Comparison of Blood Glucose Monitoring Systems for Use in Insulin Clamp Studies During Either Intralipid or Glycerol Co-infusions

Free fatty acid-induced insulin resistance can be studied in humans by performing hyperinsulinemic-euglycemic clamp studies with co-infusion of intralipids (HEC-IL).¹ As part of pilot work to establish a HEC-IL protocol in humans, we became aware of discrepancies in glucose values obtained by the HemoCue® Glucose 201 System (HemoCue America, Brea, CA, USA) during the intralipid infusion. This system measures glucose using a photometer detector of the glucose dehydrogenase reaction with demonstrated accuracy compared to reference glucose oxidase analyzer standards during research protocols.^2,3

Package insert information for this system warns that “the measured blood glucose value of lipemic specimens (triglyceride, cholesterol and intralipid) should be interpreted with caution.” As part of a quality control study of four different glucose analyzers, investigators have previously demonstrated up to 30-40% overestimation of glucose concentrations under ex-vivo conditions of increasing Intralipid concentrations using this system compared to the other glucose analyzers.⁴ To date, however, no formal analysis of accuracy of this methodology during a research HEC protocol involving coadministration of intralipids has been reported.

We therefore undertook a formal comparison of glucose measurements obtained by this system compared to the i-STAT (Abbott Point of Care Inc, Princeton, NJ, USA), an analyzer that measures whole-blood glucose using a glucose oxidase-coated amperiometric electrode not affected by blood lipid levels.⁵ Seven subjects (mean ± SD, age 30 ± 9 years, BMI 27.9 ± 3.2 kg/m²) with mitochondrial mutations causing partial fatty acid oxidation defects underwent either a 5-hour HEC-IL (20% IL, 1.5 mL/min plus heparin) or a glycerol plus heparin control infusion (HEC-CON) in a randomized, cross-over fashion on separate days. A total of 285 samples were obtained during the HEC-CON and 256 samples were obtained during the HEC-IL. Glucose values measured by the two methods were highly correlated during the HEC-CON (Spearman rank order correlation, r = 0.82, P < 0.001), but less so during the HEC-IL (r = 0.41, P < 0.001). During the HEC-CON, a total of 19.3%, 64.6%, and 93.7% of glucose measurements by dehydrogenase reaction were within 5%, 10%, and 15% of measurements by glucose oxidation, respectively, and Clarke error grid analysis⁶ (Figure 1, upper graph) showed that 97.9% of the readings were in zone A, 1.7% were in zone B, and 0.4% were in zone D. On the other hand, during the HEC-IL, only 5.1%, 13.3%, and 20.7% of glucose measurements by dehydrogenase reaction were within 5%, 10%, and 15% of values measured by the glucose oxidation method, respectively, and 39% were >50% higher. The Clarke error grid analysis (Figure 1, lower graph) showed that 31.2% of the readings were in zone A, 68.4% were in zone B, and 0.4% were in zone D.

Figure 1.

Clark error grids of glucose levels obtained by iSTAT and HemoCue Glucose 201 System during a hyperinsulinemic-euglycemic clamp and either a co-infusion of a control solution of glycerol (upper graph, n = 285) or 20% intralipid infusion (lower graph, n = 256).

This study confirms HemoCue Glucose 201 System overestimates glucose levels, often by >50%, during intralipid infusions. Documentation provided by the parent company (Angelholm, Sweden, Technical Letter No 20, December 2011) suggests this error results from intralipid-induced turbidity that interferes with the photometer detector of the glucose dehydrogenase reaction. Instead, glucose measurements utilizing the oxidase method should be used during HEC studies involving intralipid infusions.