In some patients, chronic ILD preceded the development of MPA by 1–10 years . Using the previous HRCT classification [18, 19], the most frequent pattern of ILD with MPA is that of UIP (50–57%), followed by nonspecific interstitial pneumonia (7–31%), and desquamative interstitial pneumonia (14%) . UIP is also the most common abnormal pattern in cases of MPO-ANCA-positive ILD but without generalized involvement . The present study also showed these above characteristics and findings; however, we have obtained some new knowledge.
First, HRCT pattern classification (i.e., UIP, probable UIP, indeterminate for UIP, and alternative diagnosis) based on the recent IPF guideline was useful to predict prognosis in the MPO-ANCA-positive ILD patients. In the present study, the patients with indeterminate for UIP pattern accounted for a quarter of the MPO-ANCA-positive ILD patients and had a better prognosis than those with UIP pattern, although there was not a significant difference between indeterminate for UIP and probable UIP pattern. The presence of a radiological honeycomb pattern was not a significant predictor of poor prognosis in the present study. We speculate that in previous studies, if patients had a radiological honeycomb, they most probably had UIP pattern rather than other patterns . Therefore, the position of indeterminate for UIP regardless of having a honeycomb is thought to be meaningful. In another view of this point for comparison with IPF, the present study showed UIP pattern (using the recent classification) to have a poor prognosis as patients with this pattern had a short median survival time (4.1 years) similar to that for IPF. However, previous studies showed the survival time in MPO-ANCA-positive patients with UIP pattern (using an older previous classification) tended to be longer than that for IPF [2, 3]. Importantly, 4–40% of the previously studied cases do not fit any specific HRCT pattern of the previous classification of IIPs . Taken together, the recent IPF classification might be useful for prognostic analysis and aid in the assessment of appropriate strategies of diagnosis in MPO-ANCA-positive ILD studies.
Second, the radiological finding of increased attenuation around honeycomb or traction bronchiectasis was frequently found (39.3%) as in previous studies because MPO-ANCA-positive-ILD shows more prominent inflammatory cell infiltration, lymphoid follicles with germinal centers, and cellular bronchiolitis [2, 6, 16]. More than half of the patients with indeterminate for UIP pattern had this radiological finding in the present study. Therefore, we also thought this specific radiological finding is important in the differential diagnosis of IPF.
Third, surprisingly, anterior upper lobe honeycomb-like lesion was also very frequently (29.5%) observed in the MPO-ANCA-positive ILD patients. We recently reported that the tendency for this radiological finding is higher in patients with RA-ILD (22%) rather than SSc-ILD (8%) or polymyositis/dermatomyositis ILD (8%) . This radiological finding is thought to occur due to focally destroyed lung as a result of highly inflamed airways . MPO-ANCA-positive ILD and RA-ILD may resemble diseases showing a high frequency of small airway disorder [2, 6, 15]. This radiological finding might be specific not only in patients with RA-ILD but also those with MPO-ANCA-positive ILD. Other studies have shown difficulty in distinguishing MPO-ANCA-positive ILD from IPF by HRCT [2, 21]. Some patients with ILD during follow-up developed MPO-ANCA-positive conversion or MPA [3, 4]. Therefore, the presence of anterior upper lobe honeycomb-like lesion may be an indicative predictor of MPO-ANCA-positive ILD or RA-ILD in these patients and thus can be useful information for clinicians. However, this radiological finding was a poor prognostic factor as were older age and a history of DAH. Fibrotic changes such as reticular shadows, traction bronchiectasis, and honeycomb pattern are found predominantly in the lower and outer regions of the lung [5, 22]. In the present study, 61% of the patients with anterior upper lobe honeycomb-like lesion also had honeycomb in the lower lobe as a result of the progression of lung fibrosis. Considering these results, the lesion in the anterior upper lobe may develop when MPO-ANCA-positive ILD progresses to some extent, but it is not present in the initial phase of the disease. Because we could not draw a firm conclusion, further studies of the radiological course of the disease over a long period may be needed.
Fourth, non-MPA-ILD patients had higher frequency of indeterminate for UIP pattern, whereas the MPA-ILD patients had higher frequency of UIP pattern and lower frequency of indeterminate for UIP pattern. MPO-ANCA positivity was associated with subsequent MPA development in some patients [2,3,4]. Because repeated episodes of alveolar hemorrhage due to pulmonary capillaritis could be the pathogenesis of progressive pulmonary fibrosis, and MPO-ANCA may play a direct role in the pathogenesis of progressive pulmonary fibrosis [23,24,25], to sum up, we might see non-MPA-ILD (indeterminate for UIP pattern) patients experiencing a sequence of events (i.e., pre-stage) that lead to MPA-ILD (progressive stage: UIP).
Fifth, regardless of the presence of MPA, which did not have a significant influence on prognosis, the cause of death is different between the two entities. AE of ILD occupy an important place in the outcome of ILD patients without MPA  as shown in the present study. All forms of ILD are at risk of AE and have a similar outcome to AE-IPF . Because data from studies in patients with progressive fibrotic ILD suggest that anti-fibrotic therapy could have a role in preventing AE-ILD , this therapy may also improve prognosis in MPO-ANCA-positive ILD, particularly in non-MPA-ILD. However, the presence of a history of DAH is associated with poor prognosis and mortality in MPA-ILD. Several studies suggested that anti-inflammatory therapy can reduce the risk of MPA-related DAH development . In fact, the present study showed that none of the patients with a history of DAH received anti-inflammatory therapy at the time of DAH. As Kagiyama et al. mentioned, anti-inflammatory therapy in patients with MPO-ANCA-positive ILD might offer some benefit in reducing the development of DAH because DAH is normally progressive and fatal . Moreover, Sun et al. reported poorer prognosis in MPA-ILD and non-MPA-ILD patients with elevated inflammatory markers (CRP or erythrocyte sedimentation rate) compared with non-MPA-ILD patients with normal inflammatory markers . Our study also showed higher CRP to be a poor prognostic factor, although there was no difference in survival between the MPA-ILD and non-MPA-ILD patients. However, the frequency of the patients with non-MPA-ILDs receiving anti-inflammatory agents differed greatly; more than 90% as reported by Sun et al.  versus 35.9% in our study. Therefore, anti-inflammatory agents may improve the prognosis of non-MPA-ILD patients, particularly in those having elevated inflammatory markers. Taken together, further accumulation of studies is warranted to clarify the clinical effectiveness of anti-inflammatory and anti-fibrotic therapy for MPO-ANCA-positive patients with ILD.
Our study has several limitations. First, it is a single-institution retrospective study, which introduces referral bias and limits the ability to generalize our findings. Second, it is possible that patients with incidental MPA may have been excluded from those with non-MPA-ILD because the differential diagnosis of MPA is difficult. For example, DAH may be misdiagnosed as an AE of ILD. Third, we could not quantitatively assess the HRCT patterns, and this will be an issue for further investigation. Fourth, because the protocol for ILD diagnosis is not standardized in our hospital, clinical practice varies between different clinicians, such as pulmonologists and rheumatologists. Therefore, these data may mask potential differences.