4.1. Regional Temporal Trends
The regional trends analysis show that canine seroprevalence for both Anaplasma spp. and B. burgdorferi increased broadly in the Northeastern and upper Midwestern states from 2013 to 2019. These changes are likely influenced by the same factors that are driving changes in the geographical distribution and density of I. scapularis. The distribution and density of this tick are known to be expanding across a variety of geographical ranges in the United States and Canada (36, 37). This is believed to be predominately driven by climate change that supports vector viability, increases in white-tailed deer population densities, and habitat change such as reforestation and fragmentation (38). In contrast, within the range of I. pacificus, little change was seen in Anaplasma spp. and B. burgdorferi seroprevalence. The only exception was northern California and southern Oregon, where Anaplasma spp. seroprevalence increased with relevance. This relative lack of change is supported by the stable population of I. pacificus, suggesting that this region is not experiencing the same change as that in the Northeast and Midwest (11).
In most areas, the trends in canine Anaplasma spp. and B. burgdorferi seroprevalence were similar, with B. burgdorferi having a greater geographical distribution. This is consistent with molecular surveys of ticks, molecular testing of Peromyscus leucopus (white-footed mouse) reservoirs, and serologic data from dogs. Several molecular studies on I. scapularis have shown that the prevalence of B. burgdorferi is generally higher than for A. phagocytophilum (39–41). Similarly, studies in P. leucopus have shown higher prevalences for B. burgdorferi compared to A. phagocytophilum; for example, prevalence in Wisconsin and Minnesota was 24 vs. 1.7% and 42 vs. 20%, respectively (42, 43). Finally, the seroprevalence of B. burgdorferi is much higher in dogs compared with Anaplasma spp. (1, 44). In areas extending out from the historically highly endemic regions, the prevalence of Anaplasma spp. in Ixodes spp. may be below the level of detection and thus impact resolution of change described by this model (45). This, in turn, may explain the disproportionately larger geographical distribution of increased B. burgdorferi seroprevalence. However, while lower prevalence of these pathogens in Ixodes spp. and important reservoirs helps explain some of the differences we observe in these trends, it would be an over-simplification to not consider the myriad of other factors that impact the ecologies of Anaplasma spp. and B. burgdorferi (46). In addition, the sensitivity of the currently used Anaplasma (SNAP®4Dx®Plus) antigen (APH-1) is lower compared with the B. burgorferi C6 peptide which may also explain some differences in prevalence due to missed cases (47–49).
One limitation of our Anaplasma spp. data is that the diagnostic test (SNAP®4Dx®Plus) used to measure antibody against Anaplasma spp. does not differentiate between A. phagocytophilum and A. platys. These two pathogens have different expected geographic distributions based on their vectors; however, there are areas that may overlap, particularly in the western United States. Rhiphicephalus sanguineus (brown dog tick) (50), the presumed vector of A. platys, has a broad geographical distribution that is facilitated in part by its endophilic (indoor dwelling) nature and affinity for domestic dog hosts (51). We are limited in our knowledge of the precise distribution and density of R. sanguineus and so the degree to which it overlaps with the ranges of I. scapularis and I. pacificus is not fully understood. Unlike the potentially broad distribution of A. platys, the primary Anaplasma sp. expected in dogs is A. phagocytophilum, which is restricted to the range of its vectors Ixodes scapularis and I. pacificus. Thus, A. phagocytophilum is believed to account for the trends observed in the northeastern and upper Midwestern regions, and some trends along the western coast.
In this study, areas of increased seroprevalence that may have been impacted by A. platys include small regions in California and Oregon and southern Texas. In California and Oregon, increased seroprevalence was observed for Anaplasma spp., but not B. burgdorferi. This may be due in part to the presence of A. platys, but we should also consider the variable diversities of Ixodes spp. (52) and their host communities (53) in this region compared to the Northeast and Midwest that may also impact the risk of exposure to dogs. In Texas, Qurollo et al. reported that dogs were equally likely to be exposed to A. platys and A. phagocytophilum (seroprevalences of 2.0 vs. 2.2%, respectively) (2). This is in contrast to the Northeast where dogs are more frequently exposed to A. phagocytophilum (1.5 vs. 13.0%). Given the differences between the vectors of the two Anaplasma spp. and the pathogens those vectors carry, future studies would benefit greatly from distinguishing the temporal trends of the two Anaplasma spp.
4.2. Local Temporal Trends
When the analysis of data shifts from regional trends to local trends, we observe useful information for veterinary practitioners concerned with changes at the county level. At this finer spatial resolution, we are able to see how seroprevalence trends can change over small distances, highlighting the importance of interpreting aggregated data with care. The local trends also appear to highlight areas in which exposure may be newly emergent. Note the trends for B. burgdorferi seroprevalence in West Virginia, Ohio, and Kentucky (Figure 6B).
In the Northeast and Middle Atlantic states, we observed a large cluster of counties with increasing local seroprevalence for both Anaplasma spp. and B. burgdorferi. However, the centers of these clusters were in different locations. The positive trends for B. burgdorferi were centered around West Virginia (Figure 6B). This cluster extended in most directions and positive counties were scattered throughout the Northeast, but notably not along the coast. Similar results for canine seroprevalence were obtained by (54) but the main clusters of counties with increasing trends were slightly further north which is likely because their canine serologic data set ended in 2017, 2 years prior to our final year of sampling (54). A recent report of temporal changes in human Lyme disease prevalence mirrors our observations to some extent, although the goals of the two studies differed. Using the first year of detection of human cases and select environmental and demographic factors in their analysis, Bisanzio et al. (55) identified West Virginia as a state with numerous new county reports from 2000 to 2017 and concluded the spread velocity of human Lyme disease estimated by their model was faster in the South. The approach Bisanzio et al. was to analyze the spread of human infections and the likelihood a county would become positive in a given year, but our approach differs in that our model provides data on trends in prevalence for both endemic and non-endemic regions (55). This explains why we had regions of increasing prevalence (e.g., in the Northeast and parts of upper Midwest) that do not show up on maps in Bisanzio et. al as being high risk (because they were already endemic). Thus, differences in the two analyses are likely attributable to four main factors; (1) modeling difference (two step diffusion model vs. a spatio-temporal binomial regression model with spatially varying trend parameters), (2) population under study (dog vs. human), (3) goals of study (predicting first case vs. identifying trends), and (4) the time range over which the study was conducted.
In our analysis, the cluster of counties with increasing Anaplasma spp. seroprevalence was centered around Pennsylvania and extended northward through Maine. Similar increasing trends from this same region were reported between 2010 and 2017 by (54). Differences in the prevalence of infection of B. burgdorferi and Anaplasma spp. in Ixodes spp. in these two regions might explain some of this difference, but the prevalence of infection of I. scapularis in West Virginia is not known at this time. There is evidence of the expanding range of Ixodes spp., with B. burgdorferi (8, 9, 45, 56), but as discussed with respect to regional trends, Anaplasma spp. may be lagging due to limitations in detection of accurate pathogen prevalence or other factors (45).
Local trends have direct application to veterinary medical decisions. The spatial difference in the temporal trends of B. burgdorferi and Anaplasma spp. pathogens may be related to differences in preventative practices. Duration of endemicity, awareness of acaricide products, socioeconomic factors, and client-based education vary substantially between endemic and non-endemic regions of the country, thereby affecting the frequency of preventative use (57). In addition, there are currently no protective vaccines on the market for Anaplasma spp. in dogs, and few available acaricides on the market currently repel ticks (58). Most acaricides rely on transfer of the drug during the tick bite to kill the tick within hours. This is protective against B. burgdorferi which requires a prolonged attachment time (59), but may not protect against pathogens, such as Anaplasma spp., that may transmit in a shorter time period (60). As a result, routine practices of preventative care in historically endemic regions may not fully protect against Anaplasma spp., and as a result, yield an increase in seroprevalence.
The evidence of increasing exposure to the tick-borne pathogens from this analysis is notable for several reasons. First, increasing seroprevalence suggests that utilization and compliance with recommended year-round use of preventative measures continues to be inadequate in some areas (57), particularly in established endemic regions and neighboring areas. These observations reinforce the concept that veterinarians and pet owners within these regions should recognize the persistent and growing risk of exposure, and implement appropriate preventative measures. Second, even in the presence of acaricides, prompt removal of ticks is strongly recommended to prevent pathogen transmission. Third, given the dynamic nature of tick-borne diseases, veterinarians practicing in regions proximate to endemic areas should adjust screening and preventative care protocols accordingly. Similarly, emphasis should be placed on vaccinating dogs at risk for Lyme disease prior to exposure (61), and the aforementioned areas of increasing seroprevalence provide veterinary practitioners with evidence-based recommendations for use of Lyme disease vaccines against emerging disease. Even in areas where the trends were not increasing, it is imperative that veterinarians and pet owners recognize that dogs are still at risk of exposure, particularly in endemic regions, and that preventative measures and testing should not be discontinued.
Supporting the evidence that dogs can act as sentinels in human vector-borne disease (15, 16), is the similarity in temporal trends between this study and incidence rates of human cases reported by the Centers for Disease Control and Prevention (CDC) and other researchers. Specifically, in regions with strong positive trends in canine Anaplasma spp. seroprevalence (i.e., Pennsylvania and northward, Figure 5), positive trends were observed during the same time period in the reported incidence-rate of human anaplasmosis (62, 63). Similar trends have been noted for human B. burgdorferi cases in Virginia and West Virginia (8, 64).
Although the focus of this study was identification of locations where the seroprevalence of B. burgdorferi and Anaplasma spp. was increasing, we noted several areas of that had decreases in seroprevalences. There was a remarkable cluster of counties that had a decrease in B. burgdorferi within the northeastern US, predominately along the Atlantic coast. This occurred for Anaplasma spp. seroprevalence to a lesser extent. The upper Midwest experienced very little increase, which was surprising given the historical endemicity. Instead, most counties experienced a stable or decrease in seroprevalence for both pathogens. However, similar results were obtained by Dewage et al. who analyzed canine serologic data collected during an earlier time period (2010–2017) (54). Finally, these changes are supported by trends that are observed in CDC-reported human cases of Lyme disease and anaplasmosis, lending further support to the use of dogs as sentinels for these, and possibly other, vector-borne pathogens (15, 16). Recently, several states within the Northeast have reported decreases in the number of human cases reported annually (65). In Wisconsin and Minnesota, trends of human incidence appeared to be stable during this study period (66). It is important to note that these are short term trends, and the long-term implications of these trends are unknown at this time. It is possible that public education and use of preventative practices in these endemic areas may be reducing the risk of exposure and thus reducing the incidence of infection (67). However, knowledge and use of these practices vary within and outside endemic areas (68, 69).
This study focused on two important pathogens associated with Ixodes scapularis, B. burgdorferi, and Anaplasma spp., which have large amounts of exposure data available through veterinary testing; however, this tick species is also a known vector for several human and zoonotic pathogens including Babesia microti, Borrelia miyamotoi, Ehrlichia muris subsp. eauclarensis, Powassan encephalitis virus (70), and the recently discovered B. mayonii (71). As we point out above, the seroprevalence of B. burgdorferi is, in part, a sentinel for the changing population of Ixodes spp. and, as a result, should compel both human and veterinary medical practitioners to be cognizant of the potential changes in incidence and spatial distribution of all pathogens carried by Ixodes spp. (15, 16). The known ranges for many of these pathogens, e.g., E. muris eauclarensis (72) and B. mayonii (73), are currently restricted to a small region, so future research is needed to determine what ecological factors drive the presence and distribution of these pathogens and whether there are any correlations with other pathogens transmitted by Ixodes spp.
Finally, there are limitations to this analysis. The temporal trends presented do not explicitly show spatial spread of B. burgdorferi or Anaplasma spp. seroprevalence over the study time period and the results here should not be interpreted as spatial change. The data are from a population of dogs under the care of a veterinarian, and so it is reasonable to assume that these dogs are more likely to be well-cared for and more likely to be provided preventative and medical care. As a result, these results may not reflect changes in higher risk populations of dogs (e.g., shelter and rescue dogs). The trends presented here reflect not only the change in the distribution and prevalence of the pathogens, but also changes in testing and preventative practices, both of which should be considered when interpreting these results. Finally, our data do not include the Canadian territories and provinces. However, there is no physical barrier between the northern US and southern Canadian border to prevent the movement of ticks and B. burgdorferi reservoir hosts. There is a growing recognition of canine Lyme disease in Canada along with increased geographic distribution and density of Ixodes scapularis, increased numbers of B. burgdorferi-infected ticks, increased human cases (74) and increased seroprevalence in dogs (75, 76). Canadian veterinarians and human healthcare providers should take the same precautions as those in the USA practicing in these transitional zones.