Glioblastoma is the most aggressive primary brain tumor, and even with surgery, radiotherapy, and temozolomide chemotherapy, outcomes remain poor. Monitoring these patients currently depends on imaging, which is confounded by pseudoprogression and offers limited molecular insight. A new study now reports that a highly sensitive blood test, applied repeatedly to the same patients across treatment, picked up distinct shifts in plasma proteins tied to brain injury and immune signaling at each phase of standard care.

Researchers at Wake Forest University School of Medicine analyzed archived plasma samples from 11 patients with grade 4 glioblastoma, all seen at the Atrium Health Wake Forest Baptist Comprehensive Cancer Center in Winston-Salem, North Carolina. Blood was drawn at three post-surgical time points: before radiotherapy began, at a mean of 25.4 days after surgery with a range of 12 to 39 days; after completion of radiotherapy with concomitant temozolomide; and after adjuvant temozolomide treatment. For comparison, the team included plasma from 15 healthy controls without cancer. The study was approved by the institutional review board and all participants gave written informed consent.

The technology behind the measurements is the NULISAseq CNS panel, a nucleic acid linked immuno-sandwich assay run on the Argo platform. It profiles 131 proteins related to neurodegeneration, neuroinflammation, Alzheimer’s disease, and vascular health from as little as 25 microliters of plasma. Target analytes are captured by antibodies linked to DNA barcodes, the resulting complexes are pooled into a sequencing library, and roughly 400 million single-end reads are generated on an Illumina NextSeq 2000. Protein quantities are reported in NULISA protein quantification units, which are log2-transformed, so differences between samples can be read directly as log2 fold changes. The assay reached attomolar sensitivity, with a median intra-plate coefficient of variation of 4.5 percent and mean detectability of 91.7 percent across all 48 samples, meaning most targets exceeded the limit of detection in nearly every specimen.

At baseline, before any radiation or chemotherapy, the glioblastoma plasma carried a clear signature compared with controls. Neurofilament light and neurofilament heavy chain, glial fibrillary acidic protein, brain-derived neurotrophic factor, serum amyloid A1, and tissue inhibitor of metalloproteinases-3 were among the most upregulated proteins. Chemokines and cytokines including CXCL10, CXCL8, CCL3, CCL11, IL33, IL7, and IL12p70 were also higher. Neurogranin, by contrast, was strongly reduced, with annexin A5 and BASP1 marginally lower. Hierarchical clustering of the normalized values separated glioblastoma baseline samples from controls cleanly, and box plots confirmed the differences for NEFL, GFAP, BDNF, NEFH, TIMP3, SAA1, and NRGN using unpaired t-tests with Welch’s correction.

One striking finding concerned Alzheimer’s disease markers. Brain-derived phosphorylated tau 217, pTau217, brain-derived pTau181, total tau encoded by MAPT, amyloid beta 42, and amyloid beta 38 all sat at significantly higher levels in glioblastoma plasma than in controls. Because tau and amyloid markers can rise with age, the team ran age-adjusted linear regression models, and the differences held: Aβ38 at p = 0.0008, Aβ40 at p = 0.0037, MAPT at p = 0.0036, BD-pTau217 at p = 0.0007, pTau217 at p = 0.0007, and BD-pTau181 at p = 0.0063. The authors note that amyloid deposition in glioma has been reported to localize to the tumor microenvironment rather than reflecting a systemic increase, although peripheral sources such as hyperactive platelets can also contribute to circulating amyloid beta, so the relative contributions of tumor and periphery remain unresolved.

When the same patients were profiled again after radiotherapy with concomitant temozolomide, paired comparisons against their own baseline revealed modest but significant immune shifts. Interleukin-10 rose with p = 0.017, IL33 with p = 0.012, and IL5 with p = 0.047, while BDNF declined with p = 0.027 and showed a wider distribution within the post-radiotherapy group. The authors interpret the cytokine increases as indicating systemic immune activation and caution that, because radiotherapy is always given alongside temozolomide in this setting, the observed changes reflect combined chemoradiation exposure rather than radiation alone. They also acknowledge that simultaneous low-dose dexamethasone exposure may have contributed, at least in part, to the inflammatory cytokine changes.

After adjuvant temozolomide, the picture changed again. Relative to the post-radiotherapy time point, adjuvant chemotherapy was associated with decreased levels of the inflammatory cytokines IL10, IL33, IL12p70, CX3CL1, and CXCL10, and NEFL also fell compared with the radiotherapy phase. Relative to baseline, NEFL, BDNF, GFAP, TIMP3, and CXCL1 were lower after adjuvant treatment. Longitudinal line plots for individual patients showed IL10 and IL33 climbing after radiotherapy and then declining post-temozolomide, while BDNF, NEFL, NEFH, GFAP, and TIMP3 demonstrated progressive reductions from baseline through the chemotherapy phase. The authors describe this as attenuation of the radiation-induced cytokine surge and a potential normalization of injury biomarkers with chemotherapy, while stressing that these are exploratory observations in a small cohort.

Importantly, even after both treatment phases, several proteins remained elevated above non-cancer control levels. NEFL, GFAP, NEFH, IL33, TREM2, and SAA1 stayed higher after radiotherapy, and NEFL, GFAP, IL33, TREM2, and NEFH remained higher after adjuvant temozolomide, while neurogranin and huntingtin stayed lower. Amyloid beta 38 and 40 also retained elevated levels post-chemotherapy, although MAPT and pTau no longer differed significantly from controls at that point. The authors suggest that persistence of these injury markers beyond the perioperative window, given that baseline samples were collected a mean of 25.4 days after surgery, points to ongoing disease- and treatment-associated central nervous system alterations rather than residual surgical trauma alone, though they note that baseline elevations may still partly reflect resolving perioperative injury.

The study has clear limitations the authors state directly. Volcano plots and heat maps used thresholds of absolute log2 fold change of at least 0.5 and p values below 0.05 without false discovery rate correction, and multiple testing correction for age and sex was deliberately omitted at this exploratory stage to avoid excessive filtering, so the findings are presented for hypothesis generation rather than formal hypothesis testing. With 11 patients, 15 controls, and 33 longitudinal samples, the cohort is small, and the authors emphasize that the results warrant validation in independent larger cohorts. They also note that glioblastoma currently has no blood-based marker that can accurately diagnose, stage, monitor treatment response, or detect recurrence, and that interpretation of blood biomarkers is complicated by limited release of brain tumor-derived proteins across the blood-brain barrier and by tumor characteristics such as location, size, histology, and the balance of enhancing versus non-enhancing disease.

What the work demonstrates is that a targeted, low-volume plasma panel can capture dynamic biological narratives across the standard treatment trajectory: injury and inflammation at baseline, immune modulation after chemoradiation, and dampening of inflammatory signals alongside falling neuronal and glial injury proteins after adjuvant chemotherapy. The authors propose that composite plasma signatures could eventually complement imaging to distinguish therapy-responsive biology from treatment-related injury, and that pathway-anchored targets such as the CXCL8 and CXCL10 axes, IL-5 and IL-33 circuits, TREM2-linked microglial programs, and TIMP3-mediated matrix remodeling may guide biomarker-based stratification in future trials. They call for future work to expand cohort size, integrate imaging metrics, MGMT methylation status, and IDH1 profiling, and test whether composite plasma indices predict clinical endpoints such as progression-free or overall survival. For now, the message is that the blood of glioblastoma patients carries measurable, treatment-linked protein signals, and that serial, minimally invasive sampling may offer a way to watch the disease respond in real time as larger studies put these candidate markers to the test.

Subject of Research: Longitudinal plasma protein changes in glioblastoma patients across radiotherapy and temozolomide chemotherapy measured with the NULISAseq CNS panel

Article Title: Longitudinal plasma protein profiling in glioblastoma patients using NULISAseq reveals treatment-associated changes across radiation and chemotherapy

Article References: Kumar, A., He, Y., Kim, S., Hsu, F.-C., Meade, C. S., Kiraly, D. D., Strowd, R., Lesser, G. J., & Deep, G. (2026). Longitudinal plasma protein profiling in glioblastoma patients using NULISAseq reveals treatment-associated changes across radiation and chemotherapy. Journal of Neuro-Oncology, 180(2), Article 33. https://doi.org/10.1007/s11060-026-05839-2

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05839-2

Keywords: glioblastoma, plasma proteomics, NULISAseq, radiotherapy, temozolomide, biomarkers, neuroinflammation, neurofilament light, GFAP, Alzheimer's markers, cytokines, longitudinal monitoring