Standardisation of Lung Clearance Index in a Multicentre Clinical Trial.

Armstrong DK, Bayfield KJ, Alton EW, Boyd AC, Cunningham S, Elgmati HI, Gill DR, Griesenbach U, Higgins T, Hyde SC, Innes JA, SaundersCJ, Spearing EJ, Davies JC

Thorax, Vol 69, S2, Abstract P209

British Thoracic Society Winter Meeting, London, 2014

Introduction

Lung clearance index (LCI) is a sensitive and repeatable non-invasive measure of ventilation inhomogeneity derived from the multiple breath washout (MBW) technique. It is more sensitive to early lung disease than traditional lung function measurements. Before it can be adopted as a primary endpoint in multicentre trials, it must be demonstrated that it can be applied with minimal interoperator variability. LCI is a major secondary outcome in our gene therapy multidose trial. Aim
To assess LCI achievability and intra- and inter-site agreement.

Method

136 CF patients at two sites with FEV1 50–90% predicted were randomly allocated on a 1:1 basis to receive 12 monthly nebulised doses of active gene therapy product or placebo. LCI was performed in triplicate on seven occasions for each subject using a MBW technique completed on an Innocor device using 0.2% SF6. Stringent quality control criteria have been developed, including offset calculations and minimal acceptable differences between tests. LCI was calculated using customised offline analysis software (SimpleWashout, Igor Pro) by a single operator at each site. To test inter-operator agreement, every seventh MBW from each timepoint was randomly selected, without subject duplication, and used to calculate LCI values by both operators separately.

Results

A total of 854 LCIs were performed during the trial, and technically acceptable measurements were achieved in 95.9% and 94.2% of tests at the two sites (mean 94.8%). 118 (13.8%) of LCIs were analysed independently by two operators, with a full range LCI values represented (range 7.24–19.21). The 95% limits of agreement (LoA) for LCI values were -0.04 to 0.04 (mean difference 0.00) and for FRC values were -0.01 to 0.01 (mean difference 0.00).

Conclusions

Our results demonstrate that LCI is an achievable outcome measure in a multicentre trial in 94.8% of attempts. Separate offline analysis completed by two operators, with appropriate training and knowledge of the test, produces mean LCI and FRC inter-site differences of 0.00. LCI is feasible and appropriate for use as a surrogate endpoint in multicentre clinical trials using stringent methodology.

Armstrong_BTS2014_Figure1.jpg