Bruce H. Frank

Abnormal Patterns of Insulin Secretion in Non-Insulin-Dependent Diabetes Mellitus

To determine whether non-insulin-dependent diabetes is associated with specific alterations in the pattern of insulin secretion, we studied 16 patients with untreated diabetes and 14 matched controls. The rates of insulin secretion were calculated from measurements of peripheral C-peptide in blood samples taken at 15- to 20-minute intervals during a 24-hour period in which the subjects ate three mixed meals. Incremental responses of insulin secretion to meals were significantly lower in the diabetic patients (P less than 0.005), and the increases and decreases in insulin secretion after meals were more sluggish. These disruptions in secretory response were more marked after dinner than after breakfast, and a clear secretory response to dinner often could not be identified. Both the control and diabetic subjects secreted insulin in a series of discrete pulses. In the controls, a total of seven to eight pulses were identified in the period from 9 a.m. to 11 p.m., including the three post-meal periods (an average frequency of one pulse per 105 to 120 minutes), and two to four pulses were identified in the remaining 10 hours. The number of pulses in the patients and controls did not differ significantly. However, in the patients, the pulses after meals had a smaller amplitude (P less than 0.03) and were less frequently concomitant with a glucose pulse (54.7 +/- 4.9 vs. 82.2 +/- 5.0, P less than 0.001). Pulses also appeared less regularly in the patients. During glucose clamping to produce hyperglycemia (glucose level, 16.7 mmol per liter [300 mg per deciliter]), the diabetic subjects secreted, on the average, 70 percent less insulin than matched controls (P less than 0.001). These data suggest that profound alterations in the amount and temporal organization of stimulated insulin secretion may be important in the pathophysiology of beta-cell dysfunction in diabetes.

Quantitative study of insulin secretion and clearance in normal and obese subjects.

The secretion and hepatic extraction of insulin were compared in 14 normal volunteers and 15 obese subjects using a previously validated mathematical model of insulin secretion and rate constants for C-peptide derived from analysis of individual decay curves after intravenous bolus injections of biosynthetic human C-peptide. Insulin secretion rates were substantially higher than normal in the obese subjects after an overnight fast (86.7 +/- 7.1 vs. 50.9 +/- 4.8 pmol/m2 per min, P less than 0.001, mean +/- SEM), over a 24-h period on a mixed diet (279.6 +/- 24.2 vs. 145.8 +/- 8.8 nmol/m2 per 24 h, P less than 0.001), and during a hyperglycemic intravenous glucose infusion (102.2 +/- 10.8 vs. 57.2 +/- 2.8 nmol/m2 per 180 min, P less than 0.001). Linear regression analysis revealed a highly significant relationship between insulin secretion and body mass index. Basal hepatic insulin extraction was not significantly different in the normal and obese subjects (53.1 +/- 3.8 vs. 51.6 +/- 4.0%). In the normal subjects, fasting insulin did not correlate with basal hepatic insulin extraction, but a significant negative correlation between fasting insulin and hepatic insulin extraction was seen in obesity (r = -0.63, P less than 0.02). This finding reflected a higher extraction in the six obese subjects with fasting insulin levels within the range of the normal subjects than in the nine subjects with elevated fasting insulin concentrations (61 +/- 3 vs. 45 +/- 6%, P less than 0.05). During the hyperglycemic clamp, the insulin secretion rate increased to an average maximum of 6.2-fold over baseline in the normal subjects and 5.8-fold in the obese subjects. Over the same time, the peripheral insulin concentration increased 14.1-fold over baseline in the normals and 16.6-fold over baseline in the obese, indicating a reduction in the clearance of endogenously secreted insulin. Although the fall in insulin clearance tended to be greater in the obese subjects, the differences between the two groups were not statistically significant. Thus, under basal, fasting conditions and during ingestion of a mixed diet, the hyperinsulinemia of obesity results predominantly from increased insulin secretion. In patients with more marked basal hyperinsulinemia and during intense stimulation of insulin secretion, a reduction in insulin clearance may contribute to the greater increase in peripheral insulin concentrations that are characteristic of the obese state.+

Use of biosynthetic human C-peptide in the measurement of insulin secretion rates in normal volunteers and type I diabetic patients.

We undertook this study to examine the accuracy of plasma C-peptide as a marker of insulin secretion. The peripheral kinetics of biosynthetic human C-peptide (BHCP) were studied in 10 normal volunteers and 7 insulin-dependent diabetic patients. Each subject received intravenous bolus injections of BHCP as well as constant and variable rate infusions. After intravenous bolus injections the metabolic clearance rate of BHCP (3.8 +/- 0.1 ml/kg per min, mean +/- SEM) was not significantly different from the value obtained during its constant intravenous infusion (3.9 +/- 0.1 ml/kg per min). The metabolic clearance rate of C-peptide measured during steady state intravenous infusions was constant over a wide concentration range. During experiments in which BHCP was infused at a variable rate, the peripheral concentration of C-peptide did not change in proportion to the infusion rate. Thus, the infusion rate of BHCP could not be calculated accurately as the product of the C-peptide concentration and metabolic clearance rate. However, the non-steady infusion rate of BHCP could be accurately calculated from peripheral C-peptide concentrations using a two-compartment mathematical model when model parameters were derived from the C-peptide decay curve in each subject. Application of this model to predict constant infusions of C-peptide from peripheral C-peptide concentrations resulted in model generated estimates of the C-peptide infusion rate that were 101.5 +/- 3.4% and 100.4 +/- 2.8% of low and high dose rates, respectively. Estimates of the total quantity of C-peptide infused at a variable rate over 240 min based on the two-compartment model represented 104.6 +/- 2.4% of the amount actually infused. Application of this approach to clinical studies will allow the secretion rate of insulin to be estimated with considerable accuracy. The insulin secretion rate in normal subjects after an overnight fast was 89.1 pmol/min, which corresponds with a basal 24-h secretion of 18.6 U.

Report of the American Diabetes Association's Task Force on Standardization of the Insulin Assay

Recent large-scale epidemiological studies demonstrate that blood concentrations of immunoreactive insulin predict the development of NIDDM and IDDM and are associated with the risk of several degenerative diseases, such as coronary and peripheral vessel atherosclerosis, hypertension, and dyslipidemia. The reliability of these measurements is dependent on a biological assay that has not been well standardized between laboratories. Recognizing this, the American Diabetes Association organized a task force to assess comparability of blood insulin measurements between laboratories and to suggest techniques to improve comparability. The task force found that identical serum and plasma samples measured in different laboratories produced widely disparate values that were unacceptable for population comparisons. Use of a single reference standard did little to improve comparability. Assay characteristics such as linearity, recovery, accuracy, and cross-reactivity to proinsulin and its primary conversion intermediates varied among the laboratories, and they did not readily explain differences in the measurements made from assay to assay. Use of the same assay kit in different laboratories did not always ensure comparable measurements. Linear regression of assay results from one laboratory to an arbitrarily chosen reference assay greatly improved comparability and demonstrated the potential value in comparing each assay to a reference method. The task force report defines acceptable assay characteristics and proposes a three-step process of insulin assay proficiency and comparability. A central reference assay and ongoing sample exchange will be needed to allow reliable comparisons of insulin measurements made in different laboratories. Rigorous quality control and continuous quality improvement are needed to maintain reliability of the insulin measurement.

Biosynthetic human proinsulin. Review of chemistry, in vitro and in vivo receptor binding, animal and human pharmacology studies, and clinical trial experience.

Objective To describe the rationale for the preclinical and clinical developmental course of human proinsulin (HPI), the second product after human insulin for the treatment of diabetes mellitus to be manufactured by DNA technology. Research Design and Methods The relevant and available published and unpublished preclinical and clinical information generated on pork proinsulin and human proinsulin has been integrated to demonstrate how certain clinically attractive features of pork proinsulin (a soluble intermediate-acting and possibly hepatospecific insulin agonist) led to the development of HPI. Results Clinical pharmacology studies demonstrated that HPI was definitely, although marginally, hepatospecific. More striking was the finding that the intrasubject/patient coefficient of variation of response to HPI was significantly less than that observed with NPH insulin. However, the fact that unique efficacy in controlled multicenter studies was not demonstrated suggested that these pharmacological features were not translated into clinical benefit. In one multicenter new patient study there were six myocardial infarctions, including two deaths, in patients treated for ≥1 yr with HPI and none in the control group. Conclusions To obtain an independent review of the risks and benefits of HPI, in February 1988, Lilly convened a consultant group that examined all relevant information on HPI available. These experts shared our concerns about the safety of HPI in light of the failure to demonstrate unique efficacy. Accordingly, clinical trials with HPI were suspended in February 1988. Experience with HPI demonstrates the challenge associated with the development of new drugs in general and insulin agonists in particular.

The Limitations to and Valid Use of C-Peptide as a Marker of the Secretion of Insulin

The accuracy with which the secretion rate of insulin can be calculated from peripheral concentrations of C-peptide was investigated in conscious mongrel dogs. Biosynthetic human C-peptide and insulin were infused intraportally and their concentrations measured in the femoral artery. During steady-state infusions of C-peptide, the peripheral concentration changed in proportion to the infusion rate and the metabolic clearance rate (5.2 ± 0.3 ml/kg/min) remained constant over a wide range of plasma concentrations. Application of a two-compartment mathematical model, in which the model parameters were estimated from analysis of C-peptide decay curves after intravenous bolus injections, allowed the intraportal infusion rate of C-peptide to be derived from peripheral C-peptide concentrations, even under non-steady-state conditions. Estimates of the intraportal infusion rate based on this model were 102.4 ± 2.6% of the actual infusion rate as it was increasing and 102.3 ± 5.5% of this rate as it was falling. The peripheral C-peptide : insulin molar ratio was influenced by the rate at which equimolar intraportal infusions of C-peptide and insulin were changed. The baseline C-peptide : insulin molar ratio (4.1 ± 0.9) increased to peak values of 8.2 ± 0.6,10.3 ± 2.0, and 14.9 ±1.3 when the infusion rate was increased and then decreased rapidly. Peak values of only 5.7 ±1.2 were found if the intraportal infusion rate was changed slowly. In conclusion: (1) under steady-state conditions the secretion rate of insulin can be calculated as the product of the peripheral concentration of C-peptide and its MCR; (2) under non-steady-state conditions, however, application of more complex mathematical models, such as the two-compartment model used in the present study, allows insulin secretion rates to be accurately calculated at discrete time points; and (3) under non-steady-state conditions the C-peptide:insulin molar ratio is influenced not only by the extent of hepatic insulin extraction but by other factors, including the rate of change of insulin secretion and the clearance rate of C-peptide. Changes in this ratio should therefore not be assumed to reflect changes in hepatic insulin extraction.

Chemical, Physical, and Biologic Properties of Biosynthetic Human Insulin

Human insulin derived via recombinant DNA technology was tested extensively by a complex battery of analytic procedures. This product, which is designated biosynthetic human insulin, was found to be chemically, physically, and immunologically equivalent to pancreatic human insulin and biologically equivalent to both pancreatic human insulin and purified pork insulin.

Insulin Secretion and Clearance: Comparison After Oral and Intravenous Glucose

Insulin secretion and clearance in response to the administration of oral and intravenous glucose was investigated in nine normal men. C-peptide metabolic kinetics were calculated by analysis of individual C-peptide decay curves after the bolus injection of biosynthetic human C-peptide. Glucose was administered to the subjects on three occasions: as a 75-g oral dose, a 75-g i.v. infusion, and an intravenous glucose infusion at a variable rate adjusted to mimic the peripheral glucose levels obtained after the oral glucose load (matching experiment). Glucose, insulin, and C-peptide concentrations were measured for the subsequent 5 h. The glucose level after the oral glucose load (115.9 ± 2.6 mg/dl, mean ± SE) closely approximated that after the matching experiment (120.5 ± 2.5 mg/dl) but was significantly lower than after 75 g i.v. glucose (127.7 ± 3.4 mg/dl, P < .05). Analysis of the areas under the peripheral concentration curves (60-360 min) showed that the responses of both insulin (52.7 ± 5.6 and 46.5 ± 4.5 pmol · ml−1 300 min1) and C-peptide (252.7 ± 27.5 and 267.0 ± 21.6 pmol · ml−1 · 300 min1) were not significantly different after the oral and 75-g i.v. glucose studies, respectively, whereas in the matching experiment, both the insulin (26.1 ± 3.9 pmol · ml−1 · 300 min−1) and C-peptide (178.0 ± 18.9 pmol ml−1 300 min−1) responses were lower (P < .05) than in the other two studies. Insulin secretory rates were derived from peripheral C-peptide concentrations with an open two-compartment model and individually derived model parameters. The basal insulin secretion rate was 86.8 ± 2.9 pmol/min. The insulin secretory response over the 300 min was 66.2 ± 4.8 nmol after oral glucose. This was similar to that after 75 g i.v. glucose (72.4 ± 4 . 1 nmol), whereas that secreted in response to the matching experiment was lower (47.6 ± 4.1 nmol, P ± .05). As a measure of the clearance of endogenous insulin, the ratio between the area under the insulin secretory curve and the area under the peripheral insulin concentration curve was calculated. This ratio was similar (1906 ± 149 ml/min) during the baseline period and the matching glucose infusion (2042 ± 245 ml/min) but was significantly lower after oral glucose (1330 ±112 ml/min, P < .05). The incretin effect calculated based on the insulin secretion rate (25 ± 9.2%) appeared to be less than if the calculations were based on peripheral insulin levels. These data demonstrate that equivalent doses of glucose administered orally and intravenously elicit an equivalent insulin secretory response. However, when the arterialized plasma glucose curve after 75 g oral glucose is matched by an intravenous glucose infusion, only 35.6 ± 2.9 g glucose was infused, and the intravenous glucose resulted in a lower secretory response. Furthermore, after oral administration of 75 g glucose a significant reduction in insulin clearance resulted. These data provide evidence that the hyperinsulinemia seen after oral glucose is due both to enhanced insulin secretion and diminished insulin clearance.

Biochemical and clinical implications of proinsulin conversion intermediates.

Since a complete map of insulin-related peptides in humans requires consideration of proinsulin, Arg32/Glu33-split proinsulin, Arg65/Gly66-split proinsulin, des-Arg31,Arg32-proinsulin, des-Lys64, Arg65-proinsulin, and insulin, we applied high performance liquid chromatography coupled with radioimmunoassay to investigate the formation of proinsulin conversion intermediates in vitro and in vivo. Kinetic analysis of proinsulin processing by a mixture of trypsin and carboxypeptidase B (to stimulate in vivo processes) revealed (a) a rapid decline in proinsulin concommitant with formation of conversion intermediates, (b) formation of des-Arg31, Arg32-proinsulin and des-Lys64,Arg65-proinsulin in the ratio 3.3:1 at steady state, and (c) complete conversion of the precursor to insulin during extended incubation. Studies on normal human pancreas identified a similar ratio of des-Arg31,Arg32-proinsulin to des-Lys64,Arg65-proinsulin (approximately 3:1), whereas two insulinomas contained sizable amounts of des-Arg31,Arg32-proinsulin, but barely detectable amounts of des-Lys64,Arg65-proinsulin. None of the tissues contained measurable quantities of Arg32/Glu33- or Arg65/Gly66-split proinsulin. Analysis of plasma from three diabetic subjects managed by the intravenous infusion of human proinsulin revealed less than 1% processing of the circulating precursor to conversion intermediates and no processing of the precursor to human insulin. Nevertheless, analysis of plasma from the same subjects managed by the subcutaneous infusion of proinsulin revealed 4-11% processing of the precursor to intermediates that had the properties of des-Arg31,Arg32-proinsulin and Arg65/Gly66-split proinsulin. We conclude that (a) processing of proinsulin to insulin in vivo as in vitro likely occurs by preferential cleavage at the Arg32-Glu33 peptide bond in proinsulin, (b) proinsulin is inefficiently processed in the vascular compartment, and (c) subcutaneous administration of the precursor can result in the formation of conversion intermediates with the potential for contributing to biological activity.

Research, Development, Production, and Safety of Biosynthetic Human Insulin

This paper provides some historical aspects on the research and development of Humulin® (rDNA origin), the first human health-care product derived from rDNA technology more than a decade ago. Also referred to as biosynthetic human insulin, Humulin® is currently produced via the human proinsulin route, using an Escherichia coli fermentation process. The authenticity, high purity, and safety of BHI has been investigated and verified by a complex battery of analytical and physicochemical methods. The daily treatment of more than two million diabetic patients worldwide with this rDNA human insulin not only demonstrates the value of rDNA technology in providing an important medical product, it is assurance that diabetic patients will have unlimited supplies of this vital hormone as well as potential analogue refinements.