© 2003 American Thoracic Society
Gold, Silver, and BronzeMetals, Medals, and Standards in Hypersensitivity PneumonitisDivision of Pulmonary and Critical Care Medicine Yale University School of Medicine Yale–New Haven Hospital New Haven, Connecticut As the monetary gold standard was waning, the gold standard in clinical research was gaining currency. Under a monetary gold standard, the value of money is anchored to measurable reserves of the precious metal gold. By analogy, the gold standard in clinical research anchors a patient's diagnosis to another objective measure—results of unequivocal tests. Clinical gold standards work well for conditions in which definitive diagnostic tests exist and interpretive guidelines are clear. Interestingly, the last adherent to the monetary gold standard, the United States, officially abandoned it in 1971 because its restrictiveness limited governmental monetary control and economic growth (1). Similarly, the strict use of the gold standard in clinical research becomes problematic when objective diagnostic tests or their interpretive guidelines are less than sterling. It is similarly problematic when efforts to obtain such gold standard data, broadly applied across large numbers of patients and centers, pose an unacceptable safety risk. Such was the design problem facing the Hypersensitivity Pneumonitis Study Group (2), whose results are reported in this issue (pp. 952–958). The diagnostic criteria for hypersensitivity pneumonitis are in flux. For some, only a biopsy of lung tissue demonstrating ill-formed, airway centered, non-necrotic granulomas in a patient with a plausible exposure to an offensive antigen is accepted as the gold standard, though other histologic variants have recently been described (3–6). Others, citing advances in radiographic imaging—particularly high resolution computed tomography of the chest—propose upper lobe predominant, hazy bronchocentric "ground-glass" nodular opacities with air-trapping in an appropriate clinical setting as sufficient (7, 8). Still others maintain that Th-1–polarized lymphocytic inflammation detected by bronchoalveolar lavage fluid analysis is required, though certainly not specific for hypersensitivity pneumonitis (5, 9–11). Unfortunately, more biologically appealing tests, such as the presence of circulating antibodies to specific antigens, have proven disappointing in both sensitivity and specificity (12, 13). Even the list of culprit antigens and pathologic findings continues to expand and raise fundamental questions of hypersensitivity pneumonitis as a simple host response to chronic exposure or to exposure through chronic infection (14, 15). How can progress be made in such a dust storm of uncertainty? Lacasse and colleagues have chosen an intermediate and pragmatic path (2). In their study of a total of 661 patients from seven countries with lung disease in which hypersensitivity pneumonitis was considered possible, they have used the combined savvy of a panel of clinical experts as the final arbiter of classification. This panel evaluated the composite of bronchoalveolar lavage fluid lymphocytosis, defined as at least 30% for non- or ex-smokers and at least 20% for current smokers; bilateral ground-glass or poorly defined centrilobular nodular opacities; and, in the setting of conflicting or unclear data, histopathology when available or other diagnostic tests to assign a final classification of "hypersensitivity pneumonitis" or "not hypersensitivity pneumonitis." Though perhaps not gold to the purist, this bronze—or perhaps even silver—standard was, by consensus, the best attainable across centers with diverse practices in a disease with substantial inherent diagnostic uncertainty. Once assigned, the patients' clinical histories, physical examination, and laboratory studies were incorporated into a stepwise logistic regression model to quantify the degree to which each feature predicted the assigned clinical diagnosis. This model, derived retrospectively in a 400-patient cohort and prospectively validated in a separate cohort of 261 patients, identified six significant predictors of hypersensitivity pneumonitis: exposure to a known offending antigen, positive precipitating antibodies to the offending antigen, recurrent episodes of symptoms, inspiratory crackles on physical examination, symptoms occurring 4–8 hours after exposure, and weight loss. The probability of hypersensitivity pneumonitis ranged from 0% in patients with none of these features to 98% in patients with all six. Not surprisingly, exposure to a known offending antigen—present in 97% of the patients and 33% of the control subjects—was the strongest predictor with an odds ratio of 38.8. What are the strengths of this study and what new does it tell us about hypersensitivity pneumonitis? Perhaps most important are its size and the breadth of locations throughout Europe and Japan as well as North and Central America. Using a broad patient pool with both derivation and validation cohorts and rigorous statistical analysis, this study identifies and assigns weights to six significant criteria among many that clinicians have previously used. Although similar standards have been touted for years (4, 16), this is certainly the most rigorous derivation and testing to date. What then are the caveats? The most important limitation is that the model is heavily influenced by data from a distinct subset of patients with hypersensitivity pneumonitis—those with bird fancier's disease or farmer's lung. Those two groups comprised 84% of the patients with hypersensitivity pneumonitis and therefore, not surprisingly, the clinical features so characteristic of those diseases figure prominently in the final model. However, hypersensitivity pneumonitis is increasingly recognized as a common clinical problem of diverse etiologies detected only when histologic findings of organizing pneumonia and interstitial pneumonitis, either with or without characteristic poorly formed nonnecrotizing granulomas, are found on lung biopsy in an appropriate clinical setting. Furthermore, the pool of candidates for causative organisms and antigens continues to expand more quickly than the corresponding pool of serologic tests. Finally, clinical experience suggests that, outside of bird fancier's disease and farmer's lung, classic exposure histories and clinical syndromes are few and far between; at least prospectively. Where, then, are we left? Clearly, Lacasse and colleagues (2) have nicely defined the clinical features of two common forms of hypersensitivity pneumonitis. What remains less clear is how well this predictive model will perform in atypical cases caused by rare or as-yet unidentified antigens. Although this study, using their diagnostic standard, quantifies key clinical features of hypersensitivity pneumonitis under particular clinical conditions, using the absence of these criteria alone to exclude all forms of hypersensitivity pneumonitis—particularly in patients with diffuse parenchymal lung disease of unknown etiology—is likely to be a much riskier undertaking. FOOTNOTES Conflict of Interest Statement: D.G.M. has no declared conflict of interest. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
