Abstract
Abstract
Background
The aim of this study is to compare the performance of three commonly used insulin assays with respect to the detection of exogenous human and porcine insulin added to human or rat plasma.
Methods
The DPC Immulite human insulin assay, the Mercodia rat insulin enzyme-linked immunosorbent assay and the Linco rat insulin radioimmunoassay were tested.
Results
The mean cross-reactivity of exogenous insulin ranged from 25% to 92%. The mean cross-reactivity of Actrapid in human plasma on the DPC Immulite was 56% and was independent of the endogenous insulin concentration.
Conclusions
The measurement of exogenous insulin varies according to the source of exogenous insulin, matrix and insulin assay.
Introduction
The hyperinsulinaemic euglycaemic clamp in combination with isotope dilution techniques is regarded as the gold standard for the measurement of insulin sensitivity in vivo 1 allowing for the determination of endogenous (i.e. hepatic) glucose production (EGP) and peripheral glucose utilization in the presence of various fixed plasma insulin concentrations. Performing an insulin clamp at plasma insulin concentrations in humans between 100 and 200 pmol/L gives an estimate of hepatic insulin sensitivity, whereas clamping at higher plasma insulin concentrations (>400 pmol/L) is used to assess peripheral insulin sensitivity. 2 The plasma concentration at which insulin is clamped is therefore essential to discriminate between various aspects of insulin sensitivity. To reach hyperinsulinaemia, exogenous insulin is infused using a continuous infusion technique. In human studies, synthetic regular insulin is used nowadays, whereas in rodent studies porcine insulin is used as well. Heald et al. 3 showed that most of the commercial insulin assays fail to detect recombinant insulin analogues. These considerations led us to study in more detail the performance of three insulin assays often used in clamp studies, 4 by comparing their ability to detect endogenous insulin, as well as the cross-reactivity of exogenous human and porcine insulin added to rat and human plasma in dilution series.
Methods
We tested one human assay (chemiluminescent immunoassay, Immulite 2000 system, Diagnostic Products Corporation, LA, USA) and two rat insulin assays (enzyme-linked immunosorbent assay [ELISA] Mercodia, 10-1124-01, Uppsala, Sweden and radioimmunoassay [RIA], Linco RI-13K, MA, USA). Insulin analogues used were recombinant regular human insulin (Actrapid 100 U/mL, Novo Nordisk, Copenhague, Denmark) and porcine insulin (Sigma–Aldrich, I5523, Zwijndrecht, The Netherlands). Dilutions of these insulin analogues were made with pooled human plasma using two different endogenous insulin concentrations or plasma from fasted rats. Actrapid (ready for use) was diluted with pooled human or rat plasma to give final concentrations of 722, 361, 180, 90 and 45 pmol/L. Porcine insulin (1.8 mg was dissolved in 20 mL of 0.01 mmol/L HCl containing 0.01 g/mL bovine serum albumin) was diluted with rat plasma to give final concentrations of 783, 392, 196, 98 and 49 pmol/L.
Results
The blank value of one human plasma was 131, 155 and 160 pmol/L when measured with the DPC Immulite, Mercodia ELISA and the Linco RIA, respectively. The blank value from the other when measured with the DPC Immulite was 20 pmol/L. Blank values of the rat plasma were 0, 175 and 201 pmol/L when measured with the DPC Immulite, Mercodia ELISA and the Linco RIA, respectively. In the dilution series, the assays gave a linear regression with a good correlation (range of r 2: 0.985 − 0.999). Mean cross-reactivity for the different assays are shown in Table 1. The cross-reactivity of human insulin with the DPC Immulite in human plasma was independent of the endogenous insulin concentration.
Mean cross-reactivities of insulin assays tested in three different serial dilutions
RIA = radioimmunoassay; ELISA = enzyme-linked immunosorbent assay
Conclusions
The results of this study show that all assays tested were linear. The basal insulin in human plasma does not vary much between the assays tested, whereas the human DPC Immulite system did not detect endogenous rat insulin. The mean cross-reactivity of exogenous insulin in human and rat plasma was highly variable: between 56% and 86% in human and between 25% and 92% in rat, and dependent upon the type of exogenous insulin (Actrapid or porcine) as well as the insulin assay. Furthermore, the mean cross-reactivity of Actrapid in human plasma as measured using the DPC Immulite system was 56%, irrespective of the endogenous insulin concentration. These data are in agreement with the observations reported in recent literature. 3,5,6
One possible explanation for the high variance in mean recoveries may be species differences in the amino acid sequence of the insulin forms. Furthermore, the Mercodia ELISA and DPC Immulite are solid phase two-site immunoassays, whereas the Linco RIA is a competitive immunoassay. Finally, the different assays use antibodies with different properties with respect to affinity, specificity and cross-reactivity.
Our results have implications for the interpretation of hyperinsulinaemic clamp studies in rats and humans, especially where endogenous insulin is not suppressed, because in this setting the plasma samples obtained during a clamp contain both endogenous and exogenous insulin. Assuming that the cross-reactivity of endogenous insulin is 100% using a species-specific assay, and knowing that the cross-reactivity of the exogenous insulin may range from 25% to 92%, the actual plasma insulin concentration remains unknown if endogenous insulin is not suppressed completely and if cross-reactivity of exogenous insulin has not been carefully studied.
Although the immunoreactivity of exogenous insulin is <100%, this does not necessarily mean that the same applies for the bioactivity of this insulin. Thus, correcting for analytical cross-reactivity does not per se give biologically relevant insulin concentrations.
What is the practical consequence of the present findings? It is difficult to determine plasma insulin concentrations during exogenous hyperinsulinaemia, especially when endogenous insulin is not suppressed. However, in the absence of reliable absolute insulin concentrations, the biological effects of insulin (suppression of EGP, or facilitation of glucose utilization in peripheral organs) under different experimental conditions may still be interpreted in relation to changes in plasma insulin as long as the same assay, the same exogenous insulin and the same matrix is used. Moreover, as the curve of plasma insulin versus insulin sensitivity has a sigmoid shape, there is a broad range of low plasma insulin concentrations to determine hepatic insulin sensitivity and a broad range of high plasma concentrations to determine peripheral insulin sensitivity.
In summary, the measurement of plasma insulin during a hyperinsulinaemic clamp varies according to the source of exogenous insulin, the matrix and the insulin assay. Thus, absolute insulin concentrations measured during a clamp should be interpreted carefully. When comparing different studies reporting insulin concentrations during a hyperinsulinaemic clamp, these three aspects (source of the exogenous insulin, the matrix and the assay) used for the determination of the insulin should be taken into account.
Footnotes
ACKNOWLEDGEMENT
The authors wish to acknowledge Els Johannesma-Brian from the Laboratory of Endocrinology for performing the assays.