Methods
We included patients with CF who had been using lumacaftor/ivacaftor or ivacaftor for at least 6 months, and who underwent chest CT within 18 months before or after start of therapy. All patients had been in standardized follow up since the moment before start of the CFTR modulator: just before start, patients completed a CFQ-R, underwent pilocarpine iontophoresis sweat chloride testing and pulmonary function testing and measurement of weight and height. All of these measurements were repeated after 6 months of treatment.
In patients starting ivacaftor, a low dose high resolution (HRCT) scan of the chest was performed before the start of treatment in all patients. In the patients starting lumacaftor/ivacaftor, CT scan of the chest was performed by indication only, usually for the exclusion of pulmonary embolism (CT pulmonary angiogram, CTPA: with IV contrast), for evaluation of actual bleeding in patients with hemoptysis (CTPA), or when a patient was screened for lung transplant (HRCT).
Due to the scanning protocol and indication, CTPA did not include expiration slides for evaluation of air trapping. Therefore we calculated total Brody score without air trapping for the group of patients starting with lumacaftor/ivacaftor. In the group of patients who started ivacaftor however, complete Brody score, including air trapping, was available for all patients and this has been used for analysis.
For this study, assessment of Brody score was performed by a board-certified chest radiologist. Brody score was recorded as a total score, total score excluding air trapping, and separate scores on all subdomains9. Total scores and subdomain scores were normalized to a 0-100 scale.
Spirometry was carried out according to the current European Respiratory Society (ERS) guidelines using reference data from the Global Lung Function Initiative (GLI) to calculate ppFEV110,11.
For analysis of the data, IBM SPSS 25 analysis software was used. To assess correlations, the commonly used Pearson R was calculated. In addition Spearman R was calculated. We consider the latter the best applicable method for our purpose, as this method is less sensitive to outliers and suitable for non-linear correlation. To determine statistical significance, the cut-off for p-value was set at 0.05.