Background: Cancer mutations acquired during carcinogenesis are not subject to central tolerance and thus, are cornerstone in cancer immunosurveillance. The use of mutation-derived neoepitopes to design cancer vaccines is an unprecedented opportunity to target efficient antigens, but the stochastic occurrence of mutations implies that a tailored vaccine needs to be designed and manufactured for each patient. We used next generation sequencing, machine learning based genomic data analysis and advances in genome editing to systematically identify private neoepitopes in high-risk HPV negative HNSCC and relapsing OvC patients.

Methods: Mutations were identified by DNA and RNA sequencing and putative epitopes were selected for vaccine design on their likelihood to elicit a class I or II response based on HLA binding, allelic frequency, prediction of processing and expression at RNA level. The vaccine was a recombinant virus of the modified Vaccinia Ankara (MVA) strain encoding for up to 30 sequences of 29 amino acids centred on the predicted epitope and flanked with the normal germline sequence in case of missense mutations, and of the mutated protein sequence in case of frameshift mutations. The viral vaccine was amplified under GMP conditions and administered to patients after curative intent treatment in two phase I trials. Treatment was started for OvC patients at relapse defined as elevation of CA-125 or onset of suspicious radiological findings and, after first line therapy for HNSCC patients with locally advanced disease with either extracapsular extension or positive resection margins following upfront surgery. The vaccine was administered for 6 weeks and a booster dose every three weeks over a year, in absence of other cancer therapy.

Results: Six patients (4 OvC and 2 HNSCC) were treated and experienced only mild related adverse events. After 6 doses of vaccine, we measured immunoreactive T cells in PBMCs against each of the targeted epitopes using an ex vivo IFNγ ELISPOT showed robust responses in all evaluable patients. Positive responses to vaccination were observed for a total of 46 epitopes over 115 tested with a median of 11 epitopes (6 to 18) per patient. Responses were directed against both class I and II epitopes and were either de novo responses (64%) or amplification of pre-existing responses (36%). In parallel, phenotyping of patient PBMCs showed, in all treated patients, a decline in naïve CD4 and CD8 cells and the onset of effector CD27- CD4 and CD27-/CD57+ CD8; similarly, CD56dim NK cells increased suggesting a stimulation of antitumor innate immunity. In one ovarian cancer patient, these immune changes correlated with a normalization of CA-125 that lasted 7 months.

Conclusions: These data show that the personalized approach with TG4050 is feasible, safe and immunogenic. The two phase I trials are ongoing.

Citation Format: Matthew S. Block, Jean-Pierre Delord, Christian H. Ottensmeier, Christophe Le Tourneau, Ana Lalanne, Olivier Lantz, Keith L. Knutson, Gisèle Lacoste, Annette Tavernaro, Maud Brandely, Nathalie Silvestre, Benoit Grellier, Yoshiko Yamashita, Onoue Kousuke, Naoko Yamagata, Eric Quemeneur, Kaidre Bendjama. Phase I trials of personalized cancer vaccine TG4050 in surgically treated high-risk head and neck squamous cell carcinoma (HNSCC) and relapsing ovarian cancer (OvC) patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr CT182.