EBUS TBNA Diff Quik smears have highly predictable DNA content and great feasibility for large 500 gene panel testing in lung cancer, even in smears with low cellularity.

D. Fielding*^a (Dr), A. Dalley^b (Dr), P. Simpson^b (Dr), K. Nones^c (Dr), V. Lakis^c (Dr), S. Sharma^c (Dr)

^a RBWH, Brisbane, AUSTRALIA ; ^b University of Queensland Centre for Clinical Research, Brisbane, AUSTRALIA ; ^c QIMR Berghofer, Brisbane, AUSTRALIA

* david.fielding@health.qld.gov.au

Background

Diff quik smears are commonly collected at EBUS TBNA procedures. These smears are not commonly used for molecular analysis, despite reports of insufficient cell block material. Studies to demonstrate the great potential of smears are needed in a wide range of smear cellularities. Novel extended sequencing panels need assessment also.

Methods.

This was a prospective study of patients having EBUS TBNA for malignant lymph nodes. Rapid on-site assessment (ROSE) Diff quik smears and cell block were made in all patients. Smears had microscopy for percent malignant cellularity then were extracted for DNA content. A subset of these smears were sequenced using the TSO500® panel.

Results

A total of 181 smears from 66 patients with lung malignancy were analysed. Microscopy for percent malignant cells on the smears was highly predictive of DNA amount. DNA yield ranged from 79 to 12,000ng and mean DNA yield for each quartile of percent cellularity (<25%, 25-50%, 50-75% and >75%), showed significant differences in DNA yield (p≤0.05). DNA from 52 smears underwent TSO500® sequencing: 12 had <25% malignant cells, 13 25-50%, 6 50-75% and 20 had >75%. 51/52 samples produced libraries, but 4 failed coverage on vendor Quality Control. These 4 had <25% cellularity, and smear surface area on the slide was very low (< 25%) in 3. Tumour mutation burden and Microsatellite instability was reportable for 46 smears. A total of 40 smears from 13 patients with known mutations (from cell block) confirmed mutations in all smears. Additional targetable mutations were found in smears from 2 patients (KRAS and PIK3A) where cell block was insufficient.

Conclusions

Microscopy is highly predictive of DNA content on smears. DNA from smears can successfully be sequenced with a large panel in most smears even when low in cellularity providing a reliable alternative to cell block extraction.

Disclosure of funding source(s):

Cancer Australia

Cancer Council of Queensland

Olympus Australia