Cancer vaccine graveyard—will the field rise from the dead?

My life goal is to help people live a better (healthier) life. I also need to work on problems that I can become obsessed with. My friend & I recently teamed up to explore oncology. Give us feedback.

Cancer vaccines (therapies that train the immune system to recognize and attack tumors) have represented one of oncology's most compelling yet frustrating pursuits. The concept builds on the success of traditional vaccines against infectious diseases and virus-driven cancers like HPV (helps prevent cervical cancer) and hepatitis B (helps prevent liver cancer).

The big picture appeal is that if we could vaccinate against cancer, we might achieve durable, selective immunity while sparing patients the toxicities of conventional chemotherapy, radiation, or surgery.

Context before reading: one of the most confusing parts of digging into this space has been how loosely the term ‘vaccine’ has been used. Initially I assumed it referred to a fully prophylactic (preventative vaccine), yet as I did more digging in the space I learned it has been used in both therapeutic and prophylactic cases interchangeably.

Yet despite decades of research and billions in investment, the field earned a reputation as a graveyard of failed ambitions. The immune system offers powerful, long-lasting memory—once trained to recognize a target, it can patrol for years to prevent disease recurrence. Unlike chemotherapy or radiation, which affect both healthy and malignant cells, an immune-based approach promised specificity and minimal side effects.

Scientists have been chasing this dream for over a century, starting with William Coley injecting bacterial toxins in the 1890s in a patient with inoperable cancer to try to get the “infection” to shrink the cancer (it was a little successful actually?). By the 1980s and 90s, we'd figured out the basics: cancer cells display specific proteins (tumor antigens), and T cells can potentially recognize and kill these cells. We even had proof of concept—HPV vaccines dramatically slashed cervical cancer rates by teaching the immune system to eliminate the cancer-causing virus. However, the excitement there was short-lived and early successes were a bit of a misnomer. Successful “cancer vaccines” like for HPV or Hep B reduce your risk of getting an infection that makes certain cells cancerous rather than training your immune system to get rid of cancerous cells, even though the premise is quite similar in all vaccine cases (i.e. train immune cells to get rid of bad things by using targets expressed on bad things as identifying markers of bad thing).

But here's where things get brutal. Starting in the 1990s through the 2010s, cancer vaccine after cancer vaccine crashed and burned in clinical trials. We're talking dozens of failures. Companies spent billions. Patients pinned their hopes on treatments that, time and again, showed they could trigger immune responses in the lab but couldn't actually help people live longer.

The field became known as a graveyard—and honestly, it earned that reputation.

Why most cancer vaccines have failed: biological and clinical struggles

The reasons for widespread failure have become clearer through decades of research. Cancer presents unique immunological challenges that distinguish it from infectious disease targets:

1. Immune system already has a tolerance to things that are expressed in cancer cells. Cancer cells arise from the body's own tissues, meaning many tumor antigens appear as "self" to the immune system. Powerful tolerance mechanisms that normally prevent autoimmunity can suppress responses to tumor-associated antigens. Early vaccines targeting proteins like MAGE-A3, NY-ESO-1, or PSA, which are elevated in tumors but still expressed in normal tissues, generated weak immune responses due to pre-existing tolerance.
   

2. Many first-generation vaccines targeted tumor-associated antigens that weren't truly tumor-specific. Because these proteins existed in normal tissues, immune responses were constrained to avoid autoimmunity. Additionally, tumors could evade detection by downregulating targeted antigens or overexpressing inhibitory molecules like PD-L1.
   

3. Individual tumors contain genetically diverse cell populations due to genomic instability. A vaccine targeting one antigen might eliminate certain cancer cells while resistant variants lacking that antigen continue to proliferate. This heterogeneity is particularly pronounced in advanced cancers, where multiple escape variants may already exist.
   

4. Advanced tumors create suppressive conditions that neutralize immune responses. They recruit regulatory T-cells and myeloid-derived suppressor cells, secrete immunosuppressive factors like TGF-β and IL-10, and upregulate checkpoint proteins that inhibit T-cell function. Even vaccine-primed T cells can be rendered inactive upon tumor infiltration.
   

5. Many early vaccine approaches generated only modest T-cell responses. Small peptide fragments with weak adjuvants often failed to create the robust, sustained immune activation required to overcome established tumors. Even vaccines that showed immune responses in blood often failed to correlate with tumor regression.
   

6. Trials had suboptimal patient selection and timing. Most trials enrolled patients with advanced metastatic disease who had failed other therapies. These patients typically had large tumor burdens, extensive immunosuppression, and compromised immune function from prior treatments. Testing vaccines in end-stage disease essentially created the worst possible conditions for immunotherapy success.
   

7. Clinical trials had major design issues. Early trials often focused on immunological endpoints rather than clinical outcomes. The ability to generate measurable immune responses in blood frequently failed to predict efficacy, and no validated biomarkers emerged to identify patients likely to benefit from vaccination.

High-profile failures: lessons from the graveyard

Several major clinical failures illustrate these challenges and provided crucial insights for future development:

1. MAGE-A3 Vaccine: GlaxoSmithKline's MAGE-A3 antigen vaccine reached large phase III trials in both melanoma (2013) and non-small cell lung cancer (MAGRIT study, 2014). Both trials failed to meet primary endpoints for disease-free survival or overall survival. Post-hoc biomarker analysis failed to identify any responsive patient subgroups, leading to program termination. The failures highlighted issues with tumor-associated antigen selection and immune evasion mechanisms.
   

2. PROSTVAC: This PSA-targeted viral vector vaccine for metastatic castration-resistant prostate cancer showed early promise but failed its phase III trial in 2017. An interim analysis demonstrated no improvement in overall survival compared to placebo, leading to early termination. The failure underscored that immunological responses don't necessarily translate to clinical benefit in advanced disease.
   

3. GVAX Platform: This whole-cell vaccine approach, using tumor cells engineered to secrete GM-CSF, was tested across multiple cancer types. In prostate cancer, a phase III trial was terminated for lack of efficacy. In pancreatic cancer, combining GVAX with CRS-207 (engineered Listeria) not only failed to meet endpoints but showed worse outcomes than control arms (3.8 vs 5.4 months median survival), leading to program cancellation.

By the mid-2010s, attention shifted to checkpoint inhibitors and CAR-T cells, and many researchers quietly moved on to other projects.

Strategic advances: a path forward

Recent years have seen fundamental shifts in cancer vaccine development that directly address previous limitations:

• The field has pivoted from tumor-associated antigens to personalized neoantigens—proteins arising from tumor-specific mutations. These targets are truly foreign to the immune system, avoiding tolerance issues while addressing tumor heterogeneity through multi-epitope inclusion. Advances in genomics and computational biology have made personalized neoantigen identification feasible at scale (but this is not something to overlook as trivial — it’s quite difficult to find in many cases it seems and the personalized nature of the approach makes it so that you can’t have off-the-shelf vaccines ready to go. And again, you can’t make this a truly prophylactic preventative vaccine, so it’s being used in a therapeutic context)
  

• Newer trials increasingly target minimal residual disease settings, particularly adjuvant therapy (an additional cancer treatment given after the primary treatment, like surgery or radiation therapy, to help prevent the cancer from returning. It aims to kill any remaining cancer cells and reduce the risk of recurrence) after surgical resection. This approach leverages periods of low tumor burden when immune surveillance is most effective, rather than attempting to treat extensive metastatic disease.
  

• Modern vaccine development emphasizes combination with checkpoint inhibitors and other immunomodulatory agents. Vaccines can expand tumor-specific T-cell populations while checkpoint inhibitors remove suppressive mechanisms, creating synergistic anti-tumor immunity. This addresses the immunosuppressive microenvironment that neutralized earlier monotherapy approaches.
  

• We now have better vaccine platforms. mRNA vaccine technology is just one of many examples which offer improved immunogenicity and manufacturing flexibility. These platforms can rapidly encode multiple neoantigens while generating robust T-cell and antibody responses. There have also been many recent innovations in improved peptide design & cell therapy manufacturing.
  

  Ignoring my previous complaint of the term '‘vaccine” being used in prophylactic, therapeutic (+ adjuvant) cases, there’s a lot to be said about the strategy in figuring out how to test your cancer vaccine technology/approach. How do you find a way to show it works with meaningful clinical readouts? If you wanted to do a purely prevention play, you’d have to select a patient population that was at a higher risk of getting a specific cancer, know what neoantigens were specific to that cancer and shared amongst many patients, and find readouts that show prevention without having to wait for that population to get the cancer (or intervene at a timescale that wouldn’t lead to an incredibly long clinical trial). And finally, if you wanted that vaccine to be applicable to a bigger patient population, you’d have to be pretty confident the neoantigens would extend to being shared by the greater population. Huge, if true (emphasis on “if” and “huge” — I am equally skeptical as I am hopeful because I want this to exist AND somehow be affordable or reimbursable).

Recent clinical successes: proof of concept

Jumping from my spiral to two recent trials have provided compelling evidence that properly designed cancer vaccines can achieve clinical benefit:

1. Moderna's Personalized mRNA Vaccine (mRNA-4157/V940): In the KEYNOTE-942 phase 2b trial, patients with resected high-risk melanoma received either pembrolizumab (KEYTRUDA) alone or pembrolizumab plus personalized mRNA vaccine targeting up to 34 neoantigens. The combination reduced recurrence or death risk by 44% compared to pembrolizumab monotherapy. At 18 months, recurrence-free survival was 78% vs 62%, representing the first randomized trial to demonstrate statistically significant benefit for a therapeutic cancer vaccine approach.
   

2. BioNTech's Pancreatic Cancer Vaccine (autogene cevumeran): In a phase I study of patients with resected pancreatic ductal adenocarcinoma (historically resistant to immunotherapy personalized mRNA vaccination combined with atezolizumab and chemotherapy) generated tumor-specific T-cell responses in approximately half of patients. Importantly, patients mounting vaccine-induced immune responses showed significantly longer recurrence-free survival, with durable T-cell memory persisting for years post-vaccination.

Future directions

While these results represent significant progress, cancer vaccines face continued challenges and limitations. They are unlikely to function as monotherapies for most cancers and will probably require combination with checkpoint inhibitors or other immunomodulatory agents (unless! unless! we can find neoantigens shared amongst a broad population for various cancers that can be given to the public as a fully preventable approach, showing the vaccines can mount necessary T cell response and the immune memory is maintained for long enough that re-dosing doesn’t become needed too often).

Approaching it from a realistic POV, based on current track records may be limited to specific clinical contexts, particularly adjuvant settings with minimal residual disease, and tumor types with sufficient mutational burden to generate immunogenic neoantigens. The field is expanding beyond melanoma and pancreatic cancer, with ongoing trials in lung, colorectal, ovarian, and other malignancies applying neoantigen-focused strategies. However, manufacturing and logistical challenges remain for personalized approaches, requiring sophisticated genomic analysis, rapid vaccine production, and coordinated clinical workflows.

Perhaps there’s a path forward?

Cancer vaccines have undergone a remarkable transformation from repeated clinical failures to emerging therapeutic promise. The field's history illustrates how scientific persistence, combined with systematic analysis of failure mechanisms, can eventually yield breakthrough approaches.

Contemporary vaccine development addresses the fundamental biological barriers that defeated earlier efforts: targeting truly tumor-specific neoantigens, optimizing clinical timing and patient selection, leveraging combination strategies, and employing more potent vaccine platforms. While challenges remain, recent clinical successes provide compelling evidence that cancer vaccines can contribute meaningfully to patient outcomes when properly designed and deployed.

Pls pls msg me with corrections, comments, ideas — I want cancer vaccines to exist!