Discovery of novel glioma serum biomarkers by proximity extension assay

Abstract

Background Gliomas are among the most malignant tumors, with a very poor prognosis. Early diagnosis is highly desirable since it can help implement more effective treatments for smaller tumors, which have not yet extensively metastasized. Improving early diagnosis may facilitate access of patients to clinical trials and prepare them for the future availability of new disease-modifying treatments. Methods We analyzed retrospective samples collected at diagnosis (before therapy initiation), with PEA (Olink Proteomics), quantifying about 3000 proteins. We utilized 30 plasmas from gliomas (20 glioblastomas, 5 anaplastic astrocytomas, 5 anaplastic oligodendrogliomas) and 20 meningiomas (as controls). We then analyzed the data to identify proteins which either alone, or in combination, could discriminate gliomas from meningiomas, or correlate with clinical and molecular alterations. Results We identified 8 plasma proteins which were increased in gliomas vs. meningiomas (GFAP, NEFL, EDDM3B, PROK1, MMP3, CTRL, GP2, SPINT3) and 4 proteins which were decreased in gliomas vs. meningiomas (FABP4, ALDH3A1, IL-12B and OXT). Partition algorithms and logistic regression algorithms with two biomarkers (GFAP and FABP4) achieved sensitivity of 83% and 93% at 100% and 90% specificity, respectively. The strongest single marker was GFAP with an area under the ROC curve (AUC) of 0.86. The AUC for the GFAP-FABP4 combination was 0.98. Conclusion PEA is a powerful new proteomic technology for biomarker discovery. GFAP and a handful of other plasma biomarkers may be useful for early glioma detection and probably, prognosis. Statement Detecting gliomas as early as possible is highly desirable since it can significantly improve the chances of effective treatments. Reliable glioma biomarkers can timely inform glioma patients about the efficacy of their prescribed treatment. Our results reveal some novel putative glioma markers that may prove valuable, when used alone or in combination, towards improved clinical care of gliomas. In order to better appreciate the potential usefulness of these markers, their performance needs to be further validated in a larger cohort of samples.

Glial fibrillary acidic protein, neurofilament light, matrix metalloprotease 3 and fatty acid binding protein 4 as non-invasive brain tumor biomarkers

“…Validation of data of previous ndings: We replicated our original ndings (22) (Olink data) with the present data (immunoassay) to identify gliomas and meningiomas using FABP4 and GFAP levels utilizing 1) thresholds that achieve 100% speci city, 2) using logistic regression. Because the immunoassay and Olink values do not use the same analytical units, we identi ed method-speci c thresholds.…”

“…Comparison of Olink and immunoassay data: The original Olink assay data for the four candidate biomarkers of interest, presented in our previous manuscript (22), were compared with the Meso Scale Discovery® (MSD) immunoassay data for 50 samples with both data available. Scatterplots were drawn, comparing the protein concentrations between the two methods.…”

“…Since our series of patients represent a retrospective cohort, glioma patients were categorized by using the old and the new WHO classi cation system, as appropriate (23). We used these classi cation systems to compare the current data with our previous discovery data (22). According to the old WHO classi cation system, the following categories were included: (N = number of samples in brackets): Glioblastoma multiforme (astrocytoma Grade 4; N = 71), astrocytoma grade 2 (N = 13), astrocytoma grade 3 (N = 19), oligodendroglioma, grade 2 (N = 10), oligodendroglioma, grade 3 (N = 13), pituitary adenoma (N = 15), Schwannoma, grade 1 (N = 15) meningioma, grade 1 (N = 45), meningioma, grade 2 (N = 25), metastatic adenocarcinoma to CNS (with primary sites in brackets) (colorectal; N = 5) (breast; N = 10), (lung; N = 10), (melanoma; N = 10), primary brain lymphoma (N = 9), secondary brain lymphoma (N = 7) and normal individuals (N = 30).…”

“…Despite the current plethora of candidate glioma non-invasive biomarkers, very few, if any, are used routinely, because of their low clinical sensitivity and speci city, high cost and the lack of evidence that they can contribute to improved patient survival or quality of life. In our previous work, we have undertaken the task of using new technology, the proximity extension assay (PEA) (21), to quantitatively pro le for the rst time ~ 3,000 proteins in serum of glioma and other brain malignancies to con rm (GFAP, NEFL) and discover additional glioma biomarkers, which, individually or as a small panel, could assist in glioma patient management (22). We hypothesized that a small panel of non-invasive serum/plasma biomarkers can aid in the optimal management of patients with brain tumors, in combination with current imaging modalities (CT, MRI).…”

“…In this paper, we used a different analytical technology (ultrasensitive electroluminescent immunoassay instead of PEA) and an independent, more diverse, and larger cohort of patients, to validate GFAP, NEFL, matrix metalloprotease 3 (MMP3) and fatty acid-binding protein 4 (FABP4), as non-invasive biomarkers of benign and malignant primary and secondary brain tumors. More speci cally, we sought to validate our original ndings (22), identify additional associations between plasma biomarker levels and molecular characteristics, explore the possibility of differential diagnosis based on the four biomarkers of interest, establish an association between clinical outcomes and biomarker levels (including overall survival), and explore the possibility of differential diagnosis between GBM and primary and secondary CNS lymphomas.…”