Can Dogs Detect Lung Cancer?

Lung cancer is the leading cause of cancer deaths in the United States and is the most common cancer among men and women. Since the disease often does not present symptoms until it has already reached an advanced stage, screening and early detection are both critical when it comes to lung cancer management. Not all screening techniques are created equally, however: low-dose CT scans have been proven to be more effective compared to chest X-rays in disease detection.

Meanwhile, researchers have explored another alternative detection strategy: dogs.

Lung Cancer Survival Rates: Stage Matters

When discussing lung cancer, it is important to understand how different disease stages impact predicted survival. The American Cancer Society provides data on five-year relative survival rates. The relative survival rate is a comparison of patients with the same cancer type and stage versus the overall population: “For example, if the 5-year relative survival rate for a specific stage of lung cancer is 60%, it means that people who have that cancer are, on average, about 60% as likely as people who don’t have that cancer to live for at least 5 years after being diagnosed.”

The data come from the Surveillance, Epidemiology, and End Results (SEER) database, which is maintained by the National Cancer Institute. These data do not track disease stage numerically, though, but rather as localized (no sign that the cancer has spread outside of the lung), regional (cancer has spread outside the lung to nearby structures or lymph node), or distant (cancer has spread to distant parts of the body, such as the brain, bones, liver, or the other lung).

Among patients with non-small cell lung cancer, the five-year relative survival rates are: localized, 61%; regional, 35%; distant, 6%; and overall, 24%. For small cell lung cancer, the five-year relative survival rates are: localized, 27%; regional, 16%; distant, 3%; and overall, 6%. Both datasets apply to patients diagnosed between 2009 and 2015.

Who Should Be Screened for Lung Cancer?

Since symptoms often do not manifest until the disease has progressed, screening is critical—particularly for high-risk patients. The U.S. Preventive Services Task Force recommends that adults aged between 55 and 80 years with a 30 pack-year smoking history (smoking about a pack a day for 30 years) who are either current smokers or have quit within the last 15 years should undergo low-dose computed tomography (LDCT) screening. After that person has stopped smoking for 15 years or if they develop a health problem that significantly reduces their life expectancy or willingness to undergo lung surgery.

Because lung cancer screening poses possible risks, not everyone should be screened. The Centers for Disease Control and Prevention identifies three risks that may come out of screening:

  • False-positive result: patients may test positive for lung cancer even when no cancer is present, resulting in unnecessary additional tests and surgeries that pose risks of their own
  • Overdiagnosis: a positive test result may sometimes unearth a diagnosis that if undiscovered would not have caused any problems for the patient, resulting in overtreatment.
  • Radiation exposure: repeated LDCT tests expose patients to radiation, which could cause cancer.

So What About The Dogs?

A 2019 study published in The Journal of the American Osteopathic Association assessed a very unique way to detect lung cancer.

The researchers trained four two-year-old beagles who had never been trained to detect lung cancer samples before. These dogs were chosen due to their strong sense of smell, relatively small size, and calm nature, and because they are highly trainable and sociable. One dog was ultimately removed “because she was poorly motivated by food and did not respond well to any training method used,” said the authors.

The dogs were trained over an eight-week period, using operant conditioning with clicker training, to differentiate blood serum from malignant lung cancer patients from that of healthy patients. Training was broken down into three phases. In phase one, two weeks were spent training the dogs to identify a treat in a canister. Phase two also spanned two weeks, during which time treats were paired with a positive serum sample containing lung cancer; the amount of treat paired with the sample was gradually decreased until the dogs eventually were able to identify the positive serum sample alone, without the treat present. The final phase spanned four weeks and required the dogs to identify the serum sample only with no treats present. Only when all three training phases were successfully completed were the dogs moved to the testing phase.

Ten serum samples were provided by 10 donors; six donors were female and four were male. Donor ages ranged from 26 years to 80 years. The mean age of the female donors was 64 years, and the donors were black (n=3), white (n=2), and Hispanic (n=1). The mean age of the male donors was 49.5 years, and the donors were black (n=2), white (n=1), and mixed race (n=1). Forty control samples were also used.

The results of the testing phase were recorded in terms of true positives, false positives, true negatives, and false negatives. The first dog’s outcomes were: true positives, 10; false positives, 1; true negatives, 39; and false negatives, 0. The second dog’s outcomes were: true positives, 10; false positives, 0; true negatives, 40; and false negatives, 0. The third dog’s outcomes were: true positives, 9; false positives, 2; true negatives, 39; and false negatives, 1. The total sensitivity and specificity was 96.7% and 97.5%, respectively. The proportion of lung cancer patients who tested positive for the disease was 90.6%, and the proportion of healthy patients who tested negative for the disease was 99.2%.

The study authors concluded, “Further investigation into the biochemical molecules detected by dogs could provide a foundation for the development of a highly sensitive, specific, and cost-effective method for early cancer detection. Early detection of cancer is one of the best ways to improve patient outcomes, and current methods of early detection rely on expensive imaging equipment, which can be an insurmountable obstacle for underserved and rural communities.”

This isn’t the first study to observe that dogs may be helpful in the oncology sphere—or even in lung cancer specifically. According to a 2006 study, dogs may be able to detect breast and lung cancers in the breath.

The study was born from the belief that patients with lung and breast cancers may exhibit biochemical markers in their exhaled breath. Using this knowledge, the authors trained five “ordinary household dogs” using a foods-reward method to distinguish exhaled breath samples of 55 lung cancer and 31 breast cancer patients from those of 83 healthy patients; the dogs were trained to sit/lie in front of the cancerous samples and ignore the non-cancerous ones. Similar to the newer lung cancer study, these dogs were also trained using a three-phase strategy. Each phase gradually increased the difficulty of cancer detection. When the dogs were fully trained, they were then tested using samples from subjects they had not previously encountered samples from. Both the dog handlers and experimental observers were blinded from which samples were cancerous and which were healthy.

In the lung cancer and control tests, the dogs’ overall sensitivity—compared to conventional diagnosis confirmed by biopsy—was 0.99 (95% confidence interval [CI], 0.99 to 1.00), and specificity was 0.99 (95% CI, 0.96 to 1.00). In the breast cancer and control tests, sensitivity was 0.88 (95% CI, 0.75 to 1.00) and specificity was 0.98 (95% CI, 0.90 to 0.99). Sensitivity and specificity outcomes were similar regardless of disease stage in both cancers.

In 2004, researchers reporting in The BMJ tested dogs’ ability to sniff out bladder cancer. For seven months, they trained six dogs of different ages and breeds who were previously trained in obedience commands but had no history of search or scent discrimination training. A total of 36 male and female patients with new or recurrent transitional cell carcinoma of the bladder (age range, 48 to 90 years) provided urine samples; 27 were used for training, and nine were used for formal testing. An additional 108 male and female control patients (age range, 18 to 85 years) who were a mix of both healthy and diseased were also included; 54 samples each were used for training and testing. Using samples they had never encountered before, the dogs were trained to detect the one urine sample with bladder cancer among six control samples using operant conditioning using the clicker method. Dogs were trained to lie next to the sample with bladder cancer. Training became progressively more complex: samples positive for bladder cancer were first placed against water, then healthy diluted urine, undiluted healthy urine, urine containing blood from menstruating women, and urine from patients with non-malignant active or recent urological disease or other disease. The primary outcome was the mean proportion of successes per dog compared to the value expected by chance alone (1 in 7, 14%).

When taken as a group, the dogs were able to sniff out the bladder cancer samples 22 times out of 54, for a mean success rate of 41%. “The association between presence of cancer and selection by the dogs was slightly stronger in the multivariate model, which also included presence of blood and ketones, than in the univariate model. This indicated that the association was not due to confounding with factors measured on urinalysis,” added the researchers.

Jury Still Out on Dogs’ Abilities in Some Cancers

Unfortunately, dogs may not be able to detect all cancers equally—at least not yet.

One study, published in 2017, sought to determine whether dogs could detect urinary tract transitional cell carcinoma (TCC) by smelling urine samples.

For phase one of the training, four dogs with prior scent training were trained to differentiate between a control sample and a TCC-positive urine sample. After a mean 15 sessions (20 trials per session), all four dogs were successfully trained. In phase two, the original control sample and four new ones were used in conjunction with the original TCC-positive urine sample; this task was successfully acquired after a mean 5.3 sessions. For the next training phase, four control and six TCC-positive urine samples were used, to see if the dogs had mastered categorical learning and could distinguish the samples in multiple combinations. Only one dog successfully completed this training phase.

The final phase of the study tested the dog using two novel TCC-positive urine samples and six novel healthy control samples. The dog was unable to distinguish the TCC-positive samples from the healthy ones at a greater rate than what would be expected to happen by chance.

These study authors concluded that training dogs to detect cancer using their sense of smell remains a challenge “in part because of variability in individual scent detection dog’s ability to learn an appropriate olfactory task, urine sample storage, and other confounding factors that could impair canine performance.”

While we are still a ways off from seeing dogs sniffing out diagnoses in the oncology ward, future studies with larger sample sizes and testing other types of cancers may perhaps show some merit in this unique detecting tool. In the meantime, if nothing else, the data give us one more reason to give our furry, four-legged friends an extra belly rub or two.