A Prospective Study of Correlation of Fructosamine with Continuous Glucose Monitor in Diabetic Patients

Introduction

Diabetes mellitus (DM) is a disease caused by the pancreas’ inability to produce an appropriate amount of insulin or the body’s inability to use the insulin produced effectively [1]. The resultant hyperglycemia has been shown in multiple studies to increase the risk of micro vascular and macro vascular complications in people with diabetes [2, 3, 4, 5, 6]. Achievement and maintenance of good glycemic control is therefore a critical goal in diabetes management [2, 3, 4, 5, 6].

Although hemoglobin A1C is the best validated assessment of long-term glycemic control [4, 7, 8], there are important clinical scenarios where a shorter-term estimate of glycemic control is needed. For example, in red blood cell disorders, or in situations with rapid changes in glucose homeostasis [4, 9, 10, 11], A1C does not accurately reflect recent glycemic control. The limitations of A1C present an opportunity for alternate glycemic markers, such as fructosamine, to be of clinical use in monitoring glycemic control [12, 13, 14, 15, 16]. Fructosamine’s shorter half-life may reflect average glucose changes over 2-3 weeks, in contrast to the 120 days required for A1C to stabilize. Unfortunately, fructosamine’s lack of standardization for estimating the average glucose has greatly limited its clinical utility. This objective of this study is to validate fructosamine for estimating the average glucose over a 2-week interval by comparing with average glucose obtained from continuous glucose meter.

Statistical Approaches

Univariate statistical approaches (frequency distributions and analyses of normality) were used to describe the demographic and clinical features of the patients from whom fructosamine levels were measured. Data found to be normally distributed were subjected to parametric statistics; otherwise, data were analyzed using non-parametric analyses. Given that few hematocrit values were available (n=6), this measure was excluded from the binary logistic regressions.

Bivariate Pearson correlation and Chi-square analyses and multivariable regression analyses were used to examine the linear relationships of the potential covariates of average glucose levels. Potential confounders were stratified into normal and abnormal ranges to discern if abnormal ranges were associated with higher fructosamine values; normal hemoglobin was considered to be from 12-16g/dL, normal hematocrit 36-47%, normal albumin 3.5-5g/dL and normal GFR > 60mL/min/1.73m². Covariates showing no correlative strength with CGM glucose were removed from the linear regression analyses in a backward elimination fashion. Iterative sensitivity analyses, i.e., multiple Chi- square analyses, were used to find the optimum threshold of fructosamine that gave the best specificity for predicting glucose values of < 200mg/dL.

Results

The majority of participants was male (55.05%), caucasian (88.40%), and had type II diabetes (66.90%) with a mean age of 59.74 years (+ 15.44) (Table 1). The CGM devices used by these participants were made by Librepro (76.0%), Dexcom (16%), and Medtronic (8%).

As shown in Table 2, binary logistic regressions models were constructed that first contained all potential confounders of fructosamine levels and normal/abnormal designations for hematocrit, GFR, albumin, and hemoglobin. These possible confounders were collected within 2 days of CGM download and none of these potential confounders were found to be statistically impactful on CGM glucose levels, as evidenced by their p-values of > .05 and 95% CIs that include 1. After the removal of potential confounders, binary logistic regression showed that those with a fructosamine value of 300mMol/L had 39% odds (95% CI 5.08 - 294.0) of having an average glucose value of < 200mg/dL, P < .001 (Table 3).

a. Variable(s) entered on step 1: hemoglobin2, GFR2, albumin2, fruc_grp300. Table 2: Regression of potential confounders to average CGM Glucose.

Discussion

Fructosamine is a glycated ketoamine discovered over 35 years ago. It can be measured inexpensively, precisely, and relatively free of interference [16, 17]. It can be automated for use with micro-sample volumes [17]. Fructosamine is relatively unaffected by RBC diseases, making it attractive as a suitable glycemic marker for people affected by RBC abnormalities such as sickle cell disease/trait. Compared to alternate measures of average glucose, its lower cost makes it more accessible to populations with limited resources [17], In our prospective study, we found that fructosamine has a strong relationship with average glucose as determined by continuous glucose measurement. Multiple Chi-square analysis showed that 98% of patients having a fructosamine of 300mMol/L or less had a CGM estimated 2-week average glucose of less than 200mg/dL.

There are several scenarios where this estimate of the average glucose would be clinically important. In the assessment of operative risk, knowing the average glucose is less than 200 immediately prior to scheduled surgery verifies that diabetes is under good control and thereby does not represent an increased risk for an untoward surgical outcome. For example, a recent study by Shohat et al found that preoperative fructosamine value of ≥ 292µmol/L had a significant correlation with deep infection in patients undergoing total hip and knee arthroplasty [18].

For patients with previously uncontrolled diabetes undergoing medication adjustment to get their blood sugar under control, documenting a clinical response over two weeks rather than waiting three months for the A1C to change would be useful. Although frequent point of care capillary glucose measurements can provide a reasonable estimate of average glucose, for many patients this approach is either unacceptable or not feasible. When compared to the use of a continuous glucose measurement study to establish average blood sugar control, fructosamine is substantially more convenient, quicker, and less costly.

There may be circumstances where a more precise estimate of the average glucose is needed. When the fructosamine is less than 300mg/dL (average glucose < 200mg/dL), the average glucose can be estimated: Fructosamine x 0.57=average glucose (Table 3). Unfortunately, when the fructosamine is greater than 300mMol/L, its relationship to the CGM estimated average glucose becomes unpredictable. A similar weakness is shared by the hemoglobin A1C, where it has been shown that when the A1C exceeds 8% there is no statistical correlation with the CGM-assessed average glucose [19].

Clinical conditions that affect protein metabolism such

as nephrotic syndrome, thyroid disease, and liver disease may have an influence on fructosamine level and its ability to accurately assess average glucose [4]. We did not find any impact of albumin on fructosamine’s ability to predict average glucose. We also did not find any impact of the standard deviation of coefficient of variation on fructosamine.

We are not aware of any prior study of more than 100 subjects that evaluated the correlation between fructosamine and CGM determined average BG level. Limitations of this study include lack of diversity, with the caucasian population comprising 88.4% of total participants. Potentially confounding factors, such as creatinine, hemoglobin, hematocrit albumin were not available in 50-65% of cases at the time of download.

Conclusion

On occasions where the short-term determination of average blood sugar control is important, and alternative assessment strategies are not feasible, serum fructosamine is a convenient, cost-effective, and reliable option. Knowing that the average serum glucose is less than 200mg/dL in 98% of patients with a fructosamine of less than 300mMol/L provides useful and reliable information that can significantly improve decision making in a variety of clinical situations.