A comparison of twice-daily exenatide and biphasic insulin aspart in patients with type 2 diabetes who were suboptimally controlled with sulfonylurea and metformin: a non-inferiority study.

AIMS/HYPOTHESIS: The aim of this 52-week, open-label, non-inferiority trial was to compare the safety and efficacy of exenatide (an incretin mimetic) with that of biphasic insulin aspart. MATERIALS AND METHODS: Patients on metformin and a sulfonylurea were randomised to exenatide (n = 253; 5 microg twice daily for 4 weeks, 10 microg thereafter) or biphasic insulin aspart (n = 248; twice-daily doses titrated for optimal glucose control), while continuing with metformin and sulfonylurea treatment. RESULTS: Glycaemic control achieved with exenatide was non-inferior to that achieved with biphasic insulin aspart (mean+/-SEM, HbA(1c) change: exenatide -1.04 +/- 0.07%, biphasic insulin aspart -0.89 +/- 0.06%; difference -0.15 [95% CI -0.32 to 0.01]%). Exenatide-treated patients lost weight, while patients treated with biphasic insulin aspart gained weight [between-group difference -5.4 (95% CI -5.9 to -5.0) kg]. Both treatments reduced fasting serum glucose (exenatide -1.8 +/- 0.2 mmol/l, p < 0.001; biphasic insulin aspart -1.7 +/- 0.2 mmol/l, p < 0.001). Greater reductions in postprandial glucose excursions following morning (p < 0.001), midday (p = 0.002) and evening meals (p < 0.001) were observed with exenatide. The withdrawal rate was 21.3% (54/253) for exenatide and 10.1% (25/248) for biphasic insulin aspart. Nausea (33% incidence, 3.5% discontinuation) was the most common adverse event observed with exenatide. CONCLUSIONS/INTERPRETATION: Exenatide treatment resulted in HbA(1c) reduction similar to biphasic insulin aspart and provided better postprandial glycaemic control, making it a potential alternative for the treatment of type 2 diabetes. Treatment with biphasic insulin aspart was associated with weight gain and lower risk of adverse gastrointestinal events. Although the availability of glucose-lowering agents associated with weight reduction may be considered a therapeutic advance, the long-term implications of progressive weight reduction observed with exenatide have yet to be defined.

Long-term effects of exenatide therapy over 82 weeks on glycaemic control and weight in over-weight metformin-treated patients with type 2 diabetes mellitus.

AIM: The ability of the incretin mimetic exenatide to improve glycaemic control and reduce body weight was assessed over 82 weeks in patients with type 2 diabetes failing to achieve glycaemic control with maximally effective doses of metformin. METHODS: In this interim 82-week analysis, 150 (total cohort) of an eligible population of 183 patients opted to continue exenatide treatment in an uncontrolled open-label extension of a 30-week double-blind, placebo-controlled trial. Of these, 92 patients (completer cohort) achieved 82 weeks of exenatide therapy. Patients continued metformin throughout the study. RESULTS: At the end of the placebo-controlled trial, exenatide resulted in an haemoglobin A1c (HbA1c) reduction from baseline of -1.0 +/- 0.1% (mean +/- SE) (exenatide treatment arms), with durable HbA1c reductions after 82 weeks of -1.3 +/- 0.1%. The percent of patients who achieved HbA1c < or = 7% at weeks 30 and 82 was 46 and 59% respectively. After 30 weeks, exenatide caused a reduction in weight from baseline of -3.0 +/- 0.6 kg, with a progressive reduction in weight of -5.3 +/- 0.8 kg after 82 weeks. In addition, exenatide treatment produced clinically significant improvements in cardiovascular risk factors after 82 weeks. The most frequent adverse event after 30 and 82 weeks of exenatide was nausea, which was generally of mild-or-moderate intensity. It decreased in incidence after initiation in the controlled trial and the uncontrolled open-label extension. Hypoglycaemia was rare, with no severe events. CONCLUSION: Exenatide was generally well tolerated, producing a durable reduction in HbA1c and a progressive reduction in weight over 82 weeks in patients with type 2 diabetes failing to achieve glycaemic control with metformin.

Therapy focused on lowering postprandial glucose, not fasting glucose, may be superior for lowering HbA1c. IOEZ Study Group.

OBJECTIVE: To compare the overall efficacy of combination therapies focused on fasting or postprandial blood glucose in patients with type 2 diabetes not adequately controlled with oral sulfonylurea agents alone. RESEARCH DESIGN AND METHODS: A total of 135 patients were randomly assigned for 3 months to 1 of 3 combination regimens with glyburide (G) that addressed postprandial blood glucose with insulin lispro (L+G), premeal blood glucose with metformin (M+G), or fasting blood glucose (FBG) with bedtime NPH insulin (NPH+G). RESULTS: At end point, HbA1c was significantly lower with all therapies (P = 0.001) and was significantly lower for L+G (7.68+/-0.88%) compared with either NPH+G (8.51+/-1.38%, P = 0.003) or M+G (8.31+/-1.31%, P = 0.025). FBG at end point was significantly lower for NPH+G (8.49+/-2.36 mmol/l) compared with either L+G (10.57+/-1.97 mmol/l, P = 0.001) or M+G (9.69+/-2.89 mmol/l, P = 0.029). The mean 2-h postprandial glucose after a test meal was significantly lower for L+G (10.87+/-2.88 mmol/l) versus NPH+G (12.21+/-3.12 mmol/, P = 0.052) or versus M+G (12.72+/-3.26 mmol/l, P = 0.009). The overall rate of hypoglycemia (episodes per 30 days) was low and not statistically significant between groups (P = 0.156). CONCLUSIONS: Adding a second antihyperglycemic agent, regardless of its timing of action, lowers HbA1c and glucose values. However, when insulin lispro was used to focus on postprandial blood glucose, there was a greater impact on overall metabolic control. These data support the importance of lowering postprandial blood glucose to optimize overall glycemic control and thus improve long-term outcomes.

A randomized placebo-controlled trial of repaglinide in the treatment of type 2 diabetes.

OBJECTIVE: The objective of the study was to assess the efficacy and safety of repaglinide compared with placebo in the treatment of patients with type 2 diabetes. RESEARCH DESIGN AND METHODS: This was a phase II multicenter, double-blind, placebo-controlled, randomized, dose-adjustment and maintenance trial. After screening and a 2-week washout period, 99 patients were randomized to receive either repaglinide (n = 66) or placebo (n = 33). Patients underwent 6 weeks of dose adjustment followed by 12 weeks of dose maintenance. Fasting and stimulated glycosylated hemoglobin (HbA1c), plasma glucose, insulin, and C-peptide were measured at predetermined intervals. Adverse events and hypoglycemic episodes were recorded. RESULTS: From baseline to last visit, mean HbA1c decreased from 8.5 to 7.8% in patients treated with repaglinide and increased from 8.1 to 9.3% in patients receiving placebo, with a statistically significant difference of - 1.7% (P < 0.0001) between treatment groups at the last visit. Mean fasting plasma glucose and postprandial glucose increased in patients receiving placebo and decreased in patients treated with repaglinide, with statistically significant (P < 0.01) differences between groups at the last visit. Concentrations of fasting and postprandial insulin and C-peptide were lower at the last visit compared with baseline for patients treated with placebo and higher for patients treated with repaglinide, and the differences between groups were statistically significant (P < 0.05). Overall, repaglinide was well tolerated. CONCLUSIONS: This study demonstrated that repaglinide was safe and efficacious in lowering blood glucose concentrations. In addition to overall improvement in glycemic control noted with repaglinide in both sulfonylurea-treated patients and oral hypoglycemic agent-naive patients, repaglinide had a potent glucose-lowering effect in the postprandial period.

A comparison of premixed insulin preparations in elderly patients. Efficacy of 70/30 and 50/50 human insulin mixtures.

OBJECTIVE: To compare the postprandial hyperglycemic response to a standard breakfast of two premixed humulin insulin mixtures, 50/50 (50% NPH human insulin and 50% regular human insulin) and 70/30 (70% NPH human insulin and 30% regular human insulin) in elderly non-insulin-dependent diabetes mellitus (NIDDM) patients. RESEARCH DESIGN AND METHODS: On two mornings, each patient (n = 20) consumed a standard breakfast after a single dose of 50/50 or 70/30 insulin (0.3 U/kg) was administered in a randomized crossover fashion. Plasma glucose and serum free insulin concentrations were measured before and for 4 h after insulin administration. RESULTS: Plasma glucose reached a peak at 60 min and a nadir at 240 min for both types of insulin. No differences in maximum and minimum glucose concentrations, time to maximum and minimum glucose concentrations, or areas under the curve were noted. Free insulin levels did not differ significantly. CONCLUSIONS: These results suggest that small changes in the composition of premixed insulin mixtures in NIDDM patients may not result in improved postprandial glycemic control.

Benefits and risks with glyburide and glipizide in elderly NIDDM patients.

OBJECTIVE: To compare the efficacy, benefits, and risks of glyburide and glipizide in elderly patients with non-insulin-dependent diabetes mellitus (NIDDM). RESEARCH DESIGN AND METHODS: Twenty-one elderly outpatients (mean age 70 yr) were treated for 8 wk, after being dose-titrated to achieve a fasting plasma glucose (FPG) concentration of less than 7.8 mM with glyburide or glipizide in a randomized crossover trial. FPG and postprandial plasma glucose, serum C-peptide, and HbA1c levels were measured. In 13 patients, self-monitoring of blood glucose (SMBG) with a memory meter was performed seven times per week. RESULTS: Glipizide (11.9 mg) and glyburide (8.4 mg) produced similar fasting and postprandial plasma glucose and HbA1c concentrations. No significant differences in basal or stimulated C-peptide levels were detected. Despite a few patient reports of hypoglycemia, a high incidence of SMBG readings less than 4.5 mM was attributed to the use of both drugs. CONCLUSIONS: Both treatments proved effective for glycemic control; however, both second-generation sulfonylureas are associated with a significant risk of hypoglycemia in elderly NIDDM patients. The proper use of sulfonylureas in this population should include close surveillance of ambulatory glucose monitoring and intensive and repeated patient education about the risks of hypoglycemia.

The safety and efficacy of a controlled low-energy ('very-low-calorie') diet in the treatment of non-insulin-dependent diabetes and obesity.

We evaluated the safety and efficacy of a highly supplemented controlled low-energy (1764 kJ [420 kcal]) diet in the treatment of non-insulin-dependent diabetes and obesity. Six obese, diabetic women ranging from 143% to 297% of ideal body weight were studied in a metabolic ward for 48 days. The subjects ingested a weight-maintenance diet during an eight-day control period followed by 40 days of an experimental diet containing 1764 kJ (420 kcal) of a mixture of protein (43% of energy intake), carbohydrates (51%), and fat (6%), supplemented with minerals, trace elements, and vitamins. The subjects were monitored for balances of nitrogen and minerals, as well as for the appearance of cardiac arrhythmias by 24-hour electrocardiographic recordings. Weight loss was rapid and sustained and averaged 10.1% +/- 0.8% over 40 days. Fasting plasma glucose levels declined from 16.2 +/- 1.9 mmol/L (293 +/- 36 mg/dL) to 6.9 +/- 0.8 mmol/L (126 +/- 16 mg/dL) by day 35. Similarly, hemoglobin A1c levels fell from 0.11 +/- 0.009 (11.2% +/- 0.9%) to 0.8 +/- 0.001 (8.2% +/- 1.1%). Urinary C-peptide levels declined from 62.2 +/- 15.6 nmol/48 h to 20.0 +/- 5.9 nmol/48 h by days 39 to 40 and paralleled the decline in plasma glucose values, the majority of which occurred in the first seven days. Concentrations of serum cholesterol and triglycerides decreased. Balances for nitrogen, potassium, and magnesium were negative at -1.7 g/24 h, -2.2 mEq/24 h, and -2.9 mg/dL, respectively. Blood pressure decreased without orthostasis. Resting metabolic rate fell a mean of 18% but remained within normal limits. Triiodothyronine levels also declined. Twenty-four-hour ambulatory electrocardiographic readings disclosed no significant bradyarrhythmia or tachyarrhythmia for any patient. These studies, based on a limited number of subjects, demonstrate that a highly supplemented controlled low-energy diet is a safe and efficacious treatment for diabetes and obesity, leading to significant decreases in weight, blood pressure, and levels of plasma glucose and plasma lipids. Such diets may be the optimal initial treatment of moderate to markedly obese patients with non-insulin-dependent diabetes.

Evaluation of a new radioimmunoassay for urinary albumin.

We have evaluated a new commercially available radioimmunoassay kit for albumin determination in urine. The assay is precise; within-run precision (CV) in the clinically significant ranges is 1.8-3.5%, between-run, 1.2-8.5%. The minimum detection limit was 0.6 micrograms/ml. Analytic recovery of different concentrations of albumin added to urine ranged from 98% to 103%. Samples, stored in plastic containers, were stable at room temperature for periods up to 7 days. Mean albumin excretion rates, measured in 20 normal volunteers for 3-h and 24-h periods during the same day were similar (7.1 +/- 4.6 [SD] versus 6.5 +/- 5.0 micrograms/min). In 8 normal subjects, 3-h excretion rates measured daily for 5 days showed no significant variability. In eight insulin-dependent diabetic subjects, albumin excretion measured in short periods of urine collection (3 h) were also in close agreement with 24-h collections (24.7 +/- 28.9 versus 17.6 +/- 18 micrograms/min). From these results it appears that this commercially available kit is suitable for conveniently monitoring microalbuminuria in large numbers of patients in research studies as well as for office practice. Such widespread use should make it possible to better determine the clinical usefulness of this test in the management of diabetic patients.

Use of urinary C-peptide to estimate insulin secretion during starvation.

The usefulness of measurements of urinary C-peptide excretion in indirectly assessing integrated insulin secretion during starvation was studied in eight obese subjects during a 72-h fast. Blood and urine samples were collected at 12-h intervals for measurement of insulin and C-peptide immunoreactivity. After 60 h, serum insulin and plasma C-peptide levels declined 47% and 37%, respectively, and the values were highly correlated (r = 0.8; P less than 0.001). By 72 h, urinary C-peptide excretion had declined to 70% of the level in the first 12-h period. The urinary clearance of C-peptide was not altered by starvation. A highly significant correlation was found between urinary C-peptide and C-peptide secretory rate (P less than 0.001). The molar ratio of plasma C-peptide to insulin remained constant during the fasting period. These data indicate that basal insulin secretion can be added to the list of physiological conditions in which beta-cell secretion can be effectively evaluated by urinary C-peptide measurement.

Starvation enhances the ability of insulin to inhibit its own secretion.

To examine whether decreased insulin secretion during starvation is related to a change in the ability of insulin to inhibit its own secretion, plasma C-peptide was measured after plasma insulin levels were acutely raised by intravenous (IV) insulin infusion in a dose of 40 and 80 mU/M2/min in obese subjects before and after a 72 hour fast. Plasma glucose concentration was maintained +/- 4% of basal levels by a variable glucose infusion. During the 80 mU infusion, at plasma insulin levels of 200 microU/mL, plasma C-peptide fell by 0.17 pmol/mL in the fed state. In the fasted state, despite basal levels that were 36% lower, C-peptide decreased by 0.21 pmol/mL. Highly significant increases in percent suppression after fasting were noted during both 40 mU and 80 mU studies. The plasma C-peptide response was related to the insulin infusion dose in both the fed and fasted state. In contrast, alpha cell suppression by insulin, as determined by plasma glucagon levels, was not altered by fasting. It is concluded that enhanced inhibitory influences of insulin on the beta cell during starvation may be a physiologically important mechanism for diminished insulin secretion during the transition from the fed to the fasting state.

Treatment of insulin reactions in diabetics.

Using a modification of the glucose clamp technique, we have studied the efficacy of commonly used foods to correct hypoglycemia in insulin-dependent diabetics. After lowering the plasma glucose level to 55 mg/dL at a steady-state plasma free insulin concentration of about 50 microU/mL, patients were fed 20 g of carbohydrate as milk, orange juice, or D-glucose or 40 g of carbohydrate as orange juice. The data indicate that 20 g of carbohydrate as D-glucose corrects hypoglycemia without rebound hyperglycemia. In an outpatient setting, this treatment also proved effective in spontaneous episodes of hypoglycemia. We conclude that (1) the D-glucose content of the ingested carbohydrate is an important determinant of the glycemic response, and (2) at times of moderately severe hypoglycemia, ingestion of 20 g of D-glucose provides an effective glycemic response for periods of at least 40 minutes. In view of these data, a table is provided listing some common sources of 20 g of D-glucose.

Evaluation of a direct solid-phase radioimmunoassay for progesterone, useful for monitoring luteal function.

We have evaluated a commercially available, direct, solid-phase radioimmunoassay kit for progesterone determination in serum or plasma. The assay is precise, within-run precision (CV) in the clinically significant ranges being 2.5 to 5.2%, between-run 5.5 to 5.8%. Mean analytical recovery of different concentrations of progesterone added to serum was 99.7% (range 95.3 to 102.7%). Fourteen closely related steroids showed no cross reactivity. The minimum detection limit was 0.5 microgram/L. Luteal-phase progesterone concentrations in serum were increased (greater than 3 micrograms/L) in 19 normal ovulatory menstrual cycles and decreased (less than 1.5 micrograms/L) in two nonovulatory cycles. We found this direct assay for progesterone to be analytically and clinically sound, and useful for assessing luteal-phase function.

Aspirin causes tissue insensitivity to insulin in normal man.

The effect of aspirin on glucose and insulin metabolism was examined with the hyperglycemic clamp technique in 8 normal volunteers. When the plasma glucose concentration was acutely raised and maintained at 125 mg/dl above the basal level after treatment with aspirin (3 g daily for 3 days), acute (0-10 min) and sustained (20-120 min) insulin release were 70% and 45% greater than before treatment. Despite the increased plasma insulin level, the glucose infusion rate remained unchanged (8 +/- 0.9 to 9.1 +/- 1.2 mg/kg X min). Consequently, the ratio of the glucose infusion rate to the plasma insulin level, an index of tissue sensitivity to endogenous insulin, decreased by 30%, indicative of impaired insulin action. Aspirin did not alter fasting levels of FFA. When ibuprofen, another prostaglandin synthesis inhibitor, was given to 10 normal volunteers, only an effect on acute insulin release could be demonstrated. These results demonstrate that aspirin not only enhances beta-cell sensitivity to glucose, but also impairs glucose metabolism in insulin-sensitive tissues. It is not clear whether these effects are related to aspirin's ability to inhibit prostaglandin synthesis.

Free thyroxin by radioimmunoassay: evaluation of a new direct method involving a radiolabeled thyroxin analog.

We present here the first performance evaluation of a new direct method for free thyroxin (T4) in serum by radioimmunoassay, with use of coated tubes and a radiolabeled T4 analog (Diagnostic Products Corp.). The assay is precise and robust: within-run imprecision (CV), 3.1-6.6%; between-run imprecision, 4.0-7.9%; no demonstrable variation between technologists irrespective of experience with the method. No outliers were observed when we compared the free T4 results with serum total T4. Reference values are reported for a total of 1243 euthyroid subjects; there was no significant age effect on serum free T4 in women 26 to 72 years old. The biological variation was about +/- 35% of the mean (2 SD). Free T4 results are the same for serum and plasma. The assay performs well in hypothyroidism and hyperthyroidism, and distinguishes individuals with thyroid disease from normal individuals. Free T4 values in women taking oral contraceptives are normal. Depressed results were often observed in acute nonthyroidal illness and continuing pregnancy. These results were directly comparable with those of another commercial direct radiolabeled-T4 analog kit for free T4.

Insulin action during acute starvation: evidence for selective insulin resistance in normal man.

The effects of insulin on glucose utilization, lipolysis, and potassium and phosphate metabolism were studied during short-term fasting in six lean subjects using a sequential euglycemic glucose clamp technique (two additional subjects were used in 70 mU/m2/min clamp studies). The subjects were infused with insulin for four hours at four rates ranging from 6 to 442 mU/m2/min before and after a 48-hour fast. Insulin was infused for one hour at each rate in all experiments. Fasting markedly reduced glucose utilization at all insulin infusion rates. On the other hand, the decline in levels of free fatty acids that occurred at insulin concentrations of 30 microU/ml was virtually identical before and after fasting. After insulin was infused for four hours, serum phosphate had decreased in all subjects (P less than 0.001) and strongly correlated with glucose disposal rates (r = 0.76, P less than 0.005). The plasma potassium level also declined in all subjects but did not relate to fasting or glucose disposal. These studies demonstrate that starvation produces selective insulin resistance. The biologic effect of insulin on glucose utilization and plasma phosphate shifts is clearly diminished. Free fatty acid and potassium metabolism are unaffected by starvation.

Starvation stimulates pancreatic PGE content.

Recent evidence suggests that prostaglandins may exert a tonic inhibitory tone in the pancreatic beta cell during starvation. The effects of starvation on rat pancreatic prostaglandin E (PGE) content were studied. After 72 hrs of starvation, pancreatic PGE increased 250% above that of fed controls. Administration of streptozotocin, a selective beta-cell toxin, decreased pancreatic PGE significantly (p less than 0.005), but starvation partially reversed this trend. Thus, PGE may have a physiological role in modulating insulin secretion during starvation. It appears that both beta-cell and nonbeta-cell sources of PGE are involved in this phenomenon.

Metabolic effects of prostaglandin E2 infusion in man: possible adrenergic mediation.

Infusion of prostaglandin E2 (PGE2) has been previously shown to inhibit acute insulin release and cause glucose intolerance. The present studies were undertaken to assess the effects of PGE2 on 1) pancreatic beta-cell sensitivity to glucose, 2) glucose tolerance, 3) tissue sensitivity to insulin, 4) glucose production and clearance, and 5) plasma catecholamine, glucagon, and FFA levels. Six healthy adult subjects were studied with the hyperglycemic clamp technique (plasma glucose 125 mg/dl above basal for 2 h) before and 30 min after the start of a PGE2 infusion (10 micrograms/min). Plasma epinephrine, norepinephrine, and FFA were measured during the PGE2 infusion. In additional experiments, glucose production and utilization were measured isotopically ([3-3H]glucose) during PGE2 infusion. PG infusion diminished, but not significantly, acute insulin release (0-10 min preinfusion, 172 +/- 36; postinfusion, 148 +/- 45 microunits/ml . 10 min). Late insulin release (20-120 min) was unchanged. A significant decline occurred in the amount of glucose metabolized from 9 +/- 1.1 to 7.2 +/- 1 mg/kg . min. During the initial 30 min of PGE2 infusion, plasma FFA increased by 26 +/- 6% (P less than 0.025). Plasma epinephrine and norepinephrine rose from 40 +/- 6 to 104 +/- 24 pg/ml (P less than 0.05) and 204 +/- 17 to 440 +/- 30 pg/ml (P less than 0.01), respectively. PGE2 produced a prompt 30% rise in glucose output, which declined to basal levels by 60 min. Glucose clearance decreased transiently at 45 min by 23%. We conclude that the effects on glucose homeostasis noted during PGE2 infusion occur in the face of heightened adrenergic activity. These metabolic responses closely resemble adrenergically induced changes in glucose homeostasis. As such, before any metabolic effects can be attributed directly to infused PGE2, any metabolic effects can be attributed directly to infused PGE2, the role of concomitant catecholamine release must be considered.

Adrenergic influence on glucocounterregulation in man.

To investigate the adrenergic role in glucocounterregulatory mechanisms, single-blind randomised studies were performed in 7 normal males during severe insulin-induced hypoglycaemia with or without adrenergic blockade. Intravenous phentolamine administration (5 mg stat and 0.5 mg/min) did not interfere with the restoration of euglycaemia from hypoglycaemia. However, recovery of blood glucose in the presence of propranolol (3 mg/3 min and 0.8 mg/min) was retarded when compared with control studies (mean plasma glucose levels +/- SEM , 50 +/- 6 mg/dl versus 66 +/- 4 mg/dl at 120 min after insulin administration) despite appropriate glucagon, epinephrine, cortisol, and growth hormone responses. Plasma norepinephrine response was unaffected by propranolol but augmented threefold by phentolamine. Increases in plasma lactate, pyruvate and non-esterified fatty acids were blunted with propranolol while rebound non-esterified fatty acid was observed with phentolamine infusion. These data suggest that complete recovery of blood glucose from sever hypoglycaemia requires full sympathetic nervous system activity despite the integrity of other counterregulatory mechanisms.

Indomethacin and aspirin prevent the starvation-induced fall in plasma insulin.

To investigate whether prostaglandins (PGs) play a role in the regulation of insulin secretion during starvation, we have studied the effects of two inhibitors of PG synthesis, indomethacin (INDO) and acetylsalicylic acid (ASA), on plasma insulin during a 72-h fast. Five lean males and six obese females were given 200 mg INDO daily throughout a 72-h fast during which plasma glucose was maintained at normal postabsorptive levels by a continuous infusion of glucose. In addition, four obese females were treated with 3 g ASA in a similar protocol. Another six lean males and six obese females served as a control group, receiving only iv glucose during the fast. In both the lean and obese control subjects, a significant decrease in plasma insulin was noted by 72 h (lean, 53 +/- 8% of basal insulin; obese, 69 +/- 6%; P less than 0.02). By contrast, aspirin and INDO administration prevented the decline in plasma insulin in both lean (INDO, 92 +/- 5%) and obese (INDO, 109 +/- 11%; ASA, 111 +/- 17%) subjects. These data suggest that endogenous PG production may be a controlling factor in insulin secretion during starvation.