To the Editor:

With interest, we have read the recent contribution by Xiao and colleagues in this Journal entitled “Parenchymal Airspace Profiling: Sensitive Quantification and Characterization of Lung Structure Evaluating Parenchymal Destruction” (1). We appreciate the general goal of the authors to “overcome shortcomings” of “traditional lung morphometry,” which according to the authors’ perception suffers from simplistic data interpretation, subjective data acquisition, and bias due to tissue shrinkage, just to name a few “shortcomings.”

However, many of the “shortcomings” that the authors introduce as a motivation for their study have comprehensively been addressed in the American Thoracic Society/European Respiratory Society (ATS/ERS) statement on quantitative assessment of lung structure, and clear recommendations have been made (2). The overall problem we see with the study by Xiao and colleagues is the fact that these recommendations by the ATS/ERS were not taken into consideration, so the described method is far from overcoming such “shortcomings” as mentioned above.

As an example, the study claims to overcome a bias due to “scaling, differences in degree of inflation, and tissue shrinkage during tissue collection and processing.” However, if formalin fixation and paraffin embedding are used, as in the presented study, unpredictable tissue deformation and, above all, shrinkage will occur and introduce a bias for every kind of image analysis (3).

Also, based on size analyses of distal airspaces on 2D sections, parenchymal airspace profiling was used to differentiate alveoli from alveolar ducts, and the data were further processed to determine, e.g., the “alveolar count” or “alveolar size.” However, this approach does not take into account the complexity of alveolar architecture in 3D and the fact that alveoli have entrances that connect them with alveolar ducts. When one takes a closer look at the images provided in the paper, it becomes obvious that structures that can clearly be identified as alveolar airspaces have been labeled as ductal airspaces by the parenchymal airspace profiling. Hence, simple determination of the 2D area of a distal airspace is not appropriate to separate alveoli and ducts. Moreover, for simple reasons based on stochastic geometry, it is impossible to estimate the number of objects in a 3D space (such as the number of alveoli) without bias by analyzing only single 2D thin sections. This problem can only be solved by 3D methods, such as disector (4). These considerations challenge the meaning of the data provided in the article, and it is obvious that the described methodology does not measure what it is supposed to measure.

In summary, the described method is prone to generate biased instead of accurate data, in particular because the ATS/ERS recommendations from 2010 are not taken into consideration. Unfortunately, these limitations of the study are not discussed in the paper. It is somewhat disappointing to see a paper that ignores even the most basic principles of the ATS/ERS recommendations published in this Journal, even more so because they were introduced to its readers by an editorial in 2010 (5). Quantitative assessment of lung structure is demanding and requires careful planning of experiments, regardless of whether stereology or automated image analyses are performed. The ATS/ERS guidelines from 2010 represent an extremely valuable basis for accurate assessment of lung structure, which is essential for good laboratory practice.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES