Optimizing core diameter and number for CK20 immunohistochemistry in colorectal cancer tissue microarrays: A virtual...

Cytokeratin 20 (CK20) expression in colorectal cancer (CRC) exhibits significant intratumoral heterogeneity, posing challenges for tissue microarray (TMA)-based immunohistochemical assessment. Optimal TMA design parameters (core diameter and number) remain undefined. This study aimed to determine optimal TMA parameters for CK20 detection using an optimized virtual TMA (vTMA) simulation strategy. A total of 154 CRC cases were included. Whole slide images of CK20-stained sections were used as reference standards. An optimized vTMA approach involving unrestricted candidate core generation, non-overlapping sampling constraints, and 1000 repeated simulations per parameter combination was employed. Consistency between vTMA-derived CK20 positivity and whole-slide reference was evaluated using intraclass correlation coefficients (ICC) across core diameters (0.6 mm and 1.0 mm) and numbers (1-4). A single 0.6 mm core achieved moderate consistency (ICC = 0.711, 95% CI: 0.710-0.713; RMSE = 0.188). Increasing the core diameter to 1.0 mm substantially improved consistency to a good level (ICC = 0.796, 95% CI: 0.795-0.797; RMSE = 0.150). Using three 1.0 mm cores provided excellent consistency (ICC = 0.922; RMSE = 0.086), with incremental improvements beyond three cores being minimal. Compared with conventional fixed-core pool approaches, the optimized vTMA strategy reduced confidence interval widths by up to 11.5-fold in single-core simulations and by approximately 4.3-fold in four-core simulations. A single 1.0 mm core achieves good agreement with whole-section CK20 assessment, while three 1.0 mm cores optimize consistency beyond 0.90, minimizing sampling error. The optimized vTMA strategy offers a robust framework for TMA parameter optimization in heterogeneous biomarker studies.