Page 90 - Read Online

D’Amico et al. J Cancer Metastasis Treat 2021;7:3 I http://dx.doi.org/10.20517/2394-4722.2020.93 Page 5 of 16

DNA methylation profiles were generated through reduced representation bisulfite sequencing (RRBS),
demonstrating high analytical sensitivity in detecting cancer-derived DNA. Further analyses suggested
the ability of cfMeDIP-seq in detecting tumor-specific methylation events in ctDNA and highlighted that
cfMeDIP-seq methylomes could identify active transcriptional networks in tumors or other tissues through
plasma cfDNA. CfMeDIP-seq was then explored in a discovery cohort of 189 plasma samples from healthy
controls and patients with seven tumor types. Interestingly, this approach demonstrated a high and cost-
effective performance in detecting and classifying tumors by recovering ctDNA-associated methylation
profiles and it certainly requires further investigation in larger cohorts of eBC patients.

The performance of cfDNA-based targeted methylation sequencing was also evaluated in a prospective
[20]
case-control sub-study conducted by Liu et al. , that assessed the accuracy of a panel of > 10,000
methylation regions to detect and localize a broad range of cancer types in 6,689 participants. In the
validation set, sensitivity was 18% among all cancer types, with detection increasing with higher stages, and
the specificity was 99.3%. TOO prediction was significantly accurate. This targeted methylation approach
allowed deeper sequencing of informative sites compared with whole genome sequencing (WGS).

[21]
Cristiano et al. developed a diagnostic tool to investigate genome-wide cfDNA fragmentation patterns
of 236 breast, colorectal, lung, ovarian, pancreatic, gastric, or bile duct cancer patients and 245 healthy
controls. This study provided further proof-of-concept for plasma-based screening and early cancer
detection, as cfDNA altered fragmentation profiles were detected in cancer patients compared to
healthy subjects, whose cfDNA rather reflected nucleosomal patterns of white blood cells. In addition,
fragmentation profiles were used to confine the TOO to a limited number of sites in 75% of cases.

Besides early diagnosis, plasma-based gene expression profiling allows for the characterization of
the genomics, epigenomics, and proteomics of BC patients and ctDNA-based analyses might lead to
the identification of targetable mutations at baseline, laying the foundation for very-early precision
medicine [22,23] .

Granting the value of these preliminary thrilling results, some issues need to be addressed. An accurate
blood-based cancer detection test would need a high degree of sensitivity to detect very early-stage
disease in a target, asymptomatic, but high-risk population. Notably, most of the studies have evaluated a
population of diagnosed but early-stage cancer patients so far. In fact, sensitivity can be improved by deeper
sequencing, enhanced error correction methods (including molecular barcoding technology), increased
blood amounts and longitudinal evaluation through multiple liquid biopsies. Considering that spontaneous
mutation incident rates in healthy subjects may be influenced by multiple non-cancer related biological
factors, in addition to sensitivity, specificity is crucial to minimize the risk of high-false positive rates and
subsequent unnecessary follow-up tests and to eventually guide the diagnostic workup. Given the impact of
early-stage BC diagnosis on cancer-specific survival, the implementation of cfDNA-based cancer detection
tools undoubtedly needs further validation in long-term prospective trials to establish its impact on clinical
outcomes.

ctDNA DCTECTION AT THE TIME OF PRIMARY DIAGNOSIS
The role of ctDNA detection after primary diagnosis was assessed at different timepoints with respect to
[24]
surgery, neoadjuvant, and adjuvant systemic treatments. In a pioneering prospective trial from Silva et al.
(2002), a correlation between the presence of at least 1 out of 6 different chromosomal region’s loss of
heterozygosis (LOH), commonly represented among breast carcinomas, or a mutation in TP53 before
mastectomy was a predictive factor of disease free survival. Despite the significant results, only 76% of
relapsed patients were ctDNA positive. The limited number of alterations explored and the technologies
available at that time may explain the low sensitivity.