Page 82 - SRPSKO DRUŠTVO ISTRAŽIVAČA RAKA

Serbian Association for Cancer Research SDIRSACR

in vitro for 1 – 2 weeks to ensure cell attachment, growth and stabilization prior to drug testing.
Fluorescence immunoassay. An immunofluorescence assay was developed to screen patient-derived NSCLC cell cultures
for their response to chemotherapeutics and TKIs. Drug concentrations were selected based on clinically relevant
exposure levels, with the maximum plasma concentration (Cmax) as the upper limit and four lower concentrations to
capture dose-dependent effects. Automated microscopy (ImageXpress Pico, Molecular Devices, San Jose, CA, USA) and
image analysis with CellReporterXpress software were used to quantify total cell number and differentiate cancer cells
from stromal cells based on cytokeratin (CK8/18) expression. The cytotoxicity was determined using the Cell Scoring
Analysis Protocol, which evaluates changes in the viability of cancer cells under different treatment conditions.
WES. Genomic DNA was extracted from paired tumor and corresponding normal tissue samples from NSCLC patients
using the QIAamp Fast DNA Tissue Kit. Sequencing was performed by Novogene Co (Cambridge, UK). Library preparation
was performed using the SureSelect Human All Exon V6 Capture Kit, and paired-end 150 bp reads were generated on
the Illumina NovaSeq 6000 platform. At Novogene, raw sequencing reads were subjected to initial quality control
using fastp (version 0.23.1), clean reads were aligned to the human reference genome (GRCh37) using the BWA-MEM
algorithm (version 0.7.8-r455), and post-alignment processing was performed with the Picard toolkit (version 2.6.0).
The processed BAM files containing high quality aligned reads were provided to our team. Subsequent bioinformatic
analyses were performed in-house using a pipeline developed in R/Bioconductor and shell-based tools. Somatic single
nucleotide variants (SNVs) and small insertions/deletions (InDels) were identified using VarScan2 (version 2.4.3), with
tumor normal pair analysis allowing accurate discrimination between somatic events and germline polymorphisms.
For germline variant detection, the GATK HaplotypeCaller (version 4.2) was used to generate high-confidence germline
variant call sets. Structural variants (SVs) were detected using Manta, which identifies genomic rearrangements such
as translocations, inversions and large insertions/deletions. Copy number variations (CNVs) were profiled using CNVkit,
which provides genome-wide CNV profiles at exome resolution. PureCN was used to determine tumor purity and ploidy
as well as allele-specific copy number states. Variant annotation was performed with the Variant Effect Predictor (VEP)
(ensembl-vep-release-105), which integrates several annotation resources, including RefSeq, dbSNP, COSMIC, ClinVar,
1000 Genomes, NHLBI-ESP, gnomAD, SIFT, PolyPhen, and HGMD-PUBLIC. This annotation workflow provided detailed
information on the functional consequences of each variant, pathogenicity predictions and clinical relevance. Analysis
of mutational signatures was performed using the R package signature.tools.lib, which decomposes the observed
mutational spectra into known COSMIC signatures for single base substitutions (SBS), double base substitutions
(DBS) and indel signatures (ID) and detects potential novel mutational processes active in the tumors. All results were
integrated, curated and analyzed using R (v4.3) and Bioconductor. Somatic and germline variants, CNV profiles and
mutational signatures were summarized with custom scripts and visualized with tools such as maftools and circlize. The
Cell Scoring Analysis Protocol described above was combined with the WES results to identify potential correlations
between genetic alterations and drug response.
Results: To investigate the therapeutic response of NSCLC patient cells, we performed ex vivo drug testing with a
range of chemotherapeutic agents and TKIs currently used in clinical practice. The cell response of NSCLC patients was
assessed using a modified scoring system previously described [8], which integrates multiple weighted parameters
to provide an overall drug response score. This approach allowed quantitative comparison of drug effects on cancer
and stromal cells as assessed by immunofluorescence-based cell classification. The analysis revealed considerable
variability in sensitivity profiles between patients, with some drugs showing strong cytotoxic effects in certain cultures
while being ineffective in others.
To complement the ex vivo drug sensitivity testing and gain deeper insight into the molecular features that influence
treatment response, we performed WES of tumor and corresponding normal tissue in each NSCLC patient. This
genomic profiling allowed us to identify somatic mutations, germline variants, copy number alterations and mutational
signatures that may underlie the differences in therapeutic sensitivity observed in the patient-derived cultures. The
mutational landscape of individual patients varied widely, reflecting the heterogeneity of NSCLC and underscoring the
need for individualized therapeutic approaches.
The value of this approach is demonstrated in patients TR33 and TR106. In patient TR33, a smoker, whole-exome
sequencing revealed a high number of somatic SNVs. Mutational signature analysis revealed a dominant presence of
the COSMIC signature SBS4, which is associated with tobacco exposure, while no clinically actionable point mutations
were detected. However, copy number analysis revealed amplification of the HER2 gene. Despite the absence of
actionable mutations, in ex vivo assays this patient’s cells responded selectively and robustly to the EGFR inhibitor
erlotinib, while remaining largely resistant to standard chemotherapeutics. In contrast, patient TR106 had a lower total
number of somatic SNVs but had amplification of the PIK3CA gene and a PTEN missense mutation accompanied by loss
of heterozygosity. Mutational signature analysis revealed the presence of SBS2 and SBS13, which are associated with
APOBEC-mediated mutagenesis. Ex vivo drug testing showed broad sensitivity to almost all chemotherapeutic agents
tested. The molecular profile of this patient, characterized by activation of the PI3K pathway and high APOBEC activity,
may contribute to increased susceptibility to cytotoxic agents.

67