Assay Validation of Cell-Free DNA Shallow Whole-Genome Sequencing to Determine Tumor Fraction in Advanced Cancers

Blood-based liquid biopsy is increasingly used in clinical care of patients with cancer, and fraction of tumor-derived DNA in circulation (tumor fraction; TFx) has demonstrated clinical validity across multiple cancer types. To determine TFx, shallow whole-genome sequencing of cell-free DNA (cfDNA) can be performed from a single blood sample, using an established computational pipeline (ichorCNA), without prior knowledge of tumor mutations, in a highly cost-effective manner. We describe assay validation of this approach to facilitate broad clinical application, including evaluation of assay sensitivity, precision, repeatability, reproducibility, pre-analytic factors, and DNA quality/quantity. Sensitivity to detect TFx of 3% (lower limit of detection) was 97.2% to 100% at 1× and 0.1× mean sequencing depth, respectively. Precision was demonstrated on distinct sequencing instruments (HiSeqX and NovaSeq) with no observable differences. The assay achieved prespecified 95% agreement of TFx across replicates of the same specimen (repeatability) and duplicate samples in different batches (reproducibility). Comparison of samples collected in EDTA and Streck tubes from single venipuncture in 23 patients demonstrated that EDTA or Streck tubes were comparable if processed within 8 hours. On the basis of a range of DNA inputs (1 to 50 ng), 20 ng cfDNA is the preferred input, with 5 ng minimum acceptable. Overall, this shallow whole-genome sequencing of cfDNA and ichorCNA approach offers sensitive, precise, and reproducible quantitation of TFx, facilitating assay application in clinical cancer care.

Circulating tumor DNA and homologous recombination deficiency in bone-predominant mCRPC prior to radium-223 therapy

203

Background: Radium-223 (Ra-223) is a bone-seeking alpha emitter that induces double-stranded DNA breaks, and the homologous recombination (HR) pathway is critical for repairing these breaks. While prior studies suggested that metastatic castrate-resistant prostate cancers (mCRPC) patients (pts) with HR-deficient (HRD+) tumors may be more likely to benefit from Ra-223, obtaining tissue for next generation sequencing to identify HRD+ is challenging in pts with bone-predominant disease. We hypothesized that circulating tumor DNA (ctDNA) would allow for broader identification of HRD+ to assess association with clinical outcomes in a real-world cohort. Methods: We identified 135 mCRPC pts treated with Ra-223 at our institution between 2013 and 2021. Pts who initiated another anti-tumor therapy within 60 days of Ra-223 treatment were excluded; pts continuing hormonal agents initiated >60 days prior were included. ctDNA isolated from pre-treatment plasma underwent ultra-low-pass whole genome sequencing to estimate tumor fraction (TF). Additionally, targeted panel sequencing using an institutional prostate cancer-specific panel of 319 genes with duplex sequencing (utilizing unique molecular identifiers) for error suppression was used to identify germline or somatic deleterious alterations in HR pathway genes. The primary outcome was association between HRD status and completion of fewer than 6 cycles (as a proxy for early clinical progression), assessed using logistic regression. Results: The median age was 61 (IQR, 56-67) years, median pretreatment prostate-specific antigen (PSA) level was 26.2 (IQR, 8.1-84.1) ng/mL, and median TF was 4% (IQR, 3-6%). 97% of pts (n=131) previously received a novel antiandrogen, and 63% (n=85) received prior taxane. 17% (n=23) were HRD+, and 59% (n=80) completed 6 cycles of Ra-223. On multivariable analysis, HRD+ was associated with decreased likelihood of completing 6 cycles compared to HRD- (adjusted odds ratio [AOR] 0.16, 95% confidence interval [CI] 0.05-0.48, P=0.001). 22% (n=5) of HRD+ pts completed 6 cycles compared to 67% (n=75) of HRD- pts. Additional factors associated with decreased likelihood of completing 6 cycles included a higher pretreatment TF (AOR 0.69, 95% CI, 0.48-0.97, P=0.034) and prior taxane use (AOR 0.41, 95% CI, 0.18-0.91, P=0.028), but not pretreatment PSA ( P=0.574). Conclusions: Targeted panel sequencing with error suppression from ctDNA identified HRD+ in mCRPC pts with bone-predominant disease and low median TF at a similar frequency as reported from tissue. In our cohort, HRD+ was prognostic of early clinical progression with Ra-223. Further work is in progress to understand the association of other ctDNA-derived features, including assessment of genomic signatures and transcriptional binding sites, in the setting of Ra-223 therapy. DF/HCC IRB protocol 18-135.

Abstract 2485: Somatic analysis services with best practice workflows in a cloud-based platform

With advances in high-throughput sequencing technologies and analytical tools, genomic analysis of tumors has led to the identification of various important somatic mutations that shed light on diagnosis, prevention, and treatment for cancer. However, detecting somatic variants is not a trivial task in terms of the technical aspects (e.g. filtering germline events and removing a variety of noises in tumor samples) and computational resources to handle large-scale cohort analysis. There is also a demand for maintaining stable software versions and the workflows for studies over extended periods of time, that need consistency and traceability, such as clinical trials. We introduce here, the Translational Analysis Group (TAG), a team which deploys, validates, and conducts scalable analytical workflows in a secure, cloud-based environment. We maintain 29 well-tested workflows with best practice methods and ample resources for both somatic and germline analysis. Our cloud platform, FireCloud, enables us to run workflows at any scale. Since May 2017, our team has performed nearly 10,000 analyses for mutation detection (SNV, InDel, CNV, and SV) and cohort analysis on tumor samples and cell-free DNA samples. TAG offers a range of options for somatic and germline variant detection, from legacy pipelines through recently validated contemporary pipelines to allow for continuity across long running projects.

Citation Format: Junko Tsuji, Andrew Hollinger, Alyssa MacBeth, Brian R. Grander, Micah Rickles-Young, Tera Bowers, Carrie Cibulskis, Niall Lennon. Somatic analysis services with best practice workflows in a cloud-based platform [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2485.