In vitro culture of cells exfoliated in the urine by patients with diabetes mellitus.

As an approach to facilitate the understanding of the progression of diabetic renal disease, we assessed the urine of diabetic patients and normal volunteers for the presence of cells that could be cultured in vitro. The results suggest that both normal control subjects and diabetic patients, without clinically detectable microangiopathy, exfoliate few culturable cells into the urine. In contrast, diabetics with documented retinopathy but without nephropathy exfoliate substantially higher numbers of culturable cells (5.2 cells/100 ml urine), whereas diabetics with both retinopathy and advanced nephropathy exfoliate even greater numbers of culturable cells (50.8 cells/100 ml urine). The cells that are exfoliated and culturable can be divided into five distinct cell types based on morphology at the light microscope level. The exfoliated cells proliferate at clonal density after isolation from urine and are epithelial in appearance. These data suggest that the culture of cells from urine might have diagnostic value as an early indicator of diabetic renal disease and provide a convenient, noninvasive new source of human kidney epithelial cells.

Platelet lipoproteins. A comparative study with serum lipoproteins.

The nature of human platelet lipoproteins was studied in two series of experiments. In the first series, whole platelets were utilized for extraction of lipoproteins by three different methods: chloroform/methanol/phenol; saline; or sucrose-gradient ultracentrifugation of platelet homogenates. By polyacrylamide gel electrophoresis we were able to demonstrate the existence of lipoprotein in the extracts obtained by the last two methods. These lipoproteins were found not to share antigenic determinants with alpha and beta serum lipoproteins. The second series of experiments utilized platelets solubilized either in sodium deoxycholate or sodium dodecyl sulfate. The solubilized product was characterized by double immunodiffusion and polyacrylamide gel electrophoresis. The nonidentity between plasma and platelet lipoproteins previously demonstrated in the first series of experiments was confirmed. This nonidentity was also supported by a comparison between the apoproteins of purified serum lipoproteins and platelet proteins released after solubilization with sodium dodecyl sulfate. No identical protein fractions were found. Our results suggest that, unlike erythrocyte membrane lipoproteins, the platelet lipoproteins are structurally different from plasma lipoproteins.

Pharmacological strategies to prevent macrovascular disease in NIDDM.

The pathogenesis of atherosclerosis and vascular thrombosis in NIDDM is reviewed. Evidence that suggests a role for chronic hyperglycemia, in association with other vascular risk markers, is presented. Based on this framework, a multifactorial approach to the prevention of progression of macrovascular disease in NIDDM is discussed. Results from a recent consensus conference sponsored by the American Diabetes Association regarding approaches to glycemic regulation in people with NIDDM are reviewed. It is concluded that preventive approaches will materially alter the course of macrovascular disease, reduce health care costs, and improve the quality of life for people with NIDDM.

Intensive insulin therapy in type II diabetes: rationale and collaborative clinical trial results.

The rationale for intensive insulin therapy and results from major clinical trials in diabetes are reviewed. The Diabetes Control and Complications Trial (DCCT) has shown that intensive insulin therapy will prevent or delay the onset of retinopathy, nephropathy, and neuropathy in type I diabetes. The University Group Diabetes Program (UGDP) and the U.K. Prospective Diabetes Study (UKPDS) have addressed the issue of insulin versus oral agent or diet therapy in people with recently diagnosed type II diabetes. The UGDP showed that effective glycemic control could be achieved with intensive insulin therapy, but no effect on vascular end points was seen. Early data from the UKPDS also suggest that intensive insulin therapy may be more effective in lowering HbA1c toward normal than oral agents or diet. A pressing clinical problem is the question of the use of intensive insulin therapy in type II diabetic individuals who remain hyperglycemic despite pharmacological therapy. A Veterans Affairs Cooperative Study explored the feasibility of using intensive insulin therapy in 153 male type II diabetic patients with these characteristics. A 2% lowering of HbA1c was seen, with no increase in weight gain or in hypoglycemia. However, 40 of 153 patients (26.1%) had cardiovascular events during the 27-month trial; no difference in cardiovascular event rates was seen between the two treatment groups. A long-term multicenter collaborative trial is needed to assess the benefit:risk ratio of intensive insulin therapy for type II diabetic patients in whom pharmacological therapy failed to provide glycemic management.

Inhibition of labile aggregation-stimulating substance (LASS) and platelet aggregation in diabetes mellitus.

Irreversible second-phase aggregation of platelets in diabetic patients is prevented in vitro by 5, 8, 11, 14-eicosatetraynoic acid (TYA), a competitive inhibitor of the labile aggregation-stimulating substance (LASS) which is formed from arachidonic acid. Thus, inhibition of prostaglandin synthesis inhibits platelet aggregation in diabetic subjects. These findings indicate that platelets from diabetics are subject to control by intracellular mechanisms operative in the regulation of platelet function in normal individuals.

Do platelets have anything to do with diabetic microvascular disease?

It has been postulated that abnormal platelet and endothelial function may contribute to microangiopathy in diabetes mellitus. If this proposal is correct, alterations in platelet and endothelial function should be found before the appearance of vascular disease in insulin-dependent patients and in animal models of diabetes mellitus. This appears to be the case for the following: platelet aggregation, increased platelet production of the proaggregatory prostaglandin metabolite thromboxane, decreased endothelial production of the antiaggregatory prostaglandin prostacyclin, and decreased platelet survival. Insulin therapy will return some of these findings to normal. Platelet-plasma interactions that promote platelet aggregation and increased plasma levels of the platelet-specific protein beta-thromboglobulin have been reported in insulin-dependent diabetic patients who have not manifested vascular complications as well as in those with vascular complications. It has now been demonstrated in animal models that platelet microthrombi are found in small retinal vessels after months of experimental diabetes. Collectively, these findings demonstrate that alterations in platelet and endothelial function that favor thrombosis occur early in the diabetic state and may contribute to microvascular disease. There are several ongoing studies of antiplatelet agents in diabetic vascular disease that will provide clinical evidence bearing on the major postulate. Until these and other studies are completed, the platelet-endothelial story remains an attractive hypothesis in the genesis of diabetic microvascular disease.

Elevated glucose concentrations increase factor VIIIR:Ag levels in human umbilical vein endothelial cells.

To determine if endothelial cell metabolism is affected by the elevated glucose concentrations known to occur in diabetes mellitus, we measured cellular Factor VIIIR:Ag in endothelial cells grown under conditions of increased glucose concentration. The results of this study illustrate that, under cell culture conditions, an increase in glucose concentration results in increased cellular levels of Factor VIIIR:Ag. Specifically, a glucose concentration of 300 mg/dl resulted in a 23% increase in Factor VIIIR:Ag levels while a glucose concentration of 600 mg/dl resulted in a 54% increase in Factor VIIIR:Ag levels. These in vitro findings may relate to the increase in plasma Factor VIIIR:Ag levels known to occur in diabetic patients.

Predicting insulin requirements for a portable insulin pump using the Biostator. Evidence for reversible insulin resistance in poorly controlled type I diabetics.

Glycemic control was achieved in 14 patients with insulin-dependent diabetes mellitus (IDDM) by 36-48-h treatment with a recently marketed clinical model, Biostator glucose controller (Life Science Instruments, Miles Laboratories, Elkhart, Indiana). Control was maintained by continuous subcutaneous insulin infusion with a portable pump, programmed using infusion profiles from the Biostator. Control of glycemic excursion with the Biostator was variable among patients. This control, reflected by the M-value or a blood glucose index (mean of pre-, peak, and 2-h postmeal levels for four meals) of each patient, correlated directly with their prior glycemic control, as assessed by hemoglobin A1c (HbA1c) level (r = 0.66, P less than 0.01 and r = 0.82, P less than 0.005, for M-value and blood glucose index, respectively). Total insulin infused by the Biostator/24 h overpredicted the subcutaneous infusion dose required on day 2 of pump treatment (183 +/- 11%, P less 0.001). Therefore, these data were not used to program the portable pump. Instead, total insulin dose was estimated using a dietary glucose/insulin (G/I) ratio. This ratio, derived from dietary total available glucose, urine glucose, and insulin dose/24 h during depot insulin treatment, accurately estimated total insulin for pump infusion (97 +/- 4%). The basal infusion rate of the Biostator between 2400 and 0600 h also exceeded the subcutaneous infusion requirement and was reduced to 40% for the initial pump basal rate. The remainder of the insulin (total minus basal) was distributed as premeal boluses according to the Biostator infusion profile for meals.(ABSTRACT TRUNCATED AT 250 WORDS)

Enhancement of platelet aggregation by low-density lipoproteins from IDDM patients.

Low-density lipoprotein (LDL) is known to enhance platelet sensitivity to some aggregating agents. In this study, we observed that LDL isolated from patients with insulin-dependent diabetes mellitus (IDDM) enhanced thrombin-induced platelet aggregation to a greater extent than LDL isolated from matched controls (P less than .01). Thromboxane B2 production during aggregation was also significantly more enhanced by LDL isolated from IDDM than by control LDL (P less than .01). There was no difference in the lipid composition (free and esterified cholesterol, total phospholipids, and triglycerides) of LDL isolated from diabetic and control subjects. In contrast, the extent of glycosylation of LDL isolated from diabetic patients was significantly greater than that observed in LDL from normal subjects (P less than .01), and a positive correlation (r = .605, P less than .01) between the degree of LDL glycosylation and the rate of platelet aggregation was observed. LDL glycosylated in vitro enhanced thrombin-, collagen-, and adenosine 5'-diphosphate-induced platelet aggregation to a greater extent than control LDL (P less than .01). Although LDL glycosylated in vitro was taken up by platelets to a greater extent than control LDL (P less than .05), the lipid composition (free cholesterol and phospholipid) of platelets was not significantly changed. We postulate that an increased degree of glycosylation of LDL may enhance its uptake by platelets and lead to increased platelet reactivity to aggregating agents, probably by altering the structure of the platelet membrane. The enhancement of platelet aggregation by LDL may contribute to the accelerated development of atherosclerosis in diabetes mellitus.

Altered platelet function in diabetes mellitus.

An increased sensitivity of platelets to aggregation from ADP and epinephrine is described in diabetics with or without vascular disease. This sensitivity correlates with elevated levels on von Willebrand factor (vWF), which, in turn appears to be influenced by growth hormone. VWF activity correlates with previously described "plasma factor" activity. Platelets from diabetic subjects are more sensitive than platelets from normal subjects to arachidonic acid-induced aggregation. This sensitivity is abolished by aspirin, which is a prostaglandin synthetase (cyclo-oxygenase) inhibitor. Platelets from diabetc subjects synthesize increased amounts of PGE2-like material (iPGE) in response to ADP, epinephrine, collagen, and arachidonic acid. The latter finding suggests that a fundamental mechanism for increased platelet aggregation in diabetes is increased prostaglandin synthetase activity. Therapeutic endeavors that would lower growth hormone levels, vWF activity, and/or prostaglandin synthetase activity may be of benefit in the prophylaxis of diabetic vascular disease. Prospective studies are needed to explore these hypothesis, as are more studies on the precise mechanisms and platelet aggregation in diabetes mellitus.

Increased platelet arachidonic acid metabolism in diabetes mellitus.

Platelets obtained from some diabetic patients show enhanced in vitro platelet aggregation. This study sought to determine if platelet obtained from insulin-dependent diabetic subjects synthesize increased quantities of the labile aggregating substance, thromboxane A2 (TXA2), and if it may play a role in the enhanced platelet aggregation. Arachidonic acid (1 mM)-stimulated TXA2 synthesis, as determined via radioimmunoassay of its stable metabolite TXB2, was significantly greater (P less than 0.01, N = 12) in platelet-rich plasma obtained from diabetics compared with matched controls. Arachidonic acid-stimulated TXB2 synthesis in the diabetic platelet-rich plasma was positively correlated with the ambient fasting plasma glucose (r = 0.61, P less than 0.02, N = 15). Platelet aggregation induced by arachidonic acid (0.4-0.8 mM) was inhibited significantly less by 13-azaprostanoic acid (P less than 0.04, N = 14), a competitive antagonist of the actions of prostaglandin H2 or TXA2 on platelets, compared with matched controls. The results support the notion that platelets obtained from some insulin-dependent diabetic subjects manifest increased synthesis of TXA2, which may contribute to the enhanced platelet aggregation.

Platelet function during continuous insulin infusion treatment in insulin-dependent diabetic patients.

Patients with diabetes mellitus manifest increased in vitro platelet aggregation and increased synthesis of the proaggregant and vasoconstrictor, thromboxane A2 (TXA2). We studied the effects of continuous insulin infusion treatment on platelet aggregation and arachidonic acid (AA)-stimulated platelet TXA2 synthesis (15 and 30 s post-AA, 1 mM) in 16 type I diabetic patients. Strict glycemic control was induced with the Biostator for 2 days and maintained for 12-14 days with continuous subcutaneous insulin infusion (CSII). The average premeal plasma glucose level (4/day) fell from 184 +/- 15, before treatment, to 107 +/- 6 mg/dl on the final day (P less than 0.001). After control, platelet synthesis of TXA2, measured by radioimmunoassay of its stable metabolite, immunoreactive TXB2 (iTXB2), decreased in all patients (30 s: 276 +/- 31 versus 199 +/- 28 ng iTXB2/ml/5 X 10(5) platelets; P less than 0.05). The reduction in platelet iTXB2 synthesis (15 and 30 s) was greater in poorly controlled patients (HbA1c greater than 12%; N = 8), and for all patients the decrease in iTXB2 (15 and 30 s) was correlated with the prestudy HbA1c level (15 s: r = 0.6; P less than 0.01). In contrast, platelet aggregation responses did not improve during intensive insulin treatment. The ED50 for AA (dose producing 50% maximum aggregation at 1 min) was unchanged after 2 wk of treatment and the ED50 for aggregation induced by ADP fell significantly in patients with HbA1c greater than 12% (2.8 +/- 1.3 versus 1.2 +/- 0.6 microM; P less than 0.01).(ABSTRACT TRUNCATED AT 250 WORDS)

Thyrotoxicosis without elevated serum triiodothyronine levels during diabetic ketoacidosis.

Thyrotoxicosis with a normal serum triiodothyronine (T3) concentration has been described with a variety of acute and chronic illnesses occurring in association with thyrotoxicosis. We describe the first case to our knowledge of thyroxine (T4) toxicosis in a 16-year-old boy with diabetic ketoacidosis. Although the clinical manifestations of hyperthyroidism were mild, thyromegaly and persistent tachycardia suggested thyrotoxicosis. Serum T4 levels were elevated; however, the serum T3 level was normal. Measurement of reverse T3 (rT3) initially revealed an elevated level that decreased over several days of T3 levels increased into the toxic range. Peripheral conversion of T4 to T3 was apparently inhibited by diabetic ketoacidosis and there was a concomitant increase in rT3 levels, suggesting that conversion of T4 to rT3 was increased during acute ketoacidosis. Assessment of thyroid function based on serum T3 levels in diabetics may be misleading during ketoacidosis or uncontrolled diabetes.

Effects of gemfibrozil on triglyceride levels in patients with NIDDM. Hyperlipidemia in Diabetes Investigators.

OBJECTIVE: Patients with NIDDM have a two- to fourfold increased risk of macrovascular disease. The constellation of elevated TGs and decreased HDL cholesterol are recognized as risk factors and constitute the major dyslipidemia in NIDDM. We therefore sought to determine if gemfibrozil (600 mg b.i.d.) was effective in correcting the dyslipidemia of NIDDM. RESEARCH DESIGN AND METHODS: After 8 wk of placebo stabilization, 442 patients from 46 study centers were randomized to double-blind treatment, in a designated 2:1 ratio, 295 received gemfibrozil and 147 received placebo for 20 wk. The primary end point was plasma TG; secondary end points were TC, LDL cholesterol, VLDL cholesterol, HDL cholesterol, and HbA1c. No baseline differences were noted between groups in sex, age, weight, type of diabetic therapy, fasting plasma levels of TGs, HbA1c, or C-peptide. About two-thirds received oral hypoglycemic drugs, one-third insulin. RESULTS: TG fell 26.4% in the gemfibrozil group and rose 7.4% in the placebo group (P < 0.023), by an intent-to-treat analysis. When patients who were noncompliant or with inadequate data were excluded, similar results were found--a 30.4% fall with gemfibrozil and a 4.8% increase with placebo (P < 0.0001). TG levels fell within 4 wk and remained low for 20 wk (P < 0.001). Mean HDL cholesterol rose by 4 wk and increased further at 12 wk (8-12%), P < 0.0001. TC fell. We observed a significant rise in LDL cholesterol in both gemfibrozil- and placebo-treated groups, with no significant differences between these groups. Changes in HbA1c were similar in gemfibrozil and placebo groups. No differences were observed in responses in groups treated with insulin and or oral hypoglycemic drugs. Overall AEs that were clinically important occurred in 6.1% in the gemfibrozil group vs. 2.0% in the placebo group (NS). CONCLUSIONS: We conclude that gemfibrozil is an effective and safe agent in combating the dyslipidemia of NIDDM, irrespective of type of diabetic therapy.

Pathogenesis of atherosclerosis in diabetes mellitus.

Numerous studies have shown that patients with diabetes mellitus have accelerated atherosclerotic vascular disease, and major advances in understanding it pathogenesis have been made. Current suggestions are that endothelial injury may be the initial event in the genesis of atherosclerosis, followed by platelet adhesion and aggregation at the site of injury. In diabetes, evidence of endothelial dysfunction is present. Smooth muscle cell proliferation is an important pathologic finding in atherosclerosis. This process is stimulated by a platelet mitogen, which has been partially characterized. The mitogen has not been studied in diabetes. Lipid accumulation in the area of the atherosclerotic lesion is primarily in the form of intracellular and extracellular esterified cholesterol. In uncontrolled diabetes, elevated plasma low density lipoprotein levels and decreased plasma high density lipoprotein levels favor lipid deposition in large vessels. There is evidence of a thrombotic state in certain diabetic patients. Collectively, these abnormalities of endothelial, platelet, smooth muscle, lipoprotein, and coagulation behavior may be viewed as contributing to the problem of accelerated atherosclerosis in diabetes. A thorough understanding of the pathogenesis of this process aids in designing appropriate preventive therapeutic approaches.

Release of platelet plasminogen activator inhibitor 1 in whole blood is increased in patients with type II diabetes.

OBJECTIVE: To compare platelet plasminogen activator inhibitor 1 (PAI-1) release in type II diabetic patients and healthy control subjects. RESEARCH DESIGN AND METHODS: We studied a group of 27 diabetic patients and a group of 16 nondiabetic control subjects. Whole-blood platelet aggregation, defined as a decrease in platelet count during shaking (180 rpm) of blood samples at 37 degrees C, and plasma PAI-1 antigen concentrations were measured in parallel at time 0, 7.5, 15, 30, 60, 120, and 180 min. RESULTS: Platelet aggregation did not differ significantly between the two groups at any time period. However, the increase in plasma PAI-1 antigen concentration over basal levels at time 0 was higher for the group of diabetic patients when compared with their matched control subjects. The increment of PAI-1 antigen was 61.8 +/- 29.4 vs. 35.9 +/- 13.4 ng/ml (P < 0.005, means +/- SD) after 180 min for the diabetic and control subjects, respectively. Platelet PAI-1 release was correlated to very-low-density lipoprotein cholesterol and triglyceride plasma levels, but not to HbA1c levels. CONCLUSIONS: Platelets of patients with type II diabetes release significantly more PAI-1 than platelets of healthy subjects at the same level of platelet aggregation. This may contribute to enhanced thrombosis in diabetes.

Platelet plasminogen activator inhibitor 1 in patients with type II diabetes.

OBJECTIVE: To compare platelet plasminogen activator inhibitor 1 (PAI-1) concentration in type II diabetic patients and healthy control subjects. RESEARCH DESIGN AND METHODS: We studied a group of 12 diabetic patients whose disease was controlled by diet or sulfonylurea therapy and a group of 17 nondiabetic control subjects. All subjects were free of clinically advanced vascular disease. PAI-1 antigen concentrations were measured in 5 x 10(8) isolated platelets, which were lysed by 1% Triton X-100. RESULTS: Mean platelet PAI-1 was significantly higher in diabetic patients (264 +/- 83 ng/5 x 10(8) platelets) compared with control subjects (202 +/- 71 ng/5 x 10(8) platelets) (P < 0.05). A significant independent positive correlation was found between platelet PAI-1 concentrations and fasting plasma specific insulin levels in the diabetic patients (r = 0.63, P = 0.03). CONCLUSIONS: These findings suggest that 1) a higher platelet PAI-1 concentration may contribute to enhanced thrombosis in type II diabetes and 2) megakaryocyte PAI-1 synthesis may be under the control of insulin.

Veterans Administration Cooperative Study on antiplatelet agents in diabetic patients after amputation for gangrene: II. Effects of aspirin and dipyridamole on atherosclerotic vascular disease rates.

We report the results of a randomized multicenter clinical trial on the effects of aspirin plus dipyridamole versus placebo on major vascular end points in 231 non-insulin-dependent diabetic men with either a recent amputation for gangrene or active gangrene. Primary end points were death from atherosclerotic vascular disease plus amputation of the opposite extremity for gangrene. There were 24 atherosclerotic deaths in the drug treatment group (21.8%) and 23 in the placebo group (19.0%). There were 22 patients in the drug treatment group (20.0%) and 29 patients in the placebo group (24.0%) with opposite-side amputations. Survival curve analyses revealed little difference between these groups for major vascular end points, total mortality, all amputations, or myocardial infarctions. The most noteworthy group difference was observed for cerebrovascular end points (strokes and transient ischemic attacks), with an incidence of 8.2% (9 patients) in the drug treatment group and 19.0% (23 patients) in the placebo group. We conclude from this study that antiplatelet agents have no effect on the primary vascular end points, vascular deaths and/or amputation of the opposite extremity, in this population. Similarly, no effects were seen on secondary vascular end points, except for a suggestion of protection versus strokes and transient ischemic attacks. However, this finding must be interpreted with caution, since it is a secondary end point and was found only after multiple analyses of the data.

Hemostatic alterations with exercise conditioning in NIDDM.

Various parameters of coagulation and fibrinolysis were measured in 13 men (aged 54 +/- 3 yr) with non-insulin-dependent diabetes mellitus (NIDDM) before and after 12-14 wk of exercise training. Subjects exercised for 30 min 3 times/wk at 70% of maximum O2 consumption (VO2max). Training increased VO2max by 12.5% but did not alter body weight, relative body fat, blood pressure, cholesterol, triglycerides, or high-density lipoprotein cholesterol. Slight downward trends were apparent for fasting glucose and insulin, but glycosylated hemoglobin was unchanged. There were no changes in coagulation parameters of plasminogen, hematocrit, or alpha 2-antiplasmin. Plasma fibrinogen (303 +/- 24.2 vs. 256 +/- 12.3 mg/dl) and fibronectin (380 +/- 41.9 vs. 301 +/- 22.2 micrograms/ml) were significantly reduced (P less than 0.02) by exercise conditioning. Three assays of fibrinolytic activity (tissue plasminogen activator, euglobulin lysis time, and an isotopic measure of fibrinolysis) confirmed that neither basal fibrinolysis nor the fibrinolytic responses to venous occlusion and maximal exercise were significantly altered. Exercise conditioning may have antithrombotic effects in NIDDM by reducing plasma fibrinogen and fibronectin. Although the significance of the fall in fibronectin awaits further studies, the reduction in plasma fibrinogen gives a rationale for the use of exercise training in men with NIDDM.

Response to intensive therapy steps and to glipizide dose in combination with insulin in type 2 diabetes. VA feasibility study on glycemic control and complications (VA CSDM).

OBJECTIVE: The feasibility study for the VA Cooperative Study on Glycemic Control and Complications in Type 2 Diabetes (VA CSDM) prospectively studied 153 insulin-requiring type 2 diabetes patients, randomized between an intensively treated arm and a standard treatment arm during a mean follow-up of 27 months. The glycemic response to each of the progressive, sequential phases of insulin treatment was assessed, along with the incidence of hypoglycemic reactions and the relative efficacy of different doses of glipizide in combination with fixed doses of insulin. RESEARCH DESIGN AND METHODS: Five medical centers participated; half of the patients were assigned to the intensive treatment arm aiming for normal HbA1c levels. Age of patients was 60 +/- 6 years, duration of diabetes 8 +/- 3 years, and BMI 30.7 +/- 4 kg/m2. A four-step management technique was used, with patients moving to the next step if the operational goals were not met: Phase I, evening intermediate or long-acting insulin; phase II, added day-time glipizide; phase III, two injections of insulin alone; and phase IV, multiple daily insulin injections. Home glucose monitoring measurements were done twice daily and at 3:00 A.M. once a week. Hypoglycemic reactions and home glucose monitoring results were recorded and counted in each of the treatment phases. RESULTS: Baseline HbA1c was 9.3 +/- 1.8%, and fasting plus serum glucose was 11.4 +/- 3.3 mmol/1. Fasting serum glucose fell to near normal in phase I, and remained so in the other treatment phases. An HbA1c separation of 2.1% between the arms was maintained during the course of the study, while the intensive arm kept HbA1c levels below 7.3% (P = 0.001). Most of the decrease in HbA1c occurred with one injection of insulin alone (phase I, -1.4%) or adding day-time glipizide (phase II, -1.9% compared with baseline). HbA1c did not decrease further after substituting two injections of insulin alone, with twice the insulin dose. Multiple daily injections resulted in an additional HbA1c fall (-2.4% compared with baseline). However, two-thirds of the patients were still on one or two injections a day at the end of the study. Changes in home glucose monitoring levels paralleled those of the HbA1c, as did the increments in number of reported hypoglycemic reactions, virtually all either "mild" or "moderate" in character. For the combination of glipizide and insulin (phase II), the only significant effect was obtained with daily doses up to 10 mg a day; there were no significant additional benefits with up to fourfold higher daily doses, and HbA1c levels had an upward trend with doses > 20 mg/day. CONCLUSIONS: A simple regime of a single injection of insulin, alone or with glipizide, seemed sufficient to obtain clinically acceptable levels of HbA1c for most obese, insulin-requiring type 2 diabetes patients. Further decrease of HbA1c demanded multiple daily injections at the expense of doubling the insulin dose and the rate of hypoglycemic events. In combination therapy, doses of glipizide > 20 mg/day offered no additional benefit.