Discrepancies in the bilateral intradermal test and serum tests in atopic horses.

BACKGROUND: In equine atopic patients intradermal testing (IDT) and immunoglobulin (Ig)E serology are used frequently. There is little evidence regarding the reproducibility of the IDT and IgE serology in horses.
OBJECTIVES: To compare the results of a simultaneously performed IDT on the left and right side of the neck in atopic horses, and to compare these results with allergen-specific IgE serology. ANIMALS: Ten equine patients from a university hospital population with chronic urticaria and/or pruritus. METHODS AND MATERIALS: The IDT was performed using 16 allergens and the results were evaluated after 30 min, 1, 4 and 24 h. Thirteen allergens also were analysed in duplicate with two monoclonal allergen-specific IgE enzyme-linked immunosorbent assays (ELISA).
RESULTS: Good agreement (Kappa > 0.6) between left and right IDT was found only for Dermatophagoides farinae, Lepidoglyphus destructor, birch pollen mixture and perennial rye at 30 min, birch pollen mixture at 1 h, and Acarus siro and nettle and common mugwort mixture at 4 h. The bilateral comparison of the other allergens and even the same allergens at other time points showed little or no concordance between left and right IDT. The interlaboratory comparison between both ELISAs, and the comparison between the ELISAs and IDT, showed a good agreement for two of 13 allergens: D. farinae and Dermatophagoides pteronyssinus. CONCLUSIONS AND CLINICAL IMPORTANCE: Based on these preliminary data, IDT and IgE serological test results should be interpreted with great care and further studies are needed to indicate the clinical relevance of these findings.

Introduction

Equine atopic dermatitis (AD) is considered to be an immunoglobulin (Ig)E‐mediated hypersensitivity reaction triggered by environmental allergens such as pollens of trees, weeds and grasses, fungal spores, and house dust and storage mites. , The disease is characterized clinically by seasonal or nonseasonal pruritus, and/or chronic, recurrent urticaria, and, sometimes, respiratory disorders. In atopic horses, the face, pinnae, ventral thorax, abdomen, inguinal region and limbs are reported to be the most affected cutaneous regions. ,

Equine AD is diagnosed by combination of a compatible history, clinical signs and the exclusion of infectious or other noninfectious causes of pruritus or urticaria. Unlike canine AD, the equine condition may not always be chronic and life‐long.

The commonly accepted, and preferred, method of identifying allergens in atopic horses is an intradermal test (IDT). , Healthy (disease‐free) horses as well as atopic horses could react positively to the allergens used in an IDT. This might relate to insufficient data in regard to the allergen‐specific threshold concentrations. , , However, atopic horses show positive reactions more frequently in IDT than healthy horses. Serological IgE assays are considered less reliable or even completely unreliable for identifying allergens. Studies in horses have shown either a high variability between different serological assays or only slight agreement between the IDT and the IgE serology. Nevertheless, IDT in combination with an allergen‐specific IgE serological assay is frequently used by veterinary surgeons to select allergens to define avoidance strategies or to construct allergen‐specific immunotherapy (ASIT); in studies ASIT is reported to have a success rate of 64–84%. , ,

Atopic dermatitis can be hard to manage effectively solely with symptomatic medication and, therefore, it can be the cause of serious welfare compromise and impaired athletic performance. ,

The objectives of the present study were to determine the reproducibility of the IDT by performing a simultaneous bilateral intradermal test, and to determine the concordance between the intradermal test results and two serological IgE enzyme‐linked immunosorbent assays (ELISAs) performed by two separate commercial laboratories. It was hypothesized that in the same horse similar reactions to IDT were triggered on both sides of the neck at the same time point.

Methods and materials

All procedures were approved by the Ethical Committee of the Utrecht University (10801‐2016‐1). All owners signed a written consent for approval of the study.

Animals and inclusion criteria

A total of 10 horses (two mares, seven geldings and one stallion) with a history of chronic pruritus and/or chronic, recurrent urticaria were enrolled onto the study. All horses (six Dutch warmblood horses, one Icelandic horse, one Haflinger, one Welsh cob and one cross‐breed), with a mean age of 13 years, were referral patients.

Atopic dermatitis was diagnosed by excluding ectoparasite infestations and microbial infections, along with signalment. All horses showed clinical signs especially during the winter season except for one case diagnosed with insect bite hypersensitivity. Any medications that could influence the results of the IDT or IgE serology, including systemic and topical glucocorticoids were withdrawn for ≥14 days before the study.

Allergens for the intradermal test

A total of 16 aluminum‐precipitated allergens or allergen mixtures were used for the IDT. These included the grasses, trees, weeds, fungal spores and mites commonly found in the region. Extracts from all allergens were delivered as 3 mL glass vials in an aqueous solution with 0.4% phenol (ArtuVet Animal Health B.V.; Lelystad, the Netherlands). The concentration of the allergens used was 1,000 protein nitrogen units (PNU) for pollen and maize antigen, 100 PNU for mite antigen and 100 μg/mL for fungal spore antigen. All allergens were registered for intracutaneous use in dogs. Allergens were stored in the original vials at 4⁰C. All allergens and controls were transferred to room temperature 10 min before the test. A positive control solution of 1:1,000 w/v dilution of histamine was used and as a negative control an isotonic phosphate buffered solution with 0.4% phenol was used (ArtuVet Animal Health B.V.).

Intradermal testing

All IDTs were conducted in the winter season. During the 24 h test, nine horses were stabled at the premises of Utrecht University and one horse was tested at home. Ten minutes before the IDT was carried out, jugular blood samples were collected into 5 mL vacutainer serum tubes. Three anxious horses were lightly sedated during the IDT injection phase with intravenous detomidine hydrochloride 0.01 mg/kg (Domosedan 10 mg/mL, Orion Corporation; Espoo, Finland) 5 min before testing. A rectangular test area on both sides of the neck was carefully clipped. A waterproof permanent marker pen was used to indicate 18 injection sites on each side: three horizontal rows of six sites each with a minimum of 3 cm space between the sites (see Figure 1). Intradermal injections were placed in the skin under the pen marks. Individual 1 mL syringes were preloaded with ≥0.2 mL of each of the control solutions and individual allergens. A 0.1 mL volume of each extract was injected i.d. using a 25Gx5/8 needle to form a visible bleb. Intradermal injections were consistently applied in the same order. To minimize injection technique variability, the same investigator performed all intradermal injections in all horses. The investigator reading the IDT was not informed of the specific allergens at each injection site.

Figure 1

Before the start of the intradermal test, indelible marks were placed on the neck to identify the sites of injection of allergens.

All test locations were evaluated at four different time points: 30 min, 1, 4 and 24 h after injection. The reactions were scored subjectively for turgidity using a scale of − to +++ according to Lebis and colleagues: 0, no palpable reaction; +/−, very flat reaction with a badly defined outline; +, reaction with just palpable thickness; ++, reaction with obvious thickness; and +++, reaction with same thickness as histamine reference. The reactions also were scored objectively by measuring the mean diameter of the reaction (mm) with a ruler. A reaction to an allergen was considered positive when the diameter of any wheal was equal to or greater than the mean diameter between the negative control and the positive control being measured at 30 min and 1 h, and the turgidity was ≥+. At 4 and 24 h only a subjective score could be used because the positive histamine control was no longer visible. Therefore, only a turgidity score of ≥++ was considered positive at 4 and 24 h, and the diameter of this reaction should be larger than the mean diameter of the positive and negative control at the 30 min time point.

Allergen‐specific IgE serology

Serum from each horse was separated by centrifugation for 10 min at 1,000 and four samples per horse were stored at −70°C. Samples were numbered randomly and shipped to two diagnostic laboratories, Laboratory A (LA) and LB. Two duplicate serum samples were sent to each laboratory for ELISA testing to show intraassay variability. The ELISA was performed blinded to the IDT results. Both laboratories used an ELISA technique for their assay: LA used a monoclonal antibody cocktail‐based ELISA (macELISA) for detection of allergen‐specific IgE (Aller‐g‐Complete, Greer laboratories; Lenoir, NC, USA) and LB used an ELISA based on a specific single monoclonal anti‐IgE antibody (smELISA) which was generated using equine recombinant IgE. Semi‐quantitative allergen‐specific IgE levels reported by the laboratories were interpreted according to the EAU (ELISA Absorbance Unit) for LA and optical density (OD) for LB. A positive result was defined as >300 EAU in LA and >200 OD in LB.

Statistical analysis

All statistical analyses were performed with prism v8.2.0 for Windows (GraphPad Software; San Diego, CA, USA, www.graphpad.com). All data were checked for normality with a normal probability plot. The Spearman rank correlation test (instead of the Pearson rank correlation test) was used when a nonlinear relationship was expected.

For intraassay analysis of the bilateral IDT at each time point, Spearman rank correlation and Cohen’s kappa test of concordance were used. , To calculate the Spearman rank correlation, the original wheal diameters were divided by the mean diameter of the negative and positive control of the same side. For the kappa test, the combination of wheal diameter and turgidity was assessed.

For intralaboratory analysis of the duplicate blood samples (LA: EAU and LB: OD), both Pearson rank correlation and the kappa test of concordance were used. Logarithmically transformed data of macELISA and smELISA were used to calculate the Pearson rank correlation. For interlaboratory comparison of the two ELISAs, both Pearson rank correlation test and the kappa test of concordance were used. The average of both duplicate ELISAs was transformed logarithmically before calculating the Pearson rank correlation. Average duplicate ELISA values >300 EAU in LA and >200 OD in LB were regarded as positive in the kappa test of concordance.

In order to quantify the agreement between the IDT (30 min, 1 and 4 h) and the two ELISAs, Spearman rank correlation and the kappa test of concordance were used. For the Spearman rank correlation, first, the average values of the duplicate IDT were calculated and corrected with the mean diameter of the negative and the positive control. Secondly, the average values of the duplicates of the logarithmically transformed data of macELISA and smELISA were calculated. Consecutively, the average values of duplicate IDT and duplicate ELISAs were compared using the Spearman rank correlation test. For the kappa test of concordance between duplicate IDT and ELISAs, average duplicate ELISA values >300 EAU in LA and >200 OD in LB were regarded as positive. Average duplicate IDT values greater than the average diameter of the negative and the positive control were regarded as positive.

A correlation coefficient r > 0.8 and a P‐value < 0.001 were considered indicative of a good correlation. Kappa values (κ) of positive and negative results were calculated by using the formula , where is the sum of the agreements between both tests, is the expected frequency of these agreements occurring by chance and N is the number of horses being compared. Values of κ were interpreted as follows; ≤0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, good; and 0.81–1.00, excellent (in accordance with a previous published study).

Results

Positive and negative controls

All horses showed a positive response to histamine on both sides of the neck at 30 min, with a mean difference between right and left of 1 mm. At 1 h the positive control still was clearly visible and the mean decreased by 1 mm. A broad range in the positive control diameter was seen after 4 h even in the same horse as illustrated in Table 1. Almost all positive controls were 0 mm at 24 h, except in one horse where the positive control was 12 mm on the left side only. The mean diameter of the negative control was <4 mm at all time points.

Table 1

Range, mean and standard deviation (SD) (mm) of wheal size of the positive and negative controls

Time point	Positive control range	Positive control mean	SD	Negative control range	Negative control mean	SD
30 min	14–18	16	2	0–10	4	4
1 h	12–18	15	2	0–10	3	4
4 h	0–18	10	6	0	0	2
24 h	0–12	1	3	0	0	0

Bilateral IDT results

A total of 320 IDT allergen injections were administered. Considering all four time points, these produced a total of 140 positive reactions. Of all positive reactions, 37% were positive on both sides (52 injection sites) and 63% were positive on one side only (88 injection sites). The overall difference in wheal diameter between left and right sides is displayed in Table 2.

Table 2

Summary of duplicate intradermal test results

Time point	Bilateral different results		Bilateral positive results
	Number of positive reactions (n)	Mean difference in mm (range)	Number of positive reactions (n)	Mean difference in mm (range)
30 min	23	6.6 (1–18)	10	3.4 (1–7)
1 h	40	6.2 (0–14)	12	2.0 (1–5)
4 h	24	4.3 (0–20)	30	2.8 (0–15)
24 h	1	10 (10)	0	‐

Bilateral different results: the right side showed a positive reaction and the left side showed a negative reaction or vice versa.

Bilateral positive results: both sides showed a positive reaction.

No correlation in wheal diameter between the left‐ and right‐side IDT was found for almost all allergens at all time points except for Dermatophagoides farinae at 1 h (Table 3). Good to excellent agreement (κ) between positive and negative results at 30 min was found for four of 16 allergens: Lepidoglyphus destructor, D. farinae, perennial rye and birch pollen mixture. At 1 h, good agreement (κ) was found for one of 16 allergens (birch pollen mixture); at 4 h, good κ was found for two of 16 allergens (Acarus siro and the common mugwort and nettle mixture). At 24 h after the intradermal injections, no bilateral positive reactions were noticed. When considering all time points, the highest numbers of bilateral positive reactions were noticed for D. farinae (five of 10 horses) followed by perennial rye (three of 10 horses) and Dermatophagoides pteronyssinus, birch pollen mixture and pellitory‐of‐the‐wall (two of 10 horses).

Table 3

Duplicate intradermal test results at 30 min, 1 and 4 h. Spearman rank correlation for comparison of the wheal diameters and Kappa (κ) for comparison of positive and negative results

Allergens		30 min	30 min	1 h	1 h	4 h	4 h
		Intraassay correlation (r)	Intraassay κ	Intraassay correlation (r)	Intraassay κ	Intraassay correlation (r)	Intraassay κ
Mites	Tyrophagus putrescentiae	0.22	0	0.18	0	0.61	0
	Lepidoglyphus destructor	0.26	0.6**	0.35	0.2	0.27	−0.1
	Acarus siro	0.03	0	0.61	0	0.85	0.6**
	Dermatophagoides farinae	0.55	1**	0.91*	0.5	−0.13	0.4
	Dermatophagoides pteronyssinus	0.60	0.4	0.35	0.4	0.50	0.4
Grasses	Lolium perenne (perennial rye)	0.33	1**	−0.71	0	−0.18	0.5
	Phleum pratense (timothy grass)	0.05	0	−0.38	0	0.22	neg
	Poa pratensis (Kentucky blue grass)	0.24	0	0.21	0	0.46	neg
Trees	Birch, alder and hazel mixture	0.30	0.6**	0.27	0.6**	0.41	0.4
Weeds	Parietaria officinalis (pellitory‐of‐the‐wall)	0.21	0	0.19	0	0.12	0.2
	Chenopodium album (lamb’s quarters)	0.56	0	0.51	0.3	−0.32	neg
	Plantago lanceolate (English plantain)	−0.22	0	0.031	−0.2	0.43	neg
	A combination of Artemisia vulgaris (common mugwort) & Urtica dioica (nettle)	0.09	−0.2	−0.18	−0.2	−0.09	0.7**
	Rumex acetosella (sheep sorrel)	0.40	0	0.09	−0.2	−0.20	0
Fungal spores	A combination of Alternaria alternata, Aspergillus fumigatus and Cladosporium herbarum	0.48	neg	0.18	0	0.46	neg
Grain	Zea mays (maize)	0.17	−0.2	0.34	0	−0.20	−0.1

Good correlation ( > 0.8, P < 0.001);

κ ≥ 0.6; neg, all negative.

Intra‐ and interlaboratory analysis of macELISA and smELISA

The intra‐ and interlaboratory correlation and κ of macELISA (LA) and smELISA (LB) are illustrated in Table 4. To assess intralaboratory test variability, two serum samples from the same horse were analysed in duplicate in LA and LB.

Table 4

Intra‐ and interlaboratory comparison of immunoglobulin (Ig)E serological results of Laboratory A [LA: monoclonal antibody cocktail‐based enzyme‐linked immunosorbent assay(ELISA)] and Laboratory B (LB: single monoclonal ELISA)

Allergens		LA	LA	LB	LB	LA/LB	LA/LB
		Intraassay correlation (r)	Intraassay κ	Intraassay correlation (r)	Intraassay κ	Interassay correlation (r)	Interassay κ
Mites	Tyrophagus putrescentiae	0.94*	neg	0.85*	1**	–0.79	0
	Lepidoglyphus destructor	0.83*	0	0.05	neg	–0.09	neg
	Acarus siro	0.93*	1**	0.72	0.6**	–0.82	–0.1
	Dermatophagoides farinae	0.98*	0.7**	0.97*	1**	0.89*	1**
	Dermatophagoides pteronyssinus	0.99*	1**	0.94*	1**	0.96*	1**
Grasses	Lolium perenne (perennial rye)	–0.15	1**	0.86	neg	0	0
	Phleum pratense (timothy grass)	0.57	neg	0.71	neg	–0.19	neg
	Poa pratensis (Kentucky blue grass)	0.93*	0	n.m.	n.m.	n.m.	n.m.
Trees	Birch, alder and hazel mixture	0.98*	1**	0.66	neg	0	0
Weeds	Parietaria officinalis (pellitory‐of‐the‐wall)	0.94*	1**	0.09	neg	–0.5	0
	Chenopodium album (lamb’s quarters)	0.98*	1**	0.89*	neg	0.19	0
	Plantago lanceolate (English plantain)	0.98*	0.6**	0.15	neg	0.21	0
	Artemisia vulgaris (common mugwort)	0.94*	1**	0.95*	neg	–0.53	0
	Rumex acetosella (sheep sorrel)	n.m.	n.m.	0.91*	1**	n.m.	n.m.
Fungal spores	A combination of Alternaria alternata, Aspergillus fumigatus and Cladosporium herbarum	0.46	neg	–0.01	neg	–0.30	neg

Pearson rank correlation for comparison of numerical values [ELISA Absorbance Units (EAU) and optical density (OD)] and Kappa (κ) for comparison of positive and negative results.

Good correlation ( > 0.8, P < 0.001).

κ ≥ 0.6; n.m., allergen not measured; neg, all negative.

In LA a good intralaboratory correlation in EAU of 11 of 14 allergens was found, except for perennial rye, timothy grass and fungal spores. Good to excellent intralaboratory agreement in κ between positive and negative results in LA was found for nine of 14 allergens. For three allergens (Tyrophagus putrescentiae, timothy grass and fungal spores) it was not possible to calculate the κ agreement due to complete negative results of the duplicate blood samples.

In LB a good intralaboratory correlation in OD for six of 14 allergens was found. A difference in intralaboratory correlation in LB was found for L. destructor, Acarus siro, perennial rye, timothy grass, birch pollen mixture, pellitory‐of‐the‐wall, English plantain and fungal spores. Good to excellent intralaboratory κ agreement in LB was found for five of 14 allergens. For nine of 14 allergens it was not possible to calculate the κ agreement due to complete negative results of the duplicate blood samples.

The interlaboratory analysis between the macELISA and smELISA showed both a good correlation and an excellent κ for two of 13 allergens: D. farinae and D. pteronyssinus (Table 4). As far as negative and positive results were concerned, intralaboratory differences of LA were 2.9%, intralaboratory differences of LB were 1.4%, and interlaboratory differences between macELISA and smELISA were 8.6%

Interassay analysis of IDT versus macELISA and smELISA

No correlation was found between the macELISA and IDT and smELISA and IDT at all time points for all allergens. Good to excellent κ agreement of the positive and negative outcomes of macELISA versus IDT was found for D. farinae (30 min and 1 h IDT), D. pteronyssinus (30 min, 1 and 4 h IDT) and perennial rye (4 h IDT). Good to excellent κ agreement of smELISA versus IDT was found for D. farinae (30 min and 1 h IDT) and D. pteronyssinus (30 min, 1 and 4 h IDT). Interassay correlation between IDT at 30 min and macELISA and smELISA are illustrated in Table 5.

Table 5

Comparison of the intradermal test at 30 min and IgE serology results of Laboratory A [LA: monoclonal antibody cocktail‐based enzyme‐linked immunosorbent assay(ELISA)] and Laboratory B (LB: single monoclonal ELISA)

Allergens		LA/IDT	LA/IDT	LB/IDT	LB/IDT
		Interassay correlation (r)	Interassay κ	Interassay correlation (r)	Interassay κ
Mites	Tyrophagus putrescentiae	0.09	neg	0.03	0
	Lepidoglyphus destructor	−0.02	0	−0.18	0
	Acarus siro	0.15	0	−0.25	0
	Dermatophagoides farinae	0.59	0.6**	0.66	0.6**
	Dermatophagoides pteronyssinus	0.63	0.6**	0.47	0.6**
Grasses	Lolium perenne (perennial rye)	0.10	−0.1	–0.1	0
	Phleum pratense (timothy grass)	0.33	neg	–0.28	neg
	Poa pratensis (Kentucky blue grass)	0.20	neg	n.m.	n.m.
Trees	Betula (birch)	0.35	−0.1	−0.12	0
Weeds	Parietaria officinalis (pellitory‐of‐the‐wall)	0.23	−0.1	−0.32	0
	Chenopodium album (lamb’s quarters)	−0.15	−0.1	−0.56	0
	Plantago lanceolate (English plantain)	0.21	0	0.33	neg
	Artemisia vulgaris (common mugwort)	−0.04	−0.1	0.2	0
	Rumex acetosella (sheep sorrel)	n.m.	n.m.	−0.32	0
Fungal spore	Alternaria alternata	0.21	neg	0.06	neg

Spearman rank correlation for comparison of numerical values and Kappa (κ) for comparison of positive and negative results.

Good correlation (r> 0.8, P < 0.001).

κ ≥ 0.6; n.m., allergen not measured; neg, all negative.

Discussion

To the best of the authors’ knowledge, this is the first study assessing the agreement between IDT performed on the left side and right side in atopic horses. As our study was exploratory, only a small number of horses were included and, therefore, the results should be interpreted with caution. However, the outcome of the bilateral IDT was surprising considering the extent of the differences between the left and right sides at the same time points. Previous studies in horses have shown that the repeatability of the IDT (20 min after injection) was poor when the IDT was performed with an interval of four to five months in the same horse. Furthermore, multiple studies over the years have shown that even healthy horses can react differently with intradermal tests, which is one of the reasons why the determination of thresholds for allergens is so difficult in horses. , , , However, from human studies, it is known that the repeatability of a skin prick test for birch pollen, grass pollen, D. farinae and D. pteronyssinus is high in symptomatic allergic individuals. Based on our results, only the IDT results of D. farinae, L. destructor, perennial rye and birch pollen mixture can be reliably assessed at 30 min. It is interesting to note that two of these allergens overlap with the allergens showing a good correlation in atopic people. The highest number of positive IDT reactions were noticed for D. farinae which is in agreement with previous studies. , House dust mite antigens can be found in horse rugs and because all these horses were ridden, they would most likely have been in contact with the D. farinae antigen.

However, several weaknesses of the performed IDT must be acknowledged. Technical errors could have occurred when performing the IDT because the volume and the depth of the intradermal allergen injection could have varied. It was attempted to minimize this by filling each syringe with a specified volume for each allergen. Technical errors also were minimized by always using the same batch, the same type of syringe and needle, and by ensuring that all injections were administered by the same experienced investigator.

Randomization of the allergens per injection site was not performed. However, the investigator administering and reading the IDT was blinded to the specific allergens per injection site. Therefore, interpretation biases were assumed to be minimal.

Three anxious horses were lightly sedated with detomidine hydrochloride and this sedation might have influenced the IDT outcome. However, previous studies confirmed clear positive IDT reactions to histamine and allergens in horsed sedated with detomidine hydrochloride. In atopic dogs, the use of α2‐adrenoceptor agonists during IDT is, also, regarded as a good sedative for facilitating intradermal testing with minimal to no effect on the outcome of the IDT. Especially because the sedated horses showed clear bilateral positive reactions to histamine and showed positive reactions to allergens in the IDT, it was concluded that the influence of sedation was, at most, minimal.

A lack of knowledge regarding the correct threshold concentrations of the used allergens in horses in this study could explain false positive or false negative responses to the IDT. However, to the best of the authors’ knowledge, threshold concentrations are not established in Europe. In Australia, however, one study suggested the use of >1,000 PNU/mL for pollen and fungal spore allergens. Pollen allergen extracts of 1,000 PNU/mL were used in this study and no bilateral positive reactions were identified at 30 min and 1 h for two of the tested grasses and all five weeds, indicating that for these allergens irritant reactions might not play a role in our study. Recommended threshold concentrations for IDT allergen solutions of dust mites in horses are variable. , In the present study we used the recommended allergen concentration for atopic dogs for the dust and storage mites owing to a lack of conclusive evidence for the best concentration of mite allergen extract in horses. This concentration might be too high, although if that was the case, bilaterally different results would not have been expected.

The intralaboratory variability of the macELISA and smELISA was evaluated by sending duplicate blood samples to each laboratory. The macELISA showed a good correlation and κ for the majority of the allergens and can, therefore, be regarded as an ELISA with a good intralaboratory reproducibility. The smELISA showed a good correlation for the minority of the allergens. In particular, although nine of 14 allergens showed a negative OD (OD < 200) in combination with a low correlation, it might be possible that the smELISA only shows a rather high variability with OD < 200. Larger studies are indicated to support this hypothesis. However, in atopic dogs, it is already known that another IgE serological test, the FcεRIα‐based ELISA, showed prominent variability with low OD values around the cut‐off point.

The interlaboratory variability correlation of the macELISA and smELISA was good for only two of 13 allergens: D. farinae and D. pteronyssinus. In atopic horses, an interlaboratory comparison of these two commercially available tests has never been performed before as far as the authors are aware. Due mainly to time and financial constraints, some equine practitioners prefer to use only the serological IgE assay to identify allergens for immunotherapy. Based on our report, the result of an in vitro serological IgE assay is highly dependent on the specific test chosen and therefore immunotherapy recommendations also will show undesirable high variability for storage mites, grass, tree and weed pollen, and fungal spores.

This study described a high interassay variability between the IDT and macELISA IgE serology and IDT and smELISA IgE serology, for storage mites, grass, weed and tree pollen. Previous studies in horses, investigating the agreement between IDT and serological IgE tests, have already shown little or no concordance between the tests beyond chance for the majority of allergens. , As an example, one study reported as in this report, no agreement for birch pollen, perennial rye, T. putrescentiae, A. siro, English plantain, Kentucky blue grass, common mugwort and fungal spores, when comparing IgE serology with the IDT in horses with skin hypersensitivities. Another study compared three different serological IgE tests with the IDT and found that none of their in vitro assays of allergen‐specific IgE showed good correlation with the IDT, although the FcεRI‐based ELISA showed a significantly better correlation than the other two serological IgE serological tests. Retrospectively, it would have been of value to include the FcεRIα‐based ELISA in our interlaboratory analysis to assess if this test showed a better interlaboratory correlation between the macELISA and smELISA and between the results of the IDT; another study reported a specificity > 89% for both house dust mites in atopic dogs in a comparison of the IDT and a FcεRIα‐based ELISA.

To summarize, the results presented herein demonstrate that in atopic horses the results of a bilateral intradermal test showed rather high variability. In addition, high variability was also demonstrated between the two investigated in vitro IgE serological tests and, moreover, between the IgE serological tests and the IDT at all time points. Some veterinarians dealing with equine cases use the IDT, mostly in combination with an in vitro IgE serological test, to identify allergens for avoidance or to compose ASIT in horses already clinically diagnosed as atopic. As always, evaluation of results should be performed in association with the history, clinical signs, exposure to allergens and elimination of other causes of pruritus. However, the noted high variability in both the IDT and the macELISA and smELISA could highly influence ASIT recommendations. In conclusion, our results show that there is a need for better reliability of both IDT and IgE serological tests. Randomized, independent, large‐scale, peer‐reviewed studies are needed to assess the variability of the IDT and in vitro equine IgE serological tests to provide evidence‐based guidelines for equine intradermal and IgE serological testing in atopic horses.