Page 26 - GTM-3-3

Global Translational Medicine Computational advances in cancer liquid biopsy

aneuploidy and the TF, where 3 – 5% TF is required for heterogeneity stemming from other genomic aberrations
reliable CNV analysis. Since sensitivity increases with such as SNVs and short insertions and deletions may go
increasing sequencing depth, methods are emerging to undetected. This may occur in certain hematological
pair CNV with SNV calling at an intermediate coverage cancers, often characterized by cytogenetically normal
between low-pass whole-genome sequencing (LP-WGS) genomes that do not exhibit major chromosomal
and deep sequencing to improve detection even for low- abnormalities. The application of single-cell LP-WGS,
33
burden cancers. Widman et al. recently proposed an therefore, has limited value for largely euploid malignancies.
20
ML-based approach where they integrate the denoised In such cases, alternative approaches, like deep whole
read depth information with two other sources of CNV: exome sequencing or target amplicon sequencing, could
LOH estimate (loss-of-heterozygosity perturbations where offer a better view of the underlying heterogeneity.
the B-allele frequency in plasma is known to be indicative Studying genomic variation at the single-cell level
of the presence of ctDNA in the cfDNA pool), and ctDNA allows investigators to analyze 100% pure samples and
fragment length entropy (ctDNA is associated with
shorter and more heterogenous fragment lengths than unravel cell heterogeneity, which is difficult with standard
bulk sequencing, as it averages signals over millions of cells.
normal cfDNA 21,22 ). The approach enabled the detection However, these advantages come with several single-cell-
of advanced melanoma ctDNA without the assistance of specific technical challenges: (i) Further steps are required
the matched tumor information (plasma-only) and the
detection of ctDNA shedding from precancerous, non- to isolate a cell and amplify a single-cell genome, making
invasive lesions, paving the way for WGS to complement sample preprocessing more laborious than bulk. (ii) High
targeted panels in cancer screening. 23 uniformity, low error rate, and broad coverage are three
major prerequisites of whole-genome amplification (WGA)
Like ctDNA, CTCs provide a window into cancer to accurately identify CNV in a single cell. Nevertheless,
heterogeneity and evolution. CTCs, which detach from WGA from extremely small DNA material may produce
24
the primary tumor mass, are considered the seeds of several artifacts, including failure to amplify entire
distant metastasis, and their number in the blood of a segments, GC content bias, and formation of chimeras.
35
34
cancer patient, even if strongly affected by tumor type, (iii) Amplification of exogenous DNA can occur when the
stage, and treatment, is correlated with disease prognosis. starting material is too little or poor quality, 36,37 leading
Interestingly, as with cfDNA, CTCs can be identified in to reagent waste and incorrect experimental conclusions.
patients with premalignant conditions, 25-27 and sometimes Bioinformatics tools for assessing exogenous species
may be very rare or undetectable in metastatic patients. contamination include Kraken2, Bracken, and FastQ
39
38
This is not surprising considering the challenges faced Screen. Another consideration is the tasks of tumor
40
by the very few surviving cells in circulation, including intracellular microbiota investigation, which has been
anoikis, apoptosis, filtering shear stress, and immune proven to play an important role in CTC survival and
surveillance. CTC detection power is also limited by the metastatic colonization. (iv) Noisy profiles arise from an
41
very small blood quantities that are usually analyzed, extremely low depth of sequencing. Lower signal-to-noise
the lack of sensitivity in many assays, especially marker- ratios in single-cell read count signals make CNV detection
dependent enrichment ones, which have already been more challenging compared to bulk sequencing and
proven to miss much of the heterogeneity in CTCs, 28,29 and make split-read, paired-end, or SNP density approaches
the accuracy of the tedious manual scoring, which strongly ineffective. (v) In bulk sequencing data analysis, normal
depends on the meticulousness of the operator. cells are often used to control noise, 42,43 but these are often
Nevertheless, minimally invasive and maximally unavailable in a single-cell framework. (vi) Centromeres,
informative single-cell multiomics methods have proven telomeres, repetitive regions, and poorly assembled
effective in dissecting intra and intertumor heterogeneity, regions, in general, may lead to the artificial inflation of
44
providing insight into the roles CTCs play in facilitating read counts (“bad bins”) and should be masked. (vii) The
the personalization of therapeutic options. ploidy parameter should be computationally inferred
unless previously determined experimentally. 45
Once CTCs are isolated according to their shape or
positivity to specific markers, their DNA is often amplified Among the available tools for CNV analysis, some
and sequenced by LP-WGS or ULP-WGS. Since most are designed for bulk sequencing 46,47 and are sometimes
epithelial tumors display markedly rearranged genomes, used also for CTC CNV analysis, 48-50 while others
detecting large-scale CNVs in CTC is a good and well- are specifically developed for CNV calling in single
44
accepted validation step of the isolation procedure. 30-32 cells. These include GUI-cloud-based tools, Python
On the other hand, at very low coverages, the potential packages, 51,52 and R packages. 53,54 A recently developed

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