gms | German Medical Science

Objective: A limited number of molecular biomarkers are already successfully used in clinical glioblastoma diagnostics including IDH1 mutation and MGMT hypermethylation. However, they can only be determined by analyzing tissue specimens. Consequently, we need non-invasive prognostic biomarkers that support pre-operative treatment choices. We therefore aimed at screening for prognostic serum antibodies against tumor-associated antigens (TAAs).

Method: For this purpose, we designed customized printed peptide arrays covering the complete linear amino acid sequence of six known or newly discovered glioblastoma-associated antigens (EGFR, FABP5, GLEA2, MAGEA3, PHF3, TNC). 13mer-peptides were printed as duplicates with an overlap of 9 amino acids. In order to identify the most prognostic of these 1745 peptides, we hybridized serum of 10 long-term surviving (LTS) and 14 short-term surviving (STS) GBM patients to the arrays. For validation of antibody responses, a multi-institutional serum study sample of 138 GBM patients was used.

Results: Polyclonal antibody responses were observed against TAA epitopes of LTS and STS patients. Statistical analysis revealed serum antibodies discriminating between LTS and STS patients (p<0.05) against 70 peptides. These were consequently printed together as a screening peptide array for use in a bigger study sample (70-peptide chip). A validation study on sera from 138 primary GBMs was used to assess the prognostic value of the 70-peptide chip. Employing a median cutoff for group determination in the survival analysis, multivariate analysis identified 3 prognostic antibody responses (p<0.05) against epitopes of PHF3 and GLEA2 predicting an improved survival. In addition, the prognostic performance, determined by means of receiver operating characteristics (ROC), was pronounced (AUC of 0.82, 0.79 and 0.67). To further elaborate on our finding, we used an alternative peptide array platform for technical validation. The top 30 peptides were pre-synthesized, HPLC purified and subsequently spotted on a glass surface via a PEG linker. By these means, 2 of 3 identified peptides again were revealed to be prognostic (p<0.05) using a median raw intensity cutoff.

Conclusions: Altogether our analysis identified 2 distinct epitopes on PHF3 and GLEA2 as powerful prognostic serum markers to predict better survival in glioblastoma patients. Furthermore, printed peptide microarrays may be a powerful tool for non-invasive biomarker discovery.