This study shows that TBB specimens can yield some of the findings of a UIP pattern in a portion of such patients. Two provocative findings emerge from this study. First, TBB is a useful mini-invasive tool that can identify some of the UIP criteria in about one-third of cases with a high positive predictive value and a good interobserver agreement. Second, TBB is less useful in diagnosing other fibrotic infiltrative diseases that can mimic IPF.
Our data confirm those previously published by Berbescu and co-workers  demonstrating that TBB specimens contain some (or all) of the histopathologic criteria of UIP in approximately one third of patients. Diagnostic biopsies contained more fragments and larger amount of alveolated lung parenchyma compared to UIP cases where TBB did not display UIP criteria, but even in cases with only one adequate fragment it was possible to recognize some features of UIP. One pathologist was able to detect all three features of UIP in a case with only one alveolated fragment. This shows that a very experienced pathologist can sometimes reach a confident diagnosis of UIP even with a minimal biopsy specimen. In the presence of two or more histopathologic criteria of UIP seen on TBB, the diagnosis of UIP is always confirmed by SLB. These observation supports the view that even the pathologic findings on small samples in the appropriate clinical and radiologic setting can be highly predictive of a UIP pattern. The aim of this study was limited to defining the accuracy with which histopathologic criteria of UIP can be identified in TBB. It should be recognized, however, that numerous lung diseases can manifest histological pattern of UIP and the use of multidisciplinary approach is mandatory for the final diagnosis. Therefore, further studies are needed to evaluate the value of adding TBB results to clinical and radiological findings in order to define the impact of TBB on diagnostic confidence in the multidisciplinary diagnosis of IPF and other ILDs.
The accuracy of TBB is mostly limited by its low negative predictive value and by its low sensitivity which relate to a high portion of inadequate and non diagnostic fragments. Number and size of fragments strongly influenced the accuracy of TBB in diagnosing UIP. We can hypothesize that the use of newer bronchoscopic biopsy methods, such as cryoprobes, may yield larger specimens that can highly improve the diagnostic yield . TBB in UIP cases was more informative than in non-UIP cases.
We acknowledge that the limitations of this study include the relatively small number of cases and that this study was not designed to evaluate the accuracy of TBB in diagnosing non-UIP ILDs. Therefore we are not able to draw any firm conclusion on the diagnostic utility of TBB in this other group of diseases. However, it is interesting to note that TBB was not useful to discriminate some of the ILDs that commonly enter the differential diagnosis of IPF, such as DIP, fibrotic NSIP and chronic HP [17, 18]. It has been described that in some patients non-UIP patterns, such as RB-ILD, DIP and NSIP, can coexist with UIP [19, 20] and diagnosis can be rather challenging in such cases. Our study report similar findings for TBB showing that fibrosing ILDs other than UIP present complex pathologic features that are rarely detectable by TBB and should be regarded as non diagnostic in the context of a multidisciplinary diagnosis of IPF. On the other hand UIP pattern can occur in diseases that can mimic IPF such as chronic HP, chronic sarcoidosis, asbestosis. In these cases TBB findings can be misleading. To reach the correct final diagnosis the interpretation of the UIP pattern requires a careful multidisciplinary evaluation.
For individual histological criteria of UIP evaluated on TBB we found a fair agreement between the two pathologists. Nicholson et al. [20, 21] reported levels of agreement between pathologists for individual histological criteria of UIP in SLB to be higher (mean K 0.75 for fibroblast foci and 0.76 for honeycombing) than ours that are based on TBB specimens.
We acknowledge several limitations in the retrospective, mono-centric design of this study and in the limited number of cases and controls examined. The blinded pathologic review method used represent another limit of this study. Only one slide per sample was reviewed for both SLB and TBB and special stains were not used. Ideally multiple levels should be evaluated but they were not available in all our cases. In real practice the use of deeper levels and special stains could be useful to reveal pathologic features not present in only one haematoxylin eosin stained slide. This limit the opportunity to test true diagnostic performance of TBB in a process more closely resembling standard practice and emphasize the bias that pathologists involved are experts. In current practice more than one slide per case should be evaluated and special stains are recommended in selected cases. The best study design would be a controlled, blinded, prospective study using multiple slides and special stains. However, due to the complete lack of controlled and blinded data on this topic, we postulated that this was a mandatory preliminary step. TBB were obtained in a referral centre for ILDs by experienced bronchoscopists and also this might have influenced the results. However the number and size of alveolated pieces does not differ from previous reports and is in line with feasibility data reported by others [12, 13].