← Back to Home

Evaxion Reshapes Cancer Vaccine Pipeline With AI-Designed Off-the-Shelf Brain Cancer Candidate EVX-05

The new program searches the glioblastoma “dark genome” for antigens shared across patients, aiming to eliminate the burden of producing individualized vaccines; however, the evidence remains limited to a small-sample analysis and in vitro immune responses.

By SURL BioNews

Glioblastoma has a relatively low mutational burden and offers few typical neoantigens that the immune system can recognize, making cancer vaccine design particularly difficult. Danish biotechnology company Evaxion is now turning its attention to viral remnants within the human genome that have long been dormant but may be reactivated in cancer cells, using them as the basis for the therapeutic vaccine program EVX-05.

Evaxion said EVX-05 will target conserved antigens derived from endogenous retroviruses (ERVs) that are shared by the tumors of multiple patients, positioning it as an off-the-shelf vaccine that can be prepared in advance. The company is working with a Duke University School of Medicine team to optimize the lead candidate and plans for a team led by Mustafa Khasraw to conduct initial human testing, potentially through a streamlined Phase 1 trial. No start date, trial design, or progress on regulatory submissions has yet been announced.

This AI system does not directly interpret images. Instead, it integrates DNA and RNA data from tumor and normal tissues: the PIONEER pipeline searches for mutation-derived neoantigens, while ObsERV screens for ERV-derived antigens that are abnormally expressed in tumors and less prevalent in normal tissues, then ranks them by expression level, immune presentation capability, and other criteria. EVX-05 takes this a step further by attempting to identify shared targets from personalized designs that can be used across patients, reducing the complexity of sequencing, designing, and producing a vaccine for each individual.

The Duke–Evaxion study supporting the program analyzed paired tumor and normal tissue data from 24 patients, 17 of whom survived for more than 5 years, while the other 7 survived for less than 18 months. The research team was able to generate personalized vaccine designs for all 24 tumor samples; 21 incorporated both mutation-derived neoantigens and ERV antigens, two used only neoantigens, and one used only ERV antigens. The selected antigens also triggered interferon-gamma responses in cells from three healthy donors with different HLA types, providing preliminary functional evidence.

However, these results demonstrate the feasibility of computer-based design and in vitro immune recognition, not that EVX-05 can already control tumors. The sample size was limited, the tests used cells from healthy donors, and the early research employed personalized antigen combinations. Whether the conserved off-the-shelf antigens proposed by the new program can cover enough patients, avoid normal tissues, and function in the immunosuppressive environment of glioblastoma remains to be answered through animal studies and human trials.

The pipeline adjustment also reflects resource-allocation considerations. Evaxion discontinued EVX-03 and added EVX-05, saying this would not change its existing budget or its estimate that its cash runway will extend into the second half of 2027. EVX-05 builds on the ERV-targeting concept used by EVX-04, a candidate vaccine for acute myeloid leukemia. The company has also previously said it is inclined to seek a partner for the glioblastoma program. Consequently, clinical funding, development responsibilities, and whether the candidate can enter human testing on schedule will be critical to determining whether this R&D shift can be implemented.

References

  1. marketscreener.com
  2. Evaxion
  3. Duke University School of Medicine and Evaxion (AACR 2026 poster)
  4. Evaxion
  5. Evaxion