In this analysis of an established pulmonary ILD cohort, we identified many patients who fulfilled IPAF criteria. Notably, prior to review by rheumatology, several of these patients had been considered to have CTD-ILD. IPAF patients' demographic and clinical characteristics were more similar to patients with CTD-ILD than patients with IPF. As in CTD-ILD and IPF, male sex was a negative prognostic factor in patients with IPAF. Ultimately, despite the similarities between patients with CTD-ILD and those with IPAF, only a small fraction of patients classified as IPAF developed CTD-ILD during follow-up, and those that did often developed the ASyS. These observations suggest that IPAF classification may capture a divergent entity characterized by lung-dominant autoimmunity.
Despite the heterogeneity of the IPAF classification criteria, we identified substantial overlap between the characteristics of our IPAF cohort and those described previously, with some notable exceptions.[6,7,10] Unlike previous descriptions of patients with IPAF, we did not observe any patients with mechanic's hands or Gottron's papules.[6,8–10] Although this difference could be due to limited detection of some clinical features given that nearly 30% of our IPAF cohort had not seen a rheumatologist, it is likely that patients with these findings fulfilled criteria for either ASyS or dermatomyositis within 6 months of their initial visit and were thus classified as CTD-ILD instead of IPAF. Indeed, several patients classified as CTD-ILD exhibited Gottron's papules and mechanic's hands. Importantly, prior retrospective IPAF studies have generally not incorporated features beyond the initial visit when classifying patients as IPAF or CTD-ILD.
The frequency of UIP in our IPAF cohort was lower than previously reported.[8,10] This difference may reflect the fact that, in contrast to previous cohorts in which a substantial proportion of patients classified as IPAF were previously diagnosed with IPF, our IPAF cohort predominantly comprised patients previously diagnosed with unclassifiable or CTD-ILD. Few prior IPAF studies have clarified the contribution of rheumatologists to patients' initial diagnostic evaluation, and it has been posited that pulmonary-derived IPAF cohorts may appear more IPF-like, and rheumatology-derived IPAF cohorts may appear more CTD-ILD–like, perhaps due to an overestimation of CTD-ILD in the former.[6,10,27] Indeed, in our study, almost half (n = 26 of 60) of the patients classified as IPAF after rheumatology review had initially been considered CTD-ILD, and more than half of these patients were initially believed to have true CTD-ILD rather than "undifferentiated connective tissue disease." These differences highlight the importance of a consistent multidisciplinary diagnostic approach that includes rheumatologists.
As expected, patients with CTD-ILD had longer transplant-free survival relative to patients with IPF.[10,28,29] Unlike several recent studies, however, we did not detect statistically significant differences in transplant-free survival between patients with IPAF and patients with either CTD-ILD or IPF.[7,10] Although our study may not have been adequately powered to detect such differences, it is important consider to several additional possibilities. First, in considering the heterogeneity of the IPAF classification criteria, it is entirely possible that these criteria may capture several prognostically distinct subgroups of patients. In addition, although the optimal treatment of IPAF is unknown, patients with IPAF were less likely to be receiving immunosuppression at baseline compared with patients with CTD-ILD, possibly impacting our findings. Beyond factors related to IPAF, it is also important to consider the relative composition of CTD-ILD in different studies, given that this term encapsulates rheumatic diseases with distinct prognoses. Finally, the median survival of our IPF cohort (5.6 years) was considerably longer than that of the IPF cohorts included in previous studies demonstrating survival differences between the IPAF and IPF populations, potentially precluding our ability to detect such differences between our IPAF and IPF groups.[9,10,12,13]
We identified male sex as a novel prognostic indicator among patients with IPAF. Several studies of IPF, CTD-ILD, and animal models of pulmonary fibrosis have demonstrated a reproducible association between male sex and increased mortality, which may be partly related to the role of sex hormones in pulmonary inflammation and fibrosis.[26,31–33] The Gender, Age, and Physiology model is a clinical prediction model based on sex, age, and pulmonary function that has been shown to predict mortality in different types of ILD, including IPF and CTD-ILD, and our findings suggest that this model may be useful in IPAF as well. Although associations with age and baseline FVC were not statistically significant in our study, previous studies have identified significant associations between age and pulmonary function with transplant-free survival in patients with IPAF.[7,10]
Interestingly, only a small fraction of our IPAF cohort was diagnosed with CTD-ILD during follow-up, suggesting that IPAF may not necessarily herald an inevitable trajectory towards classifiable CTD-ILD. Similar findings reported by others underscore the observation that most patients with IPAF do not seem to progress to classifiable CTD-ILD within 3 years of their ILD diagnosis.[11,13] Furthermore, none of the CTD-ILD patients who had initially presented with ILD prior to their CTD diagnosis were ever classifiable as IPAF, since the extrapulmonary manifestations that led to their CTD diagnosis developed within 6 months of their initial visit.
Half of the IPAF patients in our cohort who developed CTD-ILD during follow-up were diagnosed with ASyS. Notably, we also found that CTD-ILD patients with antisynthetase and anti-MDA5 antibodies were significantly more likely to present with ILD prior to, or concurrently with, their CTD diagnosis. These observations, and the frequency of cytoplasmic ANA staining observed in our IPAF cohort, emphasize the importance of comprehensive myositis autoantibody testing in patients classifiable as IPAF who have not already undergone such testing, regardless of the presence of myopathy. Indeed, patients with IPAF in our study were more likely to have undergone lung biopsy, and of those IPAF patients who had undergone a biopsy, 13/20 (65.0%) had not had comprehensive myositis panel testing. Collectively, our observations also reinforce the ATS's consensus for myositis autoantibody panel testing in patients with idiopathic ILD.[25,34] This is especially important considering that even recent prospective studies of IPAF have not provided a truly comprehensive description of the presence of non-Jo1 antisynthetase antibodies.[12,13]
More broadly, our observations suggest that the inclusion of antisynthetase and anti-MDA5 antibodies within the serologic domain of the IPAF classification criteria may not be appropriate, as different antisynthetase antibodies are variably associated with extrapulmonary clinical features, including myositis.[35–37] A recently published single-center retrospective study of myositis panel testing revealed that 10% of patients with isolated ILD who had undergone myositis-specific antibody (MSA) testing had a positive antibody. Moreover, there are data suggesting that patients with IPAF and MSAs may represent a distinct subgroup of IPAF with better outcomes. As a result, the transplant-free survival of a given IPAF cohort relative to CTD-ILD or IPF may be influenced by the proportion of IPAF patients in that cohort with MSAs. Furthermore, a retrospective study of patients with antisynthetase antibodies who were initially classifiable as IPAF suggests that the frequency and pace of evolution to classifiable CTD-ILD of this subpopulation may be different from the remainder of patients classified as IPAF, as 42% of these patients developed classifiable CTD-ILD, often IIM, within a median time of 1 year from their IPAF classification. Thus, clarifying both the role of MSAs in future revisions of the IPAF criteria and the definition of the ASyS could help ensure that IPAF captures a truly undifferentiated population of autoimmune ILD.
This study has several limitations. It was conducted at a single tertiary care center. Although the ILD registry included longitudinal data collection and an MDD of each patient, not all patients saw a rheumatologist, and rheumatology was not included in the MDD. We prespecified criteria to confirm the presence of clinical features such as inflammatory arthritis, but this approach is imperfect. Although serologic testing was often comprehensive, it was not standardized, and many patients were missing the full complement of "IPAF criteria" serologies, especially non-Jo1 antisynthetase antibody panels. Collectively, these factors likely affected the overall frequency and domain composition (especially the clinical and serologic domains) of patients classified as IPAF and the overall frequency of patients classified as CTD-ILD. Importantly, such limitations are not uncommon among prior IPAF studies or ILD centers, as there is not a clear recommendation regarding which serologic tests should be performed in patients with newly identified ILD. In addition, a recent international study indicated that only 37% of ILD centers reported having rheumatologists regularly participate in the MDD. Beyond this, although longitudinal data collection permitted us to measure transplant-free survival, we lacked systematically recorded follow-up measurements of pulmonary function, exercise tolerance, and imaging. Moreover, as with other studies of IPAF to date, we were unable to assess the longitudinal effects of immunosuppression on transplant-free survival or the progression of IPAF to CTD-ILD. Patients with severe disease who died before registry enrollment would not be included in our analysis, and we were unable to comprehensively identify the cause of death for patients who died.
This study also has several strengths. The inclusion of rigorous rheumatology review enabled specific definitions for each CTD including those with potential overlap with IPAF classification and, together with the prior MDD, helped to improve the accuracy of ILD diagnosis. Our specific approach of allowing a 6-month window prior to classification has not been used in other studies and likely helped distinguish IPAF from CTD-ILD. We also list the specific serologies that were not routinely obtained in patients classified as IPAF, highlighting the disproportionate variability in testing for non-Jo1 antisynthetase antibodies and emphasizing the need for a more standardized approach. By identifying patients initially classified as IPAF who progressed to CTD-ILD, we also highlight considerations related to the inclusion of antisynthetase antibodies and other MSAs (e.g., anti-MDA5 antibodies) within the serologic classification domain of IPAF.
In conclusion, we found that patients classified as IPAF had clinical characteristics that were similar to patients with CTD-ILD. As with other forms of ILD, male sex was independently associated with a worse prognosis in IPAF. Importantly, few patients classified as IPAF developed CTD-ILD during follow-up. This suggests that IPAF classification does not simply identify patients with early CTD-ILD, but instead may identify a divergent lung-dominant disease entity. It is, however, of paramount importance that practitioners exclude the possibility of myositis- and ASyS-associated ILD via comprehensive serologic testing given their variable, and occasionally amyopathic, lung-dominant clinical presentations. Ultimately, larger prospective studies using a standardized evaluation and rheumatology-inclusive multidisciplinary approach to ILD classification will enable us to confirm prognostic features, refine the IPAF classification criteria, and, most importantly, identify the clinical utility and therapeutic implications of IPAF classification.
The data underlying this article are available in the article and in its supplementary material. Additional data not included in either document may be shared upon reasonable request to the corresponding author.
J Clin Rheumatol. 2022;28(5):257-264. © 2022 Lippincott Williams & Wilkins