Article Text
PDF
Abstract
Aims The majority of pancreatic ductal adenocarcinomas (PDACs) harbour oncogenic mutations in KRAS with variants in TP53, CDKN2A and SMAD4 also prevalent. The presence of oncogenic fusions including NTRK fusions are rare but important to identify. Here we ascertain the prevalence of NTRK fusions and document their genomic characteristics in a large series of PDAC.
Methods Whole genome sequencing and RNAseq were performed on a series of patients with resected or locally advanced/metastatic PDAC collected between 2008 and 2020 at a single institution. A subset of specimens underwent immunohistochemistry (IHC) analysis. Clinical and molecular characterisation and IHC sensitivity and specificity were evaluated.
Results 400 patients were included (resected n=167; locally advanced/metastatic n=233). Three patients were identified as harbouring an NTRK fusion, two EML4-NTRK3 (KRAS-WT) and a single novel KANK1-NTRK3 fusion. The latter occurring in the presence of a subclonal KRAS mutation. Typical PDAC drivers were present including mutations in TP53 and CDKN2A. Substitution base signatures and tumour mutational burden were similar to typical PDAC. The prevalence of NTRK fusions was 0.8% (3/400), while in KRAS wild-type tumours, it was 6.25% (2/32). DNA prediction alone documented six false-positive cases. RNA analysis correctly identified the in-frame fusion transcripts. IHC analysis was negative in the KANK1-NTRK3 fusion but positive in a EML4-NTRK3 case, highlighting lower sensitivity of IHC.
Conclusion NTRK fusions are rare; however, with emerging therapeutic options targeting these fusions, detection is vital. Reflex testing for KRAS mutations and subsequent RNA-based screening could help identify these cases in PDAC.
- pancreatic neoplasms
- genes
- neoplasm
- pathology
- molecular
Data availability statement
No data are available. No data is available.
Statistics from Altmetric.com
Introduction
Pancreatic ductal adenocarcinoma (PDAC) is a leading cause of cancer-related death globally,1 with a 5-year overall survival (OS) of approximately 10%.2 Multiagent chemotherapy remains the mainstay of treatment both in early and advanced disease. In contrast to many other tumour types, targeted approaches are lacking and death rates are not falling.3 Somatic profiling in PDAC, which harbours mutations in Kirsten rat sarcoma (KRAS) in 90% of cases, identifies actionable alterations in only a small number of cases, predominately in KRAS wild-type tumours.4–8 The frequency of neurotropic tropomyosin-related kinase (NTRK) fusions in pancreatic cancer is estimated at less than 1%,9 10 with fusions considered to occur mutually exclusive of other oncogenic drivers including those of the mitogen-activated protein kinase (MAPK) signalling pathway.11
NTRK genes encode a family of genes that are integral to cell proliferation and survival.12 There are three genes, NTRK1 (located on chromosome 1q21-q22), NTRK2 (chromosome 9q22.1) and NTRK3 (chromosome 15q25) that encode TrkA, TrkB and TrkC, respectively.12–14 These genes are physiologically expressed in the central and peripheral nervous system and smooth muscle cells, and overexpressed in some cancers. While NTRK mutations were first identified in colorectal and papillary thyroid carcinoma, they are most frequently observed in secretory breast cancer, mammary analogue secretory carcinoma of the salivary glands, as well as some paediatric carcinomas.9
Oncogenic fusions occur when the 3′ region of the NTRK gene binds with the 5′ component of fusion partner genes leading to an activated and overexpressed tropomyosin-related kinase, irrespective of the fusion partner.11 13 This fusion triggers downstream oncogenic signalling pathways including PI3K-AKT, MAPK and extracellular signal-regulated kinase contributing to cellular proliferation, tumour cell survival, invasion and angiogenesis.9 15
With recent breakthroughs in efficacious treatment targeting NTRK fusions including entrectinib and larotrectinib,15 16 accurate and timely identification of NTRK fusions is paramount. Given the relative rarity of NTRK fusions in PDAC as well as there being numerous fusion partners, the current European Society of Medical Oncology (ESMO) guidelines suggest a two-step approach involving screening with immunohistochemistry (IHC) staining followed by next-generation sequencing (NGS) for cases expressing TrKA/B/C.11 17 Alternatively, upfront NGS can be performed. The advantage of IHC staining is that it is comparatively quick to obtain results and relatively inexpensive, while global access to NGS can be limited.
We sought to determine the prevalence of NTRK fusions in a large series of patients with PDAC. We further explored the clinical and molecular characteristics of patients with these fusions. We compared the sensitivity and specificity of DNA/RNA sequencing and IHC in identifying NTRK fusions in PDAC.
Materials and methods
Patient population
In this study, 400 PDAC samples were analysed from patients who underwent surgical resection of a pancreatic adenocarcinoma (n=167) between 2008 and 2015 as previously reported,18 or were enrolled in the Canadian Comprehensive Molecular Characterization of Advanced Pancreatic Ductal Adenocarcinoma for Better Treatment Selection (COMPASS; a prospective study: NCT02750657) trial (n=233).19 This trial recruited patients with either a radiological or histological diagnosis of locally advanced or metastatic PDAC. The patients consented to a fresh tumour biopsy, either from the primary tumour or from a metastatic site. Our study includes patients recruited to the COMPASS trial between December 2015 and March 2020. Biopsies and surgical specimens underwent whole genome sequencing (WGS) and RNA sequencing. All samples sequenced were enriched for tumour cells using laser capture microdissection (LCM) and processed as previously described.20 Demographics, treatment and survival outcomes were prospectively collected for both datasets. For the resected cohort of patients, results of WGS/RNAseq were not made available to the patient as this was a retrospective research study.
The study was conducted in accordance to the Declaration of Helsinki regarding clinical trials involving human subjects with additional research ethics board approval obtained from the respective institutional research ethics boards of all institutions involved in the study (University Health Network, Toronto, Ontario, Canada; MUHC Centre for Applied Ethics, Montreal, Quebec, Canada; and Queen’s University Health Services and Affiliated Teaching Hospitals Research Ethics Board, Kingston, Ontario, Canada). Each patient consented to participation in the study.
Whole genome sequencing analysis
WGS analysis involved the processing of sequenced reads as previously described18 19 21 with single nucleotide variants (SNVs) identified at the intersection of calls by Strelka22 and MuTect.23 Indels were identified by Strelka.22 Tumour cellularity, ploidy and copy number segments were called by CELLULOID.21 Structural variants (SVs) were obtained using the union of outputs from CREST24 and DELLY.25 Mutational signature proportions were calculated using a PDAC-specific subset of v2 COSMIC signatures: 1, 2, 3, 5, 6, 7, 12, 17, 18, 20 and 26.26 Tumour mutational burden (TMB) was calculated by dividing the sum of SNVs and indels over the size of the hg19 reference genome.
Fusion prediction
DNA fusion predictions were made with structural variant tools CREST and DELLY using default parameters. Somatic DNA structural variants detected by these tools with identifiable breakpoints mapped to two different genes were considered to be hypothetical fusions. RNA fusion predictions were made with ericscript and STARfusion using default parameters. These initial fusions were further refined using a minimum of three junction and one spanning fusion-supporting reads as an additional filter. Fusions predicted from DNA and RNA were combined for downstream analyses.
RNA expression
All 400 patients included had RNAseq data available. PDAC samples can contain an abundance of non-tumour cells/stroma. The relative contribution of this stroma differs between LCM-enriched and bulk sequenced samples. To account for differences in background NTRK RNA expression due to tissue of origin, we subdivided RNA sequenced samples into 227 samples from the pancreas and 173 advanced LCM-enriched samples from predominantly liver and other metastatic sites. NTRK RNA median expression was derived for these groups separately. Expression fold change values were calculated by normalising each sample’s NTRK expression against the appropriate median. We then defined samples with NTRK1/2/3 expression 10 times above the median as samples with NTRK RNA overexpression.
Fusion calls
Within our programme, we stratified candidate fusions into three tiers based on levels of evidence supporting a functional fusion. Tier 1 candidates consist of fusions predicted by at least one DNA or RNA fusion caller. Tier 2 candidates have additionally high RNA expression of the NTRK1/2/3 gene implicated in the fusion. Tier 3 candidates further express an in-frame chimeric transcript containing the NTRK kinase domain.
IHC analysis
A subset of patients’ specimens with tissue microarrays (TMAs) underwent IHC analysis. IHC staining for TrkA, TrkB and TrkC expressions was performed with the pan-Trk monoclonal antibody clone EPR17341 (Abcam, Cambridge, Massachusetts, USA) validated by Hechtman et al.27 The antibody is reactive to a homologous region of TrkA, TrkB and TrkC near the C terminus. Previous positive controls have been identified in cortical brain tissue, ganglia of the colonic plexus submucosa and testis tissue. EPR17341 was used at 6 µ/mL. All assays were performed on a Leica-Bond-3 (Leica, Buffalo Grove, Illinois, USA) automated staining platform using a heat-based antigen retrieval method and high pH buffer solution (ER2, Leica). The negative external controls used were non-neoplastic colorectal and nerve tissue.
Expression patterns of NTRK, the percentage of positive tumour cells and staining intensity were reviewed and recorded on all cases. A tumour was scored aspan-Trk IHC positive if there was any (>1+) cytoplasmic and/or nuclear staining identified in the tumour cells. The presence of nerves served as a robust internal positive control in all cases.
Sensitivity and specificity analysis
Sensitivity was determined to be the number of true NTRK-fusions (based on NGS fusion positivity with an in-frame chimeric transcript—tier 3) divided by the number of cases considered positive for a NTRK fusion by IHC. Specificity was defined as the number of true negative cases for NTRK fusions divided by the number of cases classified as negative for a NTRK fusion.
Results
In this study, 400 patients were sequenced with clinical data available between December 2008 and March 2020. The median age at sequencing was 65.3 years (range 29–87) and 56% were male. Of patients with resected PDAC, the median OS was 20.6 months (95% CI 1.8 to 140.9 months). In patients enrolled on the COMPASS trial, the median survival was 9.2 months (95% CI 0.3 to 46.1 months). The median TMB was 1.9855 muts/MB across the entire cohort, with the dominant COSMIC signatures identified signature 1 (39.55%), signature 8 (24.32%) and signature 5 (18.81%).
Predicted fusions
Twelve patients (3%) were identified as having at least one DNA/RNA breakpoint in NTRK (hypothetical fusion/tier 1 fusion prediction). Four of these 12 (1%) patients had both tier 1 and an upregulation in NTRK involved in the breakpoint (tier 2 fusion prediction), while three (0.8%) patients had tier 2 and the NTRK predicted fusion in frame with the kinase domain in the correct orientation and intact (tier 3 fusion confirmed). All patients identified as tier 3 were positive for expression of fusion RNA by RT-PCR. The three fusions identified were EML4-NTRK3 (n=2) and KANK1-NTRK3 (table 1).
Summary of NTRK fusions identified from cohort
NTRK hypothetical fusions (tier 1)
In the resectable PDAC cohort (n=167), we found 19 hypothetical fusions in 12 patients, with 3 predicted from DNA and RNA, 6 from DNA only and 10 from RNA only (table 2). The majority of the hypothetical fusions were with NTRK3 (n=12; 63%), while a smaller subset of hypothetical fusions with NTRK2 (n=5; 26%) and NTRK1 (n=2; 11%) were identified. We found one recurrent hypothetical fusion, EML4-NTRK3 (n=2); all other hypothetical fusions were only detected in one sample. In addition, of the 19 hypothetical fusions detected, 16 are novel, with only EML4-NTRK3 and PEAR1-NTRK1 28 being previously reported in other cancers.
Summary of hypothetical tier 1 NTRK fusions identified from cohort
NTRK RNA overexpression
We evaluated RNA expression two ways: (1) as an alternative independent verification for the presence of NTRK fusions (tier 2), and (2) to predetermine fusion false positives by IHC, where RNA overexpression occurs in the absence of a hypothetical fusion. A total of 46 samples were found to have high NTRK RNA expression of 10 times or more above the median. NTRK RNA overexpression was confirmed for three samples with NTRK hypothetical fusions. The remaining 43 samples had high expression in the absence of detectable fusions. RNA overexpression was mutually exclusive between NTRK1, 2 and 3.
NTRK fusion characterisation (tier 3)
We analysed the 19 hypothetical fusions with corresponding NTRK RNA overexpression for appropriate direction of transcription, preservation of NTRK kinase domain and in-frame translation of fusion protein. We found only three fusions, EML4-NTRK3 (n=2) and KANK1-NTRK3 (n=1), that met all of these criteria, which we defined as bonafide NTRK fusions. EML4-NTRK3 has been previously reported while KANK1-NTRK3 has only been described in benign renal metanephric adenomas.29 There were two samples with EML4-NTRK3 fusions, with both fusions predicted to form the same protein comprised of EML4 exons 1–6 and NTRK3 exons 13–20 (figure 1A). The N-terminal EML4 provides coil–coil domain for downstream signalling, while the C-terminal NTRK3 provides enzymatic activity through the protein kinase domain. The KANK1-NTRK3 fusion is similarly comprised of KANK1 exons 1–7 and NTRK3 exons 13–20 (figure 1B) . Interestingly, NTRK3 RNA fusion breakpoint is predicted to occur at the same location in all three samples. Tier 1 DNA fusions were associated with six false positives. Of the 400 patients analysed, 32 (8%) were KRAS wild-type (16 in the resected specimens and 16 patients with metastatic disease). The number of KRAS wild-type specimens with a predicted NTRK (tier 3) fusion was two (6.25%).
(A, B) RT-PCR validation: (A) Sanger sequencing trace of EML4-NTRK3 amplified cDNA product from patient 1 and patient 2 tumour RNA. Sequencing trace confirms expression of fusion RNA joining EML4 exon 6 to NTRK3 exon 14. The same EML4-NTRK3 RT-PCR product was detected in patients 1 and 2. A schematic of the putative fusion protein and its conserved domains is shown at the bottom. (B) Sanger sequencing trace of KANK1-NTRK3 amplified cDNA product from patient 3 tumour RNA. Sequencing trace confirms the expression of fusion RNA joining KANK1 exon 3 to NTRK3 exon 14. A schematic of the putative fusion protein and its conserved domains is shown at the bottom.
IHC analysis
IHC staining was performed in a subset of 74 resected specimens with TMAs available including two of those with predicted fusions. One case (patient 2) did not have sufficient tissue for IHC analysis. All samples also had DNA and RNA analysis. One of 74 stains was positive (patient 1, figure 2A), with DNA and RNA analysis confirming the presence of an NTRK fusion. There were no false-positive IHC stains. One stain was negative despite the presence of an NTRK fusion (patient 3, figure 2B). This is attributable to the fusion, KANK1-NTRK3, being a plasma membrane fusion. The sensitivity was therefore 50% and specificity 100%, although this must be cautioned in the presence of small numbers.
Immunohistochemical staining for patient 1 (A) and patient 3 (B) showing positive and negative staining for TrKA/B/C, respectively.
Patient characteristics
Patient 1
Patient 1 was a male in his early 80s with a resected poorly differentiated pT3N0Mx (stage IIA) PDAC. The patient received adjuvant chemoradiation with gemcitabine, with a disease-free survival (DFS) of 13.9 months and an OS of 27.0 months. The tumour transcriptional phenotype was modified Moffit classical, with DNA prediction positive (figure 3A) and a high RNA fusion fold change (FC) of 61 (figure 3B). Driver mutations identified included TP53 loss by frameshift (S17fs) with loss of heterozygosity (LOH) and CDKN2A homozygous deletion. A KRAS oncogenic mutation was not present. The tumour was considered to be proficient in homologous recombination repair (HRP) and mismatch repair (MMRp). Dominant mutational signatures were considered typical (COSMIC signature 1: 50.3%, signature 9: 23.2%, signature 5: 16.9%) (figure 4A). The tumour mutational burden was 2.6 mutations per megabase (muts/Mb). A tier 3 bonafide EML4-NTRK3 fusion was identified. Trk IHC staining was positive (figure 2A).
Patient 1: (A) DNA structural variants predicting hypothetical fusion. (B) RNA fusion reads for predicted fusion. (A) Putative EML4-NTRK3 whole genome sequencing (WGS) fusion reads viewed using IGV (Integrative Genomics Viewer from Broad Institute). WGS reads from tumour were grouped by chromosome. Reads from chr2 at the EML4 locus are shown on the left and reads from chr15 at the NTRK3 locus are shown on the right. The two reads highlighted in red are mate pairs which span the EML4-NTRK3 fusion junction. (B) Putative EML4-NTRK3 RNA fusion reads viewed using IGV. RNAseq reads were grouped by chromosome. Reads from chr2 at the EML4 locus are shown on the left and reads from chr15 at the NTRK3 locus are shown on the right. The two reads highlighted in red are mate pairs which span the EML4-NTRK3 fusion junction.
(A–C) Whole genome plot of tumour for patient 1, patient 2 and patient 3. Features displayed in the Circos plot from outermost to innermost ring are (1) the karyotype ideogram, (2) somatic base substitutions as a rainfall plot (log10 distance between mutations; C>A: blue, C>G: black, C>T: red, T>A: grey, T>A:green, T>G: pink), (3) insertions indicted by short green lines, (4) deletions indicted by short red lines, (5) copy number changes shown as blocks (gain: green, loss: red), and (6) structural variations shown as central lines (duplications: green, deletions: light red, inversions: blue, translocations: violet). NTRK and partner fusion genes are labelled around the Circos plot.
Patient 2
Patient 2 was a male in his late 50s with metastatic PDAC with synchronous liver metastases. The patient received first-line fluorouracil 2400 mg/m2 q46 hour infusion, irinotecan 150 mg/m2 and oxaliplatin 85 mg/m2 (modified FOLFIRINOX) every 2 weeks with a partial response observed (38% reduction from baseline). Treatment was discontinued due to progressive disease, with the patient then receiving a novel serine/threonine-protein kinase 4 (PLK4) inhibitor. OS was 11.8 months. The tumour transcriptional subtype was classical, with DNA prediction positive (online supplemental figure 1A) and a high RNA fusion FC of 154 (online supplemental figure 1B). Profiling identified a KRAS wild-type tumour with CDKN2A and KDM6A homozygous deletion. Dominant mutational signatures were typical (COSMIC signature 1: 29.7%, signature 17: 20.6%, signature 8: 20.2%, signature 5: 16.1%) (figure 4B) and the TMB was 4.2 muts/Mb. There was not enough tissue available for IHC testing. A EML4-NTRK3 fusion was identified; however, this was determined only after the patient had died.
Supplemental material
Patient 3
Patient 3 was a female in her early 70s with a resected poorly differentiated pT3N1Mx (stage IIB) PDAC. The patient received adjuvant chemoradiation with gemcitabine. DFS was 10.3 months, with an OS of 12.7 months. The tumour was Moffit classical, with DNA prediction positive (online supplemental figure 2A) and a high RNA fusion FC of 27 (online supplemental figure 2B). Mutational signatures were typical (COSMIC signature 1: 56.2%, signature 8: 19.8%, signature 5: 14.5%) (figure 4C) and TMB was 2.0 mut/Mb. A subclonal KRAS G12R mutation was identified, as well as a GNAS R210H mutation. Other driver mutations identified included TP53 loss by splicing variant (c.277-2A>G) with LOH, CDKN2A VUS L130P. IHC in this case was negative (figure 2B). A plasma-membrane KANK1-NTRK3 fusion was identified.
Discussion
NTRK fusions in PDAC are rare. Similar to other large datasets, the prevalence of NTRK fusions in our patient cohort was 0.8% (3/400). We document two fusions in EML4-NTRK3 in KRAS wild-type patients and a novel fusion in KANK1-NTRK3 in patients with a subclonal KRAS mutation. Aside from the absence of a KRAS mutation, mutational signatures and tumour mutational burden were similar to that of ‘typical’ KRAS-driven PDAC.
There is notably large heterogeneity in fusion partners, with the oncogenic significance and actionability of those documented in the literature unknown. Fusions previously identified in pancreatic cancer include CTRC-NTRK1 and ETV6-NTRK3.6 8 30 ETV6-NTRK3 is one of the most prevalent NTRK fusions and is present in >95% of secretory breast cancers.31 EML4-NTRK3 fusions have previously been described in numerous malignancies while KANK1 has been described as the fusion partner with NTRK2 in pilocytic astrocytoma32 and NTRK3 in benign renal metanephric adenomas. In the former case, the patient did not receive an inhibitor of TRK.
While the prevalence of NTRK fusions is low in the total population, the prevalence is much more significant when looking only at the KRAS wild-type population, which historically is considered to constitute just 10% of PDAC.7 In our population, 32 (8%) of the samples were KRAS wild type, with two of these positive for an NTRK fusion (6.25%). This strengthens the argument that cascade testing for KRAS mutations should be available for all PDAC. In the absence of a KRAS mutation, further workup could include NTRK IHC and/or expanded NGS testing to determine the presence of fusions including NTRK fusions. This consideration is particularly pertinent for health services who do not have the means or financial capacity to perform NGS on all patients with PDAC. Complicating this fact, however, as was apparent in our population, is that subclonal KRAS-mutated PDAC can also harbour an NTRK fusion. Subclonal KRAS have been reported to occur in up to 26% of KRAS-mutated lung adenocarcinomas,33 with the prevalence unknown in PDAC. This suggests that only testing for NTRK fusions based on KRAS wild type may miss a small, but definite, number of NTRK fusion-positive patients. Furthermore, in the case of this novel fusion, the responsiveness to TRK inhibitors is unknown.
Previous publications suggest IHC sensitivity for NTRK3 is 79.4% and NTRK1/2 (combined) is 96.6% with 100% specificity in PDAC.34 Our results demonstrate high specificity with lower sensitivity, although our numbers are small and results must be interpreted with caution. Despite this, it does support a high specificity for positive IHC results.
The major limitation to our study is the lack of treatment in those patients with bonafide NTRK fusions with TRK inhibitors. There are however case reports and patients with PDAC reported within trials in the literature. Specific to pancreatic cancers, two case reports were identified, one of which describes two patients with a TPR-NTRK1 fusion treated with entrectinib achieving a partial response35 and a further case report describing a CTRC-NTRK1 fusion treated with larotrectinib before switching to selitrectinib, with the best radiological response achieved being an initial partial response on larotrectinib.6 Additionally, a single patient included in a prospective single-arm basket study of lanorectinib in solid organ malignancies harboured a CTRC-NTRK1 fusion and achieved a partial response on treatment.16 While specific examples of response to TRK inhibitors in pancreatic cancer are few, the observed response in all is promising, and given the scarcity of multiple lines of effective treatments in advanced pancreatic cancers, identifying the few with NTRK fusions is paramount, as this increase the number of potential line of systemic therapy.
Conclusion
NTRK fusions in pancreatic cancer are rare and not linked to obvious unique clinical features. Reflex testing for KRAS mutations followed by RNA sequencing would provide the most accurate mechanism of in-frame fusion detection.
Take home messages
NTRK fusions in pancreatic ductal adenocarcinoma (PDAC) are rare, but targetable with novel agents.
It is therefore imperative that they are identified given the relatively few therapeutic options currently available.
The prevalence of NTRK fusions is relatively increased in KRAS wild-type tumours compared with KRAS mutant.
Reflex testing for KRAS mutations and subsequent RNA-based screening could help identify NTRK fusions cases in PDAC.
Data availability statement
No data are available. No data is available.
Ethics statements
Patient consent for publication
Ethics approval
This manuscript resulted from research approved by the University Health Network, Toronto, Ontario, Canada research ethics review board. The approval ID’s are #15-9596 and #08-0767. All participants provided consent before participating in these studies.
Acknowledgments
We thank the following individuals who provided patient samples: Gloria M. Petersen, Mayo Clinic College of Medicine, Rochester, MN, USA.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Runjan Chetty.
Correction notice This article has been corrected since it was published Online First. Author name (Jaeseung C Kim) has been spelled correctly.
Contributors Study concept and design: MJA and GMO. Data collection and assembly: MJA, AZ, PB, JCK and GMO. Data analysis and interpretation: MJA, AZ and GMO. All authors contributed to manuscript revision. Final approval of manuscript: All authors.
Funding This study was conducted with the support of the Ontario Institute for Cancer Research (PanCuRx Translational Research Initiative) through funding provided by the Government of Ontario (Grant no. I.PANC.998), the Wallace McCain Centre for Pancreatic Cancer supported by the Princess Margaret Cancer Foundation (Grant no. not supplied), the Terry Fox Research Institute (Grant no. 1078), and the Pancreatic Cancer Canada Foundation (Grant no. not supplied). The study was also supported by a charitable donation from the Canadian Friends of the Hebrew University (Alex U. Soyka) (Grant no. 800137). SG is the recipient of an Investigator Award from OICR (Award no. IA-051). We acknowledge the contributions of team members at OICR within the Diagnostic Development platform and the Genomics Program (genomics.oicr.on.ca), as well as the contributions of the University Health Network Oncology Biobank.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.