Review Papers Claim by Claim

Concise critical summary

The review systematically synthesises liquid biopsy modalities for glioblastoma including CTCs, ctDNA, nucleosomes, miRNA, extracellular vesicles, proteomics, lipidomics and AI/ML integration, and highlights the persistent barriers imposed by the blood brain barrier low biomarker abundance and lack of standardisation that currently prevent routine clinical use Primary

Paper Review detailed critique The Quest for Non-Invasive Diagnosis A Review of Liquid Biopsy in Glioblastoma

High level takeaways

• Comprehensive scope covering major liquid biopsy analytes (CTCs ctDNA nucleosomes miRNA EVs proteins lipidomics) and AI/ML approaches for integration and interpretation Scope
• Clear emphasis that cerebrospinal fluid (CSF) outperforms plasma for CNS tumours because of proximity to tumour and higher biomarker yield CSF
• Authors correctly highlight the dominant barrier is low analyte abundance because of the blood brain barrier and pre analytical/analytical variability that prevents cross study comparison and clinical translation Main

Strengths

1. Breadth and currency: the review compiles multi analyte literature across CTC ctDNA miRNA EVs nucleosomes proteomics and lipidomics and includes machine learning applications offering a timely synthesis for 2025 Breadth
2. Balanced technical evaluation: methods (ddPCR NGS WGS methylation PCR EV isolation ultracentrifugation microfluidics) and assay caveats are described rather than only reporting positive biomarker claims Methods
3. Explicit recognition of confounders including clonal hematopoiesis therapy induced changes (eg temozolomide) and postoperative effects such as nucleosome spikes following surgery which is crucial to interpret longitudinal liquid biopsy data Confounders

Limitations and critical caveats

• Predominantly narrative synthesis The review is not a systematic review or meta analysis and therefore is vulnerable to selection bias and publication bias; quantitative performance estimates across studies are limited or absent Narrative
• Heterogeneity in primary studies reduces inferential power The review aggregates studies that differ widely in sample volumes pre analytic handling EV isolation methods sequencing depth and biofluid types so direct comparison of reported sensitivity or specificity is problematic Cross
• Overemphasis risk on promising small studies Several biomarker claims especially miRNA and EV signatures derive from small cohorts sometimes n < 50 with limited external validation and perfect accuracy claims (eg report of 100% in some classifier combinations) that likely reflect overfitting and require cautious interpretation Small
• Insufficient discussion of clinical utility endpoints The review frames potential applications but stops short of rigorous discussion of how liquid biopsy would change clinical decision thresholds for surgery radiotherapy or systemic therapy or the cost effectiveness and risk benefit in routine neuro oncology care Clinical

Detailed evaluation by analyte

Circulating tumour cells CTCs

CTCs are reported as extremely rare in GBM with detection rates variable 20 40 up to 77% depending on method sample volume and timing which undermines reproducible clinical application; authors highlight need for sensitive capture methods and standardisation CTC

Circulating tumour DNA ctDNA and methylation

ctDNA in plasma is often at very low fraction due to BBB; CSF gives higher sensitivity; methylation assays including MGMT promoter methylation are clinically relevant but therapy induced changes eg temozolomide increase mutational burden complicating longitudinal interpretation ctDNA

Extracellular vesicles EVs and exosomes

EV cargo including miRNA mRNA and mutant EGFRvIII have diagnostic promise; however EV isolation methods and proteomic signatures vary and some proteomics studies failed to classify tumours reliably — stressing need for standardised EV workflows and larger cohorts EVs

MicroRNAs proteomics lipidomics nucleosomes

miRNA panels (eg miR 21 128 342 3p) show signal but frequently in small studies with variable pre analytic steps; nucleosome PTM profiling and single molecule nucleosome analysis show mechanistic promise for epigenetic signatures though still early; lipidomics and proteomic multi protein panels are promising exploratory directions that require independent validation miRNA

AI and integrative approaches

The review correctly positions ML/AI as an enabling layer to combine sparse low SNR biomarker signals into clinically meaningful classifiers and cites genome wide fragmentomics and spectroscopic liquid biopsy as promising AI enabled methods; it also appropriately notes the large high quality datasets requirement and variability in pre analytic workflows as obstacles to robust model training and external validation AI

Practical recommendations missing or underdeveloped

• Concrete pre analytical SOP recommendations (exact blood volumes anticoagulant centrifugation times storage temperatures CSF handling) are not presented as actionable standard operating procedures that would help harmonise future studies.
• No decision analytic modelling or suggested prospective trial designs are provided to show how liquid biopsy readouts could alter clinical decisions (eg re operation vs watchful waiting for suspected pseudoprogression).
• Cost and accessibility discussion is short the review notes costs but lacks concrete economic modelling or technology readiness levels for clinical labs.

Where the field should go next suggested by review and critique

1. Large prospective multicentre studies with harmonised SOPs comparing CSF and plasma multi analyte panels against gold standard pathology and imaging endpoints Need
2. Pre registered analytic plans for ML models with prespecified performance metrics and external validation cohorts to avoid overfitting and HARKing.
3. Head to head comparisons of EV isolation methods and reporting of recovery efficiency spike in controls to allow reproducible proteomic/miRNA comparisons.
4. Inclusion of therapy effect controls eg samples before during and after temozolomide and radiotherapy to characterise therapy induced genomic and epigenomic changes and to separate treatment effects from progression Therapy

Final critical appraisal and confidence

The review is a timely and useful synthesis for researchers and clinicians entering GBM liquid biopsy work because it collates multi analyte findings and flags the key biological and technical barriers that must be solved. However its narrative design and reliance on small heterogeneous primary studies limit its ability to produce strong clinical recommendations; claims of perfect classifier accuracy in small cohorts should be treated skeptically and validated prospectively. Confidence in the review conclusions is moderate reflecting appropriate emphasis on limitations and realistic translational hurdles Overall