Of genes and cancer
By Taylor McNeil
It’s a startling statistic: some 170,000 people in the United States are diagnosed with lung cancer each year, and within five years of diagnosis, 85 percent of them are dead. That mortality rate hasn’t changed since the 1970s, in part because lung cancer is hard to diagnose early. “There is an epidemic of lung cancer in the United States,” says Avi Spira (ENG’02), a pulmonary and critical care medicine physician and School of Medicine researcher. “It’s the number-one cause of cancer deaths. And the reason for the high rates of lung cancer is smoking — about one quarter of the adult U.S. population are active smokers.”
A few years ago while on rounds at Boston Medical Center, it occurred to Spira that there must be a way to improve lung cancer diagnosis rates. He had one asset that most doctors don’t: a recent master’s degree from the College of Engineering in bioinformatics, a new discipline that uses computers and computational approaches to understand complicated biological systems.
Spira wanted to use these new bioinformatics techniques to figure out how smoking causes lung cancer. A related question, he says, was, “Could we figure out which smokers are the ones more likely to get it, and therefore identify them and diagnose them at an early stage, before the cancer has spread, and be more likely to provide them with a cure?” Spira applied to the Doris Duke Charitable Foundation for a Clinical Scientist Development Award, an award given to doctors early in their clinical research careers, and was one of 10 recipients chosen nationally in 2002. With the five-year, $500,000 award, he and his research mentor, Jerome Brody, a MED professor of medicine, set up a research program initially with participants from Boston Medical Center.
What they found exceeded Spira’s expectations. Examining a group of healthy smokers and a group of healthy “never smokers,” they discovered that most of the healthy current smokers activated the expression, or production, of a set of detoxification-related genes in response to the cigarette toxins, but these same genes were inactive in a small subset of current smokers, one of whom went on a year later to develop lung cancer. “We believe there is an appropriate response that 85 percent of people who smoke mount when they are faced with this potential carcinogen, but that there is a small subset of smokers with an inherited susceptibility” to not having those genes working correctly, Spira says. “There are 10 or 15 percent of smokers who have some abnormality in their DNA coding these genes resulting in an inappropriate response to the toxin and that may be one of the reasons why they go on to develop lung cancer.” That number matches what prior epidemiological studies showed: 10 to 15 percent of all smokers develop lung cancer. (Spira quickly adds that smoking creates additional risks: other cancers, emphysema, and heart disease, for example.)
For the test group, samples were taken with a brush from the patients’ air passageway, or bronchus, in a procedure called a broncoscopy, a relatively noninvasive test, far simpler than a lung biopsy. The air passageway is exposed to toxins in smoke just as the lungs are, and Spira reasoned that the genetic response should show up in the samples. Researchers used microarrays, small glass slides containing some 22,500 genes identified by the Human Genome Project, to see which were affected by smoking. Spira and Brody, who also directs the Pulmonary Research Center, found approximately 100 genes that are altered by cigarette smoke, some of which might be markers for susceptibility to lung cancer.
A ticking clock
A group of former smokers was included in the study, too, people who quit anywhere from one month to 30 years before the test. “Surprisingly, changes in most genes are reversible,” Spira says. “Most of the changes do revert back to normal, though it takes about two years. But some of the genes, including some of the genes we think are important in cancer, such as known tumor-suppressor genes, remained altered up to 30 years after quitting smoking.” That matches epidemiological data showing the risk of lung cancer does not revert back to normal in former smokers even 20 years after quitting. The findings were published in the Proceedings of the National Academy of Sciences in July 2004 and received major media attention and calls from other researchers eager to collaborate.
From the first small, urban sample, Spira and his colleagues are now doing more extensive sampling of smokers — with and without lung cancer — from Boston suburbs, Trinity College in Ireland, and two other hospitals in Boston. “These studies are aimed at identifying genes that are differentially expressed in the airways of smokers with lung cancer as compared to smokers without lung cancer, and thus may serve as diagnostic biomarkers for lung cancer,” Spira says. If the results continue to match preliminary findings — that a set of 50 to 100 genes can distinguish smokers with and without cancer with a diagnostic predictive rate of almost 90 percent for lung cancer — Spira and his colleagues will begin larger, multicenter trials and seek the FDA’s approval of the gene expression markers as a diagnostic tool. Spira and Brody are working with Affymetrix, a biotechnology company that makes the microarrays and is a partner with Spira in the development of the new diagnostic tools.
The benefits for the 10 to 15 percent of smokers with this gene defect could be enormous, Spira says. “If we could pick up people before they got the cancer, when they’re in their 20s or 30s and just starting to smoke, and look at their genetic samples and say, ‘You’re not responding right to this toxin, and you’re on your way to getting cancer,’ that’s very powerful,” he says.