The Journal of Urology

Volume 193, Issue 4, April 2015, Pages 1382–1387

Investigative Urology

Olfactory System of Highly Trained Dogs Detects Prostate Cancer in Urine Samples

Presented at annual meeting of American Urological Association, Orlando, Florida, May 16-21, 2014.

• Gianluigi Taverna^{a, ,} ,
• Lorenzo Tidu^b,
• Fabio Grizzi^a,
• Valter Torri^c,
• Alberto Mandressi^a,
• Paolo Sardella^b,
• Giuseppe La Torre^b,
• Giampiero Cocciolone^b,
• Mauro Seveso^a,
• Guido Giusti^a,
• Rodolfo Hurle^a,
• Armando Santoro^a,
• Pierpaolo Graziotti^a

 Show more

doi:10.1016/j.juro.2014.09.099

Get rights and content

---

Purpose

We established diagnostic accuracy in terms of the sensitivity and specificity with which a rigorously trained canine olfactory system could recognize specific volatile organic compounds of prostate cancer in urine samples.

Materials and Methods

Two 3-year-old female German Shepherd Explosion Detection Dogs were trained to identify prostate cancer specific volatile organic compounds in urine samples. They were tested on 362 patients with prostate cancer (range low risk to metastatic) and on 540 healthy controls with no nonneoplastic disease or nonprostatic tumor. This cross-sectional design for diagnostic accuracy was performed at a single Italian teaching hospital and at the Italian Ministry of Defense Military Veterinary Center.

Results

For dog 1 sensitivity was 100% (95% CI 99.0–100.0) and specificity was 98.7% (95% CI 97.3–99.5). For dog 2 sensitivity was 98.6% (95% CI 96.8–99.6) and specificity was 97.6% (95% CI 95.9–98.7). When considering only men older than 45 years in the control group, dog 1 achieved 100% sensitivity and 98% specificity (95% CI 96–99.2), and dog 2 achieved 98.6% sensitivity (95% CI 96.8–99.6) and 96.4% specificity (95% CI 93.9–98.1). Analysis of false-positive cases revealed no consistent pattern in participant demographics or tumor characteristics.

Conclusions

A trained canine olfactory system can detect prostate cancer specific volatile organic compounds in urine samples with high estimated sensitivity and specificity. Further studies are needed to investigate the potential predictive value of this procedure to identify prostate cancer.

Key Words

• prostatic neoplasms;
• diagnosis;
• dogs;
• olfactory perception;
• volatile organic compounds

Abbreviations and Acronyms

• DRE, digital rectal examination;
• LR, likelihood ratio;
• PC, prostate cancer;
• PSA, prostate specific antigen;
• TURP, transurethral prostate resection;
• VOC, volatile organic compound

Prostate cancer represents the fifth most frequent cancer in the world.¹ Although PSA testing has increased PC detection, the main drawback is its lack of specificity and accuracy. High serum PSA levels can be detected in men with nonmalignant conditions.² Therefore, many men with increased PSA values undergo biopsy sampling,^{3, 4 and 5} although this procedure is invasive, offers a low level of accuracy (ie only 30% detection rate at the first biopsy) and is prone to various complications, including sepsis and death.^6 and 7 Consequently there is a need for a more sensitive diagnostic method.³ Dogs are used for detecting explosives and drugs through their olfactory system, which can perceive thresholds as low as parts per trillion.⁸ As outlined in 1971 by Pauling et al⁹ and in 2012 by Lippi and Cervellin¹⁰ VOCs can be identified in human urine samples. Several studies have shown that dogs may be trained to identify patients with cancer by tracing the presence of a unique odor signature.

In 1989 Williams and Pembroke provided the first evidence on sniffer dogs.¹¹ In 2001 Church and Williams reported on a 66-year-old man in whom a patch of eczema developed at which a pet Labrador persistently sniffed.¹² Histopathology revealed basal cell carcinoma. Since 2001, groups have reported the detection of bladder,^{13, 14 and 15} lung and breast,¹⁶ skin¹⁷ and ovarian^18 and 19 cancers, and infectious diseases²⁰ using the canine sense of smell.

Gordon²¹ and Cornu²² et al extended the use of detection dogs to PC. Although Gordon et al did not report positive results, they pointed out procedural errors that needed to be addressed by later researchers.²¹ They concluded that the study was unfortunately not successful but it provided lessons in the form of mistakes, which were presented in the hope that others might benefit from them. On the other hand, Cornu et al noted 91% sensitivity and specificity.²² Cornu et al took a step forward from Gordon et al,²¹ although they acknowledged important biases.²² A limited series of patients was enrolled, only 1 dog was used and the control group included patients older than 50 years with PSA greater than 8 ng/ml who were at high risk for undetected PC.

The opinion of Lippi and Cervellin that the most problematic issue has been the heterogeneity of performance among studies as well as in the same study¹⁰ together with the limited patient cohorts and nonstandardized training methodologies led us to design an accurate procedure to investigate whether dog olfactory detection remains a myth or could become a real clinical opportunity. We assessed diagnostic accuracy in terms of the sensitivity and specificity at which a rigorously trained canine olfactory system could recognize PC specific VOCs in urine in a large series of patients with PC of different stages and grades vs a heterogeneous control group.