Page 95 - Read Online

Page 10 of 16 D’Amico et al. J Cancer Metastasis Treat 2021;7:3 I http://dx.doi.org/10.20517/2394-4722.2020.93

response. Although only a few trials showed that increasing number of CTCs enumerated was correlated
with a higher hazard ratio for disease recurrence [29,38] , in the former meta-analysis each CTC detected
added a quantum of poor prognosis, suggesting that CTCs can be considered as a quantitative marker even
[50]
in eBC .
CAVEATS AND PITFALLS OF ctDNA IN EARLY-STAGE BREAST CANCER
Notwithstanding the potential of liquid biopsy in the metastatic setting, its deployment for early disease
[51]
diagnosis, MRD monitoring and characterization still poses multiple challenges .
As a matter of fact, ctDNA is directly associated with tumor burden, with consequently lower MAF
in early-stage, which is moreover characterized by a significantly low number of ctDNA-detectable
aberrations. In addition, low levels of ctDNA could result in low reproducibility, as high pre-analytical and
analytical variables could influence the characterization of alterations in situations with a near to the limit
of detection MAF [52,53] .

Besides tumor burden, low ctDNA levels could also be linked to the differential shedding across metastatic
[54]
sites. In a study of Serpas et al. , knocking-down the DNase1L3 in mice resulted in a change in DNA
fragment length, suggesting that different nuclease mechanisms among different tissues could provide more
information about the origin of the ctDNA, allowing increased specificity. However, Garcia-Murillas et al.
[27]
showed that single site relapses were characterized by low, undetectable, ctDNA levels and in particular
brain-only relapses were unlikely to be detected with rates of ctDNA detection similar to those of primary
brain tumors, probably due to the role of the blood-brain-barrier in hindering ctDNA shedding into the
plasma.

All these assumptions translate in the need to develop more sensible techniques, capable of detecting lower
and lower ctDNA concentrations.

A first layer of selection could be made by fragment size-based enrichment and selective sequencing that
[55]
can increase ctDNA detection and could therefore enhance downstream characterizations .

Additional mutation-agnostic ctDNA features, such as epigenetics, could potentially increase not only
sensitivity, but also specificity [56,57] , which is of pivotal importance in the early setting to avoid false
negatives with their consequent impact on healthcare costs and patients’ quality of life.

As a matter of fact, the increasingly high sensitivity of sequencing technologies can detect, as a side-effect,
somatic mutations that are present across normal tissues. The onset of such confounding somatic mutations
may depend on tissue-specific factors, exposure to mutagens as chemotherapy and age . These clones may
[58]
[59]
consequently result in a genetic drift that could potentially cause false-positive results in ctDNA analysis .
In particular, the clonal hematopoiesis of indeterminate potential (CHIP) is common with increasing age [60-62] .
Garcia-Murillas et al. prospectively assessed the detection of CHIP during MRD monitoring, showing
[27]
how persistently high levels of truncal mutations such as TP53 or PIK3CA could be detected also in the
Buffy coat as a result of CHIP and would otherwise generate false-positive ctDNA results.

Thus, breaking down the barrier of low tumor burden detection through ctDNA analysis, an increasing
need to develop highly sensitive and specific techniques has arisen. Several approaches have been
previously attempted to combine the two philosophies by applying wide targeted NGS panels or exome
sequencing to the primary tumor and consequently screen the resulting mutations on longitudinal ctDNA
samples through personalized ddPCR primers [27,32] . Such techniques include BEAMing, SafeSeqS, and
TAmSeq [63-65] .