The Continuous Glucose Monitoring System (CGMS, MiniMed Inc., Sylmar, Calif.) is a first-generation system and functions similarly to a Holter-style cardiac monitor. The patient wears the small device for one or more days, and then the stored data is downloaded at the clinician's office. In its current form, the CGMS provides 288 blood glucose readings per day, as compared to the typical 4 readings per day provided by standard fingerstick monitoring methods. Such an enormous increase in information offers health care professionals an incredibly detailed view of patients' individual glycemic variability, including patterns and underlying causes to that variability.²

Study results have established the accuracy and clinical value of this increased information across a broad range of demographic groups, including pediatric and pregnant patients,^3-5 and several late-breaking reports of clinical utility have just emerged.^6,7 In one study, therapy changes based on the CGMS reduced HbA_1c levels from 9.9 to 8.8% in 5 weeks.⁸ The CGMS is also expected to provide a far more accurate assessment of the effects of changes in therapy than has ever before been possible.

The CGMS consists of a miniature sensor placed just under the skin, typically in the abdominal area, with an automated insertion device. The procedure is generally painless and can be accomplished either at the clinician's office or by the patients themselves. The sensor is connected by a thin wire to a pager-sized monitoring device. The device acquires a sensor signal every 10 seconds and calculates, then stores, an average glucose value every 5 minutes around the clock. Individual sensors can be worn for up to 72 hours, after which they can be replaced for extended monitoring. Patients enter their fingerstick glucose measurements into the monitor a minimum of four times per day for calibration.

Since this device is for clinician use, glucose values are not displayed real time, but instead are downloaded to a computer for analysis. The output provides an immediate retrospective look, which includes graphs of glucose levels for each day, a combined modal day, and a summary table of relevant statistics for both the sensor readings and the calibration fingerstick measurements. The use of the CGMS in a private practice is illustrated in the following three cases.

Troglitazone was discontinued, and the problems persisted. While these observations suggested that the patient no longer had significant insulin resistance, they did not suggest how his now "type 1" diabetes would best be managed.

An important new concern was the progression of his diabetic retinopathy. The patient had previously undergone laser surgery on his left eye and was being monitored for retinopathy in his right eye. His most recent exam showed new proliferative changes.

We attempted to optimize the intensive therapy with 2–4 injections of lente and regular insulin per day (mealtime lispro insulin did not improve late postprandial hypoglycemia or hyperglycemia). The patient continued to self-monitor blood glucose a minimum of 4 times per day. The patient, a technical professional, actively participated in monitoring blood glucose levels and calculating average blood glucose levels.

Despite some improvement in H.B.'s blood glucose control, his right eye deteriorated further, and he underwent extensive laser surgery in March 1999. Knowing that H.B. was highly motivated to improve his blood glucose control to prevent additional complications, we invited him to participate in the clinical trial of the CGMS, which had become available in September 1999.

Figure 1. Case #1: Patient H.B.–Pre-CSII Individual Day CGMS Download.

On reviewing the blood glucose data, we discovered that his blood glucose levels often rose 100–200 mg/dl after meals and periodically during the day (Figure 1). However, his average blood glucose level was 128 mg/dl. There were significant portions of the day when lente insulin provided either too much or too little background basal insulin.

H.B. elected to begin continuing subcutaneous insulin infusion (CSII) to improve his blood glucose control by reducing the fluctuations in blood glucose and to decrease the risk of hypoglycemia during the night or during periods of significant exertion. The CGMS data was instrumental in persuading his insurance company to approve reimbursement for the cost of his insulin pump.

Figure 2. Case #1: Patient H.B.–Post-CSII Individual Day CGMS Download.

Pump therapy was initiated in February 2000. Since then, H.B.'s average blood glucose level has decreased to 111 mg/dl, and his HbA_1c has decreased to 6.0%. Postprandial hyperglycemia, as well as early morning hypoglycemia, was markedly reduced in the follow-up CGMS study. Importantly, based on the follow-up CGMS study, the patient is now meeting his target blood glucose range (70–150 mg/dl) 81% of the time (Figure 2), as compared with only 63% of the time before insulin pump therapy (based on the initial CGMS study). In addition, H.B.'s insulin usage has decreased from 45–50 U/day to 25–30 U/day with the pump, again confirming the impression from the initial CGMS study that the patient should now be considered as a patient with type 1 diabetes mellitus.

The patient's HbA_1c had increased over a 6-month period from 7.0 to 8.0%, and he complained of increased headaches in the early morning hours. His self-monitoring records showed an average blood glucose of 162 mg/dl (breakfast, 137 mg/dl; lunch, 199 mg/dl; dinner, 151 mg/dl; and bedtime, 188 mg/dl). J.T. wore the CGMS for a 48-hour period in January 1999 in an attempt to reveal the reason for the early morning headaches and the gradual increase in HbA_1c levels.

Figure 3. Case #2: Patient J.T.–Modal Day CGMS Download.

The CGMS showed a consistent pattern of normal to low blood glucose levels from 1:00–6:00 a.m., followed by a markedly high level from breakfast to lunch (>300 mg/dl after breakfast), trending back to normal in the afternoon and becoming elevated again after dinner (Figure 3). With these data, we were able to adjust J.T.'s oral hypoglycemic regimen by substituting a shorter-acting, faster-peaking oral agent (repaglinide, 2 mg with each meal) for the glyburide.

This change helped to improve the patient's post-meal blood glucose control and prevent the nocturnal hypoglycemia, which had been the primary cause of the increased headaches. CGMS monitoring, repeated for a 48-hour period in March 2000, revealed more stable blood glucose levels, with higher nocturnal and lower postprandial readings. At a follow-up visit in April 2000, HbA_1c improved to 7.2%.

Her insulin dosage schedule was as follows: 6 U regular insulin before breakfast; 3 U regular and 1 U insulin lispro before lunch, 7 U regular insulin before dinner, and 8 U NPH at bedtime. She had previously tried ultralente and lente insulins at bedtime with suboptimal results (erratic, unexplainable blood glucose readings throughout the day and night with significant hyperglycemia in the morning).

While her pre-dinner blood glucose level averaged approximately 100 mg/dl, her fasting blood glucose was significantly elevated, and her HbA_1c was 8.0%. A.F. had mild retinopathy and neuropathy, and she had a significant fear of hypoglycemia, because it precipitates migraine headaches, which reduce her ability to work efficiently for 2–3 hours.

Figure 4. Case #2: Patient A.F.–Individual Day CGMS Download.

We initiated CGMS to determine why A.F.'s fasting blood glucose and HbA_1c were elevated. The CGMS revealed an average blood glucose of 131 mg/dl. By examining the modal day calculation, we could see a pattern of normal to low blood glucose levels until 3:00 a.m., followed by a significant 8:00 a.m. rise of up to 250 mg/dl, followed by a progressive decline in blood glucose until dinner time (Figure 4). Given this profile, her concern over hypoglycemia, and her overall elevated HbA_1c, we recommended that A.F. switch to an insulin pump. She is currently considering this option and meanwhile has moved her NPH dose to later at night. A repeat CGMS analysis will be performed before her next office visit.

Case 3 presented a highly motivated professional woman whose desire for improved control was hampered by her fear of hypoglycemia and her significant dawn phenomenon. Use of an insulin pump was recommended in order to better match insulin delivery with insulin need and to avoid nocturnal hypoglycemia. Preliminary studies with the CGMS have detected a much higher incidence of hypoglycemia than previously suspected, particularly at night.⁵

As these three cases demonstrate, use of the CGMS enables clinicians to pinpoint specific causes of suboptimal glycemic control. The result is an opportunity for individualized changes in therapeutic regimen not possible with conventional blood glucose monitoring.⁶ Many more studies of the CGMS are underway and will likely demonstrate an increasing number of applications for continuous glucose sensing.