Sniffing out the signs of cancer
2022; Wiley; Volume: 130; Issue: 4 Linguagem: Inglês
10.1002/cncy.22570
ISSN1934-6638
Autores Tópico(s)Mass Spectrometry Techniques and Applications
ResumoThis immense collection, or volatilome, may act as a chemical fingerprint while providing signposts of health and disease based on telltale shifts in the VOC output in response to changes in cell metabolism and other fluctuations in the body. Although the precise cues remain unknown, dogs appear to have an unmatched ability to detect these VOC signatures as indicators of specific diseases. Because of the difficulty and variability in detection dog training, however, researchers have been asking whether the canines might help “train” or inspire the development of electronic sensors that could provide a more reliable means for sniffing out signs of trouble. Beyond the demonstrated ability of dogs to identify diseases like diabetes and COVID-19 by scent, the metabolic changes accompanying tumorigenesis have opened up windows of inquiry into the VOC-driven detection of a long list of cancers. “I absolutely have no doubt that there is a unique odor,” says Cynthia M. Otto, PhD, director of the Working Dog Center at the University of Pennsylvania in Philadelphia. Dr. Otto and her colleagues have trained detection dogs to identify the presence of ovarian cancer through a distinctive scent in the blood plasma of patients. For a study published in 2020, Dr. Otto and her collaborators took a combined approach that relied on 3 separate sensors: trained dogs, an organic and analytic chemistry method known as solid-phase microextraction gas chromatography–mass spectrometry, and a third technique using single-stranded, DNA-coated carbon nanotubes.1 All 3 of the VOC-detection techniques converged upon the same conclusion: “Plasma from ovarian cancer patients emits a volatile odor signature that can be distinguished from the VOCs of patients with benign ovarian tumors and controls.” That finding offers the tantalizing, but early, possibility that a noninvasive chemical-detection analysis might eventually provide a diagnostic alternative for ovarian cancer. Dr. Otto's program recruits canines with a demonstrated aptitude for scent detection. As part of her expansion into medical detection, she is also training dogs to detect bacterial biofilm infections after prosthetic implants; COVID-19 from urine, sweat, and saliva; and infectious prions in deer feces. Amid the growing evidence of their abilities, however, dogs also have limitations. For the ovarian cancer and COVID-19 detective work, for example, Dr. Otto and her colleagues found that identifying a discreet sample in a sterile environment was a much less demanding task for the dogs than real-world detection in more complex backgrounds. One Canadian researcher, commenting on her study's disappointing results on dog detection of Clostridioides difficile infections in a hospital, noted that a canine was distracted by food trays in patients' rooms. Researchers have learned the hard way that not every dog is well suited for a potential career in disease detection. “It just is not turnkey,” Dr. Otto says. In another small German study, 4 dogs started a training regimen for identifying the breath samples of lung cancer patients.2 Two dogs, however, were withdrawn after making “insufficient training progress.” The remaining 2 dogs correctly identified 9 of 9 and 8 of 9 cancer patients, respectively, but scored less well on specificity, identifying 8 of 10 and 4 of 10 healthy controls, respectively. “Dogs sometimes have a bad day,” Dr. Otto says. “Dogs are great at [disease detection], but if you watch these dogs, it's exhausting. It's kind of like going through spreadsheets for taxes: Some of us can do that for only so long before our head just explodes.” Some dogs, conversely, have another problem: They are so good at detecting individual odors that improper training can sometimes confound the results. For the ovarian cancer detection work, the researchers began running out of training samples and had to begin reusing them. “When we reused samples repeatedly, the dogs stopped generalizing—and stopped recognizing ovarian cancer—and started to look just for ‘Mary’ and ‘Sue’ and ‘Jane,’ because they'd seen those individual samples,” Dr. Otto says. In the German study, the authors likewise conceded that a limited number of lung cancer samples meant they had to reuse some for multiple training sessions, increasing the odds that the dogs would learn and alert to the odor of a specific patient instead of the signature scent of the cancer. Although dogs learn quickly, “they sometimes learn shortcuts that we don't realize that we're introducing,” Dr. Otto says. “I think we need to be cautiously optimistic with the dogs. I have no question that they're capable of doing it. The limitation is on us and our ability to train them and understand well enough how they're doing it so that we don't mess them up.” One key lesson is that the dogs need to encounter a broad array of training samples with few repeats. That insight, Dr. Otto says, has strongly influenced her recommendations on developing national standards for the technique. As for what the dogs might be detecting, Dr. Otto points out that the signals may be below the detection limit of existing machines. “It also is very likely that the dogs aren't detecting a single thing,” she says. Instead, they may be detecting ratios between separate compounds. Some of her experiments, in fact, suggest that different dogs may cue on different components of a chemical signature, just as dogs trained to detect explosives have been known to do. “It's so much more complex than our brains can even begin to wrap around,” she says. Researchers are trying to understand the process, however. Mangilal Agarwal, PhD, director of the Integrated Nanosystems Development Institute at Indiana University–Purdue University Indianapolis, says that his work on electronic VOC sensing was inspired by an article on a local dog detection training center. Intrigued, he met with the trainer and a collaborating endocrinologist who was interested in diabetes detection. As an engineer, Dr. Agarwal's interest was not how the dogs were sensing the odor cues, but what they were smelling and whether that signal could be detected through other means. “Without the news article, maybe I would have not been doing this research,” he says. “Sometimes it can change the way we are looking at different aspects [of research].” In a 2019 proof-of-principle study in mice, Dr. Agarwal's team used an advanced form of gas chromatography–mass spectrometry to identify a panel of VOCs in the rodents' urine that differentiated between local and metastatic breast cancer, suggesting that similar markers could be used to track disease progression.3 If the studies bear out, Dr. Agarwal says, the technology could be tested in human patients. For another project, the lab collected urine samples from human patients undergoing biopsies for prostate cancer and showed that the same sensing technique identified a panel of VOCs distinguishing between healthy patients and those subsequently diagnosed with prostate cancer. (The researchers found similar VOCs within a signature of prostate cancer in a mouse model of the disease.) “We can define those biomarkers; now, the question is, can we do it on a larger group of subjects to validate it?” –Mangilal Agarwal, PhD Dr. Otto cites a similar goal with her research. “Especially with cancer, our thought has always been, if this is something that has a clear signature, then let's get that technology so that we can have that high-throughput screening that is reproducible,” she says. The incredible sensing ability of a dog, in other words, could aid in the development of lab-based diagnostic techniques. Dr. Otto is encouraged by the results so far but concedes that it is too early to predict success. “I know what the dogs can do,” she says. “I don't know what the other side can do.”
Referência(s)