• chronic obstructive pulmonary disease
• computed tomography
• lung densitometry
• outcome measures

Forced expiratory volume in 1 second (FEV₁) is by far the most well established outcome variable in obstructive pulmonary disease. Numerous studies have documented the correlation of this parameter with clinical variables such as severity of disease and mortality,¹ and spirometric measurements have been standardised by international recommendations on lung function testing.² Nevertheless, in real life the relevance of a maximal blow through a narrow tube is not always self-evident, and the intuitive clinical meaningfulness of this surrogate parameter is therefore perhaps less obvious. FEV₁ has further limitations in chronic obstructive pulmonary disease (COPD). In general, dynamic lung volumes such as FEV₁ and forced vital capacity (FVC) are highly effort dependent. However, in emphysematous subjects an abnormally low FEV₁ is partly caused by the dynamic collapse of the airways which is also effort dependent. Therefore, in COPD the result of a more moderated manoeuvre is usually superior to the result of a maximal effort. This phenomenon adds to the variability of repeated measurements and, in COPD, the standard deviation of repeated measurements of FEV₁ is larger than the annual decline, even in heavy smokers.³ For these reasons there is an increasing interest in other measures such as number of exacerbations or disease specific questionnaire scores as alternative outcome measures for monitoring the progress of emphysema in randomised clinical trials.

Another new outcome measure is lung density as determined by computed tomographic (CT) scanning. Although the scanner has mainly been used as an imaging device, in the 1970s when Hounsfield developed CT scanning for clinical use he envisaged the scanner as a measuring device as well, because it provides precise information on the density of tissues derived from the attenuation of x rays. In this context it should be noted that pulmonary emphysema is defined pathologically as “the abnormal permanent enlargement of airspaces distal to the terminal bronchioles due to destruction of their walls, without obvious fibrosis”.⁴ In other words, loss of lung tissue is an essential and inevitable part of the emphysematous process, and no other reasonably common and diffuse lung disease shares this feature with emphysema. CT based densitometric parameters therefore have the potential to be both sensitive and specific outcome measures for monitoring the progress of emphysema.

Even in the late 1970s and early 1980s emphysema was described by CT scanning.^5,6 Since that time, several studies have compared CT and pathological findings and—with improved resolution, faster scan times, and thinner collimation—the correlation between CT scores and pathological grading has improved. Furthermore, recent studies seem to indicate that CT lung density is more reproducible than traditional spirometric variables such as FEV₁.⁷

Two papers in this issue of Thorax add important data to the validation of CT lung density as an outcome measure for monitoring the progress of emphysema. In both studies participants suffered from severe α₁-antitrypsin (AAT) deficiency (assumed PI*ZZ genotype) and both studies included St George’s Respiratory Questionnaire (SGRQ) data, lung function tests (FEV₁ and transfer coefficient (Kco)), and CT lung density measurements. Dawkins et al⁸ followed about 200 subjects for a mean of 2 years. Twenty subjects who died after enrolment in the study had lower FEV₁ (percentage predicted), Kco (percentage predicted), higher CT emphysema index (threshold –910 HU), and higher SGRQ scores indicating worse health status than survivors. Subsequent multiple regression analysis (Cox survival) showed that the CT emphysema index was the most powerful predictor of mortality followed by the SGRQ activity score, whereas age and lung function measurements had no independent influence on survival. In the study by Stolk et al⁹ 22 individuals were followed for 30 months and a significant correlation was found between changes in CT lung density (percentile density) and changes in health status (SGRQ) but, surprisingly, no correlation was seen between these variables and changes in pulmonary function measurements (FEV₁ and Kco).

Patients with severe AAT deficiency provide a good model for studies of emphysema because they develop the disease at a relatively young age when health status is less affected by co-morbidities that become more prevalent with increasing age. However, these studies do also have limitations. Although 200 subjects is a large group for studies of AAT deficiency, it is a more moderate number compared with other studies in usual COPD and the results obtained in AAT deficiency may not necessarily be applicable to the much larger population of smokers with usual COPD.

From a methodological point of view, it is interesting to note the difference between the two studies in the selection of the CT densitometric parameter and CT protocol for image acquisition. Thus, Dawkins et al used the emphysema index and an HRCT protocol (that is, thin slices and a hard reconstruction) whereas Stolk et al chose the percentile density and a volume scan protocol (that is, thick slices and a soft reconstruction algorithm). Owing to photon statistics, thick collimation and soft reconstruction usually give more reliable results and the percentile density seems to be more robust than the emphysema index, at least in longitudinal studies of the progress of emphysema.¹⁰

The radiation burden is always an important consideration when using x rays for monitoring disease, and the radiation dose of a standard CT scan of the chest is by no means negligible. However, by reducing the current of the x ray tube, it is possible to keep the total radiation dose of a full volume scan of the lung below 1 mSv without significant loss of information on lung density. This low dose technique opens up the possibility for longitudinal studies with repeated CT scans in the same subject. Volume scanning has two important advantages: (1) using modern multi-slice techniques the scan can be performed in one breath hold, and (2) the amount of air in the lungs can be calculated from the images. Lung density is obviously very dependent on inspiratory level, and the volume of air in the lungs can be used to adjust lung density for the inspiratory level.¹⁰ When analysing trends in longitudinal data such as in the study by Stolk et al, adjustment for lung volume is unavoidable because the total lung volume of a subject may vary significantly from one examination to the next.¹⁰ This kind of noise reduction may be less important in cross sectional studies such as the one by Dawkins et al,⁸ and it may even introduce new errors. An increase in total lung volume is an inherent part of the emphysematous process and, by eliminating that aspect of the disease, volume adjustment may in fact weaken the correlation between CT lung density and other measures of disease severity (unpublished data). Thus, adjustment of lung density for lung volume is not always to be recommended.

The two studies published in this issue of Thorax underline the urgent need for standardisation and international agreement on recommendations for lung density measurement based on CT scanning. However, provided CT lung density can be standardised and validated against traditional clinical outcome variables, it may prove to be a new measurement that is objective, specific, and sensitive for monitoring the effect of new drugs on the progress of emphysema in future randomised clinical trials.