A novel approach, Northstar Response™, utilizes molecule counting to precisely quantify methylated circulating tumor DNA (ctDNA) for tumor-naive cancer therapy response monitoring. The multiplex technology targets over 500 genomic locations hypermethylated in cancer, using quantitative counting templates (QCTs) to count methylated molecules at each location.
Analytical validation showed accurate detection of 0.25% absolute changes in tumor fraction (AUC > 0.94) and high accuracy across 12 tumor tissue types. The assay correlates with and is up to two times more precise than tumor-naive, targeted-panel methods measuring variant allele fraction (VAF), achieving coefficients of variation (CVs) < 10% in 1% tumor fraction samples. High concordance was observed between clinical outcomes and Northstar Response methylation measurements. The assay employs an amplicon-based, multiplex PCR approach with QCTs and next-generation sequencing (NGS) to count methylated molecules at over 500 genomic locations hypermethylated in cancer tissue. The Cancer Genome Atlas (TCGA) data was analyzed to identify these hypermethylated locations. The assay targets CpG islands, regions of DNA where a cytosine nucleotide is followed by a guanine nucleotide in the DNA sequence, known to be hypermethylated in various cancers. Theoretical assay noise calculations determined that sampling at least 1,000 methylated molecules would achieve a 3% CV, enabling accurate detection of 10% relative changes in tumor fraction. The assay design requires approximately 500 loci to achieve this CV. The targeted CpG sites are hypermethylated in multiple cancer types, and simulated measurements showed significantly increased methylation in cancer compared to normal tissue. The genomic locations covered in the assay are implicated in cancer biology. The method targets genes with DNA-binding transcription factor activity and homeobox genes, which are involved in developmental processes and oncogenicity. The selected loci are enriched for genes involved in development, morphogenesis, and cell differentiation. QCTs are used to calculate the number of input methylated molecules at each locus, accounting for amplification bias. Each QCT contains an embedded molecular identifier (EMI) with eleven random nucleotides, ensuring the uniqueness of each QCT molecule. Reads that map to each amplicon are classified as methylated based on the number of CpGs, and the total number of methylated reads is divided by the reads per molecule to calculate the number of methylated sample molecules at this amplicon prior to PCR. Non-cancer background methylation is accounted for by subtracting the number of methylated molecules measured in the buffy coat (white blood cell fraction of blood) from that measured in the paired cfDNA on a per-locus basis. Normalization loci, targeting CpG sites highly methylated in all tissues, are used to estimate the total amount of input DNA. After quantifying the number of methylated molecules, normalizing to an input of 1,000 genome equivalents (GE) of DNA, and subtracting background signal, normalized molecule counts are summed across loci to calculate the Tumor Methylation Score. Technical replicates of contrived samples, made from sheared tumor DNA spiked into sheared buffy coat DNA at various tumor fractions, were tested. The Tumor Methylation Score linearly increased with contrived tumor fraction, accurately quantifying methylated tumor DNA. The CV of the Tumor Methylation Score was < 20% for all contrived tumor fractions. The assay's ability to distinguish between different contrived tumor fractions was evaluated using receiver operating characteristic (ROC) curves.
The assay accurately called changes in Tumor Methylation Score with high sensitivity and specificity. The technology was able to distinguish whether there was a decrease to 0.5% tumor fraction or no change in tumor fraction with accuracy approaching 100%. The precision of the Northstar Response™ was compared to tumor-naive VAF-based monitoring approaches. The CV of Tumor Methylation Score was lower than the median CV and simulated CV of mean VAF at all tested VAF tumor fraction bins. Northstar Response™ achieved CVs twice as low as that measured using average VAFs with the tumor-naive, targeted-panel assay. The technology performance was tested across 54 solid tumors originating from 12 different tissues.
Tumor Methylation Scores showed high concordance with clinical outcomes. All seven patients had an initial decrease in Tumor Methylation Score after starting treatment. Changes in Tumor Methylation Scores reflected the dynamics of cancer treatment.