© 2002 American Thoracic Society
Acute Interstitial PneumoniaClues from the White StuffAcute interstitial pneumonia (AIP) is a rare but serious idiopathic illness characterized by diffuse alveolar damage with subsequent fibrotic organization and, in many cases, death (1). In the absence of an effective treatment for AIP, management of the ventilated patient is largely supportive and approximately half of patients with AIP die within 2 months. Clear-cut features that distinguish between survivors and nonsurvivors, at or near the time of presentation, in terms of either clinical status (2) or histology (3), have not been identified. The histologic pattern in AIP is indistinguishable from acute respiratory distress syndrome (ARDS), and, as such, AIP may be regarded as a useful and "pure" model for the much more commonly encountered ARDS. In this issue of the AJRCCM (pp. 1551–1556), Ichikado and coworkers report differences in morphology of the lung parenchyma on high-resolution computed tomography (HRCT) between survivors and nonsurvivors with AIP (4). At first sight, it would seem unlikely that the coarse picture provided by HRCT can do better than histologic evaluation. Before considering this study further, it is worth reviewing the concept that differences may exist on HRCT between the apparently identical histologic entities of AIP and ARDS. Tomiyama and coworkers reported that, despite an overlap in HRCT appearances, patients with AIP tended to have more honeycombing and a more symmetrical and lower zone distribution of abnormalities than patients with ARDS (5). Common to most clinical studies of patients in intensive care units, the problems of matching between groups (for example, for disease duration) cannot be overlooked. Nevertheless, the possibility of a difference between AIP and ARDS, as judged by the big picture provided by HRCT, has been mirrored in previous studies in which morphologic differences on HRCT between two apparently identical diffuse lung diseases (6) predated the identification of histologic differences (7). The appeal of HRCT in this context is the ability to get a global picture of a patchy (in time and space) disease process; indeed, HRCT has been used to reveal differences in disease pattern and distribution in patients with ARDS of direct pulmonary versus extrapulmonary cause (8, 9). The basis of the HRCT morphologic evaluation in the current study is the earlier work by Ichikado and coworkers that described a close correlation in AIP between three histologic stages of the disease (exudative, proliferative, and fibrotic phases) and HRCT patterns (increased attenuation, increased attenuation with traction bronchiectasis, and honeycomb change, respectively) (10). Given the likelihood that there will be regional variations in the progression of AIP, the ability of HRCT to provide a precise overall statement about the carbon dating of the disease is probably limited. The fundamental observation provided by this latest study is that the extent of parenchymal opacification with traction bronchiolectasis or bronchiectasis (architectural distortion) was less in survivors of AIP than in nonsurvivors; conversely, the extent of ground glass opacification without traction deformity of the airways was greater in survivors. A total score of the individual CT findings (on a scale reflecting progressive pathologic changes) taken from three lung zones was lower in survivors than nonsurvivors. In essence, these differences all point to increased established fibrosis in the nonsurvivors; this difference was not accounted for by the HRCT scans being obtained at different times in the two groups. The nature of the consolidation (not associated with traction of the airways), which was more extensive in survivors, is of interest. In the context of AIP or ARDS, it is often impossible to distinguish between airspace consolidation and the extensive atelectasis of a heavy leaking lung–both of which will be seen as areas of dense opacification (or white lung) on HRCT. Nevertheless, it is likely that at least some of the nondependent areas of dense opacification correspond to areas of organizing pneumonia. Provided that the organizing pneumonia has not become incorporated as mural fibrosis (which should be reflected on HRCT by the presence of traction bronchiectasis), this should be a reversible, potentially steroid-responsive phenomenon.
Are the findings of this study robust and relevant to clinical practice? First, a couple of methodologic issues need to be considered. Evaluation of the HRCT scans for the various patterns is probably not quite as straightforward as it seems. As presented, the data do not allow any judgment about whether there was any systematic over- or under-calling of a given pattern by the observers. More importantly, the single Can the observations be applied to individual patients to help estimate prognosis? In addition to the issue of observer variation, there are two further points to remember. First, the overlap in HRCT patterns between the survivors and nonsurvivors is such that, as is so often the case with "population" studies, application to the individual patient is hazardous: architectural distortion (the single most important HRCT feature), present in all nonsurvivors (21 of 21), was also present in most of the survivors (6 of 10). Second, it needs to be remembered that the study population consists of patients with AIP, a relative rarity in clinical practice, and not patients with ARDS. Confirmation is needed as to whether prognostication by HRCT pattern holds in patients with ARDS. Given the heterogeneous nature of precipitants and complications in ARDS, the signal may be less clear than in AIP. Nevertheless, any improvement in assessment of prognosis, which is currently largely guesswork, is worth striving for. The potential to stratify patients using macroscopic HRCT patterns could be of great benefit in evaluating new therapies. This study helps us to understand that HRCT can, on occasion, provide unexpected information, not necessarily obtainable from the more traditional parameters of clinical status or histology. DAVID M. HANSELL, M.D. Department of Radiology Royal Brompton Hospital London, United Kingdom REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
coefficient cannot reflect the degree of observer agreement for the individual HRCT patterns; one of the most significant signs, namely architectural distortion, is notoriously difficult to define and identify, particularly for occasional observers. The veracity of the overall score needs at the very least to be tested in a different population: assessment of the ability of this score to predict survival, using a cut-off derived from the same population on which it is subsequently tested, is a famously circular argument. 
