TISSUE OF ORIGIN TEST KIT FFPE

The Pathwork® Tissue of Origin Test Kit-FFPE is an in vitro diagnostic tool that employs microarray-based RNA expression profiling of approximately 2000 gene markers from FFPE tumor samples to generate similarity scores against a 15-tumor type database, aiding in the identification of a tumor's primary origin. This modification validates the test's consistent performance when using a reduced RNA input (30-150 ng) and an altered cDNA purification method involving ethanol, as confirmed by a high concordance rate in comparative studies.

The Pathwork® Tissue of Origin Test Kit-FFPE is an in vitro diagnostic system designed to assist physicians in identifying the primary site of tumors. It analyzes RNA expression patterns from a patient's formalin-fixed, paraffin-embedded (FFPE) tumor sample and compares them to a database of fifteen known tumor types. The core technology involves the Pathchip, a custom microarray that measures the transcription levels of approximately 2000 human gene markers. A proprietary algorithm, developed using machine learning from a large specimen database, processes this gene expression data to generate 15 Similarity Scores, each indicating the likelihood that the patient's tumor matches one of the tissue origins in the test panel. These scores, which sum to 100, are presented in a report for clinical interpretation alongside other diagnostic information. This document details a modification to the test kit's procedure, specifically reducing the required input of total RNA from over 150 ng to a range of 30-150 ng, and incorporating an ethanol precipitation step in the cDNA purification process. A non-clinical validation study was conducted using 43 paired FFPE tumor specimens, comparing results from the original high-RNA input method without ethanol to the modified low-RNA input method with ethanol. The study demonstrated a 97.7% concordance (42 out of 43 pairs) between the two methods. Further statistical analyses, including similarity score regression (R=0.94) and Bland-Altman plots, supported the conclusion that these procedural changes do not significantly alter the test's performance in determining tissue of origin.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. What were the specific characteristics (e.g., tumor type, initial RNA quality) of the single discordant case in the 30-150 ng RNA validation study, and how might this inform the test's limitations with low-input samples?; Given the database comprises 15 tumor types, how does the algorithm's performance vary across these types, particularly when using the modified low RNA input protocol, and what is the confidence interval for similarity scores for each type?; The Pathchip covers over 14,500 genes, but only 2000 are used as markers. What was the rationale for selecting these specific 2000 markers, and has the potential utility of the remaining ~12,500 genes for refining tissue of origin or identifying other clinically relevant features been explored?; The report mentions 'machine learning methods' for marker selection and model building. Could more specific details on the algorithms (e.g., SVM, random forest, neural network) and the validation strategy for the original model (e.g., cross-validation, independent test set performance beyond the 2196 specimens) be provided?; How does the introduction of ethanol in the cDNA purification step specifically impact RNA yield or quality from FFPE samples, and were any comparative studies done to isolate the effect of ethanol versus the reduced RNA input?