Veterinary Parasitology

Transcription

1 Veterinary Parasitology 160 (2009) Contents lists available at ScienceDirect Veterinary Parasitology journal homepage: Prevalence and geographic distribution of Dirofilaria immitis, Borrelia burgdorferi, Ehrlichia canis, and Anaplasma phagocytophilum in dogs in the United States: Results of a national clinic-based serologic survey Dwight Bowman a,1, Susan E. Little b,2, Leif Lorentzen c, James Shields c, Michael P. Sullivan c, Ellen P. Carlin a, * a Department of Microbiology and Immunology, Cornell College of Veterinary Medicine, C4-119 Veterinary Medical Center, Ithaca, NY 14853, United States b Department of Veterinary Pathobiology, 250 McElroy Hall, Center for Veterinary Health Sciences, Oklahoma State University, Stillwater, OK , United States c IDEXX Laboratories, One IDEXX Drive, Westbrook, ME 04092, United States ARTICLE INFO ABSTRACT Article history: Received 25 March 2008 Received in revised form 22 September 2008 Accepted 2 October 2008 Keywords: Anaplasma Anaplasmosis Antigen test Borrelia burgdorferi Canine Dirofilaria immitis Dirofilariasis Ehrlichia Ehrlichiosis ELISA Heartworm Lyme borreliosis We evaluated a comprehensive national database that documents canine infection with, or exposure to, four vector-borne disease agents, Dirofilaria immitis, Borrelia burgdorferi, Ehrlichia canis, and Anaplasma phagocytophilum in order to assess geographic trends in rates of positive tests. While the percent positive test results varied by agent in different regions of the United States, with D. immitis antigen and antibodies to E. canis more commonly identified in dogs from the South (3.9% and 1.3%, respectively), and antibody to B. burgdorferi and A. phagocytophilum found more frequently in dogs from the upper Midwest and Northeast ( % and %, respectively), evidence of at least one agent was found in dogs from every state considered. Furthermore, each organism also appeared to occur in endemic foci within larger areas of relatively low prevalence. Relocation of infected or previously exposed dogs from endemic regions likely accounts for some of the unexpected geographic distribution seen, although local transmission in previously underrecognized areas of endemicity could also be occurring. Although data were only available from the 48 contiguous states (Alaska and Hawaii were not included), taken together, our results suggest that these disease agents may be present over a wider geographic area, and thus pose greater animal and public health risks, than is currently recognized. Dogs can serve as sentinels to identify the presence of vector-borne disease agents of both veterinary and public health significance. Published by Elsevier B.V. 1. Introduction Concern over vector-borne disease in domestic dogs is evidenced by the common use of tick, mosquito, and * Corresponding author. Tel.: addresses: ddb3@cornell.edu (D. Bowman), susan.little@okstate.edu (S.E. Little), carlinvet@gmail.com (E.P. Carlin). 1 Tel.: ; fax: Tel.: ; fax: heartworm preventatives in small animal practice; just over half of all pet owners report administering parasite control products to the pets in their care (APPMA National Pet Owners Survey, ). The vector-borne canine disease agents of most common concern to the U.S. veterinary community are Dirofilaria immitis, Borrelia burgdorferi, Ehrlichia spp., and Anaplasma spp. While infection with these agents may be prevented to some extent through vector avoidance or other control measures, morbidity and mortality due to these diseases continue to occur in domestic dogs. Indeed, the use of /$ see front matter. Published by Elsevier B.V. doi: /j.vetpar

2 139 acaricides and insecticides alone is an ineffective means of breaking the enzootic transmission cycles of these pathogens. As the roles that these agents play in animal and human health have become elucidated over the last few decades, the need for further data on the natural history and prevalence of these infections has become apparent. Heartworm, a disease of canids caused by infection with the nematode D. immitis, is perhaps the most important helminthic disease of dogs in North America. Microfilariae circulate in the blood, where they may be ingested by a feeding mosquito. Development into infective third-stage larvae occurs in the malpighian tubules of the vector, and the parasites can then infect a new host (or reinfect the same host) upon subsequent feeding by the mosquito. Ultimately, the immature heartworms migrate to the dog s pulmonary arteries, and within 6 9 months begin producing microfilariae. It is this presence of the adults in the pulmonary arteries that causes right heart and/or pulmonary disease. Although dogs are the natural hosts, infection can also occur in coyotes and ferrets. Infections in cats and people often undergo truncated development, but these infections are sometimes associated with pathogenic manifestation, sometimes severe (Bowman, 2003; Theis, 2005). Lyme borreliosis is a bacterial disease caused by infection with the spirochete B. burgdorferi; in dogs, disease is commonly characterized by lameness, fever, anorexia, lethargy, and lymphadenopathy (Kahn, 2005). The most important vector in the eastern U.S. is Ixodes scapularis, commonly known as the black-legged tick or deer tick. On the West Coast, I. pacificus serves as the main vector for B. burgdorferi. The Ixodes spp. involved in B. burgdorferi transmission are three-host ticks that acquire spirochetes when feeding on rodents as larvae or nymphs, and then can transmit infection as nymphs or adults. The most important reservoir host is thought to be the whitefooted mouse (Peromyscus leucopus), although other rodents may also serve as a source of spirochetes to infect ticks (Schmidt and Ostfeld, 2001; LoGiudice et al., 2003; Brisson et al., 2008). Mean infection rates in I. scapularis nymphs in endemic areas ranges from about 20% to 40% (Daniels et al., 1998; Tsao et al., 2004; Wang et al., 2003). Infection of the nymphal western vector, I. pacificus, is lower, in the range of 0 14% (Eisen et al., 2004). Lyme disease is the most common tick-borne infection among people in North America and Europe (Wormser et al., 2006), with approximately 20,000 cases reported in the United States each year (CDC, 2007a). Cases in the southern U.S., however, are notably low (CDC, 2007a). Infection of I. scapularis nymphs with B. burgdorferi has not been documented in the South (Wormser et al., 2006), and infection rates in adult I. scapularis are much lower than those in the northeastern U.S., at % (Clark, 2004; Oliver et al., 2000). Confirmed human infection with B. burgdorferi is considered rare, if it occurs at all, in states south of Maryland and Virginia (Wormser et al., 2006; Dennis, 2005). Disease is characterized in people by an early set of skin-related and flu-like symptoms, and, in the absence of treatment, may be followed by arthritic or neurologic complications (Wormser et al., 2006; Steere et al., 2004). The rickettsial organisms Ehrlichia canis and Anaplasma phagocytophilum (the latter formerly known as E. phagocytophila and E. equi) are both tick-borne obligate intracellular bacteria with a tropism for leukocytes (Rikihisa, 1991). Disease caused by infection with these pathogens is typically characterized by fever, depression, myalgia, anorexia, and thrombocytopenia. Domestic and wild dogs are the natural hosts of E. canis, which has a worldwide distribution. The primary means of transmission is through the bite of Rhipicephalus sanguineus, the brown dog tick, although Dermacentor variabilis, the American dog tick, has also been shown to be a capable vector (Groves et al., 1975; Lewis et al., 1977; Johnson et al., 1998). The host range of A. phagocytophilum is significantly wider; rodents are considered the primary reservoir host, but infections also occur in dogs, sheep, cows, horses, and various species of wildlife (Rikihisa, 1991; Bown et al., 2003). Transmitted by I. scapularis in the Northeast and upper Midwest, and I. pacificus in the Western states, A. phagocytophilum is also responsible for human granulocytic anaplasmosis (HGA, formerly human granulocytic ehrlichiosis) (Rikihisa, 2006). Because A. phagocytophilum shares a vector and reservoir host system with B. burgdorferi, the geographic distribution of cases of HGA parallels that of Lyme borreliosis, and coinfections with the two agents may be seen (Daniels et al., 1998). Evidence of infection with or exposure to the causative agents of all four of these diseases can be tested for via a single-use, in-house diagnostic known as the SNAP 1 4Dx 1 Test (IDEXX Laboratories, Westbrook, ME). Many veterinarians are already familiar with this approach as the SNAP 1 3Dx 1 and SNAP 1 4Dx 1 test are widely used throughout the United States as an annual heartworm plus tick-borne disease screening tool. A portion of these results are captured through a central reporting system, which has assembled data from tests performed on several million dogs since Access to this comprehensive dataset provided an excellent opportunity to assess prevalence and distribution of these four organisms in well-cared for dogs throughout the United States. These results were then compared with previous assessments of geographic range and prevalence. In addition, because multiple agents were tested for simultaneously, the frequency of co-infection was evaluated. 2. Materials and methods 2.1. Source of data The SNAP 1 3Dx 1 test (IDEXX Laboratories, Westbrook, ME) is an in-clinic ELISA for simultaneous qualitative detection of canine antibodies to E. canis and B. burgdorferi, and to D. immitis antigen, in canine serum, plasma, or whole blood. In 2001 the test became available for commercial use as a replacement option for in-clinic heartworm only screening protocols. Starting in 2001, IDEXX began offering practice rebates toward the cost of the SNAP 1 3Dx 1 assay in exchange for practices submitting a log of all test results. The offer was extended to veterinary practices across the United States; therefore,

3 140 the clinic geographic distribution was randomly selected. These practices agreed to use the SNAP 1 3Dx 1 assay exclusively for all regular canine heartworm screenings, and committed to providing the screening data by compiling and submitting the data set on an IDEXXprovided data sheet. For the years evaluated, all pathogens were reported, except for the years 2002 and 2003 when the data collection forms did not include D. immitis. Clinic enrollments began in September of each year and typically continued through spring; clinics were required to have their data collection forms returned to IDEXX by the end of August each year in order to process the data and the clinic rebates. On average, clinics supplied 120 datapoints (unique test results from individual dogs); in some cases, in excess of 1500 datapoints per practice were supplied. Results were only reported a single time from each dog (no repeat test results reported). In September of 2006, IDEXX introduced the SNAP 1 4Dx 1 Test, which included the same analytes as the SNAP 1 3Dx 1, as well as a fourth analyte for the detection of A. phagocytophilum antibody. Similar to the study design described previously with the SNAP 1 3Dx 1, customers adopting the use of the SNAP 1 4Dx 1 were offered rebates in exchange for submitting test results. For the years 2006 and 2007, therefore, the data set is primarily comprised of results generated from the SNAP 1 4Dx 1 assay. Because dogs in the western U.S. are often not routinely tested by veterinarians in clinic for heartworm antigen or tick-borne disease exposure, additional data points were needed for this region. To increase the distribution and number of data points for the D. immitis national database, data were retrospectively included for the years 2001 to 2006 from the IDEXX Reference Laboratories network (IDEXX Laboratories, Inc., Westbrook, Maine) and added to the data set generated by clinics using the SNAP assays. Reference laboratory evaluations were performed on the IDEXX PetChek 1 microtiter plate that uses antibodies specific to heartworm antigen. Canine heartworm reference laboratory results were analyzed for duplicate test results in a given year, and subsequent results for an individual dog within the year were excluded from the analysis. This approach increased the number of results evaluated from dogs from western states for heartworm, but not tick-borne disease agents Heartworm assay The D. immitis analyte for both the SNAP assays as well as the microtiter plate is derived from antibodies specific to the heartworm antigen. Test sensitivity in necropsycategorized samples was 84% (175/208) and ranged from 64% to 93% in 1 and 4 worm burden infections, respectively (Atkins, 2003). Test specificity was 97% (30/31) in the same study B. burgdorferi assays The SNAP B. burgdorferi analyte detects antibodies specific to the C 6 peptide. Test sensitivity was 94.4% (238/ 252) when compared to a combination of immunofluorescence assay (IFA) and Western blot (WB) (O Connor et al., 2004). (The C 6 analyte has been shown, in both humans and canines, not to react to antibodies elicited following B. burgdorferi vaccination (O Connor et al., 2004; Marques et al., 2002).) SNAP specificity was 99.5% when used on field samples from 987 dogs in North Carolina (Duncan et al., 2004) Ehrlichia assay The SNAP Ehrlichia analyte detects antibody generated against peptides from the p30 and p30-1 proteins. The sensitivity of the analyte was 95.7% (134/140) when compared to IFA and/or WB (O Connor et al., 2002). Test specificity was 100% as compared to IFA and WB in two separate investigations (O Connor et al., 2004, 2006). It should be noted that some strains of E. chaffeensis have homologous proteins to those of E. canis and as a result, some E. chaffeensis infections will induce cross-reacting antibodies on the SNAP E. canis peptide (O Connor et al., 2004) Anaplasma assay The A. phagocytophilum analyte detects antibody generated against a synthetic peptide from the major surface protein (p44/msp2). In a subset population of samples, SNAP 1 4Dx 1 sensitivity and specificity were 99.1% and 100%, respectively, relative to the IFA (Chandrashekar et al., 2007). Preliminary studies indicate that the A. phagocytophilum analyte in the SNAP 1 4Dx 1 crossreacts with samples from Anaplasma platys infected dogs (unpublished data, source: SNAP 1 4Dx 1 Test kit insert). In areas where the Ixodes tick vector is less prevalent or absent, a positive Anaplasma result could be the result of A. platys exposure Data analyses Regional distribution and percent positive test calculations Data were collated by county of residence of each dog tested according to postal zip code provided with each record, and then assembled into state and regional groups as previously described (Blagburn et al., 1996). For ease of presentation, only four regional groups (Midwest, Northeast, Southeast, and West) were considered (Blagburn et al., 1996). Percent positive test results were calculated by dividing the number of dogs reported positive for each agent by the total number of dogs tested Population growth Population growth was analyzed using actual county level population data from the U.S. Bureau of the Census. Two sources of data were used for population and migration statistics. The 2000 Census long form asked respondents to report the county where they lived in The data are compiled in the Census County to County Migration Flow Files. Migration numbers can be measured as a percentage of 1 April 2000 county population for a relative penetration number of migration into a county.

4 141 Population growth data are from 1 April 2000 to 1 July 2006 and are compiled from the U.S. Census Population Estimates Program. This program publishes county level population estimates between censuses Statistical analyses Differences in the frequency of reported positive test results between counties, states, and regions were evaluated for significance with a t-test using SAS (Windows 9.1) (SAS Institute Inc., Cary, NC) with significance assigned at p < Results The number of practices that submitted samples totaled 2573, consisting of 5,626,926 datapoints from over 3 million dogs from all 48 contiguous states. Results of heartworm testing were available from 3,182,614 dogs: 1,039,295 from the Midwest, 707,875 from the Northeast, 515,157 from the Southeast, and 920,287 from the West. Results of B. burgdorferi and E. canis testing were available from 982,336 of these dogs: 373,090 from the Midwest, 271,070 from the Northeast, 290,636 from the Southeast, and 47,540 from the West. Results of A. phagocytophilum testing were available from 479,640 of these dogs: 175,829 from the Midwest, 188,438 from the Northeast, 101,148 from the Southeast, and 14,225 from the West. Evidence of at least one agent was found in dogs from every state considered (Table 1) Heartworm infection The highest percentage of D. immitis antigen-positive samples was obtained from the Southeast, at 3.9% (Table 1 and Fig. 1). The positive samples tended to occur in clusters of endemic foci, surrounded by areas of relatively low prevalence (Fig. 1). The prevalence map demonstrates the highest percent positive rates manifesting in individual counties, mostly in the Southeast, as well as clusters in northern California, where more than 9% of dogs tested were positive in Amador, Lake, Trinity, and Tehema County Lyme borreliosis Positive titers to B. burgdorferi were most common in the Northeast, with 11.6% of all samples from this region reported as positive. In contrast, of the samples that came from the Southeast, only 1.0% were positive, the majority of which were from Virginia. Like heartworm, the Lymepositive samples were often clustered in apparent hyperendemic foci. The prevalence map demonstrates the most prominent clusters in the coastal Northeast and upper Midwest (Fig. 2), where individual areas/counties recorded percent positive results as high as 44.1% (Putnam County, NY), 43.0% (Washburn County, WI), and 60.9% (Todd County, Minnesota) Ehrlichiosis Antibodies to Ehrlichia were detected most often in dogs in the Southeast, where 1.3% of samples were reported as positive (Table 1 and Fig. 3). All other regions were significantly below that rate and ranged from 0.3% to 0.6%. Again, endemic foci were seen where local percent positive test results were more than 10-fold greater than that for the state or region as a whole Anaplasmosis The highest prevalence of samples with antibodies to A. phagocytophilum were reported from the Midwest (6.7%); samples from the Northeast also were frequently found to be positive (5.5%; Table 1 and Fig. 4). In some individual counties, such as Washburn County, Wisconsin, Nantucket County, Massachusetts, Todd County, Minnesota, and Litchfield County, Connecticut, more than 50% of dogs were reported as positive for A. phagocytophilum. Dogs testing positive for both Anaplasma and B. burgdorferi were seen most commonly in the Midwest (2.0%), followed by the Northeast (1.4%; Table 1). In one county (Todd County, Minnesota), 40.8% of the dogs were reported to have antibodies to both agents. 4. Discussion Using a comprehensive data set we were able to determine prevalence values of four major vector-borne disease agents in dogs presenting to veterinary hospitals in the continental U.S. As expected, heartworm and Ehrlichiapositive samples were more commonly reported in the South, and B. burgdorferi- and Anaplasma-positive tests were more frequently found in the Northeast, upper Midwest, and in northern California (Table 1 and Figs. 1 4). However, evidence of current or previous infection with at least one agent was found in all regions of the U.S. and in every state considered (Table 1 and Figs. 1 4). For heartworm, it was not surprising that infection rates were highest overall in the southern states, as this area is well-known by veterinarians to be beset by clinical heartworm disease. Indeed, the map of percent positive tests for D. immitis generated from our data (Fig. 1) corresponds well with the map of reported cases compiled by the American Heartworm Society (Guerrero et al., 2006), including having the highest prevalence of infections in the coastal states of the southeastern U.S. (SC 5.7%, GA 2.7%, AL 3.4%, MS 7.4%, LA 6.0%, and TX 5.5%) and a relatively low prevalence of infection in pet dogs in Florida (1.8%), where veterinarians and clients may be particularly conscientious about administering preventatives yearround. Also, as with the AHS map, there is a similar indication that the infection is more common along the Mississippi River (AR 6.8%, MO 2.0%, TN 3.6%). In addition, both our analysis and the AHS map show that heartworm is present over a wider area than generally acknowledged; positive dogs were commonly reported from areas traditionally considered by many to be largely free of heartworm, including northern, central, and southern California (Table 1 and Fig. 1). The results from the western states, particularly California and Arizona, may be somewhat unexpected. However, D. immitis infections have been documented in these states (Bowman et al., 2007; Corselli and Platzer, 1982; Pensinger, 1986).

5 142 Table 1 Percent positive test results (number positive/number tested) in dogs by region and state for antigen of Dirofilaria immitis and antibody to Borrelia burgdorferi, Anaplasma phagocytophilum, Ehrlichia canis, and co-infection with B. burgdorferi and A. phagocytophilum. State D. immitis B. burgdorferi Ehrlichia Anaplasma Borrelia & Anaplasma co-infection Northeast CT 0.6% (236/37,650) 18.1% (1,846/10,209) 0.2% (21/10,209) 21.8% (1,499/6,887) 4.6% (317/6,887) DE 1.6% (79/4,986) 11.2% (516/4,595) 1.0% (48/4,595) 1.1% (48/4,315) 0.4% (17/4,315) MA 0.7% (1,657/252,281) 19.8% (6,729/33,915) 0.3% (107/33,915) 10.4% (2,803/26,911) 3.0% (811/26,911) MD 0.8% (221/28,770) 12.6% (2,882/22,945) 0.7% (165/22,945) 1.7% (282/16,307) 0.4% (63/16,307) ME 0.6% (173/27,247) 11.6% (3,269/28,230) 0.1% (39/28,230) 5.4% (1,341/24,632) 1.1% (271/24,632) NH 0.8% (168/21,056) 12.9% (2,343/18,122) 0.2% (36/18,122) 4.5% (618/13,743) 1.3% (173/13,743) NJ 0.3% (384/111,245) 14.2% (2,913/20,575) 0.4% (89/20,575) 9.8% (1,339/13,721) 3.2% (435/13,721) NY 0.5% (780/158,926) 7.1% (5,781/81,305) 0.2% (179/81,305) 3.6% (1,741/48,201) 0.9% (437/48,201) PA 0.4% (191/45,815) 9.4% (3,869/40,948) 0.2% (80/40,948) 1.6% (449/27,641) 0.6% (155/27,641) RI 0.8% (123/16,199) 14.3% (933/6,508) 0.1% (6/6,508) 4.7% (158/3,396) 0.7% (24/3,396) VT 0.7% (24/3,682) 9.9% (368/3,718) 0.2% (7/3,718) 1.7% (46/2,684) 0.5% (14/2,684) Regional mean 0.6% (4,036/707,857) 11.6% (31,449/271,070) 0.3% (777/271,070) 5.5% (10,324/188,438) 1.4% (2,717/188,438) Midwest IA 0.9% (164/19,097) 0.9% (149/17,390) 0.4% (61/17,390) 0.4% (21/4,840) 0.1% (5/4,840) IL 0.9% (2,915/337,434) 1.0% (324/31,976) 0.4% (135/31,976) 0.4% (51/11,899) 0.1% (6/11,899) IN 1.8% (428/24,290) 1.1% (231/20,515) 0.3% (54/20,515) 0.4% (26/7,084) 0.2% (11/7,084) KS 2.7% (170/6,264) 0.1% (6/5,473) 2.2% (119/5,473) 0.5% (7/1,452) 0.0% (0/1,452) MI 0.7% (2,031/292,171) 0.6% (431/67,625) 0.1% (34/67,625) 1.2% (190/16,312) 0.4% (68/16,312) MN 0.4% (332/80,810) 9.5% (7,267/76,610) 0.3% (202/76,610) 9.8% (6,002/61,374) 3.1% (1,928/61,374) MO 2.0% (457/22,673) 0.2% (59/24,095) 1.9% (462/24,095) 0.3% (14/5,250) 0.0% (0/5,250) ND 0.5% (25/4,914) 3.0% (136/4,558) 0.0% (1/4,558) 2.4% (40/1,692) 1.2% (21/1,692) NE 0.8% (34/4,387) 0.1% (5/4,282) 0.3% (13/4,282) OH 0.9% (1,242/136,548) 0.2% (140/61,138) 0.1% (79/61,138) 0.1% (13/14,414) 0.0% (0/14,414) SD 0.1% (1/962) 0.3% (1/358) 0.0% (0/358) - - WI 0.6% (616/109,745) 10.2% (6,018/59,070) 0.3% (194/59,070) 10.5% (5,409/51,512) 2.9% (1,510/51,512) Regional mean 0.8% (8,415/1,039,295) 4.0% (14,767/373,090) 0.4% (1,354/373,090) 6.7% (11,773/175,829) 2.0% (3,550/175,829) Southeast AL 3.4% (622/18,388) 0.1% (27/18,998) 0.3% (64/18,998) 0.1% (4/4,331) 0.0% (0/4,331) AR 6.8% (578/8,526) 0.1% (7/8,391) 3.9% (324/8,391) 0.6% (10/1,743) 0.0% (0/1,743) FL 1.8% (1,408/80,280) 0.5% (256/54,982) 0.8% (425/54,982) 0.5% (166/31,690) 0.1% (25/31,690) GA 2.7% (1,373/51,494) 0.3% (77/23,333) 1.9% (444/23,333) 0.2% (15/8,856) 0.0% (2/8,856) KY 1.1% (227/20,092) 0.2% (45/18,935) 0.8% (152/18,935) 0.1% (5/4,319) 0.0% (0/4,319) LA 6.0% (871/14,468) 0.1% (9/11,197) 0.2% (18/11,197) 0.1% (1/707) 0.0% (0/707) MS 7.4% (183/2,459) 0.0% (1/2,198) 3.1% (68/2,198) 0.0% (0/300) 0.0% (0/300) NC 3.0% (663/22,005) 1.3% (263/20,783) 2.1% (431/20,783) 0.4% (25/6,841) 0.1% (8/6,841) OK 2.1% (254/11,913) 0.2% (19/11,549) 3.8% (439/11,549) 1.2% (70/5,920) 0.1% (4/5,920) SC 5.7% (860/15,019) 1.3% (148/11,562) 0.8% (95/11,562) 0.1% (9/6,507) 0.0% (1/6,507) TN 3.6% (498/13,787) 0.2% (47/18,891) 2.3% (428/18,891) 0.1% (4/4,324) 0.0% (1/4,324) TX 5.5% (12,160/220,829) 0.2% (91/58,088) 0.8% (441/58,088) 0.6% (90/14,788) 0.1% (16/14,788) VA 1.1% (331/29,766) 6.7% (1,924/28,787) 1.8% (532/28,787) 0.9% (96/10,195) 0.3% (27/10,195) WV 0.8% (51/6,131) 0.3% (9/2,942) 0.1% (4/2,942) 0.2% (1/627) 0.0% (0/627) Regional mean 3.9% (20,079/515,157) 1.0% (2,923/290,636) 1.3% (3,865/290,636) 0.5% (496/101,148) 0.1% (84/101,148) West AZ 1.2% (620/53,809) 0.4% (4/992) 3.2% (32/992) 0.7% (4/583) 0.2% (1/583) CA 1.6% (8,478/530,788) 1.8% (540/29,454) 0.8% (225/29,454) 4.8% (612/12,673) 0.9% (112/12,673) CO 0.4% (1,028/261,358) 0.4% (49/11,557) 0.2% (19/11,557) 0.0% (0/86) 0.0% (0/86) ID 0.6% (32/5,748) 0.3% (1/369) 0.0% (0/369) 0.7% (2/298) 0.0% (0/298) MT 0.6% (16/2,801) NM 1.8% (427/23,429) 0.3% (7/2,060) 1.0% (21/2,060) 0.3% (1/289) 0.0% (0/289) NV 1.2% (74/6,180) OR 0.8% (235/29,176) 2.8% (77/2,798) 0.1% (2/2,798) 7.4% (22/296) 0.7% (2/296) UT 0.6% (11/1,904) 0.0% (0/93) 0.0% (0/93) WA 1.0% (39/4,099) 0.0% (0/33) 0.0% (0/33) WY 1.2% (10/700) 0.0% (0/184) 0.0% (0/184) Regional mean 1.2% (10,970/919,992) 1.4% (678/47,540) 0.6% (299/47,540) 4.5% (641/14,225) 0.8% (115/14,225) Overall mean 1.4% (43,500/3,182,301) 5.1% (49,817/982,336) 0.6% (6,295/982,336) 4.8% (23,234/479,640) 1.3% (6,466/479,640) Endemic transmission is known to occur and is routinely reported in the foothills of the Coast and Sierra Nevada mountain ranges (Roy et al., 1993). In addition, competent mosquito vectors are found along the lower Colorado River (Corselli and Platzer, 1982). Interestingly, it is in some of the California counties with high prevalence rates (i.e., Nevada, Placer, Riverside, and Shasta) in which autochthonous cases of human heartworm infections in California have been reported (Theis et al., 2001). Placer County (Auburn, CA) was found in a fairly recent national

6 143 Fig. 1. Evidence of antigen to Dirofilaria immitis in dogs by county, grouped according to percent positive tests. No results (<10) were received from counties shaded gray, precluding interpretation of the presence of the parasite in these areas. Counties depicted in white had no dogs reported as positive (0%). Remaining counties were coded as follows: % (taupe), % (salmon), % (red), % (brick red). Note: perception of a low heartworm prevalence in this area leads to lower levels of testing; anecdotally, the SNAP test is not used a lot in the West. Inclusion of laboratory data to create a more robust dataset may have contributed to this somewhat unexpected appearance. Fig. 2. Evidence of antibodies to Borrelia burgdorferi in dogs by county, grouped according to percent positive tests. No results (<10) were received from counties shaded gray, precluding interpretation of the presence of antibodies in dogs from these areas. Counties depicted in white had no dogs reported as positive (0%). Remaining counties were coded as follows: % (pale blue), % (eucalyptus), % (bright blue), % (slate blue).

7 144 Fig. 3. Evidence of antibodies to Ehrlichia canis in dogs by county, grouped according to percent positive tests. No results (<10) were received from counties shaded gray, precluding interpretation of the presence of antibodies in dogs from these areas. Counties depicted in white had no dogs reported as positive (0%). Remaining counties were coded as follows: % (pale lavender), % (lavender), % (magenta), % (intense purple). Fig. 4. Evidence of antibodies to Anaplasma phagocytophilum in dogs by county, grouped according to percent positive tests. No results (<10) were received from counties shaded gray, precluding interpretation of the presence of antibodies in dogs from these areas. Counties depicted in white had no dogs reported as positive (0%). Remaining counties were coded as follows: % (pale green), % (lichen), % (kelly green), % (dark moss).

8 145 survey to have the highest prevalence of heartworm antibody-positive cats (33% of 110 of 2190 total sampled cats from 18 states and the District of Columbia) (Miller et al., 2000). In general, a comparison of the map presented by Theis (2005) (human heartworm diagnosis as well as the number of canine cases presented by the AHS in 2001) with the map shown here (or that of the AHS 2005) seems highly suggestive that human infections tend to occur in areas of high canine prevalence. These include the east and southeastern coastal states, the midwestern states, and California where heartworms are highly endemic in the domestic dog and coyote population (Sacks et al., 2004). Moreover, when prevalence of infection was considered on a county level, we found a pattern of apparent endemic foci of heartworm infections within areas of relatively lower prevalence (Fig. 1). For example, in California, Trinity, Amador, Tehema, and Lake Counties, had dogs reported positive for heartworm (9.1%, 9.4%, 10.1%, and 12.5%, respectively) at rates significantly higher than for dogs in the rest of the state (1.6%, p < ) and the western region overall (1.2%, p < ). This region of California has experienced higher than average population growth in recent years. The Population Estimates Program shows an average growth of these counties at 11.3% compared to 7.6% between April, 2000 and July, Positive dogs in these counties may thus represent cases of heartworm imported into the region from areas considered more endemic. In addition, veterinarians may be more likely to test dogs that have a history of prior residence in a heartworm-endemic area. However, mosquitoes capable of transmitting D. immitis are present in these counties and throughout much of the western region (Corselli and Platzer, 1982; Walters and Lavoipierre, 1982), and it has been well documented in California that natural transmission occurs in coyotes in much of the state (Sacks et al., 2004). The continued importation of infected dogs, the capable vectors in the region, and the high levels of transmission occurring in coyotes in California underscore the need for regular testing and preventive use. With any test, there is a greater chance of a false positive when the population is almost certain to be all negative (Peregrine, 2005; Peregrine et al., 2007). Thus, in a study such as this one, where serological tests are being performed on a large number of animals in many areas with a low prevalence, the positive predictive value [PV+] of a test, i.e., the probability that a positive test result is correct, must be considered. If calculated for this study using the sensitivity and specificity for heartworm given in the materials and methods, then with a 1% prevalence (the case in much of the nation), the PV+ is 22.05%, the test is likely to be correct 1 in 5 times; at 3.9% prevalence (the numbers seen in Florida), the PV+ would be 53.25%, so the test is likely to be correct about half of the time; and at 6% (the prevalence in some of the counties in California), the PV+ is 64.12% or correct about two-thirds of the time. Therefore, it is important to retest dogs with positive results with a second test before treatment with an arsenical. Using the sensitivity (76%) and specificity (97%) calculated for the IDEXX in-house PetChek PF as calculated by Courtney and Zeng (2001) for dogs with low worm burdens of 1 10 worms (mean 2.3, median 3), the PV+ markedly increases when the results are validated using this as the second test. For a prevalence of 1% the PV+ improves to 87.67%; for a prevalence of 3.9% it improves to 96.51%, and at 6% it improves to 97.95%. Thus, when dogs initially test positive and appear clinically normal, they should be rechecked using this or a similar test for verification of infection status before treatment. Also, there may be occasions in areas of low prevalence when the clinical status of normal will take precedence over the two tests in series, because even they can be incorrect at a 1% prevalence about 15% of the time. Also, even in levels of high prevalence, it is possible that the diagnosis of an astute clinician will outweigh even the two positives obtained using these tests, as the tests are not infallible. As expected, dogs positive for antibodies to B. burgdorferi were most commonly reported from areas where Lyme borreliosis is known to be endemic or hyperendemic, including the Northeast, upper Midwest, and West Coast (Table 1 and Fig. 2). Prevalence of antibodies in dogs from the northeastern U.S. averaged 11.6%, with some hyperendemic localities identified in the Northeast and upper Midwest where rates of positive tests in dogs were greater than 40% (Fig. 2). We also found confirmatory evidence of exposure of dogs to the agent of Lyme disease in recognized endemic areas in California (Nevada, Humboldt, and Mendocino County, California: 7.7%, 9.2%, and 9.3%, respectively). However, exposure of dogs to B. burgdorferi was documented over a wider area than expected, particularly with regard to the southern U.S. The 2005 national map of reported human Lyme borreliosis cases by county is far sparser in this region (CDC, 2007a), and Lyme disease is considered rare, if it occurs at all, in states south of Maryland or Virginia (Wormser et al., 2006). Indeed, B. burgdorferi-specific antibodies were reported from only 1% of dogs from the South (Table 1), and even this low percent positive rate may be somewhat inflated by the inclusion of Virginia, a state where Lyme borreliosis is known to occur, into the southeastern region (CDC, 2007a). The positive rate for Virginia was 6.7%, whereas that for the rest of the southeastern region, omitting the dogs reported from Virginia, is only 0.36%. As with the distribution of D. immitis, dogs with antibodies to B. burgdorferi appeared to be reported from endemic foci within larger areas of relatively low endemnicity (Fig. 2). For example, in Montgomery County, Texas and Indian River County, Florida, 1.9% and 1.5% of dogs, respectively, were positive for B. burgdorferi, rates which were significantly higher than the prevalence in the rest of each state as a whole (TX prevalence 0.2%, p < ; FL prevalence 0.5%, p < ) and significantly higher than that in all southern border states combined (0.3%, p < ). As we saw with D. immitis distribution in unexpected areas, both of these counties have experienced large increases in population in recent years. The Population Estimates Program demonstrates growth of the Montgomery County population at 35.6% compared to 12.7% for all of Texas between April, 2000 and July, For the same time period Indian River County grew at 15.2% compared to 13.2% statewide. Montgomery

9 146 County, Texas had 1.5% of the population migrate from a state where B. burgdorferi is more endemic. This compares to 1% for the entire state. Indian River County, Florida shows a 7.5% migration from endemic areas compared to 5.4% statewide. These positive dogs may therefore represent movement of B. burgdorferi-infected dogs into the region from areas where transmission of this agent more commonly occurs. Our findings of dogs in the deep South with antibodies to the agent of Lyme disease (Fig. 2) are of particular interest because a laboratory-confirmed human case of autochthonus Lyme borreliosis has not been reported from any state south of Maryland or Virginia (Wormser et al., 2006). Published data are also lacking to support locally acquired cases of Lyme disease in dogs in the South (Duncan et al., 2004), although some veterinarians practicing in this region have witnessed seroconversion to C 6 in dogs without a travel history to endemic areas (Alleman, personal communication). Infection with B. burgdorferi sensu stricto has been described in both rodent reservoir hosts and tick vectors in the southern U.S. (Oliver et al., 2000), and the organism is known to be circulating in nature, but the relatively low prevalence and distinct natural history of the tick vector involved is thought to largely limit transmission to people in this region (Oliver et al., 2003). States that border areas where Lyme borreliosis is known to be endemic, such as North Carolina, do have a higher percent positive rate than those further south (Table 1). We do not have travel histories on the 1049 dogs from southeastern and southcentral states south of Maryland or Virginia that tested positive for B. burgdorferi in this study, and thus the data reported here cannot serve to confirm or deny the presence of endemic transmission of B. burgdorferi in the southern U.S. However, these data do suggest that the question of local transmission of B. burgdorferi to both dogs and people in areas historically thought to be non-endemic for Lyme borreliosis is one which warrants continued consideration. In North America, A. phagocytophilum is maintained largely in the same tick vector/reservoir host system as B. burgdorferi, and thus we anticipated the geographic distribution of these two agents would be similar in our study. As expected, antibodies to A. phagocytophilum were most commonly reported from dogs in the Midwest and Northeast, where 6.7% and 5.5% of dogs, respectively, tested positive for this agent (Table 1 and Fig. 3). In many locales, such as Crow Wing and Cass County, Minnesota, and in both the Midwest and Western regions as a whole, the prevalence of A. phagocytophilum in dogs was much greater than that of B. burgdorferi, and hyperendemic foci again appeared to be identified (Fig. 3). For example, in Lichtfield County, Connecticut, 50.2% of dogs were positive for A. phagocytophilum, a prevalence which was significantly higher than the rest of the state, the northeastern region, and the country as a whole (21.8%, 5.5% and 4.7%, respectively; p < ). Foci of infection were also identified in California; 19.4% of dogs in Humboldt County, California were reported as positive for A. phagocytophilum, a rate that is significantly greater than the 4.8% seen in the rest of the state, and the 4.5% for the entire West (p < ). Prevalence rates for A. phagocytophilum were significantly lower in the South (0.5%; p < ; Fig. 3), a region where human infections are also considered less common, although 509 dogs testing positive were reported from this area. Some dogs in the Southeast that were reported as positive to A. phagocytophilum may have relocated from an area where this agent is endemic or, alternatively, could be the result of infection with A. platys, which generates positives on the assay used. In studies involving dogs infected with a laboratory strain of A. platys, the SNAP 1 4Dx 1 was reactive with serum from 10 out of 10 infected animals (unpublished data, source: SNAP 1 4Dx 1 Test kit insert). A. platys is thought to be transmitted by R. sanguineus, the brown dog tick, and thus is considered widely distributed throughout the U.S. (Harvey, 2006). As was also expected, evidence of co-infection of dogs with both B. burgdorferi and A. phagocytophilum was seen, particularly in areas where prevalence of both organisms was high. In the Midwest, 2.0% of dogs had evidence of exposure to both agents, with co-infection rates in Todd County, Minnesota reaching a maximum observed level of 40.8%. Similarly, in the Northeast, 1.4% of dogs were coinfected with these two pathogens; prevalence of coinfection was as high as 19% in Nantucket County, Massachusetts. B. burgdorferi and A. phagocytophilum share a common tick vector and reservoir host, and I. scapularis ticks infected with both pathogens have been reported (Courtney et al., 2003; Adelson et al., 2004; Holman et al., 2004); thus, we were not surprised by this high prevalence of co-infection. In California, where the overall prevalence of B. burgdorferi is somewhat lower (1.8%), only 0.9% of dogs were reported as co-infected in the state as a whole. The disease agent for which we saw the lowest prevalence nationally and on a regional level was E. canis. Only 0.6% of dogs were positive for E. canis nationwide, with the highest regional rate, 1.3%, reported from the southeastern states. However, even with this relatively low overall prevalence, a number of foci of infection were identified where prevalence rates ranged from 2% to more than 10%. For some of these areas, such as Humphries County, Tennessee or Caddo County, Oklahoma, we suspect that cross-reactions with E. chaffeensis, as is known to occur with the analyte used in the assay (O Connor et al., 2006) accounted for some of the reported positive results. Several areas where prevalence rates to E. canis were higher than 2.0% (Fig. 4) correspond to areas of hyperendemicity of E. chaffeensis identified in other studies (Yabsley et al., 2005), suggesting that some of these dogs may also be exposed to or infected with E. chaffeensis. Both R. sanguineus and E. canis are considered more common in the South, although they are somewhat widely distributed across most of the country. We were surprised by an apparent hyperendemic focus of E. canis infection in Imperial County, California, where 11.1% of dogs were reported as positive, compared to 0.8% for the rest of the state as a whole, particularly because E. chaffeensis in not considered common in this area of the country (Foley et al., 1998; Kramer et al., 1999; Lane et al., 2001; Fritz et al., 2005). We were similarly surprised by foci of E. canis reactivity in dogs in Wisconsin and Minnesota (Fig. 4), another region where E. chaffeensis is not particularly common. There may be foci of intense E. canis transmission

10 147 in these particular areas. Alternatively, the test results in these dogs could represent exposure to a novel Ehrlichia sp. that has yet to be described; novel Ehrlichia and Anaplasma spp. continue to be discovered in North America (Brandsma et al., 1999; Loftis et al., 2006; Munderloh et al., 2007). There were several limitations to this study. Firstly, a positive antigen or antibody test is not equivalent to the existence of an agent in a particular locale; it is evidence only of prior exposure at some point and some location in a dog s history. Areas experiencing a population influx likely are also experiencing an influx of dogs from other regions of the country, and pets testing positive in these areas may well have been exposed elsewhere. Furthermore, only a distinct subset of canids was sampled that is, companion dogs brought to a veterinarian, and whose veterinarians and/or owners opted to test for the agents in question; our estimates of prevalence must be viewed with this bias in mind. In addition, while the overall study looked at 7 years of data, A. phagocytophilum results were only available for the last year, which could make the data vulnerable to any anomalies affecting that year, such as vector abundance. All four of these agents can also cause disease in humans. The three tick-borne diseases (Lyme borreliosis, ehrlichiosis, and anaplasmosis) are nationally reportable when diagnosed in people, and all have been documented in 2007 (CDC, 2007b,c,d). Heartworm infection has been demonstrated in people in North America both with and without a travel history to areas of high endemicity, and presents a clinical challenge due to the seriousness of the differential diagnoses for pulmonary nodules (Theis, 2005; Lagrotteria et al., 2003). The public health implications of the study are therefore significant. A physician s decision to diagnose a particular infection in a locale or region is partly influenced by the local prevalence of the agent. Therefore, people living in areas where a disease is not thought to be endemic may go undiagnosed. Conversely, some conditions may be misdiagnosed, followed by inappropriate treatment. The situation with Lyme disease, for instance, is complicated throughout the range of the lone star tick, Amblyomma americanum, which has been associated with the development of lesions similar to the cardinal erythema migrans rash characteristic of early Lyme infection (Dennis, 2005). Further prevalence studies are indicated, particularly those that focus on these newly identified areas of endemicity, and we foresee dogs (or a data set of dog infections) being used as sentinels. 5. Conclusion These vector-borne disease agents are more widely distributed than expected. Evidence of at least one agent was found in every state analyzed, and most states had evidence of all four organisms, representing all regions of the continental U.S. These data will provide veterinarians with a heightened awareness of the vector-borne disease agents common in their practice areas, and elevate their consideration of these infections when taking a travel history and choosing diagnostics. An important conclusion for the human population is that people living in areas where the disease was not thought to be endemic may also be at risk for infection. Dogs may serve as sentinels to identify the presence of vector-borne disease agents of both veterinary and public health significance. References Adelson, M.E., Rao, R.V., Tilton, R.C., Cabets, K., Eskow, E., Fein, L., Occi, J.L., Mordechai, E., Prevalence of Borrelia burgdorferi, Bartonella spp., Babesia microti, and Anaplasma phagocytophila in Ixodes scapularis ticks collected in Northern New Jersey. J. Clin. Microbiol. 42, Anon., Unpublished data. Source: SNAP 1 4Dx 1 Test kit insert. APPMA, American Pets Products Manufacturers Association National Pet Owners Survey, p. 18. Atkins, C.E., Comparison of results of three commercial heartworm antigen test kits in dogs with low heartworm burdens. JAVMA 222, Blagburn, B.L., Lindsay, D.S., Vaughn, J.K., Rippey, N.S., Wright, J.C., Lynn, R.C., Kelch, W.J., Ritchie, G.C., Hepler, D.I., Prevalence of canine parasites based on fecal flotation. Comp. Cont. Ed. Pract. Vet. 18, Bowman, D.D., Georgi s Parasitology for Veterinarians, 8th ed. Saunders, Philadelphia, PA, 422 pp. Bowman, D.D., Torre, C.J., Mannella, C., Survey of 11 Western states for Heartworm (Dirofilaria immitis) infection, Heartworm diagnostic protocols, and fecal examination protocols for gastrointestinal parasites. Vet. Ther. 8, Bown, K.J., Begon, M., Bennett, M., Woldehiwet, Z., Ogden, N.H., Seasonal dynamics of Anaplasma phagocytophila in a rodent-tick (Ixodes trianguliceps) sytem, United Kingdom. Emerg. Infect. Dis. 9, Brandsma, A.R., Little, S.E., Lockhart, J.M., Davidson, W.R., Stallknecht, D.E., Dawson, J.E., Novel Ehrlichia organism in white-tailed deer associated with lone star tick parasitism. J. Med. Entomol. 36, Brisson, D., Dykhuizen, D.E., Ostfeld, R.S., Conspicuous impacts of inconspicuous hosts on the Lyme disease epidemic. Proc. Biol. Sci. 2008, Centers for Disease Control and Prevention, 2007a. Lyme Disease, United States, Morb. Mortal. Wkly. Rpt. 56, (June 15). Centers for Disease Control and Prevention, 2007b. Table 1: provisional cases of infrequently reported notifiable diseases. Morb. Mortal. Wkly. Rpt. 56, 771 (August 3). Centers for Disease Control and Prevention, 2007c. Table 2: provisional cases of selected notifiable diseases, United States. Morb. Mortal. Wkly. Rpt. 56, 775 (August 3). Centers for Disease Control and Prevention, Nationally Notifiable Infectious Diseases, United States, 2007d. phs/infdis2007r.htm. Accessed August 24, Chandrashekar, R., Mainville, C., Daniluk, D., Cambell, J., Cyr, K., O Connor, T.O., Performance of an in-clinic test, SNAP 1 4Dx 1, for the detection of antibodies to canine granulocytic infection, Anaplasma phagocytophilium. In: Research Abstract of the 25th Annual ACVIM Forum, Seattle, WA, June. Clark, K., Borrelia species in host-seeking ticks and small mammals in Florida. J. Clin. Microbiol. 42, Corselli, N.J., Platzer, E.G., Canine heartworm: disease focus along the lower Colorado river. In: Proceedings of the 49th Annual Conference of the California Mosquito and Vetor Control Association, April 26 29, 1981, pp Courtney, J.W., Dryden, R.L., Montgomery, J., Schneider, B.S., Smith, G., Massung, R.F., Molecular characterization of Anaplasma phagocytophilum and Borrelia burgdorferi in Ixodes scapularis ticks from Pennsylvania. J. Clin. Microbiol. 41, Courtney, C.H., Zeng, Q.Y., Comparison of heartworm antigen test kit performance in dogs having low heartworm burdens. Vet. Parasitol. 96, Daniels, T.J., Boccia, T.M., Varde, S., Marcus, J., Le, J.H., Bucher, D.J., Falco, R.C., Schwartz, I., Geographic risk for Lyme disease and human granulocytic ehrlichiosis in Southern New York State. Appl. Environ. Microbiol. 64, Dennis, D.T., Rash decisions: Lyme disease, or not? Clin. Infect. Dis. 41, (Editorial Commentary). Duncan, A.W., Correa, M.T., Levine, J.F., Breitschwerdt, E.B., The dog as a sentinel for human infection: prevalence of Borrelia burgdorferi C 6 antibodies in dogs from southeastern and mid-atlantic states. Vector Borne Zoonotic Dis. 4,