Tissue Continues to Be the Issue: Role of Histopathology in the Context of Recent Updates in the Radiologic Classification of Interstitial Lung Diseases

As our understanding of the natural histories of various interstitial lung diseases (ILDs) has evolved, so has the recognition that certain radiologic patterns are correlated with specific histopathologic patterns in patients with ILDs.¹  High-resolution computed tomography (HRCT) imaging is having an increasingly more important role in clinical decision-making in patients with ILDs, but, as with any other diagnostic tool, it is important to understand its performance characteristics and to recognize its limitations.

The 2011 official American Thoracic Society and European Respiratory Society (ATS/ERS) statement on the diagnosis and treatment of idiopathic pulmonary fibrosis (IPF), among other recommendations, recognized that a histopathologic confirmation was not needed in patients with idiopathic ILD who had a usual interstitial pneumonia (UIP) pattern on HRCT of the chest.²  This recommendation was based on high-quality studies correlating HRCT findings with histopathology. Those studies showed that a “confident” HRCT read, demonstrating a “definite” UIP pattern, were 95% to 100% specific in predicting a UIP pattern on histopathology.^3–7  The 2011 IPF statement²  was very specific in stating that the need for histopathologic confirmation can be obviated only in idiopathic ILD demonstrating a UIP pattern and recommended a biopsy in patients with an ILD demonstrating other patterns.

Anecdotally, it has been our experience that there has been a tendency in the past few years in some academic and nonacademic centers treating patients with ILDs toward extrapolating the findings in a “confident read of a definite UIP pattern on HRCT chest” to other HRCT patterns, perhaps, without supporting evidence. This has translated clinically in a tendency to avoid histopathologic confirmation in many patients who have patterns other than UIP on HRCT, despite those patients being otherwise good surgical candidates. Varying proportions of those patients with idiopathic ILDs who demonstrate patterns other than a UIP pattern are labeled with an ILD diagnosis based on radiology alone. An unknown number of those patients would likely have had an alternative diagnosis had histopathology been pursued as part of the workup.

The recent timely and very comprehensive Fleischner Society white paper¹  on the diagnostic criteria of IPF highlights recent advances in our understanding of HRCT imaging in ILDs, and perhaps, importantly, the pitfalls and current gaps in our understanding of the utility of HRCT imaging in predicting histopathology in patients with ILDs. In this article, we discuss the evidence and recommendations outlined in the white paper as they pertain to the radiologic diagnosis of ILD.

The Fleischner society paper notes that, in line with the previous guidelines,²  the histopathologic confirmation is not needed in the context of a “typical UIP pattern,” which demonstrates basal-predominant (occasionally diffuse) and subpleural-predominant, often heterogeneous, reticular opacities with peripheral traction bronchiectasis or bronchiolectasis, with honeycombing, and an absence of features to suggest an alternative diagnosis¹  (Table). Corresponding histopathologic findings in UIP include temporally and spatially heterogenous fibrosis, which is predominantly subpleural, with fibroblastic foci and honeycombing.¹  UIP histopathologic pattern can be seen in various clinical contexts such as hypersensitivity pneumonitis, and connective tissue diseases. If a UIP pattern is noted on HRCT and/or histopathology, a diagnosis of IPF is made when no “secondary” causes for the ILD can be identified after a reasonably diligent history, physical examination, and laboratory investigation.

The previously described possible UIP²  pattern has been subdivided into the probable UIP pattern and an indeterminate for UIP pattern¹  (Table). The probable UIP pattern is similar to that of a UIP pattern, except honeycombing is absent.¹  Interstitial abnormalities in a probable UIP pattern are primarily reticular opacities, with peripheral traction bronchiectasis or bronchiolectasis with a predominantly basal and subpleural, and often heterogeneous, distribution.¹  Honeycombing and features to suggest an alternative diagnosis are notably absent.¹  The white paper states that “A confident diagnosis of IFP (idiopathic pulmonary fibrosis) can be made in the correct clinical context [emphasis added] when CT imaging shows a pattern of typical or probable UIP (usual interstitial pneumonia),”^1(p138) and that “The role of CT is expanded to permit diagnosis of IPF without surgical lung biopsy in select cases [emphasis added] when CT shows a probable UIP pattern.”^1(p138) One must be careful not to misinterpret this recommendation to state that histopathologic confirmation is not needed in any patient with a probable UIP pattern on HRCT. Although a probable UIP pattern is highly predictive of a UIP pattern on histopathology, exceptions exist. Raghu et al⁶  found that a probable UIP pattern on HRCT was highly predictive (94%) of finding a UIP pattern on histopathology. The patient population for that study consisted of patients who were screened for the ARTEMIS-IPF trial (a randomized, placebo-controlled study to evaluate safety and effectiveness of ambrisentan in IPF) and was, therefore, enriched for patients with IPF.⁶  It is likely that the positive predictive value for a probable UIP pattern in this study would be less if patients in this study consisted of consecutive patients with varied ILDs demonstrating a probable UIP pattern instead of the IPF-enriched patient population for the ARTEMIS-IPF trial. This very high positive predictive value was not reproduced in other studies. Chung et al⁸  in a multicenter study of 201 patients with pulmonary fibrosis found that 82% of patients with a HRCT read of “probable UIP” had UIP on histopathology. That high, but less than perfect, correlation has important clinical implications. In patients who already have a high pretest probability of IPF (eg, an older male smoker without any risk factors for ILDs) when the HRCT is read as a probable UIP pattern with a high degree of confidence, histopathologic confirmation is not needed. The decision to biopsy should be made on a case-by-case basis in a patient whose pretest probability of IPF is lower, especially if the interpretation of the HRCT pattern is not made with a high confidence. Histopathologic confirmation should be considered in those patients if they do not have risk factors that put them at a high risk for complications and mortality after a surgical lung biopsy. Therefore, in the context of a probable UIP pattern on HRCT, it may be reasonable to make a radiologic diagnosis of IPF in patients with a high pretest probability of IPF and in older patients (who have a higher pretest probability of IPF) with comorbidities, but consideration for biopsy should be given to appropriately selected, younger patients with little functional and physiologic restriction and fewer or no comorbidities; patients who are expected to have low morbidity and no mortality after a surgical lung biopsy.

The remainder of the previously described possible UIP is classified as an indeterminate UIP pattern, which is described as demonstrating fibrosis with some inconspicuous features suggestive of non-UIP pattern and in a variable or diffuse distribution.¹  Up to approximately one-half of patients with an indeterminate pattern were found to have findings other than UIP on histopathology.⁸  Biopsy should be pursued in patients with an indeterminate UIP pattern unless there are contraindications for the surgical procedure in light of the prognostic and therapeutic implications.¹ 

Another important change in the radiologic nomenclature of ILDs was the renaming of the previously described not UIP pattern²  to CT features most consistent with non-IPF diagnosis.¹  That nomenclature change was based on the recognition that, although HRCT findings suggestive of alternative diagnoses make the likelihood of finding a UIP pattern on histopathology significantly less likely, UIP can still be found on histopathology in many patients with those findings on HRCT. Chung et al,⁸  in a multicenter study of 201 patients with pulmonary fibrosis, found that 50% of patients with CT read as “inconsistent with UIP” had probable or definite UIP on histopathology. Yagihashi et al⁹  when looking at HRCT-histopathology correlation in patients with a diagnosis of IPF who were enrolled in 3 IPFnet (IPF Clinical Research Network) studies found that 31% of all patients who had histologically definite or probable UIP on histopathology had HRCT features that were “inconsistent” with a diagnosis of UIP. Other studies show that up to 60% of patients demonstrating typical UIP histopathology had CT imaging findings that were not typical for UIP.^1,10,11  Areas of decreased attenuation or mosaic attenuation, both suggesting an alternative diagnosis such as hypersensitivity pneumonitis, can be relatively common (43% in one study¹²  and 27% in another⁹ ) in patients with biopsy-proven UIP.¹  In the correct clinical setting, a diagnosis of IPF should not be excluded if the CT pattern is more suggestive of another ILD, such as nonspecific interstitial pneumonia, chronic hypersensitivity pneumonitis, or sarcoidosis.¹  The CT pattern for UIP most frequently overlaps with that of fibrotic, nonspecific interstitial pneumonia.¹  Although a confident UIP read on a HRCT is very specific and highly predictive of a UIP pattern on histopathology,^3–7 , it is not very sensitive.¹  Only about one-half of patients in an IPF study of 1061 patients were found to have a typical UIP pattern on CT.¹³  Presence of emphysema also complicates making the diagnosis of IPF on radiology alone. Moreover, HRCT accurately predicted UIP on histopathology in only 50% of 40 patients with emphysema and ILD.¹⁴  Histopathologic confirmation is, therefore, recommended when the CT demonstrates HRCT findings suggestive of a non-IPF diagnosis, to confirm the non-IPF diagnosis or to exclude IPF in patients with idiopathic ILDs. Importantly, although the final diagnosis should be based on a consensus opinion after a multidisciplinary discussion, if a typical UIP pattern is noted on histopathology, the pathologist should not be swayed by the HRCT findings (suggestive of non-IPF diagnosis) in calling out a UIP pattern on histopathology as such.

Other factors to consider when forgoing the need for histopathology based on HRCT pattern alone is that radiologists can, and often do, differ on their interpretation and classification of the radiologic findings on HRCT in patients with ILDs.¹  A large study showed that interobserver agreement on HRCT reads of various patients with ILD was only moderate for both general radiologists and thoracic radiologists with more than 20 years of experience for both UIP diagnostic categories (mean [SD] = 0.45 [0.13] and 0.48 [0.14], respectively), and for binary diagnosis score (Typical UIP, 0.41; or Possible UIP/Inconsistent with UIP, 0.40).¹⁵ 

When treating patients with ILDs, determining an accurate diagnosis has important prognostic implications because survival in patients with IPF is significantly worse than it is for patients with other ILDs.^16,17  Importantly, getting the diagnosis correct is essential because of the current dichotomy of therapeutic choices available for treating IPF and other ILDs. Corticosteroids and immunosuppressive medications, which are used to treat many non-IPF ILDs, are associated with worse outcomes in patients with IPF when compared with no treatment at all,¹⁸  and medications that slow the rate of progression of IPF^19,20  have not yet been shown to be effective in, and are not approved for, the treatment of non-IPF ILDs. Inappropriate therapy may, therefore, be worse than no treatment at all.

Additionally, it is important to recognize that although the morbidity and complication rates can appear unreasonably high in unselected patient populations with significant comorbid conditions undergoing a surgical lung biopsy, morbidity and mortality in patients undergoing surgical lung biopsy is primarily driven by the underlying comorbidities, physiologic limitations, and functional limitations, rather than the procedure itself.^21,22  Patients with fewer or no comorbidities and relatively few physiologic and functional limitations experience relatively fewer complications and no mortality after a surgical lung biopsy.^23–32  A surgical lung biopsy should, therefore, be considered in patients with ILDs when a diagnosis cannot be confidently made based on HRCT findings in the appropriate clinical context.

In summary, HRCT is highly predictive of a UIP pattern on histopathology when the HRCT shows a typical UIP pattern on a confident read by the radiologist. The previously described possible UIP pattern can be split into a probable UIP pattern and an indeterminate UIP pattern. A probable UIP pattern is also very predictive of a UIP pattern on histopathology, and histopathologic confirmation is not needed for most patients demonstrating that pattern in the appropriate clinical setting. A UIP pattern may be seen in a substantial proportion of patients with an indeterminate UIP pattern on HRCT and in a variable proportion of patients who demonstrate HRCT patterns suggestive of alternative diseases; histopathologic confirmation should be considered in patients demonstrating these patterns whenever feasible.