Background:
In persons with diabetes, glycated haemoglobin A1c (HbA1c) is used as a retrospective measurement of the patient's estimated mean blood glucose over the past 3 months. HbA1c forms in a non-enzymatic pathway when haemoglobin is exposed to blood glucose. The normal range of HbA1c and the correlation to the estimated mean blood glucose is determined from studies in persons with normal erythrocyte turnover of approximately 120 days and without severe chronic kidney disease (CKD). Several smaller studies have shown that HbA1c in persons with type 2 diabetes and CKD, especially on chronic haemodialysis, is an uncertain marker of the mean blood glucose. Generally, studies show that HbA1c underestimate the actual mean blood glucose. The reason for a false low HbA1c in persons with type 2 diabetes and with CKD has yet to be established. However, it is known that erythropoietin treatment and iron infusion increases the erythropoiesis, which results in new non-glycated erythrocytes, and this is likely to lower HbA1c. The erythrocyte life span is found to be reduced in persons on dialysis, which in combination with blood loss from the dialysis also contributes to a reduction in HbA1c. The shorter erythrocyte life span is thought to be due to the toxic uraemic environment as well as mechanical damage caused by haemodialysis.
Objective:
To investigate the correlation between the mean glucose concentration measured by CGM and the estimated mean blood glucose from HbA1c in persons with type 2 diabetes and on chronic haemodialysis compared to a control groups of persons with type 2 diabetes and normal renal function.
Method:
Prospective case-control study over 17 weeks with 40 persons in each group. The case group consist of 40 persons with type 2 diabetes and on chronic haemodialysis. The control group consists of persons with type 2 diabetes and normal renal function (defined as and estimated glomerular filtration rate (eGFR) above 60 ml/min). CGM is performed for a maximum of seven days on week 0, 4, 8, 12 and 16 of the study period with simultaneous analysis of glycaemic markers (HbA1c, glycated albumine and fructosamine). Each of the first five visit was conducted with 3 to 5 weeks interval and the final visit not more than two weeks after the final visit. The erythrocyte life span will be measured with an isotope (Crom-51) method in both the dialysis group and the control group, to establish if diminished erythrocyte life span and falsely low HbA1c is correlated.
Statistical methods:
The null hypothesis is that there is no difference between the ratio of total mean glucose from CGM and estimated mean blood glucose from HbA1c at week 17 when measured in type 2 diabetic persons on chronic haemodialysis compared to type 2 diabetic persons with normal renal function.
The alternative hypothesis is that there is a difference in the ratio between the groups which in a pilot study of persons with diabetes on chronic haemodialysis was found to be 0.16 (mean glucose from CGM/mean blood glucose from HbA1c) when compared to persons with diabetes with out nephropathy.
From one of the pilot arms ώ = 0.148 and σ = 0.128 were extracted. The following scenarios were considered; β(0) = 0 and β (1) = 0, 0.01, ..., 0.2 and n = 40. For each scenario, 5000 data sets were simulated in accordance with the above specifications. Each data set was analyzed by a mixed linear model with treatment arm as fixed effect and person as random effect. The hypothesis β(0) = β(1) was tested at a significance of 5%. The simulated effect in each scenario was calculated as the fraction of rejections in the 5000 tests. Datasets were simulated in the statistical programming language R (www.r-project.org). The smallest difference in the ratio of mean glucose from CGM and mean blood glucose from HbA1c that could be detected with a power of 80% based on the likelihood ratio test described above and using a 5% significance level was 0.1 for n=40 per arm.
