Calorie restriction is widely used as a primary therapy for obese pregnant women with gestational diabetes. To better understand the metabolic consequences of marked calorie restriction, we performed a randomized prospective trial under metabolic ward conditions. Obese gestationally diabetic women were randomized to control (n = 5) and calorie-restricted (n = 7) groups. All patients consumed an approximately 2400-kcal/day diet during the 1st wk of the study, and at the end of the 1st wk, metabolic features of the two groups were statistically indistinguishable. During the 2nd wk, the control group continued to consume approximately 2400 kcal/day, whereas the calorie-restricted group consumed approximately 1200 kcal/day. Twenty-four-hour mean glucose levels remained unchanged in the control group (6.7 +/- 0.8 mM wk 1 vs. 6.8 +/- 0.8 mM wk 2), although they dropped dramatically in the calorie-restricted group (6.7 +/- 1.0 mM wk 1 vs. 5.4 +/- 0.5 mM wk 2, P less than 0.01). Fasting plasma insulin also declined in the calorie-restricted group (265 +/- 165 pM wk 1 vs. 145 +/- 130 pM wk 2), resulting in a significant change between groups (P less than 0.02). Surprisingly, fasting plasma glucose and glucose tolerance measured by the 3-h oral glucose tolerance test did not change within or between groups. Fasting levels of beta-hydroxybutyrate rose in the calorie-restricted group (290 +/- 240 microM wk 1 vs. 780 +/- 30 microM wk 2) but not in the control group (P less than 0.01). Finally, urine ketones increased significantly (P less than 0.02) in the calorie-restricted group, whereas they remained absent in the control group.(ABSTRACT TRUNCATED AT 250 WORDS)

The effect of immunosuppression on the humoral immune response to islet autoantigens and exogenously administered insulin and the predictive value of islet cell cytoplasmic antibodies (ICAs), insulin antibodies (IAs), and HLA-DR phenotype for remission during immunosuppression were studied in a prospective randomized double-blind trial of cyclosporin administration in 98 newly diagnosed insulin-dependent diabetes mellitus (IDDM) patients. HLA-DR phenotype and glycosylated hemoglobin were determined at study entry, and insulin requirement, glucagon-stimulated C-peptide, ICAs, and IAs were measured at entry and after 1, 3, 6, 9, and 12 mo of follow-up. Cyclosporin therapy caused significant suppression of the prevalence and serum concentrations of ICAs and IAs. Cyclosporin-treated IDDM patients ICA+ at study entry had higher levels of stimulated C-peptide after 1 mo of study, but the increased beta-cell function was not associated with a higher frequency of insulin-free remission at 1 mo. ICA and IA status at entry did not predict cyclosporin-insulin-free remission as assessed by the prevalence of insulin-free remission or beta-cell function at 3-12 mo of study, and significant decrements in the titers or total disappearance of ICAs were not associated with an increased prevalence or duration of non-insulin-requiring remission or higher stimulated C-peptide values. There was no correlation between the serum levels of ICAs and IAs at entry and beta-cell function at 12 mo of follow-up.(ABSTRACT TRUNCATED AT 250 WORDS)

Fish oils, containing omega-3 fatty acids (omega 3FAs), favorably influence plasma lipoproteins in nondiabetic humans and prevent the development of insulin resistance induced by fat feeding in rats. We studied the effects of fish oils in 10 subjects (aged 42-65 yr) with mild non-insulin-dependent diabetes mellitus (NIDDM). Subjects were fed a standard diabetic diet plus 1) no supplementation (baseline), 2) 10 g fish oil concentrate (30% omega 3FAs) daily, and 3) 10 g safflower oil daily over separate 3-wk periods, the latter two supplements being given in radom order by use of a double-blind crossover design. At the end of each diet period, fasting blood glucose (FBG), insulin, and lipids were measured, and insulin sensitivity was assessed with a hyperinsulinemic-euglycemic clamp performed with [3-3H]glucose. FBG increased 14% during fish oil and 11% during safflower oil supplementation compared with baseline (P less than .05), whereas body weight, fasting serum insulin levels, and insulin sensitivity were unchanged. The absolute increase in FBG during each supplementation period correlated with the baseline FBG (fish oil, r = .83, P less than .005); safflower oil, r = .75, P = .012). Fasting plasma triglyceride levels decreased during fish oil supplementation in the 4 subjects with baseline hypertriglyceridemia (greater than 2 mM) but were not significantly reduced overall. There was no significant change in fasting plasma total, high-density lipoprotein, and low-density lipoprotein cholesterol levels. In summary, dietary fish oil supplementation adversely affected glycemic control in NIDDM subjects without producing significant beneficial effects on plasma lipids. The effect of safflower oil supplementation was not significantly different from fish oil, suggesting that the negative effects on glucose metabolism may be related to the extra energy or fat intake.(ABSTRACT TRUNCATED AT 250 WORDS)

A double-blind controlled trial of azathioprine (2 mg.kg-1.day-1) was conducted with 49 patients aged 2-20 yr (mean 10.8 yr) who had newly diagnosed type I (insulin-dependent) diabetes. Patients were randomly assigned to receive either azathioprine (n = 24) or placebo (n = 25) for 12 mo, beginning within the 20-day period after diagnosis. Baseline clinical and metabolic characteristics did not differ between the two groups. No patient experienced complete remission, defined as restoration of normal carbohydrate tolerance without other treatment. Partial remission, defined as good metabolic control (hemoglobin A1c less than or equal to 7.9%, preprandial blood glucose less than or equal to 8 mM with an insulin dose of less than 0.5 U.kg-1.day-1), occurred in 10 placebo (40%) and 7 azathioprine (29%) patients at 6 mo and in 4 placebo (16%) and 4 azathioprine (17%) patients at 12 mo (differences not significant). Fasting plasma C-peptide was significantly greater in the azathioprine-treated group at 3 and 6 mo, but this difference was not sustained. C-peptide responses to a standard meal and the frequency of islet cell and insulin antibodies did not differ between the two groups over the 12-mo period. Azathioprine caused no significant side effects. We conclude that in the dosage used, and despite early effects on endogenous insulin secretion, azathioprine alone does not influence the remission phase in children with newly diagnosed type I diabetes.

We evaluated the effectiveness of a more potent and longer-acting somatostatin analogue (SMS 201-995) as an adjunct to insulin therapy, in a double-blind placebo-controlled randomized study of 26 C-peptide-negative type I (insulin-dependent) diabetic patients (20 women, 6 men, aged 22-40 yr) on their conventional drug regimens for 12 wk. Eight patients received a low dose (10 micrograms) of the analogue, 9 received a high dose (50 micrograms) of the analogue, and 9 received placebo subcutaneously before breakfast and dinner. Twenty-four-hour serum glucose, free insulin, plasma growth hormone (GH), and glucagon profiles were obtained before and during treatment at 4-wk intervals. The mean age, duration of diabetes, daily insulin dose, and body weight were not significantly different among the groups. The mean weekly capillary blood glucose values and exogenous insulin requirements were not changed by the SMS 201-995 therapy. Mean glycosylated hemoglobin A1 levels were unchanged in both the analogue- and placebo-treated groups at wk 12. Basal and postprandial glucose, free insulin, GH, and glucagon profiles were not influenced by the SMS 201-995 therapy throughout the study. Nocturnal glucose turnover rates (D-[3-3H]glucose technique) remained unaltered by the analogue therapy. Dose-dependent gastrointestinal (GI) adverse effects (e.g., diarrhea) were documented in the analogue-treated patients. Visual acuity and fundic photomicrographs of our patients were not changed by the analogue therapy. In conclusion, the prominent adverse GI effects our patients experienced preclude the use of larger doses of the analogue that may be necessary to suppress GH and glucagon and improve glucose control in type I diabetic patients.

In a double-blind randomized controlled clinical trial conducted in two French and two United Kingdom centers, the effect of antiplatelet agents, i.e., aspirin alone (330 mg 3 times daily) or in combination with dipyridamole (75 mg 3 times daily), was tested versus placebo in 475 patients with early diabetic retinopathy. The assessment of retinopathy was based on change in the number of microaneurysms (MAs) present in the macular field, as seen on fluorescein angiograms, over 3 yr. Forty-one patients did not complete the study. At least three readable initial and yearly angiograms were available on 420 patients (266 treated with insulin and 154 not treated); the results reported are based on these patients. The mean yearly increase in definite MAs was similar in insulin-treated and non-insulin-treated diabetic patients. There was no significant difference between the aspirin-alone group (0.69 +/- 5.1 mean +/- SD, n = 145) and the aspirin-plus-dipyridamole group (0.34 +/- 3.0, n = 142). In the placebo group the mean yearly increase (1.44 +/- 4.5, n = 133) was significantly higher than in the treated group (P = .02, 1-tailed t test). There was a clear relationship between the deterioration in ophthalmological signs and the increase in mean yearly MAs (clinically stable, 0.38 +/- 3.96, n = 293; deteriorating, 1.79 +/- 4.89, n = 127; P = .002, 2-tailed t test). We conclude that aspirin alone and in conjunction with dipyridamole significantly slows the progression of MA evolution in early diabetic retinopathy.

Hypertriglyceridemic patients with non-insulin-dependent diabetes mellitus (NIDDM) have an increased risk of coronary heart disease (CHD) and acute pancreatitis. To examine the potential of hypolipidemic drugs for therapy of lipoprotein abnormalities in NIDDM, 10 patients maintaining marked (plasma triglycerides greater than 500 mg/dl) and 6 with moderate (plasma triglycerides 250-500 mg/dl) hypertriglyceridemia, despite good glycemic control, were studied in two phases. In the first phase, gemfibrozil alone (600 mg twice daily) was compared with a placebo, and in the second phase a combination of gemfibrozil and lovastatin (20 mg twice daily) was compared with gemfibrozil alone in a randomized, double-blind, placebo-controlled crossover study. In markedly hypertriglyceridemic patients, gemfibrozil reduced plasma triglycerides by 52% and very-low-density lipoprotein cholesterol (VLDL-chol) by 55% and increased high-density lipoprotein cholesterol by 23% compared with a placebo. However, low-density lipoprotein cholesterol (LDL-chol) levels increased (42%), and LDL apolipoprotein B (apoB) levels remained unchanged. Addition of lovastatin to gemfibrozil effectively reduced total cholesterol (25%), LDL-chol (30%), and LDL-apoB (19%). Lovastatin further reduced plasma triglycerides (11%) and VLDL-chol (27%). However, in moderately hypertriglyceridemic patients, gemfibrozil or the combination therapy did not seem to offer benefits over the previously reported study with lovastatin alone. Glycemic control was maintained throughout the study. In conclusion, the beneficial effects of the combination therapy on lipoprotein levels in markedly hypertriglyceridemic NIDDM patients could decrease the risk of development of both acute pancreatitis and CHD.

Seventy patients aged 15-40 yr with recent-onset insulin-dependent diabetes mellitus (IDDM) were entered into a double-blind trial, in which they were randomly assigned to either cyclosporin (7.5 mg.kg-1.day-1) or to placebo and were monitored for 1 yr for various phenotypic and functional parameters of T-lymphocyte-mediated immunity. Before treatment, the proportions of total T-lymphocytes (CD3+) and helper-inducer T-lymphocytes (CD4+) were normal, whereas significantly decreased values of suppressor/cytotoxic T-lymphocytes (CD8+), as compared with normal controls, were found in 31% of the patients. The interleukin 2 (IL-2)-receptor expression was significantly increased in IDDM patients compared with control subjects, although the single values were low: patients, 2.02 +/- 0.41%; controls, 0.88 +/- 0.25% (means +/- SE). Circulating levels of soluble IL-2 receptor were also significantly increased in IDDM patients compared with controls: patients, 372.3 +/- 25.4 U/ml; controls, 235.5 +/- 29.3 U/ml (means +/- SE). However, no major abnormalities were found in mitogen (phytohemagglutinin)-induced IL-2 production, cell proliferation, or IL-2-receptor expression. After 6 mo of cyclosporin treatment, no major modifications of any of the parameters analyzed were noted, even in patients who had cyclosporin blood trough levels greater than 300 ng/ml, i.e., the threshold value associated with clinical efficacy. One explanation for the absence of a major effect of cyclosporin, in contrast with its demonstrated clinical effectiveness, is the reversibility of its activity. Our results preclude the use of the described tests to reliably monitor IDDM patients undergoing immunosuppressive therapy.

Cyclosporin or placebo was administered in a randomized, double-blind fashion to 13 patients with multiple sclerosis for 1 yr to determine whether cyclosporin adversely affects glucose homeostasis or beta-cell function. No significant differences were observed in fasting glucose, fasting insulin, intravenous glucose tolerance, or glucose-induced insulin secretion before treatment or at 3 wk, 6 mo, or 1 yr during treatment. Longer therapeutic trials with larger patient groups will be necessary to decide whether cyclosporin can be safely given for many years without risk of developing diabetes mellitus.

Cyclosporin is an immunosuppressive drug used with increasing frequency in patients with diabetes mellitus both as experimental primary therapy for insulin-dependent diabetes mellitus and as therapy accompanying pancreatic transplantation. However, reports have appeared contending that cyclosporin causes glucose intolerance and inhibits pancreatic islet beta-cell function. Consequently, concern has been raised that the beneficial effects of immunosuppression may be offset by adverse metabolic effects of the drug. To address this issue, we examined intravenous glucose tolerance and pancreatic islet beta-cell function in a group of nondiabetic multiple sclerosis patients before and during a 2-yr course of cyclosporin or placebo therapy. Patients were randomly assigned to one of the two drug groups and followed in a double-blind manner. Basal levels of glucose, insulin, and C-peptide as well as glucose disappearance rates and pancreatic islet beta-cell function after stimulation with intravenous glucose and arginine were determined immediately before therapy and after 3 wk, 6 mo, 1 yr, and 2 yr of therapy. No abnormalities in these parameters were observed in the cyclosporin of the placebo-treated group. It appears that cyclosporin can be give in conventional doses for as long as 2 yr without encountering evidence for impaired glucose homeostasis. However, whether adverse effects will materialize over longer periods of drug use remains a question.

Proteinuria in diabetes is associated with progressive glomerular damage. We studied the effects of 3-wk dietary protein restriction on proteinuria and renal function in 10 insulin-dependent diabetic men with diabetic nephropathy. Patients were randomly assigned by a crossover design to 40-g low-protein diet (LPD) or usual-protein diet (UPD). Glomerular filtration rate and renal plasma flow were measured by inulin and p-aminohippurate clearance at the end of each period under conditions of sustained euglycemia. Total calorie intake, body weight, serum albumin and total protein concentrations, hematocrit, blood pressure, and glucose control were similar during the two diets. Achieved protein intake was 46 +/- 3 g/day during LPD and 81 +/- 4 g/day during UPD (P less than .001). Urinary urea appearance and plasma urea were significantly lower on LPD. Median total urinary protein was reduced from 3.9 g/day (range 0.5-12.3) on UPD to 2.4 (range 0.2-9.0) on LPD (P less than .006), and there was a significant fall in the median fractional clearance of albumin from 2.0 x 10(-4) (range 0.1-90.9) on UPD to 1.0 x 10(-4) (range 0.1-51.4) on LPD and IgG from 2.1 x 10(-5) (range 0.2-238) to 1.5 x 10(-5) (range 0.1-77) (P less than .006 and P less than .02, respectively). The reabsorption rate of beta 2-microglobulin was similar on the two diets and glomerular filtration rate, renal plasma flow, and filtration fraction remained unchanged. Thus, short-term dietary protein restriction reduces diabetic proteinuria independently of blood glucose or systemic blood pressure changes by improving glomerular permselectivity.

A randomized double-blind placebo-controlled trial was undertaken to determine whether cyclosporin enhances remission of insulin-dependent diabetes mellitus (IDDM) through the 1st yr after diagnosis. Dosage with insulin was minimized with target control of blood glucose levels less than or equal to 7.8 mM (140 mg/dl) before meals. Metabolic control was evaluated by serial determinations of glycosylated hemoglobin levels, and endogenous secretion of insulin was evaluated by determination of the levels of glucagon-stimulated insulin-connecting peptide (CP) in the plasma at 3-mo intervals. A compound definition of remission required a glucagon-stimulated CP level in plasma greater than or equal to 0.6 nM or a non-insulin-receiving state (NIR) in which target control of glycemia was maintained without administration of insulin. A clinical definition of remission required only the NIR state as defined. One hundred eighty-eight patients aged 10-35 yr entered the study within 6 wk of initiation of insulin therapy and within 14 wk of onset of symptoms and were studied for 1 yr. There were no significant differences in metabolic control between the two treatment groups during the study. The anticipated adverse effects of cyclosporin were not more frequent or severe than in other experience with the drug, but histological changes attributable to cyclosporin were present in some kidney biopsies obtained from selected patients after 1 yr. At 1 yr, by the compound definition, 33% of the cyclosporin-group and 21% of the placebo-group patients were in remission, when the corresponding rates for NIR remissions were 24 and 10%.(ABSTRACT TRUNCATED AT 250 WORDS)

Anti-beta-cell-specific cell-mediated immunity was studied over a 12-mo period in 65 recently diagnosed diabetic patients randomly receiving either cyclosporin or placebo. Anti-beta-cell cellular immunity was assessed by an in vitro test based on the inhibition of insulin release from cultured rat islet cells by patients' mononuclear cells. This beta-cell-suppressive effect disappeared in cyclosporin A-treated patients within 1 mo and did not reappear during 12 mo of follow-up. Conversely, the suppressive effect persisted unchanged in placebo-treated patients during 12 mo of follow-up. These changes were predictive neither of cyclosporin A-induced remission nor of relapses. Results of the insulin-release inhibition test were not correlated to islet cell autoantibodies or HLA phenotype.

Autonomic and somatosensory nerve function was studied in 24 insulin-dependent diabetic subjects (aged 29 +/- 7 yrs, diabetes duration 8 +/- 4 yr) randomly allocated to either continuous subcutaneous insulin infusion (CSII; n = 12) or unchanged conventional insulin therapy (CIT; n = 12). Measures of glycemic control and somatosensory and autonomic nerve function were comparable in the two groups at the start. Glycemic control was significantly improved in the CSII group throughout study, whereas it remained unchanged in the CIT group. In the CIT group, vibratory perception threshold (VPT) of the great toe and the medial malleolus deteriorated, as did heart rate variation (HRV) at rest, at deep breathing (.05 less than P less than .06), and at standing. In contrast, CSII patients retained their VPT and HRV. Comparison of nerve function alterations during the 2-yr trial showed better preservation in CSII than in CIT patients of VPT in the great toe (0.8 +/- 1.7 vs. -1.4 +/- 1.9 V, P less than .01) and the medial malleolus (1.5 +/- 2.9 vs. -1.4 +/- 1.8 V, P less than .05) and of HRV at rest (10 +/- 24 vs. -13 +/- 22 ms, P less than .05) and at standing (-0.01 +/- 0.13 vs. -0.15 +/- 0.16 ms, P less than .05). We conclude that intensified glycemic control can favorably influence parasympathetic and somatosensory nerve function in insulin-dependent diabetes mellitus.

The Diabetes Control and Complications Trial (DCCT) is a multicenter randomized clinical trial studying the effect of intensive insulin therapy on the early vascular and neurological complications of insulin-dependent diabetes mellitus (IDDM). During the feasibility phase of the DCCT, baseline neurological histories, physical examinations, and laboratory measurements of somatic and autonomic nerve function were obtained in 278 well-characterized IDDM subjects. Subjects were free of advanced complications, including the presence of peripheral or autonomic neuropathy sufficiently severe to require treatment. Analyses of the cross-sectional data reveal that clinically detectable peripheral neuropathy was present in 39% of the subjects. The presence of clinical neuropathy correlated with greater age, longer duration of IDDM, and male gender. The somatic and autonomic test results confirm the relationship between age, diabetes duration, and male gender and diabetic neuropathy. These results support an effect of age and gender on the development of diabetic complications.

The dawn phenomenon was evaluated in eight C-peptide-negative type I (insulin-dependent) diabetic patients on two occasions by measuring glucose concentrations every 30 min from 2400 to 0800 h while the subjects were receiving an insulin infusion (0.12 mU.kg-1.min-1). In random order at 2230 h, they orally received either a sleeping medication alone or with 5.0 mg methscopolamine bromide, an anticholinergic agent. The peak growth hormone (GH) concentrations (ng/ml +/- SE) after sleep were markedly inhibited by methscopolamine (4.7 +/- 2.6 vs. 23.0 +/- 9.2). During the control night, the late (0400-0800 h) glucose response (area under curve but above 0400 h value) was significantly higher (P less than .02) than the early (2400-0400 h) glucose response (area under curve but above 2400 h value). After methscopolamine, the early and late glucose responses were virtually identical. The anticholinergic agent did not affect glucagon levels, overnight urinary catecholamine excretion, or the diurnal cortisol concentrations. The total area under the free fatty acid (FFA) curves was significantly (P less than .05) reduced by methscopolamine. We conclude that sleep-induced GH secretion may cause the dawn phenomenon by increasing FFA levels. Oral administration of methscopolamine at bedtime is a simple pharmacological approach that could test the clinical importance of the dawn phenomenon.

To determine whether therapy with exogenous insulin or sulfonylureas results in a postprandial pattern of carbohydrate metabolism in patients with non-insulin-dependent diabetes mellitus (NIDDM) that resembles that in nondiabetic individuals, we employed a dual-isotope technique combined with forearm catheterization to examine meal disposition in NIDDM patients, before and after 3 mo of therapy with tolazamide and after 3 mo of therapy with exogenous insulin, with a randomized crossover design. Results were compared with those observed in nondiabetic subjects. Although both forms of therapy improved chronic glycemic control (glycosylated hemoglobin concentration went from 9.6 +/- 0.7 to 7.6 +/- 0.5 and 7.1 +/- 0.2%, respectively, P less than .01), exogenous insulin resulted in a lower postprandial glycemic response than tolazamide (P less than .001). Both agents comparably increased (P less than .01) fasting and integrated postprandial insulin concentrations. However, the initial rate of postprandial increase was greater with exogenous insulin (P less than .05). Tolazamide (P less than .05) but not exogenous insulin increased postprandial C-peptide concentrations. However, tolazamide did not improve the deficient early insulin release. Both agents (P less than .05) lowered postabsorptive hepatic glucose release (from 2.8 +/- 0.3 to 2.3 +/- 0.2 mg . kg-1 . min-1), but not to normal rates (1.8 +/- 0.1 mg . kg-1 . min-1).(ABSTRACT TRUNCATED AT 250 WORDS)

Fresh whole-blood buffy coats from American Red Cross volunteers were used to treat early type I diabetes. Attempts were made to adapt to human diabetic patients a protocol successfully used in prediabetic BB rats. Twenty-two type I diabetic patients (duration of disease less than 4 wk) were randomized to treatment or control groups; the treatment patients were given one buffy coat (approximately 0.6 X 10(9) T-lymphocytes) weekly for 5 wk. Plasma C-peptide (stimulated and unstimulated), insulin dose, and hemoglobin A1c were measured before and periodically after the treatment for 24 wk. The control group underwent the same studies. Although there were no significant differences for the parameters studied between the two groups, 2 of 12 patients in the treatment group underwent three complete (normal glycemia without insulin) temporary remissions. One of these patients was given a second course of transfusions after relapse from the first remission and developed a second complete remission that lasted 2 mo. No control patient had remissions during the 24-wk study. Although the future of adoptive immunotherapy in the treatment or prevention of diabetes is not known, several probable limitations of the current protocol, as discussed here, can explain the differences in results between this trial and the rodent studies.

This study addressed the controversial question of whether a negative-insulin-feedback loop exists in vivo. We utilized prehepatic insulin production, calculated by computerized deconvolution analysis of peripheral C-peptide concentration, as a measure of endogenous insulin secretion. Prehepatic insulin production was determined in 10 normal men who randomly underwent a control study and two additional studies involving different insulin infusion rates that achieved circulating insulin concentrations within the physiologic range during euglycemic clamps. The results demonstrate a dose-dependent suppression of prehepatic insulin production from 5.8 +/- 1.4 mU/min during the control study to 4.0 +/- 1.2 and 3.2 +/- 0.9 mU/min during plasma insulin levels of 34 +/- 4 and 61 +/- 6 microU/ml, respectively (P less than .05). Therefore, in contrast to recently reported results in vitro, insulin inhibits its own secretion in humans.

The aim of this study was to investigate the role of plasma glucagon levels on the blood glucose response to intravenous insulin administered continuously or in a pulsatile manner. Six type I diabetic patients proven to have no residual insulin secretion were investigated. Endogenous glucagon secretion was inhibited by a continuous intravenous infusion of somatostatin (100 micrograms/h) and replaced by exogenous infusions of the hormone at three different rates (7.5, 4.5, and 2.5 micrograms/h), resulting in three different plasma glucagon steady-state levels (i.e., approximately equal to 200, approximately equal to 130, and approximately equal to 75 pg/ml, respectively). Each subject, in random order and on different days, was infused intravenously with regular human insulin either continuously (0.17 mU X kg-1 X min-1) or with the same amount of insulin infused in a pulsatile manner (0.85 mU X kg-1 X min-1 during 2 min followed by 8 min during which no insulin was infused). At plasma glucagon levels approximately equal to 200 pg/ml, blood glucose rose from approximately 10 to approximately 13 mM without any difference between the two modalities of insulin infusion. For plasma glucagon levels approximately equal to 130 pg/ml, plasma glucose remained steady throughout the experiments, but during the last 40 min, plasma glucose levels were significantly lower when insulin was administered intermittently. This greater blood glucose-lowering effect of pulsatile insulin occurred earlier and was more pronounced for plasma glucagon levels averaging 75 pg/ml. We conclude that the greater hypoglycemic effect of insulin administered intravenously in a pulsatile manner in type I diabetics critically depends on plasma glucagon circulating levels.