Lung Diffusing Capacities Normality Thresholds in Pulmonary Fibrosis and Emphysema

The measurement of pulmonary diffusion is the key diagnostic and monitoring tool to examine the integrity of air-blood barrier. Traditionally, the lung diffusing capacity for carbon monoxide (DLco) has been widely used in both interstitial pulmonary fibrosis (IPF) and pulmonary emphysema (PE) and is currently recommended as a standard lung function measure. More recently, the measurement of lung diffusing capacity for nitric oxide (DLNO) was introduced and subsequently standardized on the premise that it is a more specific marker of alveolar membrane diffusive conductance (DM) than DLco. For DLco, several predicting equations were made available over the years, but the changes in equipment and measurement techniques made them outdated, and a new set of standardized reference equations was provided by the Global Lung Function Initiative (GLI). For the DLNO, reference equations were provided by four small studies and the data of three of them were eventually pooled by a European Respiratory Society (ERS) Task Force to derive predicting equations with suitable sample size. Major limitations of this initiative were that no sex-specific separate equations were derived, and the pooled data had been obtained by different devices without robust technical standardization. Sex- and device-specific predicting equations for DLNO have been subsequently made available by two studies, using a single device or including sex and device as covariates.

The choice of reference equations is a crucial issue in the assessment of respiratory function, as it may impact on both definition of normality and assessment of severity. Although for long time the sex-, age- and size-biased percentages of predicted were used for these purposes, the use of z-score is currently recommended as an unbiased estimator of the probability for an individual to deviate from normality. The 5th percentile (-1.645 z-score) of the reference population is currently recommended as the lower limit of normal (LLN5) for all standard lung function measurements including DLco, but for DLNO either LLN5 or the 2.5th percentile (LLN2.5, -1.960 z-score) have been also recommended.

In any case, lung function measures are a continuum and a z-score below a probabilistic threshold such as the LLN5 or LLN2.5 does not necessarily indicate a clinically meaningful abnormality to diagnose a disease but gives the probability of a false positive diagnosis in a healthy subject. By converse, z-scores above these thresholds do not allow to rule out disease with certainty due to the known overlap between healthy and diseased distributions, which may vary depending on various factors including nature and severity of disease, type and method of measures, and reference equations. These proposed limits of normality were based on the distribution in healthy population but not validated for the accuracy to separate diseased from healthy subjects.

The investigators designed the present study with the primary aim to test whether the z-score thresholds of DLco and DLNO separating IPF and PE from healthy subjects may differ from LLN5 or LLN2.5. A secondary aim of the study was to estimate the impact of the choice of test and predicting equations on severity stratification in relation to the extent of lung parenchymal abnormalities typical of IPF and PE. For these purposes, the investigators retrospectively utilized the DLco and DLNO data obtained by a single device in sixty-six subjects with IPF included in two previous studies and in fifty-four new subjects with PE, both documented by quantitative CT of the chest.