Use of nasal potential difference and sweat chloride as outcome measures in multicenter clinical trials in subjects with cystic fibrosis.

One of the goals of current research in cystic fibrosis (CF) is to develop treatments that correct or compensate for defects in function of the cystic fibrosis transmembrane regulator (CFTR) gene. The use of outcome measures that assess CFTR function such as nasal potential difference (NPD) measurements and sweat chloride determinations will be required to evaluate the efficacy of such treatments in multicenter clinical trials. The purpose of this work was to identify the sources and magnitude of variability in NPD and sweat chloride measurements when performed at multiple centers. For the variance component analysis presented here, we used NPD and sweat chloride measurements from 37 subjects with CF participating in a phase I, four-center clinical trial of CPX (8-cyclopentyl-1,3-dipropylxanthine), a drug intended to enhance trafficking of Delta F508 CFTR to the cell membrane. The specific techniques used to measure these outcomes were not standardized, and varied between the four sites. Variability of both NPD measurements (baseline potential difference during infusion with Ringer's solution; change in response to addition of 0.1 mM amiloride; and subsequent change in response to perfusion with low chloride solution containing 0.1 mM amiloride and 0.01 mM isoproterenol) and sweat chloride measurements differed significantly between study sites. For change in NPD, one study site had significantly greater variability (lower reproducibility) of measurement than the other three sites. For sweat chloride measurements, reproducibility was lower at two of the sites relative to the other two sites. Sample size calculations showed that lower reproducibility at one or more sites can substantially reduce the power of studies using NPD or sweat chloride determinations as outcome measures. Standardization of measurement protocols, careful operator training and certification, and ongoing monitoring of individual operator performance may help to improve reliability in multicenter trials.