A small pilot study raises hopes that personalized cancer vaccines might prove safer and more effective than immune-based therapies already in use or further along in development. In a paper published online in Nature on Wednesday, scientists reported that all six melanoma patients who received an experimental, custom-made vaccine seemed to benefit: their tumors did not return after treatment.

Researchers not involved in the study praised its results, but with caveats. The scientists “did a beautiful job,” said MD Anderson Cancer Center’s Greg Lizee, an expert in tumor immunology, who called the results “very encouraging.” But because the study did not include a comparison group of patients who received standard treatment and not the vaccine, he cautioned, “it’s not completely proved yet that the lack of [cancer] recurrence was due to the vaccine.”

The first cancer immunotherapies were drugs such as pembrolizumab (Keytruda) and ipilimumab (Yervoy), which interfere with molecules that block immune cells from attacking a tumor. But that’s an effective strategy only if any such immune cells are trying to reach the tumor; if not, clearing a path for something that’s nowhere to be found doesn’t help. That’s a big reason why immunotherapy drugs help in only a minority of cancers: For many tumors, immune cells are AWOL.

The second generation of cancer immunotherapies is expected to consist of living cells. Called chimeric antigen receptor T cells, they are genetically engineered to dangle a molecule that lets them grab onto a matching molecule on a tumor cell and kill it. But these CAR-T cells can incite such an immune storm that they’ve killed patients in clinical trials. And although it’s possible to engineer them to hook up to more than one matching molecule, or antigen, most CAR-Ts in development target only one. As a result, if tumor cells jettison that antigen, they might be able to evade the CAR-T cells.

Read more: Startup Spotlight: Making cancer vaccines as unique as a patient’s cancer

The shortcomings of immune-oncology drugs and CAR-Ts have persuaded some scientists to look for a better way. That’s led them to study neoantigen vaccines, which are created by taking tumor cells and sequencing their DNA. Scientists then look for mutations in the tumor DNA which are not found in healthy cells’ DNA. They use algorithms to identify which mutations make antigens (those surface molecules) that immune cells are most likely to grab, and load the vaccine with pieces of the antigens, called long peptides.

In the pilot study, scientists led by Dr. Catherine Wu of Dana-Farber Cancer Institute in Boston packed millions of copies of 13 to 20 different peptides into a vaccine; the number depended on how many antigens likely to attract T cells a patient’s tumor had.

Although ideally the immune system would attack these antigens without outside help, in reality they don’t. But flooding the body with antigen-like peptides prodded the immune system to produce T cells that swarm the tumor cells like angry hornets.

Patients got five shots in the first month and booster shots at 12 and 20 weeks. The strategy of selecting antigens on tumor cells but not healthy cells apparently worked: the T cells didn’t attack healthy cells. (Side effects included mild flu-like symptoms, redness at the injection site, rash, and fatigue.)

After two years, four patients, whose melanoma had spread to lymph nodes (called stage IIIB/C) before they were vaccinated, were cancer-free. In two patients whose melanoma had spread to the lungs (stage IVM1b), tumors returned but, after subsequent treatment with Keytruda, cancer was undetectable.

And two years out, the patients’ blood still carried anti-melanoma T cells, suggesting the vaccine’s benefits last. “The specificity of the T cells had been maintained,” Wu said.

Six patients is not enough to declare victory; sometimes melanoma keeps from recurring by itself, Lizee pointed out. In addition, two patients initially enrolled in the study had tumors with so few mutations it wasn’t possible to craft a vaccine for them. That suggests not every cancer patient has a shot at benefitting from a neoantigen vaccine.

The results for the remaining patients were nevertheless encouraging enough to justify a larger clinical trial, Dr. Cornelius Melief, of Leiden University Medical Center in the Netherlands and an expert on cancer immunotherapy, wrote in a paper accompanying the study. That should be conducted on “any type of cancer that has enough mutations to provide sufficient neoantigen targets,” he wrote.

Neon Therapeutics, a Cambridge, Mass.-based company that Wu and some of her co-authors founded in 2015, is already running a larger clinical trial of the neoantigen vaccine NEO-PV-01, used in the melanoma study, for bladder and lung cancers in addition to melanoma. With $125 million from private investors, including players such as Third Rock Ventures, the company “is well-funded to take this work forward to completion,” a spokesman said.

Several of Wu’s co-authors are also Neon employees or advisers, and have filed for patents on different elements of neoantigen vaccines. Dana-Farber, independently of Neon, is conducting clinical trials of neoantigen vaccines against glioblastoma (a brain cancer) and ovarian cancer, among others, Wu said.

Making a neoantigen vaccine for a single patient, including sequencing and analyzing tumor DNA, is expected to cost around $60,000, meaning the vaccine would likely be priced in the low-to-mid six figures. Wu said she’s optimistic that as the costs of DNA sequencing and antigen production fall, so will overall costs. If some patients also require drugs such as Keytruda, that would add another $150,000 for a year of treatment.

Republished with permission from STAT. This article originally appeared on July 5, 2017