Real World Evaluation of AI-Based Tumor Cell Content Quantification for Molecular Tumor Profiling.

PURPOSE: Repeated evidence demonstrates limited reproducibility and accuracy of the visual quantification (VQ) of the tumor cell content (TCC) by clinical pathologists for downstream molecular testing. Artificial intelligence (AI)-based digital quantification (DQ) of TCC represents a promising alternative, yet real-world evidence from routine molecular diagnostic workflows remain limited. In this study, we evaluated the analytical performance and practical aspects of analytical validation process of an AI-based DQ tool in routine molecular diagnostics. MATERIAL AND METHODS: The clinical-grade AIM-TumorCellularity (AIM-TC; PathAI©) workflow was tested in molecular diagnostics for samples analyzed by comprehensive genomic profiling (FoundationOne®CDx (F1CDx), Foundation medicine Inc.). The cohort included 300 non-paired resection, biopsy, and cytology/cell block) specimens from primary and metastatic breast (n = 66), lung (n = 117), colorectal (n = 40), pancreatic (n = 38), and prostate (n = 39) cancers, reflecting real-world diagnostic sample heterogeneity of a tertiary care center. We compared TCC estimates generated by pathologists' VQ, AI-based DQ, and molecular quantification (MQ) by bioinformatic deconvolution. RESULTS: Agreement was lowest between VQ and MQ (Spearman Rs = 0.38) and between VQ and DQ (Rs = 0.44), while DQ showed stronger concordance with MQ (Rs = 0.63). Single-cell validation against expert ground truth demonstrated high performance of DQ in tumor cell detection, with sensitivity of 0.98.5, specificity of 0.99, and accuracy of 0.99, based on 27,958 annotated cells across 60 regions of interest comparable to microscopic high-power fields. Analysis of pre-analytical and analytical factors identified specimen type and cautery/crush artifacts as the main pre-analytical contributors to DQ-VQ discrepancies, while overall variations in specimen cellularity was the dominant analytical factor. CONCLUSION: In summary, this study provides the first comprehensive real-world evaluation of AI-based TCC quantification in routine molecular pathology workflow, highlighting its robustness, accuracy, and the critical role of pre-analytical standardization, as well as pathologists` oversight for successful clinical implementation.