Table 1

Some glucose clamp advantages and limitations in the study of subcutaneous injection profiles of insulin preparations

 Ability to study PK and PD together and understand relationships
 Ability to study PK without possible disturbance of subcutaneous absorption by adrenergic hormone release through hypoglycemia
 Ability to study PD without disturbance of glucose dynamics from counterregulatory hormone changes or concentration-driven changes in glucose disposal (glucose uptake and urinary glucose disposal)
 Ability to study PD without disturbance of hepatic autoregulation of glucose production
 The technique requires research skills, in particular if conducted in people with type 1 diabetes.
 Insulin given to correct or maintain prior blood glucose control will continue to have measurable action for around 1.5 h.
 In type 1 diabetes, management of change from prior insulin (usually intravenous) can be problematic and interferes with interpretation of PK until about 0.5 h after this can be discontinued (unless a specific insulin assay is available) and with PD for about 1.5 h (due to half-time of insulin action).
 In type 1 diabetes, when loss of action of the studied insulin occurs at end of clamp, insulin must be given or glucose levels left to rise above clamp levels, both options interfering with interpretation of PD and the former with PK (unless a specific insulin assay is available).
 In people without diabetes and people with type 2 diabetes, clamp conduct must be good to prevent stimulation of endogenous insulin secretion.
 C-peptide measurements can be used to correct for gradual changes in endogenous insulin secretion, but the long plasma half-life of C-peptide (20 min) means this is problematic for studies of faster-acting insulins and that it cannot correct for rapid changes due to poor clamp technique, while in type 2 diabetes, proinsulin (which cross-reacts in many C-peptide assays) poses additional interpretation problems.
 PD (glucose infusion rate) estimation has higher variance than insulin concentration (PK) and, in particular, in type 1 diabetes often needs smoothing algorithms for determination of profiles.
 In studies of long-acting insulins, the metabolic state is increasingly that of fasting/starvation (glucose infusion rate is low), so the metabolic state changes with time (notably hepatic glucose output); the clamp level, if constant, deviates from physiological levels, and the glucose infusion rate is then artifactually disturbed.
 In studies of long-acting insulins, even with larger doses, plasma insulin levels at the beginning and end of the studies are around the LLoQ of the assay and indeed often so in some individuals for long periods: statistical handling of such undetectable levels can create very large changes and uncertainties in the shape of the published profile.
 In people with type 1 diabetes and people with type 2 diabetes exposed to insulin, insulin antibody–bound insulin will interfere with interpretation of insulin assays; the interference can be erratic and result in high variance of measured levels.