Machine Learning-Assisted Evaluation of Circulating DNA Quantitative Analysis for Cancer Screening

While the utility of circulating cell-free DNA (cfDNA) in cancer screening and early detection have recently been investigated by testing genetic and epigenetic alterations, here, an original approach by examining cfDNA quantitative and structural features is developed. First, the potential of cfDNA quantitative and structural parameters is independently demonstrated in cell culture, murine, and human plasma models. Subsequently, these variables are evaluated in a large retrospective cohort of 289 healthy individuals and 983 patients with various cancer types; after age resampling, this evaluation is done independently and the variables are combined using a machine learning approach. Implementation of a decision tree prediction model for the detection and classification of healthy and cancer patients shows unprecedented performance for 0, I, and II colorectal cancer stages (specificity, 0.89 and sensitivity, 0.72). Consequently, the methodological proof of concept of using both quantitative and structural biomarkers, and classification with a machine learning method are highlighted, as an efficient strategy for cancer screening. It is foreseen that the classification rate may even be improved by the addition of such biomarkers to fragmentomics, methylation, or the detection of genetic alterations. The optimization of such a multianalyte strategy with this machine learning method is therefore warranted.

Recent advances in circulating nucleic acids in oncology

Circulating cell-free DNA (cfDNA) is one of the fastest growing and most exciting areas in oncology in recent years. Its potential clinical uses cover now each phase of cancer patient management care (predictive information, detection of the minimal residual disease, early detection of resistance, treatment monitoring, recurrence surveillance, and cancer early detection/screening). This review relates the recent advances in the application of circulating DNA or RNA in oncology building on unpublished or initial findings/work presented at the 10th international symposium on circulating nucleic acids in plasma and serum held in Montpellier from the 20th to the 22nd of September 2017. This year, presenters revealed their latest data and crucial observations notably in relation to (i) the circulating cell-free (cfDNA) structure and implications regarding their optimal detection; (ii) their role in the metastatic or immunological processes; (iii) evaluation of miRNA panels for cancer patient follow up; (iv) the detection of the minimal residual disease; (v) the evaluation of a screening tests for cancer using cfDNA analysis; and (vi) elements of preanalytical guidelines. This work reviews the recent progresses in the field brought to light in the meeting, as well as in the most important reports from the literature, past and present. It proposes a broader picture of the basic research and its potential, and of the implementation and current challenges in the use of circulating nucleic acids in oncology.

Hypoxia differently modulates the release of mitochondrial and nuclear DNA

BACKGROUND

We investigated the influence of hypoxia on the concentration of mitochondrial and nuclear cell-free DNA (McfDNA and NcfDNA, respectively).

METHOD

By an ultra-sensitive quantitative PCR-based assay, McfDNA and NcfDNA were measured in the supernatants of different colorectal cell lines, and in the plasma of C57/Bl6 mice engrafted with TC1 tumour cells, in normoxic or hypoxic conditions.

RESULTS

Our data when setting cell culture conditions highlighted the higher stability of McfDNA as compared to NcfDNA and revealed that cancer cells released amounts of nuclear DNA equivalent to the mass of a chromosome over a 6-h duration of incubation. In cell model, hypoxia induced a great increase in NcfDNA and McfDNA concentrations within the first 24 h. After this period, cfDNA total concentrations remained stable in hypoxia consecutive to a decrease of nuclear DNA release, and noteworthy, to a complete inhibition of daily mitochondrial DNA release. In TC1-engrafted mice submitted to intermittent hypoxia, plasma NcfDNA levels are much higher than in mice bred in normoxia, unlike plasma McfDNA concentration that is not impacted by hypoxia.

CONCLUSION

This study suggests that hypoxia negatively modulates nuclear and, particularly, mitochondrial DNA releases in long-term hypoxia, and revealed that the underlying mechanisms are differently regulated.

Towards a screening test for cancer by circulating DNA analysis

e13146

Background: Following works on the clinical potentials of circulating DNA (cfDNA), our group is now particularly focused on evaluating its potential for early detection of cancer. We recently developed a test (MNR: Multi Normalized Ratio), based on various cfDNA parameters determined by a specific q-PCR based method, targeting both nuclear and mitochondrial sequences. When applied to the supernatant of cell culture, the MNR had a discriminative potential of 100% between normal and cancer cell lines. Methods: We recently developed a screening test (MNR: Multi Normalized Ratio), based on various cfDNA parameters determined by a specific q-PCR based method, targeting both nuclear and mitochondrial sequences. When applied to the supernatant of cell culture, the MNR had a discriminative potential of 100% between normal and cancer cell lines. Results: We are currently carrying out an extensive evaluation of this test in plasma samples from diagnosed and healthy individuals. Preliminary results on the plasma of 289 healthy subjects and 1,200 patients with various types of cancer (colorectal, breast, lung, liver, pancreatic, ovarian) of all stages, revealed a high potential with an AUC of 0.82 (0.78-0.84, 95% confidence interval, CI) and 70% sens. with 78% spe.. In breast cancer (N = 169), an AUC of 0.82 (0.78-0.86, 95% CI) with 72% sens. and 80% spe. were observed. In all stages CRC patients (N = 887), the results showed an AUC of 0.81 (0.78-0.84, 95%, CI) and 68% sens. with 80% spe.; for CRC stages 0/I/II (N = 249), an AUC of 0.75 (0.71-0.80, 95% CI) and 60% sens. with 79% spe.; and for CRC stage IV (N = 101), a 0.86 AUC (0.82-0.91, 95% CI) with 81% sens. and 78% spe.. When combining the MNR test to a threshold value of the total concentration of cfDNA (AUC = 0.82 (0.79-0.84, 95% CI), 70% sens. and 80% spe. in all stage cancers (N = 1078)), in CRC stages 0/I/II (N = 249) vs Healthy Individuals (N = 289), we observed a 63% sens. with 84% spe. Conclusions: Furthermore we recently discovered that cfDNA fragmentation, as determined by Whole Genome Sequencing using either double or single strand library, is also a parameter enabling discrimination between healthy and cancer individuals. Data on a multi-parametric test combining total cfDNA quantification, MNR and fragmentation biomarkers with help of a ongoing decision tree algorithm in machine learning will be presented during the meeting. Our on-going data suggest that our strategy in targeting cfDNA quantitative/structural features might be powerful for cancer screening/early detection and appears as an alternative or a synergistic combination to searching mutations.