NEW YORK – A University of Wisconsin at Madison-led team has tapped into tumor DNA fragment and nucleotide frequency profiles in blood samples to come up with an approach that could be used to detect or monitor early-stage cancer.
"Genome-wide fragmentation patterns in cell-free DNA (cfDNA) in plasma are strongly influenced by cellular origin due to variation in chromatin accessibility across cell types," senior and corresponding author Muhammed Murtaza, a University of Wisconsin-Madison researcher in surgery, human genomics, and precision medicine, and his co-authors wrote in Science Translational Medicine on Wednesday. "Such differences between healthy and cancer cells provide the opportunity for development of novel cancer diagnostics."
Based on initial results revealing distinct DNA fragments that reflect nucleosome positions and chromatin accessibility patterns, the researchers turned to whole-genome sequencing to profile "recurrently protected regions" (RPRs) in DNA isolated from blood plasma samples in 17 healthy individuals, identifying RPR and fragmentation patterns that pointed to the possibility of using DNA fragmentation and nucleotide frequency patterns for blood-based cancer diagnoses.
From there, the team used its "genome-wide analysis of fragment ends" (GALYFRE) method to assess 521 newly sequenced samples, which it analyzed in combination with 2,147 plasma samples sequenced in the past. All told, the collection spanned 2,668 blood plasma samples from 994 individuals with tumors from 11 cancer types, as well as 103 cancer-free patients with nonmalignant disease forms and 286 healthy control participants.
"Unlike earlier studies that relied on differences in fragment lengths across genomic regions or on differences in sequence motifs in individual cfDNA fragments, GALYFRE aggregates genomic positioning of break points across all sequenced fragments in a sample," the authors explained, noting that their results demonstrated that "measurement of aberrant fragmentation is a potential biomarker to distinguish blood samples from patients with cancer and healthy individuals."
In particular, the researchers found that their "information-weighted fraction of aberrant fragments" (iwFAF) method corresponded with tumor type, tumor fraction profiles, and the number of somatic point mutations and/or copy number changes found in a given region in the genome in circulating cfDNA — analyses that were possible with relatively low sequencing depths and more modest DNA inputs compared to prior approaches.
The investigators subsequently came up with a machine learning method that combines iwFAF data with fragment end nucleotide frequencies to distinguish samples from individuals with or without any stage of cancer with nearly 67 percent average sensitivity at a specificity of 95 percent. While the area under the receiver operative characteristic curve (AUC) was 0.91 across all cancer stages considered, they explained, the approach achieved an AUC of 0.87 for stage I cancers.
"Our findings remained robust with as few as 1 million fragments analyzed per sample," the authors wrote, "demonstrating that analysis of fragment ends can become a cost-effective and accessible approach for cancer detection and monitoring."