Neoepitopes as difference makers for general cancer vaccines?

SUMMARY

The cancer mutanome has been associated with disease prognosis as well as response to interventional immunotherapy and provides a substrate for development of personalized vaccines targeting tumor neoepitopes. Recent findings suggest that neoantigen-based vaccines may represent general interventional approaches for patients with solid cancers, regardless of their inherent mutational burden.

In this issue of Clinical Cancer Research, Fang and colleagues¹ report results from their single-arm, open-labeled, investigator-initiated clinical trial (NCT03662815) involving 22 advanced-stage patients treated with personalized peptide-based mutated neoantigen vaccines (designated iNeo-Vac-P01) conducted over the period of 2/2018 – 5/2019. Recruited patients had failed prior standard of care regimens and presented with one of 11 different forms of solid cancer. A general paradigm for neoantigen-based vaccine development and application in cancer patients is outlined in Fig. 1. In the report by Fang et al.¹, patient-specific mutated neoantigens and derivative MHC-presented peptide epitopes encompassing mutant sequences were determined by bioinformatics analyses (iNeo-Suite) based on comparing whole-exome sequencing (WES) data from patient-matched tumor (mutated) vs. peripheral blood specimens (genomic control). Pooled peptide vaccines (5–20 peptides with each peptide 15–35 amino acids in length) were delivered s.c. in the upper arms and para-umbilical area along with adjuvant rhGM-CSF in an accelerated priming schedule (days 1, 4, 8, 15, 22), with follow-up booster vaccines. In keeping with past trials of cancer neoantigen vaccines², the current vaccines were well-tolerated, with only 2 patients developing grade 3 or 4 acute allergic reactions but only after repeated (6) booster immunizations, necessitating treatment discontinuation. The study disease control rate (DCR) was 71.4%, with a median progression-free survival (PFS) of 4.6 months, and the median overall survival (OS) not yet reached at the time of this report (12-month OS = 55.1%). Consistent with the published clinical experience for personalized neoantigen-based vaccines, and the expectation that (somatically) mutated tumor neoantigens are not subjected to central immune tolerance mechanisms in the host², the majority of neoantigen-derived peptides (~ 80%) evaluated in the current study were capable of eliciting specific responses from patient peripheral blood T cells after vaccination based on IFN-γ ELISPOT assays. Furthermore, the vast majority (~90%) of evaluable patients exhibited on-treatment T cell responses to their personalized iNeo-Vac-P01 vaccine based on IFN-γ ELISPOT assays and the presence of high-abundance T cell clonotypes in peripheral blood, deduced by TCRBseq analyses. Multi-parameter immunofluorescence analyses of on-treatment tumor biopsy specimens from 3 patients revealed increased numbers of GZMB⁺ CD8⁺ and CD4⁺ tumor-infiltrating lymphocytes (TIL) 5.5–9 months after initial vaccine priming.

Figure 1. Operational paradigm for cancer neoantigen-based vaccines.

Patients with histologically-confirmed cancer provide tumor biopsy specimens (fresh or banked) and peripheral blood [A], from which DNA is extracted and WES performed [B]. Comparisons of tumor vs. blood sequences allow for the identification of patient-specific mutation events in translated proteins and subsequent algorithm-based definition of candidate MHC-presented CD4⁺ and/or CD8⁺ T cell epitopes based on the patient’s HLA haplotype and additional consideration of antigen-processing modifications [C]. Selected epitopes are synthesized as short (containing single epitopes) or long peptides (containing tandem epitopes) [D], then pooled and formulated with adjuvants (i.e. rhGM-CSF, TLR ligands, dendritic cells, among others) [E], prior to injection into antigen presenting cell-rich tissues, such as the skin, in prime/boost regimens [F]. Clinical trial endpoints commonly include assessment of safety (i.e. lack of severe AEs/irAEs), induction of specific T cell responses based on functional readouts such as IFN-γ release assessed in ELISPOT assays, evidence for increased prevalence of TIL (including Ag-specific TIL), tumor shrinkage based on radiographic imaging, increased DCR and extended OS/PFS (per RECIST1.1 criteria)[G]. Rather than applying WES data to develop peptide-based vaccines, this information can also be used to generate personalized DNA/RNA-based vaccines, or deduced neoantigen-derived peptides used to expand Ag-specific T cells for ACT or to isolate T cell clones from which TCR Tg T cells may be developed for ACT-based immunotherapy [H]. Although the current study implements a peptide-based monotherapy, it would be expected that the antitumor efficacy of such approaches would be enhanced by combination with alternate immunotherapeutic strategies [I] that enhance recruitment, functionality and persistence of vaccine-induced T cells within the TME and/or which broaden spreading in the evolving immune response to include a T cell repertoire reactive against tumor-associated Ag (including additional neoantigens) not inclusive in the priming vaccine. Abbreviations: ACT, adoptive cell therapy; AE, adverse event; GM-CSF, granulocyte-macrophage colony stimulating factor; DCR, disease control rate; irAE, immune-related AE; OS, overall survival; PFS, progression-free survival; TCR, T cell receptor; Tg, transgenic; TIL, tumor-infiltrating lymphocytes; TLR, toll-receptor ligands; TME, tumor microenvironment; STING, stimulator of interferon genes; WES, whole exome sequencing.

The current trial is remarkable in demonstrating immunologic, and to a lesser degree clinical, response to neoantigen-based vaccination in a comparatively large number of patients harboring extremely diverse forms of cancer. Previous trials of tumor neoantigens accrued smaller numbers of patients and were restricted to single types of cancer². Personalized iNeo-Vac-P01 vaccines were efficiently generated within a short 1.5 – 3 month period regardless of patient cancer histotype, stressing the feasibility of applying this approach in the broader cancer setting, even in cases at higher risk for rapid progression. Importantly, these vaccines were shown to be immunogenic in patients (based on iRECIST criteria) harboring consensus high (i.e. melanoma) as well as low (i.e. pancreatic carcinoma) tumor mutation burden (TMB), jibing with published results for smaller clinical trials of neoantigen vaccines in melanoma (high TMB) or glioma (low TMB) patients². Despite evidence for near universal patient immune responsiveness to vaccination with iNeo-Vac-P01 vaccines, no patients achieved complete or partial response (CR or PR) (i.e. best tumor shrinkage = ~17%), although some patients exhibited on-treatment pseudoprogression of (advanced hepatic biliary tract cancer) lesions or regression of liver metastasis (pancreatic carcinoma). These latter data are intriguing given the dismal prognosis for patients with advanced-stage pancreatic carcinoma or biliary tract cancer, suggesting that further optimization of the iNeo-Vac-P01 immunotherapeutic platform might fill the current void for effective treatment options in these patient cohorts.

A conundrum remaining to be addressed at the bench and in prospective clinical trials of tumor mutant neoantigen-based vaccines reflects the “elephant in the room”; i.e. the striking disconnect between treatment immunogenicity and evidence for therapeutic (RECIST1.1) benefit in patients with multi-focal disease. Simple considerations include insufficient numbers or quality of T cells elicited with neoepitope + GM-CSF-based vaccines in the face of dominant immunosuppression in the tumor microenvironment (TME) in the current report that might be improved in formulations using alternate adjuvants (i.e. TLR agonists, dendritic cells), or inefficient recruitment of circulating vaccine-induced T cells into the TME that could be improved by combining vaccines with a range of co-immunotherapeutic agents including checkpoint blockade, costimulatory agonists, TLR/STING agonist, oncolytic viruses, adoptive cell therapy, radiotherapy, among others (Fig. 1). A more likely consideration reflects the nightmarish logistics in correctly identifying and clinically-applying in silico-selected neoantigenic epitopes that are both naturally-processed and presented in sufficient stochastic quantities in MHC complexes on the tumor cell surface to allow for recognition by even high-avidity cognate T cells that have escaped central tolerance². For instance, even tumor infiltrating lymphocytes (TIL) and mass spectroscopy are only able to effectively identify 1–5% of in silico predicted neoepitopes as being naturally-processed and MHC-presented on the tumor cell surface³. This selection burden is made greater given known tumor heterogeneity with regards to antigenicity and immunogenicity (i.e. variance in antigen processing and MHC presentation machinery) within a given patient lesion as well as between distinct lesions in patients with multi-focal, disseminated disease, where epitopes in driver or truncal mutated neoantigens⁴ might represent particularly salient targets. In the end, while refinement in predictive algorithms may improve identification of tumor neoepitopes that serve as rejection antigens, we will likely have to depend on a strength in numbers approach. In such a scenario, even one correctly predicted “natural” neoepitope in a multi-peptide vaccine might prove sufficient to evoke therapeutic benefit, or to allow that vaccine to elicit the evolution of spreading in the T cell repertoire to reinforce reactivity against naturally-presented cancer mutated neoepitopes not present in the priming vaccine (Fig. 1). Clearly, the ability of checkpoint blockade interventions⁵ to uncover T cell responses against mutant tumor neoantigens that are prognostic of treatment outcome bodes well for the development of combined treatment modalities integrating personalized vaccines such as iNeo-Vac-P01 (Fig. 1). Hence, while there is clearly much work left to do, the report by Fang et al.¹ engenders significant enthusiasm for the prospective refinement of personalized cancer neoantigen-based vaccines as a general approach for treating patients with advanced-stage solid cancers, particularly in the context of multi-modality treatment designs.