Skip to main content
Device Global Header

Subscribe
or Renew

Advanced Search

Original Article

Comparison of Insulin Regimens in Patients with Non-Insulin-Dependent Diabetes Mellitus

List of authors.
  • Hannele Yki-Järvinen, M.D.,
  • Marjut Kauppila, M.D.,
  • Eila Kujansuu, M.D.,
  • Jorma Lahti, M.D.,
  • Tapani Marjanen, M.D.,
  • Leo Niskanen, M.D.,
  • Sulo Rajala, M.D.,
  • Leena Ryysy, M.D.,
  • Seppo Salo, M.D.,
  • Pentti Seppälä, M.D.,
  • Timo Tulokas, M.D.,
  • Jorma Viikari, M.D.,
  • Jukka Karjalainen, M.D.,
  • and Marja-Riitta Taskinen, M.D.

Abstract

Background.

Insulin is widely used to improve metabolic control in patients with non-insulindependent diabetes mellitus (NIDDM), but there is no consensus about the optimal regimen of insulin treatment.

Methods.

We treated 153 patients with NIDDM for three months with five regimens: (1) oral hypoglycemic drug therapy plus NPH insulin given at 7 a.m. (the morning-NPH group), (2) oral hypoglycemic drug therapy plus NPH insulin given at 9 p.m. (the evening-NPH group), (3) NPH and regular insulin (ratio, 70 units to 30 units) given before breakfast and dinner (the two-insulin-injection group), (4) NPH insulin at 9 p.m. and regular insulin before meals (the multiple-insulin-injection group), and (5) continued oral hypoglycemic drug therapy (the control group).

Results.

The mean (±SE) value for glycosylated hemoglobin decreased similarly in all four insulin-treatment groups (1.7±0.3, 1.9±0.2, 1.8±0.3, and 1.6±0.3 percent, respectively). The decrease was significantly greater in these four groups than in the control group (0.5±0.2 percent; P<0.001 vs. all insulin-treated groups). Weight gain was significantly less (1.2±0.5 kg) in the evening-NPH group than in the other insulin-treatment groups (2.2±0.5 kg in the morning-NPH group, 1.8±0.5 kg in the two-insulin-injection group, and 2.9±0.5 kg in the multiple-injection group; P<0.05). In addition, the increment in the mean diurnal serum free insulin concentration was 50 to 65 percent smaller in the evening-NPH group than in the other insulin-treatment groups. Subjective well-being improved significantly more in the insulin-treatment groups than in the control group (P<0.001).

Conclusions.

In patients with NIDDM who are receiving oral hypoglycemic drug therapy, the addition of NPH insulin in the evening improves glycemic control in a manner similar to combination therapy with NPH insulin in the morning, a two-insulin-injection regimen, or a multiple-insulin-injection regimen, but induces less weight gain and hyperinsulinemia. The data thus suggest that patients with NIDDM do not benefit from multiple insulin injections and that nocturnal insulin administration appears preferable to daytime administration. (N Engl J Med 1992;327:1426–33.)

Introduction

SOME patients with non-insulin-dependent diabetes mellitus (NIDDM) achieve good glycemic control when they are first treated with an oral hypoglycemic drug, but only approximately 50 percent continue to have satisfactory glycemic control after 10 years.1 These patients are then often treated with insulin alone or in combination with an oral hypoglycemic drug, but the optimal mode of insulin administration remains controversial.

The results of the available insulin-treatment studies in these patients are difficult to interpret for several reasons. First, the criteria for the selection of patients were variable, as was the duration of treatment. Second, the applicability of the proposed regimen has seldom been tested outside a specialized treatment center.1 , 2 Also, factors such as the feasibility of the regimen and the patient's subjective sense of well-being have not generally been addressed, although such factors may be particularly important in patients for whom insulin therapy is not obligatory.

The overproduction of glucose by the liver is the main cause of fasting hyperglycemia in patients with NIDDM.3 After a meal, hyperglycemia compensates for insulin resistance so that the absolute rate of glucose use does not change.4 Although insulin sensitivity improves during insulin therapy,5 the stimulatory effect of insulin on glucose use is counterbalanced by a diminished mass-action effect of glucose.4 Thus, both in the fasting state and postprandially, inhibition of excessive hepatic glucose production may be the most important effect of insulin therapy.

Since hepatic glucose production is quantitatively most abnormal at night, nocturnal insulin therapy may theoretically be more advantageous than daytime insulin therapy.6 However, the efficacy of nocturnal treatment, as compared with daytime insulin treatment or other insulin-treatment regimens, has not been determined in controlled studies.

This study was undertaken to compare four different insulin-treatment regimens with continued oral hypoglycemic drug therapy in patients with NIDDM. For this purpose, 153 patients with NIDDM were randomly treated with an oral hypoglycemic drug plus NPH insulin in the evening or the morning, a two-insulin-injection regimen, or a multiple-insulin-injection regimen without any oral hypoglycemic drug, or they were given maximal doses of oral hypoglycemic drugs.

Methods

Design

Selection of Patients

The study consisted of a six-week run-in period and a three-month treatment period. The patients were recruited at six hospitals, and the following inclusion criteria were used: (1) age from 40 through 70 years, (2) current body-mass index (calculated as the weight in kilograms divided by the square of the height in meters) less than 35 and stable weight for at least six months, (3) fasting blood glucose concentration greater than 144 mg per deciliter (8 mmol per liter), (4) the presence of diabetes for more than three years, (5) previous therapy with a maximal dose of either glipizide (15 mg per day) orglyburide (10.5 mg per day) alone or in combination with metformin (0.5 to 2.0 g per day), and (6) fasting serum C-peptide concentration greater than 0.33 nmol per liter (reference range, 0.33 to 0.69). The criteria for exclusion were congestive heart failure, myocardial infarction, or stroke during the past six months; epilepsy or other severe disease; liver disease; nephropathy as determined by a serum creatinine concentration greater than 1.3 mg per deciliter (120 μmol per liter) or by a urinary excretion of albumin ≥ 300 mg per 24 hours; proliferative retinopathy or severe maculopathy; previous insulin therapy for more than two weeks (to avoid any possible interfering effects of insulin antibodies); excessive alcohol consumption; night work; serum triglyceride concentration greater than 5 mmol per liter; and the presence of islet-cell antibodies. The study was approved by the institutional review committee at each hospital, and informed consent was obtained from the patients.

Study Protocol

Patients entering the study met with the doctor and the diabetes nurse six weeks before the start of treatment. Conventional therapy (diet and exercise) was intensified, and the patients were instructed in techniques of home glucose monitoring. The patients were instructed to measure their fasting blood glucose concentrations, to record any symptomatic hypoglycemic episodes each day, and to measure their blood glucose concentrations once each week at the following times: before and 1 1/2 hours after breakfast, lunch, and dinner, at 10 p.m., and at 4 a.m. and before breakfast the next morning. Hypoglycemia was evaluated by counting the number of episodes of symptomatic hypoglycemia and the number of times the blood glucose concentration fell below 72 mg per deciliter (4 mmol per liter) during home glucose monitoring. The patients then visited the laboratory within one week for measurements of fasting blood glucose; islet-cell antibodies; glycosylated hemoglobin; serum concentrations of creatinine, C peptide, triglycerides, and liver enzymes; and urinary albumin excretion.

Three weeks before the start of treatment, the patients met again with the doctor and the diabetes nurse. The results of the laboratory tests were checked, and if they were acceptable, the patients were assigned to a treatment group by the coordinating center.

Randomization

Table 1. Table 1. Base-Line Clinical Characteristics of the Five Groups of Patients with NIDDM.*

The patients at each hospital were assigned to one of five groups,7 as follows: (1) oral hypoglycemic drug therapy (unchanged from that given before the study) plus NPH insulin (isophane insulin suspension; Protaphan Human Pen [100 U per milliliter], Novo Nordisk, Espoo, Finland) given before breakfast (the morning-NPH group); (2) oral hypoglycemic drug therapy (likewise unchanged) plus NPH insulin given at 9 p.m. (the evening-NPH group); (3) NPH insulin and regular insulin, in a ratio of 70 units to 30 units (Demiphan Human Pen [100 U per milliliter], Novo Nordisk), given before breakfast and dinner (at 4 p.m.) with oral hypoglycemic drug therapy discontinued (the two-insulin-injection group); (4) a multiple-insulin-injection group, with NPH insulin given at 9 p.m. and regular insulin (Actrapid Human Pen [100 U per milliliter], Novo Nordisk) given before breakfast, lunch, and dinner (with oral hypoglycemic drug therapy discontinued); and (5) a control group in which oral hypoglycemic drug therapy was continued unchanged. The following characteristics were considered during randomization: age, sex, body-mass index, duration of diabetes, fasting blood glucose, type of oral hypoglycemic therapy, use of diuretic agents and beta-adrenergic—antagonist drugs, and use of other drugs. The base-line physical and clinical characteristics of the study groups are shown in Table 1.

Insulin Therapy

At the start of the study (week 0), all the patients were admitted to the hospital for five days, during which time they followed a weight-maintaining diet in which approximately 50 percent of the calories were carbohydrate, 30 percent were fat, and 20 percent were protein. On day 1, a fasting blood sample was obtained to determine concentrations of glycosylated hemoglobin and serum lipoproteins. To obtain a diurnal profile of glucose and insulin levels, blood glucose and serum free insulin concentrations were determined before and 1 1/2 hours after breakfast (7:30 a.m.), lunch (11:30 a.m.), and dinner (4:30 p.m.) and at 10 p.m. and 4 a.m. All medications, including oral hypoglycemic drugs, were continued unchanged. Blood pressure was measured twice during the first day, 1 1/2 hours after both breakfast and lunch, after the patient had been sitting for at least 15 minutes.

On day 2, insulin therapy was started. The initial dose in the combination-therapy groups was the number of units that equaled the patient's mean diurnal blood glucose concentration (in millimoles per liter). In the two-insulin-injection group, it was 0.25 U per kilogram of body weight if the diurnal blood glucose concentration was 180 mg per deciliter (10 mmol per liter), and 4 U of insulin was added for each millimole per liter by which the blood glucose concentration exceeded that value. In the multiple-insulin-injection group, two thirds of the daily dose was given before meals and one third at bedtime; in the two-injection group, two thirds was given at 7 a.m. and one third at 4 p.m. The insulin dose was adjusted subsequently to achieve normoglycemia, defined as a fasting blood glucose concentration below 126 mg per deciliter (7 mmol per liter) and a postprandial blood glucose concentration below 180 mg per deciliter. During the five-day period of hospitalization, the patients were taught about giving themselves the insulin injections and about insulin preparations, diet, hypoglycemia, exercise, and foot care. They were instructed to inject the doses of regular insulin and NPH plus regular insulin subcutaneously in the abdomen 30 minutes before meals, as appropriate, and to inject the doses of NPH insulin in the thigh at 9 p.m.

The patients were seen two weeks, one month, and two months after the start of the treatment period. At these visits, the results of the daily fasting and weekly diurnal blood glucose measurements and the episodes of hypoglycemia were reviewed, body weight and fasting blood glucose were measured, and the insulin dose was adjusted if necessary in order to achieve the blood glucose values described above. After three months, all the patients were readmitted to the hospital to repeat the tests performed on day 1. In addition, their attitudes and subjective sense of well-being were evaluated with a questionnaire (see the Results section).

Study Discontinuation

Four patients discontinued participation in the study. One patient in the multiple-injection group was found to have a renal carcinoma, one patient in the control group had a foot infection and toe amputation, one patient in the two-injection group had a psychiatric illness, and one patient declined to be included in the control group.

Analytical Methods

Home measurements of blood glucose were performed with the Hypocount Home Blood Glucose Monitor (Oriola, Espoo, Finland). Values for glycosylated hemoglobin were determined with high-pressure liquid chromatography8 (reference range, 4 to 6 percent). Serum free insulin (Phadeseph insulin radioimmunoassay kit, Pharmacia, Uppsala, Sweden) and C-peptide concentrations were measured by radioimmunoassay.9 Serum lioprotein fractions were separated by sequential flotation in an ultracentrifuge (Beckman L8–70, Beckman Instruments, Palo Alto, Calif.).10 Serum cholesterol and triglyceride concentrations were measured enzymatically with kits obtained from Boehringer–Mannheim (Mannheim, Germany). Serum liver enzymes, serum creatinine, and blood glucose were measured with standard techniques in each hospital. Islet-cell antibodies were determined with an immunofluorescent assay.11

Statistical Analysis

The distribution of data was analyzed and changes between the groups compared with use of BMDP programs12 for detailed data description (program 2nd) and analysis of variance (program 7D), followed by the Bonferroni test. Changes within a group were calculated by analysis of variance for repeated measures (program 2V). Data from questionnaires and information on the frequency of symptomatic hypoglycemia were analyzed by multiway tables (program 4F), followed by Pearson's chi-square statistic for comparison between groups. Linear regression analyses were performed by simple and multiple linear regression analysis (programs 8D and 1R, respectively). All statistical tests were two-tailed.

Results

Insulin Doses

The mean (±SE) doses of NPH insulin in the two groups receiving an oral hypoglycemic drug plus NPH insulin were similar at three months (morning-NPH group, 19±1 U per day; evening-NPH group, 20±2 U per day). Also, the total doses of insulin were comparable in the two-injection group (total, 43±2 U per day; 27±2 U per day before breakfast, and 16±1 U per day before dinner) and the multiple-injection group (total, 45±3 U per day; regular insulin, 9.5±0.7, 8.9±0.7, and 9.4+0.4 U per day before breakfast, lunch, and dinner, respectively; and 17±1 U of NPH insulin per day at 9 p.m.).

Glycemic Control

Glycosylated Hemoglobin

Figure 1. Figure 1. Glycemic Control during the Six-Week Run-in Period and during Treatment in Patients with NIDDM.

The curves for mean diurnal blood glucose levels represent 70 percent of the requested number of measurements in the five groups: the control group (+), the morning-NPH group (○), the evening-NPH group (●), the two-injection group (□), and the multiple-injection group (■). During the run-in period, the mean value for glycosylated hemoglobin decreased by 0.30±0.09 percent (P<0.001), the mean diurnal blood glucose concentration by 16±4 mg per deciliter (0.89±0.24 mmol per liter) (P<0.001), and the mean fasting blood glucose concentration by 31±4 mg per deciliter (1.7±0.2 mmol per liter) (P<0.001). Each diurnal profile included blood glucose measurements before and 1 1/2 hours after breakfast, lunch, and dinner and at 10 p.m. and 4 a.m. The mean diurnal blood glucose concentration was significantly lower during treatment in each insulin-treatment group than in the control group (P<0.01 to P<0.001). P<0.01 for the difference in values for glycosylated hemoglobin between the insulin-treatment groups and the control group at 12 weeks. To convert values for glucose to millimoles per liter, multiply by 0.05551.

The percentage of glycosylated hemoglobin and the mean diurnal blood glucose concentrations (measured at home) were comparable during the run-in period in the five groups, with glycemic control improving slightly in each group during this period. During the three months of therapy, the values obtained for these measures decreased to a similar extent in all the insulin-treatment groups but changed little in the control group (Fig. 1).

Home Glucose Monitoring

Figure 2. Figure 2. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Determined by Home Glucose Monitoring.

The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the blood glucose concentration in the insulin-treatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for glucose to millimoles per liter, multiply by 0.05551.

In the morning-NPH group, the concentrations of blood glucose measured at home before and after lunch and before dinner, but not in the evening or at night, decreased significantly after three months as compared with those in the control group. In the evening-NPH group, the blood glucose concentration decreased as compared with the control group at 4 a.m. and before breakfast (Fig. 2). The changes in blood glucose concentrations in the two-injection group and the multiple-injection group were virtually identical; the concentrations in these two groups were significantly lower before and after both lunch and dinner than in the control group (Fig. 2).

Hospital Diurnal Profile

Figure 3. Figure 3. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.

The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the blood glucose concentration in the insulin-treatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for glucose to millimoles per liter, multiply by 0.05551.

The diurnal changes in blood glucose concentrations measured in the hospital were comparable with those measured at home, except in the multiple-insulin-injection group, in which the decrease in blood glucose after lunch was more pronounced than at home (Fig. 2 and 3). At three months, the mean (±SE) diurnal blood glucose concentration averaged 204±11 mg per deciliter (11.3±0.6 mmol per liter) in the control group and 160±4 mg per deciliter (8.9±0.2 mmol per liter) in the four insulin-treated groups (P<0.001).

Diurnal Serum Free Insulin Concentrations

Table 2. Table 2. Effect of Insulin Therapy on Body Weight, Glycemic Control, Blood Pressure, and Serum Lipid Concentrations before and after Three Months of Treatment in Patients with NIDDM.

At base line, the mean diurnal serum free insulin concentrations were comparable among the groups, averaging 21±1 μU per milliliter (154±8 pmol per liter) (Table 2). During the three months of treatment, the mean diurnal free insulin concentration increased by 29 percent (in the morning-NPH group), 14 percent (in the evening-NPH group), 39 percent (in the two-injection group), and 36 percent (in the multiple-injection group), as compared with the values at base line. When the insulin-treatment groups were compared with each other, the increment in the evening-NPH group was significantly less than in the other three groups (P<0.05).

Figure 4. Figure 4. Changes in Mean Diurnal Serum Free Insulin Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.

The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the serum free insulin concentration in the insulintreatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for insulin to picomoles per liter, multiply by 7.2.

The increment in the mean diurnal serum free insulin concentration in the morning-NPH group was due to a significant increase before lunch and dinner, whereas in the evening-NPH group there was a significant increase at 4 a.m. (Fig. 4). In the two-injection group, insulin concentrations increased significantly after breakfast, before lunch, after dinner, and at 10 p.m. (Fig. 4). In the multiple-injection group, insulin concentrations also increased significantly during the day, but not during the night (Fig. 4).

Changes in Body Weight

The patients in all the insulin-treatment groups gained weight during the three-month period (Table 2). The smallest increment in body weight occurred in the evening-NPH group (mean, 1.2±0.5 kg; range, -3.0 to 6.9), and the largest in the multiple-injection group (2.9±0.5 kg; range, -2.1 to 8.3; P<0.05 for the comparison with the evening-NPH group). The control group lost 0.9±0.4 kg of body weight (range, -8.2 to 3.0; P<0.05 for the comparison with base line).

In the four insulin-treatment groups, the change in body weight was inversely correlated with the change in glycosylated hemoglobin (r = -0.20, P<0.001) and positively correlated with the change in the mean diurnal serum free insulin concentration (r = 0.23, P<0.001), suggesting that weight gain occurred concomitantly with improved glycemic control and hyperinsulinemia. In the control group, the change in weight was positively correlated with the change in glycemic control (r = 0.52, P<0.01).

Hypoglycemia

Compliance with Home Glucose Monitoring

The total number of glucose measurements during the weekly diurnal profiles measured at home (10,440 measurements) was 70 percent of the requested number. The frequency of the measurements of diurnal blood glucose concentrations was similar in all groups (69 to 79 percent for those made while fasting, after breakfast, before and after lunch, and before and after dinner; 67 percent for those made at 10 p.m.; and 46 percent for those made at 4 a.m.).

Low Blood Glucose Concentrations Measured at Home

The frequency of blood glucose concentrations below 72 mg per deciliter (4 mmol per liter) was 2 percent, with no significant differences between the insulin-treatment groups (data not shown). Low blood glucose concentrations occurred most frequently before lunch in the morning-NPH and evening-NPH groups (6 and 3 percent, respectively), after breakfast in the two-injection group (3 percent), and before dinner in the multiple-insulin-injection group (3 percent).

Symptomatic Hypoglycemia

During the three-month treatment period, the total number of symptomatic hypoglycemic episodes reported per patient was 3±1 in the morning-NPH group (P<0.05 for the comparison with the control group), 1±1 in the evening-NPH group (P not significant), 4±1 in the two-injection group (P<0.05), 2±1 in the multiple-insulin-injection group (P not significant), and none in the control group. There was no significant difference between the number of reported hypoglycemic episodes in the four insulin-treatment groups.

Changes in Blood Pressure and Serum Lipoprotein Concentrations

In the control group, the systolic and diastolic blood pressures decreased significantly by 10±4 and 7±3 mm Hg, respectively, during the three-month period. In this group, the decrease in mean blood pressure (8±3 mm Hg) correlated with weight loss (r = 0.43, P<0.05). Blood pressure did not change in any insulin-treatment group (Table 2). Serum concentrations of very-low-density lipoprotein (VLDL) triglyceride decreased 13 to 28 percent in the insulin-treatment groups and 7 percent in the control group. Serum concentrations of total, low-density lipoprotein (LDL), and high-density lipoprotein (HDL) cholesterol did not change in any group.

Attitudes and Subjective Well-Being

Most patients (from 57 to 77 percent in the various groups) considered home blood glucose monitoring easier than they had initially thought. The patients in the multiple-insulin-injection group found insulin injections most difficult, although the difference between that group and the other insulin-therapy groups was not significant (insulin injection was easier than expected for 62 percent of patients in the multiple-injection group, 94 percent in the morning-NPH group, 85 percent in the evening-NPH group, and 83 percent in the two-injection group). Adhering to diet was considered to be as difficult or more difficult than expected in all the treatment groups (data not shown).

There was significantly less improvement in the subjective sense of well-being in the control group (41 percent, P<0.001 vs. other groups) than in the insulin-treatment groups (74 percent in the multiple-injection group, 84 percent in the morning-NPH group, 100 percent in the evening-NPH group, and 86 percent in the two-injection group). The majority of the patients in all groups (89 to 100 percent) wanted to continue insulin therapy after three months.

Discussion

We compared different regimens of insulin treatment in moderately obese patients with NIDDM whose glycemic status was poorly controlled while they received oral hypoglycemic drug therapy. The results indicate that when improvement in glycemic control is used as the criterion for success, several insulin-therapy regimens are equally effective, at least during a three-month period. When other factors such as weight gain and hyperinsulinemia are considered, however, the combination of oral hypoglycemic drug plus therapy with NPH insulin in the evening appears more advantageous than the other treatment regimens.

Glycemic control improved equally well in the two-injection and the multiple-injection groups and in the two combination-therapy groups, indicating that the first two regimens cannot be regarded as superior to combination therapy. The insulin dose in the morning-NPH and evening-NPH groups was approximately 50 to 60 percent lower than those in the two-injection and the multiple-injection groups. These results are consistent with those of previous studies with a parallel or crossover design that have compared combined hypoglycemic drug and insulin therapy with insulin therapy.13 14 15 16 17 18 19 If one compares the two combination-therapy groups, the total insulin doses were similar. Despite this, the evening-NPH group had less hyperinsulinemia than the other insulin-therapy groups when determinations were made by direct measurement of serum free insulin concentrations (Fig. 4). As in the evening-NPH group, the increment in the overnight concentration of free insulin in the multiple-injection group, in which the patients also had NPH insulin injected at 9 p.m., was smaller than expected. The cause of this small overnight increment in serum free insulin remains speculative, since no studies of insulin absorption were performed. In previous studies in patients with IDDM, the total increment in serum free insulin was greater if absorption was accelerated by exercise.20 One possibility, therefore, is that a higher level of physical activity during the day than at night might explain the difference.

Weight gain is generally regarded as a side effect of insulin therapy. In this study, improvement in glycemic control was significantly correlated with weight gain in the insulin-treatment groups, whereas the opposite was true in the control group. Since patients with NIDDM on the average lose weight after diagnosis,21 part of the weight gain could reflect a correction of relative insulin deficiency. Thus, initial weight gain during insulin therapy could be a sign of the effectiveness of insulin rather than a side effect. On the other hand, there were differences between treatment regimens with respect to the magnitude of weight gain as compared with improvement in glycemic control. The patients in the evening-NPH group gained less weight (0.6 kg for each decrease of 1 percent in glycosylated hemoglobin) than those in the multiple-injection group (1.8 kg for each such decrease). Since serum free insulin also increased less in the evening-NPH group than in the multiple-injection group, it seems logical to propose that weight gain is related to the degree of peripheral hyperinsulinemia. The mechanism or mechanisms by which peripheral hyperinsulinemia may promote weight gain could involve the stimulation of lipogenesis in adipose tissue and the stimulation of appetite,22 and perhaps psychological factors, such as overeating because of a fear of hypoglycemia.

The mean blood pressure decreased significantly (by 8 mm Hg) in the control group despite only a slight improvement in glycemic control and weight loss (0.9 kg). In contrast, blood pressure did not change in any insulin-treatment group, which might be regarded as a disadvantage of insulin. These results do not contradict the concept that insulin and hypertension are linked, but they may merely reflect an unchanged net effect of factors possibly influencing blood pressure, such as weight gain, peripheral hyperinsulinemia, improved insulin sensitivity, and glycemic control.5 , 23

In keeping with previous reports,10 , 24 25 26 27 serum VLDL triglyceride concentrations fell by 13 to 28 percent during insulin therapy. In previous studies of patients in whom glycemic control was markedly improved (with a decrease in glycosylated hemoglobin of 3 percent) by multiple insulin injections, serum VLDL triglyceride concentrations decreased by 50 percent, whereas total and HDL cholesterol did not change.10 In another study serum triglycerides did not change after six months of treatment with a two-insulin-injection regimen during which glycosylated hemoglobin decreased by 2 percent and weight increased by 5 kg.28 Thus, although short-term intensive insulin therapy under carefully controlled conditions induces antiatherogenic changes in serum lipoproteins, these effects may be diminished by factors such as weight gain and suboptimal glycemic control.

A hitherto neglected but important factor in advocating a particular treatment regimen for patients with NIDDM is whether the treatment improves subjective well-being. With respect to insulin therapy, we found this to be true. A comparison of the attitudes toward particular insulin-treatment regimens revealed no significant differences, although the patients in the multiple-injection group had more negative attitudes about insulin injections, home glucose monitoring, and their willingness to continue insulin therapy. Although this study does not provide evidence that evening treatment with NPH insulin is suitable long-term therapy for patients with NIDDM, when the subjective data are combined with objective measures of glycemic control, weight gain, and the degree of hyperinsulinemia, multiple-insulin injection therapy is the least attractive, and combination therapy with evening NPH insulin the most feasible, insulin regimen for patients with NIDDM.

Funding and Disclosures

Supported by grants from the Finnish State Medical Research Council, the Sigrid Juselius Foundation (Helsinki), and Novo Nordisk (Copenhagen, Denmark).

We are indebted to Marjatta Heikkilä, Lisa Flink, Liisa Hyvärinen, Jaana Jäkälä, Margit Kalkkila, Anneli Karjalainen (deceased), Marja-Leena Kekäläinen, Riitta Nopola, and Marja Riihelä for their invaluable help as diabetes nurses; to Ms. Sirkka-Liisa Runeberg, Ms. Elisa Kostamo, Ms. Terttu Lönnblad, and Ms. Terttu Kerman for excellent technical assistance; and to Dr. Ulf-Håkan Stenman and his staff for measuring glycosylated hemoglobin.

Author Affiliations

From the Second (H.Y.-J.) and Third (M.-R.T.) Departments of Medicine, University of Helsinki, Helsinki; the Department of Medicine, University of Turku, Turku (M.K., P.S., J.V.); the Hatanpää Hospital, Tampere (E.K., S.R.); the Kuopio City Hospital, Kuopio (J.L., L.N.); the Äänekoski Health Center, Äänekoski (T.M., S.S.); the Kymenlaakso Central Hospital, Kotka (L.R.); the Lappi Central Hospital, Rovaniemi (T.T.); and the Department of Pediatrics, University of Oulu, Oulu (J.K.) — all in Finland. Address reprint requests to Dr. Yki-Jäarvinen at the Second Department of Medicine, Helsinki University, Haartmaninkatu 4, SF-00290 Helsinki, Finland.

References (28)

  1. 1. Gerich JE. . Oral hypoglycemic agents . N Engl J Med 1989;321:1231–45.

  2. 2. Lebovitz HE, Pasmantier RM. . Combination insulin-sulfonylurea therapy . Diabetes Care 1990;13:667–75.

  3. 3. Firth RG, Bell PM, Marsh HM, Hansen I, Rizza RA. . Postprandial hyperglycemia in patients with noninsulin-dependent diabetes mellitus: role of hepatic and extrahepatic tissues . J Clin Invest 1986;77:1525–32.

  4. 4. Yki-Järvinen H. . Acute and chronic effects of hyperglycaemia on glucose metabolism . Diabetologia 1990;33:579–85.

  5. 5. Andrews WJ, Vasquez B, Nagulesparan M, et al. . Insulin therapy in obese, non-insulin-dependent diabetes induces improvements in insulin action and secretion that are maintained for two weeks after insulin withdrawal . Diabetes 1984;33:634–42.

  6. 6. Riddle MC. . New tactics for type 2 diabetes: regimens based on intermediate-acting insulin taken at bedtime . Lancet 1985;1:192–5.

  7. 7. Taves DR. . Minimization: a new method of assigning patients to treatment and control groups . Clin Pharmacol Ther 1974;15:443–53.

  8. 8. Stenman U-H, Pesonen K, Ylinen K, Huhtala M-L, Teramo K. . Rapid Chromatographic quantitation of glycosylated haemoglobins . J Chromatogr 1984;297:327–32.

  9. 9. Kuzuya H, Blix PM, Horwitz DL, Steiner DF, Rubenstein AH. . Determination of free and total insulin and C-peptide in insulin-treated diabetics . Diabetes 1977;26:22–9.

  10. 10. Taskinen M-R, Kuusi T, Helve E, Nikkilä EA, Yki-Järvinen H. . Insulin therapy induces antiatherogenic changes of serum lipoproteins in noninsulin-dependent diabetes . Arteriosclerosis 1988;8:168–77.

  11. 11. Bonifacio E, Lernmark Å, Dawkins RL. . Serum exchange and use of dilutions have improved precision of measurement of islet cell antibodies . J Immunol Methods 1988;106:83–8.

  12. 12. Dixon WJ, ed. BMDP statistical software. Berkeley: University of California Press, 1983.

  13. 13. Sane T, Helve E, Yki-Järvinen H, Taskinen M-R. . One-year response to evening insulin therapy in non-insulin-dependent diabetes . J Intern Med 1992;231:253–60.

  14. 14. Lotz N, Bachmann W, Ladik T, Mehnert H. . Die Kombinationstherapie Insulin/Sulfonylharnstoff in der Langzeittherapie des Typ-II-Diabetes nach "Sekundärversagen." Klin Wochenschr 1988;66:1079–84.

  15. 15. Gutniak M, Karlander S-G, Efendic S. . Glyburide decreases insulin requirement, increases β-cell response to mixed meal, and does not affect insulin sensitivity: effects of short- and long-term combined treatment in secondary failure to sulfonylurea . Diabetes Care 1987;10:545–54.

  16. 16. Schade DS, Mitchell WJ, Griego G. . Addition of sulfonylurea to insulin treatment in poorly controlled type II diabetes: a double-blind, randomized clinical trial . JAMA 1987;257:2441–5.

  17. 17. Osei K, O'Dorisio TM, Falko JM. . Concomitant insulin and sulfonylurea therapy in patients with type II diabetes: effects of glucoregulation and lipid metabolism . Am J Med 1984;77:1002–9.

  18. 18. Lins P-E, Lundblad S, Persson-Trotzig E, Adamson U. . Glibenclamide improves the response to insulin treatment in non-insulin-dependent diabetics with second failure to sulfonylurea therapy . Acta Med Scand 1988;223:17I–9.

  19. 19. Stenman S, Groop P-H, Saloranta C, Tötterman KJ, Fyhrqvist F, Groop L. . Effects of the combination of insulin and glibenclamide in type 2 (non-insulin-dependent) diabetic patients with secondary failure to oral hypoglycaemic agents . Diabetologia 1988;31:206–13.

  20. 20. Berger M, Cuppers HJ, Hegner H, Jörgens V, Berchtold P. . Absorption kinetics and biologic effects of subcutaneously injected insulin preparations . Diabetes Care 1982;5:77–91.

  21. 21. Knowler WC, Pettitt DJ, Savage PJ, Bennett PH. . Diabetes incidence in Pima Indians: contributions of obesity and parental diabetes . Am J Epidemiol 1981;113:144–56.

  22. 22. Booth DA, Brookover T. . Hunger elicited in the rat by a single injection of bovine crystalline insulin . Physiol Behav 1968;3:439–46.

  23. 23. Hamilton BPM. . Diabetes mellitus and hypertension . Am J Kidney Dis 1990;16:Suppl 1:20–9.

  24. 24. Paisey R, Elkeles RS, Hambley J, Magill P. . The effects of chlorpropamide and insulin on serum lipids, lipoproteins and fractional triglyceride removal . Diabetologia 1978;15:81–5.

  25. 25. Agardh C-D, Nilsson-Ehle P, Schersten B. . Improvement of the plasma lipoprotein pattern after institution of insulin treatment in diabetes mellitus . Diabetes Care 1982;5:322–5.

  26. 26. Pfeifer MA, Brunzell JD, Best JD, Judzewitsch RG, Halter JB, Porte D Jr. . The response of plasma triglyceride, cholesterol, and lipoprotein lipase to treatment in non-insulin-dependent diabetic subjects without familial hypertriglyceridemia . Diabetes 1983;32:525–31.

  27. 27. Lindström T, Arnqvist HJ, Olsson AG. . Effect of different insulin regimens on plasma lipoprotein and apolipoprotein concentrations in patients with non-insulin-dependent diabetes mellitus . Atherosclerosis 1990;81:137–44.

  28. 28. Groop L, Widen E, Franssila-Kallunki A, et al. . Different effects of insulin and oral antidiabetic agents on glucose and energy metabolism in type 2 (non-insulin-dependent) diabetes mellitus . Diabetologia 1989;32:599–605.

Citing Articles (221)

    Figures/Media

    1. Table 1. Base-Line Clinical Characteristics of the Five Groups of Patients with NIDDM.*
      Table 1. Base-Line Clinical Characteristics of the Five Groups of Patients with NIDDM.*
    2. Figure 1. Glycemic Control during the Six-Week Run-in Period and during Treatment in Patients with NIDDM.
      Figure 1. Glycemic Control during the Six-Week Run-in Period and during Treatment in Patients with NIDDM.

      The curves for mean diurnal blood glucose levels represent 70 percent of the requested number of measurements in the five groups: the control group (+), the morning-NPH group (○), the evening-NPH group (●), the two-injection group (□), and the multiple-injection group (■). During the run-in period, the mean value for glycosylated hemoglobin decreased by 0.30±0.09 percent (P<0.001), the mean diurnal blood glucose concentration by 16±4 mg per deciliter (0.89±0.24 mmol per liter) (P<0.001), and the mean fasting blood glucose concentration by 31±4 mg per deciliter (1.7±0.2 mmol per liter) (P<0.001). Each diurnal profile included blood glucose measurements before and 1 1/2 hours after breakfast, lunch, and dinner and at 10 p.m. and 4 a.m. The mean diurnal blood glucose concentration was significantly lower during treatment in each insulin-treatment group than in the control group (P<0.01 to P<0.001). P<0.01 for the difference in values for glycosylated hemoglobin between the insulin-treatment groups and the control group at 12 weeks. To convert values for glucose to millimoles per liter, multiply by 0.05551.

    3. Figure 2. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Determined by Home Glucose Monitoring.
      Figure 2. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Determined by Home Glucose Monitoring.

      The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the blood glucose concentration in the insulin-treatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for glucose to millimoles per liter, multiply by 0.05551.

    4. Figure 3. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.
      Figure 3. Changes in Mean Diurnal Blood Glucose Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.

      The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the blood glucose concentration in the insulin-treatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for glucose to millimoles per liter, multiply by 0.05551.

    5. Table 2. Effect of Insulin Therapy on Body Weight, Glycemic Control, Blood Pressure, and Serum Lipid Concentrations before and after Three Months of Treatment in Patients with NIDDM.
      Table 2. Effect of Insulin Therapy on Body Weight, Glycemic Control, Blood Pressure, and Serum Lipid Concentrations before and after Three Months of Treatment in Patients with NIDDM.
    6. Figure 4. Changes in Mean Diurnal Serum Free Insulin Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.
      Figure 4. Changes in Mean Diurnal Serum Free Insulin Concentrations after Three Months in Each Insulin-Treatment Group as Compared with the Control Group, as Measured in the Hospital.

      The asterisks (P<0.05) and daggers (P<0.01) indicate a significant change in the serum free insulin concentration in the insulintreatment group shown, as compared with the control group. B denotes breakfast, L lunch, and D dinner. To convert values for insulin to picomoles per liter, multiply by 7.2.

      More Research