Bile duct brushing cytology: Cytohistologic/fine-needle aspiration correlation and diagnostic pitfalls

Cytomorphologic characteristics of next-generation sequencing-positive bile duct brushing specimens

INTRODUCTION

Cytology of bile duct brushings (BDBs) is a specific, but insensitive, test for malignancy. Next-generation sequencing (NGS) of BDBs has recently been shown to improve sensitivity. We analyzed the cytologic features of NGS-positive (NGS⁺) and NGS-negative (NGS^-) BDBs and correlated the morphology with the presence of mutations.

MATERIALS AND METHODS

A total of 96 BDBs were analyzed for 29 cytologic features by 2 pathologists who were unaware of the original diagnosis and NGS results. Clinicopathologic follow-up was used to determine the patient outcomes (ie, benign, low-grade neoplasm, malignant [carcinoma/high-grade dysplasia]).

RESULTS

We analyzed 74 NGS⁺ BDBs from 66 patients and 22 NGS^- BDBs from 22 patients. During follow-up, 58 of 66 NGS⁺ patients (88%) had malignancy compared with 0% of NGS^- patients (P < 0.001). Fewer than 50% of the malignant cases had been interpreted as malignant on cytology; however, 100% had demonstrated mutations using NGS. Within the NGS⁺ cases, 53% showed late mutations (TP53, SMAD4, and CDKN2A) supportive of a high-risk stricture. Significant morphologic differences were seen in the background appearance, presence of single cells, architectural disarray, nucleomegaly, anisonucleosis, irregular nuclear borders, increased nuclear/cytoplasmic ratio, nuclear hyperchromasia, nucleoli, abnormal groups, clusters, and/or single cells, and overall impression. Naked nuclei, nucleomegaly, anisonucleosis, and coarse chromatin were more common in BDBs with late mutations than in those with KRAS/GNAS (Kirsten rat sarcoma viral oncogene homolog/guanine nucleotide binding protein, α-stimulating complex locus) mutations only. Cytology had a sensitivity of 16% and a specificity of 100% for malignancy. In contrast, NGS had a sensitivity of 100% and a specificity of 73%. Late mutations were 100% specific for malignancy compared with mutations in KRAS/GNAS only, of which 69% were malignant.

CONCLUSIONS

We found significant overlap in the cytomorphologic features between neoplastic and non-neoplastic BDBs, and more than one half of cancer cases had been interpreted as "nonmalignant" on cytology. NGS showing late mutations was 100% specific for malignancy. Adding genetic testing to BDB cytology would be a valuable ancillary test for the detection of malignancy, and reflex testing should be considered.

Should uncertainty concerning the risk of malignancy be included in diagnostic (nongynecologic) cytopathology reports?

In diagnostic cytology, the known site-specific false positive rates at various anatomical sites for the risk malignancy (ROM) when a confirmed malignant diagnosis is made are comparatively well documented. ROM figures for diagnostic cytology specimens may vary according to the anatomical site of the specimen, the exact nature of the specimen received, the staining method(s) used, and the use of additional laboratory techniques including molecular profiling; furthermore, they often differ to some extent from institution to institution, between differing cytologists within the same institution, and over time. A brief literature review for a selected group of routine diagnostic cytology specimens shows a published ROM for a confirmed malignant diagnosis as follows: bile duct brushings, ~99% (range, 97%-100%); breast fine needle aspiration, 98.5% (range, 92%-100%); serous effusion fluid, 98.9% (range, 90%-100% although lower for squamous cell carcinoma, mesothelioma, and lymphoma), pulmonary endobronchial ultrasound cytology, ~99% (range, 86.6%-100%); thyroid FNA, 98% (range, 97%-99% if NIFTP tumors are excluded), salivary gland FNA, ~90%; (range 57%-100%) and lateral neck cyst FNA, ~99% (range, 95.5%-100%). Because most diagnostic cytology specimens have a small but accepted false-positive rate, this information is vitally important for the clinical management of patients and for shared patient decision making. In our view, the known false-positive rate for a given diagnostic cytology specimen could be included within the cytology report to assist in explaining the limitations of the diagnostic cytology interpretation and help facilitate the clinical decision-making process.