Quantitative assessment of cell block cellularity and correlation with molecular testing adequacy in lung cancer

Abstract

Determination of fine-needle aspiration (FNA) material adequacy is essential prior to performing molecular testing (MT) in order to ensure good results and maximize resources. This study investigates several quantitative measures of cellularity in FNA samples of lung carcinoma, and correlates the results with MT adequacy. A blinded retrospective analysis of 20 non-small-cell lung carcinoma cases was conducted: 13 contained "sufficient" material for EGFR/KRAS sequencing and ALK FISH studies; and 7 contained "insufficient" material for these tests. Three 400x fields-of-view (FOVs) were analyzed from digitized cell block glass slides of these cases. Cellularity in these FOVs was quantified using three methods: (1) visual estimation by cytopathologist; (2) manually annotated contours (MACs); (3) software derived, manually adjusted contours (SDMACs) using a custom segmentation script with adjustable parameters. These methods were evaluated using the Mann-Whitney-Wilcoxon test, paired t test, and receiver operating characteristic/area under the curve (AUC) analysis. There were significant differences between the insufficient/sufficient groups for each estimation method (visual P < 0.05, MAC P < 0.05, SDMAC P < 0.01). Variation of mean values was highest in the visual estimation method. AUC values were visual estimation = 0.903, MAC = 0.903, and SDMAC = 0.958. Mean variation of the 3 FOV values was found to be significantly higher for visual estimation compared with the other methods. Quantitative analysis of cellularity in digitized cell block material is feasible using different methods. In this investigation, the SDMAC method provided the highest accuracy and lowest variability. This supports image analysis as an objective and quantitative tool to assess FNA sample adequacy for guiding supplemental MT.