© 2006 American Thoracic Society doi: 10.1164/rccm.2510004
Classification of Interstitial PneumoniasWhat Do Gene Expression Profiles Tell Us?University of Michigan, Ann Arbor, Michigan
Royal Brompton Hospital, Fulham, London, United Kingdom The 2002 Joint Consensus Statement by the American Thoracic Society (ATS) and the European Respiratory Society (ERS) defined a new classification of the idiopathic interstitial pneumonias (IIPs) (1), primarily intended as a framework for studies of these difficult diseases. In some respects, the ATS/ERS initiative has been clinically helpful. Idiopathic pulmonary fibrosis (IPF), the most prevalent and malignant of the IIPs (2, 3), is now diagnosed with greater precision than ever before, based on the proposed histopathologic and clinical criteria. However, the entity of nonspecific interstitial pneumonia (NSIP) has posed major diagnostic problems for practicing clinicians. The prevailing difficulty has been the clinical heterogeneity of NSIP. High-resolution computed tomography findings in NSIP include features suggestive of organizing pneumonia, hypersensitivity pneumonitis (HP), and, occasionally, IPF (4). The response to immunosuppressive therapy of NSIP is similarly variable. The generally favorable outcomes associated with computed tomography findings overlapping with organizing pneumonia (5) contrast with the higher mortality in patients with NSIP presenting with the clinical features of IPF (2). The histologic subclassification of NSIP as cellular or fibrotic does not satisfactorily account for such diversity. In this issue of the Journal (pp. 188–198), Selman and colleagues (6) sought to compare whole-genome gene expression profiles in lung tissue obtained from patients with IPF, NSIP, and HP. First, they identified the genes that best distinguished IPF (n = 15) from HP (n = 12) using the Leave-One-Out-Cross-Validation (LOOCV) statistical method to estimate misclassification errors. Using this method, they found minimal error rates and high statistical significance for a group of 354 genes that were upregulated in IPF and 595 genes in HP. These "signature" genes were then queried in a Gene Ontology functional annotation database (DAVID [Database for Annotation, Visualization and Integrated Discovery]: http://apps1.niaid.nih.gov/david/). The HP gene expression signature was enriched in genes functionally associated with immune/inflammatory responses and T-cell activation, whereas the IPF-related genes were associated with tissue remodeling and epithelial/mesenchymal differentiation. The investigators then compared the genes that distinguished IPF from HP (in a larger set of 1,058 genes) with the gene expression profiles in patients identified with idiopathic NSIP.
This is the first published study to undertake whole-genome expression profiling comparing distinct groups of patients with idiopathic and nonidiopathic interstitial pneumonia. This work complements and adds to prior studies using similar approaches that compared IPF to normal lung (7, 8). Despite its many limitations and difficulties, microarray analyses have the potential to advance our understanding of disease pathogenesis, improve diagnostic accuracy and prognostication and, possibly, aid in determining whether specific therapies (e.g., antiinflammatory drugs) may be beneficial. Such goals are more difficult when whole tissue–based microarray technologies are applied to heterogeneous disease processes with limited numbers of patients in specific subgroups. The greatest value, at the present time, of genomewide expression analyses is in elucidating key aspects of disease pathogenesis. This has been made possible by the rapidly expanding field of bioinformatics and computational analyses that make the critical connections from vast compendia of otherwise uninterpretable sequences/genes into plausible biological mechanisms. The quality of these functional annotation databases is likely to improve as we learn more of the function(s) of specific genes within the human genome. In the study by Selman and colleagues, signature genes within the HP and IPF subgroups were attributable to specific functional classes that provide important insights into disease mechanisms. An interesting finding is the preponderance of immune/inflammation-related genes in HP and their relative absence in IPF. It is important to recognize, however, that the design of this analysis could have excluded genes (possibly some immune-related) that participate in IPF pathogenesis. Functional annotations were done only on those genes that were selected to be "discriminatory" or significantly different in expression between the two groups; thus, it is possible that a subset of genes that may have been upregulated in both diseases were not identified. Nevertheless, distinct "functional" gene signatures provide important clues to the observed differences in clinical course, prognosis, and responses to therapy in these two disease processes. Although not proven, immune-response genes (e.g., T-cell/lymphocyte activation) identified in the HP signature, but not in IPF, may account for differences in responsiveness to corticosteroid therapy. It is intriguing that there is activation of a tissue-remodeling program that resembles developmental organogenesis in IPF, but not in HP. Of particular interest is the activation of the Wnt/ The differential gene expression profile that separates IPF from HP (in the "training set"), when queried for concordance to gene expression patterns in NSIP, provided some additional insights. In three of eight NSIP cases, gene expression patterns exhibited significant overlap with IPF or HP. This observation is consistent with the accumulating clinical–radiologic–pathologic evidence for a number of clinical subtypes of NSIP (12–14). Thus, both clinical and cellular/molecular data support the concept that NSIP represents a histopathologic pattern associated with a variety of idiopathic and nonidiopathic lung disorders. Clinicians confronted with a histopathologic diagnosis of NSIP should first consider an associated clinical condition, in particular HP, IPF, organizing pneumonia, or an underlying connective tissue disease. Perhaps, gene expression signatures (or specific candidate genes) will eventually aid us in the differential diagnosis of NSIP cases in which the associated clinical condition or etiology is not known. The separation of a true "idiopathic" NSIP patient group based on the current data is more problematic. Have the authors truly identified an idiopathic NSIP subgroup that is distinct from IPF? Or, are "classical" IPF and "idiopathic" NSIP, in reality, the ends of a continuum of "fibrotic idiopathic interstitial pneumonia" (15)? Further studies with greater numbers of patients are required to determine if an NSIP gene expression signature, distinct from IPF and other NSIP-associated conditions, can be identified. If so, functional gene annotations of this signature and the identity of discriminative genes may shed light on the pathogenesis of idiopathic NSIP. The work by Selman and colleagues provides important genetic evidence that, in concert with current clinical evidence, makes a compelling argument for the future clinical subclassification of NSIP. FOOTNOTES Conflict of Interest Statement: Neither author has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-catenin pathway and epithelial–mesenchymal transition in IPF, as suggested in other recent studies (
