Potomac horse fever in Ontario: Clinical, geographic, and diagnostic aspects

Abstract

Clinical findings, geographic locations, laboratory diagnoses, and culture isolation of Neorickettsia spp. in Potomac horse fever (PHF) cases diagnosed in Ontario between 2015 and 2019 are described. Forty-six confirmed PHF cases occurred from late June to early September. Of 41 horses admitted to the Ontario Veterinary College, 28 (68%) survived and 13 (32%) were euthanized due to poor prognosis or financial constraints. Most cases were in southern Ontario along the Canada-USA border. Blood and fecal samples from 43 suspect PHF cases were submitted to 2 laboratories for polymerase chain reaction (PCR) testing for Neorickettsia risticii. Agreement between both laboratories for detection of N. risticii DNA was excellent for feces [κ = 0.932, 95% confidence interval (CI): 0.80 to 1], and fair for blood samples (κ = 0.494, 95% CI: 0.13 to 0.85). Neorickettia spp. were isolated from 16 of 41 (39%) blood samples. DNA analysis confirmed 14 isolates were N. risticii and 2 were N. findlayensis, a novel species of Neorickettsia recently demonstrated to cause PHF.

Introduction

Potomac horse fever (PHF) is an acute and potentially fatal enterotyphlocolitis of horses caused by infection with the monocytotropic rickettsia Neorickettsia risticii (formerly Ehrlichia risticii) (1). The disease is seasonal, and is characterized by depression, anorexia, fever, dehydration, diarrhea, laminitis, and occasionally abortion (2,3). Potomac horse fever was first recognized as a clinical entity in 1979 by veterinarians in an area adjacent to the Potomac River in Maryland and Virginia in the United States (3–5). Review of investigations conducted by Dr. Frank W. Schofield into an endemic disease of horses in the Kent and Essex counties of Ontario in the summer of 1924 strongly suggests that this endemic disease was PHF (6,7).Interestingly, Schofield stated that some of the oldest inhabitants in those counties suggested that the disease had existed for nearly half a century before 1924 (6,7).

The serum indirect fluorescent-antibody (IFA) test was the only test available for the diagnosis of PHF in Ontario through the 1990’s (8,9). Every summer, suspected cases of PHF are reported by veterinarians in many regions throughout Ontario and there is an apparent regional variation in the number of cases each year (7,10–11). A presumptive diagnosis of PHF is made by veterinarians based on clinical signs, month of the year, historical evidence of the disease in the area, and response to antimicrobial therapy; however, laboratory testing is often not requested due to the cost of testing and a perceived low number of positive results. Ideally, these cases should be confirmed by molecular detection and quantitation of N. risticii genomic DNA from blood and/or feces by polymerase chain reaction (PCR) (12–14). Which sample to submit (blood or feces) is a common concern of equine veterinarians, particularly those working within financial constraints. In horses that were experimentally infected with N. risticii, there was a difference between blood and feces in the time when the PCR test was positive (12,15–17). These differences in detection times have not been examined in blood and fecal samples from clinical PHF cases. In addition, there has been no inter-laboratory comparison of N. risticii PCR results previously reported.

Isolation of the agent in cell culture is the gold standard for diagnosis, but it is time-consuming and not readily available as a commercial service (14). Thus, practicing veterinarians must rely on a PCR diagnosis for PHF. However, this becomes frustrating and misleading if the PCR does not detect the organism causing disease in their area.

The aims of this study were: i) to review the clinical findings, geographic locations, and laboratory results of 46 horses with a confirmed PCR diagnosis of PHF in Ontario between 2015 and 2019, ii) to assess the level of agreement between 2 diagnostic laboratories for molecular detection of N. risticii DNA in blood and fecal samples from horses with clinical signs consistent with a diagnosis of PHF, and iii) to report on the cultural isolation of Neorickettsia spp. from blood, from horses suspected of having PHF.

The hypothesis is that PHF is widely distributed throughout Ontario with areas of increased occurrence in southwestern and eastern Ontario, and that the results of PCR testing on blood and feces between 2 laboratories will be similar.

Materials and methods

Retrospective study of horses diagnosed with PHF in Ontario (2015–2019): Geographic location and clinical findings

All horses that tested positive for N. risticii by laboratory diagnosis at the Ontario Veterinary College Veterinary Teaching Hospital (OVC-VTH) and/or the Animal Health Laboratory (AHL) at the University of Guelph, from June 2015 to September 2019, were included in this study. All horses were tested for N. risticii using either a PCR assay on samples of blood and/or feces and/or culture isolation of Neorickettsia spp. from blood. The following information was provided by the attending veterinarian: year, month, age, sex, breed, location of the farm, presenting complaint, treatments prior to admission, and time since first clinical signs were noted. Data from the OVC-VTH medical record, including clinical signs, in-hospital treatments, duration of hospitalization and outcome, were recorded. For horses that died or were euthanized, post-mortem findings were recorded. Descriptive statistics were generated for all variables in the dataset. Population characteristics were described using median and range values.

Prospective study testing the level of agreement of 2 diagnostic laboratories for detection of N. risticii by PCR on both blood and fecal samples

In 2018, equine practitioners in Ontario were contacted and asked to submit samples of blood and feces from horses suspected of having PHF. Samples were collected by the attending veterinarian at the farm and shipped on ice overnight for next day delivery. Upon arrival, the samples were aliquoted and a pair of blood and fecal samples was submitted to Laboratory A (university-based laboratory) and Laboratory B (commercial laboratory). All horses sampled were located in the province of Ontario and sample collection took place between June 1, 2018 and September 30, 2018. The geographic location of each horse was recorded. The Animal Care Committee at the University of Guelph approved the study which conformed to the standards of the Canadian Council on Animal Care.

Differences in the proportion of animals with a positive PCR for N. risticii DNA detected by each laboratory were examined using Fisher’s exact test. McNemar’s test was used to evaluate whether the PCR assays in the 2 laboratories were equally likely to identify horses with PHF. The level of agreement between the 2 laboratories in detecting PCR positive samples for PHF was explored using the Kappa coefficient test. Similarly, the level of agreement between laboratories for detecting PHF in blood and fecal samples was assessed using the Kappa coefficient test. The Kappa agreement was judged as poor when 0 ≤ κ ≤ 0.40, fair when 0.41 ≤ κ ≤ 0.59, good when 0.60 ≤ κ ≤ 0.74, and excellent when 0.75 ≤ κ ≤ 1.0 (18).

Prospective study of blood culture from horses suspected of having PHF

Blood samples were obtained in the summer months of 2015 to 2018 from 41 horses suspected by equine practitioners to have PHF based on a combination of clinical signs including fever and or diarrhea, and geographical locations that had previously confirmed PHF cases (6–9). Two of the authors (JDB, LGA) identified horses through clinical examination or a referral telephone consultation requested by the attending veterinarian. All blood samples obtained from horses that met the inclusion criteria were submitted for both PCR for N. risticii DNA and blood culture for isolation of Neorickettsia spp. Samples were not submitted for culture if the horses had been previously treated with antimicrobials. Ethylenediaminetetraacetic acid (EDTA) blood samples (~30 mL) packed in ice were submitted for Neorickettsia spp. culture by overnight delivery to the Molecular, Cellular, and Environmental Rickettsiology Laboratory, Department of Veterinary Biosciences, College of Veterinary Medicine, The Ohio State University.

Results

Retrospective study of horses diagnosed with PHF in Ontario (2015–2019): Geographic location and clinical data

During 2015 to 2019, 46 horses were confirmed by laboratory testing with a diagnosis of PHF in Ontario (Figure 1). Of the 46 horses, 22 were females, 20 were geldings, and 4 were stallions. One horse was positive in 2015, 9 in 2016, 15 in 2017, 13 in 2018, and 8 in 2019 (Figure 2). Regarding the month of presentation, 2 cases were presented in June, 21 in July, 15 in August, and 8 in September. The median age was 9 y (range: 1.2 to 27 y). Breed information was available for 26 horses, with Thoroughbreds (n = 9) being the most frequent. The medical history for seeking veterinary attention included diarrhea in 29 (63%) horses, fever in 14 (30%), and colic in 3 (7%). The duration (median) of the clinical signs before veterinarian evaluation was 2 d (range: 6 h to 7 d). Seven horses had laminitis at the initial veterinary evaluation. Outcomes were available for 41 horses, 28 (68%) horses survived and 13 (32%) were euthanized due to poor response to therapy, prognosis, or financial constraints.

Figure 1.

Map of southern Ontario showing the geographic origins of Potomac Horse Fever cases. The colored dots depict the sample tested and the method of diagnosis.

Figure 2.

Map of southern Ontario showing the geographic origins of Potomac Horse Fever cases. The colored symbols depict the year of occurrence.

The exact location of affected horses was recorded in 34/46 cases (Figure 1). Most cases occurred in southern Ontario, along the Canada-USA border, where PHF has been previously diagnosed (7); however, 1 case was identified further north in Simcoe County (Figure 1). Of the 46 PHF cases diagnosed in Ontario, 23 were treated at the Large Animal Hospital, OVC-VTH. The 23 horses had multiple clinical findings on admission, with the most common being diarrhea and fever in 10 (43%), fever in 6 (26%), diarrhea and colic in 3 (13%), diarrhea, colic, and fever in 2 (9%), and fever and colic in 2 (9%). The duration (median and range) of signs before admission (reported by the owner or referring veterinarian) was 2 d (range: 1 to 7 d). Of the 23 horses, 39% (9/23) received 1 or more antimicrobial agents before admission. Three of 9 horses received intravenous (IV) oxytetracycline, 3 IV sodium ceftiofur, 2 oral metronidazole, 2 oral trimethoprim-sulfa (TMS), and 1 a combination of IV penicillin and gentamicin. Eighteen (78%) horses received anti-inflammatory therapy; 12 (52%) received flunixin meglumine, 2 received phenylbutazone, and 4 received a combination of phenylbutazone and flunixin meglumine.

At admission, 5 horses received hypertonic sodium chloride solution (5%), and all horses received IV fluid therapy with lactated Ringer’s solution (Baxter, Mississauga, Ontario) continuous rate infusion (CRI) at a rate of 100 mL/kg body weight (BW) per day during the first 24 h, then adjusted accordingly to the hydration status. Intravenous fluid therapy was administered to the hospitalized cases for a median duration of 4 d (range: 2 to 10 d). Two horses received a polyimmune equine plasma transfusion (Equi-Plas; Plasvacc USA, Templeton, California, USA). All 23 horses were treated with oxytetracycline (Liquamycin LA-200; Zoetis Canada, Kirkland, Quebec), 6.6 mg/kg BW, IV q12h or q24h. Seventeen (74%) horses that were presented to the hospital without signs of laminitis were treated with cryotherapy using sleeve-style digital cryotherapy and equine comfort boots (Soft-ride; Bacliff, Texas, USA). Despite cryotherapy, 4 of 17 horses developed laminitis during hospitalization. Horses with laminitis were managed using pain control protocols that included non-steroidal anti-inflammatory drugs (NSAIDs) (flunixin meglumine or phenylbutazone), morphine, or CRI of butorphanol. Fourteen (61%) of the 23 horses were discharged from the hospital, whereas 9 (39%) were euthanized. Seven of 10 horses with laminitis (either identified at presentation or developed during hospitalization) were euthanized due to this complication. Autopsy results for 6 horses included a diagnosis of necro-hemorrhagic and ulcerative enterocolitis (n = 1), severe necrotizing neutrophilic colitis (n = 5) consistent with N. risticii infection, and acute laminar epidermal necrosis and fibrinous, neutrophilic laminitis with hemorrhage.

Study comparing results from 2 laboratories for PCR testing of blood and feces

Detection of N. risticii in blood samples

Paired blood samples (n = 43) were submitted to compare the level of agreement between Laboratory A and Laboratory B for detection of N. risticii by PCR assay. Overall, 8 horses tested positive for N. risticii in blood when results from both laboratories were combined. Laboratory A detected 8/43 (19%) horses positive for PHF, whereas Laboratory B identified 3/43 (7%) (P = 0.0045). Kappa coefficient analysis showed a fair agreement between both laboratories for detection of N. risticii using PCR assay (κ = 0.494, 95% CI: 0.13 to 0.85). McNemar’s test showed that both laboratories were not equally likely to detect N. risticii by PCR (P = 0.025).

Detection of N. risticii in fecal samples

Paired fecal samples (n = 42) were submitted for PCR assay for the detection of N. risticii and the level of agreement was compared between Laboratory A and Laboratory B. Overall, 11 horses tested positive for N. risticii in feces. Laboratory A detected 11/42 (26%) horses positive for N. risticii, whereas Laboratory B identified 9/42 (21%) (P < 0.001). Kappa coefficient analysis showed an excellent agreement between the laboratories for detection of N. risticii DNA (κ = 0.932, 95% CI: 0.80 to 1). McNemar’s test showed that both laboratories were equally likely to detect N. risticii DNA (P = 0.317).

Detection of N. risticii DNA in blood and fecal samples at Laboratory A

At Laboratory A, 42 blood and 43 fecal samples were analyzed. This laboratory detected 11/42 (26%) horses positive for PHF in feces and 8/43 (19%) in blood (P = 0.05). Kappa coefficient analysis showed a good agreement between both samples for detection of N. risticii DNA (κ = 0.632, 95% CI: 0.434 to 0.812).

Detection of N. risticii DNA in blood and fecal samples at Laboratory B

At Laboratory B, 42 paired blood and feces were analyzed for detection of N. risticii DNA. Three (3/42) (7%) horses were positive in blood and 9/42 (21%) were positive in feces (P < 0.001). Kappa coefficient analysis showed a poor agreement between both samples for detection of N. risticii nucleic acids (κ = 0.308, 95% CI: 0.20 to 0.53).

Culture for Neorickettsia spp. from blood samples of horses suspected of having PHF

Sixteen of 41 (39%) samples yielded a positive culture for Neorickettsia spp. organisms. Two of the 16 isolates were classified as a novel Neorickettsia spp. now designated N. findlayensis (11). Phylogenetic analysis of 12/16 of these Ontario isolates demonstrated clustering according to the geographic area of origin (11). These horses were located 200 to 6000 m (median: 2000 m) from water sources, such as lakes, rivers, or large ponds.

Discussion

This study documents the increasing number of PHF cases confirmed in Ontario over recent years and the challenges in confirming this diagnosis. In a previous report, 20 cases of PHF were confirmed at OVC over a 15-year period (7). Between 2015 and 2019, the number of cases of PHF confirmed at the Animal Health Laboratory, University of Guelph has increased. The number of PHF cases referred to OVC-VTH (n = 23) has also increased (7). There is now a greater awareness of the risk of the occurrence of PHF across southern Ontario, and this may explain, at least in part, the increased case numbers. Further, rapid and improved diagnostic tests, such as PCR, may also have played a role in the greater number of confirmed cases. However, an increased incidence of the disease should also be considered. There is an apparent year-to-year fluctuation in the number of PHF cases, and this warrants further epidemiological investigation in Ontario.

Most cases of PHF originated from areas previously recognized as endemic regions of the province for this disease (Figure 2). Clusters of cases were identified around Lake St. Clair and Lake Erie (Figure 2) as previously reported by Schofield in 1925. Culture-positive horses lived near (< 6 km) fast-running (rivers, creeks) and/or standing water (lakes, ponds). Neorickettsia infections, including those by N. risticii, have invariably been connected to lotic ecosystems; however, snails that live in lentic habitats have also tested positive for N. risticii (19,20). Both lotic and lentic ecosystems have been identified in association with this subset of Ontario horses; however, the source of infection remains unknown. One PHF-affected horse resided in a northern Ontario region (Simcoe County) where the disease has not been previously confirmed (11).

In recent years, major environmental efforts have been made to decrease water pollution in the Great Lakes and other bodies of water throughout Ontario. This may have positively impacted the number of emerging Hexagenia spp. (mayflies) (21). These aquatic insects play an important role in the life cycle of Neorickettsia spp., and after their emergence they can travel inland for up to 8 km onto farms (22). This may in part account for the increasing reports of PHF throughout Ontario; however, this hypothesis remains to be tested.

As expected, and widely reported, PHF is a seasonal disease with higher incidence during the summer months. Although most cases cluster during the months of July and August (78%), some cases may appear as early as June and as late as October (7–9). Equine veterinarians and horse owners should be made aware of the disease pattern and the clinical signs for early case identification, prompt treatment, and/or referral. Similar to other studies, there was no breed or sex predilection for the development of PHF in the present study. All horses in this study were older than a year, with a median and age range similar to previous reports (23). Although this disease is known to affect foals as young as 4 mo of age (23), PHF has not been diagnosed in foals at the OVC-VTH.

The clinical presentations of PHF in Ontario were similar to those previously reported (2,23) with diarrhea and fever among the most common signs of PHF, and when combined, these signs represent 93% of the cases. Leukopenia, characterized by neutropenia, was the most remarkable hematological abnormality. However, this finding is typical of most horses with colitis. Serum biochemistry abnormalities including hyponatremia, hypochloremia, and hypoalbuminemia are commonly observed in diarrheic horses and have been previously reported in PHF cases (23,24). Supportive therapy with IV fluids is critical for horses with enterotyphlocolitis and therefore it is not surprising that all horses referred to OVC received IV fluid therapy. Oxytetracycline is the antibiotic of choice for infections with Neorickettsia spp. and all horses admitted were treated with IV oxytetracycline at the prescribed dose of 6.6 mg/kg BW, q12 to 24h, usually for 5 d. Potomac horse fever cases generally improve markedly within 24 h to 48 h after the commencement of antimicrobial therapy; however, some horses may take longer to respond.

Molecular detection of N. risticii from clinically suspected PHF cases is the widely used and preferred diagnostic test, and was assessed simultaneously at 2 veterinary diagnostic laboratories in Ontario. The number of positive PCR tests for fecal samples was similar between the 2 laboratories; however, there was a marked discrepancy in the proportion of positive PCR tests in blood samples (Table 1). In addition, laboratory A had a good level of agreement for the detection of N. risticii in blood or feces, while Laboratory B had a low level of agreement between the results for the 2 samples. A difference in the time and duration of molecular detection of N. risticii in blood versus fecal samples has been demonstrated in experimentally infected horses (Table 2) (12,14–15). Therefore, the difference in the positive rate between blood and fecal samples was expected. However, the difference between laboratories was distinct and the reasons for this discrepancy are not known. Sample processing, methodological variations, reagents used, for example, may account for the differences in the rate of detection. It is reasonable to expect that in naturally occurring cases, however, both blood and fecal samples are adequate for molecular detection of PHF. It has been recommended that both blood and feces be submitted in order to increase the chances of detecting N. risticii (17). To the authors’ knowledge, studies comparing the N. risticii DNA positive rate between blood and fecal samples from clinical cases as well as the detection rate by different laboratories have not been previously reported.

Table 1.

Polymerase chain reaction results of blood and fecal samples tested for Neorickettsia risticii at 2 diagnostic laboratories.

Horse number	Blood		Feces
	Lab A	Lab B	Lab A	Lab B
1	Positive	Positive	Positive	Positive
2	Positive	Negative	Positive	Positive
3	Positive	Negative	Positive	Positive
4	Positive	Positive	Positive	Not tested
5	Positive	Positive	Positive	Positive
6	Positive	Negative	Positive	Positive
7	Positive	Negative	Negative	Negative
8	Positive	Negative	Positive	Positive
9	Negative	Negative	Positive	Positive
10	Negative	Negative	Positive	Negative
11	Negative	Negative	Positive	Positive
12	Negative	Negative	Positive	Positive

Table 2.

Dates of detection of Neorickettsia risticii by PCR in blood and fecal samples in experimentally infected horses.

Reference	Horse number	Blood		Feces
		First day positive	Last day positive	First day positive	Last day positive
(15)	1	7	28	11	28
	2	7	28	12	28
(12)	1	1	32	1	32
(17)	1	9	20	14	20
(16)	1	11	26	17	28
	2	13	25	21	25
	3	8	20	13	16
	4	11	20	14	16
	5	14	20	17	20
	6	7	13	NR	NR

NR — Not reported.

Between 2015 and 2018, Neorickettsia spp. was isolated from the blood of 16 horses in Ontario. Fourteen of these isolates were identified as N. risticii and 2 were classified as a novel Neorickettsia spp., now designated as N. findlayensis. Experimental transmission studies in ponies confirmed that N. findlayensis was capable of causing PHF (10). Phylogenetic analysis of 12/16 Neorickettsia cultured isolates from Ontario showed some regional clustering according to strain relatedness in southwestern and central Ontario (11). The 2 cases with the novel Neorickettsia spp. clustered together, separated from the N. risticii clades, and from other horses located in Ontario (11). With the discovery of this novel species, PHF should be investigated for both Neorickettsia spp. in Ontario in order to determine the potential infection rates for each species and/or co-infections, and other important disease variables such as disease severity, associated complications, and outcomes. Although culture isolation of causative pathogens is considered a gold standard of disease diagnosis, this approach can be challenging for intracellular microorganisms such as Neorickettsia spp. Cell culture is a sensitive method for diagnosing PHF, with comparable results to molecular testing (12). However, this approach requires laboratories that are capable of cell culture and that have the necessary technical expertise. Since culture isolation is labor-intensive and the organism may take weeks to grow, this approach cannot be used for clinical diagnostic purposes. A comparison between culture isolation and molecular diagnosis of clinical cases was originally planned but could not be performed due to the large number of samples which would require culture isolation.

In summary, PHF is endemic and widely distributed in southern Ontario. Laboratory confirmation of this disease may vary among veterinary diagnostic laboratories which warrants periodic assessment of diagnostic methods in place. It has been almost 100 y since a report was made of an endemic disease of horses in southwestern Ontario that had clinical, seasonal, and geographic similarities to PHF (6). A novel species, N. findlayensis, first isolated from a horse in Findlay, Ohio in 1991 was isolated in 2017 from 2 horses in eastern Ontario. As these 2 horses were culture positive for N. findlayensis, the negative PCR results for N. risticii were not unexpected. This species has been confirmed experimentally to cause clinical PHF (11). Future studies in Ontario should focus on the diagnostic techniques, epidemiology, and natural history of this organism as well as disease prevention strategies.