A lung cancer (arrow) grew slowly before the patient received an immunotherapy drug, as seen in images made 7 weeks apart (left and center). Three weeks after the treatment, it had ballooned.

S. CHAMPIAT ET AL., NATURE, VOL. 15, 751, (2018) ADAPTED BY C. AYCOCK/SCIENCE

New drugs that unleash the immune system on cancers may backfire, fueling tumor growth

Although the 65-year-old woman had a rare type of endometrial cancer that had spread to her liver and was expected to be fatal, she still felt well enough to work and swim. As a last hope, her doctors gave her a type of immune-stimulating drug that had had near-miraculous results in some patients with advanced cancer. But 3 weeks after she began the drug, the woman's liver tumors had grown, and her abdomen was swollen with tumors as big as oranges. "She just exploded" with tumors, says Razelle Kurzrock, an oncologist at the University of California, San Diego (UCSD), whose team treated the woman. "She was going to die anyway, but unfortunately we hastened her death."

That patient was one of a growing number of fast-progressing cancer cases that point to a dark side of so-called checkpoint inhibitors. These antibody drugs, which have put some people with advanced cancer into remission for years, block a tumor protein, PD-L1, from suppressing T cells by activating PD-1, one of their surface proteins. But only about 20% of patients respond long-term to the drugs, and Kurzrock and other oncologists in the United States and Europe are warning that in a few, they may stoke tumor growth.

Despite accumulating papers and anecdotal reports of such tumor "hyper-progression," some cancer researchers wonder whether it is simply an illusion—whether the patients' tumors were destined to grow rapidly even before checkpoint inhibitor treatment. "We are divided into believers and nonbelievers," says oncologist Marina Garassino of the Istituto Nazionale dei Tumori in Milan, Italy, who is among those gathering to discuss the issue next week in Atlanta at the annual meeting of the American Association for Cancer Research (AACR). At a session organized by the U.S. Food and Drug Administration and the National Cancer Institute (NCI), researchers will try to resolve how to pin down whether hyperprogression is real and if so, how to identify which patients should not receive these increasingly popular drugs.

Concerns first arose in late 2016 when researchers at the Gustave Roussy Institute in Paris coined the term hyperprogression to describe 12 of 131 cancer patients whose tumor growth rate doubled within 3 months of anti–PD-1 treatment. In March 2017, Kurzrock and and UCSD's Shumei Kato reported similar findings for six of 155 patients. At least a half-dozen groups have since reported evidence of checkpoint inhibitor–triggered hyperprogression, with rates ranging from 7% for several cancers to 29% for head and neck tumors.

Many cancer researchers caution that this rapid growth could have been the natural course of the patient's disease. "Anybody who's treated cancer patients, particularly with more severe cancers, has had situations where people suddenly go south," says Elad Sharon, an oncologist in NCI's Investigational Drug Branch in Bethesda, Maryland. One problem, he notes, is that groups are defining hyperprogression in different ways. At the AACR meeting, panelists will explore how to develop a standard definition that could be used across the field. Another reason for skepticism is that nobody has yet found a biological explanation for how the anti–PD-1 drugs could spur tumor growth. "As scientists we want a mechanism," Kurzrock says.

Kurzrock noted a clue in her patients with hyperprogression: Most had either mutations in EGFR or extra copies of MDM2 or MDM4, all known cancer genes. Although only some of the other teams studying hyperprogression have confirmed these findings, scientists working with Kurzrock are using cancer cell lines and mouse models that have alterations in these genes to see whether treating them with a checkpoint inhibitor somehow triggers the release of growth-promoting molecules.

Garassino and colleagues suspect a key role for immune cells called macrophages, which are often found within and around tumors and can suppress anticancer immune responses. Among 187 patients with lung cancer, all 39 who seemed to hyperprogress when they received anti–PD-1 drugs had unusually high numbers of a specific type of macrophage in their pretreatment tumor tissue, her team reported online in September 2018 in Clinical Cancer Research. Garassino's team also put either lung cancer cells or bits of tumor from some of those 39 patients into mice lacking T cells, then treated the rodents with an anti–PD-1 drug. Their tumors grew faster than in mice that didn't get the drug, and as in the patients, were swarmed with macrophages, suggesting the checkpoint inhibitor switched the cells into an immune-suppressing mode that favors tumors.

None of this work, however, has nailed down a specific mechanism, notes Jean-Charles Soria, a co-author on the original 2016 hyperprogression paper who is now at AstraZeneca, which makes a PD-L1 inhibitor. "We are working on finding baseline clinical and biological variables that can allow us to identify patients who may experience hyperprogression" and distinguish them from those who aren't benefiting for other reasons, he says.

If researchers can identify genetic changes or other biomarkers that signal a high risk of hyperprogression, patients who test positive for those markers could be advised to avoid PD-1 inhibitors in the first place. That is already Kurzrock's practice with her own patients when their tumors carry MDM2 amplifications. Or patients might be given a second drug to counter any growth-promoting effects. "We would love to be in a situation where if this is happening, we could identify on an initial biopsy what the risk is and then consider alternative therapies," Sharon says. First, however, the cancer field needs to unite on whether such drug-fueled explosions in tumors are indeed real.