study is that the increase in PWV in normotensive, normolipidemic African Americans is associated with an increase in cardiovascular mortality. What the authors fail to mention however, is that PWV has been found to be associated with an increase in cardiovascular mortality and events only in patients with end stage renal disease and/or hypertension. Studies in normotensive patients have failed to show a correlation between PWV and outcomes. It is inappropriate to extrapolate findings from one patient population (i.e. hypertension/ESRD) to another (normotensive African American women – see below). In addition, in studies performed on hypertensive patients and/or patients with ESRD, PWV has always been correlated with an increase in LVMI. Both PWV and LVMI in these populations have been identified as independent predictors of cardiovascular complications. Indeed, LVMI has been found to be directly associated with PWV as PWV is a measure of arterial stiffness and increased afterload. In the study by Stakos et al., there is a highly significant decrease in LVMI in all groups regardless of the change in PWV. The authors fail to reconcile this seeming contradiction.

  Second, this is a study that is represented as a "randomized, placebo controlled trial…" However, if we examine the distribution of the study subjects we see marked disparities between groups – glipizide with an n=25, metformin with an n=57, placebo with an n=97. How is it possible for a truly randomized trial to have such differences in subject number? Did the authors want a 1:2:4 randomization strategy? If so, why did they not describe this and the reason why in the Methods section? The authors state that since "…the shape of the active drug tablets (glipizide and metformin) is different…. We created an identical placebo tablet. Thus, the total number of placebo was at least two to three times larger for any active treatment." Also they state that at the beginning of the study, metformin was not approved by the FDA, and therefore "the metformin group was larger than the glipizide group." This implies that two separate randomizations occurred during the study, one for glipizide, and another for metformin. If so, one cannot combine the two control groups into one and compare it to each of the active groups. Rather, one should compare each active group to their respective control group. The authors do not give a reason for why this was not done.

  Third, the authors show us baseline results and give us the number of subjects in each group, however they go on to say "a small number (about 13%) of subjects dropped out." The authors fail to account for these missing subjects in their paper – were the number of dropouts evenly distributed between groups? Was one group more affected by the dropouts than another? Nowhere is this described in the body of the paper or in the associated figures and tables. This is a major lapse on the part of the authors. "About" 13% of 181 is equal to 24 patients – with group sizes of 25, 59 and 97, it is clear where and how these dropouts are distributed between the three groups will impact greatly on the interpretation of results. Should the majority of dropouts be concentrated in the active groups especially, the potential introduction of alpha and beta error becomes enormous.

  Also, in their Methods section, the authors state that "African Americans usually have high concentrations of high density lipoprotein (HDL) cholesterol and low concentrations of triglyceride. For these reasons none of the participants were considered to have metabolic syndrome." One cannot take generalizations about an entire population and apply it to a specific subgroup and expect it to apply. Indeed, according to their own findings (see Table 1 Baseline Characteristics), a significant number of their patients meet the ATP III guidelines for the Metabolic Syndrome – having >/= 3 of the following:

  1) Waist size > 40 inches (102 cm) for men or > 35 inches (88 cm) for women
  2) Fasting Triglyceride >/= 150 mg/dl (1.69 mmol/L) 3) HDL chol <40 mg/dl (1.04 mmol/L) men or <50 mg/dl (1.29 mmol/L)
  women 4) BP >/= 130/85
  5) Fasting glucose >/= 110 mg/dl (6.1 mmol/L)

  Examination of Table 1 shows another important fact - the vast majority of patients in this study are women – 89% in the Glipizide group, 74% in the Metformin group, and 76% in the Placebo group. This is not mentioned or discussed in the paper. It is extremely dangerous to make generalized statements regarding entire populations (i.e. all African Americans with insulin resistance) when the study in question is relevant to only a subgroup of that population (African American women – many with the Metabolic Syndrome).

  Finally, one must weigh in all the evidence available in the literature, and not take just one piece of information. Many studies have been published demonstrating the potential cardioprotective effects of Metformin – normalizing endothelial function, decreasing insulin resistance, decreasing PAI-1 and CRP, increasing HDL – as well as evidence supporting a mortality benefit of Metformin in patients with coronary artery disease and diabetes. Should all previous literature be tossed aside due to the findings of one small and methodologically questionable study? Even if the findings of Stakos et al are found to be true, they are true of only a subgroup of a subgroup of a subgroup of people – African American women with insulin resistance and no diabetes.

  John Kao, MD

  Assistant Professor of Medicine University of Illinois at Chicago

  Director, Cardiac Cath Lab Jesse Brown VA Hospital

  Chief, Section of Cardiology Jesse Brown VA Hospital

  References

  1. Guerin et al. "Impact of Aortic Stiffness Attenuation on Survival of Patients in End-Stage Renal Failure." Circulation. 2001;103:987-992

  2. Laurent et al. "Aortic Stiffness Is an Independent Predictor of All-Cause and Cardiovascular Mortality in Hypertensive Patients." Hypertension. 2001;37:1236-1241

  3. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III), JAMA 285 (2001), pp. 2486–2497.

  4. Kao J, Tobis J, McClellan RL, Heaton MR, Davis BR, Holmes DR Jr., Currier JW. METFORMIN TREATMENT IS ASSOCIATED WITH DECREASED CLINICAL EVENTS IN DIABETIC PATIENTS UNDERGOING PERCUTANEOUS INTERVENTION: A PRESTO SUBSTUDY. American Journal of Cardiology 2004; 93: 1347-1350.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top