Vaccines, Cancer, and Beyond

The first two lessons built a complete system: an engineered, immune-quiet mRNA, wrapped in a lipid nanoparticle that carries it into cells and releases it. Because only the coding sequence changes between products, that one system can make many different medicines. This lesson surveys what it can do, from the application that proved it to the frontiers now in trials.

The key is to keep asking a single question for each use: what protein does the mRNA tell the cell to make, and what should that protein do? A viral protein trains immunity. A tumor-specific protein directs an attack on cancer. A missing enzyme replaces one a patient lacks. Same delivery, same manufacturing, different instructions. That reframing is the entire value of a platform, and it is why one solved delivery problem opened so many doors at once.

The application that proved the platform is the preventive vaccine. Here the mRNA encodes a piece of a pathogen, a viral surface protein, for example the SARS-CoV-2 spike protein. After injection, the patient's own cells briefly make that protein and display it to the immune system, which learns to recognize it and builds antibodies and memory cells. If the real pathogen appears later, the immune system responds fast.

Crucially, the vaccine contains no virus and cannot cause the disease; it delivers only the instructions for one harmless piece. The COVID-19 vaccines Comirnaty (Pfizer-BioNTech) and Spikevax (Moderna) demonstrated this at global scale, and were the first mRNA products to reach widespread authorized use. Beyond proving safety and efficacy, they proved the platform's headline advantage: speed. Once the spike sequence was known, candidate vaccines were designed in days, because designing an mRNA is, at heart, choosing a sequence.

The most striking use of the platform turns it against cancer, and it depends on the platform's programmability in a way a conventional drug cannot match.

As tumors grow, they accumulate mutations, and some of these produce abnormal proteins found only in the cancer, never in healthy cells. These tumor-specific markers are called neoantigens. In principle they are perfect targets: an immune response against a neoantigen attacks the tumor while sparing normal tissue. The problem is that every patient's tumor carries different mutations, so there is no single off-the-shelf target. This is exactly where a programmable medicine fits. If you can read a specific tumor's mutations and encode its neoantigens into an mRNA, you can make a vaccine tailored to that one patient, using the same manufacturing process for everyone. The personalization lives entirely in the sequence.

The workflow is a loop from patient to patient. Biopsy the tumor and sequence its DNA alongside healthy tissue to find the mutations unique to the cancer. Computationally pick the neoantigens most likely to provoke an immune response. Encode them into a single mRNA, manufacture it in an LNP, and inject it. The patient's cells display the neoantigens, and T cells are trained to hunt cells bearing them, the tumor.

This is not hypothetical. The leading example is mRNA-4157 (also called V940, intismeran autogene), developed by Moderna and Merck. A single personalized transcript encodes up to 34 patient-specific neoantigens, chosen from that patient's tumor sequencing. It is given together with pembrolizumab (Keytruda), an existing immunotherapy that releases a brake on T cells, so the vaccine points the immune system at the tumor while the second drug lets it attack harder.

In the Phase 2b trial known as KEYNOTE-942, in patients with resected high-risk melanoma, the combination was tested against pembrolizumab alone. In January 2026 Moderna and Merck reported five-year follow-up data showing a 49% reduction in the risk of recurrence or death for the combination versus pembrolizumab alone. The programme has since moved into larger Phase 3 trials. It is important to be precise: this is an investigational therapy demonstrating strong results in trials, not yet a broadly approved standard of care.

Vaccines train the immune system, but the platform's logic, make any protein inside the body, extends much further. Several directions are in development:

• Protein-replacement therapy. For diseases caused by a missing or broken protein, mRNA can instruct cells to make a working copy. The liver's easy accessibility to LNPs (from lesson 2) makes liver-based metabolic disorders an early target.
• In vivo cell engineering. Rather than re-engineering a patient's immune cells in a lab, as conventional CAR-T therapy does, mRNA is being explored to reprogram those cells directly inside the body, which could make cell therapies simpler and cheaper.
• Regenerative and secreted proteins. Instructing cells to produce signaling proteins, for example to stimulate tissue or blood-vessel repair, is another active avenue.

These are at varying, earlier stages than vaccines, but they share the same engine. The transient nature of mRNA suits situations where you want a controlled, temporary burst of a protein.

The same platform, sorted by what the mRNA is told to encode and what that achieves:

Read down the middle column and the pattern is clear: the medicine is defined by a sequence, and the platform around it barely changes. That is the sense in which mRNA is less a drug than a way of making drugs, a single validated system whose output is set by software-like instructions.

The platform is powerful, not magic, and a clear-eyed view names the constraints, most of which trace back to earlier lessons.

• Delivery is still mostly liver and injection-site. Reaching specific other tissues, particular tumors, the lungs, the brain, remains hard; biodistribution limits which diseases are reachable today.
• Expression is transient. The controlled, temporary burst that suits vaccines is a drawback for chronic conditions that would need repeated dosing to keep a protein present.
• The cold chain persists. Stability and storage constrain real-world access, as lesson 2 described.
• Immunogenicity is a balancing act. You want an immune response to a vaccine antigen but not to the mRNA or lipids themselves; tuning that balance is ongoing.

None of these are dead ends, and each is an active research target. The honest summary is a validated, versatile platform with a proven first application, real clinical momentum in oncology, and a set of well-understood engineering problems standing between it and its full promise.