# Analytical validation and sequencing coverage studies suggest that performance of a liquid biopsy assay is tumor agnostic (DNA-is-DNA)

## TL;DR

This study shows that the performance of a liquid biopsy test is consistent across different cancer types, suggesting that tumor-specific sample sets may not be needed for validation.

## Contribution

The study demonstrates that analytical performance of a liquid biopsy assay is tumor-agnostic, challenging current regulatory requirements.

## Key findings

- Precision and concordance metrics showed minimal variation between tumor types.
- Coverage distribution was similar across different cancer types in clinical samples.
- Results suggest that tumor type does not significantly impact assay performance.

## Abstract

Per regulatory and standard requirements (e.g., Clinical & Laboratory Standards Institute (CLSI) guidelines, United States Food and Drug Administration (FDA) correspondence), analytical validation (AV) for each companion diagnostic (CDx) biomarker should be repeated using a clinical sample set for each cancer type listed as an indication in labelling for a CDx. Using data from AV studies and Foundation Medicine (FMI)’s clinical database, we evaluated the hypothesis that analytical performance of the FoundationOne®Liquid CDx (F1LCDx) assay is not impacted by cancer type and that large sets of clinical, tumor-specific samples might not be necessary for analytical validation of specific CDx biomarkers. We retrospectively evaluated all liquid biopsy samples from F1LCDx assay AV studies that were executed between April 2019 and November 2021 and clinical samples processed by F1LCDx between September 2020 and October 2021. For the samples from AV studies, we evaluated the precision and concordance performance by F1LCDx between tumor types; and for the clinical samples, we performed analyses comparing the distribution of coverage between tumor types. A total of 31,247 F1LCDx clinical samples and 579 samples from F1LCDx AV studies with a total of 335 disease ontologies (DOs) were included in this study. For precision: the median absolute pairwise difference of mean reproducibility between any pairs of two tumor types is 0.94% [0.01%−2.63%] and the median absolute pairwise difference of mean repeatability between any pairs of two tumor types is 0.91% [0.03%−2.98%]. For concordance: the median absolute ∂PPA between any pairs of two tumor types is 1.39% [0.1%−4.1%] and the median absolute ∂NPA between any pairs of two tumor types is 0.05% [0%−1%]. For coverage, a similar distribution was observed between tumor types using F1LCDx clinical samples. Herein, the results based on the extensive cohort of 31,826 liquid biopsy samples sequenced by the F1LCDx assay demonstrated that both analytical assessment of precision and concordance and coverage are comparable among tumor types (i.e., deoxyribonucleic acid (DNA) is DNA). The tumor type that circulating tumor DNA (ctDNA) was derived from is therefore not a vital consideration for AV studies for F1LCDx assay.