Timing of Insulin Before Meals Everyday

Data:

The Rationale for Timing of the premeal insulin administration - In type 1 diabetes, very
little, if any, endogenous insulin is available to handle the carbohydrate load which very
rapidly enters the circulation after the start of a meal. In non diabetic individuals, a
complex array of physiological events occurs prior to eating (often called the cephalic
phase of insulin secretion) which prepares the pancreas for immediate release of preformed
insulin when any meal related increase in blood glucose occurs. This rapid insulin
secretion prepares the tissues (primarily muscle and liver) to take up glucose very rapidly
and thereby prevent severe hyperglycemia in the postprandial state. Since significant
absorption of all short acting insulins (e.g. insulin lispro, aspart, or glulisine) from the
subcutaneous tissue does not occur for 15 minutes following injection (26), this delay often
results in postprandial hyperglycemia. If the patient happens to be hypoglycemic prior to
the meal, this delay can be advantageous, but if he/she is hyperglycemic, then severe
postprandial hyperglycemia may result. A rationale solution to this problem would be to
time the short acting insulin injection based on the premeal blood glucose, but aside from
our one pilot study addressing this strategy, no clinical study has been published
supporting this approach. Our grant application is designed to correct this deficiency and
provide the scientific data to support this addition to intensive insulin therapy.

We studied the effect of Timing a fixed dose of rapid acting insulin lispro (Humalogâ„¢) in
twelve type 1 diabetic subjects who were all made hyperglycemic prior to a standardized
diabetic meal (11). In all studies, the dose of insulin of 0.15U/kg (~10 units) remained
constant and represented a typical dose utilized by type 1 diabetic patients prior to an
evening meal. Each of the twelve diabetic volunteers participated in all four study arms
(48 total studies). The insulin lispro was given at the following times: 1) 30 minutes
prior to meal time (-30 minutes), 2) 15 minutes prior to the meal time (-15 minutes), 3) at
the meal (0 minutes), and 4) 15 minutes after the meal was begun (+15 minutes). As is shown
in Figure 1, a normal postprandial glucose excursion was observed when the insulin was taken
15 minutes prior to the meal. No postprandial excursion was observed when the insulin was
taken 30 minutes prior to the meal. This is the desirable outcome in a hyperglycemic
individual. Of particular interest is the fact that when the insulin lispro was taken right
at mealtime (which is the most common pattern in type 1 diabetic patients), significant
postprandial hyperglycemia was observed. Figure 2 provides the integrated postprandial
glucose excursion (above and below baseline) over the entire 5½ hour study and reflects the
data in Figure 1. This study strongly supports the injection of insulin lispro at least 15
minutes prior to the meal when premeal hyperglycemia is present.

To be certain that the absorption of insulin lispro was not different between the four
studies which might account for the difference in postprandial glucose excursion; we
measured free insulin levels during all four research studies. As is shown in Figure 3,
when the insulin excursion concentrations are superimposed on each other for each study, no
difference in insulin absorption occurred between the four insulin administration protocols.
This data strongly supports the concept that timing of the insulin injection prior to the
meal has a major effect on glucose kinetics following a meal. Utilization of Timing in an
Intensive Insulin Therapy regimen should have a major beneficial effect of suppressing
postprandial hyperglycemia and reducing A1C.

Study Design:

The proposed study will be a double-blind, randomized controlled trial of approximately
eight months duration per patient. The volunteers will be screened and then randomized to
either a Timing protocol (continuing with their Intensive Insulin Therapy regimen) or
continuing their usual Intensive Insulin Therapy. All volunteers will utilize Continuous
Glucose Monitoring every other month in order to provide data on the magnitude of
postprandial hyperglycemia and the incidence of hypoglycemia.

Insulin Timing Algorithm:

Based on our preliminary data (24), we will utilize the following algorithm to time the
premeal insulin administration dose: 1) premeal blood glucose >250 mg/dl = take insulin 30
min before the meal, 2) premeal blood glucose between 250 and 150 mg/dl = take insulin 15
minutes before the start of the meal, 3) premeal blood glucose between 150 and 75 mg/dl =
take insulin at the start of the meal, and 4) premeal blood glucose < 75 = take insulin 15
min after the start of the meal. Participants will remain on their current insulin to
determine if the the timing algorithm is indeed effective. The advantages of this algorithm
are that it is very easy to remember and practical for the typical patient with type 1
diabetes. It is designed to keep the glucose between 100 mg/dl and 150 mg/dl. It will
reduce the postprandial glucose excursion when most needed, i.e. when the premeal glucose is
abnormally high and prevent hypoglycemia when the premeal glucose is low.

Statistical Assessment

Power Analysis:

Our pilot data are the only data available in the literature that assesses the effect of
timing of the pre-prandial short acting insulin dose. As detailed above, this pilot study
was short term so did not include any assessment of changes in A1C. Since A1C is the
primary endpoint in the proposed study, we had to estimate the change in A1C that would have
occurred with the changes we observed in the integrated postprandial excursion (figure 1
above). Assuming that the postprandial glucose excursion is responsible for 50% of the A1C
value (above the normal concentration of 5.5%) as has been estimated in type 1 diabetic
patients (16,17), then the variance and mean change in integrated glucose excursion was used
to estimate the change in A1C that would be expected in the proposed study. From this data,
we calculated that an improvement of A1C of 1% (the difference in A1C between the control
Arm A and the intervention Arm B would require 16 subjects per group to provide a power of
90% to detect this difference at a p<0.01. In order to control for a maximum drop out rate
of 20%, we will recruit and randomize a total of 40 volunteers for each group (a total of 80
subjects).

Data Analysis:

The primary endpoint for this study will be the difference in A1C levels between
experimental Arm A (volunteers not utilizing Timing ) and Group B (volunteers utilizing
Timing). This comparison will provide the net effect of Timing and the direct measure of
the change in postprandial hyperglycemia. In addition, several sub analyses will be done.
This comparison will control for any positive effects that Continuous Glucose Monitoring may
have on A1C. In addition, we will integrate the area under the five hour postprandial
glucose curve as provided each day by the Continuous Glucose Monitoring Data for the primary
meal of the day (as stated by the patient - usually supper). This data will provide a
mechanism of any improvement in A1C observed. Finally, as an indirect assessment of changes
in postprandial hyperglycemia, changes in 1, 5-anhydroglucitol (Glycomarkâ„¢) will be measured
every month. We will utilize an "intention to treat analysis" to analyze the data
obtained. Analysis of data will be done using NCSS computer software (42).

Clinical Importance:

If our pilot study is validated by the proposed randomized clinical trial in this grant
application, a new parameter for control of postprandial glucose will become available for
the patient with type 1 diabetes. To date, this concept has not been utilized by healthcare
providers because no one has demonstrated that Timing will result in a reduction of A1C.
Our proposed grant application is designed to obtain the data that will make the use of
Timing an integral part of a glucose control algorithm. Our results are not restricted to
type 1 diabetic patients who utilize insulin lispro since all of the short acting insulin
analogs exhibit approximately the same absorption insulin profile (26), Participants will
remain on their current insulin regimen. If our hypothesis is proved correct, i.e., that the
correct Timing of the premeal insulin bolus will reduce the A1C by an average of 1% in type
1 diabetic individuals with A1Cs between 7.0% and 9.0%, then according to data generated by
the Diabetes Control and Complications Trial, a significant reduction in the microvascular
and macrovascular complications of diabetes will occur. In addition, the successful
completion of this study will provide a stimulus to examine the Timing of premeal insulin in
other groups of diabetic patients, including type 2 diabetic patients on insulin therapy,
pregnant women with gestational diabetes, and children with type 1 diabetes. Finally, in
this day and age of continuing increases in health care costs, the inclusion of Timing will
not add any additional financial cost to the diabetic patient's intensive insulin therapy
regimen.