Burnham et al. Parasites Vectors (2020) 13:129 https://doi.org/10.1186/s13071-020-3992-8 Parasites & Vectors RESEARCH Open Access Exploring the relationship between susceptibility to canine leishmaniosis and anti‑Phlebotomus perniciosus saliva antibodies in Ibizan hounds and dogs of other breeds in Mallorca, Spain Alexis C. Burnham1, Laura Ordeix1,2, M. Magdalena Alcover3, Pamela Martínez‑Orellana1, Sara Montserrat‑Sangrà1, Laura Willen4, Tatiana Spitzova4, Petr Volf4 and Laia Solano‑Gallego1* Abstract Background: Canine leishmaniosis caused by Leishmania infantum is a neglected zoonosis transmitted by sand flies like Phlebotomus perniciosus. Clinical signs and disease susceptibility vary according to various factors, including host immune response and breed. In particular, Ibizan hounds appear more resistant. This immunocompetence could be attributed to a more frequent exposure to uninfected sand flies, eliciting a stronger anti-sand fly saliva antibody response. Methods: This study aimed to investigate the prevalence of anti-P. perniciosus saliva antibodies in Ibizan hounds and dogs of other breeds in the Leishmania-endemic area of Mallorca, Spain, and to correlate these antibody levels with clinical, immunological and parasitological parameters. Anti-sand fly saliva IgG was examined in 47 Ibizan hounds and 45 dogs of other breeds using three methods: P. perniciosus whole salivary gland homogenate (SGH) ELISA; recombinant protein rSP03B ELISA; and rSP03B rapid tests (RT). Additionally, diagnostic performance was evaluated between methods. Results: Results indicate significantly higher anti-SGH antibodies (P = 0.0061) and a trend for more positive SGH ELISA and RT results in Ibizan hounds compared to other breeds. General linear model analysis also found breed to be a significant fac‑ tor in SGH ELISA units and a marginally significant factor in RT result. Although infection rates were similar between groups, Ibizan hounds included significantly more IFN-γ producers (P = 0.0122) and papular dermatitis cases (P < 0.0001). Older age and L. infantum seropositivity were also considered significant factors in sand fly saliva antibody levels according to at least one test. Fair agreement was found between all three tests, with the highest value between SGH and rSP03B RT. Conclusions: To our knowledge, this is the first study elaborating the relationship between anti-P. perniciosus saliva anti‑ bodies and extensive clinical data in dogs in an endemic area. Our results suggest that Ibizan hounds experience a higher frequency of exposure to sand flies and have a stronger cellular immune response to L. infantum infection than other breed dogs. Additional sampling is needed to confirm results, but anti-P. perniciosus saliva antibodies appear to negatively correlate with susceptibility to L. infantum infection and could possibly contribute to the resistance observed in Ibizan hounds. Keywords: Anti-sand fly saliva antibodies, Canine leishmaniosis, Leishmania infantum, Phlebotomus perniciosus, Ibizan hounds, Papular dermatitis, rSP03B *Correspondence: laia.solano@uab.cat 1 Departament de Medicina i Cirurgia Animals, Facultat de Veterinària, Universitat Autònoma de Barcelona, Bellaterra, Spain Full list of author information is available at the end of the article © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​ mmons​.org/licen​ses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creat​iveco​mmons​.org/publi​cdoma​in/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Burnham et al. Parasites Vectors (2020) 13:129 Page 2 of 15 Background a low or absent response more commonly occurs in Canine leishmaniosis (CanL) caused by Leishmania severe cases of CanL [14, 15]. Another method consists infantum is a neglected zoonotic disease endemic in 88 of monitoring inflammatory cytokine IFN-γ expression countries around the world, including many parts of the in sera or peripheral blood stimulated with L. infantum Mediterranean basin [1]. The disease is transmitted by soluble antigen (LSA). Lack of IFN-γ production has female phlebotomine sand flies, which introduce promas- been reported in sick dogs [16], markedly so in dogs tigotes into the host along with various salivary proteins with severe disease [17, 18], whereas greater levels of that facilitate infection and induce an immunomodula- expression have been observed in dogs with subclini- tory effect [2]. cal infections or with mild to moderate disease [19, Clinical presentation of CanL is highly variable, mani- 20]. Diagnosis of papular dermatitis also could be con- festing as a subclinical infection, a self-limiting condition, sidered an indicator, as the relatively benign condition or a debilitating, ultimately fatal disease [3]. In endemic has been associated with a marked protective cellular areas, a high proportion of dogs is resistant or subclini- immune response and a dampened humoral immune cally infected with Leishmania, with the minority devel- response [5, 6]. oping serious disease [4]. In particular, a distinctive form Susceptibility to canine leishmaniosis is known to vary of papular dermatitis has been described in infected dogs depending on a number of host-related factors, includ- living in endemic areas, characterized by a lack of other ing dog breed [21]. Certain dog breeds, such as the boxer, clinical abnormalities, control of parasite dissemina- rottweiler, cocker spaniel and German shepherd, have tion, and favorable disease prognosis [5, 6]. However, all been reported to display a more pronounced suscepti- infected dogs, even subclinical dogs, have the potential bility to the disease due to the influence of numerous to serve as a source of infection to sand flies and further loci [22–25]. In contrast, the Ibizan hound is consid- transmission [7, 8]. ered more resistant to L. infantum infection than dogs of The severity, extent, and initial onset of clinical mani- other breeds [26]. It has been proposed that the breed’s festation is inextricably linked to the host’s immune relative immunocompetence is derived from a predomi- response. Several studies have demonstrated that pro- nantly cellular, parasite-specific immune response, which tective immunity is associated with a T-cell mediated is associated with a clinically healthy status and favora- response, whereas disease susceptibility is associated ble prognosis [26, 27]. Compared to other breeds living with a primarily humoral immune response and reduced in the same area, Ibizan hounds display higher frequency cell mediated immunity [9, 10]. Although a definite T of positive LST reactions, greater IFN-γ production, and helper cell 1 (Th1) or T helper cell 2 (Th2) dichotomy was weaker or lack of humoral immune response [26, 28]. observed in the early studies of leishmaniosis in mice, However, a precise etiology for this apparent resistance the immune response to L. infantum infection in dogs has not yet been identified. has proven to be far more complex and more similar This study proposes that this distinction in clinical to the reactions observed in humans [11]. On a general development and comparable immunological compe- level, resistant dogs mount a mixed CD4+ T-cell medi- tence in the Ibizan hound can be attributed to a greater ated response that favors Th1 activity, producing the degree of sand fly exposure. A major defining feature of pro-inflammatory cytokines interferon gamma (IFN- the breed is its large upright ears, a relatively hairless γ), interleukin 2 (IL-2), and tumor necrosis factor alpha area especially vulnerable to sand fly bites. Exposure to (TNF-α) [12]. These cytokines induce activation of mac- sand fly bites or immunization with sand fly saliva or rophages, which can then kill intracellular amastigotes its derivatives has been shown to elicit a strong, specific with nitric oxide (NO), the process of which is considered anti-saliva antibody response in dogs [29, 30]. In par- the main effector mechanism in the protective response ticular, a positive correlation has been observed between [11]. Meanwhile, antibodies are incapable of neutralizing anti-sand fly saliva IgG, IgG1 and IgG2 and the number the invading parasites, thus dogs with a strong humoral of sand fly bites for two primary L. infantum vectors, the response are unable to control infection [11]. New World Lutzomyia longipalpis and the Old World The type of immune response and associated prog- Phlebotomus perniciosus; this correlation indicates that nosis can be characterized using a multitude of bio- this anti-saliva antibody response serves as a marker of markers (reviewed in [13]). A common method is the sand fly exposure [31, 32]. Moreover, recombinant sali- leishmanin skin test (LST), which involves the injection vary proteins, such as the yellow-related protein rSP03B of Leishmania antigen to elicit a delayed-type hyper- of P. perniciosus, have been shown to be valid alterna- sensitivity (DTH) reaction. A positive reaction tends to tives to traditional salivary gland homogenate (SGH) as be associated with non-existent or mild disease and is a measure of this degree of exposure [33, 34]. Although considered an indicator of favorable prognosis, whereas some studies have shown that anti-P. perniciosus saliva Burnham et al. Parasites Vectors (2020) 13:129 Page 3 of 15 antibodies correlate significantly with active canine L. (plaques) with or without an ulcerative or crusted central infantum infection, as determined by seropositivity for part (volcano sign) were considered suggestive of papular anti-Leishmania IgG [32, 33], evidence suggests that dermatitis due to leishmaniosis [5, 6]. In addition, a sam- exposure to sand fly saliva may be associated with para- ple was collected for cytological and molecular examina- site T-cell mediated immunity and favorable clinical out- tion if the size of the lesion was sufficient, with qPCR for come [35]. Thus, we hypothesize that the Ibizan hound’s L. infantum applied if presence of Leishmania amastig- inherently higher frequency of sand fly exposure due to otes was unclear [36]. its unique physical attributes could be tied to its later Additional diagnostic procedures previously performed immunocompetence during L. infantum infection. and described elsewhere consisted of in-house endpoint Therefore, the objectives of this study were to investi- indirect enzyme-linked immunosorbent assay (ELISA) gate the prevalence of anti-P. perniciosus saliva antibod- for detection of anti-L. infantum antibodies (IgG) [37], ies in Ibizan hounds and dogs of other breeds living in LST [26], L. infantum quantitative PCR (qPCR) of DNA a highly endemic area of leishmaniosis and to correlate extracted from peripheral blood [37], and IFN-γ assay these anti-saliva antibody levels with clinical, immuno- using whole peripheral blood stimulated with LSA [17]. logical and parasitological parameters. In particular, the Not all diagnostic tests were performed for all dogs due relationship between papular dermatitis suggestive of L. to lack of sample and/or lack of clinical justification. infantum infection and anti-saliva antibodies was exam- Residual blood and serum samples were used in this ined. To our knowledge, this is the first study attempting study, thus ethical approval was not needed. to elaborate the relationship between anti-P. perniciosus Fifty-seven control residual serum samples were saliva antibodies and extensive clinical data in dogs resid- obtained from the Royal Veterinary College (London, ing in an endemic area, aside from previous studies con- UK) from a random set of dogs presenting at the Queen sidering primarily serological status for L. infantum. Mother Veterinary Teaching Hospital. These dogs were assumed to be unexposed to sand flies because they were Methods obtained from a non-endemic region for CanL. No clini- Experimental design and clinical data cal information or travel history was available for the UK A cross-sectional study was conducted in December dur- dogs. ing three consecutive years (2014–2016) on the island of Mallorca, Spain. From five different herds of hunting Indirect enzyme‑linked immunosorbent assay (ELISA) dogs, 47 samples were collected from Ibizan hounds and Sera samples were analyzed by indirect ELISA measur- 45 from dogs of other breeds, all of which were living in ing anti-P. perniciosus IgG by using P. perniciosus salivary outdoor kennels (including sleeping outdoors) in inland gland homogenate (SGH) and the recombinant pro- villages under similar environmental conditions (n = 92 tein rSP03B as the antigen. ELISAs were performed in dogs). The majority of samples (71 dogs) were obtained accordance with a previous study [38]. Absorbance (OD) in 2015, whereas a small portion of the Ibizan hound was measured at 492 nm by using an automatic reader samples were taken in 2014 and 2016 (13 and 7 dogs, (Heales MB-580, Shenzhen, China). All sera samples respectively). Basic clinical characteristics were recorded, were tested in duplicate. Due to limited rSP03B availabil- including breed, sex and age. Dogs did not receive any ity, only the first 40 of the 57 UK control samples were form of treatment for leishmaniosis in the years preced- used for the rSP03B ELISA and RT, whereas all 57 UK ing the study, with the exception of a single non-Ibizan samples were subjected to the SGH ELISA. hound dog, who had been treated with allopurinol 2.5 As plates were analyzed on different days, positive con- years prior to sample collection. Ibizan hounds did not trol sera (PC, n = 2) and negative control sera (NC, n = 1) receive any insecticide treatment, but a few dogs of other were included on each plate. A conjugate control (CC), or breeds were reported to have received some topical indi- blank, containing no sera was included in each plate. To vidual insecticide treatment (permethrin spot-on or col- control for inter-plate variability, ELISA units (EU) were lar), albeit with inconsistent application and low owner calculated as the standardized optical density (SOD) as compliance with recommended guidelines. Dogs did follows: not have any history of travel outside Mallorca. All dogs Average ODSample − Average ODCC were subjected to a complete physical examination, with SOD (%) = × 100 all clinical signs recorded. In addition, specific dermato- Average ODPC − Average ODCC logical examination was performed. Raised, solid cutane- Cut-off SOD values were established with the package ous lesions in non-haired skin with a width less than one OptimalCutpoints [39] for Receiver Operating Charac- centimeter (papules) or with a greater width than height teristic (ROC) curve analysis in R software [40]. The ROC Burnham et al. Parasites Vectors (2020) 13:129 Page 4 of 15 curves were generated and compared by two common Statistical analyses methods, the Youden index [41] and “SpEqualSe”, the Statistical tests were performed with a combination of latter of which minimizes the absolute value of the dif- Minitab 17 and R software [40, 45]. Diagnostic perfor- ference between sensitivity and specificity [42–44]. The mance was evaluated by comparing degree of agreement “true” disease status of samples included in the calcula- between the three salivary antibody tests performed. tion of the ROC curve was determined by using the aver- McNemar’s test and Cohen’s kappa were calculated age SOD of the negative control UK sera samples for the for each pair of diagnostic tests, the latter of which was SHG ELISA, as this test was previously demonstrated to assessed according to the original author’s specifica- have greater sensitivity and specificity than either test tions [46]. Clinical parameters were analyzed between employing the recombinant rSP03B [38]. Samples with the two sample groups (Ibizan hounds and dogs of other an OD above three standard deviations of the mean SGH breeds) using a Mann–Whitney U-test or Fisher’s exact ELISA SOD for the UK samples were considered highly test as appropriate. Due to the small size of the data set, likely to have been exposed to sand flies and included to a normal approximation for the Fisher’s exact tests was approximate “true positive” samples in the calibration. not used in most cases. Any differences in EU for SGH All UK samples were classified as “true negative” sam- or rSP03B between groups based on clinical param- ples, except for two samples exceeding three standard eters were determined using two-sided Mann–Whitney deviations (SD) of the mean. Due to lack of clinical infor- U-tests. A one-sided Fisher’s exact test was then used mation and travel history, these two dogs were excluded to examine the significance of the trend established from the calibration. The remaining sera samples from when calculating the proportion of positive ELISA or RT Mallorcan dogs that were within three SD of the mean results for each variable. Spearman’s correlation was used UK SOD were considered “indeterminate status” and to determine any association between SGH and rSP03B thus excluded from the calculation for calibrating the EU and quantitative clinical variables. A P-value of less ROC curve. Finally, a single recommended cut-off value than 0.05 was considered significant. of 38.05 EU was established by both methods for SGH. A multivariate general linear model (GLM) was con- The analyses suggested three different cut-off values with structed for each salivary antigen test using R software, identical area under the curve (AUC) for rSP03B ELISA; including only the quantitative and qualitative variables from these values, the middle value balancing sensitivity with a P-value of less than 0.2 based on univariate anal- and specificity was chosen (33.67 EU). ysis. Five models were created in total, one each to pre- dict the following response variables: SGH ELISA result; Rapid test with rSP03B sero‑strip SGH EU; rSP03B ELISA result; rSP03B EU; and rSP03B Sero-strip tests were completed as described elsewhere RT result. All models predicting test result were con- [38]. Tests were conducted with all serum samples from structed using binomial logistic models, to predict the Mallorca, as well as with the 34 of 57 UK samples that probability being classified as positive for that particular were included in both the SGH and rSP03B ELISA analy- test (i.e. SGH ELISA result, rSP03B ELISA result and RT sis. Results were visually examined and photographed result). SGH and rSP03B EU were predicted using Gauss- after precisely 15 min of allowing the migration buffer to ian linear regression models. All EU (i.e. for the SGH, run. For comparison, a positive control serum was tested rSP03B, IFN-γ, and L. infantum ELISA) were calculated on each day that rapid tests were conducted. as the SOD using the formula previously provided and Results from the rSP03B sero-strip test were catego- ­log10-transformed before GLM analysis. To examine our rized according to the intensity of the observed band primary hypothesis, breed category (Ibizan hound or and assigned one of five values: 0 for a lack of band and other breed) was included as a variable in the final mul- negative result; 1, for a very faint band of indeterminate tivariate models irrespective of its significance, or lack nature; 2, for a weak but positive band; 3, for a very clear thereof, in the previous univariate models. When con- positive band of a weaker intensity than the control band; sidering variables with a high degree of multicollinearity, and 4, for a strong positive band comparable in intensity such as L. infantum seropositivity status and L. infantum to the control band [38]. Only samples receiving a score EU, the variable with the lowest P-value in the univari- above 1, rather than at least 1, were considered posi- ate analysis was retained in the multivariate model. Mul- tive, a modification from the original protocol suggested tivariate models were constructed in a stepwise fashion, to improve performance of the RT in the field [38]. An beginning with a full model and removing variables one- example spectrum of results is provided as part of the by-one. Multiple models for the same response variable additional files (Additional file 1: Figure S1). were compared using the Akaike’s information criterion (AIC). Goodness-of-fit was assessed by deviance of the residuals, ruling out overdispersion. After simplifying Burnham et al. Parasites Vectors (2020) 13:129 Page 5 of 15 the models, interactions between selected variables were difference was detected in the proportion of LST-posi- considered. Non-significant interactions were included tive dogs, Ibizan hounds displayed significantly greater so long as they improved model fit. Final models were LST lesion diameter and skin thickness (medians 2.1 compared using a likelihood ratio test. mm and 5.48 mm, respectively) in comparison with dogs of other breeds (medians 1.0 mm and 3.45 mm, respec- Results tively; Mann–Whitney U-test: U = 885.5 P = 0.0002 and Description of clinical data for Ibizan hounds and dogs U = 946.5, P = 0.0001, respectively). A significant dif- of other breeds ference between the two groups was observed in the Dogs proportion of dogs classified as IFN-γ producers (Fish- Samples were collected from 47 Ibizan hounds and 45 er’s exact test: P = 0.0122), with a greater proportion dogs of other breeds (n = 92), all sexually intact. The observed in Ibizan hounds (35/39 or 89.74%) than other non-Ibizan hound group included 30 purebred dogs, breeds (16/26 or 61.54%). Likewise, IFN-γ levels were sig- of which the most frequent breeds were English set- nificantly higher in Ibizan hounds (median 1115.60 EU) ter (n = 12), Brittany spaniel (n = 10), Weimaraner than in other breeds (median: 51 EU; Mann–Whitney (n = 2), and shorthaired pointer (n = 2). Fifteen mixed U-test: U = 1506, P = 0.0017). No significant difference in breed dogs were also represented, derived from crosses infection rate was found between groups, as determined between other hunting or working breeds. The median by the proportion of dogs testing positive for any of the age of the entire sample set was 24 months (range 4–132 four tests listed in Table 1. months; 7 dogs without age recorded), with 62.35% of the dogs above 18 months of age. A significant difference Diagnosis of papular dermatitis in age was detected between the two groups: the Ibizan Single or multiple papules were located in the inner hound group was characterized by a significantly lower aspect of one or both pinna (Additional file 2: Figure median age and lower proportion of adult dogs (median: S2). In the majority of dogs, these papules were centrally 12 months; 18/40 or 45% adults) in comparison with ulcerated and covered by an adherent crust. No other the group of other breeds (median: 42 months; 35/45 cutaneous lesions were observed, aside from mild crust- or 77.78% adults; Mann–Whitney U-test: U = 1432, ing dermatitis suggestive of fly dermatitis in the pinnae P = 0.0113; Fisher’s exact test: P = 0.0033). The propor- of two Ibizan hounds. Significantly more Ibizan hounds tion of dogs at least 4 years-old also was statistically dif- (30/47 or 63.83%) than dogs of other breeds (1/45 or ferent between Ibizan hounds (9/40 or 22.5%) and other 2.22%) presented with papular dermatitis suggestive of breeds (20/45 or 44.44%; Fisher’s exact test: P = 0.0408). L. infantum infection (Fisher’s exact test, P < 0.0001). A Both Ibizan hounds (12/47 or 74.47%) and other breeds single non-Ibizan hound presented with clinical signs (20/45 or 55.56%) consisted of a greater proportion of consistent with diagnosis of papular dermatitis, but later females, but the difference in the proportion of females cytological examination of the papule suggested that it between the two groups was not statistically significant could be due to an eosinophilic arthropod bite reaction, (Fisher’s exact test: P = 0.0797). as no amastigotes were detected and it did not display the typical “volcanic” appearance (Additional file 2: Figure Leishmania infantum infection status S2). This dog was confirmed as infected by low positive Table 1 summarizes the proportion of positive results for the four primary diagnostic tests for L. infantum infec- tion between breed groups. No difference was detected Table 1 Proportion of dogs testing positive for the four main in L. infantum antibody levels (EU) or in the propor- diagnostic tests for confirming L. infantum infection tion testing positive when comparing Ibizan hounds and Positive diagnostic test Proportion of Proportion of P-value other breeds. Moreover, the overwhelming majority of Ibizan hounds other breeds (n = 47) (n = 45) positive results were considered “low positive” (EU > 100 and < 500), except for three “medium positives” (EU > 500 L. infantum ELISA 20% (9/45) 22.22% (10/45) 1 and < 9000) observed in non-Ibizan hound dogs. Based L. infantum blood qPCR 35.29% (12/34) 51.22% (21/41) 0.2427 on L. infantum qPCR from blood, the proportion of Ibi- LST 93.55% (29/31) 76.19% (16/21) 0.1038 zan hounds classified as positive (35.29%) was similar to IFN-γ assay 89.74% (35/39) 61.54% (16/26) 0.0122* that observed in dogs of other breeds (51.22%). How- Any of four tests 91.11% (41/45) 75.56% (34/45) 0.0874 ever, blood parasite load was significantly greater in Note: Two-tailed Fisher’s exact test dogs of other breeds (median 0.33 parasites/ml) than in *P < 0.05 Ibizan hounds (median 0 parasites/ml; Mann–Whitney Abbreviations: ELISA, enzyme-linked immunosorbent assay; IFN-γ, interferon U-test: U = 1099, P = 0.0240). Although no significant gamma; LST, leishmanin skin test; qPCR, quantitative polymerase chain reaction Burnham et al. Parasites Vectors (2020) 13:129 Page 6 of 15 L. infantum serology and a positive reaction to LST, (75.76%, κ = 0.3846 ± 0.0840). However, a statistically sig- although it received a negative result for the IFN-γ assay nificant difference was found in the proportions of posi- and the L. infantum qPCR in blood. Subsequent statisti- tive responses between the SGH ELISA (23/92 or 25%) cal analyses examining the anti-saliva antibody response and rSP03B ELISA (33/92 or 35.87%; McNemar’s test: between papular and non-papular cases included this P = 0.0007) and between rSP03B ELISA and RT (23/92 or dog among the dogs with papular dermatitis, except 25%; McNemar’s test: P = 0.0012). Interestingly, a mod- where noted. erate but significant Spearman’s correlation of 0.4759 Of the 17 Ibizan hounds without papules, 15 were (df = 90, P < 0.0001) was obtained when evaluating the EU considered positive for L. infantum infection. Etiology for the two ELISA tests, as depicted in Table 3. of papular lesions was confirmed in 18 Ibizan hounds (definitive diagnosis) by means of visualization of amas- Relationship between P. perniciosus salivary antibodies tigotes on skin samples (n = 9) or by molecular detection and clinical parameters in stained smears (n = 9). As determined by these meth- Table 3 describes the correlation between individual ods, the proportion of dogs with a definitive diagnosis of clinical quantitative variables and SGH and rSP03B EU. papular dermatitis due to Leishmania infection was sta- SGH ELISA, rSP03B ELISA and rSP03B RT outcomes tistically significant between breed groups (Fisher’s exact (EU and/or results) are summarized by individual clinical test: P < 0.0001). parameters in Tables 4, 5 and 6, respectively. There was slight agreement between papular dermati- tis diagnosis and IFN-γ assay or LST (κ = 0.099 and 0.055, Breed respectively). However, when considering only dogs less Median SGH EU appeared higher for Ibizan hounds than 18 months-old, the degree of agreement between (28.94) than for dogs of other breeds (23.07), and further papular dermatitis diagnosis and IFN-γ assay or LST was analysis found the difference between breed groups to be moderate (κ = 0.479 and 0.42, respectively). significant (Mann–Whitney U-test: U = 2537, P = 0.0061). Evaluation of the diagnostic performance and agreement between SHG ELISA, rSP03B ELISA, and rSP03B rapid sero‑strip test Pairwise analysis was used to examine agreement between the three salivary antigen-based tests, the results of which are depicted in Table 2. Figures 1 and 2 depict the dis- tribution of SGH and rSP03B ELISA EU and of RT cat- egories, respectively, based on breed group. Correlation between SGH and rSP03B EU is shown in Table 3. Addi- tional data on test score differences between breed groups are provided in Tables 4, 5 and 6 (SGH ELISA, rSP03B ELISA and RT, respectively). A relatively high percent- age of agreement (range 73.48–81.06%) and fair degree of agreement (κ range: 0.3312–0.3846) was observed between each of the three pairs of tests. Highest percent- age of agreement was observed between SGH ELISA and Fig. 1 Distribution of SGH ELISA (a) and rSP03B ELISA (b) EU based RT (81.06%, κ = 0.3736 ± 0.1014), while highest degree of on breed. Respective cut-offs are indicated by a dotted line agreement was reported between SGH and rSP03B ELISA Table 2 Comparison of agreement between salivary antibody tests Test pair Percent agreement McNemar’s exact test κ ± SE κ ­interpretationa P-value SGH ELISA vs rSP03B ELISA 75.76% 0.0007*** 0.3846 ± 0.0840 Fair SGH ELISA vs rSP03B RT 81.06% 0.8450 0.3736 ± 0.1014 Fair rSP03B ELISA vs rSP03B RT 73.48% 0.0012** 0.3312 ± 0.0855 Fair a The interpretation for each κ value is shown in the final column according to the following scale: ≤ 0, no agreement; 0.01–0.20, none to slight; 0.21–0.40, fair; 0.41– 0.60, moderate; 0.61–0.80, substantial; and 0.81–1.00, almost perfect agreement **P < 0.01, ***P < 0.001 Abbreviations: ELISA, enzyme-linked immunosorbent assay; κ, Cohen’s kappa value; RT, rapid test; SE, standard error; SGH, salivary gland homogenate Burnham et al. Parasites Vectors (2020) 13:129 Page 7 of 15 observed in adult dogs (17/53 or 32.08%) than in juvenile dogs (3/32 or 9.38%; Fisher’s exact test: P = 0.0140). Leishmania infantum serological status Dogs seropositive for L. infantum were significantly more likely to be classified as seropositive for anti-SGH antibodies (Fisher’s exact test: P = 0.0470). A statistically significant difference in SP03B EU was found between L. infantum-positive dogs (median 34.10) and negative dogs (median 25.63; Mann–Whitney U-test: U = 3007, P = 0.0275). Additionally, a moderately strong, posi- tive correlation was described between L. infantum EU and SGH or rSP03B EU (Spearman’s correlation, SGH: rs   =  0.3585, df = 86, P = 0.0005; rSP03B: rs =  0.3888, df = 86, P = 0.0002). Multivariate general linear models (GLMs) Final GLM results are summarized in Table 7. After removing variables with a high degree of multicollinear- ity and simplifying the model to maximize the fit, only Fig. 2 Distribution of RT score categories based on breed. Only two additional variables were considered significant in scores above one are considered positive. The percentage of dogs any of the models: L. infantum EU and age in months. corresponding to each category is listed below the appropriate dot Additionally, an interaction between breed group and age (Age × Breed) was considered to account for the sta- tistically significant difference in age in Ibizan hounds Age and dogs of other breeds. Although this interaction was A significant positive correlation was reported between not considered significant in any model, it was retained age and SGH EU or rSP03B EU (Spearman’s correlation, in four of five models because it improved fit. RT was SGH: rs = 0.2360, df = 83, P = 0.0296; rSP03B: rs = 0.5212, the only test for which model fit was not improved by df  = 83 P < 0.0001). Adult dog rSP03B EU was considered including this interaction. After adjustment for the age to be significantly different from juvenile dog EU (Mann– difference between breed groups (Age × Breed), breed Whitney U-test: U = 2700, P = 0.0001). This trend cor- was considered a significant factor in predicting SGH related to a significantly higher proportion of positive EU (P = 0.0026), with Ibizan hounds more likely to have results in adults (26/53 or 45.23%) than in juvenile dogs higher EU. Breed did not quite reach significance in pre- (4/32 or 12.5%; Fisher’s exact test: P = 0.0019). Likewise, a dicting RT result (P = 0.0745). Leishmania infantum EU significantly higher proportion of positive RT results was was considered a significant variable in all of the models Table 3 Determination of the relationship between salivary antigen EU and other quantitative clinical parameters using Spearman’s correlation Relationship SGH ELISA rSP03B ELISA rs P-value rs P-value Salivary antigen EU and LST diameter (mm) 0.0625 0.6729 − 0.1177 0.4258 Salivary antigen EU and LST thickness (mm) 0.0958 0.5081 − 0.1669 0.2468 Salivary antigen EU and L. infantum EU 0.3585 0.0005*** 0.3888 0.0002*** Salivary antigen EU and IFN-γ antigen EU 0.2331 0.0616 0.0644 0.6104 Salivary antigen EU and blood parasite load (parasites/ml) 0.0974 0.4056 0.1876 0.1070 Salivary antigen EU and age (months) 0.2360 0.0296* 0.5212 < 0.0001*** SGH and rSP03B EU 0.4759 < 0.0001*** Abbreviations: ELISA, enzyme-linked immunosorbent assay; EU, ELISA units; IFN-γ, interferon gamma; LST, leishmanin skin test; rs, Spearman’s rho coefficient; SGH, salivary gland homogenate *P < 0.05, ***P < 0.001 Burnham et al. Parasites Vectors (2020) 13:129 Page 8 of 15 Table 4 Comparison of SGH EU and proportion of positive results between dogs based on clinical parameters Variable n SGH EU SGH ELISA result Median ELISA units (IQR) Mann–Whitney Proportion positive Fisher’s exact test P-value (count) P-value Breed Ibizan hounds 47 28.94 (19.76) 0.0061** 0.2979 (14) 0.1999 Other breeds 45 23.07 (19.76) 0.200 (9) Age Adult (≥ 18 months) 53 26.13 (24.81) 0.255 0.2830 (16) 0.1426 Juvenile (< 18months) 32 23.32 (17.06) 0.1563 (5) Sex Female 60 25.48 (16.52) 0.5633 0.2333 (14) 0.3957 Male 32 26.54 (35.44) 0.2813 (9) L. infantum ELISA Seropositive 19 28.94 (35.05) 0.1953 0.4211 (8) 0.0470* Seronegative 71 25.25 (16.56) 0.1972 (14) L. infantum blood qPCR Positive 33 27.62 (35.41) 0.3503 0.3030 (10) 0.1310 Negative 42 24.18 (13.93) 0.1667 (7) LST Positive 45 24.40 (13.39) 0.1715 0.1778 (8) 0.2864 Negative 7 23.69 (13.24) 0 (0) IFN-γ assay Producer 51 25.25 (20.76) 0.1088 0.2590 (13) 0.3116 Non-producer 14 19.34 (16.36) 0.1429 (2) Papule presence Papular 31 26.13 (26.35) 0.2980 0.2581 (8) 0.5450 Non-papular 61 25.70 (21.56) 0.2459 (15) L. infantum ­infectiona Yes 75 27.29 (26.70) 0.0832 0.2800 (21) 0.0689 No 15 23.69 (14.93) 0.0667 (1) a Presence of L. infantum infection defined as positive result for at least one of the following tests: L. infantum ELISA, L. infantum blood qPCR, IFN-ɣ assay, or LST. Only dogs subjected to at least one of the tests are considered Notes: Mann–Whitney tests were all conducted as two-tailed analyses, while Fisher’s exact tests were all single-tailed, examining the trend established when calculating the proportion of positive results Abbreviations: ELISA, enzyme-linked immunosorbent assay; EU, ELISA units; IFN-γ, interferon gamma; IQR, interquartile range; LST, leishmanin skin test; qPCR, quantitative polymerase chain reaction; SGH, salivary gland homogenate; n, number of dogs *P < 0.05, **P < 0.01 created except for RT. Age (months) was also considered this study examined anti-P. perniciosus saliva antibodies significantly related with EU and/or result in all models. in typically L. infantum-resistant Ibizan hounds [26] and in dogs of other breeds. Whereas previous epidemiologi- cal field studies have considered primarily seropositivity Discussion for L. infantum or a single measure of the anti-P. perni- Canine leishmaniosis caused by L. infantum is a complex ciosus saliva response [29, 33], this study examined mul- zoonosis that frequently eludes definitive diagnosis and tiple additional clinical parameters in relation to anti-P. manifests in a spectrum of disease depending on a vari- perniciosus antibodies, quantified by means of three dif- ety of host characteristics that are not yet entirely under- ferent methods. stood [47]. Although breed and host immune response Our results suggest that the anti-saliva antibody are both established as key factors in an individual’s response differed between Ibizan hounds and dogs of susceptibility to disease, it is unclear how these two are other breeds. As determined by SGH ELISA, a significant related [11, 13]. In order to address this knowledge gap, difference was detected in anti-P. perniciosus antibodies Burnham et al. Parasites Vectors (2020) 13:129 Page 9 of 15 Table 5 Comparison of recombinant protein rSP03B EU and proportion of positive results between dogs based on clinical parameters Variable n rSP03B EU rSP03B ELISA result Median ELISA units (IQR) Mann–Whitney Proportion positive Fisher’s P-value (count) exact test P-value Breed Ibizan hounds 47 25.63 (22.64) 0.1668 0.2979 (14) 0.1525 Other breeds 45 29.74 (23.27) 0.4222 (19) Age Adult (≥ 18 months) 53 33.17 (24.41) 0.0001*** 0.4528 (24) 0.0021** Juvenile (< 18months) 32 20.93 (15.56) 0.1250 (4) Sex Female 60 25.64 (22.26) 0.1672 0.3333 (20) 0.3188 Male 32 30.44 (24.20) 0.4063 (13) L. infantum ELISA Seropositive 19 34.10 (38.74) 0.0275* 0.5263 (10) 0.0709 Seronegative 71 25.63 (21.04) 0.3099 (22) L. infantum blood qPCR Positive 33 29.74 (19.26) 0.2787 0.3939 (13) 0.3016 Negative 42 24.32 (22.44) 0.3095 (13) LST Positive 45 23.69 (14.64) 0.5199 0.2222 (10) 0.5164 Negative 7 24.96 (23.73) 0.2857 (2) IFN-γ assay Producer 51 25.63 (16.83) 0.7196 0.2549 (13) 0.1742 Non-producer 14 23.19 (48.24) 0.4229 (6) Papule presence Papular 31 25.18 (20.78) 0.1554 0.3226 (10) 0.3906 Non-papular 61 26.87 (23.68) 0.3770 (23) L. infantum ­infectiona Yes 75 26.38 (19.39) 0.8795 0.3467 (26) 0.4533 No 15 24.96 (29.19) 0.4000 (6) a Presence of L. infantum infection defined as positive result for at least one of the following tests: L. infantum ELISA, L. infantum blood qPCR, IFN-γ assay, or LST. Only dogs subjected to at least one of the tests are considered Notes: Mann–Whitney tests were all conducted as two-tailed analyses, while Fisher exact tests were all single-tailed, examining the trend established when calculating the proportion of positive results Abbreviations: ELISA, enzyme-linked immunosorbent assay; EU, ELISA units; IFN-γ, interferon gamma; IQR, interquartile range; LST, leishmanin skin test; qPCR, quantitative polymerase chain reaction; n, number of dogs *P < 0.05, **P < 0.01, ***P < 0.001 in Ibizan hounds in comparison with other breeds of either no or very sparse, inconsistent insecticide treat- dogs living in the same geographical area under similar ment, both factors which individually predispose dogs environmental conditions. Although it failed to reach to a high degree of vector exposure [50]. In contrast with significance, a trend for a greater proportion of positive the more pendular ears of the other breed group, the SGH ELISA and rSP03B RT results also was observed for erected pinnae of Ibizan hounds offer an easily accessi- Ibizan hounds. Because the level of anti-sand fly saliva ble contact area devoid of hair, a characteristic that could antibodies is reflective of the intensity of vector exposure predispose the breed to phlebotomine bites. [29, 32, 34, 48, 49], these findings might imply a higher Studies in rodent and dog models have demonstrated frequency of sand fly bites in Ibizan hounds. It is unlikely that prior, repeated exposure to sand fly bites or immuni- that this difference in sand fly biting frequency is due zation with sand fly saliva or its derivatives is associated to different environmental conditions between groups; with reduced disease severity in cutaneous or visceral all dogs slept outdoors (outdoor kennels) and received Burnham et al. Parasites Vectors (2020) 13:129 Page 10 of 15 Table 6 Comparison of the proportion of positive rSP03B RT Our findings appear to support that Ibizan hounds results between dogs based on clinical parameters are more resistant to CanL and that this resistance is Variable n rSP03B RT result likely tied to a strong, parasite-specific cellular immune response. Ibizan hounds displayed a significantly higher Proportion Fisher’s positive (count) exact test rate of papular dermatitis clinically suggestive of L. infan- P-value tum infection when compared with dogs of other breeds in a highly endemic area of CanL. In particular, 63.83% Breed (30/47) of Ibizan hounds presented with single or mul- Ibizan hounds 47 0.2978 (14) 0.1999 tiple papules, whereas only 2.22% (1/45) of the other Other breeds 45 0.2000 (9) dogs had a single papulo-crusting lesion. This difference Age remained significant even when considering only papular Adult (≥ 18 months) 53 0.3208 (17) 0.0140* dermatitis cases with confirmed L. infantum infection. Juvenile (< 18 months) 32 0.0938 (3) Although this increased prevalence might be attrib- Sex uted to a higher rate of L. infantum infection, this study Female 60 0.2667 (16) 0.4053 found similar rates of infection between the two groups Male 32 0.2188 (7) (91.11% of Ibizan hounds and 75.56% of other breed L. infantum ELISA dogs), as defined by a positive result for at least one of the Seropositive 19 0.3684 (7) 0.1333 four associated tests (i.e. L. infantum ELISA, L. infantum Seronegative 71 0.2113 (15) qPCR of blood, IFN-γ assay, or LST). The host immune L. infantum blood qPCR response is known to influence clinical presentation of Positive 33 0.3030 (10) 0.1945 CanL [11], thus the higher prevalence of papular derma- Negative 42 0.1905 (8) titis observed in the Ibizan hounds could be related to a LST divergence in the immune response mounted against L. Positive 45 0.2000 (9) 0.1917 infantum infection. Indeed, in the present study, Ibizan Negative 7 0.4286 (3) hounds contained significantly more IFN-γ producers IFN-γ assay and produced significantly higher levels of IFN-γ than Producer 51 0.2353 (12) 0.3674 other breeds. This result aligns well with a previous work, Non-producer 14 0.3571 (5) which reported marked Leishmania-specific IFN-γ pro- Papule presence duction in healthy Ibizan hounds when compared to Papular 31 0.2258 (7) 0.4550 control and sick dogs [28]. In general, production of this Non-papular 61 0.2623 (16) pro-inflammatory cytokine is correlated with disease L. infantum ­infectiona resistance or mild disease [12, 19, 20, 52], while lack of 75 0.2400 (18) 0.5286 expression is described in experimentally infected symp- 15 0.2667 (4) tomatic dogs [12] and in dogs with a strong humoral a Presence of L. infantum infection defined as positive result for at least one of response and severe disease [17, 18]. Contrary to expec- the following tests: L. infantum ELISA, L. infantum blood qPCR, IFN-γ assay, or LST. tations, the proportion of positive LST reactions was not Only dogs subjected to at least one of the tests are considered substantially different between groups, although a previ- Notes: Fisher’s exact tests were conducted as single-tailed analyses, examining the trend established when calculating the proportion of positive result ous study described a higher frequency of positive results Abbreviations: ELISA, enzyme-linked immunosorbent assay; IFN-γ, interferon in Ibizan hounds [26]. However, both measures of LST gamma; IQR, interquartile range; LST, leishmanin skin test; qPCR, quantitative reaction intensity, skin thickness and lesion diameter, polymerase chain reaction; RT, rapid test; n, number of dogs were significantly higher in Ibizan hounds. *P < 0.05 In contrast, statistical analysis failed to find a significant difference between the two breed groups when rSP03B leishmaniosis [2, 35, 51]. This protective effect was later ELISA or RT were considered. However, results suggest shown by several studies to derive from the induction of a trend for a greater proportion of positive results for RT a strong Th1-dominant cellular immune response, char- in Ibizan hounds. Notably, age was considered a signifi- acterized by DTH reaction and increased production of cant factor in the results of both tests with rSP03B, and IFN-γ and/or IL-12 [2]. It could be hypothesized that the a significant moderate, positive correlation was observed repeated, high exposure to sand fly bites experienced by between age (months) and rSP03B EU (rs = 0.5212). A Ibizan hounds could induce a similar immune response significant positive relationship was also found for age upon infection with L. infantum, possibly contributing to and SGH EU, albeit to a lesser extent (rs = 0.2362). This their comparative immunocompetence. observation coincides with previous studies: older dogs possess greater cumulative exposure from experiencing Burnham et al. Parasites Vectors (2020) 13:129 Page 11 of 15 Table 7 Estimates of the multivariate models of the relationship between the outcome of the three salivary antigen tests (EU and/or result) and clinical parameters Model Factor Estimate SE P-value Missing observations SGH result Breed (Ibizan hound) 1.8057 1.6223 0.1424 0 L. infantum ­EUa 1.8612 0.7809 0.0171* 2 Age (months) 0.0282 0.0139 0.0423* 7 Age × Breedb − 0.0208 0.0201 0.3024 a SGH ­EU Breed (Ibizan hound) 0.2562 0.0824 0.0026** 0 L. infantum ­EUa 0.1738 0.0709 0.0164* 2 Age (months) 0.0032 0.0010 0.0028** 7 Age × Breedb − 0.0029 0.0017 0.1057 rSP03B ELISA result Breed (Ibizan hound) − 0.2759 0.9843 0.7793 0 L. infantum ­EUa 1.4593 0.7317 0.0461* 2 Age (months) 0.0295 0.0113 0.0093** 7 Age × Breedb − 0.0118 0.0184 0.5221 a rSP03B ­EU Breed (Ibizan hound) − 0.0018 0.0704 0.9792 0 L. infantum ­EUa 0.1812 0.0605 0.0037** 2 Age (months) 0.0037 0.0009 0.0001*** 7 Age × Breedb − 0.0013 0.0015 0.8660 RT result Breed (Ibizan hound) 1.0991 0.6163 0.0745 0 Age (months) 0.0259 0.0086 0.0003** 7 a All EU (SGH, rSP03B, and L. infantum) were ­log10 transformed prior to modeling b Interaction between age (months) and breed group Abbreviations: ELISA, enzyme-linked immunosorbent assay; EU, ELISA units; GLM, general linear model; IFN-ɣ, interferon gamma; IQR, interquartile range; LST, leishmanin skin test; qPCR, quantitative polymerase chain reaction; RT, rapid test; SE, standard error; SGH, salivary gland homogenate *P < 0.05, **P < 0.01, ***P < 0.001 multiple sand fly seasons [53], which can “prime” dogs relationship between papular dermatitis and age, the lat- for greater anti-sand fly saliva activity even at the begin- ter of which is likely a confounding variable in this study, ning of the transmission season, when lower titers would as discussed above. Papular dermatitis is more frequently be expected [33]. Because the Ibizan hound group con- diagnosed in younger dogs, and the characteristic pap- sisted of a lower percentage of adult dogs (< 18 months) ules are believed to be the site of a recent sand fly bite with a lower median age than the other breed group, age and inoculation with Leishmania parasites [5, 6], simi- is thus a likely confounding variable in the rSP03B tests’ lar to what is observed in the Old World form of human data. Considering the GLM analyses, age was determined local cutaneous leishmaniasis [54]. As such, dogs pre- to be significantly tied to outcome (EU and/or result) of senting with papular dermatitis, largely Ibizan hounds, two of the three tests, supporting this possibility. would be expected to have inherently lower anti-P. per- Interestingly, none of the other markers of a strong cel- niciosus antibody levels than their older, non-papular lular immune response alone (i.e. papule presence, IFN-γ counterparts. production or positive LST) were considered significant Leishmania infantum seropositivity appeared to be factors in anti-P. perniciosus antigenic response. Inclu- linked to a stronger anti-P. perniciosus reaction. Dogs sion of the single non-Ibizan hound categorized as pap- receiving a positive result for L. infantum ELISA exhib- ular was thought to influence these unexpected results, ited a trend for greater EU with either salivary antigen, as cytological analysis suggested it could be congruent although the difference was found significant only for with a different, eosinophilic arthropod bite reaction. Re- rSP03B. Seropositive dogs also were more likely to be classification of the animal as non-papular or its removal considered positive for anti-P. perniciosus antibodies from the sample entirely appeared to shift the trend in according to all three methods, reaching significance in favor of a stronger anti-P. perniciosus antigenic response the case of SGH ELISA. GLM results appear to support in dogs with papular dermatitis; however, no trend was this relationship: L. infantum seropositivity was shown to found to be statistically significant according to any of be a significant factor in four of five of the models. Nota- the three tests. An important factor to consider is the bly, the seropositive dogs in this study were primarily Burnham et al. Parasites Vectors (2020) 13:129 Page 12 of 15 classified as low- to medium-positive with relatively mild this discrepancy could be attributed to the inherent disease, as is common in endemic areas [3]. However, the lower sensitivity of the test, which was compared to SGH study included no cases with high sera anti-Leishmania ELISA in dogs experimentally exposed to P. perniciosus antibodies, which are associated with increased dis- bites [38]. The discrepancy between the rSP03B ELISA ease severity [47]. In order to more thoroughly exam- and RT likely derives from their respective methods ine the possible relationship between the anti-sand fly of production; the recombinant SP03B protein for the saliva antibodies, anti-Leishmania antibodies, and leish- ELISA was expressed in bacterial cells, whereas the RT maniosis prognosis, in future studies it would be ideal to protein was purified from Human Embryonic Kidney include dogs with different levels of specific antibodies (HEK) cells [59]. Future studies should include rSP03B against L. infantum and disease severity. proteins produced in HEK cells for both tests to enable Conflicting conclusions regarding the relationship a more direct comparison and to avoid undesirable non- between canine antibody response to sand fly saliva and specific reactions from co-purified bacterial proteins risk of Leishmania infection can be found in the litera- [38]. Compared to rSP03B, SGH is expected to be more ture: studies find either a positive correlation [33–55], a susceptible to cross-reactivity with saliva components negative correlation [29] or no correlation [33] between from related non-vector or vector sand fly species [31, anti-P. perniciosus saliva and anti-Leishmania antibodies. 60]. Although P. perniciosus is one of the most frequently A complicating factor is the variation in anti-Leishmania isolated and well-distributed sand fly species on Mal- antibodies during an individual dog’s course of infec- lorca, because this study was conducted with naturally tion. Anti-Leishmania antibodies are reported to reach infected dogs in an area with multiple species of sand a sustained plateau after an initial increase in sick dogs, flies [61], such cross-reactivity could very well have influ- compared with frequent fluctuations or seroconversion enced these results. in resistant dogs [56]. Additionally, a positive L. infantum The results of this study are likely influenced by sea- ELISA result is not necessarily indicative of L. infantum sonal fluctuations in anti-P. perniciosus saliva antibodies. infection, as cross-reaction can occur with other species Kinetics of anti-SGH and anti-rSP03B antibodies in dogs of Leishmania and other parasites, most notably Trypa- are known to mirror patterns in general sand fly activity, nosoma species [57, 58]. However, such cross-reaction is peaking during the warmer summer months, when vector unlikely to occur in the European Mediterranean, where density is highest, and dropping substantially during the infection with such cases occur less frequently than other colder months [33, 55]. The precise start, end, and shape endemic areas of CanL. Notably, this study employed of sand fly density patterns varies substantially between additional diagnostic techniques aside from serological sites based on numerous variables, including latitude, ele- detection of Leishmania to confirm presence of infection vation, annual temperature, activity season temperature, in some dogs. and individual year [62]. This variation in vector density, Herein, the diagnostic performance of three differ- and thus in anti-sand fly saliva antigenic response, could ent serological tests detecting P. perniciosus exposure prove challenging for cross-study comparison. In this was evaluated and compared: traditional SGH ELISA; study, all serum samples were taken during the month of recombinant protein rSP03B ELISA [34]; and the recently December, in their respective years, which is a period of developed rSP03B rapid sero-strip test [38]. SGH ELISA sand fly inactivity that would be expected to yield lower is considered the gold standard, with a higher sensitiv- levels of anti-P. perniciosus salivary antibodies. Compa- ity and specificity than either rSP03B ELISA or RT [38]. rable studies evaluate the anti-sand fly saliva antibody As in earlier studies in southern Europe [33, 34, 38, 55], response at the beginning and end of their respective a high percentage of agreement (75.76%) was observed areas’ seasons, although the precise time and total num- between SGH and rSP03B ELISA results along with ber of samplings vary [29, 33, 34, 55]. This seasonal a significant positive correlation between EU for the fluctuation also complicates the comparison of results two tests (rs = 0.4759, df = 90, P < 0.0001). The Cohen’s between salivary antigen tests. A field study using SGH kappa values comparing SGH ELISA to rSP03B ELISA or ELISA and RT to evaluate sand fly exposure concluded RT (0.3846 and 0.3736, respectively) were slightly lower that in periods of relatively low seroprevalence, such as than the values observed (0.67 and 0.83, respectively) in during the winter, the weak background signal apparent a previous study [38]. Furthermore, the present study on the RT impedes evaluation of the true degree of agree- found a statistically significant difference in the pro- ment between the tests [59]. Indeed, this study observed portion of positive results between rSP03B ELISA and the majority of dogs clustered about the cut-off for both either SGH ELISA or RT. In the case of rSP03B ELISA, Burnham et al. Parasites Vectors (2020) 13:129 Page 13 of 15 ELISA tests, with 75% of dogs producing no or faint sig- nal (score of 0 or 1) on the RT. Supplementary information Supplementary information accompanies this paper at https​://doi. Ultimately, any conclusions must take into account org/10.1186/s1307​1-020-3992-8. the small sample size, which could skew results. Expand- ing the study with an effort to control dog age is recom- Additional file 1: Figure S1. Example spectrum of rSP03B RT results. RT mended. In addition, not all clinical parameters were strips from a positive control and three experimental samples represent‑ assessed in each dog due to lack of sufficient sample or ing the spectrum of RT results. a positive control, equivalent to a score clinical justification. Overall, 47 dogs (51.09%) were of 4. b A clear positive band with lower intensity than the control band, classified as 3. c A weaker but still positive score of two. d A faint band, missing results for at least one of the four main diag- categorized as one. e A negative result, scored as zero. Notes: marked with nostic tests for L. infantum infection. Obtaining a more an asterisk, the first band on the nitrocellulose membrane is the test line, complete clinical picture of each dog would allow a the intensity of which is directly related to the amount of anti-rSP03B antibodies present in the sample. The second band is the control line to more thorough examination of the relationship between which the colloidal gold control conjugate binds, indicating that migra‑ these and other clinical parameters and the anti-sand fly tion has occurred properly and that the test strip is functional. saliva response. Evaluating other indicators of immune Additional file 2: Figure S2. Papular dermatitis in the inner aspect of the response in peripheral blood could be another possibility; pinnae of two Ibizan hounds clinically suggestive of L. infantum infection. Erythematous and slightly scaling papules (a) are indicative of acute or increased expression of the inflammatory cytokines IL-2 early-stage lesions, whereas ulcerated and/or crusted papules with the and TNF-α is associated with lower disease severity [10], characteristic “volcanic” appearance represent later-stage lesions (b) while the chemokine (C-X-C motif ) ligand 1 (CXCL1) and chemokine (C-C motif ) ligand 2 (CCL2) levels Abbreviations increase with greater disease severity [63]. Upregulation AIC: Akaike’s information criterion; AUC​: area under the curve; CanL: canine of toll-like receptor 2 (TLR2), the most commonly stud- leishmaniosis; CCL2: chemokine (C-C motif ) ligand 2; CXCL1: chemokine (C-X-C motif ) ligand 1; DNA: deoxyribonucleic acid; DTH: delayed type hyper‑ ied TLR in immunopathogenesis of canine leishmaniosis, sensitivity; ELISA: enzyme-linked immunosorbent assay; EU: ELISA units; GLM: has been revealed to correlated with disease progres- general linear model; IFN-γ: interferon gamma; IL: interleukin; IQR: interquartile sion [11, 19]. Moreover, including more data on parasite range; Ig: immunoglobulin; κ: Cohen’s kappa value; LSA: L. infantum soluble antigen; LST: leishmanin skin test; MHC: major histocompatibility complex; NO: load, which was not examined in all dogs, would enable nitric oxide; OD: optical density (absorbance); SE: standard error; SOD: stand‑ more direct comparison with similar studies with the L. ardized optical density (ELISA units); PCR: polymerase chain reaction; qPCR: infantum vector L. longipalpis that consider such param- quantitative PCR (real-time PCR); ROC curve: receiver operating characteristic curve; rs: Spearman rho coefficient; rSP03B RT: rSP03B rapid sero-strip test; eters [63]. Detecting parasite load in the skin in addition SGH: salivary gland homogenate; Th1: type 1 T helper cell (CD4+); Th2: type 2 to peripheral blood is also recommended; tissue parasite T helper cell (CD4+); TNF-α: tumor necrosis factor alpha; UK: United Kingdom. load was found to closely mirror clinical development of Acknowledgements CanL over time and to more accurately reflect host infec- The authors thank all dog owners that contributed to this study. We are very tivity than parasite load evaluated in bone marrow biop- grateful to Joan Llull, Antonio Meléndez, Dani Marí, and José Eduardo dos S. sies [64]. Silva for providing technical help during sampling. We also thank Professor Lluís Ferrer for his help with accommodation. Special thanks go to Marta Baxarias Canals and Lourdes Alarcón Sánchez for their assistance with sand Conclusions fly laboratory work. We also thank Shazia Hosein (Royal Veterinary College, When compared to other breeds living under similar University of London) for providing the control samples from UK dogs. The publication of this paper has been sponsored by Bayer Animal Health in the environmental conditions, Ibizan hounds appear to pos- framework of the 15th CVBD World Forum Symposium. Photographs in the sess higher levels of anti-P. perniciosus saliva antibodies, graphical abstract were procured under Creative Commons Licenses. Ibizan implying a higher frequency of exposure to sand fly bites. hound head image adapted from © User: Dannydulai/Wikimedia Commons/ CC-BY-SA-3.0. Ibizan hound and English setter images © User: Desaix83 / This robust response could be implicated in the stronger Wikimedia Commons/ CC-BY-SA-3.0. parasite-specific, T-cell mediated immune response observed in Ibizan hounds in this study, possibly con- Authors’ contributions LSG and LO designed the study. LO, LSG and SM collected the samples and tributing to the breed’s relative resistance to CanL. The provided clinical data. PMO performed whole blood assay, collection of anti-P. perniciosus antigenic response also was found to supernatants, and ELISA testing. SM performed the molecular work. LW, TS be stronger in older and L. infantum-seropositive dogs. and PV produced the salivary antigens, designed the salivary antigen tests, and provided technical expertise. ACB and MA performed the salivary antigen Moreover, these findings are observed to a variable extent tests. ACB conducted statistical analyses and modeling. ACB wrote the manu‑ in multiple measures, including traditional SGH ELISA script. LSG and LO revised the manuscript. All authors read and approved the as well as the less resource- and labor- intensive rSP03B final manuscript. ELISA and rapid sero-strip test. Although further inves- Funding tigation is necessary to confirm these findings, anti-P. This study was supported by Grants AGL2012-32498 and AGL2015-68477 of perniciosus saliva antibodies appear to negatively corre- the Ministerio de Economía y Competitividad (Spain), Charles University (UNCE 204072), and ERD funds (Centre for research of pathogenicity and virulence of late with susceptibility to L. infantum infection, and they parasites, CZ.02.1.01/0.0/0.0/16_019/0000759). show promise as an indicator of CanL prognosis. Burnham et al. Parasites Vectors (2020) 13:129 Page 14 of 15 Availability of data and materials 13. Maia C, Campino L. Biomarkers associated with Leishmania infantum Data supporting the conclusions of this article are provided within the article exposure, infection, and disease in Dogs. Front Cell Infect Microbiol. and its additional files. The datasets used and/or analyzed during the present 2018;8:302. study are available from the corresponding author upon reasonable request. 14. Moreira MAB, Luvizotto MCR, Garcia JF, Corbett CEP, Laurenti MD. Comparison of parasitological, immunological and molecular methods Ethics approval and consent to participate for the diagnosis of leishmaniasis in dogs with different clinical signs. Vet Signed, informed consent was obtained from all dog owners. Residual sam‑ Parasitol. 2007;145:245–52. ples of blood and serum were used for this study, thus ethical approval was 15. dos Santos WLC, Jesus EE, Paranhos-Silva M, Pereira AM, Santos JC, not needed. Baleeiro CO, et al. Associations among immunological, parasitological and clinical parameters in canine visceral leishmaniasis: emaciation, Consent for publication spleen parasitism, specific antibodies and leishmanin skin test reaction. Not applicable. Vet Immunol Immunopathol. 2008;123:251–9. 16. Carrillo E, Ahmed S, Goldsmith-Pestana K, Nieto J, Osorio Y, Travi B, et al. Competing interests Immunogenicity of the P-8 amastigote antigen in the experimental The authors declare that they have no competing interests. model of canine visceral leishmaniasis. Vaccine. 2007;25:1534–43. 17. Solano-Gallego L, Montserrrat-Sangrà S, Ordeix L, Martínez-Orellana P. Author details Leishmania infantum-specific production of IFN-γ and IL-10 in stimulated 1 Departament de Medicina i Cirurgia Animals, Facultat de Veterinària, Uni‑ blood from dogs with clinical leishmaniosis. Parasit Vectors. 2016;9:317. versitat Autònoma de Barcelona, Bellaterra, Spain. 2 Servei de Dermatologia, 18. Martínez-Orellana P, Marí-Martorell D, Montserrat-Sangrà S, Ordeix L, Fundació Hospital Clínic Veterinari, Universitat Autònoma de Barcelona, Bel‑ Baneth G, Solano-Gallego L. Leishmania infantum-specific IFN-γ produc‑ laterra, Spain. 3 Departament de Biologia, Sanitat i Medi Ambient, Facultat de tion in stimulated blood from dogs with clinical leishmaniosis at diagno‑ Farmàcia i Ciències de l’Alimentació, Universitat de Barcelona, Barcelona, Spain. sis and during treatment. Vet Parasitol. 2017;248:39–47. 4 Department of Parasitology, Faculty of Science, Charles University, Prague, 19. Montserrat-Sangrà S, Ordeix L, Martínez-Orellana P, Solano-Gallego L, Czech Republic. Montserrat-Sangrà S, Ordeix L, et al. Parasite specific antibody levels, Interferon-γ and TLR2 and TLR4 transcripts in blood from dogs with differ‑ Received: 15 December 2019 Accepted: 24 February 2020 ent clinical stages of leishmaniosis. Vet Sci. 2018;5:31. 20. Manna L, Reale S, Viola E, Vitale F, Manzillo VF, Michele PL, et al. Leishmania DNA load and cytokine expression levels in asymptomatic naturally infected dogs. Vet Parasitol. 2006;142:271–80. 21. de Vasconcelos TCB, Furtado MC, Belo VS, Morgado FN, Figueiredo FB. References Canine susceptibility to visceral leishmaniasis: a systematic review upon 1. Alvar J, Vélez ID, Bern C, Herrero M, Desjeux P, Cano J, et al. Leishma‑ genetic aspects, considering breed factors and immunological concepts. niasis worldwide and global estimates of its incidence. PLoS ONE. Infect Genet Evol. 2019;74:103293. 2012;7:e35671. 22. Quinnell RJ, Kennedy LJ, Barnes A, Courtenay O, Dye C, Garcez LM, et al. 2. Abdeladhim M, Kamhawi S, Valenzuela JG. What’s behind a sand fly bite? Susceptibility to visceral leishmaniasis in the domestic dog is associated The profound effect of sand fly saliva on host hemostasis, inflammation with MHC class II polymorphism. Immunogenet. 2003;55:23–8. and immunity. Infect Genet Evol. 2014;28:691–703. 23. Sanchez-Robert E, Altet L, Utzet-Sadurni M, Giger U, Sanchez A, Francino 3. Solano-Gallego L, Koutinas A, Miró G, Cardoso L, Pennisi MG, Ferrer L, O. Slc11a1 (formerly Nramp1) and susceptibility to canine visceral leish‑ et al. Directions for the diagnosis, clinical staging, treatment and preven‑ maniasis. Vet Res. 2008;39:36. tion of canine leishmaniosis. Vet Parasitol. 2009;165:1–18. 24. Sanchez-Robert E, Altet L, Sanchez A, Francino O. Polymorphism of 4. Solano-Gallego L, Morell P, Arboix M, Alberola J, Ferrer L. Prevalence Slc11a1 (Nramp1) gene and canine leishmaniasis in a case-control study. of Leishmania infantum infection in dogs living in an area of canine J Hered. 2005;96:755–8. leishmaniasis endemicity using PCR on several tissues and serology. J Clin 25. Quilez J, Martínez V, Woolliams JA, Sanchez A, Pong-Wong R, Kennedy LJ, Microbiol. 2001;39:560–3. et al. genetic control of canine leishmaniasis: genome-wide association 5. Lombardo G, Pennisi MG, Lupo T, Chicharro C, Solano-Gallego L. Papular study and genomic selection analysis. PLoS ONE. 2012;7:e35349. dermatitis due to Leishmania infantum infection in seventeen dogs: diag‑ 26. Solano-Gallego L, Llull J, Ramos G, Riera C, Arboix M, Alberola J, et al. nostic features, extent of the infection and treatment outcome. Parasit The Ibizian hound presents a predominantly cellular immune response Vectors. 2014;7:120. against natural Leishmania infection. Vet Parasitol. 2000;90:37–45. 6. Ordeix L, Solano-Gallego L, Fondevila D, Ferrer L, Fondati A. Papular 27. Solano-Gallego L, Llull J, Ramis A, Fernández-Bellon H, Rodríguez A, dermatitis due to Leishmania spp. infection in dogs with parasite-specific Ferrer L, et al. Longitudinal study of dogs living in an area of Spain highly cellular immune responses. Vet Dermatol. 2005;16:187–91. endemic for leishmaniasis by serologic analysis and the leishmanin skin 7. Borja LS, de Sousa OMF, da Solcà MS, Bastos LA, Bordoni M, Magalhães test. Am J Trop Med Hyg. 2005;72:815–8. JT, et al. Parasite load in the blood and skin of dogs naturally infected by 28. Martínez-Orellana P, Quirola-Amores P, Montserrat-Sangrà S, Ordeix L, Leishmania infantum is correlated with their capacity to infect sand fly Llull J, Álvarez-Fernández A, et al. The inflammatory cytokine effect of vectors. Vet Parasitol. 2016;229:110–7. Pam3CSK4 TLR2 agonist alone or in combination with Leishmania infan- 8. Molina R, Amela C, Nieto J, San-Andrés M, González F, Castillo JA, et al. tum antigen on ex-vivo whole blood from sick and resistant dogs. Parasit Infectivity of dogs naturally infected with Leishmania infantum to colo‑ Vectors. 2017;10:123. nized Phlebotomus perniciosus. Trans R Soc Trop Med Hyg. 1994;88:491–3. 29. Vlkova M, Rohousova I, Drahota J, Stanneck D, Kruedewagen EM, Mencke 9. Baneth G, Yasur-Landau D, Gilad M, Nachum-Biala Y. Canine leishmaniosis N, et al. Canine antibody response to Phlebotomus perniciosus bites nega‑ caused by Leishmania major and Leishmania tropica: comparative findings tively correlates with the risk of Leishmania infantum transmission. PLoS and serology. Parasit Vectors. 2017;10:113. Negl Trop Dis. 2011;5:e1344. 10. Pinelli E, Killick-Kendrick R, Wagenaar J, Bernadina W, del Real G, Ruiten‑ 30. Hostomska J, Rohousova I, Volfova V, Stanneck D, Mencke N, Volf P. berg J. Cellular and humoral immune responses in dogs experimen‑ Kinetics of canine antibody response to saliva of the sand fly Lutzomyia tally and naturally infected with Leishmania infantum. Infect Immun. longipalpis. Vector Borne Zoonotic Dis. 2008;8:443–50. 1994;62:229–35. 31. Lestinova T, Rohousova I, Sima M, de Oliveira CI, Volf P. Insights into the 11. Hosein S, Blake DP, Solano-Gallego L. Insights on adaptive and innate sand fly saliva: blood-feeding and immune interactions between sand immunity in canine leishmaniosis. Parasitol. 2017;144:95–115. flies, hosts, and Leishmania. PLoS Negl Trop Dis. 2017;11:e0005600. 12. Carrillo E, Moreno J. Cytokine profiles in canine visceral leishmaniasis. Vet 32. Quinnell RJ, Soremekun S, Bates PA, Rogers ME, Garcez LM, Courtenay O. Immunol Immunopathol. 2009;128:67–70. Antibody response to sand fly saliva is a marker of transmission intensity but not disease progression in dogs naturally infected with Leishmania infantum. Parasit Vectors. 2018;11:7. Burnham et al. Parasites Vectors (2020) 13:129 Page 15 of 15 33. Kostalova T, Lestinova T, Sumova P, Vlkova M, Rohousova I, Berriatua E, 52. Abbehusen MMC, dos Almeida VA, da Solcà MS, da Pereira LS, Costa DJ, et al. Canine antibodies against salivary recombinant proteins of Phle- Gil-Santana L, et al. Clinical and immunopathological findings during botomus perniciosus: a longitudinal study in an endemic focus of canine long term follow-up in Leishmania infantum experimentally infected leishmaniasis. PLoS Negl Trop Dis. 2015;9:e0003855. dogs. Sci Rep. 2017;7:15914. 34. Kostalova T, Lestinova T, Maia C, Sumova P, Vlkova M, Willen L, et al. The 53. Courtenay O, Macdonald DW, Lainson R, Shaw JJ, Dye C. Epidemiology of recombinant protein rSP03B is a valid antigen for screening dog expo‑ canine leishmaniasis: a comparative serological study of dogs and foxes sure to Phlebotomus perniciosus across foci of canine leishmaniasis. Med in Amazon Brazil. Parasitology. 1994;109:273–9. Vet Entomol. 2017;31:88–93. 54. Hepburn N. Cutaneous leishmaniasis: an overview. J Postgrad Med. 35. Gomes R, Oliveira F. The immune response to sand fly salivary proteins 2003;49:50. and its influence on Leishmania immunity. Front Immunol. 2012;3:110. 55. Velez R, Spitzova T, Domenech E, Willen L, Cairó J, Volf P, et al. Seasonal 36. Lima T, Martínez-Sogues L, Montserrat-Sangrà S, Solano-Gallego L, Ordeix dynamics of canine antibody response to Phlebotomus perniciosus saliva L. Diagnostic performance of a qPCR for Leishmania on stained cytologi‑ in an endemic area of Leishmania infantum. Parasit Vectors. 2018;11:545. cal specimens and on filter paper impressions obtained from cutaneous 56. Oliva G, Scalone A, Foglia Manzillo V, Gramiccia M, Pagano A, Di Muccio lesions suggestive of canine leishmaniosis. Vet Dermatol. 2019;30:318. T, et al. Incidence and time course of Leishmania infantum infections 37. Solano-Gallego L, Di Filippo L, Ordeix L, Planellas M, Roura X, Altet L, et al. examined by parasitological, serologic, and nested-PCR techniques in a Early reduction of Leishmania infantum-specific antibodies and blood cohort of naive dogs exposed to three consecutive transmission seasons. parasitemia during treatment in dogs with moderate or severe disease. J Clin Microbiol. 2006;44:1318–22. Parasit Vectors. 2016;9:235. 57. de Ferreira EC, de Lana M, Carneiro M, Reis AB, Paes DV, da Silva ES, et al. 38. Willen L, Mertens P, Volf P. Evaluation of the rSP03B sero-strip, a newly Comparison of serological assays for the diagnosis of canine visceral proposed rapid test for canine exposure to Phlebotomus perniciosus, vec‑ leishmaniasis in animals presenting different clinical manifestations. Vet tor of Leishmania infantum. PLoS Negl Trop Dis. 2018;12:1–14. Parasitol. 2007;146:235–41. 39. López-Ratón M, Rodríguez-Álvarez MX, Suárez CC, Sampedro FG. Optimal 58. Zanette MF, de Lima VMF, Laurenti MD, Rossi CN, Vides JP, da Vieira RFC, cutpoints: an R package for selecting optimal cutpoints in diagnostic et al. Serological cross-reactivity of Trypanosoma cruzi, Ehrlichia canis, tests. J Stat Softw. 2014;61:1–36. Toxoplasma gondii, Neospora caninum and Babesia canis to Leishmania 40. R Core Team. R: A language and environment for statistical computing. infantum chagasi tests in dogs. Rev Soc Bras Med Trop. 2014;47:105–7. Vienna: R Foundation for Statistical Computing; 2015. http://www.R-proje​ 59. Willen L, Lestinova T, Kalousková B, Sumova P, Spitzova T, Velez R, et al. ct.org. Accessed 1 Mar 2019. Field study of the improved rapid sand fly exposure test in areas endemic 41. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3:32–5. for canine leishmaniasis. PLoS Negl Trop Dis. 2019;13:e0007832. 42. Greiner M. Two-graph receiver operating characteristic (TG-ROC): a 60. Andrade BB, Teixeira CR. Biomarkers for exposure to sand flies bites as Microsoft-Excel template for the selection of cut-off values in diagnostic tools to aid control of leishmaniasis. Front Immunol. 2012;3:121. tests. J Immunol Methods. 1995;185:145–6. 61. Alcover MM, Ballart C, Martín-Sánchez J, Serra T, Castillejo S, Portús M, 43. Hosmer DW, Lemeshow S. Applied logistic regression. Hoboken: John et al. Factors influencing the presence of sand flies in Majorca (Balearic Wiley & Sons, Inc.; 2000. Islands, Spain) with special reference to Phlebotomus pernicious, vector of 44. Peng CJ, Lee KL, Ingersoll GM. An introduction to logistic regression Leishmania infantum. Parasit Vectors. 2014;7:421. analysis and reporting. J Educ Res. 2002;96:3–14. 62. Alten B, Maia C, Afonso MO, Campino L, Jiménez M, González E, et al. 45. Minitab 17 Statistical Software. State College, PA: Minitab, Inc.; 2010. Seasonal dynamics of Phlebotomine sand fly species proven vectors of http://www.minit​ab.com/. Accessed 10 Sep 2018. Mediterranean leishmaniasis caused by Leishmania infantum. PLoS Negl 46. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Trop Dis. 2016;10:e0004458. Meas. 1960;20:37–46. 63. Solcà MS, Andrade BB, Abbehusen MMC, Teixeira CR, Khouri R, Valenzuela 47. Baneth G, Koutinas AF, Solano-Gallego L, Bourdeau P, Ferrer L. Canine JG, et al. Circulating biomarkers of immune activation, oxidative stress leishmaniosis—new concepts and insights on an expanding zoonosis: and inflammation characterize severe canine visceral leishmaniasis. Sci part one. Trends Parasitol. 2008;24:324–30. Rep. 2016;6:32619. 48. Vlkova M, Rohousova I, Hostomska J, Pohankova L, Zidkova L, Drahota J, 64. Courtenay O, Carson C, Calvo-Bado L, Garcez LM, Quinnell RJ. Heteroge‑ et al. Kinetics of antibody response in BALB/c and C57BL/6 mice bitten by neities in Leishmania infantum infection: using skin parasite burdens to Phlebotomus papatasi. PLoS Negl Trop Dis. 2012;6:e1719. identify highly infectious dogs. PLoS Negl Trop Dis. 2014;8:e2583. 49. Martín-Martín I, Molina R, Jiménez M. Kinetics of Anti-Phlebotomus perni- ciosus saliva antibodies in experimentally bitten mice and rabbits. PLoS ONE. 2015;10:e0140722. Publisher’s Note 50. Martín-Martín I, Molina R, Rohoušová I, Drahota J, Volf P, Jiménez M. High Springer Nature remains neutral with regard to jurisdictional claims in pub‑ levels of anti-Phlebotomus perniciosus saliva antibodies in different verte‑ lished maps and institutional affiliations. brate hosts from the re-emerging leishmaniosis focus in Madrid. Spain. Vet Parasitol. 2014;202:201–16. 51. Ockenfels B, Michael E, McDowell MA. Meta-analysis of the effects of insect vector saliva on host immune responses and infection of vector- transmitted pathogens: a focus on leishmaniasis. PLoS Negl Trop Dis. 2014;8:e3197. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions