Comment on Piccinini and Bergman The Measurement of Insulin Clearance. Diabetes Care 2020;43:2296–2302

We found interesting the review by Piccinini and Bergman (1) about the methods for assessing insulin clearance. However, some issues deserve comment. The authors claim that single-compartment description of C-peptide kinetics may provide limited accuracy. We agree that two-compartment description is typically more appropriate, but the single-compartment approach proved to be acceptable in glucose tolerance tests characterized by low dynamics. Indeed, the study by Vølund et al. (2) reported that the C-peptide single-compartment model was acceptable for insulin and C-peptide infusion experiments in dogs and that C-peptide clearance was similar in the studied dogs and in humans from previous studies. Another study in humans showed that, even for the intravenous glucose tolerance test (IVGTT), the C-peptide single-compartment model was the most appropriate choice in the case of individuals showing low dynamics patterns (3). In fact, in 4 out of 11 patients (i.e., ∼40%), the coefficient of variation of the estimated model parameters was excessively high, indicating that in those cases the two-compartment model was not resolvable from the IVGTT data. In contrast, the single-compartment model performed satisfactorily, as was then confirmed by Bergman’s group that applied the single-compartment C-peptide model to IVGTT data in healthy females (4).

Similar to the previous study (3), in a study of ours (5) (reference 30 in the review [1]), C-peptide single-compartment description was necessary to avoid extremely high coefficients of variation for the estimated model parameters. Thus, the single-compartment approach was the most appropriate strategy. Of note, in the review (1), use of piecewise polynomials for description of insulin secretion rate is listed as a further limitation of the model in reference 30. However, unlike several other models, that of reference 30 was validated against direct measurements from the hepatic catheterization technique, and validation was satisfactory.

The review (1) also fails to report a widely used method to assess whole-body insulin clearance, that is, the ratio between the area under the curve (AUC) of the insulin secretion rate and that of plasma insulin concentration. This method may be somewhat inaccurate if plasma insulin has not returned to the basal value at the end of the glucose tolerance test. However, this limitation should be overcome by proper adjustment, as suggested for instance in study of Shah et al. (6), i.e., by using the following formula: (AUC_ISR/AUC_INS) – V_INS × (INS_FIN – INS_INIT)/AUC_INS, where ISR is insulin secretion rate, INS is plasma insulin, FIN is the final time of the test, INIT is the initial time (typically, time of the fasting condition), and V_INS represents the insulin distribution volume, usually set to 0.14 L/kg.

Finally, we noticed that the study by Sherwin et al. (7) was mentioned (reference 15 in the review) but not really presented. To our knowledge, that study was the first accurate model for also assessing insulin clearance (among other variables); thus, a brief description of that study should be interesting, at least from a historical point of view.

• © 2021 by the American Diabetes Association