Towards detection of minimal residual disease in multiple myeloma through circulating tumour DNA sequence analysis Trevor Pugh, PhD, FACMG Princess Margaret Cancer Centre, University Health Network Dept. of Medical Biophysics, University of Toronto trevor.pugh@utoronto.ca
Most cell-free DNA in blood plasma is derived from non-cancerous cells, even in myeloma patients Crowley et al., Nature Reviews Clinical Oncology, 2013
Sufficient blood volumes are needed to enable adequate sampling of low concentration ctdnas 83 ng 99.99% 65 ng 99.9% probability to sample one tumor fragment amongst 2,000 normal fragments (0.05%) 2 tubes 4 tubes In practice, we are collecting 2-4x 10 ml tubes of blood Myeloma Advanced Solid Tumours
Cell-free DNA consists of ~150 & 350 bp fragments, consistent with protection by nucleosomes DNA size reference (base-pairs) (20-60 bp) 350 150
Hybrid-capture of ctdna fragments enables full-length gene sequencing for mutation detection, as well as structural and copy number variation X TARGET EXON X
ctdna has high sensitivity (96%) and specificity (98%) to detect mutations reported in 43 bone marrows; all false positives seen seen in serial bloods Pos ctdna Neg Tumour Pos Neg True Positives False Negatives False Negatives True Negatives not covered by clinical bone marrow panel Pos ctdna Neg Tumour Pos Neg 44 2 3 172
Actionable mutations detected in 9/13 patients without matched bone marrow, serial draws consistent with prior testing
Cell-free DNA reflects clonal structure and adjacent hotspot mutations are resolvable to the nucleotide cfdna BM-derived DNA
Molecular barcoding techniques can correct for sequencer, polymerase, and DNA damage error Schmitt et al. PNAS USA. 2012 Sep 4;109(36):14508-13.
Background error rate can be overcome by molecular barcoding, if library has enough unique molecules Original reads Hemizygous TP53 mutation Single strand consensus sequence Duplex consensus sequence Read fraction Read count Dilution series MM:CRC cell line 1 10 1 10 2 10 3 10 4 10 5 10 6 1 10 1 10 2 10 3 10 4 10 5 10 6 1 10 1 10 2 10 3 10 4 10 5 10 6
Detection of structural rearrangements is more tolerant of sequencer error; breakpoints at nucleotide resolution KMS11 cell line, t(4;14) and t(14;16) Undiluted, 3,500X coverage 1/1,000 diluted, 10,000X coverage WWOX IgH WWOX IgH
T-Cell Receptor capture sequencing is quantitative and reflects expected clonal patterns applicable to Ig loci Fraction of reads mapped to unique VJ junctions 1.0 0.8 0.6 0.4 0.2 Peripheral Blood Mononuclear Cells 0 Melanoma TILs expanded in IL-2 Ovarian TILs expanded in IL-2 PBMC peptide stim. then flow sort gp100-specific TIL clone Jurkat T-lymph. cell line α β γ α β γ α β γ α β γ α β γ α β γ David Mulder, Etienne Mahe, Pam Ohashi, Naoto Hirano, Marcus Butler, Trevor Pugh
One-tube multiple myeloma ctdna capture panel design Translocations into the immunoglobulin heavy chain locus Recurrent copy number aberrations Significantly mutated genes Ig & TCR VDJ rearrangements for clonality tracking IgH partner loci including MYC Gene level; gain(1q); del(1p32); del(1p12); del(17p); αβγδkl loci 1,675 probes x 120 bp/probe 201 kb targeted Prognostic Significance Potential Therapeutic Targets Mutations associated with drug resistance DNA repair genes Transcription factors and regulatory proteins 18 genes MAPK pathway NF-kB pathway 20 genes Immunomodulatory agents, Proteosome inhibitors
Summary & Design Considerations Cell-free DNA is a non-invasive source of tumour-derived material to guide management of cancer patients over time Panel design balancing actionability, cost, & throughput Hybrid capture and ultra-deep sequencing of tumour DNA in blood is feasible as a liquid biopsy of multiple myeloma flexible mutation detection platform with sensitivity dictated by depth, error suppression strategies, volume of plasma tested, and ctdna concentration To further improve sensitivity, improvements are needed in informatics and laboratory protocols large blood draw volumes, low/no PCR cycles, higher efficiency ligation, large panels to address allele drop-out
Pugh Lab Olena Kis David Mulder Arnavaz Danesh Jessica Liu Signy Chow Tiantian Li Mark Dowar Mark Mansour Princess Margaret Cancer Centre Suzanne Trudel Rayan Kaedby Suzanne Kamel-Reid Tracy Stockley Tong Zhang Mahadeo Sukhai Scott Bratman Ming Tsao Anthony Joshua Etienne Mahe Lillian Siu Anna Spreafico Neil Winegarden & PM Genomics Centre Carl Virtanen & Bioinformatics Services
Copy number variant detection ~10% sensitivity need more probes and/or better algorithm? Undiluted 1/10 dilution Log2 [sample median/batch median] 1/100 dilution 1/1000 dilution 16
Circulating tumour DNA sequencing workflow 1. Extract Cell-Free DNA from Blood Plasma Evaluate cfdna yield and fragment size Evaluate DNA-Seq library yield and fragment size 2. Prepare DNA-Seq Libraries (ligate to unique sample barcodes) 3. Capture/PCR DNA fragments containing target genomic sequences Evaluate captured DNA yield and fragment size 4. Sequence captured DNA fragments (e.g. Illumina HiSeq) Evaluate Mean Target Coverage & Off-Target Reads 5. Analyze sequencing data to identify somatic variants (alignment, data QC, variant calling) Compare ctdna mutation calls to primary tumour