Immunotherapy with immune check point inhibitors (CPI) has dramatically transformed cancer therapy. However, only ~25% cancer patients have a positive clinical response to CPI. Moreover, the cost of the treatment and the risk of severe side effects makes it necessary to develop biomarkers to predict the benefit of the treatment.

It has been shown that the tumor neoantigen load correlates with a positive response to treatment. This indicates that pre-existing anti-tumor immune responses to neoantigens can be used for the CPI response prediction. We have discovered a new source of frameshift (FS) neoantigens created by errors in RNA production in tumor cells, including the insertion and deletion (INDEL) of microsatellite regions during the RNA transcription and the mis-splicing of exons. These errors can generate FS neoantigens, which are highly immunogenic and can elicit both T cell and B cell immune response in cancer patients. We have shown that, although most antibody reactivity to FS peptides (FSPs) are personal, there are common antibodies reactive in different cancer patients, even across different cancer types. The FSPs with positive reactive antibodies can offer protection in mouse tumor models as vaccines.

We thus hypothesize that antibodies reactive to FSPs in cancer patients can be used for predicting the clinical benefit of cancer immunotherapy. There is a total of~ 220,000 potential FS neoantigens that can be generated by INDELs of transcription and mis-splicing of genes. These neoantigens can be represented by ~400,000 FSPs, 15-amino acids peptides. We have created arrays of by in-situ synthesis of these FSPs. We used these array to test our hypothesis with pre-treatment serum of 40 cancer patients, from 26 different cancer types in clinical trials with CPI treatments. A total of 13 patients had a clinical response to CPI treatment. Similar to ELISA, diluted serum were applied to the FSP array, and total IgG were detected by fluorescent labeled antibody. IgG reactive to each FSP was measured by the fluorescent intensity and then median normalized within each array for the analysis. As predicted, there are common IgG antibodies reactive to FSPs in the response patients. By selecting 100 to 500 most significantly different reactive FSPs between two group patients, and trained with prediction models, such as SVM, our FSP array can reach up to 96% accuracy in the prediction of clinical response with leave-one-out validation. We hypothesize that the FSPs with positive IgG reactive in response patients may be related to anti-tumor immune response, which is need to be further investigated. We also showed that the FSP array can potentially predict the patients who may have high grade immune related adverse events with the CPI treatment.

Our preliminary data indicates that the FSP array is a promising technology for predicting the clinical benefit of immunotherapy. We will expand our sample size to further evaluate this technology.

Citation Format: Luhui Shen, Jianfen Chen, Stephen Albert Johnston, David Hong, Jianjun Gao, Aung Naing. A simple blood base test for predicting clinical benefit of cancer immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2231.