Observer variability in histopathological reporting of non-small cell lung carcinoma on bronchial biopsy specimens

Abstract

Aims - To evaluate the ability of histopathologists to sub-classify non-small cell lung carcinomas on bronchial biopsy material using the current World Health Organisation (WHO) classification. Methods - Twelve histopathologists each reviewed 100 randomly selected bronchial biopsy specimens which had originally been reported as showing non-small cell lung carcinoma. For each case, two sections were circulated, one stained by haematoxylin and eosin and the other by a standard method for mucin (alcian blue/periodic acid Schiff). The participants were allowed to indicate their degree of confidence in their classification of each case. A standard proforma was completed and the results were analysed using κ statistics. Results - Where the participants were confident in their classification, they were actually quite good at sub-classifying the non-small cell carcinoma sections (κ = 0 71, standard error = 0.058). Overall, however, the results were only fair (κ = 0.39, standard error = 0.034). Conclusions - The majority of non-small cell lung carcinomas can be correctly categorised on adequate bronchial biopsy material. Where a confident diagnosis was made, both squamous carcinoma (κ = 0.73) and adenocarcinoma (κ = 0.83) were well recognised.