In December 2012, an 80-year-old man in Florida went to his doctor to have a bulge on his eyelid examined. Tests showed that the bump was a rare form of skin cancer called Merkel cell carcinoma. Despite treatment, the man’s cancer spread, first to lymph nodes under his jaw, then to lymph nodes in his abdomen. So in June 2015, he enrolled in a clinical trial for Merck’s pembrolizumab (Keytruda), a checkpoint inhibitor drug that blocks programmed cell death protein 1 (PD-1) to help the immune system more effectively target and kill tumor cells.
Six months into receiving the experimental treatment, the man developed an odd-looking nodule in his lung that didn’t resemble any of his other tumors. Doctors decided to biopsy it. To their surprise, the results came back positive for tuberculosis (TB), though the man didn’t have symptoms of a TB infection. In February 2016, he started anti-TB treatment, and fortunately, after a short break from the anticancer treatment, was able to finish the full courses of both therapies in 2017.
Meanwhile, a second patient, an immigrant from Vietnam, where TB is endemic, had enrolled in a trial for Bristol-Myers Squibb’s nivolumab (Opdivo), another anti-PD-1 drug progressing through clinical trials. He didn’t have such a happy ending: in June 2016, a month after starting the experimental cancer treatment, he developed a tuberculosis infection. A month after that, he died.
“I personally am a little concerned,” says Elad Sharon, a medical oncologist at the National Cancer Institute (NCI), which has been one of the government sponsors of clinical trials globally for PD-1 and PD-L1 blockade treatments. Doctors at Emory University, where the Florida man was being treated, alerted the NCI to the man’s condition because the agency was responsible for reporting any ill effects of the treatment to the US Food and Drug Administration (FDA). But the man’s experience wasn’t an isolated case, Sharon says. Reports from other clinical trials of anti-PD-1 treatments also showed tuberculosis infections cropping up in treated patients.
The FDA approved pembrolizumab and nivolumab as therapies for a variety of cancers, starting in 2014. Anti-PD-1 therapies are also being touted as potential treatments for non-cancer diseases such as Alzheimer’s. Given the enthusiasm to expand these drugs’ use, combined with the fact that the World Health Organization estimates that nearly a quarter of the world’s population has latent TB, Sharon’s concern seems justified.
You release the brakes on T cells . . . so TB disease occurs.—Olivier Lambotte, Paris XI University Medical School
“This is something that needs a lot more attention,” says Dan Barber, an immunobiologist at the National Institute of Allergy and Infectious Diseases (NIAID) who coauthored with Sharon a study describing the two cases of TB after anti-PD-1 therapy (Sci Transl Med, 11:eaat2702, 2019). Those results, and other reports of TB following immunotherapy, suggest that “we should probably be thinking about tuberculosis more during trials and during immunotherapy for cancer and actually see if we can try and get a feel for just how likely this is,” he says. “It could be a very rare event or it could be not so rare. . . . It’s not clear.”
The idea that tuberculosis might emerge after anti-PD-1 isn’t a total surprise. In 2010, William Jacobs Jr., a microbiologist at Albert Einstein College of Medicine in New York, and colleagues reported that mice genetically engineered so that they didn’t have PD-1 protein surface receptors died extremely quickly compared to healthy mice when infected with the bacterium that causes TB, Mycobacterium tuberculosis (PNAS, 107:13402–407). Essentially, Jacobs says, the study showed that the PD-1 pathway was crucial to controlling inflammation that builds in response to the bacterium. Barber’s work as a postdoc at NIAID revealed a similar result: after exposure to M. tuberculosis in the air, PD-1 knockout mice developed a lot of M. tuberculosis–specific CD4 T cells—which play a role in triggering adaptive immune responses—and those cells actually encouraged the replication of bacteria in the mice (J Immunol, 186:1598–607, 2011).
In humans treated for cancer who develop TB, the infections are probably not new, but “the escape of an old infection kept latent thanks to the PD-1 system,” says Olivier Lambotte, an infectious disease researcher at Paris XI University Medical School who has described a case of anti-PD-1–emergent TB but was not involved with Sharon’s and Barber’s work. He explains that PD-1, which is expressed on CD4 T cells, prevents excessive inflammation and immune activation in response to M. tuberculosis. With anti-PD-1 therapy, “you release the brakes on T cells . . . so TB disease occurs.”
Barber notes that the molecular interactions that lead to illness may be more nuanced. His recent studies in PD-1 knockout mice suggest that CD4 T cells in these animals are making a lot of a cytokine called interferon gamma (PLOS Pathog, 12:e1005667, 2016). “Interferon gamma production by CD4 T cells is [the sort of thing] that people think of as being the way the adaptive immune response suppresses tuberculosis,” he says. But his work has shown that that’s not the case at all. “The outcome of boosting interferon gamma production by T cells in a mouse model of TB is a dead mouse. Not better control of the infection.”
Up until the Florida patient presented with TB, there weren’t any data on how the human immune system responded to M. tuberculosis after anti-PD-1 treatment. Samples taken from the man’s blood and analyzed by Sharon’s group, however, showed that interferon gamma–producing CD4 T cells specific to M. tuberculosis were indeed more abundant after anti-PD-1 treatment. “We were able to really prove, at least to our minds, that the data were consistent with the data that [Barber] had previously put together in the mouse,” Sharon says.
Barber, Jacobs, and Lambotte agree that the data suggest that screens for latent TB should be done before a patient undergoes any anti-PD-1 treatment. But Sharon is a bit more hesitant. “We’re still thinking about it, in part because I think that we’re just not clear on how common this is,” he says, especially when it comes to the severe cases of the patients described in last year’s Science Translational Medicine study. Clinicians and scientists don’t usually monitor for TB in clinical trials—and it’s possible the nodules that develop from the infection could be mistaken for additional cancer growths.
Still, the results are “definitely telling us something unique about the immune response,” Jacobs Jr. says. And, it could lead to innovative treatments for TB, perhaps by using anti-PD-1 treatments to encourage persistent M. tuberculosis cells to grow so they can then be killed more quickly by TB drugs—an approach that could help reduce the ability of the bacteria to become latent in the body, he explains. “That’s an idea I intend to test.”