(12) Using continuous glucose monitoring to predict disease progression and development of diabetes in youth with cystic fibrosis

Background: Cystic fibrosis-related diabetes (CFRD) is one of the most common complications of CF and is associated with undernutrition, decreased pulmonary function, and increased mortality. Clinical decline has been described up to 5 years before the diagnosis of CFRD and may be related to early dysglycemia. Continuous glucose monitoring (CGM) can detect early glucose abnormalities before they develop during a 2hr oral glucose tolerance test. However, no clear CGM based cut-points for predicting clinical decline in people with CF have been defined. Our aim was to explore CGM as a tool for prospectively predicting CF-relevant clinical outcomes in youth with CF.

Methods: This was a single center, prospective analysis of data collected as part of the GlycEmic Monitoring in CF (GEM-CF) study. Participants were ≥6 years of age, diagnosed with CF, and followed at Children’s Hospital of Colorado CF center. Participants wore a CGM (iPro®2) between 2014-2018. Participants with diagnoses of CFRD on insulin or other medications affecting glucose metabolism at time of CGM wear were excluded from this analysis. The following variables were collected at baseline (ie time of CGM wear) and prospectively through early 2020: forced expiratory volume in 1 second (FEV1), forced vital capacity (FVC), body mass index z-score (BMI-z), time to diagnosis of CFRD, and time to insulin start. Mixed effects models were used to test association between baseline CGM variables and change in BMI-z, FEV1% predicted (FEV1pp), and FVC% predicted (FVCpp) over 1 year and 3 years. Cox proportional hazards models were used to test the association between baseline CGM variables and time to diagnosis of CFRD and time to insulin start.

Results: A total of 75 individuals with CF were included. Baseline age was 13.36±3.48 years, n=42 (56%) were female, and n=5 (6.7%) were pancreatic sufficient. Baseline mean BMI-z was -0.04±0.80, mean FEV1pp was 95.9±14.7%, and mean FVCpp was 102.7±13.2%. Mean±SD for the following CGM variables were: average sensor glucose=112±15 mg/dL, %time >140 mg/dL=13.1±13.2%, %time >200 mg/dL=1.3±3.1%, coefficient of variation=20±6, mean amplitude of glycemic excursion (MAGE)=72±35 mg/dL. CGM variables did not predict change in BMI-z, FEV1pp, nor FVCpp at 1 year nor 3 years post-baseline (Table 1). For every 5 unit increase in CGM mean glucose and CGM %time spent >140 mg/dL there was a 22-23% increased risk for the development of CFRD (p< 0.05). There was also an increased odds for insulin start with every 5 unit increase in these CGM parameters (p< 0.01, Table 1). A 5 unit change in CGM mean amplitude of glycemic excursion (MAGE), a measure of glycemic variability, was also associated with increased risk for starting insulin (p< 0.05, Table 1).

Conclusions: The CGM variables explored in this group of relatively healthy, young individuals with CF who were not yet on insulin, did not predict clinical decline as measured by 1 and 3 year changes in BMI z-score, FEV1pp, nor FVCpp. However, mean glucose, %time spent >140 mg/dL, as well as a measure of glycemic variability on CGM were associated with increased odds for developing CFRD and requiring insulin over the study period.

Acknowledgements: This study was supported by funding from CFF grant CHAN16A0, CFF Student Traineeship Award LORENZ22H0, and UL1 TR001082 (CCTSI and REDCap).

References: