A non-invasive cancer diagnostic based on highly correlated regions of DNA methylation
There remains an urgent need for new diagnostic tests capable of detecting cancer earlier than existing tests to enable timely treatment before local or systemic spread has occurred. Essential characteristics of such new tests includes the need to be easily repeatable and non-invasive. This would also allow for the tumour burden to be followed in real-time enabling the assessment of the response to therapy during disease progression or after recurrence. The current gold standard in cancer diagnosis involves radiological detection followed by a tissue biopsy. Such biopsies can be painful, are invasive, can be practically impossible to do depending upon the cancer and are associated with a risk of infection. Furthermore, most tumours are genetically heterogeneous and sampling errors associated with biopsies may mean missing the full range of mutations present and this is particularly true for metastatic tumours. Finally, tumour biopsies also represent a temporal snap-shot of the tumour’s genetic composition that can change over time.
Due to the challenges stated above there has been significant research investment into finding cancer diagnostics capable of detecting tumour biomarkers in the blood and other body fluids. Most of these tests have been focused on identifying point mutations. This has proven to be technically challenging due to the fact that 1) tumours are heterogeneous in the occurrence of point mutations; 2) tumour DNA sequences may be present in body fluids at less than one genome equivalent per volume where the chances of detecting a single mutation becomes problematic; 3) normal DNA is always present in blood causing high background and; 4) error rates of PCR and sequencing limits detection to approximately 1% mutant DNA in normal DNA. To that end most studies to date have been tested on advanced metastatic disease patients were hundreds of grams of tumour exists and lack evidence to support detection of only a few grams worth of tumours.
To overcome these challenges our researchers have developed a liquid biopsy assay, called mDETECT, which leverages detection of highly correlated regions of DNA methylation using Next Generation Sequencing (NGS). In contrast to point mutations, DNA hypermethylation is a common event, and while some inter-tumoural variability is observed the large number of these events makes them much more readily detectable in the context of circulating tumour DNA. We have identified and developed a number of signatures using the mDETECT method for many tumour types including to date: various breast cancer subtypes (BC) such as triple negative (TNBC), HER2+, ER+/PR+ and signatures for prostate (PC), lung, colorectal, uveal melanoma (UM), and pancreatic cancers. To date we have validated signatures for metastatic UM, metastatic TNBC, and PC from patient blood samples with the demonstrated ability to detect very small tumours (< 0.5cm3).
It is very challenging to detect cancer at its earliest stage or during initial relapse stages within the blood. Current liquid biopsy technologies are focused on patient stratification, treatment monitoring and for personalized treatment. However, the demonstrated ability of mDETECT to detect cancer at ultralow limits of detection and accuracy make it a revolutionary step forward in cancer diagnostics.
The mDETECT method has broad applicability to many or all cancers and is performed using cell-free DNA extracted from whole blood. It is easily scalable and takes advantage of existing technologies found in many pathology laboratories unlike other diagnostic technologies in development. For example, a limited number of centres are equipped and possess the expertise to conduct diagnostic tests on circulating tumour cells.
The use of multiple probes from across the genome as well as within larger methylated regions increases the assay sensitivity. To that end we have demonstrable benefits over existing gold-standard diagnostic assays diagnostic at a very low limit of detection of < 0.03 genome equivalents. This has translated into the ability to detect tumours of very small volume < 0.5 cm3. Furthermore, the use of methylated signatures overcomes limitations in existing liquid biopsy methods in development that are focused on point mutations.
mDETECT was initially validated in uveal melanoma testing blood samples from a cohort of 40 patients with metastatic disease. The signature was found to reliably, specifically and sensitively detect tumour DNA in the blood. When compared to gold standard PCR-based test it easily outperformed it and was able to detect the presence of cancer in patients having tumours as small as 0.5 cm3 and increases in tumour sizes in the same patients could be identified.
To date mDETECT has been validated in metastatic UM and metastatic TNBC. In both UM and TNBC methylated DNA signatures were able to detect cancer with a sensitivity of 100% and a specificity of 100% from patient blood. Furthermore in UM the signature demonstrated the ability to detect < 0.5 cm3 tumours and the overall ability to detect at
mDETECT is a broad diagnostic method applicable to detecting almost any cancer at very early stages. Its non-invasiveness, repeatability and very low limit of detection will enable its use in primary diagnosis as well as for detection of recurrent cancers. These benefits also strongly support the significant potential the test method has for population-based cancer screening.