Cell-Free Testing Reliably Detects DPYD and UGT1A1 PGx Alterations, Improving Chemotherapy Dosing and Clinical Yield

Introduction – Cell-free liquid biopsies have emerged as powerful diagnostic and therapeutic tools in the management of cancer patients, offering a minimally invasive means to monitor disease and guide treatment. In addition to any tumor DNA, the plasma contains DNA from a patient’s germline, enabling identification of incidental germline alterations (any pathogenic or pharmacogenomic alterations) which is a recognized outcome from these assays. Typical germline genotyping requires obtaining a separate normal (germline) sample for comparison and introduces additional complexity and logistical burden at the time of sample collection. Pharmacogenomic testing for DPYD and UGT1A1 is essential for optimizing dosing of fluorouracil and irinotecan chemotherapy regimens, respectively, to reduce the risk of severe adverse drug reactions. Loss-of-function (LoF) variants in DPYD are particularly important for safe administration of fluorouracil, while LoF and reduced-expression variants in UGT1A1 are critical for dosing irinotecan and other SN-38 compounds. In this study, we leveraged validated germline genotype calls for DPYD and UGT1A1 in a substantial cohort of over 1,200 specimens, comparing them to matched cell-free DNA (cfDNA) assay, to evaluate the performance of genotype determination and use of cfDNA as surrogate for traditional PGx testing.

Methods – The 7 common and 3 underserved population variants in DPYD or UGT1A1: DPYD*2a, DPYD*13, DPYDHapB3, DPYDc.557A>G, DPYDc.2846A>T, UGT1A1*6, UGT1A1*27, UGT1A1*28, UGT1A1*36, and UGT1A1*37 from a CLIA/CAP validated, in-house germline assay (xT-normal) were evaluated in patient matched cfDNA assay (xF+). A cohort of 1224 clinical specimens was analyzed. The specimens contained a range of genotypic events, including homozygous reference and heterozygous and homozygous alternative SNV alleles from DPYD and UGT1A1. For UGT1A1, specimens with the short (*36) and long (*37) indels were included, as well as specimens with complex events (2 non-reference indels in a single specimen). Circulating tumor DNA tumor fraction (ctDNA TF) is an ensemble score estimated from VAF of SNVs/indels and copy number calls observed in the cfDNA population.

Results – The overall SNV/indel genotyping accuracy was 99.8%. For DPYD and UGT1A1 SNVs, the assay demonstrated 100% accuracy. Average coverage for these locations in the two genes was 15163x and indels was 3485x. For DPYD SNVs, all 5 SNVs were observed as heterozygous calls (average 245 times), with homozygous alternative calls detected at 3 positions (n=14 times). For UGT1A1, both SNVs were observed as heterozygous calls (n=4 times) and the indel was detected as a homozygous alternative in 161 instances (n=2, 158 and 1 for *36, *28, *37, respectively). While the overall agreement was excellent, two discrepancies were noted concerning UGT1A1 indels. Further analysis of these discordant calls revealed that they occurred in samples exhibiting higher tumor content (25% and 44% ctDNA TF), suggesting that while robust, the assay might encounter limitations at specific ctDNA concentrations. Despite these isolated cases, the comprehensive evaluation confirms the xF+ assay’s robustness for both germline PGx SNV and indel detection. This minimally invasive blood-based assay offers a promising and convenient alternative to traditional tissue-based testing, facilitating timely and informed treatment decisions for cancer patients by providing critical pharmacogenomic insights. The ability to apply xT algorithms to xF data further streamlines the analytical process, making this a highly valuable tool in precision oncology.

Conclusion – Our findings achieve 99.8% accuracy for detecting DPYD and UGT1A1 PGx variants. This high level of agreement underscores the reliability of cfDNA assay for PGx calling in DPYD and UGT1A1.