Materials and Methods

Samples were collected at 13 different clinical locations across the United States over 4 years (August 2019 to March 2023; Table 1). Two samples from each animal included in this study were collected in parallel, 1 for traditional culture and sensitivity testing and 1 for NGS testing. Traditional culture and sensitivity testing was carried out by the clinic’s standard culture testing lab and NGS testing by a commercial NGS-testing laboratory (MiDOG LLC). Inclusion criteria for sample collection were that the individual was an exotic pet or zoo/park wildlife animal and that the sample was taken due to a confirmed or suspected infection in the individual. Samples were collected from any relevant bodily locations using swabs from wounds, abscesses, skin, ocular, gastrointestinal, urogenital, and oral infections etc. A full list of animal species and sample types included in this study is listed (Table 2). A total of 78 matched samples were included in this study. For culture testing, aerobic bacterial incubation was ordered for all cases (78/78 [100%]) and additional anaerobic culturing was ordered for 65 patients (65/78 [83%]). For a subset of the samples (53/78 [68%]), fungal incubation was requested by the clinician. Specific details about each sample can be found elsewhere, including the type of incubation and testing laboratory (Supplementary Table S1). Sterile swabs for culturing were supplied by the clinician’s lab routinely. Fungal and bacterial NGS was performed for each sample. A sterile DNA preservation buffer (DNA/RNA Shield; catalog No. R1100-50, Zymo Research Corp; 1-ml volume) was provided by the NGS-testing laboratory that stabilized the samples at room temperature until they were shipped to the testing facility. This solution can preserve the microbial DNA present in the specimen at room temperature for up to 30 days. A sterile, DNA-free swab (HydraFlock; Puritan; catalog No. 25-3406-H) was provided as well as a standardized protocol, which instructed participants to aseptically remove the swab from the pouch and twist and twirl the swab 10 times gently back-and-forth over the sampled location. All participating clinics used the same collection devices for the NGS-based analysis.

DNA sequencing analysis was conducted as previously described, targeting the 16S ribosomal RNA taxonomic marker gene for bacterial and the ITS marker for fungal identification and quantification.²⁰ Several positive (ZymoBIOMICS microbial community standard; catalog No. D6300 and D6305; Zymo Research Corp) and negative controls are part of the standardized protocol adhered to by the NGS-testing laboratory.

The results provided by each laboratory were compared to the matched samples for the presence of microbes detected and the identified bacterial and fungal species. When no fungal cultivation was ordered for traditional culture testing, but fungal pathogens were diagnosed using NGS for that sample, this was highlighted in the results and discussion. Specific pathogens of clinical interest in exotic pets, wildlife, and zoo animals, determined by literature search^4–7,10 were highlighted if they were detected in the sample by either method. Those pathogens of special clinical interest are listed (Table 3). Routinely, commensals are either not listed in culture reports or summarized as “commensals detected.” If commensals were reported by either culture or NGS, those results were also compared, since commensals can be of clinical importance in certain samples.

Results

The study dataset included 78 samples, collected from 15 different bird species (n = 22), fish (1), 11 mammalian species (40), and 6 reptilian species (15; Table 2). The most common species investigated were rabbit (n = 22), guinea pig (7), Congo African grey (5), python (4), and sulcata tortoise (3). The most common sample collection sites were skin swabs (n = 15), feces (7), ear swabs (7), facial abscesses (6), bronchoalveolar lavage (BAL) lung fluids (4), and nasal swabs (4; Table 2; Supplementary Table S1). Samples were collected in duplicate for either traditional culture-based testing, including aerobic bacterial culture for all samples and anaerobic culture for 65 cases (83%), or NGS-based testing. A subset of the samples (n = 53, 68%) was also sent for fungal culture testing. All NGS testing was conducted by the same laboratory, but participating clinics submitted the samples for culturing at their preferred reference lab, which included 2 major US-based animal diagnostic laboratories (n = 56, 72% of samples), as well as local private laboratories (19, 24%) and university laboratories (3, 4%). Specific details on which culture laboratory analyzed which sample and the microbes detected in each test are provided (Supplementary Table S1).

Bacterial species were reported in 83% of samples by culture (n = 65), of which 63 (81%) cases included a putative pathogen, and 97% of samples by the NGS test (76), of which all cases included a putative pathogen (Figure 1; Table 3). Of the 53 samples tested both by fungal culture and NGS, growth was detected by culture in 7 cases (9%), 6 of which reported a putative pathogen and the seventh reported an unidentifiable yeast species; the NGS test detected fungal species in 58% of samples (n = 45), with 43 cases reporting a putative pathogen. Considering the sample set as a whole, including the additional samples which were not tested by fungal culture, NGS detected fungi in 88% of cases (n = 70). In summary, culture results failed to grow 15% of putative bacterial and 81% of fungal pathogens that were identified by NGS testing.

View Full Size

A—Graphical summary of study design and results for the matched sample comparisons. B—Codetection and type of organisms (bacterial and/or fungal) detected in the specimen by NGS testing (left) and culture testing (right).

Citation: American Journal of Veterinary Research 84, 8; 10.2460/ajvr.23.03.0054

• Download Figure
• Download figure as PowerPoint slide

Organisms were considered of interest to this study if they have previously been reported as either a confirmed pathogen or an opportunistic pathogen. A total of 83 bacterial genera and 29 fungal genera were identified from the 2 diagnostic approaches, with important pathogens, as listed (Table 2). The most prevalent bacterial species identified by both culture and NGS was Pseudomonas aeruginosa (culture, n = 14; NGS, 15; Table 3). Other commonly observed bacteria included Staphylococcus spp (including S aureus, S chromogenes, S cohnii, S epidermidis, S felis, S haemolyticus, S hominis, S massiliensis, S pseudintermedius, S saprophyticus, and S xylosus [culture, n = 9, 12%; NGS, 27, 35%]), Corynebacterium spp, (including C aurimucosum, C diphtheriae, C jeikeium, C kroppenstedtii, C pseudogenitalium, and C urealyticum [culture, 4, 5%; NGS, 26, 33%]), Escherichia coli (culture, 11, 14%; NGS, 7, 9%), Enterococcus spp (including E avium, E canintestini, E faecalis, and E faecium [culture, 12, 15%; NGS, 15, 19%]), and Klebsiella spp (including K oxytoca and K Pneumoniae [culture, 12, 15%; NGS, 4, 5%]). The most prevalent potentially pathogenic fungal species detected were Malassezia restricta (culture, not detected; NGS, n = 22, 42%), Alternaria sp (culture, not detected; NGS, 12, 23%), Aspergillus spp [including A fumigatus, A niger, A ostianus, A restrictus, A steynii, and A unguis [culture, 1, 2%; NGS, 18, 34%]), Trichosporon sp (culture, not detected; NGS, 11, 21%), and Candida spp (including C albicans, C catenulata, C parapsilosis, and C tropicalis [culture, 3, 4%; NGS, 16, 30%]). Significant pathogens including Mycobacterium, Mycoplasma spp (M cavipharyngis and M gallisepticums), and Helicobacter spp (H trogontum and unknown species) and the fungal genera Alternaria, Malassezia (including M caprae, M furfur, M globosa, M restricta, and M slooffiae), and Nannizziopsis chlamydospora were not detected by culture test but were detected by NGS (Table 3).

Organisms detected by each test were compared by performing a matched sample comparison (Table 4). Agreement between culture and NGS results for bacterial species varied depending on species: For Streptococcus pneumoniae and Vibrio spp, culture and NGS were in 100% concurrence, meaning these organisms were detected by both methods in all cases, and for S pseudintermedius, there was only 1 case that was positive for NGS but negative for culture. For organisms including P aeruginosa, E coli, A baumanii, E faecium, and E faecalis, concurrence was ≥ 37%; however, concordance was low (< 30%) for organisms including Corynebacterium spp, E cloacae, Fusobacterium spp, S enterica, Staphylococcus spp, and Streptococcus spp (Table 4). No concordance was reported for Citrobacter spp, Helicobacter spp, Klebsiella spp, Micrococcus spp, or Mycoplasma spp.

Interestingly, there appeared to be a high incidence of multifactorial infections, with multiple putative pathogens detected within a single sample. Out of 64 samples measured by the culture test to contain a putative bacterial pathogen, 30 (47%) had > 1 pathogens reported, and of the 76 pathogen-positive samples detected by NGS, 69 (91%) had > 1 pathogens. While fungal culture only detected 1 incidence of multifactorial fungal infection, 50% of NGS reports detected more than 1 putative fungal pathogen. While culture testing rarely reported commensal species (2/78, 2.6%), NGS testing reports all organisms present enabling the analysis of all species present as a whole. A high incidence of commensal species reported alongside pathogenic species was observed by the NGS test (Table 5). In the majority of cases, putative bacterial pathogens were detected alongside commensal species (80%), and pathogenic species were rarely reported in isolation (10%; Table 5). Fungal pathogens were also more likely to be detected together with commensal species (62%). The majority of cases reported both bacterial and fungal pathogens present in the samples according to NGS analysis (55 cases, 1%), and 8 cases found bacterial pathogens along with fungal commensals (10%).

Other rare pathogens detected in the dataset included Chelonobacter oris, which causes respiratory disease in tortoises,²⁵ Trueperella pyogenes, which causes infections in swine, cows, and domestic animals,^26,27 and Capnocytophaga canimorsus, which is zoonotic in humans but a commensal in dogs and cats²⁸ (Supplementary Table S1).

Direct comparison of the reports highlighted that the probability of getting a “no growth” diagnosis was 14% higher for bacteria and 49% higher for fungi with traditional culture versus NGS testing, even if fungal culturing was conducted. If all fungal-positive cases are taken into account, the probability of getting a no growth diagnosis is at 74%.

Discussion

Successful infection treatment depends on the correct identification of the pathogen to provide the most appropriate therapies for the patient. While culture is still considered the gold standard for clinical microorganism identification, recent investigations have highlighted that culture-free, NGS-based technologies may be less biased in characterizing microbial profiles.²⁹ In this comparative study, we investigated the results obtained from culture-based and NGS-based tests for 78 samples collected from exotic pets and zoo animals, across a range of body sites and geographical locations.

The data presented here showed that the probability of getting a no growth diagnosis was 14% higher for bacteria and 49% higher for fungi with traditional culture versus NGS testing (and 81% if culture reports that did not conduct fungal testing are taken into account). Only 1 NGS report indicated neither bacterial nor fungal species detected (1%), which could be due to the sampling method or detection limitations of the assay; however, for culture testing 8 samples (15.1%) reported no detection of any bacteria or any fungi, even though fungal culture was ordered. On average, NGS testing reported more pathogens, both bacterial and fungal, and more commensals. Specifically, several clinically relevant pathogens were detected using NGS but not reported by culture. Those pathogens included Mycobacterium, Mycoplasma, and Helicobacter for bacteria and Alternaria, Malassezia, Nannizziopsis, Trichosporon, Finegoldia magna, and Fusarium for fungi. In addition to these well-known pathogens, novel putative pathogens such as Riemerella anatipestifer,^30,31 Granulicatella adiacens,³² Corynebacterium xerosis,³³ Corynebacterium aurimucosum,³⁴ Fusarium,²⁰ Anaerosporobacter spp (noted by Vecere et al, DVM, unpublished data, 2023), Clostridium disporicum,^35,36 and Bibersteinia trehalosi^37,38 were detected by NGS and undetected by culture. A benefit of NGS diagnostics is that genomic databases describing novel bacterial and fungal pathogens isolated from patients are constantly updated by researchers worldwide, allowing for pathogen identification of even novel and rare organisms.

When added together, culture failed to identify 15% of putative bacterial pathogens. Specifically, strict and facultative anaerobic pathogens were underrepresented in culture reports, such as Enterococcus, Salmonella, and Helicobacter. Anaerobic bacteria require special handling from the moment the sample is collected to avoid cell death and an accurate representation of the sample profile in the lab.²

Interestingly, culture testing reported a higher incidence of the gram-negative bacteria E coli and Klebsiella spp than NGS, while NGS detected important gram-positive bacteria such as Corynebacterium spp. While the true composition of the samples at the time of collection is unknown due to the nature of a clinical sample, it might be possible that those bacteria grew disproportionately during transit to the culture laboratory or are contaminants. Amir et al³⁹ investigated the effects of sample storage on the microbial profile of human fecal samples and observed microbial “blooms,” in which the taxonomic profiles of samples changed over time due to growth of particular taxa, including those in the Gammaproteobacteria class. The DNA preservation buffer used for NGS-based sample collection in this study does not allow for growth or decay of any organism as the cells are efficiently inactivated, thus taking a snapshot of the sample at the time of collection.⁴⁰ Therefore, a microbial “blooming” during transit cannot be observed in this NGS-based approach.

Slow-growing and fastidious organisms provide a challenge for traditional culture testing as fast-growing organisms requiring simple growth matrices may overtake the petri dish, thus causing the slow-growing organisms to be missed in the clinical diagnosis. Two examples of such slow-growing pathogens were detected in this study, Corynebacterium urealyticum, which causes cystitis, pyelonephritis, and bacteremia,⁴¹ and Mycoplasma species (M cavipharyngis and M gallisepticum). These samples highlighted that such slow growers can be identified based on their DNA, while not being detected by culturing. Interestingly, in 1 case the slow-growing Capnocytophaga ochracea pathogen was reported in culture testing, but not using NGS. Instead, NGS reported Fusobacterium nucleatum to be present in this sample, which has a similar colony morphology to C ochracea, and may have been misidentified by culture.

Four rabbit cases presented in this study showed several pathogens of interest. One abscess showed high cell counts of Granulicatella adiacens by NGS testing but no bacterial growth in culture. This bacterium has fastidious growth requirements that make cultivation challenging. While G adiacens is considered a commensal in humans, it has also been shown to cause endocarditis, bacteremia, meningitis, and peritonitis.⁴² To date, G adiacens has not been reported as a pathogen in rabbits. Another nonculturable species detected in a rabbit nasal sample was Treponema pallidum-paraluiscuniculi, which causes venereal spirochetosis.⁴³ Furthermore, 2 rabbit cases showed Helicobacter sp in the NGS reports but not the culture reports.

In 81% of cases, important fungal pathogens were not identified by culture, thus indicating that fungal infections are more likely to be misdiagnosed as bacterial infections and in general are underreported. This dataset specifically highlighted that reptiles quite often suffer from fungal infections, which were undiagnosed with traditional methods. Severe fungal pathogens detected here included Nannizziopsis, Fusarium, Malassezia, Trichosporon, Candida, Aspergillus, and Alternaria.

Potential causes for the no growth results by culture test for both bacteria and fungi could be that the bacterial and fungal cells are no longer viable after transit due to the transport media utilized or that growth did not occur due to the type of culture media used or the incubation temperature applied. Incubation temperature is often set to 37 °C, which is optimized for samples from feline and canine patients. However, the body temperature of exotic animals is significantly higher (ie, most avian pet and small exotic pet mammals) or lower (ie, reptiles, amphibians, fish, and invertebrates)^4,6,7 than 37 °C. Therefore, microbial cells that are viable and causing infections in vivo may be unculturable in vitro due to the applied incubation temperature. It should also be noted that the use of different culture laboratories utilized in this study may have impacted the results obtained via the culture test, although all labs were accredited and no correlation between a no growth observation and the commercial laboratory that performed the testing was noted.

A commonly expressed concern about NGS-based diagnostics is the high amount of data being provided, possibly masking the clinically relevant information with commensal organisms, the fact that nonviable microbes are reported, and potential contaminants. A high amount of clinically irrelevant information is a valid concern; however, most diagnostic laboratories filter the information for the clinicians, highlighting clinically relevant organisms, as with culture testing. While culture testing may report “commensals detected” without further specification, knowing which commensals are detected could provide valuable clinical information, as under certain circumstances commensals can be opportunistic pathogens.^44,45 Microbiome research performed on a diversity of animal species and body locations has demonstrated that a diverse commensal community is significantly associated with a clinically healthy individual^{20,23,46–48} and that a lack of a diverse commensal community can be an indicator of an infection about to manifest itself. Especially in exotic medicine there is a lack of microbial reference ranges that define the microbiome profile of a clinically healthy individual, which species are commensals, and at which cell counts.¹⁰ In some cases, even high densities of otherwise nonpathogenic species can compromise a patient’s health.¹⁰ Often, traditional culturing refers to commensals based on previous knowledge obtained from human and veterinary medicine of common domestic species and this may be different for exotic pet species.¹¹ Furthermore, a commensal nonpathogenic bacterium in 1 organ system such as skin or the gut can be pathogenic once populating another organ such as the auricular bulla.¹²

NGS testing does provide the opportunity to further understand the clinically healthy microbiome of exotic animals for all body locations and guide ultimately the clinicians on which species are of clinical concern and warrant a treatment. Environmental contaminants are a concern of all microbial tests, including culture and PCR testing, and are thus not unique to NGS-based testing. As previously discussed by Eisenhofer et al,⁴⁹ NGS testing in a clinical diagnostic setting requires a high-quality control system designed to ensure immaculate sample preparation and analysis with the highest standard of precision, data accuracy, and controls.

Limitations of this study include the nature of the opportunistic sample collection. Ideally, a consistent sample host and type would have allowed a stronger signal of the pathogens reported or potentially missed by either diagnostic tool. In addition, the culture testing was not standardized, as each participating clinic submitted the samples for aerobic and anaerobic culturing at their preferred reference lab. Fungal culturing was only ordered in a subset of samples (68%), which makes a direct comparison of the results challenging. The geographic location of the patient may also have influence on the types of infections more prevalent in the given region.

The true composition of the microbial profile of the samples at the time of collection remains unknown. An unbiased comparison between the 2 diagnostic tools could be achieved by performing both tests on predefined mock community samples that contain a known mixture of pathogens and commensals at known concentrations. A limitation of this approach, however, would be that the host matrix and other confounding factors that can affect the success of the diagnostic test would not be accounted for in such an artificial setting. Opportunistic sample collection and comparison still offer an insight into the day-to-day clinical operations across the country and present a real-life perspective on the diagnostic results provided by each method. The data provided here may be used as a pilot study and provide some guidance toward a future systematic nationwide study.

In conclusion, this article provided the first study directly comparing culture- and NGS-based diagnostics for exotic animals suffering from suspected infectious diseases. Specifically, slow-growing pathogens, fungal pathogens, strict and facultatively anaerobic bacterial pathogens, novel and rare pathogens, and gram-positive bacteria were often not identified by culture but were diagnosed using NGS. Culture results failed to grow 15% of putative bacterial and 81% of putative fungal pathogens that were identified by NGS.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org.