Development of allergic sensitization and its relevance to paediatric asthma

Ceyda Oksel PhD, Adnan Custovic MD PhD*

Department of Paediatrics, Imperial College London, UK

Correspondence and requests for reprints:

Adnan Custovic MD PhDImperial College London, UK, Department of Paediatrics

St Mary’s Campus Medical School, Room 244Norfolk Place, London W2 1PG, UK

Tel: 020 7594 3274Fax: 020 7594 3984

Email: a.custovic@imperial.ac.uk

1

ABSTRACT

Purpose of review: The purpose of this review is to summarize the recent evidence on the distinct

atopic phenotypes and their relationship with childhood asthma. We start by considering definitions and

phenotypic classification of atopy and then review evidence on its association with asthma in children.

Recent findings: It is now well recognised that both asthma and atopy are complex entities

encompassing various different sub-groups that also differ in the way they interconnect. The lack of gold

standards for diagnostic markers of atopy and asthma further adds to the existing complexity over

diagnostic accuracy and definitions. Although recent statistical phenotyping studies contributed

significantly to our understanding of these heterogeneous disorders, translating these findings into

meaningful information and effective therapies requires further work on understanding underpinning

biological mechanisms.

Summary: The disaggregation of allergic sensitization may help predict how the allergic disease is likely

to progress. One of the important questions is how best to incorporate tests for the assessment of

allergic sensitisation into diagnostic algorithms for asthma, both in terms of confirming asthma

diagnosis, and the assessment of future risk.

Keywords: atopy, asthma, endotypes, disease progression, machine learning.

2

Introduction

Asthma is a complex disease characterized by a number of features, including inflammation in the

airways. This inflammation can be triggered by allergens (in atopic asthma) or non-allergic irritants (in

non-atopic asthma), but in most patients both allergic and non-allergic factors contribute to the clinical

presentation of the disease. Although allergic sensitization is a strong risk factor for asthma [1], not all

asthma is allergic in nature, and most sensitized individuals do not develop asthma or other allergic

diseases [2,3]. This complex interplay between allergic sensitization and asthma, and a series of

phenotyping studies that discriminated distinct sub-types of asthma [4,5**], eczema [6*] and atopy [7-9,

10*, 11*, 12*, 13-14], have provided support to the concept of heterogeneity of asthma and allergic

sensitization.

Heterogeneity of asthma

It is now accepted that asthma is a heterogeneous condition comprising multiple subgroups and

aetiological factors, and that it should be deconstructed into traits [15**]. Recognition of this

heterogeneity has led researchers away from a focus on a single biological marker or clinical test that

may be applicable to all patients with asthma, towards more personalised approaches that can address

specific challenges in individual patients, or groups of patients. Over the last two decades, a substantial

effort has been devoted to understanding asthma heterogeneity, and its variable clinical expression in

individuals [15**, 16, 17]. One approach to understanding disease heterogeneity involves integrating a

series of statistical models that account for the unobserved heterogeneity between individuals, with

mechanistic information about underlying pathophysiology [18*]. For example, the latent class analysis

relies on the assumption that observable characteristics are imperfect indicators of an underlying

(latent) construct [19]. Over the last two decades, many attempts have been made to identify

longitudinal trajectories of childhood wheeze [4,20-24], atopy [6*,8,10*,25,26] and asthma [27-29] by

3

means of latent class analysis. Such “phenotyping” studies have traditionally been based on clinical

characteristics that are mostly observed and recorded by physicians in the clinical/research settings.

However, features which are collected to ascertain asthma presence or severity may not be the most

informative for the discovery of true disease endotypes. A recent study has shown that a careful synergy

of data-driven methods and clinical interpretation may help better understand the heterogeneity of

asthma, and enable the discovery of true asthma endotypes [5**]. This study has identified allergic

sensitization as one of the four potentially important features for disaggregating childhood asthma, the

other three being the age of onset, asthma severity and recent exacerbations.

One key conclusion is that computational approaches can aid the interpretation of clinical and

healthcare data by identifying disease sub-groups, and their relationship to particular risk factors.

However, in the absence of mechanistic information on underlying pathophysiology, it is difficult to

identify whether or which of these sub-groups are clinically relevant. Understanding biological

mechanism underpinnings different asthma endotypes, and the effects of genetic and environmental

factors, is of critical importance to untangle the underlying pathophysiology, and is a key step to

advance personalised (or stratified) medicine in asthma [30]. However, to date, the identification of

asthma subgroups described in different studies, and how the current approach to “asthma

phenotyping” may lead to more effective and targeted asthma treatment strategies, has not been

translated to clinical setting [15**,18*,31*].

Heterogeneity of allergic sensitization and its relevance to asthma

Atopy can be defined as the genetic tendency to become sensitized to common allergens that most

people don’t react to, as a result of ordinary exposure. Atopic diseases cover a heterogeneous group of

symptoms and disorders, ranging from wheezing, coughing, breathlessness, hay fever and rhinitis, to

skin conditions such as atopic dermatitis (eczema, or atopic eczema), each of which might be triggered

4

by different factors. Different manifestations of atopic disease may co-exist in the same patient, or

develop at different times [32]. However, a history of atopy and confirmation of IgE-mediated

sensitization (e.g. using skin prick tests and/or measurement of specific serum IgE) does not necessarily

indicate the presence of an immunologically-mediated allergic reaction and/or allergic disease, since

sensitization may be asymptomatic [3].

Recent evidence has shown that similar to asthma, “atopic sensitization” is heterogeneous, and that

there are several distinct sub-groups of sensitization, differing in their risk factors, associations with

asthma progression and response to treatment [8,9,13,14,33-35**, 36*]. The disaggregation of

sensitization, and knowing which subtype a child belongs to, may help predict how the allergic disease is

likely to progress, and later-life asthma outcomes.

Patterns of sensitization were first investigated by Kurukulaaratchy et al. [7], who proposed an

investigator-defined phenotypical classification of childhood atopy based on the timing of development

of sensitization to common allergens (Early childhood atopy, Chronic childhood atopy and Delayed

childhood atopy). In an attempt to move from dichotomous (positive or negative, atopic or non-atopic)

and opinion-based classifiers, to date-driven classification of allergic sensitization, we applied machine

learning techniques to skin test and IgE data collected on multiple time points throughout childhood in a

population-based birth cohort [8]. We uncovered four different classes of sensitization, including

Predominantly dust mite, Non-dust mite, Multiple early and Multiple late sensitization, and have shown

distinct associations of these classes with asthma. We have subsequently replicated these findings in

another birth cohort [9], and have demonstrated that the association with asthma and diminished lung

function is strong for multiple early, but not other sensitisation classes [37]. In another study, Garden et

al. [14] described three latent atopic phenotypes using longitudinal data on skin prick sensitivity (Late

mixed inhalant sensitization, Mixed food and inhalant sensitization, and Dust mite monosensitization).

Havstad et al. [34] classified sensitized children at two years of age into three latent atopic sensitization

5

classes based on allergen-specific IgE patterns (Highly sensitized, Milk and egg-dominated sensitization,

and Peanut and inhalant/no milk sensitization). Stoltz et al. [13] analysed patterns of allergen-specific

sensitization based on the patterns of pet exposure, the presence of specific aeroallergen sensitization

and its quantitative degree, and described numerous different sensitization patterns (Any sensitization;

Mono-sensitization; Poly-sensitization; Cat; Dog; Dust-mite; Alternaria; Cockroach; Ragweed; Birch and

Grass). In another study [10*], atopic sensitization was classified into three phenotypes (Benign atopy,

Symptomatic atopy and Severe atopy). More recently, Dharma et al [38] revealed 4 distinct patterns of

allergic sensitization (Atopic dermatitis; Inhalant sensitization; Transient sensitization and Persistent

sensitization).

Our finding that the timing of onset and types of sensitization are the key discriminative factor for atopy

subtypes [8,9,33] has been independently confirmed by a recent study [11*], which reported the

existence of four distinct atopy phenotypes, differing in the time of onset and types of sensitising

allergens (Later sensitization to indoor allergens, Multiple early sensitization, Early sensitization to

outdoor allergens followed by indoor allergens, and Early sensitization to indoor allergens followed by

outdoor allergens). Similarly, Schoos et al [12*] explored the longitudinal patterns of atopic sensitization

in early childhood based on sIgE responses and revealed a total of 7 latent sensitization patterns:

Cat/dog/horse; Timothy grass/birch; Molds; House dust mites; Peanut/wheat flour/mugwort;

Peanut/soybean and Egg/milk/wheat flour. Although direct comparison between studies is difficult due

to differences in subject ascertainment, population structure and diagnostic criteria, there is increasing

evidence that the risk of asthma is significantly higher among children who are highly sensitized

[10*,34], sensitized early in life [8,9,33], or sensitized to multiple allergens [9,14]. Additionally,

sensitization to allergens from specific sources as dogs [13], cats or horses [12*] was reported to be

strongly associated with asthma development during childhood. Summary of cross-sectional and

6

longitudinal studies in children, which employed investigator-led and data-driven approaches to identify

distinct patterns of allergic sensitization, and their specific relationship with asthma, is shown in Table 1.

The number and type of “sensitization phenotypes” is driven by the nature of the data; however, while

different sensitization classes can be uncovered using a large amount of data collected over long periods

of time, this cannot be directly translated to a clinical situation, when a practicing physician sees a child

at any one time point. We urgently need better diagnostic markers and algorithms to help practicing

physicians differentiate between benign and clinically important allergic sensitization.

Diagnostic tests to ascertain allergic sensitisation and its relation to asthma

Skin tests and blood tests are commonly used to diagnose clinical allergies and sensitivities. However, in

the context of respiratory allergy, the interpretation of these commonly used allergy tests

remains arbitrary, since it traditionally relies on pre-defined cut-offs that have relatively poor ability to

distinguish between asymptomatic sensitizations and clinically relevant allergies, and does not take into

account either age, sex or ethnicity of the patient. A recent study has demonstrated that both age and

sex should be taken into account when interpreting the results of skin tests and sIgE measurement, and

that age- and sex-specific normative data are urgently needed [39**]. To increase the diagnostic

accuracy of current allergy tests in relation to the presence and persistence of asthma, the results

should be reported in a quantitative manner (e.g. the titre of IgE, or the size of skin test wheal

diameter). There is a clear need to develop better ways of interpreting these tests to identify patients

with clinically relevant allergies more precisely, for example by taking into consideration gender, age

and environmental exposure.

There is increasing evidence that sensitization to some, but not all allergenic proteins from different

sources is important for the expression and severity of asthma, and that assessing sensitization to

specific allergen components using component-resolved diagnostics (CRD) may be more informative

7

than standard tests using whole allergen extracts. We have shown that patterns of IgE responses to

multiple allergenic molecules have reasonable discrimination ability for childhood asthma and

rhinoconjuinctivitis [40]. In a recent publication, IgE reactivity to a limited number of allergen molecules

early in life identified children with a high risk of both asthma and rhinitis in adolescence in two birth

cohorts from the UK and Sweden [36*]. In Sweden, early-life IgE reactivity to four risk molecules (peanut

Ara h 1 , birch Bet v 1 , cat Fel d 1, and grass Phl p 1) predicted both incident and persistent asthma at 16

years, whilst in the UK similar association was observed for five allergenic molecules (dust mite

components Der p 1 and Der f 2, Phl p 1 and Phl p 5, and Fel d 1) [36*]. In a study using latent variable

modelling, we identified three cross-sectional clusters of IgE responses to multiple allergens from

different protein families in mid-school age, and each of these patterns was associated with different

magnitude of risk for having allergic airway symptoms [41]. Our subsequent study has indicated that

longitudinal trajectories of sensitization patterns to a limited number of grass and dust mite allergens

from age 5 to age 11 years had different associations with clinical outcomes, indicating that the time of

onset of specific patterns of IgE response may be critically important [33]. Similarly, Posa et al [42**]

have recently shown that IgE polysensitization to several dust mite molecules predicts current rhinitis,

and both current and future asthma. These data indicate that understanding the structure in the

developmental pathways of IgE responses to multiple allergenic molecules may facilitate development

of better diagnostic and prognostic algorithms for asthma. To address this, we have recently described

the architecture of the evolution of IgE responses to >100 allergen components from infancy to

adolescence [35**]. By applying novel machine learning techniques to CRD sensitization data

throughout childhood, we identified latent structure in the diversification of IgE responses (Figure 1),

and have shown that the timing of onset of specific patterns of sensitization may be one of the

important indicators of the subsequent risk of asthma and rhinitis [35**]. Furthermore, the results of

this study suggest that the latent structure of IgE allergenic component clusters is a reflection of the

8

source of allergens, the function of allergenic molecules, and a marker of the pathophysiological

processes leading to the development of distinct clinical presentations. Better resolution of longitudinal

patterns of sensitisation to multiple individual allergenic molecules may facilitate the development of

diagnostic algorithms, which can be used for the prediction of current and future risk. However, the

number and type of specific components which are relevant in different geographical areas, at different

ages, and for different allergic diseases, remains to be determined.

Diagnostic and prognostic test for asthma

Difficulties related to confirming the diagnosis of allergy is only a small part of a bigger diagnostic

problem in asthma. Figure 2 shows diagnostic aids for asthma in children and specific difficulties

associated with each step. The UK National Institute of Health and Care Excellence (NICE) guidance on

the diagnosis of asthma using objective tests does not recommend assessment of sensitization as an

objective test for asthma diagnosis. Rather, the NICE interim report proposes a diagnostic algorithm

which incorporates the sequential use of four measures of lung function and inflammation in children

with suggestive symptoms (spirometry, bronchodilator reversibility, fractional exhaled nitric oxide, and

peak flow variability; http://www.ngc.ac.uk/Guidelines/In-Development/). We have recently tested the

proposed algorithm amongst children aged 13–16 years, and found poor agreement between the

algorithm and asthma defined by physician diagnosis, presence of current symptoms, and regular use of

inhaled corticosteroids [43*]. These findings suggest that the proposed NICE algorithm for diagnosing

asthma should not be implemented in children until better evidence is available.

The ability to predict individuals who may become asthmatic in the long term is equally important to

prevent asthma, or minimise its impacts on many aspects of the lives of individuals. There has been a

significant amount of work carried out on developing predictive models that could be incorporated into

the decision making process [44-48], which has been systematically reviewed and critically evaluated

9

recently [49,50]. While the majority of existing asthma prediction models were developed based on

either clinical indices or regression equations [50], there is an increasing trend towards using machine

learning approaches to develop predictive models for asthma [51,52]. However, in the absence of

standardized asthma outcome definitions, diagnostic criteria for asthma, and independent validation

studies in different cohorts, the clinical value of the developed predictive models, and their applicability

to other populations and different healthcare settings, remains unclear. Moreover, the predictive

accuracy of asthma predictive indices and models decreases with the increasing prevalence of asthma in

the population studied [53], suggesting that although these tools may be useful in low-risk populations,

their prediction performance is still far from being satisfactory in clinical settings, especially for high-risk

patients. Overall, existing asthma risk prediction tools are not well suited for clinical use in their current

form, and need to be validated in different populations before their successful utilisation in daily

medical practice and decision-making.

One of the key questions going forward is how best to incorporate tests for the assessment of allergic

sensitisation into diagnostic algorithms for asthma, both in terms of confirming asthma diagnosis, and

the assessment of future risk (e.g. of asthma exacerbations, or disease persistence). It is possible that

such diagnostic algorithms may include not only the measurement of allergen-specific IgE, but also

allergen-specific IgG responses. In three birth cohort studies, we have shown that dissociation between

allergic airway symptoms and sIgE sensitisation in children is associated with a high-level coproduction

of sIgG1 [54**]. A consistent feature of children who were moderately or highly sensitized to inhalant

allergens, but who did not have allergic airway disease (asthma or rhinitis), as compared with equally

sensitized but symptomatic children, was an increased aeroallergen-specific IgG/IgE antibody ratio

[54**]. This association was strong and reproducible for sensitization-associated risk for both asthma

and rhinitis (in relation to sensitization against dust mite and grass, respectively). Furthermore, we

observed a similar relationship for asthma severity, and the risk of asthma exacerbations. These findings

10

of a sharp inverse relationship between dust mite-specific IgG1/IgE ratios to the presence, persistence

and severity of childhood wheezing and asthma, and of a similar inverse relationship between hay fever

and grass-specific IgG1/IgE ratios, suggest that sIgG1/sIgE ratio may be a potentially readily available

biomarker for the assessment of allergic airway diseases in childhood.

Taken together, the studies which we reviewed suggest that it may be possible to develop individualized

risk prediction algorithms for the diagnosis and prognosis of asthma, which should include novel

methods for assessing IgE sensitization status, likely in conjunction with a limited information on the

patterns and severity of symptoms, and other objective tests (e.g. lung function and measurement of

airway inflammation).

Conclusions

Novel computational approaches can aid disaggregation of complex phenotypes such as asthma and

allergic sensitisation, but in the absence of mechanistic knowledge, it is difficult to identify whether or

which of the discovered sub-groups are clinically relevant. The disaggregation of allergic sensitization,

and knowing which subtype a child belongs to, may help predict how the allergic disease is likely to

progress, and later-life asthma outcomes. One of the important questions going forward is how best to

incorporate tests for the assessment of allergic sensitisation into diagnostic algorithms for asthma, both

in terms of confirming asthma diagnosis, and the assessment of future risk.

Key points:

To increase the diagnostic accuracy of current allergy tests (skin tests and IgE to whole allergen

extracts) in relation to the presence and persistence of asthma, the results should be

interpreted in a quantitative manner by using the titre of IgE, or the size of skin test wheal

diameter; furthermore, age and sex of the patient should be taken into account.

11

We urgently need better diagnostic markers and algorithms to help practicing physicians

differentiate between benign and clinically important allergic sensitization.

Existing asthma risk prediction tools are not well suited for clinical use; forthcoming algorithms

for the diagnosis and prognosis of asthma may include novel ways of assessing IgE sensitization

status.

AcknowledgementsNone.

Financial support and sponsorshipCO is supported by the Wellcome Trust Strategic Award WT108818MA

Conflicts of interest The authors declare no perceived conflict of interest for the present review article.

References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: * of special interest ** of outstanding interest

12

1. Simpson B, Custovic A, Simpson A, et al. NAC Manchester Asthma and Allergy Study (NACMAAS): risk factors for asthma and allergic disorders in adults. Clinical & Experimental Allergy 2001, 31:391-399.

2. Kukhtinova NV, Kondyurina EG, Lentze MJ. Atopic and nonatopic asthma in children: two different diseases? International Journal of Biomedicine 2012, 2:214-221.

3. Custovic A, Lazic N, Simpson A. Pediatric asthma and development of atopy. Current opinion in allergy and clinical immunology 2013, 13:173-180.

4. Belgrave DC, Simpson A, Semic-Jusufagic A, et al. Joint modeling of parentally reported and physician-confirmed wheeze identifies children with persistent troublesome wheezing. Journal of Allergy and Clinical Immunology 2013, 132:575-583. e512.

**5. Deliu M, Yavuz TS, Sperrin M, et al. Features of asthma which provide meaningful insights for understanding the disease heterogeneity. Clin Exp Allergy 2018, 48:39-47.

This study has shown that a careful synergy of data-driven methods and clinical interpretation may help better understand the heterogeneity of asthma, and enable the discovery of true asthma endotypes. It has identified allergic sensitization as one of the four potentially important features for disaggregating childhood asthma, the other three being the age of onset, asthma severity and recent exacerbations.*6. Paternoster L, Savenije OE, Heron J, et al. Identification of atopic dermatitis subgroups in

children from 2 longitudinal birth cohorts. Journal of Allergy and Clinical Immunology 2017.

One of the first studies demonstrating heterogeneity of eczema using data-driven approaches.7. Kurukulaaratchy R, Matthews S, Arshad S. Defining childhood atopic phenotypes to

investigate the association of atopic sensitization with allergic disease. Allergy 2005, 60:1280-1286.

8. Simpson A, Tan VY, Winn J, et al. Beyond atopy: multiple patterns of sensitization in relation to asthma in a birth cohort study. American journal of respiratory and critical care medicine 2010, 181:1200-1206.

9. Lazic N, Roberts G, Custovic A, et al. Multiple atopy phenotypes and their associations with asthma: similar findings from two birth cohorts. Allergy 2013, 68:764-770.

*10. Hose AJ, Depner M, Illi S, et al. Latent class analysis reveals clinically relevant atopy phenotypes in 2 birth cohorts. Journal of Allergy and Clinical Immunology 2017, 139:1935-1945. e1912.

This study identified three different atopic sensitization classess:Benign atopy, Symptomatic atopy and Severe atopy.*11. Lee E, Lee SH, Kim Y-H, et al. Association of atopy phenotypes with new development of

asthma and bronchial hyperresponsiveness in school-aged children. Annals of Allergy, Asthma & Immunology 2017, 118:542-550. e541.

One of the recent studies focusing on distinguishing distinct phenotypes of allergic sensitization based on time of onset and types of allergen sources.*12. Schoos A-MM, Chawes BL, Melén E, et al. Sensitization trajectories in childhood revealed

by using a cluster analysis. Journal of Allergy and Clinical Immunology 2017.This study explores longitudinal patterns of atopic sensitization in early childhood based on sIgE responses.13. Stoltz DJ, Jackson DJ, Evans MD, et al. Specific patterns of allergic sensitization in early

childhood and asthma & rhinitis risk. Clinical & Experimental Allergy 2013, 43:233-241.14. Garden F, Simpson J, Marks G. Atopy phenotypes in the Childhood Asthma Prevention

Study (CAPS) cohort and the relationship with allergic disease. Clinical & Experimental Allergy 2013, 43:633-641.

**15. Pavord ID, Beasley R, Agusti A, et al. After asthma: redefining airways diseases. Lancet 2017.

13

Very important commission article providing a consensus expert view of where we are and where we need to go as a community of clinicians and researchers to tackle the problem of asthma.16. Anderson GP. Endotyping asthma: new insights into key pathogenic mechanisms in a

complex, heterogeneous disease. The Lancet 2008, 372:1107-1119.17. Wenzel SE. Phenotypes in asthma: useful guides for therapy, distinct biological processes,

or both? Edited by: Am Thoracic Soc; 2004. *18. Belgrave D, Henderson J, Simpson A, et al. Disaggregating asthma: Big investigation

versus big data. J Allergy Clin Immunol 2017, 139:400-407.This study highlights the importance of cross-disciplinary research and integrative approaches, whereby basic scientists, clinicians, data analysts, and epidemiologists work together to understand the heterogeneity of asthma.19. Howard R, Rattray M, Prosperi M, Custovic A. Distinguishing asthma phenotypes using

machine learning approaches. Current allergy and asthma reports 2015, 15:1-10.20. Spycher BD, Silverman M, Brooke AM, et al. Distinguishing phenotypes of childhood

wheeze and cough using latent class analysis. European Respiratory Journal 2008, 31:974-981.

21. Henderson J, Granell R, Heron J, et al. Associations of wheezing phenotypes in the first six years of life with atopy, lung function and airway responsiveness in mid childhood. Thorax 2008.

22. Chen Q, Just AC, Miller RL, et al. Using latent class growth analysis to identify childhood wheeze phenotypes in an urban birth cohort. Annals of Allergy, Asthma & Immunology 2012, 108:311-315. e311.

23. Savenije OE, Granell R, Caudri D, et al. Comparison of childhood wheezing phenotypes in 2 birth cohorts: ALSPAC and PIAMA. Journal of Allergy and Clinical Immunology 2011, 127:1505-1512. e1514.

24. Kurukulaaratchy R, Fenn M, Waterhouse L, et al. Characterization of wheezing phenotypes in the first 10 years of life. Clinical & Experimental Allergy 2003, 33:573-578.

25. Roduit C, Frei R, Depner M, et al. Phenotypes of Atopic Dermatitis Depending on the Timing of Onset and Progression in Childhood. JAMA pediatrics 2017.

26. Herr M, Just J, Nikasinovic L, et al. Risk factors and characteristics of respiratory and allergic phenotypes in early childhood. Journal of Allergy and Clinical Immunology 2012, 130:389-396. e384.

27. Weinmayr G, Keller F, Kleiner A, et al. Asthma phenotypes identified by latent class analysis in the ISAAC phase II Spain study. Clinical & Experimental Allergy 2013, 43:223-232.

28. Depner M, Fuchs O, Genuneit J, et al. Clinical and epidemiologic phenotypes of childhood asthma. American journal of respiratory and critical care medicine 2014, 189:129-138.

29. Boudier A, Curjuric I, Basagaña X, et al. Ten-year follow-up of cluster-based asthma phenotypes in adults. A pooled analysis of three cohorts. American journal of respiratory and critical care medicine 2013, 188:550-560.

30. Belgrave D, Simpson A, Custovic A. Challenges in interpreting wheeze phenotypes: the clinical implications of statistical learning techniques. Am J Respir Crit Care Med 2014, 189:121-123.

*31. Deliu M, Belgrave D, Sperrin M, et al. Asthma phenotypes in childhood. Expert review of clinical immunology 2017, 13:705-713.

A thorough review of data-driven methodologies and their implementation in paediatric asthma.32. Belgrave DC, Granell R, Simpson A, et al. Developmental profiles of eczema, wheeze, and

rhinitis: two population-based birth cohort studies. PLoS Med 2014, 11:e1001748.33. Custovic A, Sonntag HJ, Buchan IE, et al. Evolution pathways of IgE responses to grass

and mite allergens throughout childhood. J Allergy Clin Immunol 2015, 136:1645-1652 e1648.

14

34. Havstad S, Johnson CC, Kim H, et al. Atopic phenotypes identified with latent class analyses at age 2 years. Journal of Allergy and Clinical Immunology 2014, 134:722-727. e722.

**35. Howard R BD, Papastamoulis P, Simpson A, et al. Evolution of IgE responses to multiple allergen components throughout childhood. J Allergy Clin Immunol 2018.

Comprehensive description of the architecture of the developmental of IgE responses to >100 allergenic molecules from multiple sources.*36. Wickman M, Lupinek C, Andersson N, et al. Detection of IgE Reactivity to a Handful of

Allergen Molecules in Early Childhood Predicts Respiratory Allergy in Adolescence. EBioMedicine 2017, 26:91-99.

This study shows that IgE reactivity to only few allergen molecules in early childhood may help predict respiratory allergy in adolescence.37. Belgrave DC, Buchan I, Bishop C, et al. Trajectories of lung function during childhood.

American journal of respiratory and critical care medicine 2014, 189:1101-1109.38. Dharma C, Lefebvre D, Tran M, et al. Patterns of allergic sensitization and atopic dermatitis

from 1 to 3 years: Effects on allergic diseases. Clinical & Experimental Allergy 2017.**39. Mohammad HR, Belgrave D, Kopec Harding K, et al. Age, sex and the association

between skin test responses and IgE titres with asthma. Pediatr Allergy Immunol 2016, 27:313-319.

First study to highlight the importance of age and gender in interpretation of SPT and IgE test results.40. Prosperi MC, Belgrave D, Buchan I, et al. Challenges in interpreting allergen microarrays in

relation to clinical symptoms: a machine learning approach. Pediatr Allergy Immunol 2014, 25:71-79.

41. Simpson A, Lazic N, Belgrave DC, et al. Patterns of IgE responses to multiple allergen components and clinical symptoms at age 11 years. J Allergy Clin Immunol 2015, 136:1224-1231.

**42. Posa D, Perna S, Resch Y, et al. Evolution and predictive value of IgE responses toward a comprehensive panel of house dust mite allergens during the first 2 decades of life. J Allergy Clin Immunol 2017, 139:541-549 e548.

This recent study shows that IgE polysensitization to several dust mite molecules predicts current rhinitis, and both current and future asthma.*43. Murray C, Foden P, Lowe L, et al. Diagnosis of asthma in symptomatic children based on

measures of lung function: an analysis of data from a population-based birth cohort study. Lancet Child Adolesc Health 2017, 1:114-123.

A recent retrospective study testing the diagnostic algorithm proposed by UK National Institute of Health and Care Excellence (NICE) using data from a population-based cohort to simulate the proposed situation in clinical practice.44. Savenije OE, Kerkhof M, Koppelman GH, Postma DS. Predicting who will have asthma at

school age among preschool children. Journal of Allergy and Clinical Immunology 2012, 130:325-331.

45. Caudri D, Savenije OE, Smit HA, et al. Perinatal risk factors for wheezing phenotypes in the first 8 years of life. Clinical & Experimental Allergy 2013, 43:1395-1405.

46. Kurukulaaratchy R, Matthews S, Holgate S, Arshad S. Predicting persistent disease among children who wheeze during early life. European Respiratory Journal 2003, 22:767-771.

47. Castro-Rodríguez JA, Holberg CJ, Wright AL, Martinez FD. A clinical index to define risk of asthma in young children with recurrent wheezing. American journal of respiratory and critical care medicine 2000, 162:1403-1406.

48. Pescatore AM, Dogaru CM, Duembgen L, et al. A simple asthma prediction tool for preschool children with wheeze or cough. Journal of allergy and clinical immunology 2014, 133:111-118. e113.

15

49. Smit HA, Pinart M, Antó JM, et al. Childhood asthma prediction models: a systematic review. The Lancet Respiratory Medicine 2015, 3:973-984.

50. Luo G, Nkoy FL, Stone BL, et al. A systematic review of predictive models for asthma development in children. BMC medical informatics and decision making 2015, 15:99.

51. Chatzimichail E, Paraskakis E, Sitzimi M, Rigas A. An intelligent system approach for asthma prediction in symptomatic preschool children. Computational and mathematical methods in medicine 2013, 2013.

52. Finkelstein J. Machine learning approaches to personalize early prediction of asthma exacerbations. Annals of the New York Academy of Sciences 2017, 1387:153-165.

53. Brand PL. The Asthma Predictive Index: not a useful tool in clinical practice. Journal of Allergy and Clinical Immunology 2011, 127:293-294.

**54. Holt PG, Strickland D, Bosco A, et al. Distinguishing benign from pathologic TH2 immunity in atopic children. J Allergy Clin Immunol 2016, 137:379-387.

This study focuses on identifying internal control mechanisms that attenuate expression of IgE-associated responsiveness to aeroallergen in children with confirmed allergic sensitization.

16

Figure titles

Figure 1. Clustering IgE components from infancy (age 1 year) to adolescence (age 16 years) in a

population-based birth cohort study (from reference 35, with permission).

Cluster membership was determined using a Bernoulli Mixture Model applied to binarized sensitization

data from all subjects.

Figure 2. Diagnostic problems in asthma

17