Why do so many cancer drugs show promise in early trials, only to fade away into obscurity? That’s a question that has frustrated doctors, scientists, and patients alike for far too long. A new set of research out of Houston may finally have some answers. Researchers from UT Health say they’ve figured out why some cancer drugs show effective cancer-killing properties in mouse models, but fall flat when it comes time to help a human patient.
To investigate this conundrum, the research team analyzed a group of patient-derived xenografts (PDX), which is just the technical term for when human tumor tissue is implanted into an immune-deficient mouse. PDX models are frequently used in the testing and development of new cancer drugs.
“What we found is that when you put a human tumor in a mouse, that tumor is not the same as the tumor that was in the cancer patient,” says senior author W. Jim Zheng, Ph.D., professor at the School of Biomedical Informatics, in a release. “The majority of tumors we tested were compromised by mouse viruses.”
Among a total of 184 analyzed PDX models, 170 contained mouse viruses. Study authors say these viruses are “associated with significant changes” in tumors, essentially rendering any medical findings with infected models unreliable.
“When scientists are looking for a way to kill a tumor using the PDX model, they assume the tumor in the mouse is the same as cancer patients, but they are not. It makes the results of a cancer drug look promising when you think the medication kills the tumor – but in reality, it will not work in human trial, as the medication kills the virus-compromised tumor in mouse,” Zheng explains.
The entire research team believe their findings should lead to major changes in how cancer drugs are developed and tested moving forward.
“We all share the common goal of hoping to find a cure for cancer. There are 210 ongoing NIH-funded projects relevant to PDX models, with a combined annual fiscal year budget of over $116 million. We need to tighten up quality control and use models that are not compromised so that the treatments we give to future patients are effective,” Zheng adds.