Page 25 - GTM-3-3

Global Translational Medicine Computational advances in cancer liquid biopsy

allows for real-time evaluation of the evolving tumor low fraction of DNA may occur for several reasons. Both
landscape using a sample of body fluid – typically blood, normal DNA, mostly originating from the death of normal
but also including urine, saliva, cerebrospinal fluid, blood cells, and tumor DNA, reflecting both primary and
sputum, breast milk, pleural effusion, bile, and tears. By metastatic tumor biology, circulate in the plasma, with the
7
providing a comprehensive picture of tumor heterogeneity, shedding of tumor DNA into the bloodstream varying
it is especially useful for monitoring tumor sites that are over time. As a result, a typical cell-free DNA (cfDNA)
difficult to reach with traditional biopsies and for detecting biospecimen contains multiple confounding sources of
undetected sites, regardless of the disease stage. 3,8 DNA, including clonal alterations, subclonal mutations,
The great advances in sequencing technologies, clonal hematopoietic aberrations, as well as other sources
accompanied by the decrease in sequencing costs, of DNA such as germline, fetal, post-organ transplant,
12
are allowing the comprehensive characterization of concomitant malignancies, and environmental and
increasingly low amounts of cancer specimens, such as reagents contaminants. The physical fragment that supports
single circulating tumor cells (CTCs) or circulating tumor an important single nucleotide variant (SNV) may not
DNA (ctDNA), through several “omics” approaches. As a be presented in the collected plasma sample, sequencing
9
result, many computational methods are rapidly arising to depth may not be sufficient to detect the variant, or the
handle the massive volume of data generated by these high- sequencing error rate may obscure true events. The ability
throughput techniques, enabling researchers to investigate to detect somatic mutations in ctDNA depends on both
the genetic, epigenetic, and proteomic alterations assay sensitivity and the fraction of circulating DNA that
associated with cancer progression and treatment response. is tumor-derived (tumor fraction [TF]), which is directly
These computational approaches encompass a wide range related to disease burden.
of disciplines, including bioinformatics, machine learning Achieving high sensitivity and specificity is both
(ML), and statistical modeling, and play a pivotal role in challenging and crucial for accurate diagnosis and
deciphering the complex molecular profiles of individual monitoring. Hence, most developments to date have focused
cancer cells, thereby advancing our understanding of on metastatic high tumor-burden disease where cfDNA TF
the underlying mechanisms driving tumorigenesis and is high, while the major areas of future promise for cfDNA-
13
facilitating the development of personalized therapeutic based cancer studies are the detection of early-stage cancers
strategies. and post-surgery minimal residual disease (MRD) to guide
Integrating and analyzing these heterogeneous data clinical interventions in cases where ctDNA exists in very
to extract meaningful insights and ensure proper result low concentrations. 14,15 To enable mutation detection of low
interpretation is a multifaceted endeavor that demands TF cfDNA, such as when the goal is to evaluate the post-
advanced specific and multidisciplinary competencies, operative residual disease, the most obvious solution seems
computational techniques, robust statistical methods, to focus on a limited set of recurrently mutated variants of
and innovative data visualization approaches. Despite the interest using custom gene panels and increasing the depth
progress in the field, numerous challenges continue to of sequencing. 16,17 Ultra-deep sequencing, which guarantees
hinder the incorporation of liquid biopsy into standard high-accuracy detection, is also the best approach for
clinical practice. These challenges include technical selecting targeted therapy candidates in a personalized
limitations, the need for new protocol validation and medicine framework. However, it has been demonstrated
standardization, the requirement for robust clinical trials that patients with radiographically manifest disease
to demonstrate liquid biopsy reliability and clinical utility, sometimes do not show detectable ctDNA by deep targeted
physicians’ skepticism, cost and accessibility issues, and sequencing, 17,18 as sensitivity issues may occur when ctDNA
ethical and legal considerations. fragments are under-sampled or when certain mutations
are not included in the panel.
2. Genomics IchorCNA is a popular tool used to estimate a plasma
Genomics is the most established field in liquid biopsy, sample TF in an ultra-low-pass whole-genome sequencing
having been used for a longer period and more extensively (ULP-WGS) scenario (×0.1). It must be noted that while
19
developed and validated than other omics. Effective IchorCNA copy number variation (CNV) analysis is
19
mutational characterization of ctDNA, purified from effective for identifying large CNVs, it can miss smaller
the plasma of cancer patients and processed via next- events (e.g., <1 Mb) due to sparse coverage affecting
generation sequencing (NGS), is now a reality. 10,11 CNV resolution. Detection of CNVs is often higher in
However, dynamic tumor burden tracking is not an easy samples from patients with late-stage tumors compared
task, and missing critical genetic alterations present in a to early-stage patients, and it is affected by the degree of

Volume 3 Issue 3 (2024) 2 doi: 10.36922/gtm.3063