Response to Comment on: Tam et al. Defining insulin resistance from hyperinsulinemic-euglycemic clamps. Diabetes Care 2012;35:1605-1610.

We appreciate the comments of Færch and Vistisen (1) and Davidson (2) regarding our study. As outlined in our study (3), the gold standard procedure for assessing insulin resistance in humans is the hyperinsulinemic-euglycemic clamp, which is widely used in research settings to test the effect of interventions on insulin sensitivity. Such interventions include weight loss or weight gain, pharmacological treatment, or lifestyle changes. Glucose disposal rate (GDR), the main outcome measure from the clamp, varies depending on the dose and duration of insulin infusion. The motivation for our study was driven by research participants asking investigators to explain the clinical relevance behind the clamp procedure and interpretation of the results. We developed cutoff values for defining insulin resistance (and insulin sensitivity) from clamp studies performed at the Pennington Biomedical Research Center between 2001 and 2011 using insulin infusion rates of 80 or 120 mU/m2/min (3). At both insulin doses, the distribution of GDR was bimodal, as demonstrated by significantly lower GDR in subjects with diabetes compared with subjects without diabetes (Table 2 in ref. 3). The cutoff point between insulin resistance and insulin sensitivity was calculated as the intersect between the two distributions in the model. Due to the relatively small sample size of participants with type 2 diabetes, we chose this approach rather than defining insulin resistance as the lowest quartile of GDR. Our study was not designed to diagnose diabetes in the population or guide treatment in existing patients with type 2 diabetes, for which there are already well-established criteria and guidelines (4).

We recognize that the clamp technique is time-consuming, labor-intensive, and expensive and consequently not always feasible. Therefore, our study also applied commonly measured clinic and blood variables in tree models to predict GDR obtained from the clamp. As noted by Færch and Vistisen (1), the context of these models needs to be carefully considered. Our study encompassed three tree models, which were designed to mimic clinical scenarios in which the most or the least data from the volunteers was available. Model 1 was our best-case scenario model, which included data from body composition and blood chemistry. Model 2 used only blood measures, and model 3 was a minimalist model using only age, sex, and BMI.

We agree with the point made in both responses (1,2) that insulin resistance is a heterogeneous disorder with complex etiology. Insulin resistance can be caused by defects in insulin action in the target tissues, excess local and circulating concentrations of inflammatory cytokines, impaired insulin receptor signaling, or even impairments in β-cell secretion (5). Unfortunately, without a measure of β-cell function in the current analysis, we cannot dismiss the potential influence of β-cell secretion on the results. In summary, our study provides cutoff values for defining insulin resistance from hyperinsulinemic-euglycemic clamps and as such extends the clamp from a research tool to providing a clinical meaningful message for participants in research studies.