3297 - Computational model to predict response rate of clinical trials

Background

Most therapeutic vaccine clinical trials (CTs) have failed to prove efficacy, even if immunogenicity was confirmed earlier. It was already shown that immune responses generated against multiple antigens are indicative of clinical responses. We aimed to find association between the heterogeneity of immune response and clinical efficacy and, based on this develop a tool that can predict the clinical outcome of cancer vaccines.

Methods

In an extensive literature search we collected the immune- and objective response rates (IRR, ORR) of 94 CTs in which 2,338 patients were treated with 64 different vaccines. Vaccine sequences were used to predict personal epitopes (PEPIs) that bind to at least 3 HLA alleles of the same subject, for all patients of a representative model population. Then we determined the percentage of subjects with at least 1 vaccine specific PEPIs (PEPI Score) and at least 2 vaccine specific PEPIs derived from different antigens (MultiAg PEPI-Score) and compared to the published IRR and ORR.

Results

PEPI Score was able to predict the immunogenicity of therapeutic vaccines; significant correlation was found with IRR (p = 0.002). As expected, no correlation was found between ORR and IRR (p = 0.294), neither between PEPI Score and ORR (p = 0.302), suggesting, that immune response against a single epitope is not enough for efficient tumor response. However, we found that MultiAg PEPI-Score significantly correlates with ORR (p < 0.0001) consistent with earlier findings that the targeting of multiple antigens is required for tumor shrinkage.

Conclusions

Our results demonstrate that both IRR and ORR can be predicted by PEPIs. For clinical efficacy it is crucial to target and generate immune response against multiple antigens. Based on our analysis our computational model is useful and accurate for the prediction of the clinical outcome of cancer vaccines and can even be suitable for rescuing CTs with insufficient or missing responder selection.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

O. Lorincz: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. J. Toth: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. M. Megyesi: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. K. Pántya: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. E. Somogyi: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. Z. Csiszovszki: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. P. Pales: Full / Part-time employment: Treos Bio Zrt. L. Molnar: Shareholder / Stockholder / Stock options: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. E.R. Tőke: Shareholder / Stockholder / Stock options, Officer / Board of Directors: Treos Bio Ltd; Full / Part-time employment: Treos Bio Zrt. All other authors have declared no conflicts of interest.