Consistency of the personalized glycated haemoglobin (pHbA1c) methodology over time in people with type 1 diabetes (T1D) using continuous glucose monitoring

Since the discovery of the association between glycated haemoglobin (HbA1c) and glucose control in 1968, ¹ HbA1c has been adopted globally and continues to be the primary marker for overall glycaemic control due to its convenience, wide availability and evidence base for association with diabetes complications. ² , ³ , ⁴ This measure is dependent on assay performance, blood glucose levels, the glycation process itself and the lifespan of red blood cells (RBCs). ⁵ With the widespread adoption of continuous glucose monitoring (CGM), it is becoming clear that glucose metrics and HbA1c may be discordant.

Recent studies have looked at the agreement between CGM‐derived glucose management index (GMI) and HbA1c in diabetes and non‐diabetes populations and have provided insights regarding appropriate clinical interpretations, highlighting where more data are needed. ⁶ , ⁷ , ⁸ The difference between GMI and laboratory HbA1c (labHbA1c) can be clinically significant and may have implications for clinical risk. ⁹ , ¹⁰

In a group of individuals with type 1 diabetes (T1D), we recently reported that individuals in the highest tertile of reading‐to‐reading glucose change showed the greatest change in estimated glomerular filtration rate (eGFR) and that those with a higher proportion of glucose readings >18 mmol/L also showed a fall in eGFR while experiencing higher rates of sight‐threatening retinopathy, as did people with higher mean glucose.

The size of the mismatch can be significant between GMI and labHBA1c for any individual, perhaps more than has been appreciated when the association between average glucose and HbA1c was first established by the A1C‐Derived Average Glucose (ADAG) study and adopted in the American Diabetes Association Standards of Care, as well as other international guidelines. ¹¹ , ¹²

A new glycaemic measure, personalised HbA1c (pHbA1c), ¹³ was developed from first principles to calculate a value equivalent to the laboratory‐measured HbA1c from the average measured by CGM that may address the inaccuracies of HbA1c as a measure of glycaemia by accounting for interindividual variability in RBC glycation and lifespan. This is essential not only for monitoring in diabetes but also for decisions about potential treatment changes. There are also implications for the way that HbA1c is applied in screening for type 2 diabetes (T2D). ¹⁴ , ¹⁵

We here report findings from a study in individuals with T1D in which we analysed the relation between pHbA1c and labHbA1c at different time points, using FreeStyle Libre© data collected up to 18 months, and how this relationship may be modulated by individual characteristics.

CGM values were downloaded from the LibreView record for people with T1D in order to estimate pHbA1c. Their contemporaneous labHbA1c measurements were also obtained.

For a period of 81–105 days (depending on availability) prior to the labHbA1c result (expressed in DCCT units, %), the average glucose (AG) was calculated.

The formula used to calculate pHbA1c (also expressed as DCCT units, %) based on the AG in the 100‐day period prior to laboratory samples being taken was derived by Dunn et al. ¹³ establishing the apparent glycation ratio (mL/g) (AGR) = (1/AG (mg/dL) + 1/K _M) × 10⁵/(100/labHbA1C−1), where K _M = 472 mg/dL and then pHbA1C = 100/(1 + AGR/65.1) × (100/labHbA1C × 1). ¹³ These values were converted to mmol/mol to compare with the reported laboratory measured values.

Each of the derived pHbA1c values was then fitted to the respective labHbA1c values for the 35 individuals who had two HbA1c results during this period. The ratio of pHbA1c/labHbA1c was compared between these two separate time points. The pHbA1c derived from CGM was used to enhance direct patient care for each individual (Table 1).

TABLE 1.

The patient sample was balanced by age, sex, duration with diabetes and body mass index (BMI).

Total sample (of whom 2 HbA1c tests)	89 (39)
Median age (IQR)	45 (33–51)
Sex	44 males 45 females
Median BMI (IQR)	26.8 (24–30.4)
Median years duration with T1DM (IQR)	17 (10–27)
Median number CGM values recorded/patient (IQR)	39,467 (32,175–43,852)

Initial analysis of 109 laboratory HbA1c results, where each patient had more than 2000 CGM results prior to their HbA1c blood sample compared with the calculated pHbA1c showed an r ² = 0.81.

Each person's pHbA1c/labHbA1c ratio, at separate time points (separated by 192–461 days), was related to determine the consistency of the relationship over time for the same patient (Figure 1). This gave r ² = 0.52 for the ratio comparison over time, showing reasonable consistency and showing that a significant part of the variation between lab HbA1c and CGM derived HbA1c could be linked to the person rather than being systemic.

FIGURE 1.

Comparing ratio of pHbA1c to Lab HbA1c (as %) between first HbA1c sample and second HbA1c sample for the same patient (N = total days between sample dates). Personal HbA1c (pHbA1c) calculated from average taken from CGM in period (given N days) prior to blood samples compared to laboratory labHbA1c.

Reflecting on these findings, we would suggest that the adjustment of labHbA1c by incorporating recent CGM values to create a pHbA1c for each person has clinical utility. Over a period of up to 18 months, in people with T1D, there was reasonable consistency over time in the relation of pHbA1c to labHbA1c. It can be inferred that the discrepancy between labHbA1c and AG can be greatly reduced by the use of pHBA1c, which would allow more consistent clinical decision making.

The difference between CGM‐derived pHbA1c versus laboratory HbA1c is sufficiently large to influence clinical decisions regarding metabolic health management to be made ¹³ , ¹⁸ and highlights the potential value of pHbA1c.

The link between RBC and blood glucose varies due to many factors. ¹⁶ , ¹⁷ This variation might be sufficiently consistent for labHbA1c values to be personalised to a patient's individual circumstances, including age, sex and BMI. This can be achieved using a period of CGM monitoring in order to generate an apparent glycaemic ratio (AGR) ¹⁸ that can be applied to generate a pHbA1c at future points in time. This means that only once over a period of time would CGM be required for many patients.

The fact that for some individuals the ratio of pHbA1c to labHbA1c did change over time (Figure 1) implies that periodic recalibration to recalculate the AGR would be required. AGR values have been shown to differ between individuals and across various groups of patients with T1D. ¹⁹

In conclusion, adjustment of labHbA1c, by incorporating recent CGM values to create a pHbA1c for each person, has clinical utility. The discrepancy between labHbA1c and AG can be greatly reduced by the use of pHBA1c, allowing for a more consistent basis for clinical decision making.

AUTHOR CONTRIBUTIONS

A.H.H. and M.S. conceived the study. M.S. led on data analysis. A.P. provided expert input in relation to T1D management. J.M.G. and E.J. provided invaluable insight in relation to the context of the study, while M.W. led on interpretation of the glucose monitoring data and the implications of the findings for people with T1D with expert contributions from A.F. and H.H.A. All authors reviewed and approved the final version of the manuscript.

FUNDING INFORMATION

No external funding was used for this study.

CONFLICT OF INTEREST STATEMENT

None of the co‐authors have any conflict of interest.

COMPLIANCE WITH ETHICAL GUIDELINES

This study was a service evaluation exercise in a single clinic. Ethics permission was therefore not required.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.