Cell-free DNA Fragmentation Profiling for Monitoring Therapeutic Response in Metastatic Colorectal Cancer

1 2 1 1 1 1 1 Bahar Alipanahi, Iris van 't Erve, Keith Lumbard, Laurel K. Millberg, Zachary L. Skidmore, Lorenzo Rinaldi, Jacob Carey, Jennifer 1 3 1 2 4 4 2 Tom, Cornelis J. A. Punt, Nicholas C. Dracopoli, Gerrit A. Meijer, Robert B. Scharpf, Victor E. Velculescu, Remond J.A. Fijneman, 1 Alessandro Leal

1. Delfi Diagnostics, Inc., Baltimore, MD, United States

2. Department of Pathology, Netherlands Cancer Institute, Amsterdam, Netherlands.

3. Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, the Netherlands.

4. The Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD, United States

5. The Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD, United States

Disclosure Information

Bahar Alipanahi, PhD

I have the following relevant financial relationships to disclose: Delfi Diagnostics, Inc.: Employee, Stock, Patent, and Other Intellectual Property

Introduction

▪ Measurement of mutant allele fraction (MAF) in cell-free DNA (cfDNA) via expensive target panels typically relies on prior tumor tissue analyses or knowledge of specific mutations

▪ There is an unmet need to develop an effective and inexpensive test to provide predictive information about treatment outcomes in patients receiving a variety of systemic therapies

▪ The DELFI Tumor Fraction approach (DELFI-TF) is a cfDNA fragmentome-based model capable of accurately detecting circulating tumor DNA fractions without needing genetic information about the tumor of origin

▪ Here, we evaluate the first version of the DELFI-TF approach for treatment response monitoring in patients with metastatic colorectal cancer (mCRC)

Clinical Cohort Description

▪ The current study is a planned exploratory analysis of longitudinal plasma samples from 1 mCRC participants enrolled in the phase III study CAIRO5 (NCT02162563)

▪ Study participants had their tumor tissue specimens genotyped for KRAS (exon 2, 3, and 4), NRAS (exon 2 and 3), and BRAF (codon 600) mutations before enrollment

▪ All available blood draws at the pre-treatment baseline and on-treatment timepoints were retrospectively processed

▪ Cell-free DNA tumor fraction was assessed as the mutant allele frequency (MAF) of the tumor-tissue-proven RAS/BRAF mutations measured by ddPCR ▪ An external cohort of non-cancer individuals was used for analytical reference

1. Huiskens, et al. BMC Cancer, 2015. 2. Mathios, et al. Nat Commun, 2021

Model Development

▪ Utilized cfDNA aliquots used for whole genome sequencing were obtained from the same plasma volumes used for ddPCR testing

▪ After aligning the low-coverage WGS data to the reference genome, fragment-based genomic features were obtained from 504 non-overlapping 5Mb bins

▪ A Bayesian regression model was trained and cross-validated against RAS/BRAF MAF using the DELFI scores, PA scores, and fragment length distribution weights

▪ DELFI-TF slope was defined as the slope of the regression line fitted to the DELFI-TF values across timepoints associated with the response assessment by RECIST1.1

Study Flowchart

DELFI-TF strongly correlates with ddPCR MAF in patients with RAS/BRAF mutant mCRC

DELFI-TF Values Are High in Patients With Cancer and Reduced After Initiation of Therapy

Fragmentation profiles show marked heterogeneity at pre-treatment and disease progression timepoints

Fragmentation profiles exhibit aberrations even in the context of copy neutral regions in matched tissue

Tissue

Plasma

Baseline DELFI-TF and MAF are equivalent in predicting imaging response and resection status

Baseline DELFI-TF and MAF analogously predict survival outcome in RAS/BRAF mutant patients

Pre-treatment MAF and DELFI-TF are categorized as Low (< first 25th %) and High (> first 25th %)

Case study: dynamic changes of DELFI-TF and MAF compared to imaging assessment

DELFI-TF dynamics predict survival outcomes in the overall study population

DELFI-TF slope is categorized as Low (below median) and High (above median)

HR = 2.03 (95% CI: 1.247 to 3.318)

Conclusions

▪ In this proof-of-concept study, we present DELFI-TF as a model that utilizes low-coverage whole genome sequencing and cfDNA fragment-based statistics to predict tumor fraction in a non-invasive manner

▪ DELFI-TF strongly correlates with standard MAF testing in patients with mCRC, allowing equivalent prediction of imaging response and liver metastases resection status

▪ Tumor fractions at baseline and their dynamic changes predicted by DELFI-TF correlate with survival outcomes in mCRC regardless of the presence of known driver mutations

▪ The development of a tissue-independent approach that does not rely on mutation detection has the potential to expand the use of tumor-agnostic models for both early and advanced disease monitoring

Acknowledgements

All patients enrolled in the CAIRO5 study Dutch Colorectal Cancer Group Netherlands Cancer Institute University Medical Center Utrecht Johns Hopkins University