A Kinetic Model for Glucose Levels and Hemoglobin A1c Provides a Novel Tool for Individualized Diabetes Management

Abstract

Background: Regular assessment of glycated hemoglobin (HbA1c) is central to the management of patients with diabetes. Estimated HbA1c (eHbA1c) from continuous glucose monitoring (CGM) has been proposed as a measure that reflects laboratory HbA1c. However, discrepancies between the two markers are common, limiting the clinical use of eHbA1c. Therefore, developing a glycemic maker that better reflects laboratory HbA1c will be highly relevant in diabetes management.

Methods: Using CGM data from two previous clinical studies in 120 individuals with diabetes, we derived a novel kinetic model that takes into account red blood cell (RBC) turnover, cross-membrane glucose transport, and hemoglobin glycation processes to individualize the relationship between glucose levels and HbA1c.

Results: Using CGM data and two laboratory HbA1c measurements, kinetic rate constants for RBC glycation and turnover were calculated. These rate constants were used to project future HbA1c, creating a new individualized glycemic marker, termed calculated HbA1c (cHbA1c). In contrast to eHbA1c, the new glycemic marker cHbA1c gave an accurate estimation of laboratory HbA1c across individuals. The model and data demonstrated a non-linear relationship between laboratory HbA1c and steady-state glucose and also showed that glycation status is modulated by age.

Conclusion: Our kinetic model offers mechanistic insights into the relationship between glucose levels and glycated hemoglobin. Therefore, the new glycemic marker does not only accurately reflect laboratory HbA1c but also provides novel concepts to explain the mechanisms for the mismatch between HbA1c and average glucose in some individuals, which has implications for future clinical management.

Keywords: HbA1c; glycemia; kinetic model; red blood cell glycation; red blood cell turnover.

Figure 1.

Kinetic model with multiple kinetic processes. G and GI are blood glucose and intracellular glucose in RBCs, respectively. Hb and HbG are non-glycated and glycated hemoglobin in RBCs. The rate constants k_g, k_age, and k_gen are assigned to hemoglobin glycation, elimination, and generation processes, respectively.

Figure 2.

Example of data used in the method. Laboratory HbA1c values at the beginning and end of the first data section were evaluated together with daily average glucose values throughout the first data setion to determine individual kinetic constants. The second data section daily average glucose values were used to prospectively calculate a value to compare to the laboratory HbA1c at the end of the second data section.

Figure 3.

Evaluation of three methods for agreement between estimated HbA1c and laboratory HbA1c. Data are shown for study participants with both type 1 (black) and type 2 (red) diabetes. For each method, the blue lines are linear regression with 95% confidence intervals, and the black line is unity.

Figure 4.

cHbA1c trace example. Glucose trace (thin solid line), laboratory HbA1c values (blue dots), cHbA1c trace (black line), and eHbA1c traces based on 14-day average glucose (redline) from a clinical study subject. The first data section was used to calculate kinetic constants, k_gly = 8.21E−6dL/mg/day and k_age = 0.0165 day⁻¹. The prospective cHbA1c value (8.5%) agrees with the day 193 HbA1c value (8.5%).

Figure 5.

Examples of individual eHbA1c and cHbA1c relationship over time. Panels A, B, and C are type 1 diabetes study participants with absolute prediction errors at 25th, 50th, and 75th percentiles of all type 1 subjects. Similarly, panels D, E, and F are type 2 diabetes study participants with absolute prediction errors at 25th, 50th, and 75th percentiles of all type 2 subjects. Blue dots are laboratory HbA1c measurements. Black lines are cHbA1c traces. Red lines are 14-day average glucose traces.

Figure 6.

Distributions of K, kgly, and kage (A, B, and C, respectively) in 120 clinical study subjects. Subjects with age less than 48 years, 48-60 years, and more than 60 years are shown in blue, yellow, and red, respectively. K increases with longer diabetes duration, which appears to be related to larger reduction in k_age than k_gly.