Evaluation of a direct lymphocyte proliferation test for the diagnosis of canine food allergies with delayed reactions after oral food challenge

Animals

Between March and November 2023, we received samples from 55 dogs diagnosed with FA by veterinary specialists. In these dogs, the diagnosis was based on compatible cutaneous or digestive signs, a full improvement after a restrictive diet, and a subsequent flare after OFC, per current practice standards. The symptomatology, TTF of signs after OFC and food ingredients causing flares were noted in a standardised questionnaire (Data S1). Of these 55 dogs with FA, 44 (80%) had a flare of cutaneous signs that occurred ≥6 h after OFC, the longest duration expected for an IgE-mediated immediate FA in humans.¹² In 28 of the 44 (64%) dogs with delayed FA, and after verbal consent was obtained from their owners, 10 mL of peripheral blood was collected by a single venipuncture in a dry tube and either used for serum isolation for routine allergen-specific IgE determination (Pet Allergy Xplorer; Nextmune) or placed in an ethylenediaminetetraacetic acid (EDTA)-containing tube for the performance of the LPT. After securing approval from the local institutions' animal care and use committee, we also obtained 5 mL of blood in EDTA from 10 laboratory beagles that had eaten the same chicken-and-rice diet (Nutribest Adult7+ Chicken and Rice; Picart Petcare) in the preceding 3 years. Finally, a 5 mL blood sample in EDTA was collected after informed verbal consent from the owners of 25 healthy dogs from Belgium and Lithuania presented to their veterinary surgeons for vaccination or neutering. All samples were kept refrigerated until shipment to the laboratory. All results were provided back to the veterinary surgeon for sharing with pet owners.

Allergen source

Most food extracts used to stimulate canine PBMCs were produced at Nextmune (Madrid) either with phosphate-buffered saline (PBS) or 0.02% sodium hydroxide (NaOH) (for flour extracts). All were prepared from raw ingredients and, in the case of meats, with the lean parts of muscles. For plant foods, we used different parts and varieties of vegetables and fruits. The peanut extract was a 50% glycerol, 0.4% phenol and saline commercial solution for human prick testing (Inmunotek). Bovine IgG (Bos d 7; Merck KGaA) was the only molecular allergen tested.

Peripheral blood mononuclear cell isolation

Peripheral blood mononuclear cells were isolated by density gradient separation using Histopaque 1077 Hybri-Max (Sigma-Aldrich/Merck Life Sciences). First, 5 mL of blood in EDTA was diluted and mixed 1:1 in 0.9% saline before the addition of 3 mL of Histopaque 1077. This mixture was centrifuged for 30 min at 440 g without braking. The PBMC layer was aspirated and subjected to three successive cycles of saline wash. After the last centrifugation, the supernatant was removed, and PBMCs were resuspended in 5 mL of Gibco AIM-V serum-free medium (Fisher Scientific, Spain). After mixing 20 μL of the PBMC suspension with 20 μL of trypan blue (Fisher Scientific), we counted live cells on a TC20 automated cell counter (Bio-Rad); we only kept samples with >70% of viable cells for food allergen stimulation. The PBMC suspension was adjusted to 1 million cells/mL in the AIM-V medium.

Lymphocyte proliferation after food allergen stimulation

In dogs, PBMCs contain >90% of T and B cells.^{13, 14} These are the only cells with the ability to proliferate after being cultured with antigens. Therefore, the culture of PBMCs with food allergens is referred to as an LPT, just as it is for humans.¹⁵ We placed 100,000 cells in 100 μL of medium in each well of a tissue culture plate (Corning 96; Fisher Scientific). Then, 100 μL of a 10 μg/mL solution of each food allergen extract diluted in the medium was added to the PBMCs for a final allergen concentration in each well of 5 μg/mL. Food allergen stimulations were done in duplicates in 22 of 28 dogs. The three negative controls were similar volumes of saline, NaOH and 50% glycerol/0.4% phenol (all from AppliChem); these last two controls being added as some food allergen extracts contained such chemicals (see above). The positive control consisted of 100 μL of the lymphocyte mitogen concanavalin A at 10 μg/mL (Sigma-Aldrich/Merck). Food allergen-PBMC co-culture plates were then kept in an incubator at 37°C in 5% CO₂ for 5 days. On Day (D)5, 20 μL of trypan blue was added to 20 μL of cell culture to evaluate cellular viability after stimulation, and cells were counted again using the TC20 automated cell counter. Subsequently, the food allergen-induced proliferation was expressed as a stimulation index (SI), denoting the ratio of cell numbers after 5 days of allergen co-culture relative to those cultured with saline at the same time point.

Determination of interleukin-2 concentrations in cell culture supernatants

As the stimulation of sensitised lymphocytes leads to both their proliferation and the release of interleukin (IL)-2, we aimed to validate the results of the LPT by comparing the proliferation of PBMCs with the concentrations of IL-2 in the supernatant of the cells after culture with food allergens. Concentrations of this cytokine were determined using the Canine IL-2 DuoSet ELISA (DY1815; Bio-Techne R&D Systems) following the manufacturer's instructions. The IL-2 concentrations were measured every day for 5 days in the culture supernatant of eight food-allergic dogs stimulated with between five and seven food allergens, including both positive and negative stimulations. Concentrations of IL-2 were likewise measured in the supernatant of PBMCs of three healthy client-owned dogs after co-culture with nine food allergens. As for the LPT, we expressed results as SIs of IL-2 concentrations after stimulation with food allergens over those cultured with saline on the same day.

Statistical analysis

The IL-2 supernatant concentrations were compared over time by the nonparametric, repeated-measures Friedman ANOVA with Dunn's multiple comparison test. The correlation between SIs of cell proliferation and IL-2 concentrations was calculated using the Spearman test. Stimulation indices obtained with the positive control concanavalin A were compared between healthy and food-allergic dogs using the Wilcoxon–Mann–Whitney U-test. All calculations were done using Prism 10 (GraphPad Software) with a significance level set at 0.05.

Patient characteristics

Table S1 includes details on the signalment, clinical manifestations, suspected allergens and TTF of the 28 dogs with delayed FA.

These dogs belonged to different breeds, and six of them (golden retriever, Labrador retriever, Labrador retriever poodle crossbred, Border collie, French bulldog, American Staffordshire terrier) had two patients included in this cohort. Their age varied from 6 months to 12 years (average 3.5, median 3.0 years), and the female-to-male ratio was 0.75 (Table S1).

Food-induced atopic dermatitis (FIAD) and signs of gastro-enteropathy (e.g. vomiting and diarrhoea) were the main clinical manifestations of delayed FA, as they were seen in 18 of the 28 dogs (64%) each (Table S1). Less commonly, dogs exhibited pruritus without visible lesions (five dogs; 18%), urticaria (two; 7%), angioedema (two; 7%), conjunctivitis or interdigital fistulae (one dog each; 4%). Seventeen dogs (61%) presented with two or more of these clinical manifestations (Table S1).

After performing OFC, veterinary surgeons identified 66 single food ingredients and two commercial diets inducing flares ≥6 h after ingestion in their patients (i.e. 68 different food sources). The number of culprit food sources ranged from one to seven (average 2.0, median 2.4). Nine dogs (32%) only exhibited a delayed flare after eating one food item, while four others had a recurrence of signs after eating five or more ingredients. In descending order of frequencies, the main culprit food sources during OFC were chicken (16 dogs; 57%), beef (10; 36%), turkey, pork and horse (4 each; 14%), and lamb, rabbit, rice and corn (3 each; 11%). Of interest is that one dog exhibited signs after eating peanuts, a rarely reported event.

According to our selection criteria, all dogs with delayed FA exhibited signs ≥6 h after OFC. In four dogs (numbers 6, 11, 12 and 14), some food ingredients caused a delayed flare of signs (≥6 h), while others also induced a relapse before that timeframe (Table S1). Altogether, delayed reactions after OFC were observed within 6–12 h in 3 dogs (11%), between 12 and 24 h in 8 dogs (29%) and beyond 24 h in the remaining 17 (61%) (Table S1).

LPT controls

The co-culture of the cells for 5 days with the three negative controls did not lead to SIs >1.9 in any dog. By contrast, the positive control concanavalin A led to SIs ranging from 7.9 to 18.9 in the 35 healthy laboratory and client-owned dogs (average 12.3, median 11.0) and between 6.7 and 19.7 (average 11.3, median 10.5) in the 28 dogs with FA. SIs after co-culture with concanavalin A were not significantly different between these two groups of dogs (Wilcoxon–Mann–Whitney U-test; p = 0.1541).

Performance of the LPT in healthy dogs

The LPT was first performed using a set of 23 food extracts plus one milk component (Bos d 7, bovine IgG) in the 10 healthy laboratory beagles, all females aged from 4.5 to 7.5 years (average 6.5, median 6.7 years). These dogs all had eaten the same chicken-and-rice diet for the preceding 3 years. In these dogs, the lowest of the 240 possible SIs was 1.0, and the highest was 2.7 (average 1.9, median 2.0). In fact, most SIs were tightly grouped with a 95% confidence interval between 1.9 and 2.0. Importantly, the LPT had SI values <2.7 for all ingredients reported on the beagle's diet label (chicken, rice, wheat, corn, soy, beetroot and brewers' yeast). Following the methodology employed for the LPT in human subjects,¹⁵ we established the positivity threshold (i.e. the cut-off value) for the canine test as 3.0, the value corresponding to the mean plus three standard deviations of all 240 SIs obtained in laboratory beagles.

The LPT was subsequently performed in 25 pet dogs coming to their veterinary surgeon for routine interventions using 28 different food extracts plus Bos d 7. In these dogs, the SIs ranged from 0.8 to 2.9 (average and median 1.7); none of the SIs was higher than the proposed threshold of 3.0 for any of the tested food allergens.

Performance of the LPT in dogs with delayed food allergy

Finally, the 28 dogs with delayed FA were tested with the LPT. The number of allergens used for stimulation varied among dogs, as the number of co-cultures performed varied to better match food items inducing flares during OFC (range 13–30, average 26, median 27).

In contrast to healthy beagles and client-owned pets, all 28 dogs with delayed FA had at least one positive LPT with an SI ≥3.0. The number of food items with an SI above the threshold ranged from 2 to 10 (average 5.1, median 5.0). The LPT positivity rates to the various ingredients are shown in Figure 1; the five main food extracts inducing a positive LPT were chicken and turkey (both 46%), beef (39%) as well as corn (38%) and duck (36%).

In 10 dogs, allergens were tested in duplicates (291 values). The coefficients of variation (CV%) between duplicate cell counts on D5 ranged from 0.5% to 20.4% (average 4.8%, median 4.3%); overall, there was little variation between duplicate cell counts, as only 14 of 291 allergen pairs (4.8%) had a CV% >10%.

Despite the limitations of a relatively small dataset, this LPT reliably discriminated between healthy dogs and those with delayed FA and demonstrated excellent sensitivity and specificity.

Correspondence between OFC and LPT results

Altogether, 57 food ingredients with a positive LPT matched some of the 68 food sources identified as causing a delayed flare (≥6 h) after OFC (84%) (Table S1). The agreement between the OFC and LPT was full (i.e. all culprit food ingredients causing a late flare had a corresponding positive LPT) in 18 of the 28 dogs (64%); it was partial (i.e. only some of the culprit ingredients were positive in the LPT) in the other 10 (36%) (Table S1). In each one of these 28 food-allergic dogs, the LPT also was positive for food extracts for which an OFC had not been done, so the clinical relevance of these results was not assessable.

IL-2 serum concentrations in the co-culture supernatant

In the three healthy dogs whose PBMCs were cultured with nine allergens over 5 days, SIs (i.e. fold-changes) of IL-2 concentrations over those cultured without allergens reached a maximum of 2.4 on D4 and D5. The co-cultures of cells of dogs with delayed food allergies with allergens inducing a negative LPT (SI < 3.0) had a maximal IL-2 SI of 2.9 on the same final 2 days. By contrast, as early as D1, the IL-2 SIs from co-cultures of cells from allergic dogs with food allergens inducing a positive LPT were >3.0 for eight of 28 allergens (29%); by D5, all 28 offending food allergens had IL-2 SIs ranging from 3.5 to 24.8 (average 7.6, median 6.3) (Figure 2a). Altogether, the IL-2 SIs were significantly different over time (Friedman ANOVA, p < 0.0001), being significantly higher each day compared to the preceding one, except for between D4 and D5 when the IL-2 concentrations had stabilised.

Finally, the correlation between both positive and negative LPT and IL-2 SIs was highly significant (Spearman r = 0.79; p < 0.0001), suggesting that both cell proliferation and IL-2 release were likely to be the consequences of the same allergen-driven stimulation of T cells (Figure 2b).

Food allergen-specific IgE

In three dogs (1, 6 and 20), the PAX revealed elevated concentrations (≥28.0 ng/mL) of IgE against food extracts or components of the same food source that triggered a late flare during OFC. Because these dogs also had a positive LPT, they were thus suspected of having a combination of IgE- and cell-mediated FA (i.e. a mixed FA; Table S1). In three other dogs (10, 11 and 21), the PAX was positive for allergens expected to IgE-cross-react with food items that induced a flare during OFC (Table S1); these also could represent examples of mixed FA. In dogs 4, 8, 16 and 22, the PAX detected IgE against some food allergens for which an OFC had not been performed, and the clinical relevance of such sensitisation was thus unknown (Table S1).