The big push in cancer treatment these days is to sample a person's tumor, test it for mutations, and give the patient a drug tailored to a genetic weak spot in the tumor. A new study suggests one reason why this targeted drug strategy doesn't always work. A solid tumor, it turns out, is not a mass of identical cancerous cells but a mosaic of genetically different cells that aren't captured with a single biopsy. Some of these distinct cells may be resistant to the targeted drugs, allowing a tumor to persist or grow.
The classic view of how cancer develops is that a single, normal cell accumulates mutations that eventually allow or force it to divide uncontrollably. This "clone" then grows into a tumor of identical cells, which can also sow seed cells into the bloodstream that then take root somewhere else in the body, or metastasize. The assumption that tumors grow out from a single clone has spurred a rush to find drugs that block one of the clone's genetic weak spots. But although the strategy has resulted in some very effective drugs—Iressa for lung cancer and a new melanoma drug called Zelboraf, for example—these drugs often stop working within a year or two. One reason could be that solid tumors already harbor a few cells, or clones, with "resistance" mutations that take over when the cells targeted by the drug are wiped out.
Now British researchers have firmed up this idea using next-generation DNA sequencing to explore the genetic landscape of tumors in unprecedented detail. Charles Swanton of Cancer Research UK's London Research Institute and University College London, James Larkin of The Royal Marsden Hospital in London, and collaborators took samples from nine different spots within a patient's 10-centimeter-wide primary kidney tumor and from some of the patient's metastatic tumors. The researchers then put these biopsies through a battery of genomics tests, including sequencing all protein-coding genes in each sample.
The researchers found a huge amount of variation within this one patient's tumors. Only 34% of the 118 mutations they detected showed up in all of the samples. Several key cancer genes were mutated across the main tumor, but in different ways in different areas. When the researchers examined which genes in the tumor were active—these so-called gene expression signatures are thought to predict whether a kidney patient's prognosis will be poor or good—the results differed depending on what part of the tumor they tested.
The team then created a family tree of how these various clones evolved from a single clone. The good news, Swanton says, is that this analysis revealed that some mutations in the "trunk" of the tree persisted as the cancer evolved, which meant that all of the patient's tumor cells had those weak spots. Consequently, drugs targeting those mutations should work even on very heterogeneous tumors—which could explain the initial success of Iressa and Zelboraf. However, other preexisting drug-resistance mutations in the branches of the tree might eventually block the drugs from working, Swanton says. His team reports its results today in The New England Journal of Medicine.
Carlo Maley of the University of California, San Francisco, who studies the evolution of cancer, cautions that the study was small—in total, the U.K. team tested only four patients' tumors and two in detail. Still, it is "strongly suggestive that intratumor heterogeneity should be expected," he says. "It adds a big level of noise and uncertainty to what everybody's doing," Maley says. Swanton said the study's bottom line is that as researchers develop new cancer therapies, they should be aware that single biopsies can be misleading: "We need to think about ways to sequence tumors in greater depth."