Tissue Microarray vs. Whole-Slide Analysis of CD8 in Non-Small Cell Lung Carcinoma

Abstract

HYPOTHESIS: Tissue microarray sampling of CD8+ cells at the invasive margin in non-small cell lung carcinoma adequately recapitulates the CD8+ cell distribution of the whole-slide (WS). Aim 1: Quantify the spatial distribution of CD8+ cells in NSCLC. Aim 2: Evaluate the analytical accuracy of TMA sampling relative to WS samples in NSCLC. Aim 3: Identify sources of error in TMA construction and provide directives for the future design of TMAs for the study of immunological markers in NSCLC. METHODS: A TMA was constructed using approximately 3 cores from the invasive margin (IM; 50%:50% tumour:stroma) and 3 from central tumour (CT) of 35 LSC resections (N=30). The TMA and corresponding WS samples were stained with CD8 IHC. The CT and IM were digitally annotated on scans of each WS sample, and CD8 was automatically quantified using digital pathology software. The WS IM is defined as 1-mm centred on the tumour edge. The CD8 density distribution was visualized in 50-μm increments, extending 1mm on either side of the tumour. Technical core accuracy was reflected in the measured distance of each core to the tumour border. RESULTS: WS and TMA CD8 densities are higher at the IM than in CT (Wilcoxon: p=1.025e-3, MWU: p=7.140e-7, respectively), and the peak WS CD8 density occurs within 200μm beyond the tumour border. TMA cores may overestimate the WS IM CD8 density (MWU: p=8.526e-4), however, they were selected from regions of representatively high immune infiltration. Intratumoural densities are relatively constant and are adequately represented by TMA cores (MWU: p=0.959). SIGNIFICANCE: Optimizing spatial tissue core sampling is foundational to the construction of TMAs and will help direct the future of personalized cancer care in NSCLC- a single TMA may be used for thousands of analyses.