Sir,
Meticillin-resistant Staphylococcus aureus (MRSA) is a frequently reported hospital- and community-acquired pathogen. New fast and inexpensive detection tools are needed for detection. Current molecular assays are based on detection of Staphylococcal Cassette Chromosome mec (SCCmec), which contains the mecA gene [
[1]
]. Second-generation polymerase chain reaction assays are single-locus assays targeting the open reading frame X (orfX)–SCCmec junction, and continue to have specificity issues related to strains which have lost the mecA gene but retained the SCCmec cassette junction (SCCmec remnant strains, MSSAr) [1
, 2
].Canine scent detection is a rapidly expanding field of study in medicine [
3
, 4
, 5
]. Studies have shown that meticillin-resistant and -susceptible (MSSA) S. aureus may be differentiated based solely on their volatile organic compound (VOC) profiles [6
, 7
]. We hypothesized that canine scent detection could be used as a diagnostic tool for the detection of MRSA. Our objective was to determine whether dogs could distinguish MRSA from the similar strains: MSSA, mecA-positive Staphylococcus epidermidis (MRSE), and MSSAr, using bacterial cultures.The study was approved by both the North York General Hospital Research Ethics Board and the Centre for Phenogenomics Animal Care Committee.
The MRSA type strains Canadian (C)MRSA-2 (SCCmec type II, USA100) and Canadian MRSA-10 (SCCmec type IV, USA300) were obtained from the National Microbiology Laboratory (Winnipeg, Canada). Other strains used were S. aureus ATCC® 25923™ (MSSA), a clinical isolate of MRSE, and SCCmec (Type I) remnant MSSA (MSSAr) [
[8]
]. Isolates were inoculated on to fresh 5% sheep blood agar plates (BAP) in a set pattern (Supplementary Figure 1A) and incubated at 37°C in 5% CO2 in sealed plastic bags to prevent cross-contamination of odours.Dogs came to the laboratory two or three days per week for a period of 1.5 years and four or five training sessions were performed each day. Each training/testing session consisted of six trials. Each trial included one CMRSA strain (target scent, S+) and two distractors (non-target scent, S–) in a line-up of three stations positioned 50cm apart [
[4]
]. Each station consisted of the prepared plate secured in a perforated stainless steel cylinder and a secondary metal basket to prevent direct contact. Equipment was cleaned prior to each trial with water and 30% ethanol and between sessions with hot water and unscented detergent, followed by 30% ethanol, hot water, and air-drying.Several dogs were recruited initially. The dogs were tested for potential nasal MRSA colonization once monthly during the study. All the training was conducted by a certified dog trainer using science-based positive reinforcement training. Dogs were initially taught to nose target S+. After the dog indicated the S+, the dog trainer marked the response with a clicker and delivered a food reward.
The discrimination task training took 68 sessions (about three months) using samples incubated for 16–24h. The S– gradually increased complexity; initially of uninoculated BAP, then BAP inoculated with unrelated bacteria (E. faecalis/E. coli) and finally, MRSE, MSSA and MSSAr. An errorless discrimination training process was used: S– samples initially had lower numbers of bacteria than S+ samples, and gradually S– quantity was increased to the same quantity as S+ over weeks of training [
[4]
]. The dog deemed to have the greatest affinity for the task, Nimbus (a female 3–4-year-old Golden Retriever), was chosen to complete the testing phase.Three testing phases were used: samples incubated for: (1) 16–24h, (2) 4h, and (3) samples inoculated but not incubated. CMRSA-10 was used for S+ in all three testing phases, whereas CMRSA-2 was introduced later. The testing was conducted in double-blind trials, where sample positioning was assigned using a random table by an assistant (Supplementary Figure 1B). The assistant retreated to a location where she could hear but not see the dog or the trainer. The trainer, blinded to the sample position, let the dog into the room cueing the dog to search while avoiding direct eye contact. The dog was off-leash and allowed to sniff the containers in any order (Supplementary video: https://youtu.be/cF2rOoF3HkY). When the dog indicated, the trainer verbally indicated the chosen station. If the response was correct, the assistant clicked and the trainer rewarded the dog (true positive). Sniffing but not indicating S– samples were considered true negatives. Indicating S– was defined as a false positive and sniffing but not indicating S+ was considered a false negative. In all, 96 trials were conducted for each testing phase.
In the first testing phase (16–24h) the dog was able to distinguish CMRSA-10 from MSSA, MRSE, and MSSAr (presented alone and in combination) with 97% sensitivity and 92% specificity (Table I). In the second testing phase (4h) initially sensitivity and specificity decreased to 75% and 83%, respectively (Table I). We speculate that this was due to a change in VOC profile related to the change in growth phase (stationary vs exponential) as well as a substantial decrease in the number of bacteria. After additional training with 4h samples (44 training sessions, 1.5 months) discrimination improved to 93% sensitivity and 93% specificity (Table I). Interestingly, the ‘0’h incubation phase, also representing a large change in the number of bacteria, did not require additional training and the sensitivity remained constant at 92% and specificity at 96%. Lastly, the dog had no difficulty distinguishing CMRSA-2 (4h incubation) (Table I). Sensitivity for CMRSA-2 was 90% and specificity 93%.
Table ISensitivity and specificity of canine scent detection in distinguishing bacterial cultures of CMRSA from MSSA, MRSE and MSSAr in double-blind trials
| Incubation time at 37°C | Sensitivity (95% CI) | Specificity (95% CI) | |
|---|---|---|---|
| CMRSA-10 | 16–24h | 97% (91–99) | 92% (87–95) |
| 4h: baseline | 75% (53–89) | 83% (69–92) | |
| Reassessment of 4h (after brief retraining) | 93% (85–97) | 93% (88–96) | |
| 0h | 92% (84–96) | 96% (91–97) | |
| CMRSA-2 | 4h | 90% (81–95) | 93% (88–96) |
CMRSA-10 or CMRSA-2 as the target scent (S+) was placed in one of three sample stations; each of the other two included one of following non-target scents (S–): (1) MSSA, (2) MRSE, (3) MSSAr, or (4) a combination of (1 + 2) or (1 + 3). The position of each sample in the line-up was assigned by using random tables. A total of 96 trials (16 data collection sessions of 6 trials) were conducted for each testing phase. 4 h baseline collection consisted of 24 trials (4 data collection sessions of 6 trials). All trials were video recorded and after collecting the data for each set, sensitivity, specificity and associated 95% confidence intervals were calculated (http://vassastats.net).
Here we demonstrate that canines can be trained to discriminate cultures of CMRSA-10 and CMRSA-2 from MSSA, MRSE and MSSAr, individually and in combination, across a spectrum of incubation intervals. The data presented are from the single dog assessed as most adept. However, we note that throughout the study other dogs also clearly achieved the ability to discriminate MRSA from S– isolates (data not shown).
Even though the signal (bacterial quantity) was substantially decreased throughout the study, this still represents a much higher signal than that expected in diagnostic specimens. Additionally, pure cultures are not representative of diagnostic specimens, which contain a wealth of other organic compounds from the host environment including other organisms. Future work will focus on continuing to reduce the amount of bacteria and the introduction of background material using actual patient samples to investigate whether dogs are able to generalize ‘an MRSA smell-print’ in positive samples.
This proof-of-principle study lays an important foundation for the development of canine scent detection as a potential diagnostic screening tool. Whereas it is likely that the gold standard method will continue to be used in hospital settings, there may be utility in resource-limited, industrial and agricultural settings.
Conflict of interest statement
None declared.
Funding source
The study was funded by North York General Hospital Exploration Fund.
Acknowledgements
We thank B. Powell, L. Pearson, K. Delaney, and A. Girardi-Peterman of Shared Hospital Laboratory for their assistance in the experiments; Dr A. McGreer with help with Animal Care Committee approval; Dr M. Desjardins for providing the MSSAr strain; Mr Kodis and Mr I. Hormila for their expert advice on scent detection; and Dr R.S. Flanagan for critically reading the manuscript.
Appendix A. Supplementary data
The following is the supplementary data related to this article:
Supplementary Figure 1. Sample preparation and experimental set-up. (A) The density of bacterial growth on agar plates. (B) Room set-up for double-blind trials.
References
- New real-time PCR assay for rapid detection of methicillin-resistant Staphylococcus aureus directly from specimens containing a mixture of staphylococci.J Clin Microbiol. 2004; 42: 1875-1884
- Genotypic and phenotypic characterization of methicillin-susceptible Staphylococcus aureus isolates misidentified as methicillin-resistant Staphylococcus aureus by the BD GeneOhm MRSA assay.J Clin Microbiol. 2011; 49: 1240-1244
- Canine olfaction: scent, sign, and situation.in: Horowitz A. Domestic dog cognition and behavior. Springer, Berlin2014: 3-29
- Canine olfactory detection of cancer versus laboratory testing: myth or opportunity?.Clin Chem Lab Med. 2012; 50: 435-439
- Using a dog’s superior olfactory sensitivity to identify Clostridium difficile in stools and patients: proof of principle study.BMJ. 2012; 345: e7396
- Identifying methicillin-resistant Staphylococcus aureus (MRSA) lung infections in mice via breath analysis using secondary electrospray ionization-mass spectrometry (SESI-MS).J Breath Res. 2014; 8: 041001
- Distinguishing methicillin-resistant and sensitive Staphylococcus aureus using volatile headspace metabolites.IEEE Sensors J. 2010; 10: 71-75
- Characterization of Staphylococcus aureus isolates with a partial or complete absence of staphylococcal cassette chromosome elements.J Clin Microbiol. 2010; 48: 3525-3531
Article info
Publication history
Published online: March 08, 2017
Accepted:
March 2,
2017
Received:
December 16,
2016
Identification
Copyright
© 2017 The Healthcare Infection Society. Published by Elsevier Ltd. All rights reserved.
