Drugs that home in on and block cell growth signals can dramatically shrink tumors, but all too often the cancer comes back. An attention-grabbing mouse study 7 years ago suggested that these targeted drugs would work better at shrinking melanoma tumors if cancer patients got a break from their medicine every few weeks.

The first large trial to test this idea, however, has found that melanoma tumors grew back faster when patients were on such an “intermittent” dosing schedule. The study participants also lived no longer overall than those receiving the typical schedule of the drugs.

The disappointing results, to be presented today at part one of a virtual version of the American Association for Cancer Research’s annual meeting, throw cold water on a novel strategy for countering cancer drug resistance that is being tested in a number of clinical trials. “There were a lot of reasons to think this could work…It seemed like a very powerful idea,” says oncologist Alain Algazi of the University of California in San Francisco, who led the melanoma trial.

About half of melanoma patients have a gene called BRAF that contains a mutation producing an aberrant protein that drives tumor growth. Drugs blocking this mutant protein tend to shrink tumors for only a few months before resistant cells take over. This resistance is “a huge problem” not only for compounds known as BRAF inhibitors but also other targeted cancer drugs, Algazi says.

A mouse study published in Nature in 2013 suggested a simple but counter-intuitive solution. When researchers treated mice with melanoma with a BRAF inhibitor until their tumors started growing again, then took the drug away, the tumors shrank. Resistant cells that apparently had taken over the masses had become dependent on the drug. When mice were instead given the drug on an intermittent schedule—a few weeks on, then off, then on again—they lived twice as long as those getting the drug daily.

Such animal data, along with anecdotal reports of patients whose BRAF-inhibitor resistant tumors grew more slowly or shrank then they went off the drug, led the U.S. National Cancer Institute (NCI) to launch a 5-year intermittent dosing trial at multiple cancer centers in the U.S. and Canada. About 250 advanced melanoma patients received a two-drug combination of a BRAF inhibitor and a drug called a MEK inhibitor, which blocks another protein in the same growth pathway.

The 206 patients whose tumors were shrinking or stable after the first 8 weeks of daily treatment with this combo were split into two groups. One got the drugs daily until their tumors grew, while the other received them each day for 5-week stretches interspersed with 3-week breaks.

In the intermittent dosing group, patients’ tumors began growing again after just 5.5 months on average, compared to 9 months for those receiving drug continuously. “The results are the opposite of what we expected and all the preclinical data would have predicted,” says cancer researcher Antoni Ribas of the University of California, Los Angeles, co-leader of the trial. Both groups of patients, most of whom went on to get an immunotherapy drug if their tumors returned, lived about the same length of time overall—about 40% were still alive after 4 years.

Algazi’s team offers several possible explanations for why the mouse results didn’t hold up in people. Those getting intermittent therapy did not get as much drug overall as those in the standard therapy group (although that was true in the mice too). People metabolize the drugs more slowly than mice, and some animal studies suggest a sharper drop in drug levels may be needed for the drug-dependent tumor cells to die. In addition, “it’s possible there are different mechanisms of resistance in different patients,” says Algazi. (In the 2013 mouse study, tumor cells usually became resistant by ramping up production of BRAF protein.)

Algazi’s last explanation makes sense to radiologist Robert Gatenby of Moffitt Cancer Center. His team has reported that doctors can use intermittent therapy with hormone drugs for prostate cancer to extend lives — if the dose schedule is tailored to each individual patient by tracking their tumor growth and modeling its evolution. “Any fixed schedule such as the one in [the NCI trial] clearly does not incorporate the underlying evolutionary dynamics,” says Gatenby.

Other cancer researchers say the jury is still out on the NCI trial’s approach. “I don’t know that the hypothesis was adequately tested,” says cancer biologist Charles Sawyers of Memorial Sloan Kettering Cancer Center, who suggests the long half-life of the MEK inhibitor used in the trial may have been a problem.  Oncologist Mark Middleton of the University of Oxford co-leads a similar melanoma trial in the United Kingdom that is using a different dosing schedule; it is expected to have results later this year. This trial and others now underway, says Middleton, “will help us decide if there are particular circumstances that are needed for intermittent therapy to be of use, or if the idea is a bust.”