Page 32 - SRPSKO DRUŠTVO ISTRAŽIVAČA RAKA

Serbian Association for Cancer Research SDIRSACR

The isolation of extracellular vesicles is a juncture point in any diagnostic pathway because the chosen approach
determines recovery, purity, scalability, and ultimately the credibility of any subsequent biomarker claim (36, 37).
Conventional techniques such as differential ultracentrifugation remain prevalent; however, they lead to heterogenous
particle populations and co-purification of plentiful lipoproteins and protein aggregates that can mask lower-abundance,
disease-related cargo, and their inconsistent performance within laboratories—variability in g-forces, rotor k-factors,
and washing cycles—avoids comparability and reproducibility among studies (36, 40). These problems are matrix-
related in urine, where uromodulin/Tamm–Horsfall protein polymers can entrap vesicles, lower yield, and make
proteomic readouts difficult unless actively broken and removed during processing (41). With growing recognition that
method rigor reporting is as important as the biomarker signals themselves, community resources such as EV-TRACK
and its EV-METRIC have accentuated the frequency of suboptimal reporting and provided concrete checklists for
isolation parameters, particle characterization, and quality controls to enable site validation and meta-analysis (36), as
part of the broader tendency towards structured framework reporting in biomedicine to make transparency and peer
review more efficient (42).
To counter the disadvantage of bulk physical separation, many strategies of affinity-guided capture have been created
that facilitate direct improvements in selectivity, velocity, and scalability for clinical application (Table 2). Chemical affinity
capture in plate formats leverages EV surface chemistry to standardize binding and elution across large plasma cohorts,
thereby allowing for reproducible deep proteomics and discovery-to-validation pipelines for protein biomarkers with
diagnostic potential while allowing direct integration with orthogonal liquid biopsy readouts such as circulating tumor
DNA in multivariable models (37). Immunocapture takes this specificity even further via enrichment of diagnostically
significant subpopulations: CD147 recognizes an EV class that is biogenetically distinct from classic tetraspanin-positive
EVs and is selectively loaded with miRNA cargo by hnRNPA2/B1, with signal derived predominantly from cancer cells in
xenograft models; separation of circulating miRNAs by CD147 immunocapture increases detection sensitivity for tumor-
specific miRNAs and better indicates tumor miRNA signatures than conventional bulk separation(24). Simultaneously,
microarray antibodies such as the EV Array and miniaturized platforms such as ExoChip demonstrate that capture and
readout can be performed directly from unprocessed small volumes of serum or plasma with high analytical sensitivity,
thereby reducing pre-analytical handling and stabilizing turn-around time in discovery and triage settings (37, 43).
Physical and nanomaterial-facilitated selection platforms add complementary capability by combining enrichment with
functionally relevant downstream analytics (Table 2). Magnetic nanopore capture was utilized to isolate diagnostically
informative EV subsets for small-RNA sequencing and machine-learning classification; in a genetically engineered mouse
model of pancreatic ductal adenocarcinoma, this approach yielded an eleven-miRNA EV panel that classified healthy,
PanIN, and PDAC states with 88% accuracy in blinded validation, offering an early detection proof-of-concept based
on subpopulation enrichment (31). Thermophoretic assays combined with optimized filtration also readily translate
to clinical environments: an EVLET, lectin-guided thermophoretic protocol enabled fast glycan analysis of plasma EVs
and achieved 91% for the detection of triple-negative breast cancer and 96% for longitudinal therapy monitoring in a
pilot cohort, underlining the clinical value of EV surface glycomics when purification is co-designed with the resulting
readout (26).
Because the composition of biofluids has a direct impact on recovery of EVs and specificity of assays, matrix-corrected
optimization is essential for clinical-grade analysis. Disruption of networks of uromodulin polymers in expressed
prostatic secretions of urine followed by washing with alkaline condition releases entrapped vesicles and removes
co-isolated contaminants and, therefore, permits detailed proteomics analysis of EVs from prostate origin that would
otherwise be masked by matrix effects (41). Direct microarray or microfluidic immunocapture from unclarified plasma
and serum samples can minimize fractionation-induced variability as long as non-EV protein carryover and lipoprotein
contamination are evaluated with suitable negative markers and orthogonal sizing or imaging controls to verify particle
identity (36, 40). Beneath such matrix-aware strategies is the recognition that there is no single, universal isolation
strategy; rather, method choice must be explicitly matched to biofluid, analyte type, and target clinical application,
whether early diagnosis, triage, risk stratification, or monitoring (37).
New capabilities in single-particle analytics now step in as essential complements to isolation through confirmation
of enrichment, counting of heterogeneity, and enhancement of informational return per vesicle. Quantitative single-
molecule localization microscopy can measure size and biomarker density on individual particles and has demonstrated
that a pancreatic cancer–enriched EV population is present directly in patient plasma, confirming the diagnostically
exploitable potential of nanoscale phenotypic heterogeneity when correlated with appropriate capture strategies
(25). Label-free single-particle Raman spectroscopy has also shown more than 95% sensitivity and specificity in
distinguishing cancer vs. non-cancer EVs and can resolve closely related subtypes of breast cancers, allowing high-
granularity phenotyping in label-free formats that are amenable to longitudinal monitoring (29). Thermophoretic
glycan profiling provides an orthogonal surface readout that is useful both for response measurement and detection
in triple-negative breast cancer, highlighting the value of integrating surface, proteomic, and RNA cargo signals at the
single-particle or subpopulation level within a single diagnostic pathway (26).

17