Analytical validation of a 90-gene expression assay for pan-cancer tissue-of-origin classification in Southeast Asia.

16

Background: Accurate identification of the tissue of origin (TOO) is critical for enabling site-specific therapy in patients with metastatic cancer, particularly cancer of unknown or uncertain primary (CUP). The 90-gene expression assay (Canhelp-Origin) is an RT-PCR-based RNA assay developed for pan-cancer TOO classification. In a randomized clinical trial (Fudan CUP-001), site-specific treatment guided by this assay significantly improved progression-free survival compared with empirical chemotherapy. To support broader clinical implementation, we performed an analytical validation of the assay using tumor samples from Southeast Asia. Methods: Analytical validation was retrospectively conducted at a central laboratory (M Diagnostics Pte Ltd, Singapore) using total RNA extracted from formalin-fixed, paraffin-embedded (FFPE) tumor specimens collected primarily from Southeast Asian countries. Gene expression of a predefined 90-gene panel was quantified using the QuantStudio 5 real-time PCR platform. A locked 90-gene classifier generated similarity scores (0–100; sum = 100) across 21 candidate tumor types. The analytical performance metrics included classification accuracy, robustness under low RNA input and low tumor cellularity, precision, repeatability, and resistance to variations caused by necrotic tissue, tumor enrichment methods, treatment and metastatic status. Results: A total of 64 tumor samples representing 12 cancer types were evaluated for assay performance. The assay achieved a top-1 accuracy of 84.4% (54/64) and a top-3 accuracy of 93.8% (60/64). Among 20 analytically challenging sample with reduced RNA input and tumor cellularity <60%, the concordance with reference pathological diagnoses was 95.0% (19/20). Precision testing across two operators, two reagent lots, and two PCR platforms demonstrated high reproducibility, with a standard deviation of similarity scores of 1.7 and a coefficient of variation of 2.4%. Repeatability testing showed fully concordant results in duplicate analyses. Accurate TOO classification was maintained in samples containing up to 25% necrotic tissue. Conclusions: The 90-gene expression assay demonstrated robust, precise, and reproducible analytical performance in tumor samples from Southeast Asia. These findings support the assay’s analytical robustness across different populations and its reliability for clinical TOO classification in pan-cancer and CUP settings.