III. CANCER NEOANTIGEN AND CANCER VACCINE

Background and Rationale

Development of most adult cancers is initiated by genetic and/or epigenetic alterations in so-called “gatekeeper” genes. As just one model example, over 80% of all human colorectal cancers (CRC) are triggered by inactivation of a specific tumor suppressor gene (Adenomatous Polyposis Coli (APC)) in an intestinal stem cell (1). Over time, benign tumors arising from functional loss of both APC alleles then progress to malignancy by a Darwinian process of clonal cellular evolution involving accumulation of additional mutations in other key genes. Early studies highlighted the subsequent involvement of several additional so-called “driver” genes that are frequently mutated during progression to malignancy, including K-Ras, SMAD4, and p53. However, recent evidence from whole genome sequencing (WGS) of tumors has revealed an additional layer of complexity in the form of a much larger number of less-frequently mutated genes, many of which qualify as statistically significant drivers of tumor progression. In one such study, 128 different mutations were found to be distributed across different regions of a single tumor. This unexpected level of tumor heterogeneity likely explains one of the major complications associated with current approaches to personalized cancer therapy, namely, the common recurrence of tumors following treatment.

It has long been hypothesized that the immune system should be able to recognize and eliminate tumors. Unfortunately, after decades of attempts at cancer immunotherapy, the results have been extremely disappointing, leading to speculation that the immune system might not even play a role in controlling tumors at all. On the contrary, however, recent advances in understanding immune function at the molecular level have definitively confirmed the original hypothesis [reviewed in (3)]: the immune system can indeed both suppress tumor growth and destroy the tumor itself. Certain components of the immune system may also actually promote the evolution tumors that can exploit the immune surveillance system to ultimately escape detection. For example, immune checkpoints (mediated by proteins including PD-1 and CTLA-4) play an important role in down-regulating T-cell activation, which in turn reduces autoimmunity and promotes self-tolerance. Tumors frequently activate these immune checkpoints to suppress attack from the immune system. Indeed, recent therapies using checkpoint inhibitors have shown great promise in restoring the ability of the immune system to attack and destroy several types of cancer. Despite this success, long-term use of such checkpoint inhibitors might be expected to eventually lead to increased autoimmune reactions as a side effect.

As described above, the very high level of mutational heterogeneity among tumors — and even within the same tumor — represents a major obstacle for “conventional” approaches to personalized cancer therapy. However, this same heterogeneity also serves as a major advantage for a personalized cancer immunotherapy approach that is targeted to the mutations themselves, since it actually provides a large pool of potential targets for the immune system to attack. These mutations (so-called “neo-epitopes” or “neo-antigens”) should be recognized as “non-self” and attacked by the immune system, but tumors have evolved mechanisms for suppressing such attack (see above), essentially “hiding” these neo-antigens from the immune system and thereby evading detection. Over the past 4 years, a novel approach has been explored for “awakening” the immune system to the presence of these neo-antigens, involving a personalized cancer vaccine (using peptides or RNA) simultaneously directed at multiple neo-antigens present in individual tumors. Based on the extremely encouraging results from these early studies, several phase 1 clinical trials are currently being conducted.

Collectively, the conclusion from the results with both immune checkpoint inhibitors and personalized cancer vaccines against neo-antigens is that restoring immune homeostasis is sufficient to cure cancer. Especially for the personalized cancer vaccine strategy, the new vaccination methods and the ability to identify a large array of valid neo-antigens provides the tools to put the immune system back in control.

The novel personalized cancer vaccine (or “just in time” vaccine) approach involves multiple steps: (i) identification of all non-synonymous mutations in the tumor (relative to normal tissue) by sequencing DNA (whole genome or exome sequencing) and RNA (RNA-seq to determine relative expression levels for each mutation); (ii) prediction of MHC class I and class II binding efficiencies for mutations using computational tools; (iii) prioritization of the mutations based on expression level and prediction of MHC presentation efficiency; (iv) validation of neo-antigen MHC binding and immunogenicity using multiple high-throughput assays; and (v) therapeutic tumor-specific neo-antigen (TSNA) vaccination using vaccines comprised of either multiple peptide neo-antigens or synthetic RNA (encoding multiple peptide neo-antigens).

There are limitations and problems with each of the steps outlined above. For example, even though DNA and RNA sequencing costs are declining relatively rapidly, these procedures are still expensive. Furthermore, the computational tools currently available for predicting MHC binding are not very accurate. The goal of the research proposed here is to improve the personalized cancer vaccine approach by eliminating some of these limitations. First, we propose to ultimately by-pass the need for DNA/RNA sequencing by rapidly identifying non-synonymous mutations from tumors using a mass spectrometry-based proteomics approach combined with a novel de novo sequencing algorithm. Second, we will attempt to systematically improve the computational tools for predicting and prioritizing neo-antigens with optimal immunogenicity potential using a combination of results from experimental analyses coupled with machine learning techniques. Finally, the efficacy of our novel strategy and computational tools will be assessed by administration of a personalized cancer vaccine to a group of advanced-stage cancer patients in a phase-I clinical study.