Role of Transbronchial Cryobiopsy in Interstitial Lung Diseases: An Ongoing Tale.

To the Editor:

We read with great interest the study by Romagnoli and colleagues addressing the histological diagnosis of interstitial lung diseases (ILD) (1). Although transbronchial lung cryobiopsy (TBLC) has been suggested as an initial procedure to obtain tissue from patients with ILD, there is much debate regarding its diagnostic yield, and it is not supported by current guidelines (2). This is the first study to prospectively compare the results of TBLC with the “gold standard” surgical lung biopsy (SLB) in the same patients, and as such, it has been highly anticipated.

However, we have some concerns regarding the interpretation of the results, which led the researchers to conclude that “there is no role for TBLC in the vast majority of patients where histopathology is required for definitive diagnosis of diffuse ILD” (1). As implied by the authors, a distinction should be made between the pathological diagnosis per se and the “final diagnosis” as decided by a multidisciplinary assessment (MDA).

In fact, previous studies that compared the results of SLB with final diagnoses made in explanted lungs showed a poor correlation with a pathological diagnosis of usual interstitial pneumonia (3, 4).

Moreover, previously reported interobserver agreement levels for a histopathological diagnosis of ILD by SLB were not high, even in centers with extensive experience (5). The degree of interobserver agreement (i.e., concordance) between the blinded pathological review and the routine pathological reports from SLB are not reported in the current manuscript, although they are mentioned in the Methods section (1). Therefore, we used the data from Tables 2 and E1 in Reference 1 to calculate it. We found an agreement level of 57.1%, with a κ-concordance coefficient of only 0.44 (0.215–0.66). Arguably, this relatively low level of agreement between two pathologists may make one question how “golden” the SLB gold standard is.

Therefore, given that the interobserver agreement level is only 57.1%, it is no wonder that the agreement between SLB and TBLC is also poor, as the article’s title suggests. Furthermore, the authors’ statement that “patients who are able to undergo SLB should be recommended to do so” is not entirely supported by their data and does not take into account the morbidity and mortality risks of the procedure.

The level of agreement between SLB and MDA was 62% (95% confidence interval, 38–82%), which is somewhat low for a gold standard. Although this is slightly higher than the level of agreement between TBLC and MDA (48%), it is not significantly different (95% confidence interval, 26–70%). Besides overlapping confidence intervals, it is possible that a bias shifted the scale toward SLB, as SLB and TBLC were discussed simultaneously in one MDA meeting. This might be problematic, because clinicians and pathologists are more familiar with SLB than with TBLC. Furthermore, the SLB samples were on average 5–10 times larger than the TBLC samples, as would be expected. Taken together, these observations suggest that the SLB diagnosis probably influenced the MDA significantly more than the TBLC diagnosis. Therefore, the better concordance between the blinded pathological diagnosis of SLB and the MDA seems inherent to the process itself.

A better assessment would be to conduct two separate MDA discussions, one using TBLC and the other using SLB samples, and calculate the concordance between them or between each blinded assessment and its corresponding MDA. In addition, it would have been prudent to subject the samples to blinded assessments by at least two pathologists rather than one.

Thus, we believe that rejecting the role of TBLC in the assessment of ILD is premature. We agree that further prospective studies to assess the role of TBLC in the diagnostic evaluation of ILD are warranted. The ongoing prospective COLDICE (Cryobiopsy versus Open Lung Biopsy in the Diagnosis of Interstitial Lung Disease) study (6) is designed to address many of the aforementioned issues, and is expected to provide more conclusive evidence for the role of TBLC in ILD diagnosis.