Newly presenting maturity-onset diabetic subjects were put on diet and if, after 3-4 mo, their fasting plasma glucose continued greater than 6 mmol/L, they were randomized to three therapies: (1) continuing diet alone, (2) with additional sulfonylurea, or (3) with additional basal insulin supplement provided by ultralente insulin. Obese patients were also randomized to metformin therapy. The aim was to lower the fasting plasma glucose to less than 6 mmol/L and the degree to which this reduced the hemoglobin A1C (HbA1C) concentration was studied in 195 patients over 1 yr. Sulfonylurea and insulin similarly reduced (P less than 0.001) the fasting plasma glucose from 8.3 +/- 1.9 to 6.7 +/- 1.3 mmol/L (mean +/- 1 SD) and 8.6 +/- 2.2 to 6.8 +/- 1.4 mmol/L, respectively. This was accompanied by a significant reduction (P less than 0.001) of the HbA1C to the high normal range, from 9.1 +/- 2.1% to 7.8 +/- 1.2%, and from 9.1 +/- 1.9% to 8.1 +/- 1.3%, respectively, both values at 1 yr being significantly (P less than 0.05) lower than in patients randomized to diet alone. Patients randomized to diet alone had little change in fasting plasma glucose (8.6 +/- 1.8 to 9.3 +/- 2.3 mmol/L) or HbA1C (8.8 +/- 1.7% to 9.1 +/- 1.6%, respectively). Thus, the simple therapeutic aim of trying to reduce the fasting plasma glucose to less than 6 mmol/L is an effective means of reducing the HbA1C to a high-normal level. The HbA1C and fasting plasma glucose concentrations were similarly related for all three therapies (HbA1C [%] = 0.47 X fasting plasma glucose [mmol/L] + 4.7).(ABSTRACT TRUNCATED AT 250 WORDS)

Two-hundred and two pregnant women with impaired glucose tolerance were randomized to treatment with diet or diet and insulin by stratified selection. Self-monitoring of blood glucose was performed six times a day, 3 days/wk. Dietary treatment was considered inappropriate if fasting and postprandial blood glucose values exceeded 7 and 9 mmol/L, respectively, in which case insulin therapy was instituted. Insulin doses were adjusted according to blood glucose values, aiming at fasting and postprandial values below 5 and 6.5 mmol/L, respectively. There were no perinatal deaths. The two treatment regimens disclosed no differences regarding achieved degree of maternal blood glucose control, hemoglobin A1c at delivery, obstetric or neonatal complications, infant's size at birth including skin-fold thickness, or C-peptide concentration in cord serum. Routine treatment of pregnant women with mild carbohydrate intolerance with insulin seems unnecessary. However, 15 patients (14%) in the diet group needed insulin to achieve acceptable blood glucose control, underlining the importance of monitoring blood glucose to detect those who are at risk of developing overt diabetes.

In the present study, 12 patients with non-insulin-dependent diabetes mellitus (NIDDM) consumed eucaloric, mixed food diets on three consecutive days. Diets provided 50% of the calories as carbohydrate, 35% as fat, and 15% as protein. The percent of carbohydrate fed as complex (starches) and simple (mono- and disaccharides) varied among the 3 days. On day 1, the diet contained 80% of the carbohydrate as complex and 20% as simple (80/20); another contained 50% complex and 50% simple (50/50); and the final diet contained 20% of the carbohydrate as complex and 80% as simple (20/80). All simple carbohydrates represent naturally occurring sugars in fruits, vegetables, and dairy products. No refined sugars were added to any of the diets. The three experimental diets were randomly assigned using a 3 X 3 Latin square design. Blood was obtained hourly from 0800 to 1700 h for day-long glucose and insulin concentrations, and 24-h urine collections were made for the measurement of urine glucose. Mean (+/- SEM) day-long glucose concentrations were significantly greater for the 80/20 diet (2245 +/- 199 mg/dl X h, P less than 0.05) than for either the 50/50 (2030 +/- 157 mg/dl X h) or the 20/80 diets (2008 +/- 160 mg/dl X h). No significant differences were noted between the 50/50 and the 20/80 diets. In contrast, day-long insulin concentrations were not significantly different with 401 +/- 62, 370 +/- 50, and 369 +/- 60 mu U/ml X h on the 80/20, 50/50, and 20/80 diets, respectively. Twenty-four-hour urinary glucose excretion paralleled plasma glucose concentrations.(ABSTRACT TRUNCATED AT 250 WORDS)

Seven normal volunteers were studied on two different occasions during which 4-h pulsatile (PULS: 0.8 mU X kg-1 X min-1, 7.5 min of 15) and continuous (CONT: 0.4 mU X kg-1 X min-1) intravenous (i.v.) infusions of human insulin (Actrapid HM, Novo) were randomly compared. A euglycemic glucose clamp was performed and a 3-3H-glucose infusion was used for determination of endogenous glucose production (EGP) and metabolic clearance rate (MCR) of glucose. Plasma glucose was similar in both conditions; plasma insulin was stable at about 29 mU/L (CONT) and fluctuated between 10 and 45 mU/L (mean: 28, PULS). Exogenous glucose infused was 1.137 +/- 0.058 and 1.088 +/- 0.099 g X kg-1 X 4 h-1 in CONT and PULS, respectively (NS). EGP was totally suppressed in both conditions. Glucose MCR increased similarly to a maximum of 6.71 +/- 0.19 (CONT) and 6.79 +/- 0.59 (PULS) ml X kg-1 X min-1 during the fourth hour. C-peptide plasma levels remained stable, whereas plasma glucagon, free fatty acids, and 3-hydroxybutyrate were similarly suppressed in both tests. Thus, under these conditions, pulsatile and continuous insulin infusions have similar metabolic effects. These data contrast with those of Matthews et al. (1983) who reported that, at lower plasma concentrations (5-19 mU/L), pulsatile insulin had greater hypoglycemic effect than did continuous delivery. It is concluded that pulsatile insulin shows no greater activity under normoglycemic, moderately hyperinsulinemic conditions in man.

The impact on remission of normalizing blood glucose levels immediately after diagnosis of type I diabetes was studied in 14 adolescents. Accordingly, in this randomized prospective primary intervention study, 7 of the subjects (i.v. group) received insulin by continuous intravenous (i.v.) infusion via a portable preprogrammed system for 28-62 days and 7 (s.c. group) received conventional subcutaneous (s.c.) therapy. Before therapy, the two groups did not differ significantly with respect to glycosylated hemoglobin, fasting plasma C-peptide, or 24-h urinary C-peptide excretion. During the infusion period, the overall mean fasting plasma glucose (FPG) concentration for the i.v. group was 84 mg/dl with a mean coefficient of variation of 18 +/- 4% (mean +/- SD). During the comparable period for the s.c. group, the mean FPG was 253 mg/dl with a coefficient of variation of 30 +/- 20%. Twenty-four-hour urinary glucose excretions for the two groups were 0.29 +/- 0.06 (mean +/- SEM) and 59 +/- 11 g/day, respectively. Daily insulin requirements in the i.v. group decreased from 1.47 +/- 0.19 U/kg body wt/day at the start to 0.47 +/- 0.10 U/kg/day at the end of the infusion period. Notably, 10-25 days after the infusion period, 5 of 7 subjects experienced a further decrease to a low of 0.27 +/- 0.01 U/kg/day. The mean peak and low requirements in the s.c. group were 0.71 +/- 0.15 and 0.33 +/- 0.13 U/kg/day, respectively, with the only peak requirements being significantly different (P less than 0.01). No patient was able to discontinue insulin therapy.(ABSTRACT TRUNCATED AT 250 WORDS)

In a double-blind crossover study of 15 diabetic patients, elevated red blood cell (RBC) sorbitol levels were reduced by oral doses of the potent aldose reductase inhibitor, sorbinil (250 mg o.d.), to near-normal ranges. In diabetic rats with severe hyperglycemia, oral sorbinil (5 mg/kg) dramatically reduced (80-90%) sorbitol levels in tissues without affecting blood glucose; the RBC dose-response curve was similar to that in lens and sciatic nerve. In streptozotocin-treated rats with varying degrees of diabetes sorbitol levels in the lens, sciatic nerve, and RBC were elevated in proportion to the degree of hyperglycemia. RBC sorbitol levels in these animals were positively correlated with the levels in lens and sciatic nerve. These results establish that orally administered sorbinil is effective in lowering elevated sorbitol levels, and strongly suggest that the reduction seen in RBC sorbitol levels in human diabetic subjects is likely to reflect comparable effects of the drug in less accessible tissues associated with the long-term complications of diabetes.

We studied the effect of metformin therapy (1700 mg daily) on glucose tolerance, insulin secretion, and insulin binding to monocytes in 10 non-insulin-dependent diabetics (mean duration of disease 2.6 yr) who were treated for 4 wk with either metformin or placebo in a double-blind cross-over study. Metformin induced a significant decrease of glucose levels during an oral glucose load compared with placebo treatment (P less than 0.001). All patients studied showed normal or elevated basal insulin and C-peptide levels; their responses to an oral glucose load, however, were relatively hypoinsulinemic without any significant difference between metformin and placebo. Insulin binding to monocytes was nearly identical at all insulin concentrations tested in the placebo or metformin therapy phase. These data indicate that the glucose-lowering potency of metformin in non-insulin-dependent diabetics cannot be associated with changes in receptor number or affinity. It is suggested that metformin might have a positive influence on postreceptor defects in non-insulin-dependent diabetics.

The final results of a randomized controlled clinical trial of photocoagulation for diabetic maculopathy are reported, when all patients have been followed for at least 5 yr and some for as long as 7 yr. Ninety-nine patients with two similarly affected eyes had one eye chosen by a random procedure, treated with the xenon-arc photocoagulator; the untreated eyes remained as control. The mean visual acuity deteriorated by less than one line in treated eyes but by more than 2 lines in the controls (P less than 0.01). The difference in deterioration was greatest in patients whose initial vision was 6/6-6/9, and was not significant in those whose visual acuity was 6/36 or worse. Thirteen patients became blind in both eyes (visual acuity of 6/60 or less for 2 consecutive yearly assessments), 6 in the treated eye only, and 26 in the control eye only (P less than 0.01). Again the divergence between treated and control eyes was most marked in those whose initial vision was 6/6-6/9, (only one treated but 10 control eyes became blind). Hard exudates, microaneurysms, and hemorrhages improved more in the treated eyes (0.05 less than P less than 0.001) and more control eyes developed new vessels during the follow-up period. Twenty-three patients died during follow-up and another 16 failed to complete the study. Though the blood pressure of those who died was higher than those who survived (P less than 0.05 for both systolic and diastolic) no other medical abnormalities at entry had any clear effect on visual outcome or 5-yr survival. It is concluded that photocoagulation is of benefit in maintaining vision in diabetic maculopathy if the disease is not too far advanced.

A double-blind, randomized, placebo-controlled cross-over trial of the aldose reductase inhibitor sorbinil was undertaken in 15 patients (age 35-68 yr) with chronic painful diabetic neuropathy. Treatment was evaluated by subjective pain responses, clinical examination, vibration perception threshold, motor and sensory nerve electrophysiology, and cardiovascular reflex tests of autonomic nerve function. Among the many measurements, only pain, tendon reflex scores, and sural sensory potential amplitude improved significantly during sorbinil administration, while scores of clinical sensory examination deteriorated. Four patients experienced an idiosyncratic reaction that rapidly recovered on discontinuing the drug. This study suggests that aldose reductase inhibitor treatment with suggests that aldose reductase inhibitor treatment with sorbinil may have an effect on symptomatic diabetic neuropathy in man.

The immunogenicity of purified pork insulins (PPI) with and without (groups 1 and 2, respectively) trace contamination of beef insulin was contrasted with mixed beef pork insulin of lower purity (MBP, group 3) in 137 patients who had not previously been treated with insulin. Patients and physicians were blinded with regard to the species source of insulin and studies were conducted for a minimum of 1 yr. Antibody development to insulin was assessed by species-specific binding of 125I-insulin by acid charcoal extracted sera, as well as by measurement of insulin prebound to immunoglobulins by a polyethylene glycol precipitation method. NPH- and lente-treated individuals had equivalent antibody responses with regard to the rate of development of antibodies, and maximum immune responses to insulin. In all patient groups, antibody bound insulin as well as species-specific binding of 125I-insulin increased significantly over time (all P less than 0.01 for specific binding of pork and beef insulins SBP and SBB, as well as bound insulin). Maximum bound insulin and SBB as well as bound insulin and SBB over the entire course of the study were significantly greater in group 1 than in group 2 patients (both P less than 0.05). The rate of development and magnitude of antibodies' responses in both PPI-treated groups were significantly less than that seen in the MBP group (all P less than 0.01). New formation of antibeef proinsulin antibodies was seen in one patient from groups 1 and 3, but not in group 2. In all groups, insulin dose per day and fasting serum glucose concentrations increased by about 5 U/day and 10 mg/dl over 1 yr, but groups did not differ. MBP insulin used in these studies proved to be significantly less immunogenic than previously available Argentine pure beef insulin, purified by gel filtration. PPI containing even trace contamination of beef insulin was more immunogenic than PPI alone.

Cyclosporin A (CsA) is a unique immunosuppressive cyclic polypeptide that is currently being used, either alone or in combination with low-dose prednisone, to treat recipients of renal or pancreas allografts in clinical trials. CsA is very effective in preventing rejection of heart and renal allografts in rodents, but in nontoxic doses does not consistently prevent rejection of pancreas and islet allografts. Therefore, we tested low-dose CsA in various combinations with low-dose prednisone, azathioprine, or total lymphoid irradiation in rat heart, pancreas, and islet allograft models. Several combinations are synergistic and when administered continuously can indefinitely prevent rejection of heart allografts, but only delay rejection of pancreatic allografts, transplanted across a major histocompatibility barrier, CsA by itself prolonged the survival of islet allografts transplanted across a minor, but not a major, histocompatibility barrier. CsA and azathioprine had a synergistic effect in the minor histocompatibility barrier islet transplant model, but, in the nontoxic combinations tested, could not prevent rejection indefinitely. A randomized prospective trial comparing standard immunosuppressive therapy (ALG, prednisone, and azathioprine), with CsA and low-dose prednisone for clinical renal allotransplantation is ongoing at the University of Minnesota. Current actuarial 1-yr graft survival is 93% for CsA-treated patients (N = 48) and 81% for conventionally treated patients (N = 52). Patient survival is 98% for CsA and 100% for conventionally treated patients. A pilot trial of CsA in the clinical pancreas transplant program at the University of Minnesota is also underway. Since 1978, 46 pancreas transplants have been performed in 43 patients. Of 30 technically successful pancreatic allografts, 5 of 12 recipients treated with conventional immunosuppression and 6 of 18 recipients treated with CsA currently have functioning grafts and are insulin independent between 1 and 44 months after transplantation. The results of metabolic studies are similar in conventional and CsA-treated patients with functioning pancreas grafts. Since pancreas grafts may fail for reasons other than rejection, further observations are needed to ascertain the role of CsA in clinical pancreas transplantation.

Semisynthetic human Actrapid insulin (HAI, Novo, Copenhagen) was tested against porcine Actrapid insulin (PAI, Novo, Copenhagen) in 12 type I diabetic patients treated with continuous subcutaneous insulin infusion (CSII). In a double-blind crossover trial each patient received both types of insulin over a 3-wk period, respectively. No significant differences between HAI and PAI were observed in the following parameters: mean blood glucose levels (MBG) of 3-6 measurements per day (129 +/- 5 versus 125 +/- 4 mg/dl, means +/- SEM), mean maximal excursions of blood glucose during the day (107 +/- 6.6 versus 107 +/- 6.9 mg/dl), total daily insulin requirements (sum of basal and premeal insulin doses, 45.7 +/- 1.4 versus 42.7 +/- 1.4 U/day), and a mean of weekly determined hemoglobin A1c values (7.77 +/- 0.13 versus 7.83 +/- 0.14% of total hemoglobin); the frequency of mild hypoglycemic episodes was similar with the two insulins. Thus, under the controlled conditions of CSII, semisynthetic human and porcine regular insulin preparations are of identical efficacy in type I diabetic patients at near normoglycemia.

Twenty insulin-dependent diabetic patients participated in a 1-yr prospective randomized cross-over study comparing multiple subcutaneous injections (MSI) and continuous subcutaneous insulin infusion (CSII) complemented by home blood glucose monitoring. While 4 patients dropped out early, 16 patients completed the study. Patients had severe insulin deficiency documented by absent C-peptide response to glucagon stimulation. A marked improvement in control was observed when mean blood glucose and glycosylated hemoglobin A1 were compared with conventional therapy. No significant differences in the degree of metabolic control achieved, as measured by mean fasting, preprandial, and postprandial capillary blood glucose (CBG), M values, glycosylated hemoglobin A1 concentration, cholesterol and triglyceride levels were seen between MSI and CSII in the sixteen patients who completed the study. However, individual comparisons showed that fasting CBG and M-values were lower under CSII than MSI in seven patients (P less than 0.05). In contrast, two patients exhibited lower M values under MSI than under CSII (P less than 0.01), while for the remaining seven patients the results were similar. After completion of the study, two patients went back to conventional insulin therapy, seven patients remained on the pump, and seven patients chose to stay on MSI. In conclusion, on a long-term basis, the two methods can produce comparable levels of blood glucose and glycosylated hemoglobin in ambulatory insulin-dependent diabetics.

A single-blind, nonrandomized, placebo crossover clinical trial of an aldose reductase inhibitor, Alrestatin (AY 22, 284) was performed over a 4-mo period in nine patients with diabetic peripheral neuropathy. Most patients had subjective benefit, but objective measures of conduction were essentially unchanged. Substantial toxicity was evident, particularly photosensitive skin rash.

The use of electrophysiological (EP) tests as the primary basis for determining outcome in clinical trials of therapy for symptomatic diabetic polyneuropathy, and the frequently short duration of such trials, is based on assumptions at variance with the pathology and natural history of this disorder and with the evidence that the commonly employed EP tests predominantly reflect the status of the large myelinated nerve fibers. The course of painful, distal symmetrical, primarily sensory polyneuropathy was studied in nine chronic diabetics, aged 21--59 yr, selected for the absence of other forms of diabetic neuropathy, other causes of neuropathy, and other significant illness. All were treated with modifications of diet, insulin, and a daily multivitamin tablet, and, on a randomized basis, also received either placebo or myo-inositol tablets. Initially, and after 2, 4, and 6 mo, a standardized questionnaire was used to assess symptoms, and a standardized neurological examination and battery of EP tests were performed. A minimum of 6 mo was found necessary to assess the clinical course of this syndrome. Clinical improvement occurred in both legs and arms in four patients, as judged by improvement both in symptoms and in the extent of deficits in pinprick and temperature perception; abnormalities in sensory modalities mediated by large myelinated fibers, however, were generally unaltered after 6 mo. A nonuniform distribution of abnormal EP tests of sensory components of the commonly studied nerves of the leg and arm was demonstrated in the study group at the outset, and clinical improvement was not accompanied by evidence of any consistent pattern of improvement in the initially abnormal EP tests. A significant fraction of chronic diabetics with painful, distal symmetrical, primarily sensory polyneuropathy selected by standard criteria appear to have potential for clinical improvement over 6 mo, but primarily in sensory modalities that make it inappropriate to use the common EP tests as the primary basis of judging outcome.

We have studied the effects of mixed meals and dextrose intake on blood glucose and insulin delivery by the artificial pancreas in 24 insulin-dependent diabetics. A group of 12 patients had 3 mixed meals containing at random 20, 40, and 60 g of complex carbohydrate along with protein and fat; another group of 12 diabetics, comparable in weight, age, and duration of diabetes, received at random 20, 40, and 60 g of dextrose. Dextrose ingestion led to a higher initial blood glucose increase than did the mixed meal, but the duration of blood glucose increase lasted significantly longer after the mixed meal than after the dextrose load. The areas under the curves of hyperglycemia were not significantly different. There was a high (but not linear) correlation between the total amount of insulin delivered in order to restore initial blood glucose values and the amount of CHO consumed. There was no correlation with age, body weight, duration of diabetes, nor with the nature and order of administration of the CHO load; 5.1 +/- 1.6 to 13.7 +/- 2.1 units of insulin were needed for a period of 94 +/- 11 to 132 +/- 11 min. It is suggested that some of the data obtained in this study might be useful in the programming of an open-loop insulin infusion system.

Twenty-four volunteers, mean age 78, including eight mildly non-insulin-dependent diabetics, were randomly allocated to one of two groups and were fed (daily for 8 wk) 9 g of either chromium-rich brewers' yeast (experimental) or chromium-poor torula yeast (control). Before and after yeast supplementation, the serum glucose and insulin response to 100 g oral glucose was measured at 30 min intervals for 2 h. Fasting serum cholesterol, total lipids, and triglycerides were also determined. In the total experimental group (normals + diabetics) and in both the diabetic and nondiabetic experimental subgroups, glucose tolerance improved significantly and insulin output decreased after supplementation. Cholesterol and total lipids fell significantly after supplementation in the total experimental group. The cholesterol decrease was particularly marked in hypercholesterolemic subjects (cholesterol > 300 mg/dl). In the control group, no significant change in glucose tolerance, insulin, triglycerides, or total lipids was found. Cholesterol was significantly lowered in the nondiabetic but not in the diabetic group. Thus, chromium-rich brewers' yeast improved glucose tolerance and total lipids in elderly subjects, while chromium-poor torula yeast did not. An improvement in insulin sensitivity also occurred with brewers' yeast supplementation. This supports the thesis that elderly people may have a low level of chromium and that an effective source for chromium repletion, such as brewers' yeast, may improve their carbohydrate tolerance and total lipids. The improvement in serum cholesterol in some control subjects, as well as in the total experimental group, also suggests the presence of a hypocholesterolemic factor other than chromium in both brewers' and torula yeast.

In a diabetes detection survey carried out between 1962 and 1965, 2477 (1.1%) of 228,883 subjects had Clinistix-positive glucosuria after a carbohydrate-rich luncheon meal. Of these 2477, 578 displayed impaired tolerance to oral glucose without having manifest diabetes. From this group, 267 men were divided into five groups and subjected to the following treatments and controls: (a) diet regulation and 0.5 g tolbutamide t.i.d. (N = 49), annual oral glucose tolerance test (OGTT); (b) diet regulation and one placebo tablet t.i.d. (N = 48), annual OGTT; (c) diet regulation only (N = 50), annual OGTT; (d) no treatment (N = 61), annual OGTT; and (e) no treatment, OGTT at follow-up (N = 59 at follow-up). In addition, a control group was included comprised of men with normal OGTT (N = 52). At follow-up, 29% of those without diet regulation and medication (group e: N = 59) had developed diabetes. Of those on diet regulation, but without active medication (group b plus group c, N = 98), 13% had diabetes. No individual maintaining tolbutamide and diet regulation (N = 23) had progressed to diabetes. In this group, 80% of those later examined (N = 11) had serum tolbutamide concentrations in the therapeutic range. No individual with initially normal OGTT developed diabetes or impaired OGTT. The findings suggest that normal oral glucose tolerance signifies little risk of progress to impaired glucose tolerance and manifest diabetes, whereas impaired glucose tolerance is associated with a high risk of progression to diabetes. In addition, it seems possible that treatment with diet regulation, in combination with tolbutamide, may prevent or postpone progression from impaired glucose tolerance to manifest diabetes.

The kinetics of unlabeled porcine insulin were studied in 69 nondiabetic male subjects aged 18-83 yr with obesity indexes of 0.93 - 1.51 and in 12 maturity-onset diabetics age 46-78 yr with obesity indexes of 0.95-1.56 by using the euglycemic clamp technique. Analysis of the insulin kinetic data by using a mathematical model permitted the determination, for each individual, of steady state distribution masses and degradation rate constants. The individuals were grouped to allow comparison of the results on the basis of age, obesity index, or diabetes. The responses over a period of 120 min to an infusion and wash out of insulin show some transient as well as steady state differences with age, obesity, or diabetes. Analysis of these data by use of compartmental models leads to the conclusion that in the steady state the ratio of insulin in extravascular spaces to that in plasma (T/P) is decreased in the moderately obese group (26%) and in the diabetic group (17%) but increased in the older group (13%) when each is compared with the appropriate control. Since extravascular insulin includes both insulin bound to receptors and insulin in the interstitial fluid, the observed changes in the extravascular to plasma mass ratio most likely reflect changes in in vivo binding to receptors, although the magnitude of the change would be modified somewhat by changes in the size of the interstitial spaces relative to plasma. In addition, the rate of entry of new insulin into plasma (BSDR) was increased in the diabetic population (45%; P less than 0.02) as well as in the moderately obese group (27%) but was decreased somewhat in the older group (11%). The following general conclusions can be drawn from the results: The pattern of parameter changes seen with obesity is similar to that seen with maturity-onset diabetes. The decrease in T/P seen with obesity and with maturity-onset diabetes cannot be accounted for solely by changes in fasting plasma insulin levels in these populations. The pattern of changes seen in the older subjects is opposite that seen in the maturity-onset diabetics, which suggests that diabetes is a perturbation distinct from the normal aging process. Finally, the changes in the metabolism of insulin are not large, making it unlikely that they are the sole cause of the major alterations in glucose tolerance seen with aging, obesity, or diabetes.