The current cell free DNA assays are designed primarily for precision medicine. These assays, also called liquid biopsy assays, focus on detection of actionable alterations to inform treatment strategies or monitoring disease [?? 0:15]. They generally cover smaller portions of the genome suitable for detection of driver mutations or require prior tumour tissue information for targeted sequencing of individual mutations. By contrast, the ultimate goal of our efforts is the evaluation of ctDNA potentials in early detection of cancer which requires a different approach. Obviously the genomic information of tumour is not available prior to diagnosis so it is critical for the assay to be able to detect the mutations without any prior knowledge of the tumour, what we call de novo mutation coding. There is extensive tumour heterogeneity within each patient and also across the patients in the population. The assay should cover large proportions of the genome in order to increase the likelihood of finding mutations that derive from the tumour.
The concentration of ctDNA can also be very low, especially in patients with early stage cancers so it’s very critical for the assay to sequence the genome very deeply in order to increase the sensitivity for finding mutations at very low levels.
As a first step towards this goal we utilised the novel ctDNA assay that combines ultra-deep sequencing with broad genomic coverage. We call this approach a high intensity approach and we aim to determine the concordance of alterations detected in plasma cell free DNA versus detected in tissue using tissue as our reference and assess the cell-free DNA alterations detection rate by patient based on finding at least one tissue alteration in the plasma.
These are our methods. Patients with metastatic breast, lung and prostate cancer were prospectively enrolled. The patients should have de novo metastatic disease or progressive disease at the time of enrolment. Our final cohort included 124 eligible patients. Blood and tissue were collected within six weeks of each other with no intervening therapy changes. Our ultimate goal is early cancer detection but we selected specific patients with advanced disease as we thought that these are the population that are more likely to have detectable cell free DNA. In other words, this population acts as our positive control to assess the performance of this high intensity approach.
This is our workflow. I’d like to highlight that to perform this analysis for each patient we had to perform four separate sequencings to complete the cell free DNA assay and tissue assay sequence genomic profiling. The high intensity sequencing assay incorporated cell free DNA extracted from blood and also genomic DNA extracted from white blood cells from the same exact tubes. The library construction for next gen sequencing comprised molecular bar codes, also known as unique molecular identifiers for error suppression and noise reduction. The assay covers a large proportion of the genome that includes all the protein coding regions and selected non-protein coding regions of 508 genes. The panel size is 2.1 megabases which is at least ten times more than the current cell free DNA assays that are used. Selected regions of the genome were sequenced in both cell free DNA and white blood cell genomic DNA altered deeply with each position sequenced on average at least 60,000 times.
This combined approach resulted in approximately a hundred times more data compared to commonly used cell free DNA sequencing methods. A novel analytical pipeline was developed for de novo cell free DNA mutation coding, combining this sequencing data from the cell free DNA and the white cells. The rationale for sequencing the cell free DNA and the white blood cell combination approach, which is distinguishing between tumour driver mutations from mutations arising from clonal hematopoiesis is described in another poster that we will be presenting later today.
For tumour sequencing we utilised the Memorial Sloan Kettering clinically validated MSK-IMPACT assay which includes sequencing the tumour tissue and normal DNA of 410 genes to a depth of 500-1,000x. The mutation coding for ctDNA and tissue was performed independently and the data was merged for concordance analysis.
These are our main results. In 89% of the patients we were able to find at least one tissue detected mutation in cell free DNA including 38 out of 39, or 97%, of breast cancer patients, 85% of the lung cancer patients and 84% of the prostate cancer patients. The table shows the cell free DNA detection rate when we looked broadly across all types of mutations, including copy number changes and fusions, and also including the mutations that were found in the tissue in very low levels. Overall 73% of the mutations were also detected in cell free DNA without any prior knowledge of the tumour genomic profile. When we limited our analysis to actionable alterations our detection rate was 76%.
As you may know, there is extensive tumour heterogeneity within each tumour and across different tumour sites. This may affect the detectability of cell free DNA in blood as some of the mutations, mostly early events, exist in virtually all cancer cells. We call these mutations clonal mutations. In contrast, some of the other mutations are only found in a subset of cancer cells, what we call sub-clonal mutations. So we hypothesise the detection rate of cell-free DNA is associated with clonality of the mutations and, indeed, across all tumour types we found a very strong association between the fraction of the cancer cells that harbour the mutation and detection rate of that particular mutation in the plasma, as we see in these box bloods.
In conclusion, this novel high intensity sequence assay incorporates an unprecedented combination of depth and breadth of coverage compared to the previous assays. High levels of concordance for variants between plasma and tissue provides evidence for high rates of tumour DNA detection in plasma. Although we didn’t have time to present these additional results another important finding that we report is that this high intensity approach provided greater insight into tumour biology, including first exploration of mutational signatures in the plasma.
I would like to highlight that this study is part of a larger programme to evaluate high intensity sequencing approaches such as whole genome sequencing methylation patterns to characterise potential cancer defining signals in cell-free nucleic acid with the ultimate goal of hopefully enabling detection of cancer in early curable stages. Thank you.