The University Group Diabetes Program and the United Kingdom Prospective Diabetes Study (UKPDS; see "United Kingdom Prospective Diabetes Study 17: A 9-Year Update of a Randomized, Controlled Trial on the Effect of Improved Metabolic Control on Complications in NIDDM") have both addressed the issue of a comparison of glycemic regulation by diet, oral agents, or insulin in patients with newly diagnosed NIDDM [3-7]. Controversy exists, however, about whether intensive insulin therapy can be safely and effectively used in patients with NIDDM of longer duration who have failed to achieve glycemic control by diet, exercise, or pharmacologic therapy or a combination of these therapies.

Available data suggest that patients with NIDDM in the United States, who are being treated with oral agent or insulin (pharmacologic) therapy, or both, usually have only fair to poor metabolic control. In general, either in cross-sectional studies or in broadly based epidemiologic surveys, such patients are found to have fasting serum glucose levels of 11.1 mmol/L (200 mg/dL) or higher and hemoglobin A_1c (HbA_1c) levels of 9% to 10% [8, 9]. Presumably, physicians do not raise insulin doses to produce normoglycemia but rely more often on diet, exercise, and suboptimal pharmacologic therapy. The rationale for this approach probably includes the insulin resistance of NIDDM (which usually demands large doses of insulin to produce near-normal glycemia); the paucity of information that can clearly identify chronic hyperglycemia as a major causative factor for macrovascular disease; and the fear of hypoglycemia or weight gain, or both, with intensive insulin therapy.

To begin to explore this important clinical issue, the Planning Committee of the Department of Veterans Affairs Cooperative Study on Glycemic Control and Complications in NIDDM (VACSDM) designed a randomized, controlled clinical trial to address the relative benefits and risks of standard as compared with intensive insulin therapy in patients with NIDDM who had continued hyperglycemia despite being on oral agent or insulin therapy. The Planning Committee recognized that a short-term feasibility trial should be done before seriously considering a long-term study. Accordingly, the VACSDM investigators designed such a study for the Veterans Affairs Cooperative Studies program and activated it in January 1990.

Methods

Top Methods Discussion Author & Article Info References

The design of the feasibility trial is reported in detail in another publication [9]. The study was a randomized clinical trial, done with 153 male patients at five Veterans Affairs medical centers. The first randomization occurred in May 1990, and all 153 patients were enrolled by October 1991. Patients were followed for an average of 27 months (range, 18 to 35 months). The last patient visit was on 30 April 1993. The patients' ages ranged from 40 to 69 years (mean, 60 ± 6), and they had elevated HbA_1c values (> 3 standard deviations above the mean normal of 5.05%, or more than 6.55%) while receiving their usual sulfonylurea or insulin therapy. Major exclusion criteria included a duration of known diabetes for more than 15 years, myocardial infarction or cerebrovascular accident in the past 6 months, congestive heart failure, malignancy, autonomic neuropathy, albuminuria greater than 520 mg/d, a serum creatinine level greater than 141 µmol/L (1.6 mg/dL), a history of amputation or gangrene, a low fasting C-peptide level (< 0.21 pmol/mL), and body weight greater than 60% above ideal.

The primary objective of the study was to determine, with at least a 1-year follow-up of each patient, if a statistically and clinically significant difference in the HbA_1c level could safely be achieved between a standard and an intensive insulin therapy group. This difference was defined as a mean fall of the HbA_1c level of 1.5% or more, plus an absolute target of less than 7.5% in the intensively treated group.

We evaluated retinal microvascular disease by yearly seven-field stereo fundus photographs and sent them for masked reading to the Retinopathic Reading Center in Madison, Wisconsin. We evaluated renal function by yearly determinations of creatinine clearance and albumin excretion rates. One study objective was the assessment of specified major cardiovascular events for a long-term trial. To this end, we assessed the occurrence of new myocardial infarction, congestive heart failure, stroke, amputation for ischemic gangrene, and cardiovascular mortality. We also recorded other cardiovascular events, including angina or documented coronary disease, angioplasty or bypass graft, transitory ischemic attacks, claudication, or ischemic ulcers. An End-Points Committee of consultants external to the study and masked to treatment assignment used predetermined criteria to decide whether an event occurred and to categorize it.

Finally, side effects of therapy were to be determined. If these major feasibility goals could be met, it was proposed that a long-term trial could then be designed, with a primary goal of determining if intensive insulin therapy affects the progression rate of macrovascular disease in patients with NIDDM who had chronic hyperglycemia on pharmacologic therapy.

Treatment Strategies

In both the standard and intensive insulin therapy groups, comparable criteria for diet, exercise, smoking cessation, and treatment for hypertension and hyperlipidemia were defined. Patients who were randomly assigned to the standard therapy group were seen quarterly, could take one or two insulin injections daily, and did not adjust insulin doses according to blood glucose level tests. A stepwise approach was used in the intensively treated group as follows: 1) single intermediate or long-acting evening insulin; 2) evening insulin plus daytime glipizide before meals; 3) insulin twice daily with no glipizide; and 4) multiple-dose insulin (three or four times daily) with no glipizide. The goal in each phase was to achieve a normal fasting serum glucose level (4.4 to 6.4 mmol/L, or 80 to 115 mg/dL) and a HbA_1c level between 4% and 6%. The intensively treated group was seen monthly, and if this goal was not achieved by 6 to 8 weeks in one phase, the protocol called for a transfer to the next phase.

Results of the Feasibility Trial

The results are given in detail elsewhere [10] and were presented at the Annual Scientific Meeting of the American Diabetes Association. Major study results are summarized in Table 1, which shows that the major goals of the trial were met. The stepwise approach to insulin therapy produced a significant lowering of HbA_1c and fasting serum glucose levels without undue weight gain or increased hypoglycemia. The difference in the mean HbA_1c level between the two groups was significant (P < 0.001) and averaged 2.07 percentage points from 6 months onward. With intensive therapy, 41.3% of patients reached the normal range of HbA_1c (4% to 6%), but only 9.3% maintained normal levels from the ninth month to the end of the trial. With standard therapy, only 3.9% briefly reached the normal range, but none maintained a normal HbA_1c level.

The mean fasting serum glucose level showed little change with standard therapy, although with intensive therapy, it was reduced to close to the normal range from 3 months onward. The difference in the mean fasting serum glucose level between the two groups after baseline was 5.46 mmol/L (98.3 mg/dL; P < 0.0001).

Seventy-five of the intensively treated patients were treated in phase 1 for 8.7 ± 7 months (mean ±SD), 66 in phase 2 for 10.6 ± 7 months, 48 in phase 3 for 7.4 ± 5 months, and 25 in phase 4 for 5.2 ± 3 months.

At the end of the first year of follow-up for each patient, 27% of the intensive insulin therapy group were still in phase 1, and 60% were in phase 2. At the time of the last visit, 10% were in phase 1, 23% in phase 2, 31% in phase 3, and 36% in phase 4. Final HbA_1c values (mean ±SD) for the patients completing each phase were 7.9% ± 1.5% in phase 1, 7.4% ± 0.8% in phase 2, 7.4% ± 0.9% in phase 3, and 6.9% ± 0.7% in phase 4.

The mean insulin dose during the trial was less than 100 units per day in the intensive group as a whole. A nearly maximal fall in the HbA_1c level occurred after phase 2, with a mean insulin dose (±SD) of 64 ± 42 units per day. For those patients who were in phase 4, mean insulin dose was 133 ± 80 units per day.

The vascular risk factors of hypertension and dyslipidemia were treated equally and effectively in both groups. Adherence was excellent. The two groups were well matched regarding cardiovascular risk factors at baseline. Most patients in both groups had preexisting cardiovascular disease, as assessed at baseline by history, electrocardiography, 24-hour ambulatory electrocardiography, or radionuclide ventriculography.

Nine of the 78 (11.5%) patients in the standard therapy group had 10 major cardiovascular events, and 16 of the 75 (21.3%) patients in the intensive therapy group had 18 major cardiovascular events. Sixteen patients (20.5%) in the standard therapy group had a total of 26 cardiovascular events, and 24 patients (32.0%) in the intensive therapy group had a total of 35 cardiovascular events. There was no statistical difference in event rates for major or for total cardiovascular events between the two groups. Overall, 40 out of 153 (26.1%) patients had cardiovascular events in this 27-month study. The subject of cardiovascular events in this study will be covered in detail in a separate manuscript.

Discussion

Top Methods Discussion Author & Article Info References

A characteristic of this trial was the use of a stepwise approach to insulin therapy. The study showed that evening insulin administration was a good initial strategy for glycemic regulation. Supplementation by oral agent therapy during the day was effective as a second step. Substantial lowering of the HbA_1c level was accomplished with these first two steps of therapy.

The VACSDM investigators chose evening insulin administration as the first step for various reasons. First, the overnight fasting glucose level is a stable and consistent indicator of glycemic regulation in NIDDM [11]. It depends on an increased glucose output by the liver, and this increased output can be made normal by insulin administration [12, 13]. Although postprandial glucose level increases may then occur, normalization of the blood glucose level can lead to a recovery of ß-cell secretory response and may improve insulin action [12, 14]. The use of intermediate-acting insulin at bedtime usually lessens the occurrence of hypoglycemia in the early morning hours (0300 to 0400 h) and should theoretically result in peak insulin action near the time of the dawn phenomenon and the breakfast meal. Several studies have shown that giving intermediate insulin at bedtime improves glycemic regulation more than giving it in the morning [15-20]. Evening injections do not increase the risk for nocturnal hypoglycemia nor do they lower glucose levels at 0300 to 0400 h [21].

The investigators chose a combination of evening intermediate insulin and an oral sulfonylurea agent (glipizide) as the second step in therapy. The rationale was to provide an agent to stimulate endogenous insulin release after a period of good overnight glucose level control. The investigators selected glipizide because it is rapidly absorbed, has a relatively short half-life, and appears to exert a postprandial insulinogenic action [22]. Combined insulin and sulfonylurea therapy has been shown to modestly improve glucose level regulation and to limit use of the very high doses of insulin often needed for optimal glycemic control in NIDDM [23-26]. It was reasoned that this stepped approach could help to avoid the hyperinsulinemia that has been suspected of accelerating atherosclerosis [27, 28].

The pooled total cardiovascular event rate in the VACSDM study was 26.1%, with an average follow-up of 27 months. The major cardiovascular event rate (cardiovascular death, nonfatal myocardial infarction, congestive heart failure, stroke, and amputation) was 16.3%, or 7.3% per year. These are high event rates when compared with recently diagnosed patients with NIDDM and show the poor prognosis at this stage of the disease. In contrast, in recently diagnosed patients with NIDDM, event rates of about 2% to 4% per year have been reported [3-728, 29]. For physicians, it is critical to determine definitively if intensive insulin therapy modifies the high cardiovascular event rate seen in patients with NIDDM in whom pharmacologic therapy has failed to achieve glycemic regulation.

A review of approximate vascular event rates in persons with NIDDM of over 5 years' duration is given in Table 2. The rates of major cardiovascular events, defined here as cardiovascular death, nonfatal myocardial infarction, stroke, or amputation, range from 5.0% to 5.8% per year [30-33]. In the VACSDM study, if all cardiovascular events of clinical significance are pooled, a very high cardiovascular event rate of approximately 26% in 2.25 years (11.6% per year) is seen. The significance of these findings is supported by studies that clearly show that cardiovascular mortality rates in patients with NIDDM increase linearly with duration of diabetes in men and women [29].

Thus, risk for cardiovascular events increases with increasing duration of NIDDM. Because current practice is to try diet, exercise, and oral agent therapy before considering insulin in most patients with newly diagnosed NIDDM, those survivors in whom glycemic regulation fails will clearly have a long duration of unregulated diabetes and will therefore carry a high risk for subsequent cardiovascular events. A recent study [30] shows that metabolic control and duration of diabetes are important predictors of coronary heart disease in elderly patients with diabetes (see "Glycemic Control and the Risk for Coronary Heart Disease in Patients with NIDDM: The Finnish Studies"). These findings lend support to the view that intensive glycemic control could reduce the risk for coronary heart disease in diabetes. However, whether intensive insulin therapy will change this risk, and in which direction, is simply not known. Further, and perhaps more significant, we do not know when to start intensive insulin therapy during the natural progression of NIDDM. It is anticipated that the UKPDS will provide important information on the benefits and risks of starting insulin therapy at the time of first diagnosis of NIDDM (see "United Kingdom Prospective Diabetes Study 17: A 9-Year Update of a Randomized, Controlled Trial on the Effect of Improved Metabolic Control on Complications in NIDDM").

As a corollary, it is important to determine whether an adult patient has NIDDM or IDDM. Presumably, the results from the Diabetes Control and Complications Trial may be extrapolated to adults who require insulin for control of IDDM. Although this differentiation may be difficult, IDDM may be suspected in nonobese adults who have required insulin since the diagnosis of diabetes and who have low or absent plasma C-peptide levels. Because the natural history of retinopathy, nephropathy, neuropathy, and macrovascular disease differs in patients with IDDM and NIDDM, efforts to define which type of diabetes is present are indicated to guide prognosis and therapy [34].

In view of the results from this feasibility trial, a planning committee developed a protocol for a long-term trial. This protocol was reviewed by the Veterans Affairs Cooperative Studies Evaluation Committee and was revised according to their recommendations. The final protocol was reviewed and approved for funding by that committee at its October 1992 meeting. The primary objective of this long-term trial is to determine whether a clinically and statistically significant reduction in major cardiovascular events [myocardial infarction, cardiovascular death, congestive heart failure, stroke, and amputation] can be observed in a group of persons with NIDDM in whom glycemic regulation on pharmacologic therapy has failed and who are randomly assigned to an intensive therapy glycemic control group as compared with a standard therapy group. Eligibility for the trial will be the same as for the feasibility trial, except that women will also be included. Therapeutic goals are to maintain a HbA_1c separation of 1.5 or more percentage points between the two treatment arms and a mean HbA_1c value of less than 7.5% for the intensive arm. A stepwise approach to therapy will be used. On the recommendation of the Cooperative Studies Evaluation Committee, the combined insulin-sulfonylurea step will be dropped, resulting in three phases: 1) evening insulin, 2) two injections a day, and 3) multiple-dose insulin. The sulfonylurea step was dropped after long discussion; the major rationale was to maintain comparable pharmacologic treatment strategies for diabetes between the two study arms. Secondary end points will be monitored and will include total cardiovascular events, retinopathy, albuminuria, and neuropathy. Protocols to ensure equal treatment for vascular risk factors (dyslipidemia, hypertension, weight, and smoking) in the two groups have been developed, following the feasibility trial experiences.

A sample size of 1463 patients among 20 participating Department of Veterans Affairs hospitals is projected. Patient accrual will be for 2 years, and follow-up will be for an additional 5 years. Sample size is based on a reduction in major cardiovascular events from 35% after 6 years of follow-up to 26% in the intensive therapy group, or a reduction of 25%. This yields a power of 0.90 at a two-sided significance level of 0.05. A reduction from 35% to 23% (35% reduction) has a power of 0.99, equals 0.05.

In conclusion, the VACSDM group has successfully completed a feasibility trial of intensive as compared with standard insulin therapy in men with NIDDM, in whom glycemic control on pharmacologic therapy has failed. The experiences and results from the feasibility trial have led to a protocol for a 7-year, long-term trial with 1460 men and women to explore the critical issue of the risks and benefits of intensive insulin therapy in NIDDM.