Ronald Levy looks at orange fluid in a container in his lab

Ronald Levy is a cancer researcher at Stanford University in California.Credit: Norbert von der Groeben/Stanford Medicine

In 2017, Bill Morrison was working hard and feeling fine. But swelling on his neck began to bother him. “It didn’t hurt but it was kind of puffed out on one side,” he says. “It just seemed a little strange.”

Initial blood tests led to scans, which led to a biopsy, and finally to the revelation that Morrison had a B-cell lymphoma that had spread throughout his body. “It just came out of nowhere,” Morrison says.

Searching for a lymphoma specialist, one of Morrison’s two sons identified Joshua Brody, an oncologist at the Mount Sinai Hospital in New York City — about 80 kilometres from Morrison’s home on Long Island. For almost 20 years, Brody has been studying the possibility of vaccinating people who have cancer against their own tumours.

The concept stems from the fact that cancers contain genetic mutations that make them molecularly distinct from healthy tissues — just as pathogens are distinct from human cells. It should, therefore, be possible to treat people with cancer using a vaccine, tailored to their tumour, that trains their immune system to better recognize and attack the malignant cells in their body.

These therapeutic vaccines have been pursued for decades, with many false starts. But several oncologists think that the strategy is finally about to be validated, thanks to personalized messenger RNA (mRNA) vaccines.

Based on the mRNA technology made famous during the COVID-19 pandemic, these individualized cancer vaccines are a technical tour de force. First, a sample of a person’s tumour is removed and fully genetically characterized to identify unique antigens — proteins that the immune system should respond to because it recognizes them as foreign. Then, machine-learning algorithms predict which antigens will generate the strongest immunity. Finally, mRNA that contains instructions for building 20–40 of the most promising antigens is incorporated into a unique vaccine, personalized to an individual.

These mRNA vaccines are progressing through increasingly large clinical trials, even though designing and producing a bespoke formulation for each person makes them complex and costly undertakings. But Brody is taking a radically different approach that — despite sounding almost paradoxical — generates an immune response that is highly personalized to the recipient’s tumour, even though it is made up of entirely pre-prepared, off-the-shelf components.

Called in situ vaccination (ISV), it involves delivering a combination of drugs to stimulate immune cells along with an intervention that kills cancer cells, such as radiotherapy, directly to an individual’s primary tumour.

The goal is for the radiotherapy to make dying cancer cells release their cargo of unique antigens. The stimulated immune cells then mop up these antigens and trigger a body-wide immune response. Just as with mRNA vaccines, recipients are immunized against their own cancers — it just happens through a different route. “A vaccine is whatever educates your immune system,” Brody says. “We are making the vaccine at the site of the tumour.”

Under Brody’s care, Morrison has participated in two early-stage clinical trials of ISV. He entered the first — which involved radiotherapy and two immune-stimulating drugs — when his lymphoma was a low-grade, ‘watch and wait’ cancer. For him, the intervention failed, but the stakes were low.

Morrison’s entry into the second trial was more urgent. His lymphoma had become more aggressive, unchecked by chemotherapy. This trial — for which Morrison finished treatment in January 2021 — included a third drug. Most participants in the trial did not see big benefits, but for him, it worked spectacularly. He entered complete remission. And although he still gets nervous ahead of each six-monthly check-up scan, he remains in remission today.

ISV is still in early investigational studies, with various research groups assessing which drug combinations achieve the strongest vaccinal responses1. Furthermore, the current need to inject drugs directly into tumours — to avoid dangerously activating the entire immune system — limits which hospitals offer it and precludes treating inaccessible tumours.

Joshua Brody talks to a woman in an office

Oncologist Joshua Brody (left) is pursuing in situ vaccines to treat cancer.Credit: Mount Sinai Health System

From a development perspective, this requirement to carefully administer a series of interventions, rather than a single — preferably systemically administered — product, poses a challenge. “There’s a problem with selling the whole concept to investors and to industry,” says Ronald Levy, a cancer researcher at Stanford University in California, who worked with Brody when he first started researching ISV.

But, because ISV has the potential to evoke a powerful anti-cancer immune response without having to create a personalized vaccine for every recipient, Levy, Brody and others continue to push this unusual approach to vaccination forwards.

Making an in situ vaccine

Cancer immunotherapy, in all its variations, is predicated on the immune system’s natural ability to destroy cancerous cells. This happens all the time, and routinely stops full blown cancer from developing. Cancers that do become established are those that have acquired mechanisms to escape the immune system — either by preventing the immune system from learning to recognize them or by stopping immune cells from attacking them. Therapeutic cancer vaccines aim to solve the recognition problem.

Most such vaccines follow the conventional method of first identifying appropriate antigens, then packaging them in a vaccine (see Nature 627, S34–S35; 2024). But in the 2000s, ISV, which typically operates without anyone ever knowing which antigens drive immunization, emerged as an alternative approach.

One foundational 2004 study2 was led by Sandra Demaria, who is now a cancer researcher at Weill Cornell Medicine in New York City. Demaria started with the observation that sometimes in animal models — and rarely in the clinic — radiotherapy does not just kill the directly irradiated cancer cells, but also triggers regression of distant metastatic tumours.

She hypothesized that radiation-killed tumour cells sometimes liberate cancer antigens that stimulate the immune system to fight the remaining, metastasized cancer. If this were true, Demaria and her team thought they should be able to boost this process.

Before irradiating the primary tumours of mice that had metastatic mammary carcinomas, the researchers treated some animals with a drug that activated a receptor called FLT3 — a growth stimulant for dendritic cells.

Modulating dendritic cells — a class of antigen-presenting cell — lies at the heart of ISV. These cells move through the body, ingesting antigens wherever they encounter them, before travelling to lymph nodes and presenting the antigens to cytotoxic T cells — which, when activated, can bind to and destroy cancer cells bearing the relevant antigens. For cancer therapy, it is crucial that dendritic cells stimulate cytotoxic T cells — which, when activated, can bind to and destroy cancer cells bearing the relevant antigens.

In Demaria’s control mice, only directly irradiated cancers were eliminated. But in mice that had received the drug to stimulate the production of extra dendritic cells, metastatic tumours in non-irradiated parts of the body also shrank.

To confirm this was a vaccinal effect, Demaria gave the FLT3 activator to mice with both breast cancer and lymphoma, then irradiated just their primary mammary tumour. If the mice were being immunized only against breast cancer, rather than having their immune function boosted generally, their lymphoma should be unaffected. And that’s exactly what happened. “It showed we can really use radiation to generate immune responses,” Demaria says, “you just need to add something to it.”

Shortly after Demaria’s work, Levy began conceptually similar animal experiments aimed at treating lymphomas. He, too, added a dendritic-cell-targeting drug to a conventional cancer-killing intervention. However, instead of using a growth factor, Levy’s group included a drug that directly activated dendritic cells3,4 to kill the primary tumour.

Dendritic cells contain numerous receptors that alert them to infections and inflammation, including toll-like receptors (TLRs). Stimulated TLRs propel dendritic cells into an activated state that is required for efficient antigen presentation. The drug Levy’s group chose acted on TLR9. The combination again caused tumours that were far away from the irradiated site to regress.

Iteratively does it

Emboldened by these results, Levy and Brody launched a clinical trial. The results were published in 2010. The trial tested the effect of radiotherapy and TLR9 activation on metastatic lymphoma in 15 people5. One participant entered complete remission, three had partial regressions and the disease stabilized in two more. A 2012 trial that focused on a different cancer type yielded similar results6.

These outcomes meant that other, more-conventional lymphoma therapies were more effective, but convinced Levy and Brody to continue pursuing ISV. The trial, Brody says, “showed pretty plainly that you can inject something into one tumour and make distant tumours melt away. And some of those remissions were pretty durable.”

By the time Morrison joined a trial in 2017, Brody’s approach had evolved to include two drugs plus radiotherapy — three elements for achieving what Brody sees as the three crucial actions on dendritic cells for successful ISV: “Mobilize, load, activate”7.

First, he explains, you must mobilize the cells. To this end, a FLT3-activating drug is given to an individual, to stimulate dendritic-cell proliferation. The second step is radiotherapy, to release the antigens that will load onto the waiting dendritic cells. Finally, a TLR3 activator is provided to further activate the antigen-bearing dendritic cells. Although this combination did not shrink Morrison’s tumours, one trial volunteer entered complete remission, and 2 more of the 11 participants had sizable regressions. Still, Brody thought he could do better — and he was right.

Before and after tumour biopsy images showing increase in CD8+ T cells

Pre- and post-vaccination tumour biopsies show an influx of immune cells (CD8+ T cells; brown).Credit: Ref. 7

First, in mice, he added a fourth element: an immune checkpoint inhibitor. This extra element would make all the difference for Morrison in the next trial. Checkpoint inhibitors suppress a mechanism that cancers use to stop T cells attacking them. These drugs have transformed the treatment of several cancer types, although most tumours don’t respond to checkpoint inhibitors alone.

Brody’s idea was to use checkpoint inhibitors after the ISV protocol, to allow the full potential of the vaccine’s effects to unfold. “You have to teach the T cells who’s the bad guy,” Brody says, “but even after you teach them, there’s a huge amount of nurturing required to get those T cells to do their job.” The ISV triple combination alone cured 40% of mice. Adding the checkpoint inhibitor doubled this to about 80% — the nurturing appeared to work7.

In his continuing human trial of this approach in lymphoma and breast cancer, only some participants have achieved significant remissions — out of ten recipients Morrison and two others saw marked improvements8. Still, Brody thinks it is a major step forwards. “These were tumours that were absolutely refractory to every type of chemotherapy, hormonal therapy, surgery and targeted therapies,” he says.

Adding checkpoint inhibitors and other drugs that boost T-cell function might turn out to be a crucial addition to all cancer vaccine. (Notably, most ongoing mRNA vaccine trials include checkpoint blockade, too.) Levy has now added an antibody drug — anti-OX40 — to his TLR9-based ISV approach that directly stimulates T cells. Human trials have run into difficulties, because, Levy says, most OX40 antibodies for human T cells do not yet work well, but Levy describes results in mice as “spectacular”. In experiments in which mice with breast cancer were treated intratumourally with a TLR9 activator and anti-OX40, more than 30% lived until the end of the six-month experiment following tumour resection, compared with less than two months for resected but untreated mice. Adding a checkpoint inhibitor increased survival further9. Levy pictures a clinical practice where people with operable primary tumours are vaccinated against their disease before undergoing surgery, “to leave behind the trained immune system to go after metastases — wouldn’t that be great?” he says.

Demaria is part of a team that is already exploring this strategy for breast cancer. Trial participants are first given a FLT3 activator, targeted radiotherapy and a checkpoint inhibitor, and then receive a mastectomy to remove their primary tumour (see Having to monitor the long-term incidence of post-mastectomy metastatic disease entails a long wait to see if the presurgical treatment makes a difference. But nearer term, Demaria says that by examining the tissue removed during surgery, she will be able to look for evidence of an active vaccinal process.

Small academic-led trials such as these could continue for some time. The mushrooming of immunotherapy research has led to the availability of more and more drugs that modulate dendritic and other immune cells10. Plus, radiotherapy is not the only way to make cancer cells release antigens — oncolytic viruses and certain forms of chemotherapy are also being tested in ISV protocols11.

But, overcoming the challenge of selling this approach to industry partners is arguably the most important next step for ISV to progress towards routine clinical care.

Next steps

“The proof-of-concept scientific phase has been promising,” says Christine Moussion, head of the cancer immunotherapy discovery group at the biotechnology company Genentech in South San Francisco, California. What’s needed next, she says, is for ISV to show its readiness for commercial development.

Genentech is testing a FLT3-targeting drug as a dendritic-cell growth stimulator, which could potentially be used in ISV12. But Moussion highlights three barriers to overall ISV development. First, she says, there are the practical difficulties of advancing multi-drug therapies: trials can quickly become complex, especially when custom-developed components must be characterized individually before being studied in ensembles.

Second, she says, not knowing which specific antigens are driving immunity makes it difficult to directly test whether ISV induces a T-cell response. Without such assays, trials will have to rely solely on clinical outcomes — and not surrogate endpoints such as the development of immunity — meaning they will probably take longer. Moreover, if clinical benefits are absent, researchers will be uncertain of whether the vaccination failed or whether something downstream prevented tumour regression.

The third issue is that these drugs must be delivered directly into tumours to limit systemic toxicity. This constrains the use of ISV to treating easily accessible tumours such as breast cancer and lymphoma — and makes treatment more complicated and probably more variable than administering drugs systemically.

Brody concedes that establishing immunity assays is challenging. But he and others think that intratumoural injections could soon be avoidable. The solution, they say, is to chemically link dendritic-cell-targeting drugs to antibodies or other molecular partners that accumulate inside tumours after being given systemically.

Such drug conjugates are in widespread development. In 2022, Levy and his colleagues completed work that showed that a TLR9-activating drug coupled to a peptide travelled to tumours in mice and shrank these cancers without systemic side effects13. “It avoids the whole logistical difficulty of tumour injection,” he says. So compelling were the results, a start-up company — TwoStep Therapeutics in San Carlos, California — was founded in 2023 to develop such conjugates.

And Moussion agrees with ISV’s pioneers that the approach has two key appeals. One is its off-the-shelf nature. Although the developers of the personalized mRNA cancer vaccines that currently hold centre stage see their bespoke approach as viable, creating bespoke therapies for every patient is expensive. The beauty of ISV, Brody says, is that the treatment is identical for everyone, but the result is still immunization against each person’s unique cancer.

What’s more, ISV theoretically exposes each treated individual to a much greater diversity of tumour antigens than the few dozen selected for personalized mRNA vaccines — including some malformed proteins that would not be detected by the genetic sequencing used for those vaccines. This could induce a stronger, wider-ranging immune response that powerfully attacks tumours initially and potentially reduces relapses.

For now, the remission experienced by Morrison and a handful of other individuals — all of whom had extensive metastatic disease — stands as testament to the potential of ISV. These positive results motivate Brody to keep going, and he is buoyed by increasing commercial interest. “Good ideas float to the top,” he says. “But there can be a lot of obstacles to them. So, they don’t always float quickly.”