Cancer DNA tests give only a partial picture of the genes driving the disease, according to a study that throws cold water on the idea that scanning may quickly lead to highly effective personalized treatments.
Multiple scans on kidney tumors determined that mutations in each portion of the malignancy varied wildly, with only one-third of anomalies found in all parts, according to a report in the New England Journal of Medicine.
If confirmed in other cancer types, the finding suggests doctors and companies have only a limited ability to precisely identify individual mutations that can be targeted with drugs, since their tests are typically based on a single biopsy sample. It may also explain why cancer becomes resistant to gene-targeted medicines, the U.K-based investigators said.
The result points to “a serious flaw in the imagined future of oncology,” wrote Dan L. Longo, a cancer researcher and deputy editor of the journal, in an accompanying editorial.
The researchers also performed an experimental prognostic test that looked at levels of 110 genes on different samples from a patient’s cancer, according to the report. When the test was done on one part of a tumor, the disease was judged to have a favorable prognosis. When the same scan was performed on another section, the cancer had an adverse prognosis.
Devising nonstandard drug regimens based on the result of a single gene test “should be applied with great caution until more is known,” said Charles Swanton, an oncologist at Cancer Research U.K. London Research Institute and the senior study author. It could lead to harm if the patient gets prescribed a drug that hits a gene mutation that turns out not to be present in most of his tumor, he said.
The tumor variability could also explain why cancer becomes resistant to gene-targeted drugs, according to Swanton. The resistance mutations may be present from the start in some parts of the tumor.
Barrett Rollins, chief scientific officer of the Harvard University-affiliated Dana-Farber Cancer Institute in Boston, said the finding doesn’t undercut current practice. While variability is an issue, many cancers now respond to drugs targeted at specific genetic features, he said.
“I’d be disappointed if the inference from this study is that targeted therapies are useless,” Rollins said in a telephone interview. “They’re not useless, they’re just not useful enough.”
Cancer genetics is so complex that researchers may never completely grasp it, Rollins said.
“Until we understand that whole universe of mutations, we’ll have to treat what we find and test to see whether it works,” he said.
‘Better Tools’ Needed
Brian Druker, the doctor and cancer researcher who helped develop Gleevec, an early gene-based leukemia medicine made by Novartis AG of Switzerland, said the analysis shows the need for “better tools to identify cancer at earlier stages, when it is far less complex.” Druker directs the Knight Cancer Institute at the Oregon Health & Science University in Portland.
Personalized medicine “is going to be much harder than we hoped,” said Andrew Futreal, a study co-author who is starting a new job at the University of Texas MD Anderson Cancer Center in Houston later this month, in a telephone interview.
“If you stick a needle in the right side of the tumor, you could miss a key mutation in the left side,” he said.
One crucial question is whether lung cancer, breast cancer, and other common cancer types are as variable as kidney cancer was found to be, said Futreal, who did the research for the tumor variability study in his previous job at the Wellcome Trust Sanger Institute near Cambridge, England.
Ongoing large-scale cancer genome studies could provide clearer answers in a year or two, he said.
Begun in 2010
The latest research began in 2010 when Swanton decided to investigate previous research hinting that individual tumors were genetically variable, he said. That could mean scanning from a single biopsy wouldn’t produce results reliable enough to guide therapy, he said.
Swanton and colleagues from other U.K. hospitals and research institutes obtained tumor specimens that had been removed from four advanced kidney cancer patients. They then used DNA sequencing machines from San Diego-based Illumina Inc. in the U.S. and other tests to evaluate genetic variability.
Instead of evaluating the DNA from just one tumor sample, the typical practice today in hospitals, the researchers scanned eight sections of the first patient’s primary tumor, and four more sections from the patient’s metastases, malignant cells that have escaped the main nesting area.
They found that only 40 of the 128 total gene mutations found were present in all of the tumor samples.
“It was disturbing, sobering,” said Swanton, who is also affiliated with University College London Hospital, in a telephone interview. “There were overall more differences than similarities between biopsies.”
A variety of DNA and chromosome tests on samples from three other kidney cancer patients found the same pattern, he said.
In the future, researchers and physicians will have to carefully validate cancer gene findings to make sure they understand which mutations are likely found everywhere inside a tumor, Swanton said.
Sequencers made by Illumina, Life Technologies Corp., of Carlsbad, California, and other companies are allowing scientists to analyze large swaths of DNA from numerous tumors for the first time. Biotechnology companies are also selling and developing tests that analyze tumor cell DNA to determine how fast cancers will grow, or whether they may be treatable with available medications.
Scientists at these companies are well aware that different tumors in the same patient, and different parts of those tumors, may harbor a variety of genetic mutations, said Michael Pellini, president and chief executive officer of Foundation Medicine, a cancer diagnostics company in Cambridge, Massachusetts.
“As an industry, we have to make sure we have very robust tests,” he said in a telephone interview. Among other changes, techniques for performing biopsies, the sampling of diseased cells to characterize tumors and guide treatment, can be improved to make sure cells that best represent the entire tumor are analyzed, Pellini said.
Druker, the Gleevec developer, said the finding highlights a problem that may soon be fixed.
“As the price of sequencing plummets, it will actually be possible to sequence multiple areas of a tumor, or primary and metastatic tumors from the same patient,” Druker said.
A second positive finding that can be drawn from the report, he said, is that “while the primary tumor diverges, there appears to be a convergence of mutations along specific pathways. If we identify the right pathways to target we should be able to use combination therapies to improve outcomes.”
Currently, Foundation Medicine charges $5,800 to search DNA from patient tumors for cancer-related changes that may be treated by existing drugs.