mRNA cancer vaccines: What cell therapy can teach us about commercialisation

The recent success of INTerpath-001 feels like an important inflection point for personalised mRNA cancer vaccines. Until now, much of the debate has been about whether an individualised vaccine could generate a meaningful and durable anti-tumour response. A positive Phase 3 study in melanoma starts to move that conversation on.

The next question is more practical: can this type of therapy move from a successful clinical-development programme into a scalable, accessible and commercially sustainable treatment model? Oncology has seen this challenge before. CAR-T and tumour-infiltrating lymphocyte (TIL) therapies have shown that once a product is made specifically for an individual patient, commercialisation starts well before launch and the treatment pathway itself becomes part of the product.

Personalised mRNA is different and, in principle, should avoid some of the intrinsic manufacturing and administration burden of living-cell therapies. The useful lesson is therefore not that mRNA inherits the biology of CAR-T or TIL; it is that it may inherit parts of the same one-patient-one-product operating model.

INTerpath-001: from scientific promise to a commercialisation question
On 19 August 2026, Merck and Moderna announced that the Phase 3 INTerpath-001 study of intismeran autogene (V940) plus pembrolizumab met its primary endpoint of recurrence-free survival and key secondary endpoint of distant metastasis-free survival in patients with completely resected stage IIB-IV cutaneous melanoma. The trial enrolled 1,137 patients. Detailed hazard ratios, absolute event rates, subgroup analyses and overall-survival data have not yet been released.

The result builds on the randomised Phase 2b KEYNOTE-942 study, which enrolled 157 patients 2:1 to V940 plus pembrolizumab or pembrolizumab alone. At approximately five years of follow-up, the combination was associated with a 49% reduction in the risk of recurrence or death and a 59% reduction in the risk of distant metastasis or death. Overall survival remains exploratory and immature.

What makes the Phase 3 result commercially interesting is that V940 is not a conventional vaccine manufactured in bulk. It is designed from mutations in each patient’s tumour and can encode up to 34 patient-specific neoantigens. That changes the operating model before the product ever reaches the clinic.

A recently published description of the manufacturing process gives us a first indication of what that could look like. Once adequate tumour and blood samples are available, delivery to the clinic has taken approximately six weeks in trials, while manufacturing success has exceeded 99% among patients in whom neoantigen selection was successful. That qualification matters: we still do not know the full attrition from tumour sampling through sequencing, neoantigen selection, manufacture and eventual treatment.

So the challenge is becoming less about whether mRNA can be synthesised reproducibly for one patient and more about whether thousands of personalised workflows can be run in parallel, reliably, quickly and at sustainable cost. COVID proved mRNA could manufacture the same product at extraordinary scale; oncology now asks almost the opposite question — can the same platform support mass customisation, potentially producing a different product for every patient?

Not all mRNA cancer vaccines are the same
It is tempting to talk about mRNA cancer vaccines as a single modality, but commercially that is too simplistic. Fixed or off-the-shelf approaches such as BioNTech’s BNT111, BNT113 and BNT116, and Moderna’s mRNA-4359, use predefined antigens and avoid one-patient-one-product manufacturing. Fully individualised approaches such as V940 and autogene cevumeran start with the patient’s tumour and create a unique product from it.

There is also an interesting middle ground. BNT113 targets HPV16 E6 and E7 in HPV16-positive head and neck cancer. The product is standardised, but the registrational study selects patients using both HPV16 and PD-L1, so the operational burden has not disappeared; some of it has simply moved from manufacturing into diagnostics and patient identification.

The important point is that architecture determines much more than the biology. As you move from a fixed vaccine towards an individualised product, complexity progressively shifts away from inventory and distribution and into patient identification, testing, design, manufacturing and coordination.

Figure 1 | Illustrative mRNA cancer-vaccine architectures and their different operating models.

Navigating the personalised treatment journey
With an off-the-shelf medicine, the commercial supply chain broadly starts with finished inventory. With an individualised medicine, it starts with the patient. For V940, tumour material has to be obtained, sequenced and analysed; relevant neoantigens selected; an individual construct designed; the vaccine manufactured and released; and the correct product returned for treatment.

Every step creates a potential point of delay or attrition. Tissue quality and tumour cellularity matter before manufacturing even begins, sequencing may need repeating, capacity can constrain turnaround, and a patient may recur before the vaccine is ready. mRNA should remove some of the biological variability of expanding living cells, but it does not remove biological variability from the pathway altogether — some of it moves upstream into tissue acquisition and product design.

This is why I think personalised mRNA is better viewed as mass customisation than simply bespoke manufacturing. The sequence changes from patient to patient, but much of the production platform around it could potentially be standardised and increasingly automated. The commercial challenge may therefore sit less in making the RNA itself and more in orchestrating thousands of parallel patient journeys.

Figure 2 | The personalised mRNA pathway: the product begins with the patient.

What cell therapy has already taught us
CAR-T and TIL are imperfect analogues because they are living-cell products, with variability in starting material, expansion, yield and manufacturing success. They also carry far greater treatment-centre and administration burden. Personalised mRNA should avoid many of those constraints, which is precisely why the comparison is useful only if we are clear about what actually transfers.

The more useful comparison is the operating model. Each patient still triggers an individual workflow, with a turnaround clock, a product that must remain linked to that patient and a treatment pathway that has to stay coordinated while manufacturing takes place. Cell therapy has already shown that strong efficacy does not automatically translate into broad uptake if the surrounding system cannot move patients through it efficiently.

Interestingly, cell therapy itself is now trying to engineer away those constraints. In-vivo CAR-T aims to generate CAR-T cells inside the patient, potentially removing leukapheresis and central ex-vivo manufacturing. Whether through targeted viral vectors or mRNA/LNP delivery, the strategic objective is familiar: preserve the biological value of personalisation while standardising as much of the delivery system as possible. Personalised mRNA is approaching the same problem from the other direction.

Melanoma provides a useful commercial test case
AMTAGVI is a useful real-world comparison because it is another personalised therapy operating within melanoma. As of Q2 2026, Iovance reported manufacturing turnaround of 31 days or less and more than 95 authorised treatment centres across the U.S., Canada and Australia. Those are impressive operational achievements — but the launch also shows how much infrastructure has to mature around a personalised product before clinical demand consistently converts into treated patients.

Iovance entered 2025 guiding to $450–475 million in total product revenue, then reduced that expectation to $250–300 million as treatment-centre growth and treatment timelines developed more slowly than expected; it ultimately reported $264 million for the year. By Q2 2026, momentum had improved, with record total revenue of about $99 million and approximately $91 million from U.S. AMTAGVI. The lesson is not that personalised therapy cannot scale. It is that centre activation, referrals and workflow maturity can shape the adoption curve as much as the size of the eligible population.

The read-across needs one important caveat. AMTAGVI is used in advanced melanoma after anti-PD-1 failure, whereas V940 is being developed in resected adjuvant disease — different populations, different urgency and different referral routes. So the lesson transfers at the level of the operating model, not the uptake curve. Personalised mRNA may be much easier to scale, but it will still need a scalable system around it.

Personalisation versus scalability: what is optimal?
The Moderna and BioNTech programmes show why the answer is unlikely to be straightforward. BioNTech’s BNT111 demonstrated activity in advanced melanoma but is not being taken forward in that setting. Autogene cevumeran generated neoantigen-specific immune responses, yet its randomised Phase 2 study in previously untreated advanced melanoma did not significantly improve progression-free survival.

More recently, BioNTech and Genentech stopped the Phase 2 BNT122-01 study of autogene cevumeran in ctDNA-positive resected colorectal cancer. The study had crossed a futility boundary, and a later monitoring review concluded that further follow-up was unlikely to change the efficacy outcome; a numerical imbalance in overall survival was also reported, although no new safety signal was identified. Coming within days of the positive INTerpath-001 announcement, it is a useful reminder that personalisation itself is not the value proposition.

So the strategic question is not whether personalised vaccines are inherently better than fixed vaccines. It is whether the extra biological precision created by personalisation delivers enough additional clinical value to justify the complexity that comes with it. The answer may look very different by tumour type and treatment setting.

Figure 3 | Conceptual trade-off between biological precision and operational complexity; positioning is illustrative, not a comparison of clinical efficacy.

Indication selection becomes part of commercial strategy
For a personalised therapy, indication selection is not only about prevalence, unmet need and probability of clinical success. It is also about whether the product can actually fit into the disease pathway. Tumour accessibility matters because tissue is required; disease tempo matters because the patient has to remain suitable during manufacturing; and treatment setting determines how much turnaround time is clinically acceptable.

Melanoma illustrates the point well. Surgery provides tissue, and the adjuvant setting creates a window in which an individual vaccine can be designed and manufactured. A six-week process may be entirely workable there and completely unacceptable in a patient with rapidly progressing metastatic disease. Manufacturing time therefore becomes more than a CMC metric — it can influence which indications make sense first.

Geography, treatment-centre concentration and the existing standard of care matter for the same reason. For personalised platforms, indication prioritisation should combine biology, clinical value and operational feasibility rather than treating commercial attractiveness as something to assess after the clinical plan has already been built.

The COVID halo: an additional variable
mRNA enters oncology with something most new cancer modalities do not have: widespread public recognition. COVID accelerated manufacturing investment, lipid nanoparticle capability, analytical development and regulatory familiarity, potentially removing years of platform-development friction before cancer vaccines reach routine practice.

It also left patients with pre-existing associations with the term ‘mRNA vaccine’. Whether those attitudes translate into oncology is unknown and should be tested rather than assumed. A patient deciding whether to receive another COVID booster is making a very different decision from someone considering a therapy designed from mutations in their own tumour. That makes perception a commercial variable, but not the central story.

What developers should do differently
If personalised mRNA continues to progress, the most useful lesson from cell therapy is simple: the treatment pathway becomes part of the product.

1. Design the commercial pathway in parallel with the clinical pathway
Map where patients are identified, where tissue is collected, who orders sequencing, what turnaround is clinically acceptable, who manages the patient while the product is being made and where treatment takes place. Clinical trials should generate evidence against these questions while the operating model can still change.

2. Model the patient journey, not just the eligible population
For a personalised therapy, the relevant funnel is closer to: eligible → identified → referred → sampled → successfully sequenced → neoantigens selected → manufactured → quality released → patient remains eligible → treated. Every transition can leak, and that attrition can matter as much as market share.

3. Treat manufacturing, economics and regulation as product variables
Turnaround time, neoantigen-selection success, manufacturing success, release, capacity, automation, cost of goods and geographic reach are not just CMC metrics if they determine whether a patient can actually be treated. The same is true of the regulatory model: when the sequence changes for every patient, release specifications, comparability and the scope of the marketing authorisation can directly shape commercial capacity.
Cost and reimbursement sit inside the same problem. In a one-patient-one-product model, the economics are shaped by capacity utilisation, success rate and the surrounding treatment pathway rather than a simple fixed unit cost. The relevant commercial question is therefore not just ‘what is the price of the product?’ but ‘what does it cost the system to get one patient successfully treated?’

4. Include operational feasibility when selecting lead indications
Tumour accessibility, disease tempo, treatment setting, manufacturing window, centre concentration and fit with the existing standard of care can materially alter the attractiveness of an indication before full manufacturing scale has been achieved.

5. Build evidence for adoption before launch
A clinical trial demonstrates efficacy and safety under controlled conditions; it does not prove that the future healthcare system can identify, manufacture and treat the intended population at scale. Development programmes should measure tissue success, turnaround, patient drop-off, referral behaviour, centre capability, payer requirements and patient perception before launch.

The road ahead
INTerpath-001 may ultimately prove to be a landmark study for cancer vaccination, but in many ways its Phase 3 success starts the more interesting experiment. The detailed efficacy and safety data still matter, and BioNTech’s recent colorectal setback reminds us that the platform alone will not determine success. What has changed is that commercialisation is no longer theoretical.

The opportunity for personalised mRNA is to make individualisation operationally routine: the sequence can change for every patient while the manufacturing and treatment system around it becomes increasingly standardised. Cell therapy has already shown the value — and the cost — of personalisation. In-vivo CAR-T and personalised mRNA now point towards the same broader ambition: preserve the biology while progressively removing the operational burden.

If that can be achieved, the innovation may be as much in how the therapy is delivered as in the vaccine itself.

For anyone who has lived through a CAR-T or TIL launch: what is the one thing you wish your team had modelled two years earlier?

I’m Executive Director, Innovation & Strategy at Inovia Bio. The views expressed here are my own.