Thoughts on hypersensitivity pneumonitis

 Thoughts on pathology hypersensitivity pneumonitis from the newish ATS, ACCP consensus statement, and a review article by Andrew Churg 

https://pubmed.ncbi.nlm.nih.gov/34531525

https://pubmed.ncbi.nlm.nih.gov/35738345

https://pubmed.ncbi.nlm.nih.gov/32706311

These criteria are designed primarily to provide a uniform language for clinical trials.  Nonetheless, they provide a platform to air general problems with current approaches.

Both ATS and ACCP emphasize that HP should be thought of as non-fibrotic vs fibrotic rather than acute, subacute or chronic. This seems like a good thing but is basically relabeling how I (and presumably others) always thought of them with a more logical label. A precedent would be acute (cellular only) vs chronic/fibrotic interstitial nephritis. No attempt though is made to define how much fibrosis is needed to be labeled fibrotic e.g. is one reticulation on HRCT enough?  Lack of quantitation for all criteria is a problem throughout 

For both fibrotic and non-fibrotic HP, the ATS and ACCP consensus statements use an approach modeled after the UIP pathology scheme with relative levels of certainty. ATS uses the categories of definite, probable, and indeterminate (with alternative diagnoses implied but not listed) while ACCP uses typical, compatible, indeterminate, or alternative diagnosis. The indeterminate category could also be labeled non-specific since findings could be found in any inflammatory or fibrotic ILD. Churg (despite being a co-author of the ACCP guidelines) dismisses this approach as formalistic and not helpful in real life.

The basic approach for cellular HP is non-controversial: Lung should show lymphoplasmacytic inflammation +/- granulomas. However, when we get to the details, things get dicey. 

All three emphasize that the inflammation needs to be bronchocentric. For some reason, ATS/ACCP distinguish brochocentric inflammation from cellular bronchiolitis. The distinction of the two seems arbitrary. Are there any cases in which cases have one but not the other and it matters? Given that the lung is a three-dimensional structure, how do we know that any lymphoid aggregate is not related to an airway without serial sectioning, a technique which is not mentioned in any protocol?

ATS emphasizes that the HP granulomas need to be poorly formed.  While poorly formed granulomas are a hallmark of HP, daily practice HP has granulomas with a range in compactness from very ill formed to readily identified. In fact, the classic teaching is that they key difference of HP and sarcoidosis is the presence of pneumonitis away from granuloma, not the granuloma itself.  It is true that sarcoidal granulomas but not HP granulomas have concentric hyalinized fibrosis.

Churg emphasizes that Schaumann bodies count as granulomas while ACCP considers them “minor findings” different from granulomas.  Curiously, ACCP does not state as far as I can tell what the significance of “minor findings” are. Finally, Churg makes the unexpected claim that organizing pneumonitis is rare in HP relative to inflammation alone, contrary to decades of experience. ACCP requires areas of OP to be “small foci”. The definition and rationale for that is not discussed. 

For definitive/ typical diagnosis of fibrotic HP, ATS / ACCP require requires a granulomatous inflammatory component.  This is a step forwards towards clarity although why granulomas are needed rather than any inflammatory component in setting of bronchocentric fibrosis is unclear.  Note that in setting of requirement for poorly defined granuloma, a cluster of lymphocytes with one macrophage is entirely non-specific while adding one or two more macrophages become definitive HP. 

The criteria for the remaining category of “probable/compatible” fibrotic HP is problematic.  As Churg notes, “fibrotic HP is the problem child of ILD”. For ATS “probable” depends on recognizing a pattern of fibrosis which is 1) severe and patchy 2) without an inflammatory component and 3) with fibroblast foci and 4) which is different from UIP. The key is to identify the fibrosis as airway centered rather than subpleural and paraseptal.  However, as fibrosis becomes more extensive, the ability to determine bronchocentricity is markedly reduced.  That borderline is not addressed by ATS or ACCP. It is unclear why ATS included “chronic fibrosing interstitial pneumonia” as a required element for this category since the pattern or extent of fibrosis seems irrelevant to the diagnosis if clear bronchocentricity is seen elsewhere. ACCP seems to recognize this problem by indicating that areas of bronchocentric fibrosis should be in different areas than other types / patterns of fibrosis. This is not clearly stated, however.

Under airway centered fibrosis, ATS accepts as criteria either so-called bridging fibrosis which extends from airway to airway or peribronchiolar metaplasia (PBM). However, the ACCP consensus excludes airway bridging as non-specific and considers PBM non-specific unless present in at least 50% of airways. That number is based on a single center study by Churg of relatively few individuals.

Note that bronchocentricity, whether for inflammation or fibrosis, is usually not applicable to cryobiopsies since less than the full secondary lobule is sampled. 

The striking thing about all three of these papers is that there is no sense that there might be criteria required to generate criteria.  These are by and large simply declarative statements rather than data driven. 

One of two approaches could be considered here. One is to use some kind of external gold standard against which to compare criteria. In this case, one could try to find cases that are clinical high probability (obvious exposure history with appropriate time course which got biopsied for some other reason). In the absence of that, one could look to see if there is a distinct cluster of findings such that a particular histologic pattern stands out, a kind of unsupervised cluster analysis which is familiar from machine learning. In both cases, it should be a requirement that any criteria be reproducible in real life among pathologists trained to recognize the features. This latter requires large numbers of cases that can be reviewed by large numbers of pathologists. 

Finally, the real goal here is to separate probable/compatible fibrotic HP from UIP.  The authors provide no data that these criteria improve upon (presumably heterogeneous) existing practice to distinguish these entities more reproducibly.  Other data suggest that the criteria overlap with previous IPF criteria such that in any given series, a quarter of patients might fight both chronic fibrotic HP and IPF. Given that the prognosis of highly fibrotic HP is quite similar to IPF, this might not matter in practice. I understand further work is ongoing to reconcile / integrate the IPF and HP criteria into a single standard.