On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Ovarian cancer is relatively rare, ranking as the eighth-most frequent cancer, but it is the fifth-leading cause of cancer deaths among U.S. women. It is disproportionately deadly because ovarian tumors tend to flourish while producing few obvious symptoms. And no reliable methods exist to detect the cancer at early stages, when treatments are most effective. But this situation may soon change if researchers can extend the promise of a recent study, in which scientists detected ovarian cancer from blood samples with near 100 percent accuracy.
Researchers at the Georgia Institute of Technology collected blood from 44 women diagnosed with ovarian cancer at various stages, along with that from 50 healthy controls. After removing red blood cells and clotting factors, they analyzed the remaining serum to assess each patient's metabolome—the body's unique chemistry, which results from everyday cellular processes and is made up of hormones and by-products of chemical reactions. A machine called a mass spectrometer separated the 20,000 types of metabolites and quantified the amount of each one. Computer algorithms then identified features of the metabolic profiles that distinguished cancer patients from controls.
The team used this information to build a predictive computer model capable of classifying the cancer status of unknown samples. In its first test the model flawlessly distinguished between cancer and control samples, with no false-positives or negatives. A second evaluation showed the model was 99 percent accurate—it incorrectly placed only one control sample in the cancer group. The results were published online August 10 in Cancer Epidemiology, Biomarkers & Prevention.
Cancer cells may produce unique metabolic profiles, in part because they grow very rapidly and have metabolic activity very different from normal cells. For years scientists have thought that metabolites might be a good way to detect disease, says biologist John McDonald, co-author of the study, but "the technology is what has been holding people back." McDonald explains that his team's new approach requires less preprocessing of samples, making it "more plausible that it can be done on a high-throughput basis, at a reasonable cost."
The National Cancer Institute estimates that at least 21,000 new cases of ovarian cancer will be diagnosed in the U.S. in 2010, and nearly 14,000 women will die of the disease, making it the deadliest of the female reproductive cancers. If, however, physicians spot the disease early—before the cancer spreads to other organs—the five-year survival rate exceeds 90 percent. Early detection is hampered by the cancer's vague and nonspecific symptoms, such as abdominal pain, bloating, back pain and fatigue. The malignancy often remains undiagnosed until the cancer has spread, leaving patients with a bleak prognosis and a five-year survival rate of less than 30 percent.
Although imaging techniques such as ultrasound or CT scans may aid the diagnosis, neither is very good at distinguishing between benign growths and cancer. A blood test for CA-125, a protein that is elevated in some women with ovarian cancer, is frequently used to monitor tumor recurrence, but the test lacks sufficient accuracy for screening purposes. "Part of the reason for that," McDonald says, "is that cancer is not a single disease—there's a lot of variability. A given patient may have more of one protein, while another won't."
MacDonald's approach, which effectively monitors 20,000 biomarkers in the form of metabolites, represents a culmination of sorts in recent thinking in biomedicine. "This finding follows the general theme, emerging over the past decade, of identifying patterns of biomarkers in [bodily] fluids such as blood, saliva and tears," says Emanuel Petricoin, co-director of the Center for Applied Proteomics and Molecular Medicine at George Mason University, who was not involved in the study. He adds that these patterns give a better diagnostic predictive value than simply focusing on any one protein or gene. "What's really intriguing here is that they're doing this with metabolites. This suggests that there is metabolic information that ovarian cancer, and perhaps other cancers, give off into the blood."
Petricoin says the findings are not yet ready for the doctor's office, however. "Most of the samples they used came from women with later stages of cancer. Clinically, a technique that distinguishes late-stage cancer from controls is not that useful." He adds that a more immediate application may be in tests for women with an elevated risk of developing ovarian cancer or for cancer recurrence in women who have already undergone treatment.
McDonald echoes this concern: "One limitation is that so far we've only [run] 94 patient samples. We'll have more confidence when we run thousands."
And for any detection method to be clinically useful, Petricoin emphasizes, it needs to have 100 percent sensitivity and specificity. Because ovarian cancer is relatively rare, occurring in approximately one out of every 2,500 women, a test with only 99 percent specificity would result in false-positive diagnoses for 25 women, leading to unnecessary and risky surgeries and procedures.
McDonald's team also wants to determine whether this technique distinguishes between ovarian and other types of malignancies. "We're running pancreatic and lung cancer samples now," McDonald says. He envisions storing these cancer-specific profiles in a database so that one day a single drop of a patient's blood could be quickly and easily scanned for multiple diseases.
