Effect of Intensive Therapy on Residual β-Cell Function in Patients with Type 1 Diabetes in the Diabetes Control and Complications Trial

Methods

The design, selection criteria, and methods used in the DCCT have been completely described elsewhere [8]. The C-peptide testing protocol and preliminary results from the feasibility phase of the DCCT have been presented in an earlier report [9]. The following eligibility and methodologic details summarize our approach to evaluating C-peptide secretion.

The eligibility criteria were created to enroll a group of patients between 13 and 39 years of age who had type 1 diabetes. During the screening process (1983 to 1989), more than 3000 patients were tested with the DCCT C-peptide protocol. Of the prospective DCCT participants who had had type 1 diabetes for 1 to 5 years, the upper limit of stimulated C-peptide for eligibility in the DCCT was designated as 0.50 pmol/mL. Of the 1441 patients who were ultimately enrolled in the DCCT, 303 had had type 1 diabetes for 1 to 5 years and had stimulated C-peptide values between 0.20 and 0.50 pmol/L. In postrandomization analysis, these patients were designated as C-peptide responders; 138 had been randomly assigned to intensive therapy, and 165 had been assigned to conventional therapy (Table 1). In contrast, 552 of the DCCT participants who had had type 1 diabetes for 1 to 5 years had stimulated C-peptide levels less than 0.20 pmol/mL and thus were designated as C-peptide nonresponders; 274 of these had been assigned to intensive therapy, and 278 had been assigned to conventional therapy. Of the 303 patients who responded at baseline, 7 (5%) in the intensive therapy group and 15 (9%) in the conventional therapy group did not have repeated C-peptide testing at 1 year or thereafter.

To assess the persistence of C-peptide response and to examine the effect of intensive therapy on residual insulin secretory capacity, repeated C-peptide stimulation tests were done annually for up to 6 years in patients who maintained stimulated C-peptide levels of 0.20 pmol/mL or greater. A patient was no longer tested when the stimulated C-peptide level was less than 0.20 pmol/mL. The choice of 0.20 pmol/mL as the cutoff for distinguishing responders from nonresponders was based on our previous study [9], which showed a distinct difference in the pattern of residual β-cell secretion in adolescents and adults and clearly defined the degree of endogenous secretion (≥ 20 pmol/mL) that may help control metabolism (hemoglobin A_1c) in type 1 diabetes.

All eligible volunteers were randomly assigned to intensive or conventional therapy. Randomization sequences were stratified by clinic. In the intensive therapy group, the goal was to achieve near-normal glycemic levels. Intensive therapy consisted of insulin injections given three to four times per day or continuous subcutaneous infusion of insulin; blood glucose monitoring at least four times per day with algorithms for dosage adjustment; blood glucose testing done at 3:00 a.m. once a week; close attention to the effects of diet and exercise on blood glucose; clinic visits for medical supervision and support held at least once a month; and frequent telephone contact with the patient to review and adjust treatment [8, 11]. Patients in the conventional therapy group were managed with one or two insulin injections per day, daily monitoring of blood or urine glucose levels, dietary instruction, and follow-up visits at 3-month intervals [8, 11].

C-Peptide Testing

All C-peptide stimulation tests were done after an 8- to 12-hour overnight fast. The usual morning insulin injection was delayed until test completion. When patients arrived at the center, a baseline blood sample was obtained to determine serum glucose and C-peptide levels. The patient then drank 6 mL of Sustacal per kg of body weight to a maximum of 360 mL (Mead-Johnson, Evansville, Indiana; 1 calorie/mL; 55% carbohydrate, 24% protein, and 21% fat) in a period not exceeding 10 minutes. Ninety minutes after patients completely ingested Sustacal, another blood sample was obtained to determine glucose and C-peptide levels.

Laboratory Analyses

Blood samples were kept at 4 °C; serum was separated and frozen at −20°C within 30 minutes of collection and shipped on dry ice within 7 days of collection to the Central Biochemistry Laboratory at the University of Minnesota. Serum glucose was measured within 1 week of sample collection by a glucose oxidase method using a glucose analyzer (Beckman Instruments, Inc., Fullerton, California). C-peptide was measured within 4 weeks of sample collection by immunoassay using the M-1230 antiserum and other reagents obtained from Novo Industries (Bagsvaerd, Denmark). The procedure followed that recommended by Novo, with the additional step of first precipitating serum samples, blanks, and standards with polyethylene glycol (25%, weight/volume) in distilled water. The interassay coefficient of variation was 0.3% at 0.28 pmol/mL, 3.9% at 0.15 pmol/mL, and 12% at 0.07 pmol/mL. The assay's lower limit of detection was 0.03 pmol/mL. Undetectable values were reported as 0.03 pmol/mL. From 1983 to 1986, hemoglobin A_1c levels were measured by the Central Hemoglobin Laboratory at the Joslin Diabetes Center (Boston, Massachusetts) by a high-performance liquid chromatographic method [12]; after that 3-year period, measurements were done at the Central Biochemistry Laboratory by using the BioRad Diamat high-performance liquid chromatography method (BioRad, Inc., Richmond, California). Duplicate samples gave identical results in both laboratories. All laboratory analyses were done by personnel who were blinded to treatment assignment.

Statistical Analysis

To compare the baseline characteristics of the C-peptide responders between treatment groups, we used a chi-square test for categorical variables and a Wilcoxon rank-sum test for continuous and ordinal variables. To test the distribution of C-peptide and hemoglobin A (_1c) values between treatment groups, we used a Wilcoxon rank-sum test for each time point and an N-weighted test of stochastic ordering adjusted for baseline variables for the entire distribution over time [13]. We defined C-peptide levels of 0.20 pmol/mL or greater as “survival” to evaluate the probability of maintaining C-peptide responses by using life-table analysis. A test of differences of cumulative survival was obtained from the Mantel-Cox log-rank test [14]. The average relative risk comparing the responders and the nonresponders within each treatment group during the complete observation period was estimated by using a proportional hazards regression model [14]. The adjusted relative risk was stratified by baseline retinopathy and adjusted for the stimulated C-peptide level at study entry. The percentage risk reduction from the nonresponders was calculated from the relative risk of intensive compared with conventional therapy as (1 −relative risk) × 100%. The rate of severe hypoglycemia was determined as the number of events per 100 patient-years. The differences in these rates were tested by analysis of overdispersed Poisson count data [15]. The Spearman rank correlation coefficient was used to evaluate the relation between stimulated C-peptide and hemoglobin A (_1c) values at each year.

Role of Study Sponsor

An independent, external data monitoring committee regularly reviewed the outcome data and advised the sponsoring institute (National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health) about when to terminate the DCCT. The DCCT Research Group analyzed and interpreted the data and prepared the manuscripts.

Results

Baseline Characteristics

At entry into the DCCT, responders who were randomly assigned to intensive therapy had clinical characteristics similar to those of responders assigned to conventional therapy. The only exceptions were the greater proportion of male patients and the higher albumin excretion rate in responders assigned to the conventional therapy group (Table 1). Nonresponders assigned to conventional therapy had baseline characteristics similar to those of nonresponders assigned to intensive therapy. On the other hand, differences in baseline clinical characteristics emerged when patients were divided by C-peptide levels at study entry. Responders in both groups tended to be slightly older and to have had a shorter duration of diabetes than the respective nonresponders (Table 1). As in previous studies [6], residual endogenous insulin secretion in responders was associated with mean plasma glucose and hemoglobin A_1c values at baseline that were lower than those in nonresponders, even though responders were receiving substantially less insulin.

Effect of Therapy on β-Cell Function

Despite similar stimulated C-peptide levels at study entry, the responders randomly assigned to intensive therapy had higher stimulated C-peptide levels during the first 5 years of the study than the responders randomly assigned to conventional therapy (P < 0.001) (Figure 1). The life-table-estimated probability of maintaining β-cell function (stimulated C-peptide levels ≥ 0.20 pmol/mL) was significantly higher in the intensive therapy group after 1 year (Figure 2). For the first 2 years of the study, the maintenance of β-cell function in the intensive therapy group was one third greater than that in the conventional therapy group. After 2 years, the prevalence of stimulated C-peptide levels of 0.20 pmol/mL or greater in the intensive therapy group began to decrease toward the prevalence in the conventional therapy group. Nevertheless, intensive therapy reduced the adjusted mean risk for loss of C-peptide response (that is, <0.20 pmol/mL) by 57% (95% CI, 39% to 71%; P < 0.001) over the mean 6.5 years of study.

View larger version:

• In this window
• In a new window

Figure 1. Box plots present the annual distributions for each treatment group. The ends of each box correspond to the lower and upper quartiles. Brackets represent the 5% and 95% points of the distribution. The horizontal line in each plot represents the median, and the asterisk in each plot represents the mean. > 0.2 for eligibility period, > 0.2 at baseline, = 0.003 at year 1, = 0.01 at year 2, = 0.006 at year 3, = 0.001 at year 4, = 0.124 at year 5, and > 0.2 at year 6. Striped bars = intensive therapy group; white bars = conventional therapy group. Distribution of stimulated C-peptide levels by treatment group over 6 years of the study among patients who were C-peptide responders at baseline.PPPPPPPP

View larger version:

• In this window
• In a new window

Figure 2. 20 pmol/mL) with intensive therapy (solid line) compared with conventional therapy ( ) ( < 0.001). Probability of maintaining C-peptide secretion (stimulated C-peptide level ≥ 0.dotted lineP

Clinical Outcomes

As was seen for the DCCT study as a whole, intensively treated responders and nonresponders had lower hemoglobin A_1c values than their conventionally treated counterparts (Table 2). Intensively treated responders also had reduced risks for development and progression of retinopathy and microalbuminuria and increased risks for severe hypoglycemia compared with conventionally treated responders (Table 3). Nevertheless, differences were seen within the intensive therapy group that were based on baseline C-peptide status (Table 2 and Table 3). For the first 4 years of the study, hemoglobin A_1c levels remained lower (P < 0.01) in intensively treated responders than in intensively treated non-responders; this difference was associated with a 50% reduced risk for retinopathy progression (CI, 12% to 72%). After adjustment for the current mean hemoglobin A_1c value, the risk reduction was not significant (Table 3). A similar trend was seen with a lower risk for the development of microalbuminuria (relative risk, 0.77 [CI, 0.37 to 1.62]). Despite lower hemoglobin A_1c values in intensively treated responders, the risk for severe hypoglycemia with seizure or coma was 65% (CI, 53% to 74%) less in this group than in intensively treated nonresponders (Table 3). In contrast, conventionally treated responders and nonresponders had similar hemoglobin A_1c levels after the first 2 years (by which time C-peptide secretion had largely waned) and until year 8; the small number of patients with residual C-peptide secretion at that point (n = 17) makes interpretation of the difference in HbA_1c values difficult (Table 2). Responders and nonresponders in the conventional therapy group had similar risks for progression of retinopathy and for development of microalbuminuria.

Discussion

The large number of patients with type 1 diabetes who entered the DCCT with a modest amount of residual endogenous insulin secretion provided a unique opportunity to examine the influence of treatment on residual β-cell function in type 1 diabetes. Intensive insulin therapy in responders did not prevent the inexorable decline in insulin secretion that occurs with type 1 diabetes, but it markedly slowed the loss of β-cell function compared with that in responders who received conventional therapy. Although it is difficult to estimate the amount and duration of residual β-cell function sustained by responders who received intensive therapy, islet function was probably extended for at least 2 years. The residual insulin secretory capacity allowed patients to manage themselves intensively in a more effective manner than could their C-peptide-deficient counterparts. Intensively treated responders maintained substantially lower mean hemoglobin A_1c levels than did similarly treated nonresponders until the fifth and sixth years of the study, when most previous responders had also lost residual β-cell function.

The lower hemoglobin A_1c values in intensively treated responders presumably explains their lower risks for progression of retinopathy and development of microalbuminuria [16-18]. In the conventional therapy group, on the other hand (in which early loss of C-peptide secretion led to increased hemoglobin A_1c values in patients who were previously responsive to C-peptide), no differences in the development of complications were seen between the previous responders and nonresponders.

Prolongation of β-cell function also affected the rate of hypoglycemia. Among responders, the rate of seizure or coma (6.6 episodes per 100 patient-years of follow-up) in the intensive therapy group was higher than the rate (3.0 episodes/100 patient-years) in the conventional therapy group but was much lower than that seen in the intensively treated nonresponders (17.3 episodes/100 patient-years) (Table 3).

The mechanisms underlying the prolongation of islet function with intensive therapy in the DCCT remain unknown. A reduction of the glucotoxic effects of hyperglycemia on islet-cell viability may spare the islets from one putative factor contributing to their destruction [19]. Another possibility is that the autoimmune process that destroys islets may be slowed with intensive therapy. Under more benign conditions of intensive therapy, the islets that still survive the continuing autoimmune destructive process may function more effectively, thereby permitting better glycemic control. In this scenario, intensive control has no effect on the auto-immune or other process destroying the islets [2-4], but it does allow an enhanced insulin secretory capacity of the islets that still maintain function. We could not determine which of these mechanisms, if any, preserved C-peptide secretion.

Another potential benefit of sustaining islet function lies in sustained exposure to C-peptide of the organs at risk for microvascular complications. Thus, C-peptide may also influence retinopathy [20] and nephropathy [21, 22] in addition to the effect of preserved insulin secretion on glycemic control. The reduced effect of residual β-cell function on retinopathy in our study after adjustment for current hemoglobin A_1c values (Table 3) makes this possibility less likely [23]; however, it does not completely rule out innate C-peptide activity as a separate beneficial mechanism. Additional studies are needed to address this issue.

Because the DCCT was not specifically designed to examine the preservation of β-cell function, our observations have several limitations. We noted the beneficial effect of intensive therapy on residual β-cell function even though all patients had had diabetes for at least 1 year (mean duration, 2.7 years) before starting intensive therapy. Moreover, patients with better preserved insulin secretion during eligibility screening (stimulated C-peptide levels > 0.50 pmol/mL) were excluded from the DCCT. We can only speculate about the effects of intensive therapy on β-cell function in patients with shorter duration of type 1 diabetes and greater C-peptide secretion.

The loss of residual C-peptide secretion was described as a permanent loss of function, and the rates were calculated by life-table analysis. More intensively treated responders might have recovered secretion of 0.20 pmol/mL or more after losing it compared with conventionally treated responders. Because we did not continue to test C-peptide secretion after levels less than 0.20 pmol/mL occurred, this is speculative.

Our observations further support the implementation of intensive therapy as early in the course of type 1 diabetes as is practical and safe. Such treatment helps sustain residual endogenous insulin secretion, which, in turn, allows better metabolic control for longer periods with fewer acute and chronic complications. The ability to maintain lower hemoglobin A (_1c) values with relatively fewer episodes of severe hypoglycemia (the major adverse event associated with intensive therapy) makes earlier intervention especially appealing.