Maltose interference-free test strips for blood glucose testing at point-of-care: a laboratory performance evaluation

Maltose interference is a concern with blood glucose testing at point-of-care. We evaluated a maltose interference-free test strip (with a modified recombinant glucose dehydrogenase-pyrroloquinoline quinone system) for the Accu-Chek(®) Performa glucose meter (Roche Diagnostics, Mannheim, Germany).

Blood specimens (n = 120) sent for clinical laboratory glucose testing were used in assessing performance characteristics, including imprecision, linearity, clinical impact analysis, and method comparison, of the test strips. To evaluate sugar interference, two heparinized blood specimens were spiked with maltose, xylose, and galactose (up to 500 mg/dL) followed by testing with modified Performa, Accutrend(®) (Roche Diagnostics), and Advantage II (Roche Diagnostics) test strips and by the laboratory method.

Test strips demonstrated total laboratory coefficients of variation of <7%; within-run coefficients of variation were 2.7-5.4% for blood glucose at 2.5-19.7 mmol/L. Clarke Error Grid analysis of the 120 results (0.8-27.6 mmol/L) showed all values to be within critical clinical limits. Comparison with laboratory results gave 0.960 correlation (Spearman's r(2)) with a Deming regression y (Performa) = 0.95x (laboratory) - 0.11 mmol/L (SEy|x0.06 mmol/L). A slight negative bias (-0.5 mmol/L) was demonstrated with the Bland-Altman difference plot. Maltose (up to 13.9 mmol/L) and xylose (33.3 mmol/L) had no effect, but galactose (2.2 mmol/L) showed interference. The sugars also affected test strips for Advantage II but not Accutrend glucose meters. With International Organization for Standardization ISO 15197:2003 criteria, 99% of the 120 results determined by the test strips were within the minimal acceptable performance; only one of 106 (5.9 mmol/L) was >20% from the laboratory result.

The modified and improved Performa test strips were not affected by maltose and xylose. They meet ISO 15197:2003 requirements with a slight bias (-0.5 mmol/L) compared to the laboratory method.