Even as a tumor is shrinking in response to a drug, some tumor cells may be undeterred by the drug's effects, eventually causing the tumor to grow back. By analyzing tumors of patients who have relapsed, researchers have discovered that tumors harbor different populations of cells carrying distinct therapy-escaping mutations, and that these vary from tumor to tumor and—even within a single tumor—can change over time. Eradicating tumors may therefore require combinations of drugs tailored to their changing mix of cells.

Recent research has focused on lung cancer, a leading cause of death globally that has been difficult to treat. Some types of lung cancer are driven by an overly active epidermal growth factor receptor (EGFR) and respond dramatically to drugs such as erlotinib (Tarceva; OSI Pharmaceuticals) and gefitinib (Iressa; AstraZeneca) that bind to EGFR and block its function. However, the response to these targeted therapies is, in most cases, short lived. Several studies have shown that about half of tumors that become resistant to erlotinib and gefitinib contain in their mix cells in which EGFR has a mutation (T790M) that prevents the drugs from binding properly. In other patients, some cells have turned on a different growth receptor, called MET, to allow them to grow when EGFR is blocked. Yet other cells use other mechanisms to escape treatment.

The picture beginning to emerge is that treating a cancer with a drug targeted to one type of cancer-causing mutation may impact the growth of the majority of cells, as intended, but also allows populations of cells with additional mutations, which may only have been present in small numbers in the original tumor, to grow and take over. Depending on the type of drug used in treatment, different cell populations will dominate. “There is a competition among populations of cells,” says Jeffrey Engelman of Harvard Medical School. By taking biopsies of a patient's tumor before, during, and after treatment, Engelman's group has shown that the cellular landscape of the cancer looks much different depending on the stage of treatment. “The implication is that we may have to continuously reassess a tumor,” he says. Researchers are now starting to test combinations of drugs that can target not only the mutation driving cancer growth, but also the tumor's potential resistance mechanisms, all at once. The idea of using combination therapy is not new, but now that researchers know many of the molecular drivers of cancer and of resistance, they can select drug combinations in a more rational way, according to William Pao, professor at Vanderbilt-Ingram Cancer Center in Nashville. The results of several clinical trials for combination therapies should start to become available this summer.