Glioblastoma (GBM) is a lethal and aggressive form of brain cancer, with a median survival of only 14.6 months. GBM accounts for 15-20% of brain cancer-related deaths, underscoring the critical need for an early intervention to improve patient prognosis. While various diagnostic and prognostic tools exist – gene expression profiling, circulating tumor cell (CTC) detection, and patient-derived xenografts (PDX) – the high costs, long processing times, and specificity and sensitivity of these methods limit their clinical utility. Furthermore, given the rapid progress of the disease and the short median survival span of GBM patients, there’s a need for a faster, inexpensive, and more accurate method to guide patient prognosis and treatment.
Methods
We developed a microfluidic assay for the quantitative assessment of cell invasion, termed MAqCI (Microfluidic Assay for Quantification of Cell Invasion). MAqCI is a Y-shaped PDMS-based microfluidic device featuring a 20 μm-wide feeder channel that bifurcates into two narrower branches, 3 μm and 10 μm in width. Because migration, deformability, and proliferation are key hallmarks of metastasis, primary glioblastoma (GBM) cells are introduced into the device to simultaneously evaluate these behaviors. We quantify three metrics: (1) the percentage of cells that enter one of the narrower branch channels (indicative of high motility and deformability), (2) the percentage of cells entering the 3 μm-narrow branch channel (narrow entry), and (3) the proportion of highly motile cells that are Ki67-positive (proliferative). These values are integrated into a normalized MAqCI score.
Through optimization with primary patient-derived GBM samples, we established a prognostic threshold: a MAqCI score >0.6 correlates with short-term survival (<14.6 months), whereas a score <0.6 predicts long-term survival (>14.6 months).
Results
Previously, our lab validated the prognostic utility of the MAqCI assay using 28 primary patient-derived GBM samples, demonstrating a predictive accuracy of ~86% for patient survival outcomes. In the current study, we expanded our cohort to include an additional 32 GBM patient samples. Using the MAqCI platform, we quantified three key cellular behaviors: highly motile cells, cells capable of entering the narrow 3 μm channels (narrow entry), and highly motile Ki67⁺ cells. Incorporation of these data into the MAqCI scoring algorithm yielded a similar predictive accuracy of ≥90%.
To investigate potential molecular drivers of invasive behavior, we performed transcriptomic analysis via RNA sequencing (RNA-seq), which revealed significant upregulation of the adrenomedullin (ADM) gene in highly motile cells. In view of this finding, we evaluated the therapeutic potential of targeting ADM using different concentrations of the ADM inhibitor, paroxetine (PTX). PTX is an FDA-approved selective serotonin reuptake inhibitor (SSRI) that has been demonstrated to have potential therapeutic value in conditions such as panic and major depressive disorder. Notably, treatment with 1 μM PTX reduced the MAqCI score below the 0.6 threshold in three short-term survivor–derived GBM lines, effectively reclassifying them as long-term survivors. These results suggest that PTX may also have therapeutic potential in mitigating invasive behavior in GBM and improving clinical outcomes.
Conclusion
In conclusion, our novel MAqCI platform provides a cost-effective and efficient approach for accurately predicting survival outcomes in GBM patients. Beyond its prognostic power, we demonstrate that MAqCI also functions as a robust tool for evaluating therapeutic responses, highlighted here through the assessment of SSRI efficacy. These findings underscore the dual utility of MAqCI as both a predictive and functional assay. With its ability to inform patient-specific invasive potential and drug responsiveness, MAqCI holds significant promise for clinical translation as a diagnostic and therapeutic platform to guide personalized treatment strategies in GBM.