Validation of Interstitial Fluid Continuous Glucose Monitoring in Cystic Fibrosis

Diabetes is a complication of cystic fibrosis (CF) that is of growing clinical importance. The recognition of diabetes complications in CF subjects (1) has emphasized the need for more accurate monitoring of glycemia. This is complicated by the many factors affecting glycemia in CF subjects, ranging from the consequences of malabsorption to the caloric burden of supplemental nutrition, as well as the metabolic effects of infection and drugs (2). The recent introduction of devices that provide a continuous glucose profile has revealed clinically relevant excursions in glycemia previously overlooked by conventional measures (3). They are able to provide the detailed glucose profile required in CF patients before and after an established diagnosis of cystic fibrosis-related diabetes (CFRD). Many of these new methods rely on sampling glucose levels in interstitial fluid and its correlation with plasma glucose. They have been shown in non-CF subjects to be strongly correlated with capillary and plasma glucose values (4–8). However, an individual continuous glucose monitoring system (CGMS) value may differ considerably from plasma glucose measured simultaneously (9). Altered subcutaneous tissue composition in CF might affect cellular and interstitial fluid kinetics and reduce the reliability of these devices. To date, the validity of these devices in CF patients has not, to our knowledge, been tested.

We recruited 21 (14 male, age 27 ± 12 years [mean ± SD]) nondiabetic CF subjects age-matched with 21 (8 male, age 29 ± 8 years) nondiabetic non-CF control subjects. After an overnight fast each subject underwent insertion of a CGMS (MiniMed, Sylmar, CA) followed by an Oral Glucose Tolerance Test (OGTT). The CGMS remained in place for another 48 h before it was removed, and the data were downloaded. During this time capillary blood glucose (CBG) samples were performed four times each day using a Precision Glucose Sensor (MediSense), and the values were entered into the CGMS. Any subjects with <24 h of data were excluded from analysis. There were no adverse events and tolerability of the device was excellent in both groups.

Comparison of paired CBG/CGMS values revealed a correlation coefficient of 0.77 (P < 0.001) for the CF group and 0.70 (P < 0.001) for the control group. The mean absolute difference (mean ± SD) between the CBG/CGMS values was 10.7 ± 8.7% for the CF group and 10.5 ± 8.7% for the control group.

Paired venous/CGMS values, obtained in 14 CF subjects and 15 control subjects, revealed comparable yet weaker correlation coefficients of 0.57 (P < 0.001) and 0.36 (P < 0.001) for the CF and control groups, respectively. The mean absolute difference between the venous/CGMS samples was 24.9 ± 21.0% for the CF subjects and 28.8 ± 22.9% for the control subjects. Individual data revealed stronger correlation coefficients, with a median of 0.94 (range 0.42–0.99) in CF subjects and 0.86 in control subjects (0.44–0.98). Ten of 14 CF subjects and 11 of 15 control subjects had a correlation coefficient >0.8. The mean CGMS values were highly correlated with the mean plasma glucose values at each of the five points of the OGTT (correlation coefficient CF 0.94 [P = 0.005], control 0.93 [P = 0.008]). These results are consistent with previous studies in non-CF subjects, which have shown that CGMS in an individual reflects the trend but not the absolute value in plasma glucose (9). However, the the mean values obtained by CGMS for a group of subjects will closely reflect the mean of simultaneous plasma values.

We conclude that this CGMS method is well tolerated in CF and that the correlations seen between capillary and plasma glucose values are similar to those seen in the non-CF patients. The CGMS is therefore appropriate to use in CF. However, as in non-CF patients, the mean absolute difference between CGMS and plasma means caution is needed when interpreting solitary CGMS values.