Validation and further insight into the International Severe Asthma Registry (ISAR) eosinophil gradient algorithm in the Wessex AsThma CoHort of difficult asthma (WATCH) using historical blood eosinophil counts and induced sputum

CHI3L1/YKL- 40 in serum of patients with normal sputum eosino phil levels, 9 and CHI3L1/YKL- 40 was also identified in our previous severe asthma cohort as a sub- phenotype in severe asthma. 1 Thus, monitoring a broad spectrum of airway inflammatory markers in asthma should be an important future consideration in addition to the use of classifications like the ISAR algorithm. In conclusion, though the term “non- eosinophilic asthma” is used to describe severe asthma patients without current evidence of raised eosinophils while on high dose steroids. It is imprecise: many of these patients will have an underlying eosinophilic phenotype. Nevertheless, though, they may underestimate the presence of an eosinophilic phenotype. It is important to reiterate that contempo raneous blood and sputum measurements are proven theragnostic biomarkers, predicting response to anti IL- 5 and steroid treatment for reducing asthma exacerbations.


Validation and further insight into the International Severe Asthma Registry (ISAR) eosinophil gradient algorithm in the Wessex AsThma CoHort of difficult asthma (WATCH) using historical blood eosinophil counts and induced sputum
To the Editor, Severe asthma is a heterogeneous disease comprising numerous endophenotypes. 1 In recent years, there has been increased focus on eosinophils and type 2 (T2) inflammatory pathways as treatable traits in severe asthma, as these can be effectively targeted by an emerging portfolio of biologic therapies. 2 Traditionally, the presence of increased eosinophils in induced sputum was used to define eosinophilic asthma. Sputum induction and analysis, however, require specialist expertise, preventing its use in routine clinical care. Consequently, the classification of severe eosinophilic asthma by blood eosinophil status has become commonplace in clinical practice.
There are, however, limitations in defining patients as eosinophilic or not, as severe asthma patients have high oral or inhaled corticosteroid burden which may suppress blood eosinophils and mask the underlying eosinophilic nature of the disease. Particularly if this trait is defined by limited timespan assessment in this variable disease. 1,3,4 Consistent with this, we recently reported that repeated blood eosinophil counts (BECs) over time can help unmask the underlying T2 asthma status not revealed by a single snapshot measure. 3 This longitudinal approach demonstrated a higher burden of T2 asthma status than previously reported. Thus, rigorous approaches are needed to correctly define non-eosinophilic and eosinophilic forms of severe asthma to accurately guide clinical management. The ISAR eosinophil gradient algorithm report identified two points that merit further understanding and which could be addressed by our WATCH cohort data: firstly, how longitudinal BEC assessment might consolidate these algorithm-derived phenotypes; secondly, how the identification of eosinophilic asthma via the ISAR algorithm aligns with the traditional method of sputum analysis.

| ME THODS
Wessex AsThma CoHort of difficult asthma is a real-world study of patients with difficult-to-treat/severe asthma attending a tertiary centre (Southampton, United Kingdom). The study had national research ethics committee (REC) approval (reference 14/WM/1226), and all patients provided written informed consent. The WATCH study protocol and methodology has been described previously. 6 The WATCH dataset includes current and historic BEC data plus induced sputum data in a subset of patients with severe asthma.
Sputum was induced in poorly controlled patients who were at least 4 weeks clear of an exacerbation. We applied the ISAR algorithm to the WATCH population to explore its generalizability, using both a single BEC taken at enrolment and, where available, observations of BECs up to 15 years prior. Additionally, we correlated induced sputum granulocyte phenotypes with the ISAR-defined phenotype in a subset of WATCH patients. and ISAR was found when a single snapshot BEC measure was used: though 70% of patients considered at least "likely eosinophilic" (Grade 2), only 45% of the population being considered "most likely eosinophilic" (Grade 3). Progressively greater alignment in the proportions of ISAR Grades was found by increasing the retrospective reach of BEC observation in the WATCH cohort (Table 1) Table 2). Only 1 patient (0.8%) was classed as Grade 0; this patient presented with a pauci granular sputum phenotype ( Table 2). When sputum eosinophilia (≥2% eosinophils) alone is used to define eosinophilic asthma, less than half of patients are found to be eosinophilic (36% patients with ≥2% sputum eosinophils and 12% with mixed granular sputum [≥2% eosinophils, ≥61% sputum neutrophils]). The pauci granular population had fewer patients reported to have CT scan evidence of bronchiectasis than those with sputum granular disease; however, no significant difference in the presence of bronchiectasis was identified between sputum neutrophilic and sputum eosinophilic patients. 46% of patients classified as Grade 3 did not have evidence of sputum eosinophilia.

| DISCUSS ION
We validated the ISAR eosinophil gradient algorithm in the WATCH cohort and demonstrated that extended longitudinal BEC monitoring within that framework increases the probability of identifying Grade 3 patients. It is notable that in both ISAR and WATCH datasets, a very high percentage of subjects were Grade 3 "most likely eosinophilic" using this algorithm. We previously showed remarkably similar overwhelming prevalence of underlying eosinophilic status in WATCH using an alternative perspective of longitudinal BEC monitoring. While there is potential selection bias with that latter approach (given increased propensity to perform blood counts when patients are exacerbating), the ISAR algorithm corroborates that observation by including clinical characteristics, which is particularly relevant when longitudinal BEC data are limited. When present, however, the inclusion of multiple historic BECs, mitigates against granulocyte count instability and treatment effects, in the detection of an underlying eosinophilic phenotype. Our WATCH data thus support Heaney et al's findings and those of others, 7 that severe asthma is mostly an eosinophilic disease. 5 It is worth noting that while the terms eosinophilic and T2 asthma are often applied interchangeably, the ISAR algorithm does not incorporate measures of atopic predisposition such as total IgE or specific allergen sensitization. Since allergy is a T2-associated process that omission might raise concerns that the ISAR algorithm could miss a proportion of T2 patients. However, the ISAR algorithm still finds overwhelming prevalence of eosinophilic status. As no differences were identified in eczema, atopy or total IgE between patients considered Grade 2 and above vs. those considered Grade 0 these definitions cannot be considered interchangeable using the ISAR algorithm in the WATCH cohort.  1 (0.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (100.0%) Note: ISAR grade based on highest blood eosinophil count ever.