Diagnosis of EP as the cause of clinical disease can be challenging in endemic areas, where the percentage of serologically and molecularly positive horses is high, and the detection of parasites does not necessarily imply on the cause of non-specific clinical signs . Therefore, in clinical cases suspected as EP, quantitative evaluation of parasite load, using molecular tools, may assist in determining a threshold for cause of disease decision. Here we demonstrate that clinically infected horses with either parasite of EP have significantly higher parasite loads and lower PCV than subclinically infected horses. This is intuitive, as merozoite replication in erythrocytes causes hemolysis, the main clinical manifestation of EP. Thus, higher parasite loads may induce increased hemolysis that will be reflected in lower PCV. Parasite loads of both clinical and subclinical horses were generally lower in cases of B. caballi infection than in cases of T. equi infection. This may explain the milder clinical disease in B. caballi infections compared to T. equi infections, and to the possible natural clearance of B. caballi parasitemia without treatment, while T. equi carriage is usually life-long [1, 2].
To determination whether T. equi is the probable cause of disease in suspected clinical cases we established a clear cutoff (P < 0.001) between clinical (0.12–5.3% PE) and subclinical (3*10− 4-5*10− 3% PE) cases. The parasitemia values in our study concur with published subclinical range (1.99–1000 parasites per µl blood , equivalent to 2.2*10− 5 to 0.011% PE). In clinical cases T. equi parasitemia ranges between 1–7% PE, and may reach up to 95% [1, 34], however, we had clinical cases with parasitemia as low as 0.12% PE, which also manifested in low PCV.
Three of six clinically B. caballi infected horses showed parasitemia below the documented range (0.1–10% PE [1, 34]). Parasitemia in subclinical carriers of B. caballi ranged between 0.0001 and 0.0012% PE, which was significantly lower than the clinical cases (P < 0.001). To the best of our knowledge, no previous study quantified B. caballi parasitemia in subclinical horses. Although in this group the difference between clinical and subclinical parasitemia was less distinct, it still manifested in lower PCV, and allowed to establish a cutoff value to identify B. caballi as the probable cause of clinical disease.
Despite the limited number of cases included in the quantitative analysis, the highly significant results may serve as first indication that qPCR may serve as a diagnostic tool. Additional data should be collected to validate this method for clinical use.
All seven clinical cases which originated from different farms and geographical locations were classified as T. equi 18S rRNA genotype A. The A genotype is not the most prevalent in our area (30% of subclinical horses in this study, 33% in a previous study), and is rarely found in highly endemic farms . Thus, although the number of cases was limited, genotype A may be associated with clinical disease. Genotype A was previously isolated from horses in both endemic and non-endemic countries [5, 10, 31, 35], it was isolated in two outbreaks in the US , and was found to be associated with clinical and seropositive cases in Italy . Interestingly, genotype A was the predominant genotype isolated from ticks collected from horses in Israel, including in farms in which this genotype was not isolated from horses (unpublished data). It is possible that this genotype is more adapted to the tick vector environment and encounter an active barrier at the horse stage, meaning that genotype A is more likely to lead to clinical disease, while genotypes B, C and D are more likely to result in subclinical infection . With the recent concerns regarding the classification of Theileria species according to the 18S rRNA gene [6, 7], it is possible that this “A genotype” is the cause of the “classic” equine theileriosis, while other genotypes may represent closely related, less pathogenic, species or subspecies. More comprehensive genetic investigation of different genotypes is required to support this hypothesis.
In an attempt to partially address these issues, we classified T. equi according to three different genes: 18S rRNA, ema-1 and ema-2, as the last two loci had sufficient number of published sequences for comparative analysis. However, we could not amplify ema-1 and ema-2 from all samples, probably due to polymorphism in the primer sites or the sensitivity of the PCR assay, and most of the successfully sequenced amplicons were from isolates of 18S rRNA genotype A, as previously reported . The over represented 18S rRNA genotype A may be the result of the higher parasitemia in the clinically infected horses, enabling better detection in PCR. Nevertheless, using qPCR, ema-1 gene was detected in all samples, strengthening parasite identification.
The 18S rRNA classification in subclinical horses resulted in prevalence of genotypes D (57.5%), A (30%) and C (12.5%) as previously described in our area  strengthen the statistical power of the data with a larger sample size. Sequence analysis of both ema-1 and ema-2 did not reveal much polymorphism in these loci within a geographic area, as was previously demonstrated [10, 14, 16–18, 19]. Only 16 ema-2 sequences were available for classification, mostly from India and the US along with four sequences from Nigeria (generated by our group, Acc. No. MN519202-MN519205). Although this gene had low variability, we identified three distinct genotypes, which also differ in their amino acid sequences. This variability may be important if it affects immune response and may lower the sensitivity of ema-2 based ELISA assays .
Genetic classification of B. caballi is limited, with two 18S rRNA genotypes identified in South Africa . We were unable to amplify this gene from the subclinical horses, since all were co-infected with T. equi, and the primers are not species specific. Therefore, we used the rap-1 gene which is specific to B. caballi and is fairly conserved, with some heterogenicity between American and Asian-African strains [12–14, 19]. Two 18S rRNA sub-genotypes were identified in clinical cases, which correlated with the rap-1 sub-genotypes of the same samples. Comparison of the rap-1 gene between clinical and subclinical cases did not reveal differences in parasite genotypes in relation to clinical disease.