Skip Navigation
National Cancer Institute U.S. National Institutes of Health
Personal tools
You are here: Home News Study Raises Concerns about Using Cancer Cell Lines to Test Drug Resistance
Document Actions

Study Raises Concerns about Using Cancer Cell Lines to Test Drug Resistance

Cancer cells grown in culture and stored for use in laboratory studies—known as cell lines—are the workhorses of basic cancer research. A new study from NIH and NCI researchers suggests, however, that established cancer cell lines may have important limitations when used to investigate resistance to chemotherapy drugs. The findings indicate the need for better in vitro models of cancer that more accurately represent how tumors behave in the body, they said.

In the study, published November 15 in the Proceedings of the National Academy of Sciences, Dr. Michael Gottesman, NIH deputy director for Intramural Research, and his colleagues showed that, in various cancer types, the expression of a specific set of genes associated with drug resistance was very different in cell lines than it was in tumor samples representing the same cancer types.

The study included cell lines from NCI's Human Tumor Cell Line Screen, commonly called the NCI-60 panel, a collection of 60 cancer cell lines derived from 9 common tumor types. Cancer cell lines from the NCI-60 panel and other sources are used for many types of studies, such as testing experimental drugs for anticancer effects or investigating how specific genes influence tumor development.

The researchers focused on 380 genes associated with drug resistance. They compared the activity of these genes in cancer cell lines representing six tumor types with the activity of the same genes in clinical samples of the same six tumor types. There was "no correlation between clinical samples and established cancer cell lines," the researchers found.

In addition, when the researchers profiled the activity of these genes associated with drug resistance across all the cell lines of the NCI-60 panel, they made "the striking observation that all of the cell lines…bear more resemblance to each other, regardless of the tissue of origin, than to the clinical samples that they are supposed to model," the researchers wrote.

"We thought we might see some similarities among the cancer cell lines," said the study's lead author, Dr. Jean-Pierre Gillet of the Laboratory of Cell Biology in NCI's Center for Cancer Research. But the extent of the differences in the activity of these drug-resistance genes between the cell lines and tumor samples was "impressive" and "definitely surprising," Dr. Gillet added.

In cell lines, the researchers found changes in gene expression that may be the result of "selection pressure and culture conditions" that the cells acquired during extended growth in culture so they can thrive in an artificial environment. These same genes are also involved in drug resistance. "In other words, the cancer cell lines are highly selected during their establishment for expression of genes associated with [multidrug resistance]," the researchers wrote.

The research team initially focused on ovarian cancer. They compared the gene expression profile of the drug-resistance genes for 15 ovarian cancer cell lines—including five from the NCI-60 panel and 10 other commonly used lines—with the profiles of 80 tumor samples obtained from patients before treatment began. Not only were the cell line and tumor sample profiles "strikingly different," the researchers reported, but the ovarian cancer cell line profiles were similar.

In various cancer types, the expression of a specific set of genes associated with drug resistance was very different in cell lines than it was in tumor samples representing the same cancer types.

The research team also found that the drug-resistance profiles of the other cell lines in the NCI-60 panel were very similar to those of the ovarian cancer cell lines. Moreover, the profiles of the cell lines for five other cancer types were also different from those of the tumor types they are meant to model.

These findings should be interpreted with caution, noted Dr. Cyril Benes, director of the Center for Molecular Therapeutics at Massachusetts General Hospital. "The functional significance" of the similar gene expression profiles, or clustering, "is unclear," Dr. Benes wrote in an e-mail. "The transcriptional profile is different, but does that translate into an important biological outcome in terms of drug response?"

No one believes that cancer cell lines are perfect, said Dr. Edward Sausville, associate director for clinical research at the University of Maryland Greenebaum Cancer Center. Other studies have hinted at these types of issues with cell lines, and it's not surprising that cells grown in a plastic dish and stored for long periods evolve to handle the stress of the conditions, he continued.

"Whether it's a set of cell lines or other models, they're not one size fits all," Dr. Sausville said. "They are good for some things and not good for others."

Studies like this one, that molecularly characterize the NCI-60 panel and other commonly used cell lines, are welcome and needed, said Dr. Susan Holbeck of NCI's Developmental Therapeutics Program, the program through which the NCI-60 panel is managed. But even with their shortcomings, she stressed, cancer cell lines are still highly valuable tools.

For example, some of the early suggestions that the targeted therapies vemurafenib (Zelboraf) and crizotinib (Xalkori)—both of which were recently approved by the Food and Drug Administration—may be highly effective in patients with mutations in the BRAF and ALK genes, respectively, came from studies of cell lines with those mutations, she noted.

"Each cell line can be seen as a bag full of molecular targets," Dr. Holbeck continued. "They're excellent tools for generating hypotheses that can then be tested in other models," such as genetically engineered mouse models that more closely mimic how tumors develop and progress in humans.

Studies like this "are certainly valuable," Dr. Benes said, in part because they demonstrate that the entire undertaking of modeling human cancer "is a learning process…. Understanding what predicts translation from results in models to patients is also an important part of it, as well as making different, possibly better models."

Carmen Phillips

Source: NCI Bulletin, November 29, 2011