The current COVID-19 pandemic requires accurate diagnosis, both to isolate the infected and to treat the symptomatic patients. However, the supply-demand gap is widening all the time. A new Colombian study published on the preprint server bioRxiv* in June 2020 describes the use of trained dogs to identify severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infected humans by scent alone. This would allow uninfected people to work together safely, rescuing the economy without driving up medical costs.
The most important requirement of COVID-19 testing is that it must accurately detect the presence of infection at any stage, whether the individual is asymptomatic, presymptomatic, or symptomatic. Any diagnostic test newly developed must be at least as effective as the gold standard, which is the reverse transcriptase-polymerase chain reaction (RT-PCR). This is unfortunately not as widely accessible to large segments of the world’s population.
Antibody testing is not useful in limiting infection, because they reach a peak positive only after the infectious period is over. Without a universal, safe, and readily available means of diagnosis, non-pharmacological interventions, namely, lockdowns, social distancing, and quarantines, have become the methods of choice. However, these do not prevent infection, but limit the viral spread, flatten the curve, and provide time for public health responses to catch up with the acceleration of the pandemic peak.
At the same time, the exit of billions of workers from the economy is a move fraught with economic danger, or even catastrophe, especially if this withdrawal from overt economic activity is prolonged, and occurs in a developing country. Thus, the researchers say, “Balancing quarantine measures with the safe operation of the economy is a necessity.”
In the absence of supportive measures, lockdowns in a developing or undeveloped economy can well lead to increased poverty, uprisings, malnutrition, and social disobedience. The financial means to provide such support, meanwhile, is often simply not there.
Therefore, says the study, “It is imperative to implement new, fast, and reliable diagnostic tests that offer novel opportunities for infection control and for safely re-opening the economy.”
Training Dogs to Detect Pathogens and Dangers
For the current study, the researchers turned to man’s best friend, the domestic dog, Canis lupus familiaris. After all, they are known to sniff out minute traces of explosives, blood, fragrance, pollutants, currency, and even insects. It is noteworthy that the most sensitive detection tool has not been able to exceed the sensitivity of scent detection of dogs, mainly because of the demand for “perfect performance and complete reliability.”
Thus, reasoned the researchers, why not train them to detect SARS-CoV-2 infection by scent? Some earlier studies have shown that training with a strictly scientific method can allow dogs to perform extremely consistently with respect to obtaining excellent results with diseases like multidrug-resistant Clostridiodes difficile. Some not so well-validated studies have also reported that dogs can be trained to pick up specific cancers, infections, migraine aura, impending seizures, or drastic drops in blood sugar.
The agricultural pest Candidatum Liberibacter asiaticus was also detected by dogs trained to sniff it out, according to the US Department of Agriculture (USDA), who reported that these dogs found infected trees much better than the quantitative PCR test did.
How Do Dogs Detect SARS-CoV-2 Scent?
The ability of dogs to detect different scents is due to their picking up and distinguishing various odors by the volatile organic compounds (VOCs) contained in the substances that emit those odors. These VOCs mix with breath, sweat, saliva, feces, urine, mucus, or skin, or any combination of these, to create a “smell print” of the organism.
Pictures and identification of the six dogs trained for the scent-detection of SARS-CoV-2. (1) Andromeda, intact female, 6-mo, Belgian Malinois. (2) Nina, intact female, 25-mo, Belgian Malinois, (3) Niño, castrated male, unknown age, American Pit Bull Terrier. (4) Timo, intact male, 31-mo, Belgian Malinois. (5) Vika, intact female, 36- mo, Belgian Malinois. (6) Vita, intact female, 36-mo, first generation Alaskan Malamute x Siberian Husky.
Extending this to SARS-CoV-2, a collaborative project discovered over 330 interactions between human and viral proteins during the COVID-19 infection. A healthy human breathes out air containing VOCs, of which almost half is composed of nitrogen, which indicates protein. This means that detecting infection with this virus could be simpler than thought since these breath proteins will show specific changes that produce specific odors.
These odors will make COVID-19 patients quickly distinguishable by their breath and respiratory secretions to a dog’s nose. The current study describes how the researchers produced a dog training protocol to make it possible for dogs to detect COVID-19 patients in a safe, accurate and inexpensive way, without using chemical reagents, or putting anyone at risk.
The Smell Source
The first step was to create a safe saliva container where sealed flasks of respiratory secretions from these patients were placed, to protect both dogs and people from viral transmission. The VOCs evaporate over time, producing an odor perceptible to the dog. Six dogs were trained, and the results validated by the USDA protocol.
The training part of the project proceeded in three phases, the in vitro recognition phase, the in vitro diagnostic phase, and the in vivo diagnosis phase. The first two, which took about 28 + 21 days to accomplish, respectively, are being reported here. These two phases involved investigated well over 3,000 and 6,000 samples by scent, respectively. The third phase is still going on.
Phase I Results
The low positive predictive value (PPV), at 74%, meant that about 26% of diagnosed positives would be actually negative. This is characteristic of a low-prevalence condition when the PPV is low, but the negative predictive value (NPV) is high. To resolve this, the researchers increased the prevalence and found that the dogs identified all the specimens with complete accuracy.
To bring the results to a more realistic context, they then carried out phase II training with a prevalence between 1% and 5% (typical prevalence is at present below 1%), and discouraged their detection of false positives by not giving frequent rewards unless they stopped detecting false positives. The dogs achieved better performance, with the PPV 86% (up by 12 points), and the NPV almost 100%, indicating that any person not detected by the dogs would probably not have the virus at all.
The researchers point out that there are well over 135 million dogs on earth, and if one million of them were appropriately trained, the pandemic would be over. “Self-isolation, simultaneously with rapid detection of contacts by the nearest dog, could be the best alternative to solve this problem because the needed technology is not available to most countries.” Many issues remain, not least the possibility that dogs may trick the trainer into getting more rewards. A centralized accreditation might help to standardize the results.
bioRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.