• Breast pathology
• cytopathology

The paper by Tran et al on the histological correlates of atypia in fine needle aspirates (FNAs) of the breast was useful in suggesting which cytological features are likely to indicate malignancy.1 However, their conclusion that 67% of atypical FNAs are from benign lesions and 33% malignant tumours may be misleading.

In their review of 5340 cases, 363 (7%) of breast FNAs were diagnosed as atypical, but the authors include in their analysis only those 98 atypical aspirates that had a histological follow-up. The reason why the remaining 265 lesions were not biopsied was probably because, in most cases, they were clinically and radiologically benign; omitting these cases from analysis therefore constitutes a significant workup (verification) bias.2 The true proportion of atypical FNAs that were from benign lesions is therefore likely to be significantly higher than their figure of 67% suggests, and a review of these patients' medical records would allow a fairly accurate estimate of the true figure. Including all 363 atypical FNAs in their analysis would also, by increasing the size of the cohort, address one of the criticisms of the study made by the authors themselves, thereby increasing the accuracy of the results by narrowing their CIs.

Furthermore, given that the assessment of cytological atypia is subjective, it is also rather surprising that none of the 98 aspirates originally reported as atypical was changed to either benign or malignant on review. This suggests that a further bias, namely review bias, may be present, as these FNAs were not subject to blinded review (by being mixed in with an unknown number of benign and malignant aspirates, for example). Knowing that all the FNAs reviewed had already been considered atypical might have unconsciously biased the reviewers in their interpretation of the cytological characteristics.