Subclinical Interstitial Lung Abnormalities: Toward the Early Detection of Idiopathic Pulmonary Fibrosis?

The beneficial treatment effects on pulmonary function trends with the use of antifibrotic agents in idiopathic pulmonary fibrosis (IPF) have prompted a focus on key IPF subgroups, including patients with “mild disease” as judged by FVC levels. In recent analyses, treatment effects have been virtually identical above and below FVC thresholds of 70% (1) and 80% (2). These findings have led to increasing interest in the identification of subclinical IPF, with a view to early intervention. Recent studies of subclinical “interstitial lung abnormalities” (ILAs) on computed tomography (CT) scans in large cohorts provide a conceptual basis for a possible screening strategy. In this issue of the Journal, Araki and colleagues (pp. 1514–1522) have extended their previous observations and data reported by other groups in the first report of the prevalence of ILA and patterns of progression (as judged by serial high-resolution CT [HRCT] and pulmonary function trends) in a general population, consisting of adult participants in the Framingham Heart Study (3). They report that in a large Framingham Heart Study cohort, progression of ILA was associated with an increased risk of pulmonary function decline and death.

The first studies of subclinical ILAs were performed by subjective HRCT evaluation in family members of patients with familial IPF (4) and healthy elderly patients (5). In the latter study, there was a high prevalence of limited subpleural reticular abnormalities in participants older than 75 years of age, raising the possibility that ILA might represent a normal aging phenomenon. However, in subsequent much larger cohorts (6–16), mostly evaluated by automated CT methodologies, a very different picture emerged. Clinical associations with ILAs and with IPF were strikingly similar. ILAs increased in prevalence with advancing age (8–10, 12, 15), were more prevalent in smokers (6, 9, 13, 15), and, in studies confined to smokers, increased in prevalence with increasing tobacco exposure (7, 8, 10) or with active smoking (3). ILAs were associated with reductions in resting pulmonary function indices (6, 9, 15, 16) and a diminution in exercise capacity (11). The 6% prevalence of ILA in the general population, reported by Araki and colleagues (3), was similar to that in other recent cohorts (15). Importantly, in the Framingham Heart Study population, there was a significant association between ILAs and the MUC5B promoter polymorphism in participants older than 50 years of age (12), similar to the association previously reported in IPF. In the study of Araki and colleagues (3), the MUC5B genotype was associated with serial progression of ILAs.

Taken together, these observations provide strong indirect support for the view that screening for ILAs might eventually provide a means for the early identification of IPF. However, a number of problems, exemplified by the findings of Araki and colleagues (3), need to be addressed if this goal is to be achieved. The grouping together of ILAs as a single entity in the hope of early IPF diagnosis is unlikely to be fruitful. The prevalence of ILAs is substantially higher than the prevalence of IPF. Limited CT abnormalities due to respiratory bronchiolitis are well recognized in smokers and must necessarily contribute to the higher prevalence of ILAs in smoking subjects. Given that ILAs can, in fact, be characterized morphologically, exactly as is routine practice in established interstitial lung disease, it is increasingly important that abnormalities compatible with early IPF (i.e., subpleural reticular ILA) be evaluated more selectively than has been the case in many reports. The separation between fibrotic and nonfibrotic ILAs, made in a number of studies (3, 8–14), is not sufficient in itself, as limited fibrosis is variably present in smoking-related respiratory bronchiolitis and in many other non-IPF contexts. Even in studies in which ILA subgroups have been clearly described, important clinical associations have mostly been defined for ILAs as a single amalgamated entity.

It is also important that terminology be standardized, especially regarding the important distinction between fibrotic and nonfibrotic ILAs. In the study of Araki and colleagues (3), for example, more than 80% of ILAs were reticular and subpleural but less than 10% of ILAs were categorized as “definite fibrosis.” By contrast, 39% of ILAs were considered to be wholly or partially fibrotic in an earlier report (13). Given the problem of powering analyses in ILA subgroups, it is essential that agreement on ILA definition be reached to allow amalgamation of clinical associations in a number of cohorts. The accurate identification of subclinical IPF may require equal weighting to be given to the automated detection of ILAs and, once identified, their subjective evaluation by experienced radiologists.

However, even the standardized selective evaluation of fibrotic subpleural ILAs in higher-risk populations does not address the problem that the prevalence of ILAs is an order of magnitude higher than the prevalence of IPF. An important additional observation made by Araki and colleagues (3) was that although serial progression of ILAs on CT was associated with greater serial decline in FVC levels, this difference was small (an excess of only 25 ml per year, compared with subjects without ILAs). The small excess in FVC decline over decline expected with normal aging is difficult to interpret. It is possible that subclinical IPF has a long lead time and that the majority of subpleural reticular ILAs would, in reality, progress to IPF over many years, in patients surviving to an advanced age. However, it is equally possible that subpleural reticular ILAs consist of a smaller subgroup with subclinical IPF and a larger subgroup with other forms of limited fibrosis or “benign” age-related interstitial abnormalities. In either event, it seems likely that the accurate identification of subclinical IPF within key ILA subgroups will require the use of serum biomarker data predictive of the progression of fibrosis. Although no current data exist in this regard, “proof of concept” is provided by the recent observation that in a Framingham Heart Study cohort, higher serum galectin-3 concentrations were associated with an increased likelihood of ILAs and statistically significant reductions in FVC and carbon monoxide diffusing capacity (16).

In four recent cohorts, the presence of ILAs was associated with a significant increase in mortality (15), and in the study of Araki and colleagues, ILA progression was predictive of a fourfold increase in the risk of death (3). It is these findings that indicate that the goal of achieving earlier IPF diagnosis is plausible and should be pursued, but it is essential that the correct ILA subgroup be selected for further study. It is clear that workers in this field recognize the importance of subclassifying ILAs, but clearly the amalgamation of data from a number of large cohorts is necessary in order for the adequate powering of analyses in ILA subgroups of interest. Given the likely disparities between individual studies in the definition of ILA subgroups, a consensus position statement on ILA subcategorization, similar to the Fleischner Society statement on the HRCT features of interstitial lung disease (17), is surely required. Consensus statements on acute exacerbations in IPF (18) and on “interstitial pneumonitis with autoimmune features” (19) were constructed because the use of different definitional criteria in different studies necessitated a fresh start. Without this approach and the integration of biomarker data, studies of ILAs will continue to add to “the gaiety of nations” but are unlikely to lead to efficient algorithms for the early diagnosis of IPF.