You Smell

There is a notorious research freezer at the Monell Chemical Senses Center, on Market Street in Philadelphia. It belongs to George Preti, an organic chemist who has been collecting and studying the olfactory essence of humanity for years. His freezer is brimming with glass jars, ziplock plastic bags, and test tubes charged with swabs of sweat and skin secretions, fabric-captured breath, and urine extracts. The freezer is a cryogenic curio of human emanation, and if you open it, it stinks to high heaven. Running away or slamming the door shut are about the only options you consider.

The freezer represents what it takes to answer this question: Does each human being have an odorprint as personalized and identifying as a fingerprint? Or this question: Does body odor bear discernible biomarkers that reflect your state of health, such as whether you have cancer or diabetes? Or this one: Is it possible to smell anxiety or deception, maybe even love and hate, in a person’s personal vapors?

For those immersed in the world of trailing bloodhounds, the answers to those queries are obvious because each of the 6.7 billion people on Earth has a signature body odor. Milica Kokot-Nichols, cofounder of the Alaska-based Law Enforcement Bloodhound Association, recalls an emblematic, albeit nonscientific, demonstration of odorprint detection at one of her association’s training sessions. In trials in which the dogs were given sniffs of clothing from each one of a set of triplets (the children of an officer attending the seminar), veteran trailing dogs were able to distinguish between the otherwise carbon-copy babies.

“The dogs had no trouble,” Nichols says. “If you ask me how they do it, I could not tell you—not until my dog learns to speak and tells me how.”

Human odorprint researchers aren’t waiting on talking bloodhounds for help. But the dogs’ olfactory feats provide a major justification for the scientists’ work. “We know for sure that the odorprint exists, based on the dog evidence and our research,” says Gary Beauchamp, director of Monell, where he, Preti, and colleagues constitute one of the more established and accomplished scientific teams trying to understand, codify, monitor, and exploit the human odorprint.

Another scientist on the hunt is Kenneth G. Furton of Florida International University. One of his goals is to establish a scientific basis for dogs’ abilities to tell one person from another by smell. Investigators, mainly in Europe but also in the U.S., have long used the canine-based forensic technique known as a scent lineup. Investigators swab crime-scene evidence to capture a criminal’s scent and then line up the collected scent with “decoy scents.” A dog previously presented with a pad that had been swiped on a suspect is then set onto the scents to see whether the canine zeros in on the one associated with the crime.

The forensic technique has been used for years, but, Furton says, “courts now are asking for the scientific validation.”

In an effort to provide that validation, he and his colleagues have been searching for machine-detectable patterns in the volatile chemicals emitted by people. In a study funded by the Netherlands’ National Police Agency, Furton swabbed the hands of 60 individuals with specially cleaned pads and placed these inside glass vials. Then, using a solid-phase microextraction technique, the volatile compounds in the vials’ headspaces were taken up by treated polymeric fibers, and those compounds were eluted into a gas chromatography-mass spectrometry (GC/MS) system. The analyses turned up 63 compounds: a mix of acids, alcohols, aldehydes, hydrocarbons, esters, ketones, and nitrogenous molecules. By applying a pattern-recognition technique, the researchers reported that they could distinguish individuals on the basis of the GC/MS data. They are the sort of data, Furton says, that could help validate the use of dogs in scent lineups.

Validating scent lineups is just one of the incentives driving odorprint researchers. The U.S. federal government has also been adding some momentum by funding efforts like those at Monell and supporting in-house researchers such as analytical chemist Brian A. Eckenrode of the Federal Bureau of Investigation’s Counterterrorism & Forensic Science Research Unit at the FBI Academy, in Quantico, Va.

Eckenrode, whose papers have titles like “Odor Analysis of Decomposing Buried Human Remains,” couldn’t reveal much when C&EN asked him how the FBI might put reliable odorprint detection and tracking technology to work. But he did talk generally about “human ID and surveillance.”

The idea here would be to place odorprint-sniffing devices at airports and train stations and eventually—depending on the paranoia level—all over. These would relentlessly collect air samples bearing people’s personal bouquets of volatile organic chemicals (VOCs). Then, with GC/MS or VOC-detecting chips now under development, the system would produce an odorprint for each person sampled. The overall ID system would have to include a stored database of odorprints, including ones associated with worrisome people, much like the FBI’s massive fingerprint database, in Clarksburg, W.Va.

This sort of government surveillance is not far-fetched. An exhibit at the Stasi Museum, in Berlin, chronicles the practice of the former East German secret police organization of surreptitiously collecting odor samples—from specially designed seat cushions, for example—with the notion of using these stored odors to identify and track suspect citizens by way of sniffer dogs.

The U.S. is taking steps in this direction. In August, for instance, the Department of Justice released a solicitation titled “Human Emanation Study—Air Shower,” for “a ‘phone booth’-like pod that should be capable of ‘showering’ a human or item of interest with air.” The synopsis states that “the FBI’s desire is to sample the air emanating from a living human being.” These would go beyond the puffers already deployed to pick up signs of explosives, a simpler analytic task.

Back in March, the Department of Homeland Security’s Science & Technology Directorate floated a research solicitation that presages the kinds of information the security community hopes to glean from odorprint detection and surveillance. “The DHS S&T Directorate is requesting approval for a contracted, outsourced, proof-of-principle study to determine if human odor signatures can serve as an indicator of deception,” the solicitation states.

And the Defense Advanced Research Projects Agency, with a case of dog olfaction envy, has doled out tens of millions of dollars for its RealNose program. “The key to the program concept,” DARPA says, is that by “simulating the entire canine olfactory system (from air intake to pattern recognition), revolutionary detection capabilities will be created, demonstrating canine-comparable sensitivity, distance, and detection thresholds.”

Monell’s Alan Gelperin, a former Bell Laboratories researcher who once worked on a never-made supermarket checkout system that would have detected fruits and vegetables by way of chemical signatures, is part of the RealNose project.

His focus is to build sensor arrays that can, for example, detect odors of land mines in an environment with all kinds of relevant and irrelevant chemical signals. The chips should work for odorprint sensing, too, Gelperin surmises. “The odorprint would show up as a pattern of information from a large group of sensors whose signals would vary in details from reading to reading but in which there would be a constancy that a clever pattern-recognition algorithm would pick out,” Gelperin says. “You would associate that constancy with a particular person.”

The technical basis for this work, done in collaboration with A. T. (Charlie) Johnson Jr. of the University of Pennsylvania, is a sensor that combines the transistor function of single-wall carbon nanotubes with the selective uptake of gaseous molecules by different single-stranded DNA molecules adsorbed onto the nanotubes. Volatile molecules that bind to DNA snippets with differing affinities alter the electronic signals of the underlying nanotubes, providing the basis of the chip’s ability to smell.

With enough single-stranded DNA and sophisticated-enough pattern-recognition algorithms, Gelperin suspects human odorprint detection will become possible with these devices. There are already air-sampling units at airports, subway stations, and elsewhere, he points out. “If you had better sensor technologies, you could take samples, upload them to a government site,” and check patterns against those in an odorprint database, Gelperin suggests.

What Gelperin’s sensor arrays cannot do is reveal much about the chemistry or behavioral relevance of individuals’ odors and the chemical senses. He’s been leaving those knowledge gaps to others, including his Monell colleagues.

“Different body parts produce different profiles of odorants,” Preti says, referring, for example, to underarms, breath, skin, and genitals. “That is because of different types and amounts of bacteria that live there, the amount of oxygen available, and the different types and numbers of skin glands and secretions. All of these go into creating different groups of volatiles that influence the odorprints that are emanating from you.”

Because the composition of volatile chemicals in breath and urine is more sensitive to diet than that in underarm sweat and secretions, Preti has long leaned 
toward the latter as the pragmatic source for odorprint detection. In contrast to Gelperin’s electronic-signal-based view, Preti sees an odorprint “as a group of molecules present in certain ratios that might be quantitatively different for each individual on the planet.” In his mind, Preti says, “it would look like a GC/MS trace.”

Typical studies of human odor involve collecting VOCs and other chemicals in human emanations with absorbent material and then applying a technique such as solid-phase microextraction to prepare the VOCs for analysis, usually by GC/MS. The end results often are tables listing dozens of odor chemicals and their relative 
concentrations.

One of the more extensive and revealing recent studies was undertaken by an international research team led by Dustin Penn of the Austrian Academy of Sciences, in Vienna (J. R. Soc. Interface 2007, 4, 331). The team identified 373 peaks that it could detect repeatedly in the underarm sweat of individuals during a 10-week sampling period. The researchers collected samples from 197 adults in a small town in the Austrian Alps, and they found 4,941 distinct GC/MS peaks across all samples. Predominant in the complex mix of VOCs were short- and long-chain hydrocarbons, alcohols, carboxylic acids, ketones, and aldehydes.

Applying data analysis and pattern-recognition techniques to the 373 most repeatedly detectable compounds, the researchers reported finding “individually distinct and reproducible GC/MS fingerprints, reproducible differences between the sexes, and ... the chemical structures of 44 individual and 12 gender-specific volatile compounds.” Because none of the compounds was found exclusively in certain individuals or in just men or women, it was the “multivariate fingerprint”—the components’ relative proportions—that was most telling. “Odor may be analogous to facial features, in that no single measurement on a face can easily be used to recognize an individual,” the researchers suggested.

Adding to the challenge for anyone in search of human odorprints is the cacophony of volatile chemicals that modern living brings into the body-odor context, Preti says. A person’s primary portfolio of odor chemicals includes volatiles that depend on diet and drug regimens as well as discretionary odiferous additions from products such as shampoo and perfume. “My emphasis always has been to try to discriminate the endogenous metabolites from those that are being added to an individual’s environment,” Preti says.

For more than 30 years, he and others have been looking for a molecular biological basis for the primary, and presumably most stable, olfactory notes of odorprints. One of the earliest and most long-lived suggestions was made in 1974 by the late cancer researcher and science writer Lewis Thomas. He speculated that the genes and cell-surface proteins that comprise the immune system’s major histocompatability complex (MHC) could do double duty by somehow generating individually specific odor profiles. The MHC differs subtly across individuals and is the basis for such biomedical phenomena as tissue rejection.

Subsequent research at Monell and elsewhere, much of it based on behavioral studies involving mice that differ only in MHC genotype, has substantiated Lewis’ speculation. Yet even after decades of searching, Monell’s Beauchamp laments, “we still have no idea what the MHC odorants are.” Some researchers have suggested that the MHC peptides themselves, even though they are nonvolatile, are what the mice somehow detect as distinguishing factors. Or it could be that MHC proteins play an enzymatic role that could help generate individualized profiles of odor molecules. Another theory is that different MHCs lead to different body microflora, which in turn influence the mix of each animal’s body-odor chemicals.

Despite these large knowledge gaps, the quest to identify and characterize odorprints has been yielding interim discoveries with potential near-term payoffs.

Perhaps foremost among these is the identification of disease biomarkers in breath and skin emanations for diabetes, skin and lung cancer, and other maladies (C&EN, Sept. 22, 2008, page 81). “This is the sort of thing that could help doctors choose antibiotics and other treatments more effectively,” says Gelperin, who has worked on sniffing out the biomarkers of sinus infections. The vision here is that patients might breathe into a collector or that their skin emanations might be harvested with vacuum-based devices so that the volatile components could be analyzed to help with diagnoses.

Then there’s that “smell of deception” that DHS would like to be able to detect. In theory, this should be possible. “Anything that causes physiological changes is likely to change body odor,” Beauchamp notes. And stress, including stress associated with lying, has physiological effects.

Experimental psychologist Pamela Dalton of Monell, along with Preti and others, has found that the composition of underarm sweat is altered by stressful experiences, such as being tasked to count backward by intervals of 13 quickly and without error lest you need to start over. “It is a humbling experience, and our experiments show that particular kinds of stressors will make levels of stress hormones, such as epinephrine and cortisol, increase rapidly,” Dalton says.

This stress response does not require a GC/MS to detect. A human nose will do. “We found that people can readily distinguish the smell of someone who is stressed versus the same person when not stressed,” Dalton tells C&EN. “Slightly oniony, where you have left an onion out awhile, and you get a lot of those sulfur volatiles emanating from it” is how she describes the smell of stress.

Dalton can’t say precisely what chemicals are responsible or whether these are normally at indiscernible levels in the odorprint baseline or are unique signals produced under stress. Even if she can uncover the chemical details, it will be tricky to put such findings to use for, say, security applications. “There are people who go to the airport every day who are terrified of flying,” yet they might smell stressed out, like a person planning to blow up the plane, Dalton cautions.

Another tantalizing trait of an odorprint is that it lingers in an environment. “You might be able to detect that there was, in the recent past, a moment of stress in a particular location,” Dalton says. “You might be able to exploit that to determine if a bank robbery had taken place.”

You hear a lot of “you might be able to do this and that” from odorprint researchers, says Preti, proud guardian of the human-odor freezer, but “bottom line—we ain’t there yet.”

Chemical & Engineering News

ISSN 0009-2347

Copyright © 2011 American Chemical Society