New minimal invasive diagnostic methods for early detection of lung cancer are urgently needed. It is known that the immune system responds to tumors with production of tumor-autoantibodies. Protein microarrays are a suitable highly multiplexed platform for identification of autoantibody signatures against tumor-associated antigens (TAA). These microarrays can be probed using 0.1 mg immunoglobulin G (IgG), purified from 10 µL of plasma. We used a microarray comprising recombinant proteins derived from 15,417 cDNA clones for the screening of 100 lung cancer samples, including 25 samples of each main histological entity of lung cancer, and 100 controls. Since this number of samples cannot be processed at once, the resulting data showed non-biological variances due to “batch effects”. Our aim was to evaluate quantile normalization, “distance-weighted discrimination” (DWD), and “ComBat” for their effectiveness in data pre-processing for elucidating diagnostic immune‑signatures. “ComBat” data adjustment outperformed the other methods and allowed us to identify classifiers for all lung cancer cases versus controls and small-cell, squamous cell, large-cell, and adenocarcinoma of the lung with an accuracy of 85%, 94%, 96%, 92%, and 83% (sensitivity of 0.85, 0.92, 0.96, 0.88, 0.83; specificity of 0.85, 0.96, 0.96, 0.96, 0.83), respectively. These promising data would be the basis for further validation using targeted autoantibody tests.

For the identification of antigenic protein biomarkers for rheumatoid arthritis (RA), we conducted IgG profiling on high density protein microarrays. Plasma IgG of 96 human samples (healthy controls, osteoarthritis, seropositive and seronegative RA, n = 24 each) and time-series plasma of a pristane-induced arthritis (PIA) rat model (n = 24 total) were probed on AIT’s 16k protein microarray. To investigate the analogy of underlying disease pathways, differential reactivity analysis was conducted. A total of n = 602 differentially reactive antigens (DIRAGs) at a significance cutoff of p < 0.05 were identified between seropositive and seronegative RA for the human samples. Correlation with the clinical disease activity index revealed an inverse correlation of antibodies against self-proteins found in pathways relevant for antigen presentation and immune regulation. The PIA model showed n = 1291 significant DIRAGs within acute disease. Significant DIRAGs for (I) seropositive, (II) seronegative and (III) PIA were subjected to the Reactome pathway browser which also revealed pathways relevant for antigen presentation and immune regulation; of these, seven overlapping pathways had high significance. We therefore conclude that the PIA model reflects the biological similarities of the disease pathogenesis. Our data show that protein array analysis can elucidate biological differences and pathways relevant in disease as well be a useful additional layer of omics information.