Glioblastoma (GBM) is the most common and aggressive stage IV brain
cancer with a poor prognosis and survival rate. The blood–brain
barrier (BBB) in GBM prevents the entry and exit of biomarkers, limiting
its treatment options. Hence, GBM diagnosis is pivotal for timely
clinical management. Currently, there exists no clinically validated
biomarker for GBM diagnosis. T cells exhibit the potential to escape
a leaky BBB in GBM patients. These T cells infiltrating the GBM interact
with the heterogeneous population of tumor cells, display a symbiotic
interaction resulting in intertwined molecular crosstalk, and display
a GBM-associated signature while entering the peripheral circulation.
Therefore, we hypothesize that studying these distinct molecular changes
is critical to enable T cells to be a diagnostic marker for accurate
detection of GBM from patient blood. We demonstrated this by utilizing
the phenotypic and immunological landscape changes in T cells associated
with glioblastoma tumors. GBM exhibits a high level of heterogeneity
with diverse subtypes of cells within the tumor, enabling immune infiltration
and different degrees of interactions with the tumor. To accurately
detect these subtle molecular differences in T cells, we designed
an immunosensor with a high detection sensitivity and repeatability.
Hence in this study, we investigated the characteristic behavior of
T cells to establish two preclinically validated biomarkers: GBM-associated
T cells (GBMAT) and GBM stem cell-associated T cells (GSCAT). A comprehensive
investigation was conducted by mimicking the tumor microenvironment in vitro by coculturing T cells with cancer cells and cancer
stem cells to study the distinct variation in GBMAT and GSCAT. Preclinical
investigation of T cells from GBM patient blood shows similar characteristics
to our established biomarkers (GBMAT, GSCAT). Further evaluating the
relative attributes of T cells in patient blood and tissue biopsy
confirms the infiltrating ability of T cells across the BBB. A pilot
validation using a SERS-based machine learning algorithm was accomplished
by training the model with GBMAT and GSCAT as diagnostic markers.
Using GBMAT as a biomarker, we achieved a sensitivity and specificity
of 93.3% and 97.4%, respectively, whereas applying GSCAT yielded a
sensitivity and specificity of 100% and 98.7%, respectively. We also
validated this diagnostic methodology by using conventional biological
assays to study the change in expression levels of T cell surface
markers (CD4 and CD8) and cytokine levels in T cells (IL6, IL10, TNFα,
INFγ) from GBM patients. This study introduces T cells as GBM-specific
immune biomarkers to diagnose GBM using patient liquid biopsy. This
preclinical validation study presents a better translatability into
clinical reality that will enable rapid and noninvasive glioblastoma
detection from patient blood.