Constructing Diurnal Profiles
To simulate 24-hour profiles, the diurnal study data were assembled end-to-end. At the connections between meal cycles, time 0 minutes for the subsequent cycle was used in place of time 360 minutes for the previous cycle. Concentrations between hours 18 and 24 were estimated based on a straight-line extrapolation and shown as dotted lines.
Correlation analyses are based on 38 diurnal data points from -0.5 to 18 hours.
Meal profile misalignments
caused by experimental factors
For the diurnal study, the profile alignments between breakfast/lunch and lunch/dinner were evaluated by comparing the analyte levels at times 360 and 0 minutes to the peak analyte levels measured during each meal cycle. Alignment for nondiabetics was quite good: the starting and ending plasma concentrations of all three analytes were close together, with misalignments of less than 4% of peak concentrations.
Alignment for T1D subjects was not as close as for nondiabetics:
Time 0 average insulin concentrations were elevated 10-27% of peak concentrations above time 360 concentrations, presumably in part because of the pre-meal adjustment boluses by some subjects.
Time 0 average glucose concentrations were depressed 22-25% of peak concentrations below time 360 concentrations, perhaps because of the pre-meal insulin adjustments.
Time 0 average glucagon concentrations were elevated about 11% concentrations over time 360 concentrations; we can offer no mechanistic hypothesis to explain this.
These alignment differences should be considered when evaluating the composite 24-hour diurnal profiles. For the T1D subjects, they distort somewhat the linkages between breakfast/lunch and lunch/dinner, so chart connections between meals for T1D subjects are shown as dotted lines. However, the alignment differences should be irrelevant for the correlation analyses.
Perspective on healthy,
nondiabetic diurnal patterns
Exhibit 5 combines all three analytes for healthy nondiabetics on a scale indexed to levels at thirty minutes before breakfast (the basal level). The delta in insulin from basal has been divided by seven to render visually comparable spikes to glucose and insulin.
As expected, there is a direct relationship between glucose and insulin concentrations. Also as expected, the glucagon profile is largely contrary to the glucose (and insulin) profile: rising glucose causes a postprandial depression in glucagon, then falling glucose causes a rise in post absorption glucagon, followed by a return of glucagon toward basal levels before the next meal.
Perspective on
T1D diurnal patterns
Exhibit 6 combines the three analytes for T1D subjects on the same scale as Exhibit 5.
Mealtime glucose spikes are proportionally similar to those in nondiabetics, albeit from a higher starting concentration (Exhibit 7). Mealtime insulin spikes are relatively muted because of constraints associated with subcutaneous delivery. Also, insulin returns toward basal more slowly than in nondiabetics, probably because of continuing diffusion from the subcutaneous infusion site.
