Issue Date: September 22, 2008
Sniffing Out Cancer
LONG BEFORE BLOOD TESTS and MRI scans were invented, physicians relied on their sense of smell to help diagnose patients' ailments. Researchers are now updating this practice so doctors can use sensors and other devices to analyze the volatile compounds their patients emit. Such techniques could conceivably be used as the basis for rapid, noninvasive procedures for early detection of diseases such as cancer.
The use of odors as diagnostic clues has ancient roots, according to George Preti, who studies the nature and origin of human odors at the Monell Chemical Senses Center, a nonprofit research institute based in Philadelphia. "Hippocrates—commonly called the father of modern medicine—is reported to have smelled the breath of patients as part of his assessment," Preti said. "And you can go to the ancient Roman literature and find references to the use of olfaction and taste to evaluate a patient's urine and other excretory matter." Preti made his remarks at last month's American Chemical Society national meeting in Philadelphia during a Division of Agricultural & Food Chemistry symposium on chemical senses and health.
"Prior to the 20th century and the widespread use of scented soaps and consumer products, it was much more readily apparent that humans emit a variety of volatile odorous metabolites," Preti noted. "And prior to the development of instrumental analysis and clinical medical testing, a physician had to rely more on his or her observational skills than quantitative tests. Consequently, many diseases became associated with their chief presenting complaint, which is a characteristic odor."
Patients with scurvy, for instance, have a sweet, putrid odor, whereas those with typhoid give off a smell reminiscent of fresh brown bread. Preti, who was speaking in a large hall, said the emanations from a single patient suffering from a severe form of trimethylaminuria could "be strong enough to fill this auditorium with the smell of rotting fish." The odor associated with this genetic disorder derives from trimethylamine, a compound produced by intestinal bacteria to aid digestion. Unlike healthy people, patients with the disease are unable to metabolize trimethylamine into trimethylamine oxide, which is odorless.
Although severe forms of trimethylaminuria are easy to detect with a sniff, some forms of the disorder produce more subtle olfactory clues. Other diseases alter the body's volatile emissions in ways that can't be detected by the human nose but can be sensed by other animals. For example, dogs can sniff out skin tumors because the tumors smell different than healthy skin, according to Michelle Gallagher, a former Monell postdoc who is now an analytical chemist at Rohm and Haas in Spring House, Pa.
At the national meeting, Gallagher described work she conducted with Preti and others to determine how the profile of volatile compounds released from the skin changes in people with skin cancer. The researchers also wanted to know whether this alteration could be used to detect early-stage skin cancer, which strikes one in five Americans.
First, though, they had to characterize normal skin emissions while avoiding cross-contamination from modern hygiene practices. The Monell researchers asked their subjects to refrain from using any scented products such as perfumes, to wash with fragrance-free soap and shampoo, and to wear new unwashed T-shirts for seven to 10 days before coming to the lab for testing.
USING TWO DIFFERENT techniques, Gallagher and her colleagues then collected samples of emanations from the volunteers' skin. In one method, the scientists used solid-phase microextraction (SPME) to collect volatiles inside a funnel placed over each subject's skin. In the other, the researchers applied a mixture of ethanol and hexane to the skin of the volunteers to collect less-volatile compounds. The researchers analyzed the identity and concentration of collected compounds by gas chromatography (GC) and mass spectrometry (MS).
When Gallagher and her coworkers compared the profile of volatile organic compounds emitted from the skin of healthy patients and skin cancer patients, they found notable differences in just two of the compounds. Emissions of dimethyl sulfone were higher and those of 6-methyl-5-hepten-2-one were lower in cancer patients. Dimethyl sulfone, a metabolite of sulfur-containing amino acids, is odorless to most people. The other compound is reminiscent of "nail polish remover, with some pungency," Preti told C&EN, and its biological source is unknown.
"Our findings may someday allow doctors to screen for and diagnose skin cancers at very early stages," Gallagher said. Physicians could move a handheld scanner over a patient's skin to check for the telltale chemical signature of cancer even before any visible signs, such as moles, appear.
The researchers collected urine from healthy "placebo" mice and from mice that serve as models for two different types of lung cancer. The Monell team then trained other mice to distinguish between urine from the healthy mice and urine from the mice with tumors. The fact that this experiment succeeded "proves that there is a volatile signal that can be detected by the nose that signals the presence of cancer," Beauchamp told C&EN.
The researchers used SPME followed by GC/MS to analyze the differences in volatile biomarkers between the two types of urine samples. They discovered that tumor growth altered the amounts of several of these volatile metabolites. However, individual compounds weren't particularly effective as diagnostic markers for the presence of cancer, Beauchamp explained. "Instead, we found that certain combinations of compounds generated high-accuracy scores for discriminating tumor from placebo groups. Thus, it appears that cancer is signaled by alterations in the pattern of volatile metabolites."
In many, but not all, cases the amounts of metabolites decreased in mice with tumors compared with those in healthy mice, Beauchamp noted. His team is now investigating the metabolic pathways for the affected compounds.
The results indicate that "it is possible to identify mice with lung cancer tumors based on volatile biomarkers," Matsumura said. The Monell group also found that urinary biomarker profiles differ for mice with early-stage versus late-stage cancer, but the researchers declined to divulge more details.
"The successful results of this mouse work encourage us to conduct similar studies on human patients," Beauchamp told C&EN. "Of course, the ultimate goal is to develop reliable, noninvasive, inexpensive tools to help in the diagnosis of human lung and other cancers."
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society