Optimization of performance of Dutch newborn screening for cystic fibrosis

Background: Dutch newborn screening (NBS) for Cystic Fibrosis (CF) introduced in 2011 showed a sensitivity of 90% and a positive predictive value (PPV) of 63%. We describe a study including an optimization phase and evaluation of the modified protocol. Methods: Dutch protocol consists of four steps: determination of immunoreactive trypsinogen (IRT) and pancreatitis-associated protein (PAP), DNA analysis by INNO-LiPA and extended gene analysis (EGA). For the optimization phase we used results of 556,952 newborns screened between April 2011 and June 2014 to calculate effects of 13 alternative protocols on sensitivity, specificity, PPV, ratios of CF to other diagnoses, and costs. One alternative protocol was selected based on calculated sensitivity, PPV and costs and was implemented on 1st July 2016. In this modified protocol DNA analysis is performed in samples with a combination of IRT ≥60 µg/l and PAP ≥3.0 µg/l, IRT ≥100 µg/l and PAP ≥1.2 µg/l or IRT ≥124 µg/l and PAP not relevant. Results of 599,137 newborns screened between 1st July 2016 and 31st December 2019 were similarly evaluated as in the optimization phase. Results: The modified protocol showed a sensitivity of 95%, PPV of 76%, CF to CF transmembrane conductance regulator-related metabolic syndrome/CF screen positive, inconclusive diagnoses (CRMS/CFSPID) ratio 12/1, CF/CF carrier ratio 4/1. Costs per screened newborn were slightly higher. Eleven children, of whom five with classic CF, would not have been referred with the previous protocol. Conclusions: The modified protocol results in acceptable sensitivity (95%) and good PPV of 76% with minimal increase in costs.