“Every patient has the potential to eradicate their cancer within their own immune system”: New Frontiers in Cancer Vaccines

From my perspective, the field faces three important hurdles:

1. The identification of therapeutically relevant targets in the form of naturally presented antigens by which cancer cells can be specifically and exclusively recognized.
2. The availability of vaccine platforms that reliably generate T cell responses to most or all the target antigens in patients.
3. Immune monitoring approaches that monitor T cell responses against cells expressing the targeted protein rather than simply the target peptide.

Of these, I think advances in immunopeptidomics, especially around the identification of non-canonical “dark” antigens, could be the most consequential as they open the door to an entire new class of targets.

Although much more fundamental research is needed to validate these as being truly tumor-specific and, importantly, whether any might be shared across specific tumor types.

The vaccine formulation problem could be meaningfully advanced by recent advances in artificial intelligence (AI)-assisted protein design, but I continue to believe that the target identification problem is going to require real-world data from patients. 

Cancer vaccines are intended to induce potent immune responses to the target antigens they encode or contain. As every patient’s cancer possesses a set of distinct mutations that can be targeted, their vaccine will similarly be “personalized” to their particular tumor and immune response potential.

Although we have a range of genetic and protein vaccine platforms that can generate immunity against “one size fits all” antigens, such as viral proteins, we simply lack robust, cost-effective and scalable platforms for delivering individualized multi-valent vaccines to large patient populations.

The peripheral immune system retains a cellular memory of the antigens it has encountered, whether through natural encounter or vaccination, in the form of B and T cells. Most personalized antigen discovery efforts are based on predictive approaches that seek to model which peptides may be available on cancer cells for recognition by T cells, but these do not address the higher-order biology by which such T cells are generated under physiological conditions.

My colleagues and I reasoned some years ago that, rather than guessing at which tumor mutations might be valid targets, letting the patient’s immune system report on which antigens it had generated T cells against might be an interesting alternative approach to identify therapeutically actionable targets.

My colleagues and I developed a platform called IPV (Identify-Prioritize-Validate), which combines bioinformatic prioritization of expressed mutations with functional testing of patient PBMC for spontaneous pre-existing T cell responses. IPV was trained on biology rather than prediction and has successfully detected natural T cell responses against expressed mutations (referred to as “neoantigens”) in more than 25 types of cancer, suggesting that most, if not all, cancer patients possess the cellular potential to benefit from a personalized vaccine, provided it targets the right antigens.

We performed a Phase 1b clinical trial of personalized cancer vaccines against target antigens identified with the IPV platform in poorly mutated solid tumor patients in the metastatic and adjuvant settings, which has demonstrated impressive clinical effects, including protection from recurrence and death.

The idea that every patient has the potential to eradicate their cancer within their own immune system is especially compelling and worth pursuing.