Patterns of allergic sensitization in adults with severe asthma: the ATLAS non-interventional study

Abstract Objectives Severe asthma is heterogeneous, with childhood-onset asthma believed more likely to be allergic, whereas adult-onset asthma is considered typically non-allergic. However, the allergic diagnosis is typically by exclusion: if patients do not react to an allergen panel, which is not standardized and often limited to few allergens, they are considered non-allergic. The overall aim of the ATLAS study was to characterize the sensitization to allergens in severe asthma (independent of phenotype). Methods Single-visit, cross-sectional, non-interventional study in adults with severe asthma. Analyses were conducted for total and specific immunoglobulin E against 53 allergens, overall and in subgroups, including age at asthma onset (<20 [childhood-onset] and >40 years of age). Results Among 1010 recruited patients, 28.4% reported childhood-onset asthma and 33.6% onset >40 years of age. After excluding patients receiving omalizumab/anti-IL5 therapy, 27.6% were not sensitized to any tested allergens, whereas 19.1% were sensitized to >10 allergens. All allergens triggered sensitization in some patients. Baseline characteristics in the two onset subgroups were similar; 23.2% with childhood-onset asthma were not sensitized to any allergen, compared to 32.0% with onset >40 years of age. Conclusion When a broad panel of allergens is used for sensitization testing, as many as three quarters of patients with severe asthma display sensitivity to at least one allergen, with substantial overlaps in all characteristics between the two age-at-onset subgroups. All of the tested allergens triggered a response in at least some patients, emphasizing the importance of including a broad range of allergens in any testing panel.


Introduction
Although patients with severe asthma only comprise between 5% and 10% of the adult asthma population (1)(2)(3)(4), these patients drive much of the overall cost of asthma management (5,6), and so being able to target effective therapies to an individual patient's phenotype is important.Severe asthma is not a single phenotype, but is a heterogeneous disease (1), with a number of different phenotypes having been identified.Childhood-onset asthma is believed to be more likely to be allergic (i.e., immunoglobulin E [IgE] mediated), whereas in adult-onset asthma IgE-mediated sensitization is often regarded either to be absent or to play a lesser role (7).This distinction has become more relevant with the availability of monoclonal antibodies, since omalizumab is indicated for patients with severe allergic asthma (7), whereas the anti-interleukin-5 (IL5) and IL5 receptor alpha monoclonal antibodies mepolizumab, benralizumab and reslizumab are indicated for non-allergic asthma, characterized by raised blood eosinophil cell counts (8).However, there is substantial overlap in these phenotypes (with the allergic response in asthma involving both cross-linking of mast cell-bound IgE and eosinophil recruitment (9)).Indeed, eosinophil counts alone are not effective in distinguishing between allergic and non-allergic asthma: in one study over 50% of adults with severe allergic asthma being treated with omalizumab had blood eosinophil counts ≥300 cells/ µL, yet omalizumab was similarly effective in the high and low eosinophil groups (10).
One challenge is that the diagnosis of allergic asthma is typically one of exclusion-if a patient has severe asthma and does not react to a panel of allergens, then they are considered to have severe, non-allergic asthma.This is an issue, as in an analysis of patients being managed in clinical practice in Germany as having non-allergic asthma (the IDENTIFY project), 43.6% subsequently tested positive to at least one of 35 perennial allergens (11).In the current ATLAS study, we sought to extend the results of IDENTIFY by recruiting adults with severe asthma, independently of phenotype, and applying a broader panel of 53 allergens, both perennial and seasonal (plus a food screen).The overall aim of ATLAS was to characterize the spectrum of sensitization to allergens in patients with severe asthma recruited throughout Germany.

Methods
ATLAS was a single-visit, cross-sectional study conducted in private practices by pulmonologists in Germany, and was non-interventional in terms of the medications that patients received.Demographics and disease characteristics were captured in a questionnaire, along with details on possible allergen exposure, and serum and blood samples were taken.The serum samples were sent to a central laboratory for processing (the Institute of Molecular and Clinical Immunology at the Otto-von-Guericke University, Magdeburg, Germany), with total and specific IgE analyzed using the Phadia TM 250 system (ThermoFisher Scientific).The list of allergens tested is in Supplementary Table 1, with patients considered sensitized to an allergen if they had CAP > 0. In addition to testing for seasonal and perennial allergens, a combined test for specific IgE toward the most frequent food allergens was applied (egg protein, milk protein, fish, peanut, soya, and wheat protein).Automated differential blood counts were performed at the Institute of Clinical Chemistry and Pathobiochemistry (Otto-von-Guericke University Magdeburg, Germany).
Recruited patients were aged 18-80 years, male or female, with no restriction on body mass index (BMI), and had a diagnosis of severe asthma, defined as Global Initiative for Asthma (GINA) Stage 4 or 5 (12).Patients were excluded only if they had a diagnosis of chronic obstructive pulmonary disease, then or a connective tissue or autoimmune disease.Physicians were asked not to recruit patients currently receiving omalizumab or one of the anti-IL5 treatments (the study was conducted prior to the approval of the anti-IL4/IL13 monoclonal antibody dupilumab in Germany).All patients provided written informed consent prior to any study-related procedure.
The study was approved by the ethics committee of the Medical Faculty of the Otto-von-Guericke University Magdeburg, and was performed in accordance with the principles of the Declaration of Helsinki, and the International Conference on Harmonization notes for guidance on Good Clinical Practice (ICH/CPMP/135/95).

Outcomes
The analyses described in this manuscript sought to answer the following: • What proportion of adults with severe asthma has any sensitization to allergens (both common and less common)?• Are there any differences in blood eosinophil counts between those without IgE-mediated sensitization and those with ≥1 sensitization?• What proportion of adult-onset severe asthma shows an IgE-mediated sensitization, and does this differ from childhood-onset asthma?• Is there an association between total IgE concentration and the presence of sensitization? • Are there any regional differences in IgE-mediated sensitization, in terms of city vs. countryside residential location?

Sample size and statistical methods
The study was not formally powered.It was estimated that 1000 patients distributed throughout Germany would provide a representative sample of the German severe asthma population, and that approximately 500 pulmonologists in private practice would be needed to recruit this number of patients.Statistical evaluations were performed using IBM SPSS.Descriptive analyses were performed overall and in a series of subgroups, with data presented as number (percent) or mean (standard deviation).Chi-square tests, mean value comparisons, rank tests and correlation and regression analyses were used for statistical inference, with significance level set at 0.05.

Participants
The study was conducted between January 2017 and December 2019.Of 1026 patients recruited, 16 did not have any information recorded.The demographics and disease characteristics of the remaining 1010 patients are presented in Table 1.Of note, 80% of patients had asthma that was not fully controlled, 69% had experienced ≥2 exacerbations in the previous 12 months, and 38.8% were using maintenance oral corticosteroid (OCS).Although the physicians were asked not to recruit patients receiving omalizumab or anti-IL5 treatments, 90 of the recruited patients were receiving omalizumab and 112 were receiving anti-IL5 antibodies.
Approximately, two-thirds of the patients were female; compared to the male participants, females were slightly more likely to be obese (BMI ≥30 kg/ m 2 ) and to have more severe asthma (GINA 5 and ≥3 exacerbations in the previous 12 months).They were also more likely to have a history of allergic rhinitis or dermatitis, and slightly less likely to have a history of nasal polyps.When split by age at asthma onset, 28.4% reported onset in childhood (which we defined as <20 years of age).In comparison to the 33.6% of participants with onset >40 years of age, the childhood onset group was slightly more likely to have less severe asthma (physician-reported GINA 4).However, this group also had a higher proportion of patients with uncontrolled asthma and with ≥3 exacerbations in the previous 12 months, and was more likely to receive leukotriene receptor antagonists or theophylline.

IgE sensitization
Excluding the patients receiving omalizumab or an anti-IL5 therapy, the population mean ± SD IgE was 447 ± 808.2 (range <2, >5000) kU/L, with blood eosinophil count 315 ± 496.1 (range 0, 11, and 500) cells/ µL (Table 2).The most common allergens in this group were pollens-specifically birch, Timothy grass and rye (Table 3).However, at least some of the patients were sensitized to each of the allergens-with the least common (to duck feathers) triggering a sensitization in 8 (1.0%; Supplementary Table 1).A total of 27.6% of the patients were not sensitized to any of the tested allergens, whereas 19.1% were sensitized   to >10 allergens.The mean blood eosinophil count was 283 cells/µL in patients without sensitization (i.e., all CAP 0), compared to 326 cells/µL in those with ≥1 sensitization (Table 2).Furthermore, when sub-grouped by concomitant maintenance OCS use, 28.4% of the patients receiving OCS had no sensitization, compared with 27.1% that did not receive OCS.When patients not receiving omalizumab or an anti-IL5 therapy were analyzed by age at asthma onset, results in the <20 years of age subgroup were similar to the overall population.However, the mean blood eosinophil count in this group was lower than the overall population, and the count was also lower than the overall population in those with no sensitization and ≥1 sensitization.In the >40 years of age subgroup, all three mean eosinophil counts were higher than in the overall population, and the most common allergen was Staphylococcus aureus (Enterotoxin toxic shock syndrome toxin [TSST]), followed by birch and Candida albicans, with no patients sensitized to duck feathers.In this subgroup, 32.0% were not sensitized to any of the tested allergens.
As would be expected from the mode of action, patients receiving an anti-IL5 had lower mean eosinophil values compared to patients not receiving omalizumab or an anti-IL5 therapy.In this subgroup, 29.8% of patients were sensitized to the most common allergen (birch), with 31.9% not sensitized to any of the tested allergens and 18.1% sensitized to >10 allergens.

Association between total IgE and sensitization
The total IgE concentration was statistically significantly lower in patients who were not sensitized to any of the tested allergens than in those sensitized to ≥1 allergen (mean [SD] 95.1 (170.70) vs. 581.4[909.52]kU/L; p < 0.001).When analyzed by IgE concentration categories, 96.9% of patients with a high value (>500 kU/L) were sensitized to at least one of the tested allergens (Figure 1).However, 48.3% of the patients in the lowest category (<100 kU/L) were also sensitized to at least one allergen.

Regional differences
When divided by residential location (city vs. countryside) there were few differences in the overall demographic or disease characteristics (Table 1).Patients residing in a city were slightly less likely to be obese (BMI ≥30 kg/m 2 ), and were more likely to have asthma that was fully controlled and less dependent on maintenance OCS compared to patients residing in the countryside, but were also more likely to have more severe asthma (GINA 5).In terms of the comorbidity profile, city-based patients were slightly less likely to have nasal polyps or allergic dermatitis, but were more likely to have allergic rhinitis.The mean blood eosinophil count was higher in city-based patients, when analyzed overall and in the two sensitization subgroups (Table 2).
The most common allergens were similar in the two residential location subgroups, as were the proportions of patients with no sensitization and with >10 sensitization, and the similar proportion of patients in the two subgroups were sensitized to storage mites (Table 3).All tested allergens triggered a response in some patients (Supplementary Table 1).

Discussion
Overall, our analyses clearly demonstrate that the 'pure' non-allergic asthma phenotype is relatively uncommon (only 27.6% of the population not receiving omalizumab or an anti-IL5 therapy were not sensitized to any of the tested allergens), and that sensitization is frequent in all subgroups.Note: Data are for the number (%) of patients with CaP >0.# is position in top 10 for each allergen, with na indicating that the allergen is not in the top 10 for that group.rows are sorted by percentage of patients in the overall population.Il, interleukin; tsst, toxic shock syndrome toxin.
The study population that we recruited was representative of patients with severe asthma, given they were recruited throughout Germany.Their characteristics are typical of a severe asthma population, in that they were predominantly female, few had well-controlled asthma, and more than half experienced ≥2 exacerbations in the previous year.There was high variability in blood eosinophil count and IgE concentration, as indicated by the wide SDs and ranges, and approximately three quarters of the patients were sensitized to ≥1 of the tested allergens, with nearly 20% sensitized to >10.The most common sensitization was to pollens, but all allergens triggered a response in some of the patients.
There was a statistically significant association between total IgE concentration and the likelihood of sensitization, with almost all of the patients who had concentrations >500 kU/L sensitized to >1 of the tested allergens.However, in the lowest IgE category, approximately half of the patients were sensitized to >1 tested allergen.This suggests that whilst a high total IgE value is a good positive predictor of sensitization, a low value cannot be taken as a surrogate for a lack of sensitization to a specific allergen.
Childhood-onset asthma is typically considered to be allergic asthma (i.e., IgE-mediated), whereas adult-onset asthma is typically considered to be non-allergic (7).Although in our analyses the mean IgE concentration was slightly higher and the mean blood eosinophil count slightly lower in those with onset <20 years of age than in those with onset >40 years of age, there was substantial overlap between age-onset subgroups in both parameters.In addition, the proportion of patients with no sensitization was similar in the two populations (23.2% and 32.0% in <20 and >40 years, respectively).However, a notable difference between these two subgroups was in the most common allergens.Whereas the three most common allergens in the group with onset <20 years of age was the same as in the overall population (Timothy grass, rye and birch), this was not the case for the group with onset >40 years of age.In this latter group, the most common allergen was S. aureus (Enterotoxin TSST), followed by birch and C. albicans-with overall a smaller proportion of patients testing positive to each allergen than in the group with onset <20 years of age.This emphasizes the importance of either using a broad panel of allergens for sensitization testing, or taking a careful patient history to identify likely allergens if it is only possible to use a narrower panel.These results are consistent with those of the ESSAy project, which evaluated allergic sensitization in a group of German adults with severe, uncontrolled asthma (13).Even though most had been diagnosed with asthma in adulthood, 72.2% were sensitized to at least one of the tested allergens, including S. aureus enterotoxins.
In the previous IDENTIFY project, S. aureus enterotoxins were also among the most common allergens, and patients who were sensitized to S. aureus were more likely to also be sensitized to other allergens than those who were not sensitized to S. aureus (11,14).These data support the role of S. aureus enterotoxins as "superantigens" that are potentially involved in the pathophysiology of patients with severe asthma (15).The belief that patients who are newly diagnosed with asthma in adulthood are more likely to have a nonallergic phenotype may be because the sensitization testing does not include S. aureus enterotoxins.Alternatively, patients could develop upper airway involvement later in life due to S. aureus colonization during the course of the disease.If this would be the case, then it would be worth considering repeat allergen testing even if an individual has previously tested negative.Indeed, in a previous observational study the proportion of patients with asthma who were sensitized to an allergen (in that case Aspergillus fumigatus) increased over a mean 19.1 year follow-up period, from 9% at baseline to 31% (16).In contrast to IDENTIFY, which only tested sensitivity to perennial aeroallergens (consistent with the indication for omalizumab), we included both perennial and seasonal (pollen) allergens.Our rationale for including pollens (in addition to perennial allergens) is that the aim of ATLAS was to characterize the spectrum of sensitization in patients with severe asthma.This meant that the panel should be as broad as possible, and should comprise all relevant allergens-including pollens, which are known to be involved in the pathogenesis of asthma (17).
In the analysis of residential location, there were few differences between the two groups of patients.In particular, the sensitization patterns were similar in the two groups-most notably for the proportion of patients who are sensitized to storage mites.Since storage mites are typically found on poorly stored food and grains, this sensitization could be more relevant to those who live in the country (18).However, in a previous analysis of data from northern Europe, storage mite sensitization was as common as sensitization to house dust mites, and was not associated with place of upbringing (farm vs. village vs. city) (19), so consistent with our results.
Given investigators were asked not to recruit patients receiving omalizumab or an anti-IL5 monoclonal antibody, it is surprising that 9% of the recruited patients had omalizumab listed as an asthma medication, with 11% listing an anti-IL5.It is possible that these were prior medication (i.e., that the patients were no longer receiving at the time of the visit).However, it means that the population recruited provides no information on the use of biologics for the management of severe asthma in Germany.As a conservative approach we excluded these patients from our detailed analyses of blood eosinophil count and IgE-mediated sensitization.However, since they had been recruited, we decided to evaluate sensitization in the patients receiving an anti-IL5 monoclonal antibody (but not omalizumab).Interestingly, the IgE concentration was slightly higher in this population than in the overall population.The blood eosinophil count was lower, although with a wide range-and given these therapies have been shown to dramatically reduce eosinophil counts the relatively high counts in some patients support the theory that this was prior and not current medication in at least some (highlighting one of the challenges of conducting 'real world' studies with potentially inexperienced investigators).It is notable, however, that in these 94 patients (who were presumably being managed by their physicians as having non-allergic asthma) sensitization to the broad panel of allergens was common, with only 31.9% not sensitized to any allergen.
The main limitations of the study are associated with the non-interventional status.We did not perform skin prick or provocation tests to confirm functional reactivity to allergens, and so we cannot comment on the clinical relevance of the sensitization (sensitization does not necessarily mean an individual will experience symptoms on exposure to the positively tested allergens).Furthermore, eosinophil counts were assessed on a single occasion-although the use of a central laboratory helped to minimize variability.In addition, regional differences are based only on self-declared residence, and do not take into account the duration of residence, or location of work/school.Finally, given all of patients were recruited in Germany, the results cannot necessarily be extrapolated to other countries.

Conclusion
When a broad panel of allergens is used for sensitization testing, as many as three quarters of patients with severe asthma display sensitivity to at least one allergen.As might be expected, patients who are sensitized to allergens have higher IgE concentrations than those who are not sensitized; however, eosinophil counts were also higher in these patients, with substantial overlaps in all characteristics between the two ages at onset subgroups, clearly demonstrating the substantial overlap between the allergic and non-allergic severe asthma phenotypes.
The overall spectrum of sensitization was remarkably broad.Although the most common allergens were pollens, all of the tested allergens triggered a response in at least some of the recruited patients.This emphasizes the importance of including a broad range of allergens (informed by the patient's history and exposure) in any testing panel for patients with severe asthma.

Figure 1 .
Figure 1.association between immunoglobulin E concentration category and sensitization in patients not receiving omalizumab or an anti-Il-5 therapy.

Table 1 .
Demographics and disease characteristics.

Table 2 .
Immunoglobulin E concentration, and blood eosinophil counts.