Untargeted profiling of cell-free tumour DNA in plasma and urine of patients with renal tumours - 11th International Symposium on Circulating Nucleic Acids in Plasma and Serum (CNAPS)

Tina Moser ¹ Christopher G. Smith ² Florent Mouliere ² Dineika Chandrananda ² Martin Pichler ³ Maximilian Seles ⁴ Gabriel Wcislo ⁵ Nitzan Rosenfeld ² Michael R. Speicher ¹ Grant D. Stewart ⁶ Ellen Heitzer ¹

¹Institute of Human Genetics, Diagnostic and Research Center for Molecular Biomedicine, Medical University of Graz, Graz, Austria
²Cancer Research UK Cambridge Institute, CRUK Major Centre, University of Cambridge, Cambridge, UK
³Department of Internal Medicine, Division of Oncology, Medical University of Graz, Graz, Austria
⁴Department of Urology, Medical University Graz, Graz, Austria
⁵Department of Oncology, Military Institute of Medicine, Warsaw, Poland
⁶Department of Surgery, University of Cambridge, Cambridge, UK

Background
Extensive pathological and genetic heterogeneity are inherent to renal cell carcinoma (RCC), which confounds treatment decisions and is a significant hurdle in precision medicine. The analysis of circulating tumour DNA (ctDNA) offers a minimally-invasive tool for successive profiling of tumour-specific alterations. Despite showing great promise in various cancers, there is little and often contradictory data of ctDNA in RCC.

Methods
We employed established untargeted sequencing methods to determine the presence, levels and predictive ability of ctDNA in plasma and urine in two independent cohorts of RCC patients (DIAMOND and MonReC; n=90) with a broad range of renal tumour subtypes. For DIAMOND, we utilised the trimmed Median Absolute Deviation (tMAD) score of shallow whole genome sequencing (sWGS) data for ctDNA detection. For MonReC, we applied a combination of modified Fast Aneuploidy Screen Test Sequencing System (mFAST-SeqS) and ichorCNA analysis of sWGS data.

Results
ctDNA was detected in plasma of 8/43 (18.6%) MonReC patients. For DIAMOND, detection rates in plasma and urine samples were similarly low with 6.4% (3/47) and 19% (4/21), respectively. However, in silico size selection for sequencing reads from plasma DNA of 90-150 bp in length to enrich for tumour-derived fragments improved detection rate in 11/47 (23.4%) DIAMOND and in 14/43 (32.6%) MonReC patients. Furthermore, a machine-learning based model, applied to untargeted data, was capable of triaging patients with sufficiently high levels of ctDNA for targeted sequencing.

Conclusion
Although enrichment for tumour-derived fragments based on fragment length improved the ctDNA detection rate, our data indicate low ctDNA levels in RCC. Even the use of sensitive targeted approaches revealed similar findings (see abstract, Smith et al.). Our data suggests that the mechanisms that determine the release and levels of ctDNA in plasma and urine of patients with renal tumours vary, and further developments are needed for clinical utility.