Immunotherapies have transformed cancer treatment by helping the immune system recognize and attack tumors. They work for only about 20% of patients, though, and doctors still struggle to predict who will benefit.
A Rutgers Cancer Institute study in Cell Reports Medicine promises help with both those problems. It identifies a protein that appears to predict drug response, and it shows that pairing immunotherapy with a second type of drug dramatically improves survival in mice.
In those mice experiments, roughly half the animals receiving the two-drug combination survived to the end of the 60-day study and showed no signs of side effects. With either drug alone, only approximately 20% of mice survived that long, and none of the untreated mice did.
“The effect was substantial compared with single drug treatment,” said Zhaohui Feng , the senior author of the study and a professor at Rutgers Cancer Institute, New Jersey’s only National Cancer Institute-designated Comprehensive Cancer Center, together with RWJBarnabas Health.
The protein at the center of the study is an enzyme called PHGDH. It helps cells produce the amino acid serine, a building block that fast-growing cancer cells consume in large quantities to fuel their rapid growth. PHGDH is found at abnormally high levels in roughly half of all colorectal cancers and 40% of breast cancers and drug companies have been developing compounds to shut it down.
The Rutgers team, led by Feng and first author Juan Liu , an assistant professor at Rutgers Cancer Institute, discovered that PHGDH does more than its known metabolic role: It also ramps up production of a protein called PD-L1, which allows cancer cells to evade the immune system. PD-L1 is the specific target of widely used immunotherapy drugs, which work by blocking PD-L1’s ability to shut down immune cells.
Crucially, the researchers found that PHGDH’s effect on PD-L1 is independent of its metabolic function. When they tested versions of the enzyme genetically altered so they could no longer produce serine, those dead enzymes still drove PD-L1 levels up.
“We found PHGDH can increase PD-L1 levels, and it is independent of the metabolic enzyme function – it is not related to serine,” said Liu. “It’s not the traditional function of the metabolic enzyme.”
That distinction matters for treatment. Experimental drugs now in development to block PHGDH work by shutting down its metabolic activity and cutting off the tumor’s serine supply. However, because those drugs don’t eliminate the PHGDH protein itself, they cannot prevent it from helping tumors evade the immune system. The researchers reasoned that adding immunotherapy to attack the problem from both sides – one drug to cut off the tumor’s fuel and another to strip away its immune disguise – should work better than either alone. The mouse experiments bore that out.
Combination therapy in humans would require the U.S. Food and Drug Administration to approve a PHGDH inhibitor, which hasn’t yet happened, followed by clinical trials testing the two-drug approach. But the study’s results could affect patient care sooner through a different route. The researchers analyzed data from cancer patients who had already received immunotherapy and found that those whose tumors had high levels of PHGDH responded significantly better. If confirmed in a large prospective trial, measuring PHGDH levels could help oncologists decide which patients are most likely to benefit from drugs that bind to PD-1 or PD-L1.
“Those patients who have high PHGDH are more sensitive to anti-PD-1 therapy,” said Feng, who also holds appointments at the medical school and the Rutgers Institute for Infectious and Inflammatory Diseases . “This could be a potential biomarker to predict the efficacy of checkpoint inhibitor therapy.”
The research received support from the National Institutes of Health, the New Jersey Commission on Cancer Research, the New Jersey Health Foundation and Ludwig Research Support.
Cell Reports Medicine
Experimental study
Animals
Targeting the noncanonical function of metabolic enzyme PHGDH in driving PD-L1 expression and cancer immune evasion
28-Mar-2026