UvA-DARE (Digital Academic Repository)
Precision medicine in childhood asthma The role of genetic variations in treatment response Farzan, N.
Publication date 2019 Document Version Final published version License Other Link to publication
Citation for published version (APA): Farzan, N. (2019). Precision medicine in childhood asthma: The role of genetic variations in treatment response.
General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)
Download date:04 Oct 2021 Precision Medicine in Childhood Asthma
The role of genetic variations in treatment response
Niloufar Farzan
Precision Medicine in Childhood Asthma
The role of genetic variations in treatment response
Niloufar Farzan
1
Het printen van dit proefschrift werd mede mogelijk gemaakt door AMC Amsterdam en Longfonds (Lungfoundation Nertherlands).
Printed by: Ipskamp
Precision Medicine in Childhood Asthma: The role of genetic variations in treatment response
2019 Niloufar Farzan, Amsterdam.
2
Precision Medicine in Childhood Asthma
The role of genetic variations in treatment response
ter verkrijging van de graad van doctor
aan de Universiteit van Amsterdam
op gezag van de Rector Magnificus
prof. dr. ir. K.I.J. Maex ten overstaan van een door het College voor Promoties ingestelde commissie,
in het openbaar te verdedigen in de Agnietenkapel
op dinsdag 12 februari 2019, te 10.00 uur
door
Niloufar Farzan
geboren te Tabriz
3
PROMOTIECOMMISSIE
Promotor
prof. dr. A.H. Maitland-van der Zee AMC-UvA
Copromotor
dr. S.J.H. Vijverberg AMC-UvA
Overige leden:
prof. dr. E.H.D. Bel AMC-UvA
prof. dr. P.J. Sterk AMC-UvA
dr. S.W.J. Terheggen AMC-UvA
prof. dr. G. Bartlett-Esquilant McGill University, Montreal Canada
prof. dr. B. Wilffert Rijksuniversiteit Groningen
Faculteit der Geneeskunde
4
Table of contents
Chapter 1: General introduction 7
Chapter 2: Reviewing evidence of (pharmaco)genomics effects in asthma 19
Chapter 2.1: The use of pharmacogenomics, epigenomics, and transcriptomics to improve 21 childhood asthma management: Where do we stand?
Chapter 2.2: Pharmacogenomics of inhaled corticosteroids and leukotriene modifiers: 45 a systematic review
Chapter 3: Pharmacogenomics in childhood asthma 113
Chapter 3.1: Rationale and design of the multi-ethnic Pharmacogenomics in Childhood 115 Asthma Consortium
Chapter 3.2: 17q21 variant increases risk of exacerbations in asthmatic children despite using 145 inhaled corticosteroids
Chapter 3.3: Genome-wide association study of inhaled corticosteroid response in 171 African-admixed children with asthma
Chapter 4: Risk factors associated with asthma exacerbations despite ICS use 213
Chapter 4.1: Risk factors of asthma exacerbations in asthmatic children treated with ICS: 215 is there an added value of genetic risk factors?
Chapter 5: Cost-effectiveness of pharmacogenetic-guided treatment 251
Chapter 5.1: Cost-effectiveness of genotyping before starting LABA therapy in asthmatic 253 children and young adults.
Chapter 6: General discussion 277
Scientific Summary & samenvatting 311
Appendices 326
5
6
CHAPTER 1 General Introduction
7
Evolution of disease management towards precision medicine
Since prehistoric times, humans have used natural products such as extracts from plants, animals and even microorganisms to treat different diseases (1). However, only since the 19th century, pharmacology evolved as an independent principle when scientists began to develop and apply experiments to understand physiological mechanisms and drug actions (1). These advances were followed by the technological breakthroughs and development of synthetic techniques at the beginning of the 20th century which revolutionized the pharmaceutical industry (1). However, parallel to the development of highly effective pharmacological agents, evidence started to pile up regarding the inter- and intraindividual variability in response to most drugs (2). Studies show that response to most drugs ranges between 50 and 75 percent (2). This fact has triggered scientists to find out why an effective and safe medication for one group of patients has a decreased or an exaggerated effect in another group. Decades of research have shown that factors such as age (3), gender (4), diet (5) and concomitant disease or medication use (6) can influence response to treatment. Additionally, over the past three decades advances in genomic technologies have helped scientists discover the role of genomic variations in treatment response (7,8). Altogether, these findings indicated that a combination of clinical, environmental and genomic factors could result in treatment response heterogeneity. In fact, this idea has ultimately paved the way for a new era in medicine called Precision Medicine (9). According to National Institutes of Health (NIH), precision medicine is "an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person" (9). The ultimate aim of precision medicine is to increase health on the population level by preventing diseases and giving the right treatment to the right patient at the right time (9).
Pharmacogenomics
The study of the relation between variations in the genome and response to a specific medication in terms of efficacy and safety is called pharmacogenomics/pharmacogenetics (10). Although these two terms are used interchangeably, their definitions differ. Pharmacogenetics mainly refers to genes determining drug metabolism, pharmaco- genomics, on the other hand, studies all genes in the genome that might determine drug response (11). In this thesis, we will use the term pharmacogenomics.
8
Single Nucleotide Polymorphisms (SNPs), the most abundant genetic variations in the genome, are the most commonly investigated genetic variations in pharmacogenomics studies (12). A SNP is a single base pair change at a specific position in the DNA sequence that results in different alleles (12). The human genome contains approximately 10 million SNPs (13). SNPs can occur in both protein-coding and non-coding regions that might contain enhancers and promoters (13). SNPs that fall within the protein-coding regions might change the amino acid sequence which can subsequently result in an altered amount or function of the protein. SNPs within the non-coding regions may modify the expression of one or several genes (13). Any alterations in gene expression amount or function of the protein could ultimately influence the mechanisms related to absorption, distribution, metabolism or excretion/elimination (ADME) of the drug (pharmacokinetics) or target (mostly receptors) of the drugs (pharmacodynamics) (14). These alterations subsequently can influence the efficacy of the drug and treatment response (12,15).
In pharmacogenomics studies, the association between genetic variations and treatment response is typically investigated by two main approaches: hypothesis-driven and hypothesis-generating approaches (16). The most common hypothesis-driven pharmaco- genomics studies are candidate gene studies (16). In these studies, ‘candidate genes’ are often selected based on prior knowledge of the biological mechanisms related to the disease or drug signaling pathway. The number of pre-selected SNPs can range from one to several hundred (16).
In contrast to hypothesis-driven approaches, in hypothesis-generating approaches usually, thousands or even millions of SNPs are studied simultaneously without pre-selection (16). Hypothesis-generating approaches are suitable for identifying (novel) genetic markers in complex diseases and traits where not only genes but also environmental factors, gene- gene interactions, and gene-environment interactions play a role in the development of disease (17). Genome-Wide Association Studies (GWAS) are one of these high throughput methods that study 500,000 to millions of common SNPs (frequency 5%) across the genome (18). The SNPs that are not assayed on the genotyping chips, can be estimated using reference panels and based on the knowledge about the correlation between a large set of SNPs (linkage disequilibrium). This process which is called genotype imputation increases the number of common variants with great precision (17).
The result of the pharmacogenomics studies could ultimately help to predict patients’ response to treatment based on their genetic makeup and could help to select the optimal drug and dose for each patient. Several clinical trials have shown promising results for
9
genotype-guided dosing algorithms of anticoagulants such as acenocoumarol, phencopromon, and warfarin (19,20) and genotype-guided drug selection for carbamazepine (an anti-epileptic drug) (21) and abacavir (an antiretroviral drug) (22). Pharmacogenomic markers have been investigated in different types of cancers (23), diabetes (24) and respiratory disorders such as asthma (25) as well.
Asthma and its pharmacological management in children
Asthma is the most common chronic disease in childhood and affects 10-15% of the pediatric population worldwide (26). Asthma can be particularly challenging in children because of the day-to-day fluctuations in its symptoms and severity (27). Although asthma cannot be cured, effective treatments are available for its management (26). Clinical asthma guidelines suggest a step-wise treatment approach to opt for the most appropriate treatment based on disease symptoms, severity and future risk of exacerbations (26). The first step of asthma management consists of treatment with short-acting β2 agonists (SABA) as needed to relieve bronchoconstriction. If symptoms continue despite SABA use, a low dosage of inhaled corticosteroids (ICS) are added to as needed SABA to reduce symptoms by suppressing inflammation. Subsequently, as a third step, the dosage of ICS may be increased or a long-acting β2 agonist (LABA) might be added. Alternatively, instead of a LABA, a leukotriene modifier (LTM) can be prescribed. In step 4, the dosage of ICS is further increased, or an LTM may be added to the ICS/LABA combination (26). If the symptoms remain uncontrolled, anti-IgE treatment might be considered for children ≥ 6 years of age with allergic asthma in the fifth step of asthma management (26).
Asthma exacerbations and response to ICS
Asthma exacerbations are defined as progressive worsening of the symptoms and lung function that needs urgent actions such as hospital admissions (26). Asthma exacerbations not only have a high impact on quality of life of the children and their parents/caregivers, but they also impose high expenses to patients and society (28,29). More than half of the direct costs of asthma are related to the use of emergency room facilities and hospitalizations (28). Therefore, prevention of asthma exacerbations is of utmost importance to increase the quality of life and reduce the financial burden. It is well known by now that asthma exacerbations are a result of interactions between genetic and environmental factors (26).
10
ICS are the cornerstone of asthma management in children with persistent asthma (26). ICS prevent asthma exacerbations by influencing a wide range of cells and genes involved in the inflammatory machinery (30,31). Although the majority of children respond well to ICS, 30-40% of the children continue to experience exacerbations despite good therapy adherence and correct inhalation technique (32). Genetic variations within the genes encoding for the proteins involved in inflammatory machinery and ICS signaling pathway could alter the treatment response. Indeed, heritability estimates suggest that approximately 40-60% of the heterogeneity in therapeutic response to ICS could be explained by genetic variations (33,34).
Pharmacogenomics of asthma
The first pharmacogenomics study of asthma was published in 1999 by Drazen et al. (35). In this study, researchers found a significant difference in response to an LTM (montelukast) between adult patients with different tandem repeats in the promoter of the ALOX5 gene. ALOX5 encodes for 5-lipoxygenase (5-LO), a limiting enzyme that is part of the leukotriene signaling pathway (36). In most individuals, ALOX5 contains a transcription factor binding region with five binding sites (Sp1 repeats); however, deletion and addition of binding sites result in a variation in this gene (36).
After this pioneering study, more than 80 pharmacogenomics studies have investigated the effect of genetic variations on response to three different asthma medication (ICS, LTM, SABA and LABA)(37,38). Like in other diseases, earlier pharmacogenomics studies of asthma have applied candidate gene approaches. The first GWAS of treatment response in asthma was performed in 2011 to identify genetic markers associated with ICS treatment response in children (39).
So far none of the pharmacogenomic markers of ICS found in GWAS or candidate genes, have reached clinical practice. First, most genetic variants identified in these studies have not been successfully replicated in additional populations. Validation of the association results across independent study populations is the first step towards translation of findings of a pharmacogenomics study into clinical practice (40). However, it seems that most of the potential pharmacogenomics markers of ICS response have remained in the first step of this process. Overestimation of the findings (the so-called winner’s curse) in the original studies due to the small sample sizes is one of the main reasons for unsuccessful replications in
11
other populations (40). Lack of consistency in definitions of the response outcome is another factor that results in a lack of replication. Further, heterogeneity in study populations (e.g. ethnic background and age) has made the generalizability of the findings to other populations even more challenging. For example, most of the previous studies have included non-Hispanic white patients and African-admixed populations seem to be underrepresented in pharmacogenomics studies of asthma (41). Heterogeneity between studies can result in a failed replication even if there is a true association (40). Therefore, there is a need for large sample sizes and standardization of the methodology in pharmacogenomics of asthma especially in asthmatic children, since the influence of genetic variations on treatment response might be more evident because of the shorter duration of chronic inflammation and airway remodeling in this group.
The Pharmacogenomics in childhood asthma consortium (PiCA)
One of the options to overcome some of the previously mentioned limitations is international collaborative efforts (40). Over the past decades, successful international consortia have been established to investigate the genetic factors of cardiovascular diseases (42), and asthma (43). International collaborations can build sufficient sample sizes to detect a true pharmacogenomic effect. The PiCA consortium was initiated in 2013 to boost international collaborations and to perform large-scale pharmacogenomics studies of asthma using standardized outcome definitions in well-characterized asthmatic children across different ethnicities.
12
Objectives of the thesis
The overall objective of the thesis is to identify and validate genetic variants that could help better asthma management in children. The first objective of the thesis is to evaluate the influence of the genetic variations on the risk of asthma exacerbations in children/young adults treated with ICS by performing large-scale genetic association analyses within the PiCA consortium. The second objective is to investigate clinical and/or environmental risk factors of exacerbations and the added value of the single and joint genetic information to identify children at higher risk of asthma exacerbations while being treated by ICS. The third and last objective is to assess whether pharmacogenetic-guided treatment can be cost- effective, using ADRB2-guided LABA treatment as a case study.
13
Thesis outline
This thesis is divided into four parts, which are preceded by a general introduction (chapter 1) and concluded by a general discussion (chapter 6).
Chapter 2 starts with a review of the current and potential future role of the three most commonly studied omics fields (pharmacogenomics, epigenomics, and transcriptomics) in childhood asthma (chapter 2.1). This part is followed by a systematic review that provides an in-depth and an up-to-date scientific evidence of pharmacogenomics studies of anti- inflammatory treatment (ICS and LTM) in asthma (chapter 2.2). in addition to reviewing the scientific evidence, the methods of the studies are evaluated to address the unmet needs of the asthma management.
Chapter 3 is dedicated to the design and studies of the international PiCA consortium. Chapter 3.1 describes the design and rationale of the PiCA consortium. In Chapter 3.2, a large-scale meta-analysis is performed to evaluate the influence of a previously identified asthma-related genetic variant (rs7216389) within 17q21, on asthma exacerbations despite ICS in children and young adults participated in PiCA. Chapter 3.3, focuses on a GWAS meta-analysis conducted in African-admixed children to identify genetic variants associated with asthma exacerbations despite ICS. The findings of the GWAS are further replicated in PiCA studies including non-Hispanic white subjects.
In Chapter 4, the focus shifts toward detecting clinical and environmental risk factors of asthma exacerbations in children despite ICS use in one of the PiCA studies: the PACMAN cohort. Furthermore, it assesses the added value of genetic markers to these factors in order to identify children at higher risk of exacerbations.
Chapter 5, is dedicated to an economic evaluation to examine if a pharmacogenomic- guided strategy to prevent exacerbations is cost-effective for asthmatic children who need a step up from low dose ICS.
Finally, Chapter 6 provides a general discussion of the findings from a broader perspective and makes recommendations for future pharmacogenomics and precision medicine studies of childhood asthma.
14
References
1. Yuan H, Ma Q, Ye L, Piao G. The 14. B.G. Katzung and A. J. Trevor: Basic and Traditional Medicine and Modern Medicine clinical pharmacology. 13th edition, Mc Graw from Natural Products. Molecules 2016;21(5): Hill, 2015. Accessible online via access 559. medicine.mhmedical.com.
2. Spear BB, Heath-Chiozzi M, Huff J. Clinical 15. Li H, Achour I, Bastarache L, Berghout J, application of pharmacogenetics. Trends Mol Gardeux V, Li J, et al. Integrative genomics Med. 2001;7(5):201–4. analyses unveil downstream biological effectors of disease-specific polymorphisms 3. Mangoni A.A., Jackson S.H.D. Age-related buried in intergenic regions. NPJ genomic changes in pharmacokinetics and pharmaco- Med. 2016;1(1):16006. dynamics: basic principles and practical applications. Br J Clin Pharmacol. 2004;57(1) 16. Hattersley AT, McCarthy MI. What makes 6–14. a good genetic association study? Lancet. 2005;366(9493):1315–23. 4. Schwartz JB. The current state of know- ledge on age, sex, and their Interactions on 17. Bush W.S., Moore J.H. Chapter 11: clinical pharmacology. Clin Pharmacol Ther. Genome-wide association studies. Lewitter 2007;82(1):87–96. F, Kann M, editors. PLoS Comput Bio. 2012;8(12):e1002822. 5. de Boer A, van Hunsel F, Bast A. Adverse food–drug interactions. Regulatory Toxicol 18. Cordell H.J., Clayton D.G. Genetic Pharmacol. 2015;73(3):859–65. association studies. Lancet. 2005;366 (9491) :1121–31. 6. Palleria C, Di Paolo A, Giofrè C, Caglioti C, Leuzzi G, Siniscalchi A, et al. Pharmaco- 19. Stitham J, Vanichakarn P, Ying L, Hwa J. kinetic drug-drug interaction and their imp- Cardiovascular pharmacogenetics of anti- lication in clinical management. J Res Med thrombotic agents and non-steroidal anti- Sci. 2013;18(7):601–10. inflammatory drugs. Curr Mol Med. 2014; 14(7):909–31. 7. Price AL, Spencer CCA, Donnelly P. Progress and promise in understanding the 20. Vanichakarn P, Hwa J, Stitham J. genetic basis of common diseases. Proc R Cardiovascular pharmacogenetics of anti- Soc B Biol Sci.2015;282(1821): 20151684. hypertensive and lipid-lowering therapies. Curr Mol Med. 2014;14(7):849–79. 8. Pirmohamed M. Pharmacogenetics: past, present and future. Drug Discov Today. 2011; 21. Chen P, Lin J.J, Lu C.S, Ong C.T, Hsieh 16(19–20):852–61. PF, Yang C.C, et al. Carbamazepine-Induced Toxic Effects and HLA-B*1502 Screening in 9. NIH. US National Library of Medicine. Taiwan. N Engl J Med. 2011;364(12):1126– https://ghr.nlm.nih.gov/primer/ 33. precisionmedicine/definition. 22. Mallal S, Phillips E, Carosi G, Molina J-M, 10. NIH. US National Library of Medicine. Workman C, Tomažič J, et al. HLA-B*5701 https://ghr.nlm.nih.gov/primer/ Screening for Hypersensitivity to Abacavir. N genomicresearch/pharmacogenomics. Engl J Med. 2008;358(6):568–79.
11. Pirmohamed M. Pharmacogenetics and 23. Hertz D.L., Rae J. Pharmacogenetics of pharmacogenomics. Br J Clin Pharmacol. Cancer Drugs. Annu Rev Med. 2015; 66(1): 2001;52(4):345–7. 65–81.
12. Burton PR, Tobin MD, Hopper JL. Key 24. Maruthur NM, Gribble MO, Bennett WL, concepts in genetic epidemiology. Lancet. Bolen S, Wilson LM, Balakrishnan P, et al. 2005;366(9489):941–51. The Pharmacogenetics of Type 2 Diabetes: A Systematic Review. Diabetes Care. 2014;37 13. NIH. https://hgr.nlm.nih.gov/primer/ (3):876–86. genomeresearch/snp.
15
25. Tavares LC, Marcatto LR, Santos PC. ALOX5 promoter genotype and the response Genotype-guided warfarin therapy: current to anti-asthma treatment. Nat Genet. 1999; status. Pharmacogenomics. 2018;19(7):667– 22:168–70. 85. 36. In KH, Asano K, Beier D, Grobholz J, Finn 26. Global Strategy for Asthma Management PW, Silverman EK, et al. Naturally occurring and Prevention 2018: www.ginasthma.org. mutations in the human 5-lipoxygenase gene promoter that modify transcription factor 27. Lodrup Carlsen KC, Pijnenburg MW, ERS binding and reporter gene transcription. J Clin Task Force Monitoring Asthma in Children. Invest. 1997;99(5):1130–7. Monitoring asthma in childhood. Eur Respir Rev. 2015;24(136):178–86. 37. Farzan N, Vijverberg S.J.H., Arets H.G., Raaijmakers J.A.M, Maitland-van der Zee 28. Williams AE, Lloyd AC, Watson L, Rabe A.H. Pharmacogenomics of inhaled cortico- KF. Cost of scheduled and unscheduled steroids and leukotriene modifiers: A sys- asthma management in seven European tematic review. Clin Exp Allergy J. 2017; 47 Union countries. Eur Respir Rev. 2006; (2):271–93. 15(98):4–9. 38. Slob E.M.A, Vijverberg S.J.H, Palmer 29. Williams A.E., Lloyd A.C, Watson L, Rabe C.N.A, Zazuli Z, Farzan N, Oliveri N.M.B, et KF. Cost of scheduled and unscheduled al. Pharmacogenetics of inhaled long-acting asthma management in seven European beta 2 agonists in asthma: a systematic Union countries. Eur Respir Rev. 2006; review. Pediatr Allergy Immunol. 2018. 15(98):4–9. 39. Tantisira KG, Lasky-Su J, Harada M, 30. Stahn C, Löwenberg M, Hommes D.W., Murphy A, Litonjua AA, Himes BE, et al. Buttgereit F. Molecular mechanisms of gluco- Genomewide association between GLCCI1 corticoid action and selective glucocorticoid and response to glucocorticoid therapy in receptor agonists. Mol Cell Endocrinol. asthma. N Engl J Med. 2011;365(13):1173– 2007;275(1–2):71–8. 83.
31. Jiang C.L, Liu L, Li Z, Buttgereit F. The 40. Ioannidis J.P.A., Thomas G, Daly M.J. novel strategy of glucocorticoid drug develop- Validating, augmenting and refining genome- ment via targeting nongenomic mechanisms. wide association signals. Nat Rev Genet. Steroids. 2015;102:27–31. 2009;10(5):318–29.
32. Szefler SJ, Phillips BR, Martinez FD, 41. Ortega E., Meyers A. Pharmacogenetics: Chinchilli VM, Lemanske RF, Strunk RC, et implications of race and ethnicity on defining al. Characterization of within-subject resp- genetic profiles for personalized medicine. J onses to fluticasone and montelukast in Allergy Clin Immunol. 2014;133(1):16–26. childhood asthma. J Allergy Clin Immunol. 2005;115(2):233–42. 42. Manson L.E., van der Wouden C.H., Swen JJ, Guchelaar HJ. The Ubiquitous 33. Anderson W.H., Koshy B.T., Huang L, Pharmacogenomics consortium: making Mosteller M, Stinnett S.W., Condreay L.D., et effective treatment optimization accessible to al. Genetic analysis of asthma exacerbations. every European citizen. Pharmacogenomics. Ann Allergy, Asthma Immunol. 2013;110(6): 2017;18(11):1041–5. 416–422.e2. 43. Moffatt M.F., Gut I.G., Demenais F, 34. Palmer LJ, Silverman ES, Weiss ST, Strachan DP, Bouzigon E, Heath S et al. A Drazen JM. Pharmacogenetics of asthma. large-scale, consortium-based genomewide Am J Respir Crit Care Med. 2002;165(7):861 association study of asthma. N Engl J Med. –6. 2010;363(13):1211–21.
35. Drazen JM, Yandava CN, Dubé L, Szczerback N, Hippensteel R, Pillari a, et al. Pharmacogenetic association between
16
17
18
Reviewing evidence of CHAPTER 2 (pharmaco)genomics effects in asthma
19
20
CHAPTER 2.1
The use of pharmacogenomics, epigenomics and transcriptomics to improve childhood asthma management: where do we stand?
Farzan N
Vijverberg SJ
Kabesch M
Sterk PJ
Maitland van der Zee AH
Pediatric Pulmonology, 2018;53(6):836-845.
21
Abstract
Asthma is a complex multifactorial disease and it is the most common chronic disease in children. There is a high variability in response to asthma treatment, even in patients with good adherence to maintenance treatment and a correct inhalation technique. Distinct underlying disease mechanisms in childhood asthma might be the reason of the heterogeneity in response to treatment. A deeper knowledge of the underlying molecular mechanisms of asthma has led to the recent development of advanced and mechanism- based treatments such as biologicals. However, biologicals are recommended only for patients with specific asthma phenotypes who remain uncontrolled despite high dosages of conventional asthma treatment. One of the main unmet needs in their application is lack of clinically available biomarkers to individualize pediatric asthma management and guide treatment. Pharmacogenomics, epigenomics and transcriptomics are three omics fields that are rapidly advancing and can provide tools to identify novel asthma mechanisms and biomarkers to guide treatment. Pharmacogenomics focuses on variants in the DNA, epigenomics studies heritable changes that do not involve changes in the DNA sequence but lead to alteration of gene expression, and transcriptomics investigates gene expression by studying the complete set of mRNA transcripts in a cell or a population of cells. Advances in high-throughput technologies and statistical tools together with well- phenotyped patient inclusion and collaborations between different centers will expand our knowledge of underlying molecular mechanisms involved in disease onset and progress. Furthermore, it could help to select and stratify appropriate therapeutic strategies for subgroups of patients and hopefully bring precision medicine to daily practice.
22
Introduction
Asthma affects more than 300 million individuals worldwide and is the most common chronic disease in children (1). According to the GINA guidelines, inhaled corticosteroids (ICS) combined with short-acting β2 agonists (SABA) or long-acting β2 agonists (LABA) are the preferred treatment for persistent asthma symptoms (1). However, even in asthmatic children with a good adherence to maintenance treatment and an adequate inhalation technique, there is a high variability in response to treatment (2–4); approximately ten percent of the patients still have serious complaints despite high doses of ICS (5).
Precision medicine focuses on identifying approaches for treatment and prevention of the disease based on the personal genetic profile, environment and lifestyle of the patients. Understanding the causes of asthma heterogeneity and the underlying molecular mechanisms is of utmost importance to improve treatment outcomes. However, information merely based on clinically recognizable features, does not sufficiently explain the observed heterogeneity in asthma treatment outcomes (6). A deeper knowledge of the molecular disease mechanisms has led to the development of more targeted asthma treatments such as omalizumab (anti-IgE monoclonal antibody) for patients (≥6 years of age) with allergic asthma and mepolizumab (anti-interleukin-5 monoclonal antibody) for patients (≥12 years of age) with severe eosinophilic asthma. These medications (also called biologicals) are recommended for patients with these specific phenotypes when they remain uncontrolled despite treatment with high doses of ICS (1). However, biologicals are expensive and debates on their efficacy are ongoing (7,8). One of the main unmet needs in their application is lack of clinically available biomarkers to individualize pediatric asthma management and guide treatment.
Omics is a broad term that refers to various research fields within biology (such as genomics, transcriptomics, proteomics etc.)(Fig.1) that investigate biological mechanisms, molecules and molecule states in a broad and general manner (9). These approaches, especially when applied in combination, have the potential to improve diagnosis, phenotyping and disease monitoring by identifying novel biomarkers that can be used to characterize subgroups of patients and therefore select the most appropriate therapy for stratified groups of patients (10). In this review, we will provide an overview of three most validated ‘omics approaches in relation to pediatric asthma management; pharmaco- genomics, epigenomics and transcriptomics, we will provide examples of how these three
23
fields can interact with each other and finally, we will discuss limitations of these fields in relation to the clinical implementation of recent findings.
Figure1. Illustration of the different omics layers. A, Influence of genetic (SNP) and epigenomic variations (1.DNA methylation, 2.Histone modification, and 3.non-coding miRNAs) on gene expression (transcriptome), protein synthesis (proteome), and finally on metabolites (metabolome). B, interaction between different omics layers. Integration of omics data has the potential to improve diagnosis, phenotyping, and disease monitoring by identifying novel biomarkers that can be used to characterize subgroups of patients and therefore select the most appropriate therapy for stratified groups of patients. miRNA; microRNA, SNP; Single Nucleotide Polymorphism
Genomics and Pharmacogenomics
Genomics refers to the genome-wide study of variants in the deoxyribonucleic acid (DNA). Pharmacogenomics is a subfield of genomics which assesses the effect of DNA variants on the patient’s response to medication (11). Modulating effects of genetic variations, both focusing on asthma onset (genomics) as well as response to therapeutic agents
24
(pharmacogenomics), have been reported in numerous studies (4, 12–15). Single nucleotide polymorphisms (SNPs) are the most common type of genetic variation in the human genome and are commonly investigated in pharmacogenomics studies using candidate gene and genome wide association studies (GWAS).
A large number of loci have been identified to be associated with childhood and adult onset asthma in GWA studies such as genes encoding for cytokines such as ILRL1/IL18R1, IL33, IL2RB, and IL18RB1 and several other genes involved in inflammation and immunity such as SMAD3, TSLP and the HLA region (also known as the human MHC region) (16– 20). The strongest and most consistent genetic signals associated with childhood of asthma is located in the 17q21 locus (18). The most frequently assessed SNP in the locus, rs7216389, influences asthma risk (odds ratio (OR)= 1.84; 95% confidence interval (CI), 1.43–2.42) and the expression of several genes on chromosome 17 (ORMDL3, GSDMA and CDK12) (18). These associations with asthma susceptibility have been confirmed across diverse ethnic populations (13,21–23) but the genetic signal from that region may even be broader. Some of the genes in 17q21 (ORMDL3, GSDMA, GSDMB, ZPBP2, IKZF3) also associate with severe asthma in children (12,24,25), asthma exacerbations (26), treatment response (27–29) but not with atopy (23,24).
Pharmacogenomics effects related to asthma medication, such as inhaled corticosteroids (ICS), long-acting beta2-agonists (LABA) and leukotriene modifiers (LTMs), have been studied intensely in the last decades (4,30). For LABA pharmacogenomics, the most clinical relevant variant seems to result in the substitution of amino-acid Glycine by Arginine at position 16 in the gene (ADRB2) encoding for beta-2 receptor. This variant was shown to be associated with poor response to LABA in children (14,31,32). In a recent meta-analysis of 4226 asthmatic children included in the multi-ethnic Pharmacogenomics in Childhood Asthma (PiCA) consortium, patients treated with ICS plus LABA showed an increased risk of asthma exacerbations for each copy of the ADRB2 risk allele (OR 1.52, 95 % CI 1.17–1.99; p = 0.002) (33). These findings suggest that ADRB2 genotype-guided treatment can improve disease management.
ICS response has been investigated in a few GWA studies (34–39) as described extensively in a recent systematic review (4). In short; the most significant findings have been reported for FBXL7, (rs10044254, chromosome 5) (37), and SNPs located in chromosomes 6 (rs6924808) (38), 11 (rs1353649) (38) and 16 (rs2388639) (37), close to LOC728792). Several other SNPs located in ALLC (34), GLCCI1 (35) and T gene (36), have been suggestively associated with response to inhaled corticosteroids but they did
25
not reach genome-wide significance. In a recent publication including 2,627 patients (age >12 years) with different ethnic backgrounds, no genetic variation was associated with the response to ICS (39). In candidate gene studies (>25 studies), CRHR1, FCER2, GLCCI1, TBX21, STIP1 and NR3C1 are among the most commonly investigated candidate genes (4). Despite the large number of studies, few findings have been replicated and the clinical applicability of these associations remains unclear. The most promising genetic determinants of ICS in response in asthmatic children may be found in the FCER2 gene (40–42). Variants in this gene have been found to be associated with severe asthma exacerbations, poor lung function and asthma symptoms despite ICS in pediatric populations.
GWAS and candidate gene studies of LTMs also delivered inconsistent results (43–47). To date, two GWAS of LTMs have been performed (47,48). SNPs within MRPP3 (rs12436663, chr 14), GLT1D1 (rs517020, chr12) and MLLT3 (rs6475448, chr 9) were identified in the GWAS to be associated with responses to LTMs. These studies included adults and adolescents but GWAS including pediatric populations are missing. In candidate gene studies of LTMs, LTC4S, LTA4H, ALOX5, ALOXAP and CysLTR1 (genes involved in the cysteine leukotriene pathway) and MRP1 have been studied most (4).
Although genomics studies have identified genetic variants which contribute to asthma development and pharmacogenomics have identified genetic variants that contribute to treatment heterogeneity, these markers are currently not ready for clinical application. Regarding pharmacogenomics of LABA, genetic variant in the ADRB2 gene is the most clinical relevant biomarker that has been replicated in pediatric asthma populations. In order to implement genotype-guided treatment, randomized clinical trials are needed to validate the finding (32,49). There is still not enough evidence to implement genetic testing in clinic before starting ICS or LTMs. Small sample sizes and different outcome measures in the pharmacogenomics studies of ICS and LTMs resulted in lack of replication of the findings. Furthermore, heterogeneity between these studies has made the comparison of the findings challenging. International and nation-wide collaborations have been established to perform large-scale GWAS to identify asthma susceptibility genes (13,50). To assess the clinical value of genetic markers for asthma management, international collaboration and large-scale meta-analyses with well-phenotyped asthmatic children are needed (51). Hopefully, the recent PiCA consortium that combines existing studies to obtain larger sample sizes will deliver new opportunities to study the clinical importance of pharmacogenetics associations in large observational studies (52).
26
Epigenomics
Epigenomics is the study of heritable changes that lead to alteration of gene expression that do not involve changes in the DNA sequence (53). In contrast to genetics, environmental factors can change epigenomic signatures. Three epigenomic mechanisms in humans are recognized. The first epigenomic mechanism is post transcriptional histone modification. The double helix DNA strand is wrapped around proteins called histones. These histones play an important role in the regulation of DNA expression, by influencing the DNA accessibility for the transcription machinery. Structural modification of the histones can change how densely the DNA is packed and how easily DNA polymerase can reach the DNA. There are four main histone modification processes: histone methylation, acetylation, phosphorylation and ubiquitination. The second epigenomic mechanism is expression of non-coding single-stranded microRNAs (miRNAs). These miRNAs regulate transcriptions by neutralizing messenger RNA (mRNAs) (54).
The third and most studied epigenetic mechanism so far is DNA methylation. DNA methylation occurs when a methyl group is added to the carbon in the cytosine nucleic acid. Approximately 70-80% of C-G dinucleotides (CpG) in human DNA are methylated (55). DNA methylation inhibits gene transcription through altering the accessibility of the DNA for transcription (56). DNA methylation profiling studies can be conducted using candidate genes or (epi)genome-wide markers in CpG sites, ranging from couple of thousands to millions of CpG sites. DNA methylation and histone modification are interacting processes (57).
Unlike pharmacogenomics, only a limited number of studies have so far investigated the effect of epigenomics in response to asthma medication therapy (58,59). Most studies have focused on the effect of in utero or early life epigenomic changes on asthma development and asthma like symptoms in childhood. Most of these studies have investigated influence of different environmental factors such as in utero exposure to farm environment (60–62), exposure to air pollutants (63–65) and tobacco smoke exposure (66–71) on DNA methylation patterns and several longitudinal studies have studied the persistence of the epigenomics changes over time (60,66,72).
As epigenomic profiles are cell- and tissue-specific and many cell types are involved in asthma pathogenesis, various sample types have been used to study epigenomic mechanisms in children with asthma; from epithelial and nasal brushings (59,73,74) to peripheral blood samples (64,73,75–79) and cord blood (80–82). In this section, we will
27
summarize and discuss most recent findings of epigenomic studies in childhood, their possible contribution to childhood asthma onset, and the effect of ICS on epigenomic profiles and thereby on treatment response.
It has been shown that in utero or early life (before 1 year of age) exposure to farm environment influences asthma development and affects different phenotypes of asthma in children (83,84). Early life exposure to various microbes in a farm environment (e.g. Acinetobacter lwoffii F78), due to the contact with livestock and consumption of unpasteurized milk, may have a protective effect towards asthma development (85,86). This protective effect is in line with the well-known ‘hygiene hypothesis’ which was proposed in 1989 (87). The hygiene hypothesis suggests that children exposed to germs early in life are protected from developing asthma and other atopic diseases. However, it seems that hygiene hypothesis cannot explain the high prevalence of asthma in low- income urban areas (88). This discrepancy may be due to the differences in microbial composition of different environments (88).
Epigenomic modulation of different cell types involved in immunity and inflammation may underlie this observed protective effect of farm environment on asthma development (60,72,89). An example of such a mechanism is epigenomic modulation of the CD14 receptor. This receptor exists in a soluble form or on the innate immune cells such as macrophages and monocytes and has a role in bacterial lipopolysaccharide (LPS) detection and signaling along with Toll-like receptor (TLR)2 and TLR4 (90–92). Gene- environment interactions mediated by CD14 has been previously proposed to influence susceptibility to asthma (93). Samples of placenta’s of mothers living in a farm environment showed low DNA methylation of the CD14 promoter and higher CD14 expression compared to mothers not living in a farm environment (89,94). Furthermore, genetic and epigenetic mechanisms may interact over time. For example, expression of CD14 may change over childhood due to genetic and epigenomic variations (72). Genetic variations in the promoter of CD14 influence sCD14 levels in early childhood (at birth and 2 years of age) but not later (measured at age 10) (72).
In another longitudinal study, methylation changes in 23 potentially regulatory regions related to 10 genes previously associated with asthma and atopy mechanisms were investigated between birth and the age of 4.5 years in asthmatic and non-asthmatic children growing up in farming and non-farming families. While farm exposure had a strong impact on methylation at birth, asthma status was more related to methylation changes at age 4.5 years. Changes in methylation over time occurred in 15 gene regions and these
28 differences clustered in the genes highly associated with asthma (ORMDL family) and IgE regulation (RAD50, IL13, and IL4), but not in the T-regulatory genes (FOXP3, RUNX3) (60).
Tobacco smoke is another factor that has long been known to increase the risk of asthma development (95). It has been shown that in utero exposure to tobacco smoke results in impaired lung function and wheeze early in life (68–71). The role of epigenomic alterations due to in utero exposure to smoking has been reported in several studies (66–71). Some studies have also investigated the effect of timing (before or during pregnancy), and intensity (sustained or occasional) of mother’s smoking on epigenomic changes in cord blood of newborns (96,97). In a recent large-scale DNA methylation meta-analysis including 13 cohorts (6,685 newborns), the effect of sustained maternal smoking during pregnancy on DNA methylation of infants has been reported (97). Approximately 3,000 CpGs were differentially methylated in newborn cord blood of smoking mothers compared to newborns of non-smoking mothers. The most significant result was found for a methylation in AHRR with a p value < 1.6 × 10-193. Influence of smoking on methylation patterns of this gene has been previously reported in several studies (98,99). All CpGs in the meta-analysis were identified in older children as well. The results from this study indicate that epigenomic changes in newborns related to in utero smoke exposure persist into late childhood (100).
In a genome-wide DNA methylation analysis that assessed atopic asthmatic children from inner-city areas several genes, including two T helper type 2 (Th2) immunity genes IL13 and RUNX3, were hypo-methylated in peripheral blood mononuclear cells (PBMCs) of asthmatic children when compared with healthy controls (101). Methylation of 11 genes was associated with higher serum concentrations of IgE and methylation of 16 genes were associated with FEV1% predicted.
The effect of epigenomic signatures on corticosteroid response in asthmatic patients has been assessed in a limited number of studies (58,59). Corticosteroids themselves are strong epigenetic modifiers that influence histone acetylation of inflammatory genes (102). In one study, DNA methylation of the VNN1 gene has been found to be associated with corticosteroid response (59). Xiao et al. showed that asthmatic children who respond better to systemic corticosteroids when experiencing severe exacerbations have increased VNN1 methylation compared to non-responders. These results suggest that VNN1 methylation may regulate corticosteroid response in patients with severe asthma
29
exacerbations. VNN1 encodes for pantetheinase which is an epithelial ectoenzyme and might play a role in oxidative-stress response (103).
Transcriptomics
Transcriptomics is the investigation of all RNA transcripts such as mRNAs, non-coding RNAs and small RNAs (so-called transcriptome) in a cell or cell population using high- throughput methods. The transcriptome reflects the active expression of genes at a given time (104). RNA sequencing, microarrays and real-time PCR are the most common methods used for gene expression studies. In contrast to microarrays, which cannot detect novel transcripts, RNA sequencing is an unbiased method that provides all transcripts for the whole-genome including non-coding RNAs (105). Gene expression patterns are highly influenced by biological (i.e. age, gender, tissue/cell types, time of sampling etc.), environmental (i.e. allergens, infections, medication etc.) and technical factors (sampling methods, analytical methods and gene expression platforms) (106). Since a massive amount of data is generated in transcriptomics analyses, cluster analyses such as hierarchical clustering, k-means, self-organizing maps and Gene Set Variation Analyses (GSVA) are commonly used to reduce false-positive (107–110). Sophisticated large-scale analytical methods have been applied in projects such as the Unbiased Biomarkers for the Prediction of Respiratory Disease Outcomes (U-BIOPRED) consortium, to identify gene expression patterns in adults and children with severe asthma (111–115).
Findings of the U-BIOPRED project, show a marginal overlap between conventional definition of severe asthma based solely on clinical manifestations and definitions of disease based on unbiased molecular mechanisms. In a recent study by Hekking et al., significant difference between gene expression profiles of childhood-onset severe asthma (onset before 18 years of age) and adult-onset severe asthma indicated distinct underlying mechanisms of the disease in these two age groups of patients (116). Differences in gene signatures between adult-onset and childhood-onset severe asthma treated with ICS indicate an age dependent difference in response to this medication. These results show that findings in adult-onset asthma cannot be translated directly to the childhood asthma population due to the different underlying molecular mechanisms responsible for disease onset and progress.
30
Several childhood asthma studies have investigated transcriptome and gene expression in severe asthma (117) and asthma exacerbations (118–121). A transcriptomic analysis of asthma using peripheral blood leukocytes (117) of patients with uncontrolled and therapy resistant asthma (despite high doses of ICS and other therapeutic agents) identified 816 differentially expressed genes previously known to be related to asthma and 489 novel sites. E.g. there was a decrease in glucocorticoid receptor (NR3C1) signaling in children with therapy-resistant asthma compared to children with controlled persistent asthma and healthy individuals. Furthermore, the RORA gene, previously identified in GWAS of asthma, (13) was significantly upregulated in severe asthmatics. Additional gene network analysis demonstrated enhanced activity of mitogen-activated protein kinase (MAPK) and Jun kinase cascades in severe asthmatics. This cascade regulates cell activities such as mitosis, cell survival, differentiation, proliferation and gene expression in response to a range of stimuli (e.g. heat shock and inflammatory cytokines) (122).
It has previously been shown that atopic asthmatic patients are more prone to severe exacerbations in the presence of viral infections suggesting an interaction between viral and allergen-triggered immune responses resulting in increased risk of progression to persistent asthma in children (123).
Several studies have attempted to discover underlying mechanism of asthma exacerbations in children by conducting transcriptomics (118–120,124). One of these studies compared genome-wide expression profiling of PBMCs from 67 atopic asthmatics during hospitalization for viral-induced severe exacerbations to their gene expression during a subsequent recovery period (118). During exacerbations, expression of CCR2, a key regulator of dendritic cell (DC) and monocyte trafficking to the lung during airway inflammation (125), and FECR1 (high-affinity IgE receptor) were significantly upregulated in innate immune cells and particularly in monocytes and DCs. Furthermore, allergen- specific and total IgE levels, monocytes and DCs were significantly higher during exacerbations compared with convalescence phase. On the other hand, Aoki et al. (119) have identified several differentially expressed genes during exacerbation that are not related to immune responses against viral infections which reveals other underlying pathways in asthma exacerbations (119). They identified 153 differentially expressed genes from PBMCs during asthma exacerbations (12 patients) compared to stable asthmatic children (11 patients) using microarray analysis. In another study (120), genome-wide gene expression analysis in response to environment dust mite allergen showed higher expression of IL9, IL5 (a cytokine produced mainly by mast cells and Th2
31
cells) (126) and PRG2 (or natural killer cell activator) in PBMCs from dust mite sensitized children compared to non-sensitized patients (120). Furthermore, in a gene by environment interaction model, three genetic variations in IL9 were significantly associated with severe asthma exacerbations in patients exposed to high dust mite levels. IL9 cytokine is a production of mast cells, Th2 and it regulates several biological processes such as cell proliferation (127).
Although the main findings of these studies differ, one main conclusion is that they all have illustrated the complexity of asthma exacerbations. Moreover, they have identified several novel mechanisms involved in exacerbations in addition to the known immunologic biomarkers such as IL9. Association of IL9 with asthma has been reported in animal models and research on therapeutic potential of this cytokine is still ongoing (128).
Putting omics together
While many studies so far have applied hypothesis-free omics approaches of one or another kind, studies that combine different omics techniques are still rare. These so- called systems biology approaches where different layers are sophistically analyzed, are thought to be the next big thing in medical science.
One of these rare studies has been published in the field of asthma research (66). In that study, researchers showed that DNA methylation sites remain stable up to 4 years after birth in children and their mothers. Interestingly, there was a marginal overlap between DNA methylation patterns between mothers and their children indicating a distinct response pattern on the DNA methylation level to the same environmental exposure. Approximately 20% of the differentially methylated regions (DMRs) were influenced by environmental factors (non-genetic DMRs) and the remaining were influenced by genetic variations. Non-genetic DMRs remained stable after 4 years in children.
In a systems approach (129), combination of DNA methylation data with genotype, histone modification patterns and transcription data revealed a joint effect of the genotype and tobacco smoke on DNA methylation of an enhancer (JNK2) located in the GFPT2 gene. Furthermore, hypo-methylation of this enhancer was associated with an increased risk of asthma-like symptoms such as wheezing in children (129).
32
Discussion and conclusion
Recent studies in pediatric asthma demonstrate the importance of environmental, genetic and epigenetic factors on disease development and their complex interactions with each other that result in different gene expression patterns and distinct disease phenotypes. Asthma diagnosis in childhood, especially in pre-school children, remains challenging (130). Asthma symptoms such as coughing and wheezing are also common in children after an upper airway infection (131). Recent findings show that early life or in utero exposure to different environmental factors together with genetic and epigenetic markers play a crucial role in asthma development later in childhood. Omics approaches have already started to contribute to a deeper understanding of underlying molecular mechanisms of asthma.
High-throughput technologies and advanced statistical tools have enabled large scale analyses using mathematical models that integrate different genomic layers. This can be done by Similarity Network Fusion (SNF) that has recently been developed and validated (132). Finding ‘biomarker fingerprints’ using this technology can reveal the underlying molecular mechanisms of disease, new pathways related to disease onset and progress and finally it can lead to precision medicine (133).
Molecular fingerprinting could also help to decide which asthma treatment benefits which patient groups most. It has been shown that prescription of ICS in pre-school children experiencing wheezing helps to improve symptoms, reduce exacerbations and maintain lung function. However, not all children with wheeze respond to this medication and it remains unclear which children will experience persistent asthma symptoms (134). However, no clinical pharmacogenomics, epigenomics or transcriptomics marker has currently reached clinical asthma practice. This can be due to the heterogeneity between study designs, patient selection, clinical outcomes and sampling tissues.
The heterogeneity between studies makes the comparison of findings difficult. The clinically most relevant biomarker that has been replicated positively in pediatric asthma populations is the substitution of amino-acid Glycine by Arginine at position 16 of the ADRB2 gene. Furthermore, the most consistent results in the pharmacogenomics of ICS have been reported for the FCER2 gene. Future clinical trials are needed to validate genotype-guided treatment in patients carrying the risk variants. In contrast to genomics, epigenomics and transcriptomics profiles are tissue-specific and in complex diseases such as asthma, several types of tissues and cells are involved in the disease pathology
33
(116,135). Hence, to gain a better understanding of the disease, samples from different tissues must be collected. Furthermore, epigenomic and transcriptomic profiles can be influenced by various factors such as time of tissue sampling, sampling technique, environmental factors, which can cause problems in large sample sizes and longitudinal study design. In high-throughput analyses, large sample sizes are required to reduce false positive results. Therefore, it is essential to externally validate all studies in independent populations. In order to do so, standardized step-wise procedures have to be followed in order to reach the required statistical robustness (136). Collaboration of different research centers with well-phenotyped patients could help to overcome problems with the small sample sizes. There have been successful collaborations in the field of genomics/ pharmacogenomics of asthma in pediatric populations (52,137). These collaborations in a systems approach are needed to disentangle the complexity of asthma, to identify phenotypes that are predictive of therapy response and to identify clinically available biomarkers to guide asthma management.
In conclusion, complex interactions between physiological and environmental factors and biomarkers play role in the development of asthma and its clinical phenotypes. Deeper understanding of genomic, epigenomic and transcriptomic mechanisms and their interaction can lead to discovery of new drug targets. A Systems Medicine approach investigates these interactions between different dimensions and it can help to identify phenotypes that are predictive of therapy response.
34
References
1. www.ginasthma.org. Global Iniative for Application of ’omics technologies to bio- Asthma. Global Strategy for Asthma marker discovery in inflammatory lung Management and Prevention. (Update diseases. Eur Respir J. 2013;42(3):802–25. 2016). 11. Available from: https://ghr.nlm.nih.gov/ 2. Lemanske RF, Mauger DT, Sorkness primer/genomicresearch/pharmacogenomic CA, Jackson DJ, Boehmer SJ, Martinez FD, s. et al. Step-up therapy for children with uncontrolled asthma receiving inhaled 12. Halapi E, Gudbjartsson DF, Jonsdottir corticosteroids. N Engl J Med. 2010; 362 G.M., Bjornsdottir U.S., Thorleifsson G, (11):975–85. Helgadottir H et al. A sequence variant on 17q21 is associated with age at onset and 3. Szefler S.J., Phillips B.R., Martinez F.D., severity of asthma. Eur J Hum Genet. 2010; Chinchilli V.M., Lemanske R.F., Strunk 18(8):902–8. R.C., et al. Characterization of within- subject responses to fluticasone and 13. Moffatt MF, Gut IG, Demenais F, montelukast in childhood asthma. J Allergy Strachan DP, Bouzigon E, Heath S et al. A Clin Immunol. 2005;115(2):233–42. large-scale, consortium-based genomewide association study of asthma. N Engl J Med. 4. Farzan N, Vijverberg S.J.H, Arets H.G., 2010;363(13):1211–21. Raaijmakers J.A.M, Maitland-van der Zee A.H. Pharmacogenomics of inhaled cortico- 14. Zuurhout M.J., Vijverberg S.J., steroids and leukotriene modifiers: A sys- Raaijmakers J.A., Koenderman L, Postma tematic review. Clin Exp Allergy J. 2017; DS, Koppelman GH, et al. Arg16 ADRB2 47(2):271–93. genotype increases the risk of asthma exacerbation in children with a reported use 5. Holgate ST, Polosa R. The mechanisms, of long-acting 2-agonists: Results of the diagnosis, and management of severe pacman cohort. Pharmacogenomics. 2013; asthma in adults. Lancet. 2006;368 (9537): 14(16):1965–71. 780–93. 15. Wang Y, Tong C, Wang Z, Mauger D, 6. Haldar P, Pavord ID, Shaw DE, Berry Tantisira K.G., Israel E, et al. MA, Thomas M, Brightling CE, et al. Cluster Pharmacodynamic genomewide association Analysis and Clinical Asthma Phenotypes. study identifies new responsive loci for Am J Respir Crit Care Med. 2008; 178 (3): glucocorticoid intervention in asthma. 218–24. Pharmacogenomics J. 2015;15(5):422–9.
7. O’Byrne PM. POINT: Will New Anti- 16. García-Sánchez A, Isidoro-García M, eosinophilic Drugs Be Useful in Asthma García-Solaesa V, Sanz C, Hernández- Management? Yes. Chest. 2017;151(1): Hernández L, Padrón-Morales J, et al. 14–7. Genome-wide association studies (GWAS) and their importance in asthma. Allergol 8. Barnes PJ. COUNTERPOINT: Will New Immunopathol (Madr). 2015;43(6):601–8. Antieosinophilic Drugs Be Useful in Asthma Management? No. Chest. 2017;151(1): 17– 17. Zhang Y, Moffatt M.F., Cookson W.O.C. 20. genetic and genomic approaches to asthma: new insights for the origins. Curr Opin Pulm 9. Wanichthanarak K, Fahrmann J.F., Med. 2012;18(1):6–13. Grapov D. genomic, proteomic, and meta- bolomic data integration strategies. Bio- 18. Moffatt M.F., Kabesch M, Liang L, mark insights. SAGE publications; 2015; Dixon A.L et al. Genetic variants regulating 10(Suppl 4):1–6. ORMDL3 expression contribute to the risk of childhood asthma. Nature. 2007; 448(7152): 10. Wheelock CE, Goss VM, Balgoma D, 470–3. Nicholas B, Brandsma J, Skipp PJ, et al.
35
19. Ramasamy A, Kuokkanen M, 28. Leusink M, Vijverberg SJH, Koenderman Vedantam S, Gajdos ZK, Couto Alves A, L, Raaijmakers JAM, de Jongste JC, Sterk Lyon HN, et al. Genome-wide association PJ et al. Genetic variation in uncontrolled studies of asthma in population based childhood asthma despite ICS treatment. cohorts confirm known and suggested loci Pharmacogenomics J. 2015;1–6. and identify an additional association near HLA. Perry JR, editor. PLoS One. 29. Berce V, Kozmus C.E.P, Potočnik U. 2012;7(9): e44008. Association among ORMDL3 gene exp- ression, 17q21 polymorphism and response 20. Ferreira MA, Vonk JM, Baurecht H, to treatment with inhaled corticosteroids in Marenholz I, Tian C, Hoffman JD, et al. children with asthma. Pharmacogenomics J. Shared genetic origin of asthma, hay fever 2013;13(6):523–9. and eczema elucidates allergic disease biology. Nat Genet. 2017;49(12):1752–7. 30. Vijverberg S.J.H, Farzan N, Slob E.M.A, Neerincx A.H., Maitland-van der Zee A.H. 21. Yan Q, Brehm J, Pino-Yanes M, Forno Treatment response heterogeneity in E, Lin J, Oh S.S., et al. A meta-analysis of asthma: the role of genetic variation. Expert genomewide association studies of asthma Rev Respir Med. 2018;12(1):55–65. in Puerto Ricans. Eur Respir J. 2017;49(5): 1601505. 31. Basu K, Palmer C.N.A, Tavendale R, Lipworth B.J, Mukhopadhyay S. Adrenergic 22. Galanter J, Choudhry S, Eng C, Nazario β2-receptor genotype predisposes to S, Rodríguez-Santana JR, Casal J et al. exacerbations in steroid-treated asthmatic ORMDL3 gene is associated with asthma in patients taking frequent albuterol or three ethnically diverse populations. Am J salmeterol. J Allergy Clin Immunol. 2009; Respir Crit Care Med. 2008;177(11):1194– 124(6):1188–1194.e3. 200. 32. Lipworth B.J., Basu K, Donald H.P., 23. Andiappan AK, Sio YY, Lee B, Suri BK, Tavendale R, Macgregor DF, Ogston SA, et Matta SA, Lum J et al. Functional variants al. Tailored second-line therapy in asthmatic of 17q12-21 are associated with allergic children with the Arg(16) genotype. Clin Sci asthma but not allergic rhinitis. J Allergy (Lond). 2013;124(8):521–8. Clin Immunol. 2016;137(3):758–766.e3. 33. Turner S, Francis B, Vijverberg S, Pino- 24. Bisgaard H, Bønnelykke K, Sleiman Yanes M, Maitland-van der Zee A.H., Basu P.M.A., Brasholt M, Chawes B, Kreiner- K et al. Childhood asthma exacerbations Møller E et al. Chromosome 17q21 gene and the Arg16 β2-receptor polymorphism: A variants are associated with asthma and meta-analysis stratified by treatment. J exacerbations but not atopy in early Allergy Clin Immunol. 2016;138 (1):107- childhood. Am J Respir Crit Care Med. 113.e5. 2009;179(3):179–85. 34. Park T.J, Park J.S, Cheong H.S., Park 25. Binia A, Khorasani N, Bhavsar PK, B.L, Kim L.H., Heo J.S., et al. Genome-wide Adcock I, Brightling CE, Chung KF et al. association study identifies ALLC poly- Chromosome 17q21 SNP and severe morphisms correlated with FEV₁ change by asthma. J Hum Genet. 2011;56(1):97–8. corticosteroid. Clin Chim Acta. 2014;436: 20–6. 26. Smit LAM, Bouzigon E, Pin I, Siroux V, Monier F, Aschard H, et al. 17q21 variants 35. Tantisira KG, Lasky-Su J, Harada M, modify the association between early Murphy A, Litonjua AA, Himes BE, et al. respiratory infections and asthma. Eur Genomewide association between GLCCI1 Respir J. 2010;36(1):57–64. and response to glucocorticoid therapy in asthma. N Engl J Med. 2011;365(13):1173– 27. Tavendale R, Macgregor DF, Mukho- 83. padhyay S, Palmer CNA. A polymorphism controlling ORMDL3 expression is 36. Tantisira KG, Damask A, Szefler SJ, associated with asthma that is poorly Schuemann B, Markezich A, Su J, et al. controlled by current medications. J Allergy Genome-wide association identifies the T Clin Immunol. 2008;121(4):860–3.
36 gene as a novel asthma pharmacogenetic 45. Sampson A.P., Siddiqui S, Buchanan locus. Am J Respir Crit Care Med. 2012; D, Howarth P.H., Holgate S.T., Holloway 185(12):1286–91. J.W., et al. Variant LTC(4) synthase allele modifies cysteinyl leukotriene synthesis in 37. Park H.W, Dahlin A, Tse S, Duan Q., eosinophils and predicts clinical response Schuemann B, Martinez F.D, et al. Genetic to zafirlukast. Thorax. 2000; 55 Suppl 2: predictors associated with improvement of S28-31. asthma symptoms in response to inhaled corticosteroids. J Allergy Clin Immunol. 46. Dahlin A, Litonjua A, Irvin CG, Peters 2014;133(3):664–669.e5. SP, Lima JJ, Kubo M, et al. Genome-wide association study of leukotriene modifier 38. Wang Y, Tong C, Wang Z, Wang Z, response in asthma. Pharmacogenomics J. Mauger D, Tantisira K.G., et al. Pharmaco- 2015;1–7. dynamic genome wide association study identifies new responsive loci for gluco- 47. Dahlin A, Litonjua A, Lima J.J, Tamari corticoid intervention in asthma. Pharmaco- M, Kubo M, Irvin C.G., et al. Genome- genomics J. 2015;15(5):422–9. wide association study identifies novel pharmacogenomic Loci For Therapeutic 39. Mosteller M, Hosking L, Murphy K, Response to Montelukast in Asthma. PLoS Shen J, Song K, Nelson M, et al. No One. 2015;10(6):e0129385. evidence of large genetic effects on steroid response in asthma patients. J Allergy Clin 48. Park H.W., Dahlin A, Tse S, Duan Q.L., Immunol. 2016;139(3):797-803.e7. Schuemann B, Martinez FD, et al. Genome wide association study of leukotriene 40. Koster ES, Maitland-van der Zee A-H, modifier response in asthma. J Allergy Clin Tavendale R, Mukhopadhyay S, Vijverberg Immunol. 2014;133(3):664–9.e5. SJH, Raaijmakers JAM, et al. FCER2 T2206C variant associated with chronic 49. Vijverberg S.J., Pijnenburg M.W, symptoms and exacerbations in steroid- Hövels A.M, Koppelman G.H., Maitland- treated asthmatic children. Allergy. 2011; van der Zee A.H. The need for precision 66(12):1546–52. medicine clinical trials in childhood asthma: rationale and design of the PUFFIN trial. 41. Wu AC, Tantisira K, Li L, Schuemann Pharmacogenomics. 2017;18(4):393–401. B, Weiss ST, Fuhlbrigge AL. Predictors of symptoms are different from predictors of 50. Myers R.A., Himes B.E., Gignoux C.R., severe exacerbations from asthma in Yang J.J., Gauderman W.J., Rebordosa C, children. Chest. 2011;140(1):100–7. et al. Further replication studies of the EVE Consortium meta analysis identifies 2 42. Tantisira KG, Silverman ES, Mariani TJ, asthma risk loci in European Americans. J Xu J, Richter BG, Klanderman BJ, et al. Allergy Clin Immunol. 2012;130(6):1294– FCER2: a pharmacogenetic basis for 301. severe exacerbations in children with asthma. J Allergy Clin Immunol. 2007; 51. Vijverberg S.J., Turner S.W., Palmer C. 120(6):1285–91. N, Tantisira K.G, Maitland van der Zee A.H. Realizing Personalized Medicine in 43. Tantisira KG, Lima J, Sylvia J, Asthmatic Children Requires Large-Scale Klanderman B, Weiss S.T. 5-lipoxygenase Collaboration. Pediatr Ther. OMICS pharmacogenetics in asthma: overlap with International; 2015;05(02). Cys-leukotriene receptor antagonist loci. Pharmacogenet Genomics. 2009; 19(3): 52. Farzan N, Vijverberg SJ, Andiappan 244–7. AK, Arianto L, Berce V, Blanca-López N et al. Rationale and design of the multiethnic 44. Telleria JJ, Blanco-Quiros A, Varillas D, Pharmacogenomics in Childhood Asthma Armentia A, Fernandez-Carvajal I, Jesus consortium. Pharmacogenomics. 2017;18 Alonso M, et al. ALOX5 promoter genotype (10):931–43. and response to montelukast in moderate persistent asthma. Respir Med. 2008; 102 53. Berger SL, Kouzarides T, Shiekhattar (6):857–61. R, Shilatifard A. An operational definition
37
of epigenetics. Genes Dev. 2009;23(7): human airway epithelium. Proc Natl Acad 781–3. Sci. 2009;106(7):2319–24.
54. Lewis BP, Burge CB, Bartel DP. 63. Jahreis S, Trump S, Bauer M, Bauer T, Conserved seed pairing, often flanked by Thürmann L, Feltens R, et al. Maternal adenosines, indicates that thousands of phthalate exposure promotes allergic human genes are microRNA targets. Cell. airway inflammation over 2 generations 2005;120(1):15–20. through epigenetic modifications. J Allergy Clin Immunol. 2017; 141(2):741-753. 55. Jones P.A. Functions of DNA methyla- tion: islands, start sites, gene bodies and 64. Rossnerova A, Tulupova E, Tabashidze beyond. Nat Rev Genet. 2012; 13(7):484– N, Schmuczerova J, Dostal M, Rossner P, 92. et al. Factors affecting the 27K DNA methyl- ation pattern in asthmatic and healthy 56. Law J.A., Jacobsen S.E. Establishing, children from locations with various en- maintaining and modifying DNA methyl- vironments. Mutat Res. 2013;741–742: 18– ation patterns in plants and animals. Nat 26. Rev Genet. 2010;11(3):204–20. 65. Perera F, Tang W, Herbstman J, Tang 57. Arzenani M.K., Zade A.E., Ming Y, D, Levin L, Miller R, et al. Relation of DNA Vijverberg S.J.H, Zhang Z, Khan Z, et al. methylation of 5’-CpG island of ACSL3 to Genomic DNA hypomethylation by histone transplacental exposure to airborne deacetylase inhibition implicates DNMT1 polycyclic aromatic hydrocarbons and nuclear dynamics. Mol Cell Biol. 2011;31 childhood asthma. Suter CM, editor. PLoS (19):4119–28. One. 2009;4(2):e4488.
58. Ito K, Caramori G, Lim S, Oates T, 66. Bauer T, Trump S, Ishaque N, Chung KF, Barnes PJ, et al. Expression Thürmann L, Gu L, Bauer M, et al. Environ- and Activity of Histone Deacetylases in ment induced epigenetic re-programming in Human Asthmatic Airways. Am J Respir genomic regulatory elements in smoking Crit Care Med. 2002;166(3):392–6. mothers and their children. Mol Syst Biol. 2016;12(3):861–861. 59. Xiao C, Biagini Myers JM, Ji H, Metz K, Martin LJ, Lindsey M, et al. Vanin-1 67. Breton C V, Byun H-M, Wenten M, Pan expression and methylation discriminate F, Yang A, Gilliland FD. Prenatal tobacco pediatric asthma corticosteroid treatment smoke exposure affects global and gene- response. J Allergy Clin Immunol. 2015;136 specific DNA methylation. Am J Respir Crit (4):923–931.e3. Care Med. 2009;180(5):462–7.
60. Michel S, Busato F, Genuneit J, 68. Tager I.B., Ngo L, Hanrahan J.P. Pekkanen J, Dalphin J.C, Riedler J, et al. Maternal smoking during pregnancy. Farm exposure and time trends in early Effects on lung function during the first 18 childhood may influence DNA methylation months of life. Am J Respir Crit Care Med. in genes related to asthma and allergy. 1995;152(3):977–83. Allergy. 2013;68(3):355–64. 69. Burke H, Leonardi-Bee J, Hashim A, 61. Liu F, Killian JK, Yang M, Walker RL, Pine-Abata H, Chen Y, Cook D.G, et al. Hong JA, Zhang M, et al. Epigenomic Prenatal and passive smoke exposure and alterations and gene expression profiles in incidence of asthma and wheeze: respiratory epithelia exposed to cigarette systematic review and meta-analysis. smoke condensate. Oncogene. 2010;29 Pediatrics. 2012;129(4):735–44. (25): 3650–64. 70. Leung D.Y.M., Szefler S.J., Associate 62. Schembri F, Sridhar S, Perdomo C, Editors of the JACI. In utero smoke (IUS) Gustafson A.M., Zhang X, Ergun A, et al. exposure has been associated with MicroRNAs as modulators of smoking- increased prevalence of asthma and induced gene expression changes in reduced lung function in healthy children. J
38
Allergy Clin Immunol. 2010;126(3):481–2. 79. Gaffin JM, Raby BA, Petty CR, Hoffman EB, Baccarelli AA, Gold DR, et al. β-2 71. Zacharasiewicz A. Maternal smoking in Adrenergic receptor gene methylation is pregnancy and its influence on childhood associated with decreased asthma severity asthma. ERJ Open Res. 2016;2(3):00042- in inner city Schoolchildren. Clin Exp 2016. Allergy. 2014;44(5):681–9.
72. Munthe-Kaas M.C., Torjussen T.M., 80. Curtin JA, Simpson A, Belgrave D, Gervin K, Lødrup Carlsen K.C., Carlsen Semic-Jusufagic A, Custovic A, Martinez KH, Granum B, et al. CD14 polymorphisms FD. Methylation of IL-2 promoter at birth and serum CD14 levels through childhood: alters the risk of asthma exacerbations a role for gene methylation? J Allergy Clin during childhood. Clin Exp Allergy. 2013; Immunol. 2010;125(6):1361–8. 43(3):304–11.
73. Somineni H.K., Zhang X, Biagini Myers 81. Michel S, Busato F, Genuneit J, J.M., Kovacic M.B., Ulm A, Jurcak N, et al. Pekkanen J, Dalphin J-C, Riedler J, et al. Ten-eleven translocation 1 (TET1) methyl- Farm exposure and time trends in early ation is associated with childhood asthma childhood may influence DNA methylation and traffic related air pollution. J Allergy in genes related to asthma and allergy. Clin Immunol. 2016;137(3):797–805.e5. Allergy. 2013;68(3):355–64.
74. Stefanowicz D, Hackett TL, Garmaroudi 82. Morales E, Bustamante M, Vilahur N, FS, Günther OP, Neumann S, Sutanto EN, Escaramis G, Montfort M, de Cid R, et al. et al. DNA methylation profiles of airway DNA hypomethylation at ALOX-12 is epithelial cells and PBMCs from healthy, associated with persistent wheezing in atopic and asthmatic children. Christensen childhood. Am J Respir Crit Care Med. BC, editor. PLoS One. 2012;7(9):e44213. 2012;185(9):937–43.
75. Yang I.V., Pedersen B.S, Liu A, 83. Riedler J, Braun-Fahrländer C, Eder W, O’Connor GT, Teach SJ, Kattan M, et al. Schreuer M, Waser M, Maisch S, et al. DNA methylation and childhood asthma in Exposure to farming in early life and the inner city. J Allergy Clin Immunol. 2015; development of asthma and allergy: a 136(1):69–80. cross-sectional survey. Lancet. 2001;358 (9288):1129–33. 76. Harb H, Raedler D, Ballenberger N, Böck A, Kesper DA, Renz H, et al. 84. von Mutius E, Vercelli D. Farm living: Childhood allergic asthma is associated effects on childhood asthma and allergy. with increased IL-13 and FOXP3 histone Nat Rev Immunol. 2010;10(12):861–8. acetylation. J Allergy Clin Immunol. 2015; 136(1):200–2. 85. Sleiman PMA, Flory J, Imielinski M, Bradfield JP, Annaiah K, Willis-Owen SAG, 77. Hew K.M., Walker A.I., Kohli A, Garcia et al. Variants of DENND1B associated with M, Syed A, Mc-Donald-Hyman C, et al. asthma in children. N Engl J Med. 2010;362 Childhood exposure to ambient polycyclic (1):36–44. aromatic hydrocarbons is linked to epigenetic modifications and impaired 86. Ege MJ, Bieli C, Frei R, van Strien RT, systemic immunity in T cells. Clin Exp Riedler J, Ublagger E, et al. Prenatal farm Allergy. 2015;45(1):238–48. exposure is related to the expression of receptors of the innate immunity and to 78. Naumova A.K., Al Tuwaijri A, Morin A, atopic sensitization in school-age children. Vaillancourt V.T., Vaillancout V.T., Madore J Allergy Clin Immunol. 2006;117(4):817– A.M., et al. Sex- and age-dependent DNA 23. methylation at the 17q12-q21 locus as- sociated with childhood asthma. Hum 87. Strachan DP. Hay fever, hygiene, and Genet. 2013;132(7):811–22. household size. BMJ. 1989;299 (6710): 1259–60.
39
88. van Tilburg Bernardes E, Arrieta M.C. 98. Joubert BR, Håberg SE, Nilsen RM, Hygiene hypothesis in asthma develop- Wang X, Vollset SE, Murphy SK, et al. 450K ment: Is hygiene to blame? Arch Med Res. epigenome-wide scan identifies differential 2017;48(8):717-726. DNA methylation in newborns related to maternal smoking during pregnancy. 89. Slaats GGG, Reinius LE, Alm J, Kere J, Environ Health Perspect. 2012;120(10): Scheynius A, Joerink M. DNA methylation 1425–31. levels within the CD14 promoter region are lower in placentas of mothers living on a 99. Richmond RC, Simpkin AJ, Woodward farm. Allergy. 2012;67(7):895–903. G, Gaunt TR, Lyttleton O, McArdle WL, et al. Prenatal exposure to maternal smoking 90. Wright SD, Ramos RA, Tobias PS, and offspring DNA methylation across the Ulevitch RJ, Mathison JC. CD14, a receptor lifecourse: findings from the Avon for complexes of lipopolysaccharide (LPS) Longitudinal Study of Parents and Children and LPS binding protein. Science. 1990; (ALSPAC). Hum Mol Genet. 2015;24(8): 249(4975):1431–3. 2201–17.
91. Kitchens RL. Role of CD14 in cellular 100. Rehan V.K., Liu J, Naeem E, Tian J, recognition of bacterial lipopolysac- Sakurai R, Kwong K, et al. Perinatal charides. Chem Immunol. 2000; 74: 61–82. nicotine exposure induces asthma in second generation offspring. BMC Med. 92. Tapping RI, Tobias PS. Soluble CD14- 2012;10(1):129. mediated cellular responses to lipopolysac- charide. Chem Immunol. 2000; 74:108–21. 101. Yang I.V., Pedersen B.S, Liu A, O’Connor G.T., Teach S.J., Kattan M, et al. 93. Vercelli D. Gene environment inter- DNA methylation and childhood asthma in actions in asthma and allergy: the end of the inner city. J Allergy Clin Immunol. 2015; the beginning? Curr Opin Allergy Clin 136(1):69–80. Immunol. 2010;10(2):145–8. 102. Barnes P.J. How corticosteroids 94. Joerink M, Oortveld MAW, Stenius F, control inflammation: quintiles prize Lecture Rindsjö E, Alm J, Scheynius A. Lifestyle 2005. Br J Pharmacol. 2006;148 (3):245– and parental allergen sensitization are 54. reflected in the intrauterine environment at gene expression level. Allergy. 2010;65 103. Pitari G, Malergue F, Martin F, Philippe (10):1282–9. JM, Massucci M.T., Chabret C, et al. Pantetheinase activity of membrane-bound 95. Cook DG, Strachan DP. Health effects Vanin-1: lack of free cysteamine in tissues of passive smoking. 3. Parental smoking of Vanin-1 deficient mice. FEBS Lett. 2000; and prevalence of respiratory symptoms 483(2–3):149–54. and asthma in school age children. Thorax. 1997;52(12):1081–94. 104. Katayama S, Tomaru Y, Kasukawa T, Waki K, Nakanishi M, Nakamura M, et al. 96. Joubert BR, Haberg SE, Bell DA, Nilsen Antisense transcription in the mammalian RM, Vollset SE, Midttun O, et al. Maternal transcriptome. Science. 2005;309 (5740): smoking and DNA methylation in newborns 1564–6. In utero effect or epigenetic Inheritance? Cancer Epidemiol Biomarkers Prev. 2014; 105. Wang Z, Gerstein M, Snyder M. RNA- 23(6):1007–17. Seq: a revolutionary tool for transcript- omics. Nat Rev Genet. 2009;10 (1):57–63. 97. Joubert B.R., Felix J.F., Yousefi P, Bakulski K.M., Just A.C., Breton C, et al. 106. Bérubé J-C, Bossé Y. Future clinical DNA methylation in newborns and Maternal implications emerging from recent genome- Smoking in Pregnancy: Genome-wide wide expression studies in asthma. Expert Consortium Meta-analysis. Am J Hum Rev Clin Immunol. 2014;10(8):985–1004. Genet. 2016;98(4):680–96.
40
107. Hänzelmann S, Castelo R, Guinney J. sputum transcriptomics in U-BIOPRED. Eur GSVA: gene set variation analysis for Respir J. 2017;49(2):1602135. microarray and RNA-seq data. BMC Bio- informatics. 2013;14(1):7. 116. Hekking PP, Loza M, Pavlidis S, De Meulder B, Al. E. Pathway discovery using 108. Wagener A.H., Yick C.Y., Brinkman P, transcriptomic profiles in adult-onset severe van der Schee MP, Fens N, Sterk PJ. asthma. Toward composite molecular signatures in the phenotyping of asthma. Ann Am Thorac 117. Persson H, Kwon A.T., Ramilowski Soc. 2013;10 Suppl:S 197-205. J.A., Silberberg G, Söderhäll C, Orsmark- Pietras C, et al. Transcriptome analysis of 109. McCarthy D.J., Smyth G.K. Testing controlled and therapy-resistant childhood significance relative to a fold- change asthma reveals distinct gene expression threshold is a TREAT. Bioinformatics. 2009 profiles. J Allergy Clin Immunol. 2015;136 ;25 (6):765–71. (3):638–48.
110. Michiels S, Kramar A, Koscielny S. 118. Subrata LS, Bizzintino J, Mamessier Multi-dimensionality of microarrays: statisti- E, Bosco A, McKenna K.L, Wikström ME, et cal challenges and (im)possible solutions. al. Interactions between innate antiviral and Mol Oncol. 2011;5(2):190–6. atopic immunoinflammatory pathways precipitate and sustain asthma exacerba- 111. Hekking PP, Loza MJ, Pavlidis S, De tions in children. J Immunol. 2009; 183(4): Meulder B, Lefaudeux D, Baribaud F, et al. 2793–800. Transcriptomic gene signatures associated with persistent airflow limitation in patients 119. Sordillo J.E., Kelly R, Bunyavanich S, with severe asthma. Eur Respir J. 2017;50 Mc Geachie M, Qiu W, Croteau-Chonka (3): pii:1602298. DC, et al. Genome-wide expression profiles identify potential targets for gene environ- 112. Fleming L, Murray C, Bansal A.T., ment interactions in asthma severity. J Hashimoto S, Bisgaard H, Bush A, et al. Allergy Clin Immunol. 2015; 136(4):885– The burden of severe asthma in childhood 892.e2. and adolescence: results from the paedia- tric U-BIOPRED cohorts. Eur Respir J. 120. Sordillo J.E., Kelly R, Bunyavanich S, 2015;46(5):1322–33. McGeachie M, Qiu W, Croteau-Chonka DC, et al. Genome-wide expression profiles 113. Bigler J, Boedigheimer M, Schofield identify potential targets for gene environ- JPR, Skipp PJ, Corfield J, Rowe A, et al. A ment interactions in asthma severity. J Severe Asthma Disease Signature from Allergy Clin Immunol. 2015;136 (4):885– Gene Expression Profiling of Peripheral 892.e2. Blood from U-BIOPRED Cohorts. Am J Respir Crit Care Med. 2017;195(10):1311– 121. Bosco A, Ehteshami S, Panyala S, 20. Martinez F.D. Interferon regulatory factor 7 is a major hub connecting interferon 114. Kuo C.H.S, Pavlidis S, Loza M, mediated responses in virus induced Baribaud F, Rowe A, Pandis I, et al. A asthma exacerbations in vivo. J Allergy Clin Transcriptome-driven Analysis of Epithelial Immunol. 2012;129(1):88–94. Brushings and Bronchial Biopsies to Define Asthma Phenotypes in U-BIOPRED. Am J 122. Pearson G, Robinson F, Beers Gibson Respir Crit Care Med. 2017;195(4):443–55. T, Xu B, Karandikar M, Berman K, et al. Mitogen-Activated Protein (MAP) Kinase 115. Kuo C.H.S, Pavlidis S, Loza M, Pathways: Regulation and Physiological Baribaud F, Rowe A, Pandis I, et al. T- Functions. Endocr Rev. 2001;22(2):153– helper cell type 2 (Th2) and non-Th2 83. molecular phenotypes of asthma using respiratory diseases. Paediatr Respir Rev. 123. Holt P.G., Sly P.D. Viral infections and 2014;15(4):348–54. atopy in asthma pathogenesis:
41
new rationales for asthma prevention and 133. Gruzieva O, Merid SK, Melén E. An treatment. Nat Med. 2012;18(5):726–35. update on epigenetics and childhood respiratory diseases. Paediatr Respir Rev. 124. Bosco A, Ehteshami S, Panyala S, 2014;15(4):348–54. Martinez F.D. Interferon regulatory factor 7 is a major hub connecting interferon- 134. Castro-Rodriguez JA, Rodrigo GJ. mediated responses in virus-induced Efficacy of inhaled corticosteroids in infants asthma exacerbations in vivo. J Allergy Clin and preschoolers with recurrent wheezing Immunol. 2012;129(1):88–94. and asthma: a systematic review with meta- analysis. Pediatrics. 2009;123(3):e519-25. 125. Robays L.J, Maes T, Lebecque S, Lira S.A, Kuziel W.A., Brusselle G.G., et al. 135. Gustafsson M, Nestor CE, Zhang H, Chemokine receptor CCR2 but not CCR5 Barabási A.L, Baranzini S, Brunak S, et al. or CCR6 mediates the increase in Modules, networks and systems medicine pulmonary dendritic cells during allergic for understanding disease and aiding airway inflammation. J Immunol. 2007;178 diagnosis. Genome Med. 2014;6(10):82. (8):5305–11. 136. McShane L.M., Cavenagh M.M, Lively 126. Lalani T, Simmons RK, Ahmed AR. T.G., Eberhard D.A., Bigbee W.L., Williams Biology of IL-5 in health and disease. Ann P.M, et al. Criteria for the use of omics- Allergy Asthma Immunol. 1999;82(4):317- based predictors in clinical trials: ex- 32. planation and elaboration. BMC Med. 2013; 11(1):220. 127. Goswami R, Kaplan M.H. A brief history of IL-9. J Immunol. 2011;186(6): 137. Custovic A, Ainsworth J, Arshad H, 3283–8. Bishop C, Buchan I, Cullinan P, et al. The Study Team for Early Life Asthma 128. Gong F, Pan Y.H, Huang X, Zhu H.Y, Research (STELAR) consortium “Asthma Jiang D.L. From bench to bedside: e-lab”: team science bringing data, Therapeutic potential of interleukin-9 in the methods and investigators together. treatment of asthma. Exp Ther Med. 2017; Thorax. 2015;70(8):799–801. 13(2):389–94.
129. Bauer T, Trump S, Ishaque N, Thürmann L, Gu L, Bauer M, et al. Environment-induced epigenetic repro- gramming in genomic regulatory elements in smoking mothers and their children. Mol Syst Biol. 2016;12(3):861.
130. Pedersen SE, Hurd SS, Lemanske RF, Becker A, Zar HJ, Sly PD, et al. Global strategy for the diagnosis and management of asthma in children 5 years and younger. Pediatr Pulmonol. 2011;46(1):1–17.
131. Martinez FD, Wright AL, Taussig LM, Holberg CJ, Halonen M, Morgan WJ. Asthma and Wheezing in the First Six Years of Life. N Engl J Med. 1995;332 (3): 133–8.
132. Wang B, Mezlini AM, Demir F, Fiume M, Tu Z, Brudno M, et al. Similarity network fusion for aggregating data types on a genomic scale. Nat Methods. 2014;11(3): 333–7.
42
43
44
CHAPTER 2.2
Pharmacogenomics of inhaled corticosteroids and leukotriene modifiers: a systematic review
Farzan N
Vijverberg SJ
Arets HG
Raaijmakers JAM
Maitland-van der Zee AH
Clinical and Experimental Allergy (CEA), 2017;47(2):271-293
45
Abstract
Background: Pharmacogenetics studies of anti-inflammatory medication of asthma have expanded rapidly in recent decades, but the clinical value of their findings remains limited.
Objective: To perform a systematic review of pharmacogenomics and pharmacogenetics of inhaled corticosteroids (ICS) and leukotriene modifiers (LTMs) in asthmatic patients.
Methods: Articles published between 1999 and June 2015 were searched using PubMed and Embase. Pharmacogenomics/genetics studies of asthmatic patients using ICS or LTMs were included if ≥1 of the following outcomes were studied: lung function, exacerbation rates or asthma symptoms. The studies of SNPs that had been replicated at least once were assessed in more detail.
Results: In total, 59 publications were included in the systematic review: 26 addressed LTMs (including two Genome-Wide Association Studies [GWAS]) and 33 addressed ICS (including four GWAS). None of the GWAS reported similar results. Furthermore, none of the SNPs assessed in candidate gene studies were identified in a GWAS. No consistent reports were found for candidate gene studies of LTMs. In candidate gene studies of ICS, the most consistent results were found for rs28364072 in FCER2. This Single Nucleotide Polymorphism [SNP] was associated with all three outcomes of poor response and the largest effect was reported with the risk of exacerbations (hazard ratio, 3.95; 95% CI, 1.64- 9.51).
Conclusions: There is a lack of replication of genetic variants associated with poor ICS or LTM response. The most consistent results were found for the FCER2 gene (encoding for a low affinity IgE receptor (CD23)) and poor ICS response. Larger studies with well phenotyped patients are needed to assess the clinical applicability of ICS and LTM pharmacogenomics/genetics.
46
Introduction
Asthma is a chronic inflammatory disease that affects approximately 1-18% of the world’s population (1). A wide variability in patient response to asthma control medications, such as inhaled corticosteroids (ICS) and leukotriene modifiers (LTM) has been reported (2–4). Approximately 35 to 40% of the patients receiving ICS for 8-12 weeks do not show an improvement in lung function (2,3). Moreover, approximately 10% of the asthma patients receiving ICS experience severe asthma symptoms despite regular use of this medication (5). In the United States, the mean annual proportion of adult asthmatic patients using ICS has increased from 39.9% (1998-1999) to 51.2% (2008-2009) and the proportion of LTM users has increased from 12.4% to 20.4% (6).
ICS are the preferred first-line treatment for persistent asthma (GINA step 2) (1). Glucocorticoids (GCs) suppress inflammation in the airways by binding to the glucocorticoid receptor (GR). Subsequently, the receptor complex translocates to nucleus and represses the activation of inflammatory genes, for example by the recruitment of histone deacetylase-2 (HDAC2) and reversing histone acetylation (7). GCs also modulate the activity of genes encoding proteins involved in airway epithelial barrier and airway remodeling (8) and enhance the transcription of several anti-inflammatory genes (9).
LTMs are recommended for patients with moderate to severe persistent asthma, as one of the possible add-on therapies to ICS in the step 3 of the GINA guideline (4,10). These drugs have anti-inflammatory effects and reduce bronchoconstriction as well. LTMs consist of two groups of medications: cysLT1 receptor antagonists (i.e. montelukast, pranlukast and zafirlukast) and inhibitors of the 5-lipoxygenase pathway (i.e. zileuton, ABT-761) (11). It has previously been reported in various studies that 35 to 78% of patients do not respond to LTMs with regard to lung function improvements (2,3,12).
Continued environmental exposures, poor adherence to medication, and misdiagnosis influence treatment response in asthmatic patients (13,14). However, genetic variation might also be an important factor (4,10). Pharmacogenetics assesses the relation between variations in single genes and treatment response. Pharmacogenomics was defined by the National Institutes of Health (NIH) in 2016 as “the influence of genetic variation on the individual’s response to medication to design and develop safer treatment options with higher efficacy and less adverse effects based on the genetic profile of the patients” (15). Pharmacogenomics and pharmacogenetics are often used interchangeably in literature. In this paper, we will use the term pharmacogenomics.
47
In the last decade, advanced genotyping technologies have yielded a large number of pharmacogenomics studies. Nevertheless, the potential clinical value of the findings from these studies remains unclear. In this systematic review, we provide an up-to-date assessment of the scientific evidence regarding the use of genetic testing before starting ICS or LTMs by comparing the study designs, outcome definitions, study populations and reported findings in this field.
Methods
Articles published from 1999 until the end of June 2015 were searched in PubMed and Embase using specified search terms (Table S1 and S2). Publications were included based upon keywords in their titles and abstracts that indicated that they had investigated genetic effects on the response to ICS or LTM based on lung function tests (i.e. forced expiratory volume in one second [FEV1], either as absolute values or as a percentage of predicted [FEV1%pred] and airway hyper-responsiveness to methacholine), exacerbation rates and asthma symptoms (i.e. Asthma Control Test [ACT], and Asthma Control Questionnaire [ACQ]). Conference abstracts, articles assessing adverse drug effects and articles that had not published in English were excluded. We also screened review articles for possible missed publications. From all the publications selected the following data were extracted by author FN: year of publication, journal name, authors, ICS and LTM type, genes and Single Nucleotide Polymorphisms [SNPs], outcome, study design, ethnic backgrounds of patients, sample size, age range and duration of the study. We assessed the outcomes, population structure and the effect measures of the studies investigating the SNPs that had been replicated at least once. This systematic review is reported according to PRISMA guidance (16).
Pharmacogenomics of ICS
In the primary search, 146 studies assessing ICS pharmacogenomics were identified in PubMed and 315 studies were identified in Embase. Of these studies, 33 met the inclusion criteria (Figure1), with 29 being candidate gene studies (Table S3)(17,18,27–36,19,37– 45,20–26) and four being GWAS (46–49).
48
Records identified through database searching (PubMed [146] and EMBASE [315])
(n = 461) Identification
Records excluded because of duplication (n = 175)
Records screened
Screening (n = 286)
Records excluded (n = 254)
Articles assessed for eligibility
(n = 32)
Articles identified from
Eligibility references of articles
(n = 1)
Studies included after reading the full-text
(n = 33) Included
Figure 1. Inclusion flowchart for pharmacogenomics studies of ICS.
GWAS studies of ICS
Three out of the four GWAS (46–49) were conducted by the same research group. In all three publications (46–48), different numbers of Caucasian patients from the Single-
49
Nucleotide Polymorphism Health Association-Asthma Resource Project (SHARP) were included. SHARP included patients from three studies: The Childhood Asthma Manage- ment Program (CAMP) study (50), Childhood Asthma Research and Education (CARE) and Asthma Clinical Research Network (ACRN). Despite similarities in the study populations, there were differences in the study designs, outcome measurements, genotyping platforms and sample sizes (Table1). The overlap in the populations included in the three GWAS was not reported. Different genes and SNPs were identified to be associated with response to ICS in these three studies. Changes in the lung function measurements were assessed as the outcome in two out of the three studies (46,47). The first of the GWAS of pharmacogenomics of ICS was published in 2011. In this study, 118 asthmatic child-parent trios from the CAMP study were studied and a family-based association analysis was conducted (46). A SNP (rs37972 C>T) located in the GLCCI1 (Glucocorticoid-induced transcript 1 protein) gene was associated with changes in FEV1% predicted in children treated with budesonide. Subsequently, this SNP was genotyped in four separate populations (935 adult patients) and in three of them, patients homozygous for the wild-type allele (C) had an approximately 12% increase in FEV1% predicted as compared to the patients homozygous for the mutant allele (T) with an approximately 4% increase (pooled P value=7 x 10-4) (46). This SNP (rs37972) is in complete linkage disequilibrium (LD) with another rs37973 SNP in the same gene, which was found to influence GLCCI1 gene expression (46), yet the function of GLCCI1 remains largely unknown.
The second study by the same research group appeared in 2012. In it, researchers assessed the response to ICS by changes in FEV1% predicted in patients from the SHARP project (in total, 418 Caucasian children and adults) (47). In the GWAS analysis of the 418 patients, none of the SNPs reached a genome wide threshold. Subsequently, 47 SNPs identified in the GWAS with the lowest p values were genotyped in a replication population of asthmatic adults (n=407). Two SNPs, rs6456042 and rs3127412, had the lowest p values from this analysis (P=1.06 x 10-5 and P=6.13 x 10-6, respectively). These SNPs were found to be in high LD with three SNPs within functional regions of the T gene (rs3099266, rs1134481 and rs2305089). The most significant result was found for rs1134481 with wild-type homozygotes improving> 10% in FEV1% at the end of week eight compared to an improvement of only 4% in FEV1% in variant allele homozygotes (P=1.57 x 10-5). The GLCCI1 gene, which was identified in the previous study, was not found in the GWAS analysis. Differences in the study designs and genotyping platforms could have played a role here.
50
In the third GWAS (48), in 2014, three SNPs (rs2388639 in the gene LOC728792, rs10044254 in the gene FBXL7 [F-Box And Leucine-Rich Repeat Protein7] and rs1558726 in the gene RMST [Rhabdomyosarcoma 2 associated transcript]) were associated with self-reported asthma symptoms (symptom scores ranged from 0 to 3 [severe]) in children treated with ICS from discovery and replication phases (combined P values were 8.56 x 10-9 for rs2388639, 9.12 x 10-8 for rs10044254 and 1.02 x 10-5 for rs1558726). These associations were not replicated in adult patients (48). Alterations in the FBXL7 gene expression were shown in the presence of rs10044254 SNP in immortalized B cells(48). This gene encodes a member of the F-box protein family, which plays a role in phosphorylation-dependent ubiquitination of proteins (51). Homozygotes for the mutant allele for rs10044254 had significantly poorer responses compared with heterozygotes or homozygotes for the wild-type allele (median score of 1.14 in homozygotes for mutant allele vs -0.28 in homozygotes for the reference allele).
In the most recent GWAS of ICS in 2014, the changes in FEV1% from baseline measure was assessed in 189 Korean asthmatic patients (15-75 years) (49). A SNP (rs11123610) located within the ALLC (allantoicase) gene was associated with lung function improvement (P=3.6 x 10-7). Subsequently, 25 additional SNPs within ALLC were selected for genotyping. Among the selected SNPs, six (rs17017879, rs7558370, rs11123610, rs6754459, rs17445240 and rs13418767) were statistically significantly associated with the change in FEV1% in response to ICS treatment (p value < 1.0 x 10-5). The largest effect was found for rs11123610 with the differences in mean FEV1% higher than 20.3% among its genotypes. ALLC is located in chromosome 2q35 and encodes for the allantoicase enzyme, which is involved in the uric acid degradation pathway. However, this enzyme does not exist in mammals and its function is unclear (52).
51
52
Table 1. GWAS analysis of response to ICS and LTMs.
GWAS analysis of response to ICS Study Discovery Study Medication in Replication Definition SNPs chosen for Study outcome phase design & the discovery population of replication (gene, population duration phase response chromosome position)
Tantisira Caucasian Clinical Budesonide Caucasian Changes in Rs37972 (GLCCI1, Patients homozygous for et al. children trial 200 children FEV1 from 7p21.3) the wild-type allele (C) 2011 (CAMP)118 (16 µg twice daily (n=935): baseline had approximately 12% child-parent months) (CARE trial, improvements in FEV1% trios n=101) pred compared with the Adults: 4% increase in TT (Adult study, carriers after 4 to 8 n=385) weeks of treatment with (LOCCS, n=185) ICS (combined p value = (SOCS/SLIC, 7 x 10-4). n=264) Tantisira Caucasians Clinical CAMP: adults (n=407) Changes in rs3099266, rs1134481 Patients homozygous for et al. (n=418): trial (6-8 budesonide FEV1% and rs2305089 (T gene, the wild type allele of all 3 2012 children: weeks) 200 µg twice pred from 6q27) SNPs had a two to three- (CAMP and daily baseline fold increase in CARE trials) CARE: FEV1%pred compared to (n=239) fluticasone homozygotes for the Adults: propionate 100 mutant allele. Combined (ACRN trial) μg twice daily P values of study (n=179) ACRN: populations for the Triamcinolone rs1134481, rs2305089 400 μg twice and rs3099266 were 1.57 daily x 10-5, 2.3 x 10-4 and 1.1 x10-4 respectively.
52
H. Park Caucasian Clinical Budesonide Caucasian Asthma rs1558726 The combined P values et al. children trial (8 200 µg twice Children: symptoms (RMST,12q21), of rs2388639, 2014 (CAMP) weeks) daily (CARE, n= 77) rs2388639 (LOC728792) rs10044254 and (n=124) Adults:(LOCCS, and rs1558726 SNPs for the n=110) rs10044254 (FBXL7, pediatric CAMP and (ACRN, n=110) 5p15.1) CARE subjects were 8.56 × 10−9, 9.12 × 10−8and 1.02 x 10-5 respectively. Homozygotes for the mutant allele for rs10044254 had significantly poorer responses to treatment compared to the patients homozygous or heterozygous for the wild- type allele (increase of 1.14 (as median score) in homozygotes for the mutant allele vs -0.28 in homozygotes for the reference allele).
T. Park Korean Clinical 1000 μg of Same population Changes in 14 SNPs within ALLC rs17017879, rs7558370, et al. adults trial (4 fluticasone with the discovery FEV1% (from GWAS) and 11 rs11123610, rs6754459, 2014 (n=189) weeks) propionate daily phase additional SNPs in ALLC rs17445240 and (2q35) rs13418767 were significantly associated with change in FEV1% (P value < 1.0 x 10-5).
53
53
54
GWAS analysis of response to LTMs Dahlin African- Clinical Zileuton CR African-American Changes in rs12436663 (A/G) Patients carrying two et al. American trial (12 1200 mg twice and Caucasians: FEV1 (MRPP3, 14q13.2) was copies of the A allele in 2015 and weeks) daily Abbott 2 (n=149) associated with response rs12436663 had a mean Caucasians LOCCS, (n=69) to ziletuton. change of -123mL and (>12 years and LODO, rs517020 (G/T) compared to patients age) (n=160) (n=64) (GLT1D1,12q24.33) was carrying at least one G associated with response allele 283mL (β= -1.85L) to montelukast and (combined P value from zileuton. discovery and replication phase = 6.28 x 10-8).
In the zileuton trial, a mean decrease of 435 mL in FEV1 and in the montelukast trial (LOCCS) a mean decrease of 152mL in FEV1 in rs517020 carriers was estimated (combined P value = 1.25 x 10-7) Dahlin African- Clinical Montelukast 184 African- Changes in rs6475448 (A/G) Patients homozygous for et al. American trial (8 5 or 10 mg American and FEV1 (MLLT3, 9p22) the A allele had a higher 2015 and weeks) daily Caucasians increase of FEV1 from Caucasians children (CLIC, 5- baseline in LOCCS adolescents 10 mg (changes in FEV1 of 344 and adults montelukast and mL compared to -4.66 mL (n=133) PACT, 5 mg for homozygotes for G (LOCCS and montelukast) allele) (β= 1.87) LODO) (combined P value = 1.97 x 10-9).
54
ACRN, Asthma Clinical Research Network; CAMP, Childhood Asthma Management Programme; CARE, Childhood Asthma Research and Education; CR, controlled release; FEV1, forced expiratory volume in 1 second; FEV1% pred, percentage of predicted FEV1; GWAS, genome-wide association studies; LOCCS, the Leukotriene Modifier or Corticosteroid or Corticosteroid–Salmeterol trial; LODO, Effectiveness of Low Dose Theophylline as Add On Therapy for the Treatment of Asthma; MAF, minor allele frequency; SLIC, the Salmeterol ± Inhaled Corticosteroids trial; SOCS, Salmeterol or Corticosteroids trial
5
5
55
Candidate gene studies of ICS
In the 29 candidate gene studies on pharmacogenomics of ICS, more than 500 SNPs in 120 different genes were studied. The most commonly studied genes were CRHR1 (Corticotropin Releasing Hormone Receptor 1), which was studied five times in a total of 843 patients; FCER2 (Fc Fragment Of IgE, Low Affinity II, Receptor For [CD23]) was studied five times in a total of 2275 patients; NR3C1 (Nuclear Receptor Subfamily 3, Group C, Member 1) was studied four times in a total of 953 patients; GLCC1 was studied three times in a total of 3931 patients; TBX21 (T-Box 21) was studied three times in a total of 400 patients; and STIP1 (stress induced phosphoprotein 1) was studied three times in a total in 671 patients. Only four SNPs within three genes (CRHR1: rs242941, rs1876828; GLCCI1: rs37973 and FCER2: rs28364072) were positively replicated at least once (Table 2). Response to medication was assessed using different outcomes: lung function measurements such as changes in FEV1 over time (17,18,27–30,19–26), annual decline in FEV1 (31–33) and bronchial hyper responsiveness [BHR] to methacholine(27,30,34,35), asthma symptoms based on questionnaires (26,36–41) and exacerbations despite treatment (39,42–45).
ICS Pharmacogenomics: replicated genes and SNPs
The replication of SNPs previously associated with ICS response remains complicated. Three candidate gene studies of GLCCI1 have tried to replicate the results of the GWAS analysis for rs37972 conducted by Tantisira et al., but only one study could replicate the results (31). In the successful replication study, a decline of over 30ml/year in FEV1 was used as the cutoff marker for poor response in 224 adult Japanese asthmatics. Patients with GG genotypes had a decrease in FEV1 by 1.10 units (95% CI0.02 -2.18 p value< 0.05) (1 unit = 30 ml/year) (31). The two other replication studies (including Caucasian patients) could not replicate the results of the GWAS study(18,43). In one study (n=1916 adult asthmatics), the same outcome of interest as the GWAS (changes in FEV1 from baseline) was measured in a 8 to12 weeks trial (18).The second unsuccessful replication study was a meta-analysis of three pediatric asthma cohorts (n= 1791 patients), and the risk of asthma exacerbations despite ICS use was assessed as the outcome (43). In total, the association between GLCCI1 and ICS response has been studied in 3916 patients, and in 3692 patients no statistically significant associations were found (18,43).
56
Differences in outcome measures and ethnic backgrounds could play a role in the lack of replication.
Another gene that has been associated with response to ICS treatment in various studies is CRHR1. This gene encodes a receptor that binds to Corticotropin Releasing Hormone (CRH), which is a major regulator of the hypothalamic-pituitary-adrenal pathway. This protein is needed for the activation of signal transduction pathways that regulate various physiological processes, including stress and immune responses (53,54). It has been hypothesized that the change in the level of the endogenous corticosteroids secretion due to variations within CRHR1, might change the response to the exogenous corticosteroids (55). In total, 20 SNPs within this gene have been studied in four different studies (40,55–57). Two SNPs have been replicated (rs1876828 and rs242941).
Rs1876828 has been studied in three studies (40,48,49) in five different populations (a total of 1003 patients). In two populations, a statistically significant association with lung function was found in adult and pediatric asthmatics (289 patient in total) (40,55). In the study by Tantisira et al., patients homozygous for the minor allele (A) had a four- times greater improvement in their FEV1%(23.72 ± 9.75) compared to homozygotes for the common G allele (5.14 ± 1.31%) (55). A comparable association was found in a later study by Mougey et al., in which AA genotype carriers had an 8-fold greater improvement in FEV1% predicted as compared to the GG carriers (24.2±9.10% vs 2.98±1.88%) (40). However, in three other populations (including more than 700 adult and pediatric patients in total) there was no interactions between this SNP and response to ICS (55,56).
The other replicated SNP within CRHR1 is rs242941. Interaction between rs242941 and ICS response (based on lung function measurements) has been studied in four studies (a total of 1314 patients) (40,55–57); however, there was a discrepancy in the findings of these studies. In two separate populations (616 Caucasian pediatric and adult asthmatic patients in total) within one study, the variant (T allele) was significantly associated with improvements in FEV1% in 6-8 weeks treatment with ICS (55). Patients from the CAMP study were included in one of the populations. In both populations, patients homozygous for the minor allele had a higher improvement in mean FEV1% (13.3±3.1% and 17.8±6.8%) than homozygotes for the wild-type allele (5.5±1.4% and 7.6±1.5%) (55). However, in two other studies (40,57) (including a total of 375 pediatric and adult asthmatics with African-American and Caucasian
57
backgrounds), the minor allele was significantly associated with poor lung function improvement. Mougey et al. found a mean decrease of 10±3.2 in FEV1% predicted after 16 weeks of treatment with ICS (P= 0,003) in 65 patients with minor allele (40). On the other hand, Rogers et al. categorized children of the CAMP trial to non- responders (improvement of FEV1% predicted <7.5% from baseline) and responders (improvement of ≥7.5%) (57). Children carrying the T allele of rs242941 had a higher risk of a poor response to ICS (OR: 1.9, P=0.02). Inconsistencies in the findings of these studies might be the result of differences in the definitions of outcome measures and the duration of the studies. Moreover, in two other populations (322 patients in total) there was no statistically significant interactions between CRHR1 rs242941 and lung function measurement (55,56). Definition of the outcome measure and the methods used for patient inclusion were also different in these studies. No statistically significant association was found between rs242941 and other response outcomes (40,57).
The FCER2 gene encodes for a low affinity IgE receptor (CD23) (58). In total, 25 SNPs within this gene have been studied, and one (rs28364072) has been successfully replicated in three studies (39,45,57). The association between this SNP and ICS response was initially identified in Caucasian and African-American children from the CAMP study (45). The SNP was associated with a risk of severe exacerbations despite ICS use (Hazard Ratio [HR]: 3.95; 95% CI: 1.64-9.51 in white children; and HR: 3.08; 95% CI: 1.00-9.47 in African-Americans) (45). A second study found statistically significant associations with ICS response when categorizing the patients to responders (change in FEV1% >7.5%) and non-responders (change in FEV1% <7.5%) (OR: 2.1, 95%CI: 1.2–3.5) (57). A meta-analysis of three pediatric asthma cohorts also found a statistically significant association with the risk of asthma-related hospital visits (OR: 2.38, 95%CI: 1.47–3.85) (39). It has been shown that variation in rs28364072 is associated with an altered FCER2 expression (45). Consistent results across these studies demonstrate a considerable amount of evidence for a gene-drug interaction. Genetic variants in the TBX21 (26,27,34) and 17q21 loci (25,30) have also been found to influence ICS response in several studies; however, since the associations were found using different outcome measures, it is not further discussed in the manuscript.
58
Table 2. Replicated genes of ICS studies.
SNPs Study Study Design Definition of response Study outcome population
GLCCI1 (7p21.3)
Rs37973 Hosking White Pooled data of Changes in FEV1 from baseline There was no significant association (A/G) et al. adolescents six to eight-week after 8 weeks in 6 studies and at between changes in FEV1 and rs37973 2014 and adults trials and one week 12 in one study. genotypes. (n=1916) 12-week clinical trial
Rs37972 Vijverberg North Meta-analysis of Exacerbations: There was no significant association (T/C) et al. European three pediatric - Hospital visits between increased risk of OCS use, 2014 children asthma cohorts - OCS use increased risk of asthma exacerbations and rs37972 genotypes BREATHE Poor asthma symptoms (n=1037) - ACT scores ≤ 19 - ACQ-scores ≥ 1.5 PACMAN (n=431)
PAGES
(n=323)
5
9
59
60
Rs37973 Izuhara Adult Asthma cohort Annual decline in FEV1 rs37973 GG was associated with a decline (A/G) et al. Japanese in FEV1 of 30ml/year or more (estimated 2014 (n=224) effect: 1.10: 0.02 to 2.18, P=0.047). There was no association between rs37973 genotypes and the outcome, when decline in FEV1 was analyzed as a continuous variable.
CRHR1 (17q21.31)
rs242941 Tantisira Caucasian Three Changes in FEV1% from baseline rs242941: for the minor allele homozygotes rs1876828 et al. children and independent the percent change was 13.28 ± 3.11 in 2004 adults: Adult study and 17.80 ± 6.77 in CAMP 6-8-week compared with 5.49 ± 1.40 and 7.57 ± 1.50 Adult study Clinical trials for those homozygous for the wild-type allele (Adults, (P= 0.025 and P=0.006, respectively). n=415) rs1876828: Homozygotes for the minor T CAMP allele had an average increase in their FEV1 (Children, of 23.72 ± 9.75 compared with 5.14 ± 1.31% n=201) for homozygotes for the common allele in patients from ACRN. ACRN (Adults, n=224)
rs242941 Dijkstra Adults Asthma cohort Changes in FEV1 from baseline No associations were found between SNPs rs1876828 et al. (n=98) (immediate effect) and the outcomes. 2008 Rate of decline in FEV1 annually (long-term effect)
60
rs242941 Rogers African- 4-year clinical Exacerbations: Emergency room The minor allele (T) showed a significant et al. American trial visit Hospitalization Oral association with poor response based on 2009 and prednisone burst. lung function (OR: 1.6, CI 95% 1–2.7, Caucasian P=0.05). children Lung function measurements: (CAMP, change in FEV1% pred: ≤7.5% n=311) considered as poor responders
rs1876828 Mougey Caucasian 16 weeks -Change FEV1% pred Minor allele of rs1876828 was associated rs242941 et al. children, Clinical trial with improvements in FEV1% pred (P=1.89 x 2013 adolescence Asthma symptoms: 10 -4). and adults (n=65) - Slopes of plots of ACQ scores Major allele of rs242941 was associated with versus time improvements in FEV1% pred (P=2.07 x 10- 3).
FCER2 (19p13.3)
Rs28364072 Tantisira African- 4-year clinical Severe exacerbations: There was significant association between (A/G) et al. American trial rs28364072 variant and severe 2007 and -Emergency room visits and/or exacerbations in both white and African Caucasian hospitalizations. American children (hazard ratio, 3.95; 95% children CI, 1.64-9.51; and hazard ratio, 3.08; 95% (CAMP, CI, 1.00-9.47). n=311)
Rogers African- 4-year clinical Exacerbations: Emergency room The minor allele of rs28364072 in FCER2 et al. American trial visit Hospitalization Oral was associated with recurrent exacerbations 2009 and prednisone burst. in white subjects (OR: 1.9 for minor allele, P Caucasian < 0.05) but it did not reach significance in
children multivariate analysis (OR: 1.4, P = 0.18).
61
61
62
(CAMP, Lung function measurements: The minor allele was also associated with n=311) change in FEV1% pred: ≤7.5% poor response considering lung function considered as poor responders changes (OR: 2.1, 95%CI: 1.2–3.5, P=0.006).
Koster et Caucasian PACMAN & Exacerbations: The rs28364072 variant was associated with al. 2011 children BREATHE: Emergency room visits increased risk of asthma-related hospital Hospitalization visits in the meta-analysis (OR: 2.38, 95%CI: PACMAN asthma cohorts 1.47–3.85, P= 0.0004). (n=386) Asthma symptoms: -ACQ- scores The variant was associated with increased BREATHE -Respiratory symptoms (wheeze, risk of uncontrolled asthma measured by (n=939) CAMP: shortness of breath and cough ACQ scores (OR: 2.64, 95%CI: 1.00–6.98) -Asthma-related sleep and was associated with increased daily CAMP 4-year clinical disturbances - steroid dose (OR: 2.46, 95%CI: 1.38–4.39). (n=311) trial Asthma-related limitations in daily activities -Additional (airway) medication use during the preceding 12 months
* In observational studies doses and types of ICS varied between patients. ACQ, Asthma Control Questionnaire; ACRN, Asthma Clinical Research Network; ACT, Asthma Control Test; AHR, airway hyper-responsiveness; AQLQ, Asthma quality of life questionnaire; CAMP, Childhood Asthma Management Programme; CARE, Childhood Asthma Research and Education; FEV1, forced expiratory volume in 1 second; FEV1% pred, percentage of predicted FEV1; FVC, Forced vital capacity; ICS, inhaled corticosteroids; OCS, oral corticosteroid; OR, odds ratio; PACMAN, Pharmacogenetics of Asthma Medication in Children: Medication with Anti-Inflammatory effects; PAGES, Paediatric Asthma Gene Environment Study; PC20, provocation concentration causing a 20% fall in FEV1; PD20, provocation dose causing a 20% decline in FEV1.
62
Pharmacogenomics of Leukotriene Modifiers In total, 106 studies on LTM pharmacogenomics were identified. Twenty-six studies met the inclusion criteria (Figure 2). Of these studies, 24 were candidate gene studies (Supplementary Table S4) (22,40,67–76,59,77–80,60–66) and two were GWAS (81,82). None of the SNPs that had been associated with treatment response in the
candidate gene studies were identified in GWAS studies of LTMs.
Records identified through database searching (PubMed [106] and EMBASE [186])
(n = 292) Identification
Records excluded because of duplication (n = 42)
Records screened
(n = 250) Screening
Records excluded (n = 220)
Articles assessed for eligibility
(n = 28)
Eligibility Full-text articles excluded, (n = 2)
Studies included after reading the full-
uded text
(n = 26) Incl
Figure 2. Inclusion flowchart for pharmacogenomics studies of LTMs.
63
GWAS studies of LTMs
Two GWAS have been published recently by one research group (Table 1) (81,82). Both studies assessed the changes in FEV1 from baseline to the end of trial (8-12 weeks) in white and non-white asthmatic patients. The first, a GWAS analysis of LTMs by Dahlin et al. (81) included 160 patients >12 years age receiving zileuton. This study found that after combining the analyses of discovery and replication cohorts, one SNP (rs12436663) within the MRPP3 gene (also known as KIAA0391) was significantly associated with the changes in FEV1 from baseline after 12 weeks of treatment (in 309 patients). The mean change of FEV1 for patients carrying AA genotype was -123mL, and the mean change for patients carrying at least one G allele was 283mL. This gene encodes for MRPP3 protein that functions in the maturation of mitochondrial tRNA (83) and has previously been found to be associated with autoimmune diseases such as Crohn’s disease and psoriasis (84). In this first GWAS, additional replication was also performed in two other clinical trials consisting of 133 adolescent/adult asthmatic patients treated with montelukast in order to find the SNP associations shared between zileuton and montelukast treated patients. Rs517020 SNP within the GLT1D1 (glycosyltransferase 1 domain containing 1) gene was associated with a poorer response to both montelukast and zileuton (combined P value 1.25 x 10-7). A mean decrease of 152 mL in FEV1 for montelukast trial and 435 mL for zileuton trial was estimated in the additive model. The product of GLT1D1 has glycosyl transferase activity (85), but its role in asthma is still unknown. In the second GWAS study (82), the discovery cohorts included 133 asthmatic adolescents/adults from two clinical trials of montelukast (used for replication analysis in the first GWAS (81)). The replication cohort included two other montelukast clinical trials (including a total of 184 asthmatic children). The change in FEV1 from baseline to 8 weeks of LTM treatment was measured in all studies. Rs6475448 within MLLT3 (myeloid/lymphoid or mixed-lineage leukemia; translocated to, 3), exceeded genome-wide significance (P= 1.97x10-9) after combining the p-values of the four trials. This gene encodes for a protein that is a part of a complex that increases the catalytic rate of RNA polymerase ll transcription (86). Patients homozygous for the variant allele (A) had a mean increase in FEV1 of 344 mL compared to -4.66 mL for patients with GG genotype. The SNPs identified in the first GWAS were not found to be associated with response to montelukast in the second GWAS. Moreover, SNPs identified in LTM candidate gene studies were not identified in these GWAS.
64
Candidate gene studies of LTMs
More than 250 SNPs in 30 different genes were studied in candidate gene studies of LTMs. The majority of the candidate gene studies of response to LTMs (22 out of 24) assessed the response to CysLT1 receptor antagonists, with only two assessing the response to 5-lipoxygenase inhibitors. Arachidonate 5-lipoxygenase (ALOX5), ALOX5 activating protein (ALOX5AP), leukotriene A4 hydrolase (LTA4H), LTC4 synthase (LTC4S), the ATP-binding cassette family member ABCC1/multidrug resistance protein 1 (MRP1), and cysteinyl leukotriene receptor 1 and 2 (CysLTR1 and CysLTR2) are the genes that encode the main proteins in LT pathway (Figure 3). Among these genes, LTC4s (investigated in 12 studies, 1345 patients in total), ALOX5 (8 studies, 1277 patients in total), LTA4H (5 studies, 1030 patients in total), CysLTR1 (5 studies, 991 patients in total) and MRP1 (3 studies, 360 patients in total) were studied most often. However, only three variants in two genes (ALOX5, Sp1 repeats in the core promotor and rs2115819, and MRP1 rs119774) were found to be statistically significantly associated with response to LTMs and positively replicated for the same outcome at least once (Table 3). In 21 of the 24 studies, lung function measurements were considered as the primary outcome. However, the definition of this outcome was different among the various studies having been defined as follows: change in FEV1% from baseline over time (22,59–65), change in FEV1 after using short acting beta2 agonists [SABA] (66,67), changes in PEF (63,65,66,68) and BHR to adenosine monophosphate (AMP), Methacholine or aspirin (in patients with Aspirin-intolerant Asthma (AIA)) (69–71). Furthermore, an exercise challenge test was performed in study patients with exercise induced bronchoconstriction (EIB) as the primary outcome parameter (72–76). Asthma-related symptoms were measured in four studies (40,68,77,78), and asthma exacerbation rates were studied in two studies (59,60). The definition and assessment method of uncontrolled asthma symptoms and exacerbations rates varied between the studies.
65
Figure 3. Leukotriene synthesis pathway. PLA2, Phospholipase A2; 5-LO, 5-lipoxygenase; FLAP, 5-lipoxygenase–activating protein; LT; Leukotriene; LTA4H, Leukotriene A4 hydrolase; LTC4S, Leukotriene C4 Synthase; MRP1, multidrug resistance protein 1; CysLT1, Cysteinyl leukotriene receptor 1; BLT receptor, Leukotriene B4 receptor. *proteins and enzymes encoded by genes studied in pharmacogenomics studies of LTMs.
LTM pharmacogenomics: replicated genes and SNPs
Two genes in the leukotriene pathway have frequently been studied and successfully replicated: ALOX5 and MRP1. The products of these genes play important roles in the production and transport of Cysteinyl Leukotriens (LTC4, LTD4 and LTE4) that cause inflammation and constriction in the airways(87,88). Two genetic variants in ALOX5 (Sp1 repeats, and rs2115819) and one genetic variant in MRP1(rs119774) have been frequently studied and successfully replicated. Five studies have investigated the association between response to LTMs and genetic variation in the core promoter SNP of ALOX5(59–61,63,69). Normally, this locus contains a transcription factor binding region with five binding sites (Sp1 repeats); however, deletion and addition of binding
66 sites results in the genetic variation(89). Changes in FEV1 were assessed in four separate clinical trials, including 402 asthmatic patients in total, in order to investigate the interaction between ALOX5 Sp1 repeats and response to LTMs (59–61,63). Two studies (with a total of 175 patients) found statistically significant improvements in FEV1% (61) and increase in FEV1%predicted (60) in patients carrying at least one wild-type allele as compared to patients homozygous for the mutant allele while they were on treatment with LTMs. In the first pharmacogenomics study of asthma medication, published in 1999, Drazen et al. reported a change of 18.8±4% in FEV1 in patients homozygous for the wild-type allele (5 tandem repeats) compared with −1.2±3% in patients homozygous for the mutant genotype (tandem repeats other than 5) (61). In a later study, patients with at least five tandem repeats had an increase of 8.6%±8 compared with -1.4%±8.04 in patients homozygous for the mutant allele (60). These two confirmative studies both used a different LTM in their studies (ABT-761 and montelukast).Two other studies (with a total of 218 patients) did not report any significant association between Sp1 repeats and response to montelukast when considering changes in FEV1% predicted as the outcome (59,63). There are inconclusive results regarding the effect of ALOX5 Sp1 repeats on the occurrence of exacerbations(59,60). In a study by Lima et al., (59), patients homozygous for the tandem repeats other than five showed a 73% lower risk of exacerbations. However, in a study by Telleria et al. patients homozygous for the wild-type allele had a higher reduction in the amount of exacerbations compared to patients homozygous for the mutant allele receiving six months of treatment with montelukast (reduction in exacerbation rate: -4.4 and -1.3, respectively) (60). Asthma exacerbation was defined as a fall in PEF rate in the morning of at least 25% compared with the baseline (60). In contrast, Lima et al. defined asthma exacerbation according to the presence of one or more of the following items: an increase in the use of a rescue inhaler with up to four puffs in one day, a decrease in PEF rate of more than 30% for two consecutive days, or an unscheduled visit to emergency room/clinic/hospital or an OCS course (59). In the fifth study, BHR to AMP was considered as the treatment response in a retrospective analysis of four placebo-controlled trials including 52 adult asthmatics treated with montelukast or zafirlukast. Researchers did not find a statistically significant difference between heterozygotes and wild-type homozygotes in terms of response to LTMs and in this study there was no homozygotes for the mutant allele (69).
67
The association of ALOX5 rs2115819 A>G and MRP1 rs119774 C>T with response to LTMs has been studied in four (40,59,66,90) and three studies (40,59,90), respectively. Two earlier studies by one research group investigating the association of SNPs within five candidate genes of LT pathway (MRP1, ALOX5, LTA4H, LTC4S and CysLTR1) with changes in FEV1 reported statistically significant associations for both SNPs in Caucasian patients treated with montelukast and zileuton. Different study populations (300 asthmatic children and adults in total) were included in these studies (59,90). In the first study, patients with the GG genotype for rs2115819 showed greater improvements in FEV1 with 6 months of treatment with montelukast (FEV1 improvement: 30% [95% CI: -0.017-1.21]), compared to the patients with AA and AG genotypes (FEV1 improvement 4.4% [95% CI: 0.025-0.66] and2% [95% CI: 0.013- 0.075], respectively). Likewise, patients homozygous for the C allele at rs119774 had greater improvements in FEV1 24% (95% CI: -0.105-0.577), compared to patients heterozygous for this allele, who showed improvements of 2.2% (95% CI: -0.005- 0.049); there were no patients homozygous for the T allele. In the second study, these findings were replicated in patients treated with zileuton for 12 weeks. Patients carrying G allele of rs2115819 had more than 20% improvement in the FEV1% predicted compared with the AA carriers that had approximately 15% improvement in the FEV1% at the end of the trial (P=0.01) (90). Improvements of FEV1% was not shown for the rs119774 in the study. However, no associations were found for these SNPs with the changes in FEV1 and ACQ scores in a later study by this research group in which 60 asthmatic adults and children with Caucasian ethnicity who had been treated with montelukast for four months were studied (40). Furthermore, rs2115819 was not associated with the changes in morning PEF in 81 patients treated with montelukast (66) and exacerbation rates (59).
68
Table 3. Candidate gene studies of response to LTMs
Study Study Design Definition of response Study outcome population
ALOX5 (10q11.2) tandem repeats of the Sp1-binding domain
Drazen et al. Adults (n=114) 12 weeks Clinical Changes in FEV1% from The mean change in FEV1% from baseline was 1999 trial baseline 18.8±3.6% in wild-type (5/5) patients and 23.3±6.0% in heterozygous (5/X) patients compared to the patients with the mutant genotype (X/X) with the change in FEV1% of −1.2±2.9% (p < 0.0001 and p < 0.0006).
Fowler SJ et Adult atopic Retrospective Bronchial hyper- There was no association between different al. 2002 asthmatics analysis of four responsiveness (BHR) to genotypes and the outcomes. (n=52) placebo-controlled adenosine monophosphate trials (AMP), changes in FEV1 (l), FEF25-75 (l/s) and PEFR (l/min)
Lima et al. Caucasian 6 months Clinical Changes in FEV1% pred from Homozygotes for the mutant allele for tandem 2006 adults (n=61) trial baseline repeats (repeats other than 5) had a 73% decrease in the risk of exacerbations (P= 0.04). Asthma exacerbation rate
- One or more of the following:
69 a more than 30% decrease in
69
70
peak expiratory flow rate for two consecutive days, a course of oral steroids, an unscheduled visit to the clinic, the emergency room, or hospital or an increase of four puffs of rescue inhaler use in one day
Klotsman Caucasian, 12 weeks Clinical Changes in FEV1% pred from No associations were found between this SNP and et al. 2007 Hispanic and trial baseline changes in FEV1% pred. African- American
adolescents and adults (n=166)
Telleria et al. Adolescent and 6 months clinical Changes in FEV1% pred from Homozygotes for the mutant allele (4/4) had 2008 adult atopic trial baseline 1.88±0.92 rate of asthma exacerbations, with a asthmatics mean reduction of 1.33±1.22 and patients (n=61) Number of exacerbations: homozygous for the common allele (5/5) and heterozygotes (4/5) had an exacerbation rate of -A decrease in peak expiratory 0.4±0.21 with a reduction of 4.41±2.76 (P= 0.001). flow rate in the morning of at least 25% compared with Homozygotes for the mutant allele had a mean baseline. decrease of 1.4%±8.04 in FEV1% pred compared to the homozygotes for the common allele and heterozygotes with a mean increase of 8.6%±7.6 in FEV1% pred (P=0.0006).
70
ALOX5, rs2115819 (A/G)
Lima et al. Caucasian 6 months Clinical Changes in FEV1% pred from There was a significant association between ALOX5 2006 adults (n=61) trial baseline (rs2115819) and changes in FEV1% pred (p<0.05)
Asthma exacerbation rate Patients homozygous for the G allele had 30% improvements in the FEV1%pred (95% CI = -0.017 - One or more of the following: to 1.21) compared to 4.4% (95% CI, -0.025 to 0.66) a more than 30% decrease in and 2.0% (95% CI, 0.013–0.075) improvements in peak expiratory flow rate for the AA and AG genotype groups, respectively. two consecutive days, a course of oral steroids, an unscheduled visit to the clinic, the emergency room, or hospital or an increase of four puffs of rescue inhaler use in one day
Tantisira et al. Caucasian 12 week Clinical Changes in FEV1% pred from Significant association was found between 2009 children, trial baseline rs2115819 changes in FEV1% from baseline adolescence (P=0.01). G allele carriers had higher improvements and adults in the FEV1% pred at the end of the 12 weeks and (n=239) improvement in FEV1% pred from baseline was more than 20% in GG carriers compared with the AA carriers with mean improvement of 15% from the baseline.
Mougey et al. Caucasian 16 weeks Clinical Change FEV1% pred No association was found between rs2115819 and 2013 children, trial the outcomes. adolescence Asthma symptoms:
71
71
72
and adults - Slopes of plots of ACQ (n=65) scores versus time
Kotani et al. Japanese adults 4-8 week Clinical Self-measured changes in No significant association was found between the 2012 (n=21) trial PEF after SABA rs2115819 and changes in PEF and FEV1.
MRP1 (16p13.1) rs119774 (C/T)
Lima et Caucasian 6 months Clinical Changes in FEV1% pred from There was a significant association between MRP1 al.2006 adults (n=61) trial baseline (rs119774) and changes in FEV1% pred (p<0.05). Heterozygotes for the T allele had 24% improvement Asthma exacerbation rate (95% CI: -0.105 to 0.577) in the FEV1% pred compared with the patients homozygous for the C - One or more of the following: allele with 2.2% improvements in the FEV1%pred a more than 30% decrease in (95% CI: -0.005 to 0.049). peak expiratory flow rate for two consecutive days, a course of oral steroids, an unscheduled visit to the clinic, the emergency room, or hospital or an increase of four puffs of rescue inhaler use in one day.
Tantisira et al. Caucasian 12 weeks Clinical Changes in FEV1% pred from Significant association was found between rs119774 2009 children, trial baseline and changes in FEV1% from baseline (P= 0.05). No adolescence data on improvements in FEV1% predicted was and adults shown. (n=239)
72
Mougey et al. Caucasian 16 weeks Clinical Change FEV1% pred No association was found between rs119774 and 2013 children, trial the outcomes adolescence Asthma symptoms: and adults (n=65) - Slopes of plots of ACQ scores versus time
ACQ, Asthma Control Questionnaire; ACT, Asthma Control Test; ACRN, Asthma Clinical Research Network; AHR, airway hyper-responsiveness; AMP, adenosine monophosphate; CAMP, Childhood Asthma Management Programme; CARE, Childhood Asthma Research and Education; EIB, Exercise-induced bronchoconstriction; FEV1, forced expiratory volume in 1 second; FEV1% pred, percentage of predicted FEV1; FVC, Forced vital capacity; OCS, oral corticosteroid; PACMAN, Pharmacogenetics of Asthma Medication in Children: Medication with Anti-Inflammatory effects;
PAGES, Paediatric Asthma Gene Environment Study; PEF, forced expiratory flow; SABA, short acting beta agonist; SP1, specificity protein 1.
73
73
Discussion
Since 1999, over fifty studies have investigated if and how genetic variation can explain why response to asthma maintenance therapy varies between asthmatic individuals. However, despite the number of studies, the clinical applicability of these pharmacogenomics findings remains unclear. Few variations have been replicated and effect estimates are relatively low. The most consistent results were found for a SNP (rs28364072) within FCER2 in relation to ICS response. All three studies assessing the association between this SNP and response to ICS, reported significant associations between the SNP and different outcomes (exacerbations, lung function measurements and asthma symptoms) in asthmatic children treated with ICS(39,45,57). More evidence was provided in the in vitro study of FCER2 that confirmed altered gene expression in patients carrying the variant allele (45). There have been other successful replications in the pharmacogenomics of ICS (i.e. CRHR1 and GLCCI1) and pharmacogenomics of LTMs (i.e. MRP1 and ALOX5), but the results were not consistent. Differences in study design and reported outcomes made it impossible to perform meta-analyses. Interestingly, GWAS and candidate gene studies delivered different results. Some of these SNPs might not have been genotyped or imputed in GWAS studies. Another reason that none of the SNPs associated with treatment response in candidate gene studies were identified in GWAS could be the relatively small sample sizes in GWAS (<500). One of the important measures of quality in genetic association studies is the sample size (91). It has been shown that to detect a relative risk between 1.7-2.0, a sample size of 1000 cases and 1000 controls in GWAS analysis is required (92,93). Since effect sizes in complex diseases like asthma are even smaller (1.1-2.0), larger number of cases and controls are needed. Various candidate gene studies on LTM pharmacogenomics were performed in small study populations (<100 patients), and this might have resulted in a low power to detect genetic variations with a low or moderate effect influencing treatment response. There is also a large heterogeneity in the patient populations studied within asthma pharmacogenomics. In our review, we did not find remarkable differences between the different ethnic groups studied and associations between genetic variants and response to medication. Differences in response to medication among different ethnicities have been previously shown (94–96). The majority of the study populations in our review included only adults or a combination of adults and children. The biological factors influencing the response to treatment in children might differ from adults, and disease phenotypes vary between asthmatic children and adults (97). This
74 could lead to failed replication when combining adult and pediatric data. For example, a genetic variation influencing FBXL7 expression has been found to be associated with an improvement in asthma symptoms in response to ICS in two independent pediatric asthma cohorts, but failed to be replicated in an adult asthmatic population (48). In addition, rs28364072 within FCER2 was found to be associated with response to ICS in pediatric patients (34,39,45), but since this SNP has only been studied in pediatric populations, its influence on response to ICS in adult patients remains unclear. These examples imply that findings from the studies in pediatrics cannot easily be generalized to the adult asthma population and vice versa. The comparability of results is further complicated by the heterogeneity in outcome definitions. Different outcome phenotypes might reflect different aspects of the disease (27,57). Even within one dimension of the disease, such as exacerbations, there was no consistency in how this disease aspect was defined (i.e. emergency room visits, hospitalizations, need for treatment with oral corticosteroids, and a decrease in peak expiratory flow rate in the morning of at least 25%) (59,60). The most often used outcome parameters were lung function measurements. However, exacerbations cause stress and high expenses to both the patients and the health care system (98) and, therefore, considering these parameters as response phenotypes is important. Patient-centered outcomes, such as asthma symptoms were not frequently studied. In summary, the consideration of different outcome phenotypes on the pharmacogenomics analysis of asthma is crucial; yet, there is also a need to standardize the definitions of response phenotypes.
In conclusion, there is still not enough evidence to implement genetic testing in clinics before starting a patient on medication. In order to provide a larger body of evidence, large sample sizes of well-phenotyped asthmatic patients and different standardized outcome phenotypes should be taken into account in pharmacogenomics studies. Collaborations and large meta-analysis might show more evidence of the influence of genetic markers on the response of asthma to anti-inflammatory medication.
75
References
1. Global Initiative for Asthma (GINA). 10. Lima JJ, Blake K V, Tantisira KG, Weiss From the Global Strategy for Asthma ST. Pharmacogenetics of asthma. Curr Opin Management and Prevention, Global Pulm Med. 2009;15(1):57–62. Initiative for Asthma (GINA) 2015. http://www.ginasthma.org/. 11. Drazen JM, Israel E, O’Byrne PM. Treatment of asthma with drugs modifying the 2. Szefler S.J., Phillips B.R., Martinez F.D., leukotriene pathway. N Engl J Med. 1999;340 Chinchilli V.M., Lemanske R.F., Strunk (3):197–206. R.C., et al. Characterization of within- subject responses to fluticasone and 12. Israel E, Chervinsky P.S., Friedman B, montelukast in childhood asthma. J Allergy Van Bavel J, Skalky C.S., Ghannam AF, et al. Clin Immunol. 2005;115(2):233–42. Effects of montelukast and beclomethasone on airway function and asthma control. J Allergy Clin Immunol. 2002;110(6):847–54. 3. Malmstrom K, Rodriguez-Gomez G,
Guerra J, Villaran C, Piñeiro A, Wei L.X, et 13. Haughney J, Price D, Kaplan A, Chrystyn al. Oral montelukast, inhaled beclo- H, Horne R, May N, et al. Achieving asthma methasone, and placebo for chronic control in practice: understanding the reasons asthma. A randomized, controlled trial. M- for poor control. Respir Med. 2008;102(12): ontelukast/Beclomethasone Study Group. 1681–93. Ann Intern Med. 1999;16;130(6): 487–95.
14. McGhan SL, MacDonald C, James DE, 4. Drazen J.M., Silverman E.K., Lee T.H. Naidu P, Wong E, Sharpe H, et al. Factors Heterogeneity of therapeutic responses in associated with poor asthma control in asthma. Br Med Bull. 2000;56(4):1054–70. children aged five to 13 years. Can Respir J. 2006;13(1):23–9.
5. Chung KF, Godard P, Adelroth E, Ayres 15. Available from: https://ghr.nlm.nih.gov/ J, Barnes N, Barnes P et al. Difficult/ primer/genomicresearch/pharmacogenomics. therapy-resistant asthma: the need for an
integrated approach to define clinical ph- 16. Moher D, Liberati A, Tetzlaff J, Altman DG, enotypes, evaluate risk factors, understand PRISMA Group. Preferred reporting items for patho-physiology and find novel therapies. systematic reviews and meta-analyses: the ERS Task Force on Difficult/Therapy-Re- PRISMA statement. J Clin Epidemiol. 2009;62 sistant Asthma. European Respiratory S. (10):1006–12. Eur Respir J. 1999;13(5):1198–208.
17. Hawkins GA, Lazarus R, Smith RS, 6. Sarpong, E. Changes in Adult Asthma Tantisira KG, Meyers DA, Peters SP, et al. Medication Use and Expenditures, United The glucocorticoid receptor heterocomplex States, 1998–1999 to 2008–2009. 2012;18 gene STIP1 is associated with improved lung (9):499-504. function in asthmatic subjects treated with inhaled corticosteroids. J Allergy Clin Im- 7. Rhen T, Cidlowski JA. Antiinflammatory munol. 2009;123(6):1376–1383.e7. action of glucocorticoids--new mechanisms for old drugs. N Engl J Med. 2005;353(16): 18. Hosking L, Bleecker E, Ghosh S, Yeo A, 1711–23. Jacques L, Mosteller M, et al. GLCCI1- rs37973 does not influence treatment 8. Horvath G, Wanner A. Inhaled cortico- response to inhaled corticosteroids in white steroids: effects on the airway vasculature subjects with asthma. J Allergy Clin Immunol. in bronchial asthma. Eur Respir J. 2006; 2014;133(2):587–9. 27(1): 172–87. 19. Perin P, Berce V, Potocnik U. CD14 gene 9. Barnes P.J., Adcock I.M. How do cortico- polymorphism is not associated with steroids work in asthma? Ann Intern Med. 2003;139(5 Pt 1):359–70.
76 asthma but rather with bronchial obstruction the histone deacetylase 1 gene is associated and hyperreactivity in Slovenian children with with the response to corticosteroids in non-atopic asthma. Respir Med. 2011;105 asthmatics. Korean J Intern Med. 2013;28 (6): Suppl:S54-9. 708–14.
20. Chang HS, Heo J-S, Shin S-W, Bae D-J, 29. Raeiszadeh Jahromi S, Mahesh PA, Song HJ, Jun JA, et al. Association between Jayaraj BS, Holla AD, Vishweswaraiah S, TAAR6 polymorphisms and airway respon- Ramachandra NB. IL-10 and IL-17F Promoter siveness to inhaled corticosteroids in asthma- Single Nucleotide Polymorphism and Asthma tic patients. Pharmacogenetics Genomics. A Case-Control Study in South India. Lung. 2015;25(7):334–42. 2015;193(5):739–47.
21. Cho S-H, Park B-L, Shin SW, Heo J-S, 30. Leusink M, Vijverberg SJH, Koenderman Park J-S, ParkS.W, et al. Association L, Raaijmakers JAM, de Jongste JC, Sterk PJ between WDR21A polymorphisms and et al. Genetic variation in uncontrolled airway responsiveness to inhaled cortico- childhood asthma despite ICS treatment. steroids in asthmatic patients. Pharmaco- Pharmacogenomics J. 2015;1–6. genet Genomics. 2012;22(5):327–35. 31. Izuhara Y, Matsumoto H, Kanemitsu Y, 22. Balantic M, Rijavec M, Skerbinjek Kavalar Izuhara K, Tohda Y, Horiguchi T, et al. M, Suskovic S, Silar M, Kosnik M, et al. GLCCI1 variant accelerates pulmonary Asthma treatment outcome in children is function decline in patients with asthma associated with vascular endothelial growth receiving inhaled corticosteroids. Allergy. factor A (VEGFA) polymorphisms. Mol Diagn 2014;69(5):668–73. Ther. 2012;16(3):173–80. 32. Vonk JM, Postma DS, Maarsingh H, 23. Dijkstra A, Postma DS, Bruinenberg M, Bruinenberg M, Koppelman GH, Meurs H. van Diemen CC, Boezen HM, Koppelman Arginase 1 and arginase 2 variations GH, et al. SERPINE1 -675 4G/5G poly- associate with asthma, asthma severity and morphism is associated with asthma severity β2 agonist and steroid response. Pharmaco- and inhaled corticosteroid response. Eur genet Genomics. 2010;20(3):179–86. Respir J. 2011;38(5):1036–43. 33. Ierodiakonou D, Postma DS, Koppelman 24. Berce V, Potocnik U. Functional poly- GH, Boezen HM, Gerritsen J, Ten Hacken N, morphism in CTLA4 gene influences the et al. E-cadherin gene polymorphisms in response to therapy with inhaled cortico- asthma patients using inhaled cortico- steroids in Slovenian children with atopic steroids. Eur Respir J. 2011;38(5): 1044–52. asthma. Biomarkers. 2010;15(2):158–66. 34. Tantisira KG, Hwang ES, Raby BA, 25. Berce V, Kozmus C.E.P., Potočnik U. Silverman ES, Lake SL, Richter BG, et al. Association among ORMDL3 gene ex- TBX21: a functional variant predicts improve- pression, 17q21 polymorphism and response ment in asthma with the use of inhaled to treatment with inhaled corticosteroids in corticosteroids. Proc Natl Acad Sci USA. children with asthma. Pharmacogenomics J. 2004;28;101(52):18099–104. 2013;13(6):523–9. 35. Raby BA, Van Steen K, Lasky-Su J, 26. YM Y, HY L, Kim S. Pharmacogenetic Tantisira K, Kaplan F, Weiss ST. Importin-13 study of the effects of NK2R G231E G>A and genetic variation is associated with improved TBX21 H33Q C>G polymorphisms on airway responsiveness in childhood asthma. asthma control during inhaled corticosteroid Respir Res. 2009;10:67. treatment. Clin Pharmacol Ther. 2009;85:S 76. 36. Jin Y, Hu D, Peterson E.L., Eng C, Levin A.M., Wells K, et al. Dual-specificity phos- 27. Lopert A, Rijavec M, Zavbi M, Korošec P, phatase 1 as a pharmacogenetic modifier of Fležar M. Asthma treatment outcome in inhaled steroid response among asthmatic adults is associated with rs9910408 in TBX21 patients. J Allergy Clin Immunol. 2010;126(3): gene. Sci Rep. 2013;3:2915. 618-25.e1-2.
28. Kim M-H, Kim S-H, Kim Y-K, Hong S-J, 37. Stockmann C, Fassl B, Gaedigk R, Min K-U, Cho S-H, et al. A polymorphism in
77
Nkoy F, Uchida D.A., Monson S, et al. 46. Tantisira KG, Lasky-Su J, Harada M, Fluticasone propionate pharmacogenetics: Murphy A, Litonjua AA, Himes BE, et al. CYP3A4*22 polymorphism and pediatric Genomewide association between GLCCI1 asthma control. J Pediatr. 2013;162(6):1222– and response to glucocorticoid therapy in 7, 1227.e1-2. asthma. N Engl J Med. 2011;365(13):1173– 83. 38. Vijverberg S.J.H, Tavendale R, Leusink M, Koenderman L, Raaijmakers J.A.M, 47. Tantisira KG, Damask A, Szefler SJ, Koppelman GH, et al. Pharmacogenetic Schuemann B, Markezich A, Su J, et al. analysis of GLCCI1 in three north European Genome-wide association identifies the T pediatric asthma populations with a reported gene as a novel asthma pharmacogenetic use of inhaled corticosteroids. Pharmaco- locus. Am J Respir Crit Care Med. 2012; genomics. 2014;15(6):799–806. 185(12):1286–91.
39. Koster ES, Maitland-van der Zee A.H., 48. Park H-W, Dahlin A, Tse S, Duan QL, Tavendale R, Mukhopadhyay S, Vijverberg Schuemann B, Martinez FD, et al. Genome- S.J.H, Raaijmakers J.A.M, et al. FCER2 wide association study of leukotriene modifier T2206C variant associated with chronic response in asthma. J Allergy Clin Immunol. symptoms and exacerbations in steroid- 2014;133(3):664–9.e5. treated asthmatic children. Allergy. 2011;66 (12):1546–52. 49. Park T-J, Park J-S, Cheong HS, Park B- L, Kim LH, Heo JS, et al. Genome-wide 40. Mougey E.B., Chen C, Tantisira K.G., association study identifies ALLC poly- Blake K.V., Peters S.P., Wise R.A., et al. morphisms correlated with FEV₁ change by Pharmacogenetics of asthma controller treat- corticosteroid. Clin Chim Acta. 2014;436:20– ment. Pharmacogenomics J. 2013;13(3):242 6. –50. 50. The Childhood Asthma Management 41. Szczepankiewicz A, Breborowicz A, Program (CAMP): design, rationale, and Sobkowiak P, Popiel A. No association of methods. Childhood Asthma Management glucocorticoid receptor polymorphisms with Program Research Group. Control Clin asthma and response to glucocorticoids. Adv Trials. 1999;20(1):91–120. Med Sci. 2008 ;53(2):245–50. 51. Nandi D, Tahiliani P, Kumar A, Chandu D. 42. Hunninghake G.M., Soto-Quirós M.E., The ubiquitin-proteasome system. J Biosci. Avila L, Su J, Murphy A, Demeo D.L., et al. 2006 ;31(1):137–55. Polymorphisms in IL13, total IgE, eosino- philia, and asthma exacerbations in child- 52. Vigetti D, Monetti C, Acquati F, Taramelli hood. J Allergy Clin Immunol. 2007;120(1): R, Bernardini G. Human allantoicase gene: 84–90. cDNA cloning, genomic organization and chromosome localization. Gene. 2000;256 43. Vijverberg SJH, Tavendale R, Leusink M, (1–2):253–60. Koenderman L, Raaijmakers JAM, Postma D.S., et al. Pharmacogenetic analysis of 53. Dautzenberg FM, Hauger RL. The CRF GLCCI1 in three north European pediatric peptide family and their receptors: yet more asthma populations with a reported use of partners discovered. Trends Pharmacol Sci. inhaled corticosteroids. Pharmacogenomics. 2002;23(2):71–7. 2014;15(6):799–806. 54. Chrousos G.P. The hypothalamic pituitary 44. Rogers AJ, Tantisira K., Fuhlbrigge AL, adrenal axis and immune-mediated inflamm- Litonjua A., Lasky-Su JA, Szefler SJ, et al. ation. N Engl J Med. 1995;332(20): 1351–62. Predictors of poor response during asthma 55. Tantisira K..G, Lake S, Silverman E.S, therapy differ with definition of outcome. Palmer L.J., Lazarus R, Silverman E.K, et al. Pharmacogenomics. 2009;10(8):1231–42. Corticosteroid pharmacogenetics: associa- 45. Tantisira KG, Silverman ES, Mariani TJ, tion of sequence variants in CRHR1 with Xu J, Richter BG, Klanderman BJ, et al. improved lung function in asthmatics treated FCER2: a pharmacogenetic basis for severe with inhaled corticosteroids. Hum Mol Genet. exacerbations in children with asthma. J 2004;13(13):1353–9. Allergy Clin Immunol. 2007;120(6):1285–91.
78
56. Dijkstra A, Koppelman GH, Vonk JM, 65. Sampson AP, Siddiqui S, Buchanan D, Bruinenberg M, Schouten JP, Postma DS. Howarth PH, Holgate ST, Holloway JW, et al. Pharmacogenomics and outcome of asthma: Variant LTC(4) synthase allele modifies no clinical application for long-term steroid cysteinyl leukotriene synthesis in eosinophils effects by CRHR1 polymorphisms. J Allergy and predicts clinical response to zafirlukast. Clin Immunol. 2008;121(6):1510–3. Thorax. 2000;55 Suppl 2:S28-31.
57. Rogers A.J., Tantisira K.G., Fuhlbrigge 66. Kotani H, Kishi R, Mouri A, Sashio T, A.L., Litonjua A.A., Lasky-Su J.A., Szefler Shindo J, Shiraki A, et al. Influence of leuko- S.J., et al. Predictors of poor response during triene pathway polymorphisms on clinical asthma therapy differ with definition of responses to montelukast in Japanese outcome.Pharmacogenomics. 2009;10 (8): patients with asthma. J Clin Pharm Ther. 1231–42. 2012;37(1):112–6.
58. Acharya M, Borland G, Edkins AL, 67. Tcheurekdjian H, Via M, Giacomo, A.D, Maclellan LM, Matheson J, Ozanne BW, et al. Corvol H, Eng C, Thyne S, et al. ALOX5-AP CD23/FcεRII: molecular multi-tasking. Clin and LTA4H polymorphisms modify augment- Exp Immunol. 2010;162(1):12–23. ation of bronchodilator responsive-ness by leukotriene modifiers in Latinos. J Allergy Clin 59. Lima JJ, Zhang S, Grant A, Shao L, Immunol. 2010;126(4):853–8. Tantisira KG, Allayee H, et al. Influence of leukotriene pathway polymorphisms on 68. Mastalerz L, Nizankowska E, Sanak M, response to montelukast in asthma. Am J Mejza F, Pierzchalska M, Bazan-Socha S, et Respir Crit Care Med. 2006;173(4):379–85. al. Clinical and genetic features underlying the response of patients with bronchial 60. Telleria JJ, Blanco-Quiros A, Varillas D, asthma to treatment with a leukotriene Armentia A, Fernandez-Carvajal I, Jesus receptor antagonist. Eur J Clin Invest. 2002; Alonso M, et al. ALOX5 promoter genotype 32(12):949–55. and response to montelukast in moderate persistent asthma. Respir Med. 2008;102(6): 69. Fowler S.J., Hall IP, Wilson AM, Wheatley 857–61. A.P, Lipworth B.J. 5-Lipoxygenase poly- morphism and in-vivo response to leukotriene 61. Drazen J.M., Yandava C.N., Dubé L, receptor antagonists. Eur J Clin Pharmacol. Szczerback N, Hippensteel R, Pillari a, et al. 2002;58(3):187–90. Pharmacogenetic association between ALOX5 promoter genotype and the response 70. Currie G.P., Lima J.J., Sylvester J.E., Lee to anti-asthma treatment. Nat Genet. 1999; DKC., Cockburn WJR., Lipworth B.J. Leuko- 168–70. triene C4 synthase polymorphisms and responsiveness to leukotriene antagonists in 62. Tantisira KG, Lima J, Sylvia J, asthma. Br J Clin Pharmacol. 2003;56(4): Klanderman B, Weiss ST. 5-Lipoxygenase 422–6. pharmacogenetics in asthma: Overlap with Cys-leukotriene receptor antagonist loci. 71. Pasaje CFA, Kim J.H, Park B.L, Cheong Pharmacogenet Genomics. 2009;19(3):244– HS, Park T.J, Lee J.S, et al. Association of 7. the variants in AGT gene with modified drug response in Korean aspirin-intolerant asthma 63. Klotsman M, York TP, Pillai SG, Vargas- patients. Pulm Pharmacol Ther. 2011;24(5): Irwin C, Sharma SS, van den Oord EJCG, et 595–601. al. Pharmacogenetics of the 5-lipoxygenase biosynthetic pathway and variable clinical 72. Kim J.H, Lee S.Y, Kim H.B, Jin H.S, Yu response to montelukast. Pharmacogenet J.H, Kim B.J, et al. TBXA2R gene poly- Genomics. 2007;17(3):189–96. morphism and responsiveness to leukotriene receptor antagonist in children with asthma. 64. Asano K, Nakade S, Shiomi T, Nakajima Clin Exp Allergy. 2008;38 (1):51–9. T, Suzuki Y, Fukunaga K, et al. Impact of pharmacokinetics and pharmacogenetics on 73. Kang M.J, Kwon J.W, Kim B.J, Yu J, Choi the efficacy of pranlukast in Japanese W.A, Shin Y.J, et al. Polymorphisms of the asthmatics. Respirology. 2009;14(6):822–7. PTGDR and LTC4S influence
79
Responsiveness to leukotriene receptor anta- Lima JJ, Kubo M, et al. Genome-wide gonists in Korean children with asthma. J association study of leukotriene modifier Hum Genet. 2011;56(4):284–9. response in asthma. Pharmacogenomics J. 2015;1–7. 74. Lee S.Y, Kim H.B, Kim J., Kim B.S, Kang M.J, Jang S.O, et al. Responsiveness to 82. Dahlin A, Litonjua A, Lima JJ, Tamari M, montelukast is associated with bronchial Kubo M, Irvin CG, et al. Genome Wide hyperresponsiveness and total immuno- Association Study Identifies Novel Pharm- globulin E but not polymorphisms in the acogenomic Loci For Therapeutic Response leukotriene C4 synthase and cysteinyl leuko- to Montelukast in Asthma. PLoS One. triene receptor 1 genes in Korean children 2015;10(6):e0129385. with exercise-induced asthma (EIA). Clin Exp Allergy. 2007;37(10):1487–93. 83. Holzmann J, Frank P, Löffler E, Bennett K.L., Gerner C, Rossmanith W. RNase P 75. Kang M.J, Lee S.Y, Kim H.B, Yu J, Kim without RNA: identification and functional B.J, Choi W.A, et al. Association of IL-13 reconstitution of the human mitochondrial polymorphisms with leukotriene receptor tRNA processing enzyme. Cell. 2008;135(3): antagonist drug responsiveness in Korean 462–74. children with exercise-induced broncho- constriction. Pharmacogenetics Genomics. 84. Ellinghaus D, Ellinghaus E, Nair RP, 2008;18(7):551–8. Stuart PE, Esko T, Metspalu A, et al. Combined analysis of genome-wide associa- 76. Asano K, Shiomi T, Hasegawa N, tion studies for Crohn disease and psoriasis Nakamura H, Kudo H, Matsuzaki T, et al. identifies seven shared susceptibility loci. Am Leukotriene C4 synthase gene A (-444) C J Hum Genet. 2012;90 (4):636–47. polymorphism and clinical response to a CYS-LT (1) antagonist, pranlukast, in 85. Mizuochi T, Matthews TJ, Kato M, Japanese patients with moderate asthma. Hamako J, Titani K, Solomon J, et al. Pharmacogenetics. 2002;12(7):565–70. Diversity of oligosaccharide structures on the envelope glycoprotein gp 120 of human 77. Mougey EB, Feng H, Castro M, Irvin CG, immunodeficiency virus 1 from the lympho- Lima JJ. Absorption of montelukast is blastoid cell line H9. Presence of complex- transporter mediated: A common variant of type oligosaccharides with bisecting N- OATP2B1 is associated with reduced plasma acetylglucosamine residues. J Biol Chem. concentrations and poor response. 1990;265(15):8519–24. Pharmaco-genet Genomics. 2009;19(2):129 –38. 86. Lin C, Smith ER, Takahashi H, Lai KC, Martin-Brown S, Florens L, et al. AFF4, a 78. Balantic M, Rijavec M, Kavalar Skerbinjek component of the ELL/P-TEFb elongation M, Suskovic S, Silar M, Kosnik M, et al. complex and a shared subunit of MLL Asthma treatment outcome in children is chimeras, can link transcription elongation to associated with vascular endothelial growth leukemia. Mol Cell. 2010;37(3):429–37. factor A (VEGFA) polymorphisms. Mol Diagnosis Ther. 2012;16(3):173–80. 87. Peters-Golden M, Henderson W.R. Leukotrienes. N Engl J Med. 2007;357(18): 79. Kim K.A, Lee H.M, Joo H.J, Park I.B, Park 1841–54. J.Y. Effects of polymorphisms of the SLCO- 2B1 transporter gene on the pharmaco- 88. Lam B.K., Owen W.F., Austen K.F., kinetics of montelukast in humans. J Clin Soberman RJ. The identification of a distinct Pharmacol. 2013;53(11):1186–93. export step following the biosynthesis of leukotriene C4 by human eosinophils. J Biol 80. Tapaninen T, Karonen T, Backman JT, Chem. 1989;264(22):12885–9. Neuvonen PJ, Niemi M. SLCO2B1 c.935G>A single nucleotide polymorphism 89. In KH, Asano K, Beier D, Grobholz J, Finn has no effect on the pharmacokinetics of PW, Silverman EK, et al. Naturally occurring montelukast and aliskiren. Pharmacogenet mutations in the human 5-lipoxygenase gene Genomics. 2013;23(1):19–24. promoter that modify transcription factor binding and reporter gene transcription. J Clin 81. Dahlin A, Litonjua A, Irvin CG, Peters SP, Invest. 1997;99(5):1130–7.
80
90. Tantisira KG, Lima J, Sylvia J, 96. Ortega VE, Meyers DA. Pharmaco- Klanderman B, Weiss ST. 5-lipoxygenase genetics: implications of race and ethnicity pharmacogenetics in asthma: overlap with on defining genetic profiles for personalized Cys-leukotriene receptor antagonist loci. medicine. J Allergy Clin Immunol. 2014;133 Pharmacogenet Genomics. 2009;19(3):244– (1):16–26. 7. 97. Fleming L, Murray C, Bansal AT, 91. Hall IP, Blakey JD. Genetic association Hashimoto S, Bisgaard H, Bush A, et al. The studies in Thorax. Thorax. 2005;60(5):357–9. burden of severe asthma in childhood and adolescence: results from the paediatric U- 92. Gail MH, Pfeiffer RM, Wheeler W, Pee D. BIOPRED cohorts. Eur Respir J. 2015;46(5): Probability of detecting disease-associated 1322–33. single nucleotide polymorphisms in case- control genome-wide association studies. 98. Lane S, Molina J, Plusa T. An inter- Biostatistics. 2008;9(2):201–15. national observational prospective study to determine the cost of asthma exacerbations 93. Eberle M.A., Ng P.C., Kuhn K, Zhou L, (COAX). Respir Med. 2006;100(3):434–50. Peiffer DA, Galver L, et al. Power to detect risk alleles using genome-wide tag SNP panels. PLoS Genet. 2007;3(10):1827–37.
94. Freedman M.L., Reich D, Penney K.L., McDonald GJ, Mignault AA, Patterson N, et al. Assessing the impact of population stratification on genetic association studies. Nat Genet. 2004;36(4):388–93.
95. Price A.L., Patterson N.J., Plenge R.M., Weinblatt M.E., Shadick N.A., Reich D. Prin- cipal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38(8):904–9.
81
Online supplementary
Pharmacogenomics of inhaled corticosteroids and leukotriene modifiers: A systematic review
82
Table S1. PubMed search terms for LTMs.
"pharmacogenetics"[MeSH Terms] OR "pharmacogenetics"[All Fields]) OR "pharmacogenomics"[All Fields] OR "polymorphism, genetic"[MeSH Terms] OR "polymorphism"[All Fields] AND "genetic"[All Fields]) OR "genetic polymorphism"[All Fields] OR "polymorphism"[All Fields]) OR "polymorphism, genetic"[MeSH Terms]
AND
("leukotriene antagonists"[All Fields] OR "leukotriene modifiers"[All Fields] OR ("leukotriene antagonists"[Pharmacological Action] OR "leukotriene antagonists"[MeSH Terms] OR ("leukotriene"[All Fields] AND "antagonists"[All Fields]) OR "leukotriene antagonists"[All Fields] OR "antileukotrienes"[All Fields]))
Table S2. PubMed search terms for ICS
"pharmacogenetics"[MeSH Terms] OR "pharmacogenetics"[All Fields] OR "pharmacogenomics"[All Fields] OR "polymorphism, genetic"[MeSH Terms] OR "polymorphism"[All Fields] AND "genetic"[All Fields] OR "genetic polymorphism"[All Fields] OR "polymorphism"[All Fields] OR "polymorphism, genetic"[MeSH Terms]
AND
"inhaled corticosteroid"[All Fields] OR "inhaled glucocorticoids"[All Fields] OR "inhaled glucocorticosteroid"[All Fields]
83
84
Table 3S. Candidate gene studies of response to ICS.
Gene and SNPs Study population/ Ethnicity/ Definition of response Study outcome Study Design country
TBX21 (chr11) Pediatric Caucasian Change in FEV1% The estimated mean Tantisira et al. rs2240017 population increase in log-transformed 2004 (n=139)/ 4-year AHR to methacholine (PC20) PC20 was 3.5 fold more in clinical trial patients carrying mutant allele (MAF (G) = 0.04) compared to wild-type homozygotes (P=0.0001).
There was no association found with changes in FEV1%.
131 SNPs in 14 children and adults Caucasian Changes in FEV1% from The FEV1 % change was Tantisira et al. genes from: baseline 13.28 ± 3.11 and 17.80 ± 2004 6.77 for the minor allele of Ref [53]: Adult study rs242941 (MAF (T) = 0.3) (Adults, n=415) homozygous compared with ALOX15 5.49 ± 1.40 and 7.57 ± 1.50 (17p13.3) CAMP (Children, for those homozygous for n=201) the wild-type allele in adult CRH (8q13) study (P=0.02) and CAMP ACRN (Adults, (P=0.006), respectively. CRHBP (5q13.3) n=224)/ Three independent
84
CRHR1 6-8 week Clinical Homozygotes for the minor (17q21.31) trials T allele of rs1876828 had an average increase in their FCER2 (19p13.3) FEV1 of 23.72 ± 9.75 compared with 5.14 ± 1.31% GATA3 (10p15) for homozygotes for the common allele in patients HSD11B1 from ACRN. (1q32-q41)
IL18BP (11q13)
MAPK8 (10q11.22)
NFATC4 (14q11.2)
NR3C1 (5q31.3)
POMC (2p23.3)
STAT3 (17q21.31)
STAT5A (17q11.2)
IL-13 (5q31) Pediatric Caucasian ER visits and Hospitalizations (at Rs1800925 (MAF (T) = 0.2) Hunninghake et rs1800925 population least once during the first 4 was significantly associated al. 2007 rs20541 (n=503)/ years of the trial) with increased risk of
rs848 asthma exacerbations in
85
85
86
rs2243204 CAMP: Clinical white (non-Hispanic) rs1881457 trial children on ICS (P=0.02).
Costa Rica study: asthma cohort
FCER2 (19p13.3) Children from African- Severe exacerbations: Significant associations Tantisira et al. rs28364072 CAMP (n=311)/4- American and found between rs28364072 2007 year clinical trial Caucasian -Emergency room visits and/or and severe exacerbations in hospitalizations. both white and African- American children (hazard ratio, 3.95; 95% CI, 1.64- 9.51; and hazard ratio, 3.08; 95% CI, 1.00-9.47).
NR3C1 (5q31.3) Pediatric Caucasian Asthma control There was no association Szczepankiewicz population between these SNPs and et al. 2008 rs6190 (n=113)/ -Assessment of the increased response to ICS. rs41423247 observational demand for ICS. rs6195 study rs10052957
CRHR1 Adults (n=98)/ Not mentioned Changes in FEV1 from baseline No associations were found Dijkstra et al. (17q21.31) Observational (immediate effect) between SNPs and the 2008 rs1876828 study outcomes. rs242941 Rate of decline in FEV1 annually rs242939 (long-term effect)
IPO13 (1p34.1) Pediatric Caucasian changes in methacholine airway There was no association Raby et al. rs6671164 population (CAMP hyper-responsiveness (AHR- between these SNPs and 2009 rs4448553 PC20). change in methacholine rs1990150
86
rs2240447 children, n=654)/ PC20 in patients treated with rs2486014 8-month budesonide. rs2301993 rs2301992 Clinical trial rs1636879 rs7412307 rs2428953
TBX21 (chr11) Adults (n=53)/ Korean Asthma symptoms control based Among mutant allele carriers Y-M Ye et al. 12-week clinical on GINA guidelines for rs2240017 (MAF (G) = 2009 rs2240017 C>G trial 0.12) and G231E G>A (MAF Change in FEV1% predicted (A) = 0.4) higher proportion NK2R (10q22.1) of patients had not well- Changes in PEFR% predicted controlled asthma compared G231E G>A to wild-type carriers (P= Changes in AQLQ 0.006 and 0.041, ADRB2 respectively). (5q31-q32): There was a significant 46 A>G association between G231E G>A and changes in ADCY9 FEV1%pred. Patients (16p13.3): carrying G allele had 19.53% improvement in 1772M T>C FEV1% pred compared to A allele carries with 15.75%
improvement in FEV1% pred (P= 0.03). There was no
statistically significant association between changes in FEV1%pred and
87 rs2240017. None of the
87
88
other SNPs showed any statistically significant associations with the outcomes.
CRHR1 Children from African- Exacerbations: The minor allele (MAF Rogers et al. (17q21.31) CAMP (n=311)/4- American (T)=0.3) of rs242941 2009 rs242941 year clinical trial and - Emergency Department visit showed a significant Caucasian association with poor - Hospitalization response based on lung function (OR: 1.6, CI 95% 1– - Oral prednisone burst. 2.7, P=0.05).
FCER2 (19p13.3) Lung function measurements: The minor allele of rs28364072 change in FEV1% predicted: rs28364072 in FCER2 was associated with recurrent ≤7.5% considered as poor exacerbations in patients responders with Caucasian ethnicity (OR: 1.9 for minor allele, P < 0.05) but it did not reach significance in multivariate analysis (OR: 1.4, p = 0.18).
The minor allele of rs28364072 was also associated with poor lung function (OR 2.1 1.2–3.5, P= 0.006).
HSPCB (6p12) Adults (n=382)/ Caucasian Changes FEV1% from baseline FEV1% change at 4 weeks Hawkins et al. for rs6591838 (MAF (G) = 2009 0.2) (18.41±26.15 for GG vs
88
HSPCA 4-8 week Clinical 4.57±16.82 for AA) and (14q32.33) trial rs2236647 (MAF (T) = 0.4) (12.05±23.99 for TT vs HSPA8 5.23±17.52 for CC) within (11q24.1) STIP1 were significantly DNAJB associated with the (19p13.2) response to ICS (P= 0.02 and 0.04, respectively). PTGES3 (12q13.3) rs6591838 (MAF (G) = 0.2)
(20.7 ±28.29 for GG vs FKBP5 (6p21.31) 5.5±17.63 for AA) and rs1011219 (MAF (A) = 0.12) FKBP4 (6.92±18.74 for GG vs - (12p13.33) 36.85±18.59 for AA) were associated with FEV1% STIP1 change at 8 weeks (P= 0.01 (11q13) and 0.005, respectively).
DUSP1 (5q34) Adults (n=165)/ African- Change in FEV1% from There was no association Jin et al. rs881152 6-week clinical trial American baseline. between rs881152 (MAF (A) 2010 = 0.14) and changes in Changes in self-reported asthma FEV1% but the SNP was control (ACT) significantly associated with ACT score changes over 6 weeks of treatment with ICS. Mean change for each genotype: GG: 2.81±6.22, AG: 1.38±7.36 and AA: -
3.5±9.13 (P= 0.01)
89
89
90
CTLA4 (2q33) Pediatric Slovenian Changes in %FEV1 pred Atopic asthmatics Berce et al. rs3087243 population homozygous for the A allele 2010 (n=102)/ 4-week AHR to Methacholine PC20 and of rs3087243 (MAF (A) = clinical trial logPC20 FEV1/FVC ratio. 0.4), had 21.7% increase in FEV1%pred compared with an 8.6% increase in heterozygotes and a 5.8% increase in G allele homozygotes (P <0.01).
ARG1 (6q23) Adults (n=200)/ The Annual FEV1 decline Annual FEV1 decline was Vonk et al. Observational Netherlands significantly higher in 2010 rs2781667 study changes in FEV1 upon homozygotes for the C allele rs2781668 bronchodilator. of the ARG1 rs2781667 rs17599586 (MAF (T)= 0.28) compared to the T allele carriers. (19.9ml/year, 95% CI: 4.8– 35.0) (P <0.01). ARG2 (14q24.1) Patients carrying mutant rs2145467 allele for rs7140310 (MAF rs2295643 (G) = 0.13) and rs10483801 rs17249437 (MAF (A) = 0.16) within rs17249444 ARG2 had higher rs12896052 bronchodilator response rs7140310 (P=0.007 and P=0.03, rs3742879 respectively). The other rs10483801 investigated SNPs in ARG1 and ARG2 did not show a significant association with the FEV1 decline.
90
WDR21A adolescents and Korean Changes in FEV1% pred from Patients homozygous for Cho et al. (14q24.3) adults (n=230)/ 4- baseline variant allele of rs7160796 2011 rs7142086 week clinical trial (MAF (C) = 0.24) had a rs3213729 greater change in FEV1 rs7160796 (58.04%) compared to rs12892806 homozygotes for wild-type rs2302588 (24.51%) or heterozygotes rs12885397 (24.51%) (P= 0.04). rs3742832 rs2240981 rs2806039 rs7143683 rs7144738
CDH1 (16q22.1) Two separate Northern Annual decline in FEV1 The minor alleles of Ierodiakonou et al. study populations region of the rs8056633 (G), rs16958383 2011 [ref 31] rs2902185 including 57 and Netherlands post bronchodilator FEV1/VC (A), rs7203904 (C) and rs11075699 146 adults using ratio rs17690554 (G) SNPs were rs1125557 ICS/ significantly associated with rs121259718 Observational less FEV1 decline. rs7199991 study Rs7188750 (A), rs7186053 rs4783573(A), rs3785078 rs10431924 (C), rs8056633 (G), rs4783573 rs16958383 (A), rs7203904 rs7188750 (C) and rs17690554 (G) rs8056633 SNPs were associated with rs4783689 higher post bronchodilator rs16958383 FEV1/VC in the presence of rs2276330 ICS in population 2. rs1801552 Rs7203904 had a rs3785078 statistically significant
91 association with post
91
92
rs7203904 bronchodilator FEV1/VC in rs17690554 population 1.
The minor alleles of rs1125557 (G), rs7199991 (C) and rs7186053 (A) were significantly associated with accelerated FEV1 decline in population 2.
SERPINE1 Adults (n=281)/ The Changes in FEV1 from baseline ICS treatment showed an Dijkstra et al. (7q22.1) Observational Netherlands immediate improvement in 2011 study FEV1 in asthmatics carrying rs1799889: the 5G allele (4G/5G or 5G/5G) (increase in mean 4G/4G FEV1: 189.9 mL, 95% CI 30.5–349.2 mL) this 4G/5G improvement was absent in 4G/4G (-63.1 mL, 95% CI - 5G/5G 298.3–172.1 mL) (P= 0.04).
CD14 (5q31.1) Pediatric Slovenian Changes in FEV1% pred from No association between the Perin et al. rs2569190 population baseline. SNP and the response to 2011 (-159C/T) medication. (n=247)/ 4-week clinical trial
FCER2 (19p13.3) Children and Caucasian Exacerbations: In the meta-analysis, the Koster et al. rs28364072 adolescents from Emergency department visits rs28364072 variant was 2011 three study Hospitalization associated with increased populations: risk of asthma-related hospital visits (OR: 2.38,
92
PACMAN Asthma symptoms: 95%CI: 1.47–3.85, P= (n=386) ACQ- scores 0.0004). Respiratory symptoms (wheeze, BREATHE shortness of breath and cough) The FCER2 variant was (n=939) Asthma-related sleep associated with increased disturbances risk of uncontrolled asthma CAMP Asthma-related limitations in measured by ACQ scores (n=311) daily activities (OR: 2.64, 95%CI: 1.00– Additional (airway) medication 6.98) and was associated use during the preceding 12 with increased daily steroid months. dose (OR: 2.46, 95%CI: 1.38–4.39).
VEGFA (6p12) Pediatric Slovenia Changes in FEV1% pred Patients with AA genotype Balantic et al. rs2146323 population for rs2146323 (MAF (A)= 2012 rs833058 (n=131)/12-month Changes in FEV1/FVC 0.3) showed a greater clinical trial improvement in the FEV1% ACT (scores <20 defined as pred (9% increase) than uncontrolled) patients with AC or CC (5% increase) after 12 months (P= 0.03).
No association was found with ACT and FEV1/FVC.
HDAC1 (1p34) Pediatric Korean Changes in FEV1% from Subjects with the CC M. Kim et al. rs1741981 population (n=70) baseline genotype of rs1741981 2013 /8-week clinical (MAF (C) = 0.47) showed HDAC2 (6q21) trial significantly lower FEV1% rs58677352 (14.1% ± 5.9%) increases in response to ICS compared
93 with the CT or TT genotype
93
94
CYP3A4 (7q21.1) carriers (19.4% ± 8.9%) (P = rs35599367 0.03). rs2246709 rs4646437 There was no association between rs58677352 and lung function.
CYP3A5 (7q21.1) Pediatric Hispanics and Asthma control scores ranged Among children with Stockmann et al rs776746 population non-Hispanics from 0 (well-controlled) to 15 rs35599367 (CYP3A4*22) 2013 rs10264272 (n=268)/ (poorly-controlled). variant allele (T, rs15524 observational frequency=0.04), average study asthma control scores were CYP3A7 (7q22.1) significantly lower (2.9 ± 2.2) rs2687133 compared with the CC rs2257401 carriers (5.0 ± 3.7) (P=0.02). rs2740565
ORMDL3 (17q12) Pediatric Caucasian Change in FEV1% pred Patients with AA genotype of Berce et al. rs2872507 population rs2872507 (MAF (G) = 0.38) (n=311)/ AHR to methacholine (PC20) showed statistically 2013 observational significant improvements in study Changes in FEV1/FVC FEV1% pred (13.3%) compared to GG carriers (4.9%) (P= 0.02).
94
TBX21 (chr11) Adults (n=208)/ White Change in FEV1% pred Patients with AA genotype of Lopert et al. rs9910408 observational asthmatics rs9910408 (MAF (G) = 0.4) 2013 study AHR to methacholine (PD20) had a greater decrease of AHR compared to GG Changes in ACT and AQLQ genotype (P= 0.049, scores OR:2.74, 95%CI 1.06-7.06).
Patients with AA genotype had a greater improvement of FEV1 compared to GG carriers (P= 0.01, OR:5.78, 95%CL 1.49-22.37)
Non-atopic patients with AA had higher improvements in AQLQ scores compared to GG (P= 0.04, OR:23.82, 95% CI: 1.05-542.70). In total study population (atopic + non-atopic), no association was found between rs9910408 and ACT/AQLQ scores.
GLCCI1 (7p21.3) adolescents and Caucasian Changes in FEV1 from baseline There was no significant Hosking et al. rs37973 adults (n=1916)/ association between 2014 Seven 8-12 week changes in FEV1 and
clinical trials rs37973 genotypes.
95
95
96
GLCCI1 (7p21.3) Adults (n=224)/ Japanese Annual decline in FEV1 Rs37973 GG was Izuhara et al. rs37973 Observational associated with a decline in 2014 study FEV1 of 30ml/year or more STIP1 (11q13) (estimated effect: 1.10, 0.02 rs1011219 to 2.18 0.047) rs6591838 There was no association T gene (6q27) between rs37973 genotypes rs3127412 and the outcome, when decline in FEV1 was analyzed as a continuous variable.
There was no association between annual decline in FEV1 and STIP1 and T gene SNPs.
GLCCI1 (7p21.3) Children and Caucasian Exacerbations: There was no significant Vijverberg et al. rs37972 adolescents from Hospital visits association between 2014 three study OCS use increased risk of OCS use, populations: increased risk of asthma BREATHE Asthma symptoms: exacerbations and rs37972 (n=1037) ACT scores ≤ 19 (not well genotypes PACMAN (n=431) controlled) PAGES (n=323)/ - ACQ-scores ≥ 1.5 (not well Meta-analysis of controlled) three pediatric asthma cohorts
96
IL-10 (1q31-q32) Adolescents and South India Changes in FVC (mL) from No statistically significant Raeiszadeh rs1800871 adults (n=419)/ baseline. difference among SNPs Jahromi et al. rs1800896 variants and lung function 2015 2 months Case- Changes in FEV1 (mL) from measurements. P < 0.016 IL-17F (6p12) Control baseline. was considered as rs1889570 statistically significant by applying Bonferroni correction.
TAAR6 (6q23.2) Adolescents and Korean Changes in %FEV1 from The changes in FEV1% was Chang et al. 15 SNPs adults (n=246)/ 4- baseline higher for the GG carriers of 2015 week Prospective rs7772821 (MAF (G) = 0.22) controlled trial (60.77%) than the patients with the T/G or T/T genotypes (21.32 and 31.60%, respectively) (P=0.002 in the co-dominant model).
NR3C1 (5q31.3): children and Caucasians Asthma exacerbations: Major allele (G) of the Vijverberg et al. 9 SNPs adolescents: and Hispanics rs138335 SNP within the 2015 PACMAN (n=357) Asthma-related hospital visits ST13 gene was associated FKBP4 with increased risk of (12p13.33): BREATHE OCS use as a rescue medication asthma-related hospital 2 SNPs visits OR = 1.22 (P = 0.013) (n=820) and OCS use OR = 1.22 (P ST13 (22q13.2): = 0.0017), in the past 12 2 SNPs PAGES months.
(n=391)
97
97
98
SERPINA6 CAMP (n=172) (14q32.1): 8 SNPs PASS (n=172)
CREBBP GALA 2 (n=745) (16p13.3): 4 SNPs
TBP (6q27): 2 SNPs
NCOA3 (10q11.2): 5 SNPs
SMAD3 (15q22.33): rs744910
ARG1 (6q23): rs2781667
17q21 locus: rs7216389
IL2RB (22q13.1): rs2284033
PDE4D (5q12): rs1544791 rs1588265
98
17q12-21 locus Pediatric Caucasian Changes in FEV1%pred and rs72821893 (MAF (T) = Leusink et al. population AHR to Mch (PD20) 0.03) in KRT25 was 2015 47 SNPs ref [28] (n=110)/Clinical significantly associated with trial Poor treatment response: FEV1%pred (P = 3.75x10-5), Mch PD20 (P = 0.00095) Decrease in FEV1%pred and and poor treatment low PD20 response based on Mch PD20 (P = 0.006). despite high levels of treatment during the trial despite high levels of treatment.
ACQ, Asthma Control Questionnaire; ACRN, Asthma Clinical Research Network; ACT, Asthma Control Test; AHR, airway hyper-responsiveness; AQLQ, Asthma quality of life questionnaire; CAMP, Childhood Asthma Management Programme; ER, emergency room; FEV1, forced expiratory volume in 1 second; FEV1% pred, percentage of predicted FEV1; FVC, Forced vital capacity; ICS, inhaled corticosteroids; OCS, oral corticosteroid; OR, odds ratio; PACMAN, Pharmacogenetics of Asthma Medication in Children: Medication with Anti-Inflammatory effects; PC20, provocation
concentration causing a 20% fall in FEV1; PD20, provocation dose causing a 20% decline in FEV1.
99
99
100 Table S4. Candidate gene studies of response to LTMs
Gene and SNPs Study population/ Ethnicity Definition of Study outcome Study Design /country response
ALOX5 (10q11.2) adult asthmatics Not Changes in The mean change in FEV1 from baseline Drazen et al. tandem repeats of (n=114)/12-week mentioned FEV1% from was 18.8±3.6% in wild-type (5/5) patients 1999 the Sp1-binding Clinical trial baseline and 23.3±6.0% in heterozygous (5/X) domain patients compared to the patients with the mutant genotype (X/X) with the change in FEV1 of −1.2±2.9% (p < 0.0001 and p < 0.0006) (MAF (X) = 0.23).
LTC4S (5q35) Adults (n=23)/2-week Not FEV1% change No significant association found Sampson rs730012 clinical trial mentioned from baseline et al. 2000
FVC % change from baseline
PEF% change from baseline.
ALOX5 (10q11.2) adult atopic Scotland AHR to AMP There were no statistically significant Fowler et al. tandem repeats of asthmatics (n=52)/ differences between heterozygotes and 2002 the Sp1-binding Retrospective analysis homozygotes for the wild-type allele (MAF domain of 4 placebo-controlled (X) = 0.11). trials
100
LTC4S (5q35) Adults with EIB Japanese Changes in C allele carriers responded better to the Asano et al. rs730012 (n=50)/4-week maximal % fall in treatment compared with the A allele 2002 clinical trial FEV1 after homozygotes (P < 0.01). exercise challenge test from baseline.
Responders were defined as children showing 10% post- treatment improvement.
LTC4S (5q35) Adults (n=26)/8-week Not Responders we No significant association found. Mastalerz rs730012 clinical trial mentioned defined as: et al. 2002 Increase in PEF ≥7%
Improvement of daytime and nocturnal symptoms by a score of ≥0·5
Decrease in the use of b2-agonists by ≥50%,
Improvement in the Quality of Life Questionnaire
(QLQ) by ≥1
101
101
102
LTC4S (5q35) Children/adolescences Caucasian Changes in FENO No statistically significant results Whelan et al.
rs730012 (n=12)/3-week clinical and African- from baseline 2003 trial American Changes in FEV1% pred from baseline
LTC4S (5q35) (n=78) Not mentioned/ Not AHR to AMP and No association significant found Currie et al. rs730012 retrospective analysis mentioned Methacholine. 2003 of 8 randomized, placebo-controlled trials
ALOX5 (10q11.2): adult asthmatics Caucasian Changes in Significant associations found between Lima et al. 6 SNPs (n=61)/6-month FEV1% pred from rs2115819 (MAF (G) = 0.47) in ALOX5 2006 Clinical trial baseline and rs119774 (MAF (T) = 0.07) in MRP1 LTA4H (12q22): and the changes in FEV1% pred p<0.05. 3 SNPs Asthma exacerbation rate Significant association found between LTC4S (5q35): rs730012 (MAF (C) = 0.3) (LTC4S), 3 SNPs - One or more of rs2660845 (MAF (G) = 0.29) (LTA4H) and the following: a exacerbation rates (p values 0.023 and MRP1 (16p13.1): more than 30% <0.001 respectively) 13 SNPs decrease in peak expiratory flow rate Patients homozygous for he mutant allele CYSLTR1 for two consecutive of tandem repeats (MAF (X) = 0.2) in (Xq13.2-q21.1): days, a course of ALOX5 had significant reduction in 4 SNPs oral steroids, an exacerbation rates (P=0.04). unscheduled visit to the clinic, the emergency room, or hospital or an increase of four
102
puffs of rescue inhaler use in one day
LTC4S (5q35): Asthmatic children Korean Changes in No significant association found S-Y Lee et al. rs730012 with EIB (n=100)/8- maximal % fall in 2007 week clinical trial FEV1 after CYSLTR1 exercise challenge (Xq13.2-q21.1): test from baseline. rs320995 Responders were defined as children showing 10% post- treatment improvement
PLA2G4A Adolescents and Caucasian, Changes in Significant associations found between Klotsman et al. (1q31.1): 2 SNPs adults (n=166)/12- Hispanic and morning PEF 2007 week Clinical trial African- change in morning PEF and rs912277 LTC4S (5q35): American Changes in (MAF (G) 0.07) (P=0.02) and rs912278 rs730012 FEV1% pred from (MAF (G) = 0.41) within CYSLTR2 baseline (P=0.02), rs4987105 (MAF (T) = 0.19) CYSLTR1 (P=0.01), rs4986832 (MAF (A) = 0.19) (Xq13.2-q21.1): (P=0.01) and tandem repeats of the core rs320995 promoter (MAF (X) = 0.24) (P=0.05) within ALOX5. CYSLTR2 (13q14.2): 2 SNPs Significant associations found between the changes in FEV1% pred and rs6188 (MAF ALOX5AP (A) = 0.29) (P=0.01) and rs6196 (MAF (G)
(13q12): 3 SNPs = 0.14) (P=0.02) within NR3C1, rs1042714
103
103
104
CYP3A4 (7q21.1): (MAF (G) = 0.39) (P= 0.02) and rs1042711
2 SNPs (MAF (C) = 0.39) (P=0.02) within ADRB2.
CYP2C9 (10q21): rs1799853
ADRB2 (5q31- q32): 3 SNPs
NR3C1 (5q31.3): 4 SNPs
ALOX5 (10q11.2): 5 SNPs
ALOX5 (10q11.2) Adolescent and adult Spain Changes in Homozygotes for the mutant allele had a Telleria et al. tandem repeats of atopic asthmatics FEV1% pred from rate of asthma exacerbations of 1.88±0.92, 2008 the Sp1-binding (n=61)/6-month baseline with a mean reduction of 1.33±1.22 and domain clinical trial patients homozygous for the common Number of allele and heterozygotes had an exacerbations: exacerbation rate of 0.4±0.21 with a reduction of 4.41±2.76 (P= 0.001) (MAF (4 -A decrease in tandem repeats) = 0.4) peak expiratory flow rate in the Homozygotes for the common allele and morning of at least heterozygotes had a mean increase of 25% compared 8.6%±7.6 in FEV1%pred compared to the with baseline. homozygotes for the mutant allele with a mean decrease of 1.4%±8.04 in FEV1% pred (P=0.0006).
104
IL-13 (5q31) Asthmatic children Korean Changes in Significant association found between - Kang M-J et with EIB (n=80)/ maximal % fall in 112C/T (MAF (T) = 0.2) and the outcome al. 2008 – 1512A/C 8-week clinical trial FEV1 after (P=0.02). exercise challenge – 1112C/T test from baseline.
+ 2044G/A Responders were defined as children showing 20% post- treatment improvement.
TBXA2R Asthmatic children Korean Changes in No significant associations were found Kim J-H et al. (19p13.3): with EIB (n=100)/8- maximal % fall in between any of the SNPs and the 2008 +795T>C week clinical trial FEV1 after outcome. +924T>C exercise challenge test from baseline. In combination of 2 SNP, patients with the +924T>C TT (MAF (C) = 0.2) homozygote Responders were and +795T>C (MAF (C) = 0.36) hetero- or defined as children homozygote (CT or CC) had a 3.67-fold showing 10% post- poorer response compared with the treatment patients with the common alleles (OR: improvement. 3.67; 95% CI: 1.15–11.15).
LTC4S (5q35): Adults (n=50)/4-weeks Japanese Changes in Significant association found between Asano et al. rs730012 clinical trial %FEV1 from increases in FEV1% and C allele carriers 2009 baseline (MAF (C) = 0.34) (P = 0.03)
ALOX5 (10q11.2): Children, adolescence Caucasian Changes in There were significant associations with Tantisira et al. 4 SNPs and adults (n=239)/ FEV1% pred from the outcome for several SNPs: ALOX5 2009 12-week clinical trial baseline (rs2115819) (P=0.01), (rs892690)
105 (P=0.01), (rs2029253) (P=0.03). MRP1
105
106
CysLTR1 (rs119774, rs215066) (P=0.05). LTC4S
(Xq13.2-q21.1): (rs272431) (P=0.005). None of these 4 SNPs findings met the Bonferroni correction adjusted P values. LTA4H (12q22): 3 SNPs
LTC4S (5q35): 2 SNPs
MRP1 (16p13.1): 13 SNPs
ALOX5AP Children and Hispanic Change in FEV1% Significant associations were found for two Tcheurekdjian (13q12): adolescents (n=649)/ (Puerto upon SABA SNPs within the LTA4H gene (rs2540491 et al. 2010 rs10507391 Clinical trial Ricans and (albuterol) (MAF (A) = 0.37) (P=0.008) and rs9551963 Mexicans) rs2540487 (MAF (A) = 0.2) (P=0.03)) and the outcome. After stratifying by ethnicity, LTA4H (12q22): these associations were found only for rs17525488 Puerto Ricans. rs2540493 rs2540491 No significant association between SNPs rs2540487 within ALOX5AP and other SNPs of LTA4H and the outcome.
LTC4S (5q35): Asthmatic children Korean Changes in Statistically significant association with Kang et al. rs730012 with EIB (n=92)/ maximal % fall in rs803010 (PTGDR, MAF (C)= 0.25)) 2011 8-week clinical trial FEV1 after (P=0.04). Higher number of non- PTGDR (14q22.1): exercise challenge responders were heterozygous and rs803010 test from baseline. homozygous for the C allele. Two SNPs rs8004654 within PTGDR are in high LD (D’=0.978). Responders were defined as children
106
showing 10% post- No association found for rs730012 treatment (LTC4S, (MAF(C)=0.2). improvement.
AGT (1q42.2): Adults with Aspirin Korean Aspirin induced Significant associations found between 2 Pasaje et al. 38 SNPs Induced Asthmatics rate of FEV1% SNPS (+2401C>G (rs2478544, MAF (G)= 2011 Ref [69] decline (>15% 0.25) (P=0.0009, after correction 0.02) and (AIA) (n=56)/12-week non-responders, +2476C>T (rs2478545, MAF (T) = 0.29) Clinical trial <15% responders) (P= 0.002 after correction 0.03) with aspirin induced rate of FEV1% decline.
VEGFA (6p12): Two pediatric Slovenia Changes in Among regular montelukast users, Balantic et al. rs2146323 populations (n=91)/ %FEV1 pred from rs2146323 AA genotype was associated 2012 rs833058 12-month clinical trial baseline with uncontrolled asthma (OR 0.07, 95% CI: 0.009-0.57 (P=0.02)) and for patients Changes in using montelukast episodically, AA FEV1/FVC genotype was associated with reduced FEV1/FVC ratio (P=0.04). ACT scores (<20 uncontrolled) Homozygotes for the T allele of rs833058 had improvements in FEV1% pred. However, CC or CT carriers had no improvement in FEV1% pred (P = 0.03)
ALOX5 (10q11.2) Adult (n=21)/ 4-8 week Japanese Self-measured Significant associations were found Kotani et al. rs2115819 Clinical trial changes in PEF between genotypes of the rs2660845 2012 (L/min) (MAF(A)=0.42) and changes in PEF and LTA4H (12q22) FEV1 (P < 0.05). In AA genotype carriers, rs2660845 Changes in FEV1 PEF significantly improved (more than 50 (L) after SABA L/min) compared with G allele carriers with
a very low improvement in PEF (P < 0.05).
107
107
108
G allele carriers had no improvements in FEV1 compared with AA carriers that showed more than 0.2 L improvements (P<0.05) No significant association between the rs2115819 and the outcomes were reported.
169 SNPs within Children, adolescents Caucasian Changes in Homozygotes for the major alleles of Mougey et al. 25 genes: and adult (n=60)/16- FEV1% pred from rs739645, rs1876831 (MAF (T) = 0.2), 2013 week clinical trial baseline rs1876829 (MAF (C)= 0.2) and rs1876828 ABCC1 (16p13.1) (MAF (T) =0.2) within CRHR1 had ABCC6 (16p13.1) Slopes of plots of improvements in FEV1% pred (4.6 ±1.8%) ADRB2 (5q31- ACQ scores compared to homozygotes for the minor q32) ALOX5 versus time alleles with no improvements in FEV1% (10q11.2) pred from baseline (P=8.6 x 10-3). ALOX5AP (13q12) CHRM2 (7q31- Rs242950 (MAF (T) = 0.09) within CRHR1 q35) COL2A1 was significantly associated with ACQ (12q13.11) slope. Heterozygotes had reduced asthma CPAMD8 symptoms whereas homozygotes for the (19p13.11) CREM major allele had an increase in symptoms (10p11.21) (P=3.57 x 10-3). CRHR1 (17q21.31) Homozygotes for the minor allele and CYSLTR1 heterozygotes had significant (Xq13.2-q21.1) improvements in FEV1% pred from F2RL3 (19p12) baseline compared to homozygotes for the FBN2 (5q23.3) major allele of rs3757016 (MAF (T) = 0.4) HAL (12q23.1) within HDAC2. (P=8.63 x 10-3). HDAC2 (6q21) HSPA8 (11q24.1) IMP5 (13q32.2)
108
LTA4H (12q22) LTC4S (5q35) MGAT4B (5q35) NR3C1 (5q31.3) STIP1 (11q13) TFCP2 (12q13) TNF (6p21.3) URP2 (11q13.1)
SLCO2B1 (11q13) 20 African-American, African- Montelukast Individuals with GG genotype for Mougey et al. rs12422149 55 Caucasian and 5 American, plasma rs12422149 had significant increases in 2009 rs2306168 Hispanic adult Caucasian concentration ASUI after one and 6 months of therapy asthmatics (in total 80 and Hispanic compared to heterozygotes (P<0.0001). patients)/clinical trial Asthma symptoms There was no homozygous for A (minor utility index (ASUI) allele) in the study population (MAF (A) was used to ranged between 0.08 – 0.2 for different measure response ethnicities). Heterozygotes for rs12422149 to montelukast had significantly lower plasma concentration (3±1 ng/ml) compared to GG (7±0.9 ng/ml) carriers (P=0.02). No significant associations found between rs2306168 and plasma concentrations of montelukast nor asthma symptoms.
SLCO2B1 (11q13) Healthy individuals Finland Montelukast The rs12422149 genotype had no Tapaninen et Rs12422149 (n=33)/clinical trial plasma significant effect on the plasma al. 2013 concentration concentrations of montelukast.
SLCO2B1 (11q13) Healthy individuals Korean Montelukast No association found between these SNPs K. Kim et al. Rs12422149 (n=24)/clinical trial plasma and plasma concentrations of montelukast. 2013
c.1175C>T concentration 109
c.1457C>T
109
110
c.43C>T c.601G>A c.644A>T g.‐282G>A
ACT, Asthma Control Test; AHR, airway hyper-responsiveness; AMP, adenosine monophosphate; EIB, Exercise-induced bronchoconstriction; FEV1, forced expiratory volume in 1 second; FEV1% pred, percentage of predicted FEV1; FVC, forced vital capacity; OR, odds ratio; PEF, forced expiratory flow; SABA, short acting beta agonist; SP1, specificity protein 1.
110
111
112
Pharmacogenomics CHAPTER 3 in childhood asthma
113
114
CHAPTER 3.1
Rationale and design of the multi-ethnic Pharmacogenomics in Childhood Asthma (PiCA) consortium
Farzan N, Vijverberg SJ, Andiappan AK, Arianto L, Berce V, Blanca-López N, Bisgaard H, Bønnelykke K, Burchard EG, Campo P, Canino G, Carleton B, Celedón JC, Chew FT, Chiang WC, Cloutier MM, Daley D, Den Dekker HT, Dijk FN, Duijts L, Flores C, Forno E, Hawcutt DB, Hernandez-Pacheco N, de Jongste JC, Kabesch M, Koppelman GH, Manolopoulos VG, Melén E, Mukhopadhyay S, Nilsson S, Palmer CN, Pino-Yanes M, Pirmohamed M, Potočnik U, Raaijmakers JA, Repnik K, Schieck M, Sio YY, Smyth RL, Szalai C, Tantisira KG, Turner S, van der Schee MP, Verhamme KM, Maitland-van der Zee AH
Pharmacogenomics, 2017;18(10):931-943
115
Abstract
Objective: International collaboration is needed to enable large-scale pharmaco- genomics studies in childhood asthma. Here, we describe the design of the Pharmaco- genomics in Childhood Asthma (PiCA) consortium.
Material & Methods: Investigators of each study participating in PiCA provided data on the study characteristics by answering an online questionnaire.
Results: A total of 21 studies, including 14,227 children/young persons (58% male), from 12 different countries are currently enrolled in the PiCA consortium. Fifty-six percent of the patients are Caucasians. In total 7,619 were inhaled corticosteroid (ICS) users. Among patients from 13 studies with available data on asthma exacerbations, a third reported exacerbations despite ICS use. In the future pharmacogenomics studies within the consortium, the pharmacogenomics analyses will be performed separately in each center and the results will be meta-analyzed.
Conclusions: PiCA is a valuable platform to perform pharmacogenetics studies within a multi-ethnic pediatric asthma population.
116
Introduction
Asthma is the most common chronic disease in childhood. Although it cannot be cured, effective treatments are available to decrease the symptoms, maintain lung function and prevent future exacerbations (1). Standard treatment regimens for persistent asthma include regular use of inhaled corticosteroids (ICS) combined with long-acting β2 agonists (LABA) and short-acting β2 agonists (SABA) as needed (2). There is heterogeneity in response to treatment; approximately 30-40% of the patients receiving ICS, do not show an improvement in lung function and remain uncontrolled (3–6). Uncontrolled asthma is associated with low quality of life for patients and can be life threatening (7,8). Furthermore, unscheduled physician visits and hospital admissions due to exacerbations are responsible for almost half of the costs of asthma management (9,10). Poor adherence to medication, ongoing environmental exposures, disease severity and misdiagnosis influence response to treatment in asthmatic patients. In addition, it has been shown that genetic variation contributes to the heterogeneity in treatment response (11). To date, a large number of candidate gene studies and several genome-wide association studies (GWAS) have been conducted to study the pharmacogenomics of childhood asthma (12,13). However, one of the main unmet needs for pediatric asthma management is the lack of clinically available biomarkers (for example pharmacogenetic markers) to guide asthma treatment. Genetic associations have been reported with three commonly used outcome measures (i.e. asthma exacerbations, asthma symptoms and lung function) (14,15). Different outcomes might reflect different aspects of asthma control and the heterogeneity in the outcome measures complicates the comparison of study results. In addition, most studies have been performed in relatively small study populations. There is a need for international collaboration in the field of pharmacogenomics of asthma to obtain large sample sizes of well-phenotyped asthmatic children to perform large scale meta- analyses to assess the clinical value of genetic markers for asthma management and identify markers that can guide asthma treatment (16,17). There have been successful efforts to establish consensus on diagnosis and management of asthma (18,19). The Pharmacogenomics in Childhood Asthma (PiCA) consortium was initiated in December 2013 and brings together asthma studies that have genetic data and treatment outcome measures. The main goals of the PiCA consortium are to create a platform to identify new pharmacogenomic markers in asthma by conducting GWAS meta-analyses. To replicate these new and also previously identified loci that are
117
associated with treatment response, and finally, to develop pharmacogenetics-guided algorithms to guide asthma therapy to improve symptoms and reduce/prevent future exacerbations. This is the first consortium that focuses on pharmacogenomics in child- hood asthma. In this study, we describe the characteristics of the study populations currently included in the PiCA consortium, assess the outcome measures that can be used to study treatment response within the consortium and describe the design of the pharmacogenomics studies that will be performed within PiCA.
Methods PiCA consortium The PiCA consortium was established in December 2013 by the pharmacogenomics research group of Prof. dr. AH Maitland van der Zee (Utrecht University, The Netherlands) by expanding existing and new collaborations. Studies were identified from the literature, at conferences and by references of other PiCA collaborators. Studies were eligible to participate in the PiCA consortium if: • Data on asthmatic children or young persons were collected; • DNA samples were collected or could be collected; • Data were collected on asthma drug use; • Data were collected on treatment outcome. PiCA is a growing consortium and new studies can join the consortium if they meet the inclusion criteria (www.pica-consortium.org).
Data collection: An online questionnaire (created using www.surveymonkey.com) was sent to the investigators of each cohort to collect information about the patients and design of the studies.
Characteristics of the studies and study populations Information was collected on the following characteristics of the studies: study design (i.e. asthma cohort, clinical trial and (high risk) birth cohort), country where the study was conducted and location of patient enrollment (type of health care center: primary, secondary or tertiary care). Per study, the following data were collected on the study populations: the age range (in yrs.), number of male asthmatics, and the number of
118 patients in distinct ethnic groups (i.e. Caucasians, African-Americans, Hispanics and Asians). In order to assess the potential of PiCA to perform pharmacogenomics studies, the numbers of patients with a reported use of asthma medication (ICS, SABA, LABA, Leukotriene modifiers (LTMs), Anti-IgE and Oral corticosteroids (OCS)) were collected per study. It was also ascertained whether data regarding environmental exposures and atopy were collected. The source for the DNA collection (i.e. blood, saliva) and availability of whole genome genotyping data were assessed.
Outcome measures and treatment response The presence of information on exacerbations, asthma symptoms and lung function was assessed for each study. A severe exacerbation was considered as a short course (3-5 days) OCS use or a hospitalization/emergency room (ER) visit according to the American Thoracic Society/European Respiratory Society (ATS/ERS) 2009 statement (20). The presence of information on unscheduled General Practitioner (GP) visits or asthma-related absences from school was also assessed. The two outcomes have been used as indicators of exacerbations in several pharmacogenomics studies. For asthma symptoms, presence of information on validated asthma symptom questionnaires (asthma control questionnaire (ACQ) or Asthma Control Test (ACT)) was assessed within the studies. The comparability of the results of these two questionnaires has been shown previously (21). Patients with ACQ scores ≥0.75 and ACT scores <20 were considered to have poor asthma control. In addition, availability of information on asthma symptoms based on guidelines (i.e. Global initiative for Asthma (GINA) and ATS/ERS) was also assessed. According to the availability of data in each study, the number of patients with exacerbations despite regular use of ICS was collected. For observational studies, the presence of any of these outcomes in the preceding 6 or 12 months was gathered. Asthma diagnosis is difficult in infants and pre-school children. Hence from birth cohorts within the PiCA consortium, we collected outcomes of children ≥ 6 years of age with physician-diagnosed asthma. Cohen’s Kappa statistic was calculated per study, to show the overlap between patients experiencing exacerbations and asthma symptoms (22). This was calculated for those studies in which both outcomes were available. The analysis was performed in R (Package ‘irr’) (23). Furthermore, since lung function measures are widely used as a response outcome in asthma, it was ascertained whether data regarding lung function measurements, especially changes in FEV1 from baseline over time (before and after treatment) and
119
changes in FEV1 after SABA use were also collected within the studies included in the consortium.
Results
Baseline characteristics of the studies and patients Currently, 21 asthma studies from 12 different countries are enrolled in the PiCA consortium. PiCA includes 15 asthma cohorts, three birth cohorts, two high-risk birth cohorts (inclusion of infants based on allergic history of the mother) and one clinical trial (Table 1). In total, PiCA includes data of 14,227 asthmatic patients up to 25 years of age. In 17 studies (80%), asthma was based on physician-diagnosis and/or hospital records. For three studies asthma diagnosis was based on parental-reported asthma diagnosis. PACMAN included children with a regular use of asthma medication. Analysis of PiCA children showed that 58% are male. From almost all patients within PiCA (97%) information was available on ethnic background. The majority of the asthmatic patients in PiCA are Caucasian (56%), 12% are Asian, 22% are Hispanic, and 8% have an African/African-American background and the remaining (2%) have mixed/other ethnic backgrounds (Figure 1). In the PiCA consortium studies, data on medication use was collected based on parental/patient reports (17 studies), pharmacy records (nine studies), and physician’s prescriptions (five studies). Medication data was available for 12,736 patients. Most of the patients in the studies were treated with ICS (n=7,619) and SABA (n=8,571). Furthermore, 2,050 patients received LABA, and 2,132 used LTM. OCS as a maintenance medication was used in 568 patients (Figure 2). In line with clinical asthma guidelines, most patients were treated with a combination of different asthma medications.
120
2% 8%
12%
56%
22%
Caucasians Hispanics Asians African-Americans mixed/other
Figure 1. Ethnic backgrounds of the asthmatic patients included in the PiCA consortium
Figure 2. Physician or patient/parental reported medication use in PiCA. *OCS considered as long-term therapy
121
Table 1. PiCA characteristics: study design and patient characteristics Study name Country Study Recruiting Asthmatic Age design centers patients (Range, (n) yrs.)
BAMSE Sweden General Primary care 420 0-16 birth cohort
BREATHE UK Asthma Primary and 1570 3-22 cohort secondary care
British Columbia Canada Asthma Tertiary/ 343 1-18 Childhood cohort quaternary Asthma Cohort referral center
CAMP USA RCT Tertiary care 1041 5-12
COPSAC2000 Denmark High risk Written invitation 43 0-7 birth cohort
COPSAC2010 Denmark General Written invitation 90 0-5 birth cohort
COPSACSevere* Denmark Asthma Registry based 1173 0-25 cohort
DUCHA Greece Asthma Tertiary care 193 5-14 cohort
ESTATe Netherlands Case-control Primary care 111 4-19
122
Table 1. PiCA characteristics
Male n, Mean (SD) FEV1% Medication DNA source (%) predicted baseline
ICS LABA SABA LTM
242 (57.6) 103 (11.0) 226 57 218 - Peripheral blood§
1017 (64) 96.6 (15.5) 959 62 1505 210 Saliva§
223 (65) - 343 54 343 79 Buccal cell and Saliva
621 (59) 95.6±18 311 - 418 - Peripheral blood§
22(51) 94.4 (12.1) 43 * 43 * Peripheral blood§
52(57) 97.1 (12.1) 90 0 90 * Peripheral blood§
791 (67) - * * * * Peripheral blood
179 (92) 101.2 (12.8) 193 56 18 25 Peripheral blood
67 (60) - 110 42 111 2 Saliva§
123
Table 1. PiCA characteristics Study name Country Study Recruiting Asthmatic Age design centers patients (Range, (n) yrs.)
follow Germany/ Asthma Secondary and 313 7-25 MAGICS Austria cohort tertiary care
GALA II# USA Case-control Secondary care, 2377 8-21 community and clinic-based recruitment
Generation Netherlands Population- Primary, secondary 399 fetal- R#2 based birth and tertiary care ongoing cohort
GOASC# Spain Asthma Secondary 125 2-18 cohort and tertiary care
PACMAN Netherlands Asthma Primary care 995 4-12 cohort
PAGES UK Asthma Primary, secondary 701 2-18 cohort and tertiary care
PASS UK Asthma Tertiary care 525 5-18 cohort
PIAMA Netherlands General Primary care 428 8 birth cohort/ high risk birth cohort
124
Table 1. PiCA characteristics Male n, (%) Mean (SD) FEV1% Medication DNA source predicted baseline ICS LABA SABA LTM
194 (62) - 150 104 107 27 Peripheral blood§
1288 (54) 90.8 (16.2) 1174 368 1900 610 Peripheral blood§
249 (62.4) 100 (12.8) 200 50 280 10 Umbilical cord blood§
76 (60) 94.6 (15.2) 125 78 14 107 Peripheral blood and Saliva
616 (61) - 844 229 819 87 Saliva§
519 (74) 94 (16) 648 347 696 286 Saliva
307 (58) - 525 395 525 369 Peripheral blood and Saliva§
254 (59.3) 105.4 (12.2) 208 28 210 5
125
Table 1. PiCA characteristics Study name Country Study Recruiting centers Asthmatic Age design patients (Range (n) yrs.)
SAGE II # USA Case-control Secondary care, 987 8-21 community and clinic-based recruitment
SCSGES# Singapore Asthma Tertiary care 1450 18-25 cohort
SLOVENIA Slovenia Asthma Tertiary care 350 5-19 cohort
Study of USA Case-control Tertiary care and 593 6-14 asthma in population based Puerto Rican probabilistic sampling children (HPR) design
Total: 12 countries 21 studies
126
Table 1. PiCA characteristics
Male n, (%) Mean (SD) FEV1% Medication DNA source predicted baseline ICS LABA SABA LTM
503 (51) 98.7 (14.1) 670 171 822 96 Peripheral blood and Saliva§
600 (41) 76.9 (12.8) 394* * * * *
162 (46) 89.9 (14.85) 193 * * 86 Peripheral blood
320 (53.9) 88.5 (16.5) 213 9 452 133 Peripheral blood§
- Data not available, * Data collection ongoing, # patient inclusion ongoing. §Studies with GWAS data available. #2 Patient follow-up ongoing, numbers based on participation until April 1st, 2015, aged 9 years. BAMSE, Swedish abbreviation for Children, Allergy, Milieu, Stockholm, Epidemiology. CAMP, Childhood Asthma Management Program. COPSAC, The Copenhagen Prospective Study on Asthma in Childhood. ESTATe, Effectiveness and Safety of Treatment with Asthma Therapy in children. GALA II, Genes-Environment and Admixture in Latino Americans. GOASC, Genetics of Asthma in Spanish Children. ICS, inhaled corticosteroids. LABA, Long-acting Beta2 agonist. LTM, Leukotriene Modifier. MAGICS, Multicenter Asthma Genetics in Childhood Study. PACMAN, Pharmacogenetics of Asthma Medication in Children: Medication with Anti- inflammatory effects. PAGES, Paediatric Asthma Gene Environment Study. PASS, Pharmacogenetics of adrenal suppression. PIAMA, The Prevention and Incidence of Asthma and Mite Allergy. SABA, Short-acting Beta2 agonists. SAGE II, Study of African Americans, Asthma, Genes Environments. SCSGES, Singapore Cross Sectional Genetic Epidemiology Study. RCT, randomized controlled trial
127
Outcome measures and treatment response Thirteen studies had information on exacerbations and approximately one third of the patients had severe exacerbations despite ICS treatment. In eleven studies (including 5,769 patients) data were available on OCS use as rescue medication despite ICS treatment. The prevalence of OCS use ranged between 7-67% in different studies, and in total 1,929 (33%) PiCA patients on ICS had received rescue OCS in the preceding 6-12 months of the study visit. Thirteen studies had data available on asthma-related ER visits or hospitalizations despite ICS (n=6,095). The prevalence of ER visits/hospitalizations ranged between 7-67%. In total 1,806 (29%) patients reported asthma-related ER visits or hospitalizations. Data on asthma-related school absences despite ICS use were available for 2,587 patients in six studies. Furthermore, data on unscheduled general practitioner (GP) visits were available for 1,479 patients in six studies (Figure 3). The total number of patients experiencing exacerbations in each study is shown in supplementary table 1. Validated scaled questionnaires to assess current asthma symptoms (ACQ and ACT) were used in five studies (DUCHA, ESTATE, PACMAN, PAGES and Singapore Cross Sectional Genetic Epidemiology Study) (in a total of 2,070 patients). In this population, 37% (n=766) of the patients had ACQ scores ≥ 0.75 or ACT scores <20 indicating poor asthma control. Furthermore, a modified version of the 1978 American Thoracic Society–Division of Lung Diseases Epidemiology Questionnaire (24) was used to assess current asthma control in GALA II and SAGE II in 1,725 patients; 41% had uncontrolled asthma symptoms based on this questionnaire. In addition to these scaled questionnaires, several other categorical measures of symptoms were used in studies. Modified GINA definition for long-term asthma control was used in BAMSE (n=226), with 34% of the patients having poor asthma symptoms. In the PIAMA study (n=110), 43% of the patients using inhaled steroids had uncontrolled asthma at age eight. Guidelines of the Dutch Pediatric Society (NVK), which follow the GINA guidelines, were used to define uncontrolled asthma (25). Regarding lung function measurements, changes in FEV1 after bronchodilator were measured in seven studies and changes in
FEV1 from baseline were measured in four studies. Information on asthma severity was available for 5,608 PiCA patients. The number of severe asthmatics according to ATS/ERS, GINA and British Thoracic Society/Scottish Intercollegiate Guidelines Network (BTS/SIGN) (step 4 or higher) guidelines was 838.
128
Figure 3. Exacerbations despite regular use of ICS in the preceiding six months or year. A) Percentage of OCS users as a rescue medication in 11 PiCA studies. B) Percentage of patients with ER visit/hospitalization in 13 PiCA studies. C) Percentage of patients with asthma-related school absences in 6 PiCA studies. D) Percentage of patients with unscheduled GP visits in 6 studies. BCCAC; British Columbia Childhood Asthma Cohort, Gen.R; Generation R, SCSGES; Singapore Cross Sectional Genetic Epidemiology
Study. In PASS and BREATHE exacerbation data were available in the preceding 6 months.
129
129
Overlap between exacerbations and asthma symptoms: In three studies (GALA II, PACMAN and SAGE II), we could assess the overlap between exacerbations (defined by OCS use) and patients with asthma symptoms (Figure 4). In all three patient populations, there was only a slight to fair agreement between these two outcomes (kappa: 0.03-0.21); 46-72% of the patients with reported OCS use as a rescue medication also had uncontrolled asthma symptoms according to the asthma questionnaire. The overlap between patients with ER visits/ hospitalizations in the past 6/12 months and uncontrolled asthma symptoms in four studies (BAMSE, GALA II, PACMAN, and SAGE II) was also poor (Kappa: 0.03 to 0.22); 41-55% of the patients with ER visits/hospitalizations had uncontrolled asthma symptoms. The overlap between OCS users and ER visits/hospitalizations in five different studies (BREATHE, GALA II, HPR, PACMAN and SAGE II) was fair to substantial (Kappa: 0.22 to 0.62). Kappa value for the asthma-related school absences and OCS use in two studies (BREATHE and HPR) was moderate to substantial (0.48- 0.64) and fair for the school absences and hospitalizations in the last six months of the same two studies (0.26 and 0.36). Kappa values for the different outcome definitions are shown in the supplementary table 2.
130
Figure 4. Overlap between different outcome measures from six studies within the PiCA consortium. The overlap between distinct definitions of exacerbations and uncontrolled asthma symptoms calculated in PACMAN, GALA II, SAGE II and BAMSE. For patients included in BAMSE uncontrolled asthma symptoms were assessed in the last 12 months. In PACMAN, GALA II and SAGE II, uncontrolled asthma symptoms at study visit was considered: 46% to 72% of the patients with OCS use as a rescue medication also had uncontrolled asthma symptoms. 41 to 55% of the patients with ER visits/hospitalizations had uncontrolled asthma symptoms. The overlap between OCS use and ER visits/hospitalizations in different studies varied between 19% and 55%. Furthermore, overlap between school absences and hospitalizations/OCS use was calculated in HPR and BREATHE. The overlap between hospitalizations and OCS use with school absences was 100% in HPR. In the BREATHE cohort, the overlap between OCS and school absences was 76% and the overlap between hospitalizations and school absences was 80%. Hosp; hospitalizations, OCS; oral corticosteroids, ER; emergency
Pharmacogenomic studies in PiCA:
DNA samples have been collected in 20 studies, and for one study the DNA collection is still ongoing. The source of DNA per study is shown in table 1. A protocol written by the research center interested in a specific research question will be sent to the Principal investigators (PI) of the consortium for review. Next, the
131
protocol will be sent to all PiCA studies. Centers that are willing to participate will perform the association analysis and the results will be sent to research center that initially initiated the research proposal. In case, individual study lack resources or expertise to perform the analyses, other PiCA collaborators will help to perform the analysis.
GWAS in PiCA:
Currently GWAS data is available for 13 studies (n= 6,743). In addition, 1,967 DNA samples from 5 studies will be genotyped; BAMSE (n=400), BREATHE (n=92), PAGES (n=514), GoShare (n=561) and SLOVENIA (n=400).
In the discovery phase of the GWAS, genotyped samples will be imputed with the Michigan imputation server (Available at: https://imputationserver.sph.umich.edu). After imputation and quality check association analysis will be performed with EPACS (efficient and parallelizable association container toolbox. Available at: http://genome.sph.umich.edu/wiki/EPACTS). Principal component analysis and adjustment for gender and age will be performed when necessary. GWAS meta- analysis will be performed by METASOFT (26). In the replication phase, association analysis will be performed for the top hits identified in the discovery phase.
Candidate gene approach in PiCA:
Candidate gene studies will be conducted for newly identified SNPs from GWAS meta- analyses and for previously identified SNPs in GWAS of childhood asthma onset and pharmacogenomics of asthma and SNPs that might associate with treatment response based on biological pathways.
Association analysis will be performed in the studies that have genotype or imputed data with high quality. The results of the association analysis will be meta-analyzed.
132
Discussion
The PiCA consortium is a unique initiative that brings together data from 14,227 asthmatic children/young adults from 12 different countries worldwide. In genetic association studies, replication of the results across populations with different ethnic backgrounds is of high importance in order to support the findings of the pharmacogenomics analysis (27). The PiCA consortium is a novel platform to study the pharmacogenomics of uncontrolled childhood asthma despite asthma treatment.
It is important to study pharmacogenomics of childhood asthma in addition to adult asthma, since asthma phenotypes differ between children and adults and findings in adult studies cannot be translated directly to the pediatric asthma population (28). For example, a genetic variant influencing FBXL7 expression has been found by the CAMP group to associate with improvement in asthma symptoms in response to ICS in two pediatric populations, but it failed to replicate in adults (13). Several GWAS of response to asthma medication have been published by the CAMP study group (29–31) and they can be found in the National Human Genome Research Institute (NHGRI) and the European Bioinformatics Institute [EMBL-EBI] GWAS catalog (32). In addition, variation in the ADRB2 gene has been associated with altered LABA response, but mainly in pediatric populations (33–36). Hence, it is important to study treatment response in asthmatic children. However, assessing treatment response in asthmatic children remains a challenging subject, as symptoms may vary over time. Different measures of uncontrolled asthma (i.e. exacerbations, symptoms, or lung function) might reflect distinct dimensions of the disease. It has been previously shown that demographic characteristics and biomarker profiles of children with severe exacerbations were different from children with persistent symptoms (15), and children without asthma symptoms can be prone to severe exacerbations (37). Furthermore, It has been shown that the definition of treatment response influences the genetic risk profile associated with drug response (38,29,39). Calculated Kappa values showed only minimal to moderate agreements between asthma symptoms and exacerbations. Since different dimensions of uncontrolled asthma include different patient populations and overlap only partly, distinct outcome measures need to be studied separately. An important strength of PiCA is the collection of well-defined asthma outcomes in > 14.000 individuals for future pharmacogenomics studies within the PiCA consortium, we will perform analyses using distinct measures of poor treatment response that reflect different dimensions of asthma.
133
Within the PiCA consortium, we included study designs such as observational asthma cohorts and (high risk) birth cohorts. An observational study (cohort or case-control) is a common approach to assess pharmacogenomics and should not be undervalued. Observational studies can provide valuable evidence for clinically relevant pharmaco- genomics markers. Once identified, the next step would be further replication and developing a prognostic biomarker test with additional replication for generalizability and investigating the functional biology to interrogate the mechanistic aspect of the replicated findings.
Major strengths of the design of the PiCA consortium are inclusion of patients from mild to severe asthmatics with thoroughly investigated outcome and phenotype data (i.e. exacerbations and asthma symptoms), and the coverage of the broad spectrum of pediatric asthmatic medication users, which will make it possible to assess the value of pharmacogenetics for subgroups of patients. Study heterogeneity makes it possible to assess the generalizability of findings across multiple designs and/or multiple ethnicities. Sensitivity analyses can be used to assess for which group a certain marker might have the highest clinical value.
In addition to large-scale pharmacogenomics studies, which are the main goal of this consortium, PiCA also has potential to study other factors influencing treatment outcomes, such as continued exposure to allergens or epigenomics. However, obtaining additional biological samples or data might be complicated for some PiCA studies; this might only be possible in part of the PiCA population. Several potential limitations of this consortium should be acknowledged. One of the limitations of PiCA could be population stratification. However, this heterogeneity will help us to identify different genetic markers associated with the treatment response in patients with different ethnicities. Furthermore, it will help us to discover pharmacogenomics markers that are associated with the treatment response in asthmatics regardless of the ethnic background of the patients. In genome-wide association analyses, we will adjust the results of each cohort by principal components when necessary. In candidate gene studies, the analyses will be performed separately for each study and the results will be meta-analyzed. Furthermore, we will also perform sensitivity analysis by conducing separate analysis for patients with different ethnic backgrounds. The results of these analyses will be compared and in the presence of a significant difference, they will be reported. Another limitation could be the wide age range of the patients included in PiCA, although this does reflect the general asthma population in
134 clinical practice, infant onset asthma might be a different phenotype from asthma in teenagers (40). In addition, asthma diagnosis is complicated at a young age, and infants and pre-school children can suffer from symptoms (such as wheezing) similar to those caused by asthma. In PiCA we will only include children who were still experiencing asthma symptoms at ≥ 6 years of age. In the majority of the PiCA studies (17 out of 21), asthma was based on physician-diagnosis and/or hospital records. Although criteria for physician-diagnosis might differ between countries, this difference reflects current clinical practice.
This is the first large effort to unite childhood asthma studies with a common interest in pharmacogenetics. Various studies within PiCA have collected detailed information on asthmatic children and followed children prospectively, making PiCA a unique platform for collaboration and validation. Several other studies (Asthma Genetics in Hungary (AGH), EUROPA from the Netherlands, GoShare from the UK and the Canadian asthma cohort) are still in the stage of recruiting patients, data and genotyping DNA samples, and will participate in the future projects of the PiCA consortium. In other fields, such as in cardiovascular pharmacogenomics, large research consortia have delivered key discoveries (41–44). PiCA is a growing consortium and it provides the opportunity to study pharmacogenetics on a large scale, paving the way for precision medicine in asthma.
135
References
1. Global Initiative for Asthma. Global 2007. J Allergy Clin Immunol. 2011;127(1): Strategy for Asthma Management and 145–52. Prevention, 2016: www.ginasthma.org. 11. Fleming L, Wilson N, Bush A. Difficult to 2. Masoli M, Fabian D, Holt S, Beasley R. control asthma in children. Curr Opin The global burden of asthma: executive Allergy Clin Immunol. 2007;7(2):190–5. summary of the GINA Dissemination Committee report. Allergy. 2004;59(5): 469 12. Pijnenburg MW, Szefler S. Per- –78. sonalized medicine in children with asthma. Paediatr Respir Rev. 2015;16 (2):101–7. 3. Szefler SJ, Martin RJ, King TS, Boushey HA, Cherniack RM, Chinchilli VM, et al. 13. Park H-W, Dahlin A, Tse S, Duan QL, Significant variability in response to inhaled Schuemann B, Martinez FD, et al. Genetic corticosteroids for persistent asthma. J predictors associated with improvement of Allergy Clin Immunol. 2002;109(3):410–8. asthma symptoms in response to inhaled corticosteroids. J Allergy Clin Immunol. 4. Szefler S.J., Phillips B.R., Martinez F.D, 2014;133(3):664–9.e5. Chinchilli V.M, Lemanske R.F., Strunk R.C, et al. Characterization of within-subject 14. Rogers AJ, Tantisira KG, Fuhlbrigge responses to fluticasone and montelukast AL, Litonjua AA, Lasky-Su JA, Szefler SJ, in childhood asthma. J Allergy Clin Immu- et al. Predictors of poor response during nol. 2005;115(2):233–42. asthma therapy differ with definition of outcome. Pharmacogenomics. 2009;10(8): 5. Malmstrom K, Rodriguez-Gomez G, 1231–42. Guerra J, Villaran C, Piñeiro A, Wei L.X, et al. Oral montelukast, inhaled beclo- 15. Wu AC, Tantisira K, Li L, Schuemann methasone, and placebo for chronic B, Weiss ST, Fuhlbrigge AL. Predictors of asthma. A randomized, controlled trial. symptoms are different from predictors of Montelukast /Beclomethasone Study severe exacerbations from asthma in Group. Ann Intern Med. 1999;130(6):487– children. Chest. 2011;140(1):100–7. 95. 16. Evans WE, Relling M V. Moving 6. Langmack EL, Martin RJ. Heterogeneity towards individualized medicine with phar- of response to asthma controller therapy: macogenomics. Nature. 2004;429 (6990): clinical implications. Curr Opin Pulm Med. 464–8. 2010;16(1):13–8. 17. Yip VLM, Hawcutt DB, Pirmohamed M. 7. Bahadori K, Doyle-Waters MM, Marra C, Pharmacogenetic Markers of Drug Efficacy Lynd L, Alasaly K, Swiston J, et al. Econ- and Toxicity. Clin Pharmacol Ther. 2015;98 omic burden of asthma: a systematic (1):61–70. review. BMC Pulm Med. 2009;9:24. 18. Custovic A, Ainsworth J, Arshad H, 8. Chapman KR, Boulet LP, Rea RM, Bishop C, Buchan I, Cullinan P, et al. The Franssen E. Suboptimal asthma control: Study Team for Early Life Asthma prevalence, detection and consequences in Research (STELAR) consortium “Asthma general practice. Eur Respir J. 2008;31(2): e-lab”: team science bringing data, 320–5. methods and investigators together. Thorax. 2015;70(8):799–801. 9. Williams AE, Lloyd AC, Watson L, Rabe KF. Cost of scheduled and unscheduled 19. Bacharier LB, Boner A, Carlsen K-H, asthma management in seven European Eigenmann PA, Frischer T, Götz M, et al. Union countries. Eur Respir Rev. 2006;15 Diagnosis and treatment of asthma in (98):4–9. childhood: a PRACTALL consensus report. Allergy. 2008;63(1):5-34. 10. Barnett SBL, Nurmagambetov TA. Costs of asthma in the United States: 2002- 20. Reddel HK, Taylor DR, Bateman ED,
136
Boulet L-P, Boushey HA, Busse WW, et al. 30. Park H-W, Dahlin A, Tse S, Duan Q., An official American Thoracic Society/ Schuemann B, Martinez FD, et al. Genetic European Respiratory Society statement: predictors associated with improvement of asthma control and exacerbations: stan- asthma symptoms in response to inhaled dardizing endpoints for clinical asthma trials corticosteroids. J Allergy Clin Immunol. and clinical practice. Am J Respir Crit Care 2014;133(3):664–669.e5. Med. 2009;180(1):59–99. 31. Tantisira KG, Damask A, Szefler SJ, 21. Khalili B, Boggs PB, Shi R, Bahna SL. Schuemann B, Markezich A, Su J, et al. Discrepancy between clinical asthma Genome-wide association identifies the T control assessment tools and fractional gene as a novel asthma pharmacogenetic exhaled nitric oxide. Ann Allergy Asthma locus. Am J Respir Crit Care Med. 2012; Immunol. 2008;101(2):124–9. 185(12):1286–91.
22. Viera AJ, Garrett JM. Understanding 32. Hindorff L, Parkinson H, Welter D, interobserver agreement: the kappa MacArthur J, Morales J, Burdett T, et al. statistic. Fam Med. 2005;37(5):360–3. The NHGRI GWAS Catalog, a curated resource of SNP-trait associations. Nucleic 23. R Development Core Team (2008). R: Acids Research, 2014, Vol. 42 (Database A language and environment for statistical issue): D1001-D1006. computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900- 33. Palmer CNA, Lipworth BJ, Lee S, Ismail 051-07-0, http://www.R-project.org. T, Macgregor DF, Mukhopadhyay S. Arginine-16 beta2 adrenoceptor genotype 24. Ferris BG. Epidemiology Standardiza- predisposes to exacerbations in young tion Project (American Thoracic Society). asthmatics taking regular salmeterol. Am Rev Respir Dis. 1978;118(6 Pt 2):1– Thorax. 2006;61(11):940–4. 120. 34. Zuurhout MJL, Vijverberg SJH, 25. de Jongste J, Vrijlandt EJLE. Astma Raaijmakers JAM, Koenderman L, Postma bijkinderen: samenvatting van de herziene DS, Koppelman GH, et al. Arg16 ADRB2 rich-tlijnen van de Sectie Kinderlongziekten genotype increases the risk of asthma vande NVK. [Guideline “Asthma in exacerbation in children with a reported use Children” for pediatric pulmonologists]. of long-acting β2-agonists: results of the PACMAN cohort. Pharmacogenomics. 26. Han B, Eskin E. Random-effects model 2013;14(16):1965–71. aimed at discovering associations in meta- analysis of genome wide association 35. Wechsler ME, Kunselman SJ, Chinchilli studies. Am J Hum Genet. 2011;88(5):586– VM, Bleecker E, Boushey HA, Calhoun WJ, 98. et al. Effect of beta2-adrenergic receptor polymorphism on response to longacting 27. Hall IP, Blakey JD. Genetic association beta2 agonist in asthma (LARGE trial): a studies in Thorax. Thorax. 2005;60(5):357– genotype-stratified, randomised, placebo- 9. controlled, crossover trial. Lancet (London, 28. Fleming L, Murray C, Bansal AT, England). 2009;374(9703):1754–64. Hashimoto S, Bisgaard H, Bush A, et al. 36. Turner S, Francis B, Vijverberg S, Pino- The burden of severe asthma in childhood Yanes M, Maitland-van der Zee AH, Basu and adolescence: results from the paed- K, et al. Childhood asthma exacerbations iatric U-BIOPRED cohorts. Eur Respir J. and the Arg16 β2-receptor polymorphism: 2015;46(5):1322–33. A meta-analysis stratified by treatment. J 29. Tantisira KG, Lasky-Su J, Harada M, Allergy Clin Immunol. 2016; 138(1):107- Murphy A, Litonjua AA, Himes BE, et al. 113.e5. Genomewide association between GLCCI1 37. Carroll CL, Schramm CM, Zucker AR. and response to glucocorticoid therapy in Severe exacerbations in children with mild asthma. N Engl J Med. 2011;365(13): 1173 –83.
137
asthma: characterizing a pediatric pheno- 42. Paternoster L, Standl M, Chen C.M, type. J Asthma. 2008;45(6):513–7. Ramasamy A, Bønnelykke K, Duijts L, et al. Meta-analysis of genome wide association 38. Leusink M, Vijverberg S.J.H., studies identifies three new risk loci for Koenderman L, Raaijmakers J.A.M, de JC, atopic dermatitis. Nat Genet. 2012;44 (2): Sterk PJ, et al. Genetic variation in un- 187–92. controlled childhood asthma despite ICS treatment. Pharmacogenomics J. 2015;1– 43. Moffatt M.F., Gut IG, Demenais F, 6. Strachan D.P., Bouzigon E, Heath S, et al. A large-scale, consortium based genome- 39. Vijverberg S.J.H, Tavendale R, Leusink wide association study of asthma. N Engl J M, Koenderman L, Raaijmakers J.A.M, Med. 2010;363(13): 1211–21. Postma D.S., et al. Pharmacogenetic analysis of GLCCI1 in three north European 44. Ferreira M.A.R., Matheson M.C., Duffy pediatric asthma populations with a report- DL, Marks GB, Hui J, Le Souëf P, et al. ed use of inhaled corticosteroids. Phar- Identification of IL6R and chromosome macogenomics. 2014;15(6):799–806. 11q13.5 as risk loci for asthma. Lancet (London,England). 2011;378(9795):1006– 40. Depner M, Fuchs O, Genuneit J, 14. Karvonen AM, Hyvärinen A, Kaulek V, et al. Clinical and Epidemiologic Phenotypes of Childhood Asthma. Am J Respir Crit Care Med. 2013;189(2):129-38.
41. Owen R.P., Altman R.B., Klein T.E. PharmGKB and the International Warfarin Pharmacogenetics Consortium: the chang- ing role for pharmacogenomic data-bases and single-drug pharmacogenetics. Hum Mutat. 2008; 29(4):456–60.
138
Online supplementary
Rationale and design of the multi-ethnic Pharmacogenomics in Childhood Asthma (PiCA) consortium
139
140
Supplementary Table 1. PICA characteristics: Treatment outcomes within the total population Study Asthma exacerbations Poor asthma symptoms
OCS use1 ER1 Hospitalizations1 School GP visits1 ACQ ACT n (%) n (%) n (%) absences1 n (%) n (%) n (%) n (%)
BAMSE - 32 (7.6) 4 (1.0) - - - -
PIAMA 2 (0.4) 3 (0.7) 2 (0.4) - 75 (17) - -
PAGES 668 (95) - 289 (41) - - - 64 (9)
PACMAN 60 (6) 61 (6.1) - - 61/953 (6.4) 406 (40) -
followMAGICS - 12 (4) 6 (2) 75 (24) 108 (35) - -
GALA II 745 (31) 1144 (48) 1195 (50) - - - -
SAGE II 187 (18) 335 (33) 48 (4) - - - -
GOASC 11 (8) 16 (12) 2 (1.6) - - - -
140
BREATHE 468 (29) - 299(19) 605 (38) - - -
Singapore Cross 38 (2.6) 120 (8) 34 (2.3) 170 (11) 309 (21) - 92 (6) Sectional Genetic Epidemiology Study
DUCHA ------56 (29)
British Columbia 232 (67) 214 (62) 67 (19) - 91 (26) - - Childhood Asthma Cohort
PASS 309 (58) - 141 (26) - - - -
CAMP 538 (51) 183 (17) 564 (54) 479 946) - -
HPR 229 (38) 279 (47) 98 (16) 355 (59) 467 (78) - -
ESTATE 39 (35.1) 13 (11.7) - - 37 (33) - 32 (29)
- Data not available, * data not analyzed. 1 number of children with this outcome during past 6 months (Italic font) or past 12 months of cohort studies (bold font), currently in the birth cohort (underlined), or during the trial (standard font). ACQ-score ≥ 0.75 and ACT–score ≤ 19 is considered not well controlled asthma. #2 patient follow-up ongoing, numbers based on participation until April 2015, aged 9 years.
141
141
142
Supplementary table 2. Cohen’s Kappa values for different definitions of uncontrolled asthma
Uncontrolled symptoms Uncontrolled OCS use & School absences School absences & & OCS use symptoms & ER visits and/or & hospitalizations OCS use ER visits and/or hospitalizations hospitalizations
Calculated values of Cohen’s Kappa
HPR 0.22 0.26 0.64
BREATHE 0.62 0.36 0.48
GALA II 0.21 0.19 0.32
PACMAN 0.03 0.03 0.29
SAGE II 0.16 0.2 0.35
BAMSE 0.22
142
143
144
CHAPTER 3.2
17q21 variant increases risk of exacerbations in asthmatic children despite using inhaled corticosteroids
Farzan N*, Vijverberg SJ*, Hernandez-Pacheco N, Bel E.H.D, Berce V, Bønnelykke K, Bisgaard H, Burchard E.G, Canino G, Celedón J.C, Chew F.T, Chiang WC, Cloutier MM, Forno E, Francis B, Hawcutt DB, Kabesch M, Karimi L, Melén E, Mukhopadhyay S, Merid S.K, Palmer CN, Pino-Yanes M, Pirmohamed M, Potočnik U, Repnik K, Schieck M, Sevelsted A, Sio YY, Smyth RL, Soares P, Söderhäll C, Tantisira KG, Tavendale R, Tse SM, Turner S, Verhamme K.M, Maitland-van der Zee AH
* Authors contributed equally to this manuscript.
Allergy, 2018;73(10):2083-2088
145
Abstract
Background: Genetic variants in the 17q21 locus are the strongest known genetic determinants for early onset childhood asthma. A limited number of studies including mainly non-Hispanic white, have investigated the effect of these variations on asthma treatment response.
Objective: In this study, we aimed to investigate the association between genetic variation in the 17q21 locus and asthma exacerbations despite inhaled corticosteroids (ICS) use in 14 studies of the multi-ethnic Pharmacogenomics in Childhood asthma (PiCA) consortium.
Methods: We conducted a trans-ethnic meta-analysis of 4,529 children and/young adults (non-Hispanic whites, Hispanics, African-Americans, East-Asians) with asthma and a reported use of ICS. The association between a single nucleotide polymorphism (SNP) in the 17q21 locus (rs7216389) and asthma exacerbations was assessed per study using multivariate logistic regression (adjusted for age, gender and treatment step). Two exacerbations measures were studied: asthma-related hospitalization/ emergency room (ER) visits (13 studies, n= 4,454 patients) and short-course of oral corticosteroid (OCS) use (11 studies, n=4,050 patients) in the past 6/12 months. Analyses were performed assuming an additive genetic model. Results were meta- analyzed with the inverse variance weighting method assuming random-effects.
Results: Variation in the 17q21 locus at rs7216389 was significantly associated with both outcomes. Odds Ratio (OR) per increase in risk allele (T) was 1.32 (95%CI: 1.17- 1.49, p<0.0001, I2=3.9%) for hospitalization/ER visits. Regarding the OCS use, OR per increase in risk allele was 1.19 (95%CI: 1.04-1.36, p=0.01, I2=22.8%).
Conclusions: Variation in the 17q21 locus contributes to an increased risk of asthma exacerbations despite the use of ICS in children/young adults.
146
Introduction
Asthma is the most common chronic disease in children and allergy-induced asthma is the most common type of asthma that affects most of the asthmatic children (1). According to the Global Initiative for Asthma guidelines (GINA), inhaled corticosteroids (ICS) are the preferred first-line treatment for persistent asthma (2). However, approximately 25% of the pediatric asthma patients suffer from uncontrolled asthma despite regular use of ICS (3). Continuous environmental and allergen exposures, poor medication adherence, severe disease and misdiagnosis, influence treatment response in asthmatic patients. Genetic variation might also be an important factor in the individual’s response to treatment (4–6). Genetic variations located in the 17q21 locus were found to be associated with asthma susceptibility in children in a large consortium-based meta-analysis (7) and this association has been replicated successfully in studies with different asthma sub- phenotypes within different ethnic groups (8–10). The most significant association with childhood onset asthma was reported for a single nucleotide polymorphism (SNP) at rs7216389 (7). This SNP was shown to be a risk factor for early onset childhood asthma (7), and childhood-onset severe asthma (11–14). The SNP affects the expression of multiple genes, including ORMDL3 (encoding orosomucoid 1-like 3), GSDMA (encoding gasdermin A), GSDMB (encoding gasdermin B) and CDK12 (also known as CRKRS, encoding cell division cycle 2-related protein kinase 7) (14). It has previously been shown that SNPs within the 17q21 locus might also influence the response to ICS in asthmatic children (15–17). However, the interaction between these SNPs and response to ICS has only been investigated in a limited number of studies, which exclusively included non-Hispanic white patients. To generalize the findings of the pharmacogenomics analysis, the SNP should be studied in populations with different ethnic backgrounds (18). The Pharmacogenomics in Childhood Asthma (PiCA) consortium is a multi-ethnic consortium that brings together data from ≥14,000 asthmatic children/young adults from 12 different countries to study the pharmaco- genomics of uncontrolled asthma despite treatment (19).
In this study, we aimed to assess the association between rs7216389 SNP in the 17q21 locus and asthma exacerbations despite ICS use in 14 distinct PiCA study populations with diverse ethnic backgrounds.
147
Methods Studies were eligible to participate in the current study if:
• Included patients were on British Thoracic Society (BTS) treatment step 2 or higher (20). • Data were available on exacerbations in the last 6 or 12 months of the study. • Genotype data of rs7216389 were available in their study population. From 22 studies currently participating in PiCA (19), 14 met the inclusion criteria and were included in the meta-analyses.
Study populations
European cohorts
BAMSE (Stockholm, Sweden) is a prospective birth cohort including children born between 1994 and 1996 in Stockholm. In the current study, Non-Hispanic white children at the age of 8 years with physician diagnosed and parental-reported asthma were included in the analysis (21). The ethnicity was defined based on the principle component analysis from genome-wide genotyping data. BREATHE is an observational study and included physician diagnosed asthmatic children/young adults (age: 3-22 years) (15), through primary and secondary health centers in either Tayside or Dumfries (Scotland, United Kingdom). The majority of the patients were considered to be non-Hispanic whites. COPSAC2000 is a prospective birth cohort including children (born between 1998 and 2001) of asthmatic mothers (22). For this study, self- reported non-Hispanic whites were included in the analyses. ESTATE (Effectiveness and Safety of Treatment with Asthma Therapy in children) is a case-control study including children and young adults with physician diagnosed asthma recruited from primary care units in the Netherlands (age: 4-19 years). Self-reported non-Hispanic whites were included in the analyses. FollowMAGICS (Germany, age: 7-25 years) is the follow-up study of the MAGICS cohort approximately 5 years after recruitment (Multicentre Asthma Genetics in Childhood Study), which is an observational study and includes physician diagnosed asthmatic children and young adults recruited from secondary and tertiary centers in Germany and Austria. The patients were self- reported non-Hispanic whites and their ethnicity was further validated through principle component analysis using genome-wide genotyping data. The Pharmacogenetics of
148
Asthma Medication in Children: Medication with Anti-inflammatory effects (PACMAN) study (the Netherlands, age: 4-12 years) (23), is an observational cohort study including children with a reported regular use of asthma medication through community pharmacies. In PACMAN, self-reported ethnicity was categorized as African, Asian, white, Hispanic, or mixed. In this study, non-Hispanic white patients were included in the analyses. Given that genome-wide genotyping data was available for PACMAN, principal component analyses was performed (Figure S1) to check whether the results could be affected by differences in the ancestral composition among individuals. The effect sizes of the association analyses in PACMAN with and without including three principal components as covariates, varied slightly (data not shown). Furthermore, since genome-wide analysis of ICS response did not result in genomic inflation due to population stratification (lambda value = 0.98) (Figure S2), and the model was not adjusted by ancestry. The Paediatric Asthma Gene Environment Study (PAGES) (Scotland, United Kingdom, age: 2-16 years) (24), is an observational cohort study including children/young adults with physician diagnosed asthma through primary, secondary and tertiary centers. For this study, patients with self-reported ancestry other than non-Hispanic white were excluded. Pharmacogenetics of adrenal suppression (PASS) (United Kingdom, age: 5-18 years) is a multicenter cohort of asthmatic children (25). Children were included in the study based on asthma diagnosis by a secondary care paediatrician experienced in the treatment of asthma, severe asthma that requires inhaled steroids and clinical concern about adrenal suppression sufficient to warrant a Low Dose Short Synacthen Test (LDSST). The ethnicity of two patients was self-reported mixture of non-Hispanic white and another ancestry, the remaining were self-reported non-Hispanic whites, all validated through principal component analysis from genome-wide genotyping data. The SLOVENIA cohort (Slovenia, age: 5-19) is an observational study and includes asthmatic children/young adults recruited from tertiary health centers. Asthma was defined by physician diagnosis and hospital records. Patients were self-identified Central European Caucasians of Slavic origin.
North American clinical trial
CAMP is a multi-center trial including 1,041 children (age: 5-12 years at the start of the trial) with mild-to-moderate asthma randomized to ICS (budesonide), nedocromil, or
149
placebo twice daily (26). In this study, patients with self-identified non-Hispanic white of European origin were analyzed.
Hispanic/Latino and African American cohorts
The Genes-Environment and Admixture in Latino Americans (GALA II) study (United States, age: 8-21 years) (27) is an ongoing multi-center study of Hispanic children/ young adults with and without asthma. Asthma was defined by physician diagnosis and report of symptoms in the two years preceding enrollment. Patients were self-identified Hispanics with four Hispanic grandparents and their ethnicity was validated with principal component analysis from genome-wide genotyping data. The Study of asthma in Puerto Rican children (HPR) (United States, age: 6-17 years) (28,29), is a case-control study including patients from tertiary care. Asthma was defined based on physician diagnosis and symptoms or medication use in the past 12 months. All participants had to have four Puerto Rican grandparents. The study of African Americans, Asthma, Genes Environments (SAGE II) study (United States, age: 8-21 years) (27), is an ongoing multi-center study of children and young adults with African American ancestry with and without asthma. Asthma was defined by physician diagnosis and report of symptoms in the two years preceding enrollment. Patients had four African-American grandparents and their ethnicity was validated with principal component analysis from genome-wide genotyping data.
East Asian cohort:
The Singapore Cross Sectional Genetic Epidemiology Study (SCSGES) (Singapore, age: 6-31 years) is an on-going cross-sectional genetic epidemiology study on allergic diseases among Singapore Chinese individuals (10). Self-identified Chinese ethnicity was further validated through principal component analysis from genome-wide genotyping data. Volunteers, including asthmatic and non-asthmatic individuals, were recruited from two different recruitment centers within Singapore. Asthma was defined by physician diagnosis and report of symptoms in the year preceding enrollment. In this study we included patients up to 25 years of age.
150
All studies participating in this study received ethical approval from their local medical ethics committees.
Medication data
Data collection on ICS use in the last 6/12 months of the study or first year of the trial:
BAMSE/COPSAC2000/followMAGICS/PACMAN:
Pharmacy records and parent/patient reported medication use at the study visit or based on completed study questionnaires.
BREATHE/GALAII/HPR/PAGES/SAGEII/SCSGES:
Parental/patient reported medication use at the study visit or based on completed study questionnaires.
ESTATE/PASS/SLOVENIA: Physician prescriptions and pharmacy records.
In CAMP patients received budesonide 200 μg twice daily.
Outcomes
The following outcomes were studied:
1) Any asthma-related hospitalizations/emergency room (ER) visits reported by the parent/child at the study visit or based on completed study questionnaires: ▪ BREATHE/PAGES/PASS: hospitalizations in the past 6 months preceding the study visit. ▪ PACMAN: ER visits in the past 12 months preceding the study visit. ▪ BAMSE/eSTATE/followMAGICS/GALAII/HPR/SAGEII/SCSGES/ SLOVENIA: ER visits/hospitalizations in the past 12 months preceding the study visit or filling in the questionnaire. ▪ CAMP: hospitalizations/ER visits in first 12 months of the trial.
2) Any course of oral corticosteroid use (OCS) reported by the parent or child at the study visit or based on completed study questionnaires:
151
▪ BREATHE/PAGES/ PASS: in the past 6 months preceding the study visit. ▪ ESTATE/GALAII/HPR/PACMAN/SAGEII/SCSGES: in the past 12 months preceding the study visit. ▪ CAMP: in the first 12 months of the trial ▪ COPSAC2000: in the past 6 months preceding the study visit. In COPSAC2000, OCS and acute treatment with high-dose inhaled steroids (4 times normal dose) were used interchangeably for treatment of severe asthma exacerbations.
Genotyping
Genotyping in BAMSE, BREATHE, PACMAN and PAGES was performed using the Sequenom Mass Array platform (Sequenom, San Diego, California, USA). Genotyping was performed using Axiom LAT1 array World Array 4 (Affymetrix, Santa Clara, CA, USA) in GALA II and SAGE II. In COPSAC2000 genotyping was performed using Illumina Infinium HumanOmniExpressExome chip. Genotyping in CAMP was performed using Illumina Infinium II 550 K SNP Chips and 610 Quad Chip (Illumina, Inc, San Diego, California) and rs7216389 was imputed based on 1000 Genomes. Genotyping in eSTATE was performed using Human Core-24 BeadChip Marker information and the SNP was imputed from the Haplotype Reference Consortium. For followMAGICS the Illumina Sentrix HumanHap300 BeadChip used for genotyping. In HPR genotyping was performed using Illumina HumanOmni2.5 BeadChip. Genotyping in PACMAN was performed using Human Core-24 BeadChip Marker information and the SNP was imputed from the Haplotype Reference Consortium. Furthermore, a subset of the PACMAN samples was previously genotyped using the Sequenom Mass Array platform (Sequenom, San Diego, California, USA). For this study, the genotype data from these two methods were pooled together for the analyses. Genotyping in PASS was performed using Illumina Omni Express 8v1 and the SNP was imputed with ShapeIT and Impute2. The imputation quality was estimated by “INFO” score ( ≥0.5 indicating high imputation quality). Genotyping in SCSGES was performed using 2- staged method, with each of discovery and validation stages focuses on separate sets of human participants. In discovery stage of genotyping, Illumina Bead Xpress Assay was performed, while validation stage used Sequenom platform with MassARRAY and iPLEX technology. For SLOVENIA, real time PCR instrument LightCycler 480 (Roche) used for genotyping.
152
Statistical analysis
Logistic regression analysis was used to assess the risk of exacerbations when carrying rs7216389. Odds ratios (OR) were calculated per study. The model was adjusted for age (in years), gender and for an adjusted form of the BTS guideline treatment steps (step 2: ICS plus short-acting beta agonist as needed, step 3: step 2 plus long-acting beta agonist, step 4: step 3 plus leukotriene antagonist). The ORs were meta-analyzed with the inverse variance weighting method assuming random- effects, this was due to consideration of potential heterogeneity between cohorts and their sampling. Two statistical tests of heterogeneity, I2 and Cochran’s Q-test, are reported along with other results of the meta-analysis (30). Forest plots were made with R (version 3.3.3) and the ‘metafor’ package (31). An additive model was assumed for the analysis. In addition, separate meta-analyses were performed for European studies and Hispanics. Furthermore, since younger children might have a different phenotype of the disease and a different response to ICS, a sensitivity analysis was performed to investigate whether the association would be different by categorizing the patients based on their age at the time of the outcome measurement (2-4 years of age and ≥5 years). Deviation from Hardy-Weinberg equilibrium (HWE) in the studies with genome-wide genotyping data was assessed with standard quality control for GWAS. For the remaining studies (BAMSE, BREATHE, PAGES, SLOVENIA and SCSGES), we used a web program (http://www.oege.org/software/hwe-mr-calc.shtml) which uses Pearson chi-squared test for HWE testing (32). The results are reported according to the STREGA guidelines.
Results
Study characteristics Age, gender, genotype data and exacerbation data were available for 4,529 steroid- treated children and young adults. The characteristics of the study populations are listed in Table 1. The mean age of the patients ranged between 3.3 (1.0) years for COPSAC2000 and 17.13 (3.03) years for followMAGICS. Data on BTS treatment steps was available for eight studies and in five of them more than 50% of the patients were on step 2. The risk allele (T) frequency was highest in East-Asians (n= 182, T = 0.81), followed with African-Americans (T=0.79, n=468) and Hispanics (T=0.66, total n=916)
153
and it was less frequent in patients with European ancestry (ranged between 0.54- 0.62, total n = 2,963). The SNP was in HWE for all cohorts in the analyses.
154
Table 1. Characteristics of the study populations.
BAMSEa BREATHE CAMPa COPSAC2000a,b eSTATE followMAGICSa GALA IIa
(n=122) (n=808) (n=172) (n=54) (n=95) (n=150) (n=744)
Patient Characteristic:
Age (yrs), mean (SD) 8.37 (0.41) 9.8 (4.0) 8.8 (2.1) 3.3 (1.0) 10.8 (4.3) 17.13 (3.03) 12.1 (3.2)
Male gender, % (n) 79 (96) 60.8 (491) 55.2 (95) 54 (29) 57.9 (55) 59.3 (89) 56.8 (423)
Asthma exacerbations in past
year:
Asthma-related ER visit/hospital 14.7 (18) 19.0 (154) 13.4 (23) - 10.5 (10) 8.6 (13) 58.3 (434)
admission, % (n)a
Oral corticosteroid use, % (n) - 31.7 (256) 47.1 (81) 11.1b (6) 36.8 (35) - 42.3 (315)
BTS treatment step:
2, % - 65.6 c - 60 28.7 41.1
3, % - 18.4 - - 37.9 60.7 43.6
4, % - 16.0 - - 2.1 10.6 15.3
Rs7216389, T-allele frequency 0.57 0.56 0.58 0.62 0.60 0.58 0.66
155
155
156
a a a a a
HPR PACMAN PAGES PASS SAGE II SCSGES SLOVENIA
(n=172) (n=665) (n=308) (n=390) (n=468) (182) (n=199)
Patient Characteristic:
Age (yrs), mean (SD) 9.8 (2.7) 8.7 (2.3) 9.2 (3.8) 11.1 (4.0) 13.5 (3.4) 13.35 (5.09) 10.9 (3.41)
Male gender, % (n) 49.4 (85) 61.1 (406) 56.8 (175) 55.9 (218) 54.1 (253) 67.6 (123) 54.8 (109)
Asthma exacerbations in past
year:
Asthma-related ER visit/hospital 58.1 (100) 6.0(39/644) 15.6 (48) 76 (296) 44.7 (209) 20.3 (37) 35.6 (71)
admission, % (n)a
Oral corticosteroid use, % (n) 77.3 (133) 6.5 (43) 43.2 (133) 52 (203) 29.3 (137) 20.3 (37) -
BTS treatment step:
2, % 60.7 72 25.6 - 68.6 - -
3, % 36.4 22.3 61 - 25 - -
4, % 2.9 5.7 13.3 - 6.4 - -
Rs7216389, T-allele frequency 0.66 0.59 0.58 0.60 0.79 0.81 0.54
–, data not available; BTS, British Thoracic Society; ER, emergency room; SD, standard deviation; y, years. b Patients with exacerbations were treated with oral corticosteroid or high‐dose inhaled corticosteroids. c all children were on 200 μg of budesonide (ICS) plus SABA as needed.
156
17q21 and risk of asthma related hospitalizations/ER visits:
In 13 studies (n= 4,454), data were available on asthma related hospitalizations and/or ER visits. In total 1,378 (30%) patients reported hospitalizations/ER visits in the 6 or 12 months prior to the study visit or during the first year of the trial. Rs7216389 was statistically significantly associated with asthma-related ER visits/hospitalizations, (summary OR per increase in risk allele: 1.32, 95%CI: 1.17- 1.49, p<0.0001, I2=3.9%) (Fig 1). The SNP was also associated with the outcome when performing the analyses separately for European (n= 2,888, OR:1.33, 95%CI:1.10-1.61, p=0.004, I2=30.2%) and Hispanic (n= 916, OR: 1.31, 95%CI:1.06-1.63, p=0.01, I2=0.00%) populations.
Figure 1. Forest Plot of the association between rs7216389 and asthma-related hospitalizations/ER visits in thirteen studies. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals with rs7216389, controlling for age, gender and BTS treatment step.
157
17q21 and risk of OCS use: Data on OCS use/high dose ICS as a rescue medication for asthma exacerbations were available for 4,050 children/young adults. Among these patients, 1,269 (31%) had OCS use/high dose ICS in the last 6 or 12 months of the study or first year of the trial. In the meta-analysis of the nine studies the rs7216389-T was statistically significantly associated with OCS use/ high dose ICS (summary OR per increase in variant allele:1.19, 95%CI: 1.04-1.36, p=0.01, I2=22.8%) (Fig 2). Rs7216389 was associated with the outcome in the meta-analysis of seven European studies (n=2,492, OR:1.26, 95%CI:1.09-1.45, p=0.002, I2=6.2%) but not in Hispanics (n= 916, OR: 0.96, 95%CI:0.76-1.22, p=0.7, I2=0.00%).
Figure 2. Forest Plot of the association between rs7216389 and OCS/high dose ICS use in eleven studies. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals with rs7216389, controlling for age, gender and BTS treatment step.
158
17q21 and risk of asthma exacerbations stratified by age Sensitivity analysis was performed to assess the effect of the SNP on treatment response in preschool children (2-4 years of age) and ≥5 years of age. When excluding children <5 years of age in the meta-analysis, the results remained significant. Regarding asthma-related hospitalization/ER visits, data of patients ≥5 years of age were available in 13 studies (n= 4,254) and the SNP was associated with the outcome (summary OR: 1.32, 95%CI: 1.18-1.49, p<0.0001, I2 =22.8%) (figure S3). Regarding OCS use, 10 studies collected data on patients ≥5 years of age (n= 3,771). The meta- analysis of 10 studies rs7216389 was associated with the OCS use (summary OR: 1.20, 95%CI: 1.05-1.38, p=0.01, I2 = 21.7%) (figure S4). We also performed a meta-analysis for the studies that had sufficient data available on preschool children (2-4 years of age). Although, the effect estimates in younger children were in the same direction for both OCS/high dose ICS use (n= 223, summary OR: 1.11, 95%CI: 0.60-2.04, p=0.7, I2=28.9%) and hospitalizations/ER visits (n=169, summary OR: 1.44, 95%CI:0.83-2.49, p=0.1, I2=28.9%), the results were not statistically significant. All preschool studies included European children (figure S5 & figure S6).
Discussion In this large trans-ethnic meta-analysis, we found rs7216389 in the 17q21 locus to be associated with higher rates of short-term OCS use and asthma-related hospitalization/ ER visits in children/young adults treated with ICS. The SNP was also correlated with higher risks of exacerbations in children ≥5 years of age but not in younger children. In subgroup analysis, the effect estimates for hospitalizations/ER visits were the same for both non-Hispanic whites and Hispanics. There was a low to moderated heterogeneity between studies. Large differences were seen among different studies regarding the exacerbation rates. Exacerbation rates ranged between 6.5% (PACMAN) and 77.2% (HPR) for OCS use and 6% (PACMAN) and 58% (GALA II and HPR) for hospitalizations/ER visits. Recruitment of the patients in different health care settings (e.g. primary/secondary/tertiary care, community pharmacies) might be the reason of differences in exacerbation rates. Moreover, cultural differences such as differences in the willingness of the physicians to prescribe OCS, socioeconomic status or healthcare utilization by patients in different countries might be another reason of the differences between exacerbation rates among studies (33,34). Due to the low number of OCS
159
users in COPSAC2000 (n=54), two weeks of high-dose ICS (4 times normal dose= 1600 mcg budesonide/day) was also considered as rescue medication. In COPSAC2000, OCS/two weeks of high-dose ICS were used interchangeably for treatment of exacerbations in preschool children. Despite these differences, the effect estimates of the different studies were mostly in the same direction.
The effect of the genetic variants in the 17q21 locus on response to asthma treatment has been reported previously in studies with European children, however these studies did not restrict analyses to ICS users (15) or used different outcomes for treatment response (16-17). An association between rs7216389 and the risk of exacerbations in patients on different BTS treatment steps (from step 0 to 4) was initially identified in the BREATHE cohort including 1054 asthmatic children/young adults (age: 3-22 years) (15). In this study, an increased risk for exacerbation for each copy of the T allele (1.30 (95% CI, 1.07-1.59; P =0.008)) was reported (15). However, this analysis was not restricted to steroid-treated patients. In another study including 300 children from the SLOVENIA cohort (16), rs2872507, a functional SNP in high linkage disequilibrium with rs7216389, was found to be significantly associated with improvements in FEV1% predicted in atopic children treated with ICS.
Rs7216389 was found to be the most significantly associated SNP with childhood onset asthma in the discovery GWAS study by Moffat et al. (7) and has subsequently been replicated by others (8–10). It remains unclear which SNP or gene in the 17q21 locus might play a causal role in disease onset, severity and treatment response. Different genes located in this locus (i.e. ORMDL3, GSDMA, GSDMB, ZPBP2, IKZF3) were found to be associated with asthma development (7,8,12,17,35–38). ORMDL3, seems to be involved in sphingolipid biosynthesis and metabolism (39). Altered metabolism of sphingolipids by ORMDL3 has been proposed to be one of the key pathways of asthma susceptibility (40,41). Studies on mice have shown that ORMDL3 is mainly expressed in airway epithelial cells that can play a role in allergic inflammation, remodeling, sensitivity, proliferation and contractility of bronchial smooth muscle and antiviral responses by regulating the expression of various metallo- proteases, chemokines and oligoadenylate synthetases antiviral genes (42–44). On a molecular level, ORMDL3 has been found to be involved in the regulation of eosinophil trafficking, recruitment and degranulation (45). In more recent studies, correlations between ORMDL3 expression and T cell cytokines levels (i.e. IL-2, IL-4 and IL-13) have been reported (46,47). Since genetic variants in the 17q21 locus also regulate
160 other gene transcripts than ORMDL3, such as GSDMA and GSDMB, other mechanisms may underlie asthma pathogenesis (48). The function of the genes in the GSDM family remains largely unknown, though in a recent study, asthma variants in GSDMB have been shown to alter the pyroptosis activity of GSDMB in epithelial cells which restrains the release of inflammatory cytokine (49).
Limitations of the study include the use of retrospective reporting of exacerbations in the observational cohort studies. However, the effect was also observed in a clinical trial population (CAMP), where exacerbations were reported prospectively with the use of diaries. Hence, we do not believe that using retrospective data has significantly influenced the results. Asthma-related hospitalizations/ER visits and OCS use are both proxies for exacerbations (50). Since not all studies had data available on both outcomes, we did not combine the two outcomes in our analysis. Furthermore, since information on treatment adherence was not available in all included studies, patients’ compliance to treatment was not considered in the analysis.
In summary, we show that 17q21, a widely replicated asthma susceptibility locus, is also associated with an increased risk of exacerbations in children/young adults treated with ICS. This means that children who carry the risk allele of rs7216389 could be less responsive to ICS. Rs7216389 seems to increase bronchial responsiveness and therefore exacerbation rates in children (12), suggesting that carriers of rs7216389 might have a more severe form of asthma. One possible option to resolve whether this locus is an asthma severity locus independent of ICS therapy or a pharmacogenomics locus, would be to perform the analyses in ICS-free subgroup. However, due to the small sample sizes and low frequencies of exacerbations in ICS-free groups, we could not perform the analyses. However, we used medication use (BTS treatment steps) in our analyses as a marker to control for disease severity. Although the molecular mechanisms underlying exacerbation-prone phenotype of pediatric asthma need to be elucidated, the incorporation of genetic information in clinical algorithms could eventually facilitate diagnosis of asthma phenotypes and guide treatment.
161
References
1. Fuhlbrigge A.L., Adams R.J., Guilbert T.W, J Respir Crit Care Med.2008;177(11):1194– Grant E, Lozano P, Janson SL et al. The 200. burden of asthma in the United States: level and distribution are dependent on inter- 9. Moffatt MF, Gut IG, Demenais F, Strachan pretation of the national asthma education DP, Bouzigon E, Heath S et al. A large-scale, and prevention program guidelines. Am J consortium-based genomewide association Respir Crit Care Med. 2002; 166(8):1044–9. study of asthma. N Engl J Med. 2010;363(13): 1211–21. 2. From the Global Strategy for Asthma Management and Prevention, Global In- 10. Andiappan AK, Sio YY, Lee B, Suri BK, itiative for Asthma (GINA) 2015. Global Matta SA, Lum J et al. Functional variants of Initiative for Asthma (GINA). 17q12-21 are associated with allergic asthma but not allergic rhinitis. J Allergy Clin Immunol. 2016;137(3):758–766.e3. 3. Chung KF, Godard P, Adelroth E, Ayres J, Barnes N, Barnes P et al. Difficult/therapy- 11. Halapi E, Gudbjartsson D.F., Jonsdottir GM, resistant asthma: the need for an integrated Bjornsdottir U.S., Thorleifsson G, Helgadottir H approach to define clinical phenotypes, et al. A sequence variant on 17q21 is associat- evaluate risk factors, understand patho- ed with age at onset and severity of asthma. Eur physiology and find novel therapies. ERS J Hum Genet. 2010;18(8):902–8. Task Force on Difficult/Therapy-Resistant Asthma. Eur Respir J. 1999;13(5):1198– 208. 12. Bisgaard H, Bønnelykke K, Sleiman PMA, Brasholt M, Chawes B, Kreiner-Møller E et al. Chromosome 17q21 gene variants are as- 4. Drazen J.M., Silverman E.K., Lee T.H. sociated with asthma and exacerbations but not Heterogeneity of therapeutic responses in atopy in early childhood. Am J Respir Crit Care asthma. Br Med Bull. 2000;56(4):1054–70. Med. 2009;179(3):179–85.
13. Binia A, Khorasani N, Bhavsar PK, Adcock 5. Lima JJ, Blake K V, Tantisira KG, Weiss I, Brightling CE, Chung KF et al. Chromosome ST. Pharmacogenetics of asthma. Curr Opin 17q21 SNP and severe asthma. J Hum Genet. Pulm Med. 2009;15(1):57–62. 2011;56(1):97–8.
6. Farzan N, Vijverberg S.J.H, Arets H.G., 14. Hao K, Bossé Y, Nickle DC, Paré PD, Raaijmakers J.A.M, Maitland-van der Zee Postma DS, Laviolette M et al. Lung eQTLs to AH. Pharmacogenomics of inhaled cortico- help reveal the molecular underpinnings of steroids and leukotriene modifiers: A sys- asthma. PLoS Genet. 2012;8(11):e1003029. tematic review. Clin Exp Allergy J. 2017;47 (2):271–93. 15. Tavendale R, Macgregor D.F., Mukhopadhyay S, Palmer C.N.A. A poly- 7. Moffatt MF, Kabesch M, Liang L, Dixon AL morphism controlling ORMDL3 expression is et al. Genetic variants regulating ORMDL3 associated with asthma that is poorly expression contribute to the risk of childhood controlled by current medications. J Allergy asthma. Nature. 2007;448(7152):470–3. Clin Immunol. 2008;121(4):860–3.
8. Galanter J, Choudhry S, Eng C, Nazario 16. Berce V, Kozmus C.E.P, Potočnik U. S, Rodríguez-Santana J.R., Casal J et al. Association among ORMDL3 gene express- ORMDL3 gene is associated with asthma in ion, 17q21 polymorphism and response to three ethnically diverse populations. Am treatment with inhaled corticosteroids in children with asthma. Pharmacogenomics J. 2013;13(6):523–9.
162
17. Leusink M, Vijverberg SJH, Koenderman 26. The Childhood Asthma Management L, Raaijmakers J.A.M, de Jongste J.C., Sterk Program (CAMP): design, rationale, and P.J. et al. Genetic variation in uncontrolled methods. Childhood Asthma Management childhood asthma despite ICS treatment. Program Research Group. Control Clin Pharmacogenomics J. 2015;1–6. Trials. 1999;20(1):91–120.
18. Hall I.P., Blakey J.D. Genetic association 27. Nishimura KK, Galanter JM, Roth LA, Oh studies in Thorax. Thorax. BMJ Publishing SS, Thakur N, Nguyen EA et al. Early-life air Group Ltd and British Thoracic Society; 2005; pollution and asthma risk in minority children. 60(5):357–9. The GALA II and SAGE II studies. Am J Respir Crit Care Med. 2013;188(3):309–18. 19. Farzan N, Vijverberg SJ, Andiappan AK, Arianto L, Berce V, Blanca-López N et al. 28. Forno E, Cloutier MM, Datta S, Paul K, Rationale and design of the multiethnic Sylvia J, Calvert D et al. Mouse allergen, lung Pharmacogenomics in Childhood Asthma function, and atopy in Puerto Rican children. consortium. Pharmacogenomics. 2017;18 PLoS One. 2012;7(7):e40383. (10):931–43. 29. Brehm JM, Man Tse S, Croteau-Chonka 20. British Thoracic Society (BTS). BTS/ DC, Forno E, Litonjua AA, Raby BA et al. A SIGN British guideline on the management of Genome-Wide Association Study of Post- asthma 2016: https://www.brit-thoracic.org. bronchodilator Lung Function in Children with uk/document-library/clinical-information/ Asthma. Am J Respir Crit Care Med. 2015; asthma/btssign-asthma-guideline-2016. 192(5):634–7.
21. Ballardini N, Kull I, Lind T, Hallner E, 30. Higgins JPT, Thompson SG, Deeks JJ, Almqvist C, Ostblom E et al. Development Altman DG. Measuring inconsistency in and comorbidity of eczema, asthma and meta-analyses. BMJ. 2003;327(7414):557– rhinitis to age 12: data from the BAMSE birth 60. cohort. Allergy. 2012;67(4):537–44. 31. Softw WV.JS, 2010 U. Conducting meta- 22. Bisgaard H. The Copenhagen Pros- analyses in R with the metafor package. pective Study on Asthma in Childhood brieger.esalq.usp.br. (COPSAC): design, rationale, and baseline data from a longitudinal birth cohort study. 32. Rodriguez S, Gaunt TR, Day INM. Hardy- Ann Allergy Asthma Immunol. 2004;93(4): Weinberg equilibrium testing of biological 381–9. ascertainment for Mendelian randomization studies. Am J Epidemiol. Oxford University 23. Koster E.S., Raaijmakers J.A, Koppelman Press; 2009;169(4):505–14. G.H., Postma D.S., van der Ent C.K., Koenderman L et al. Pharmacogenetics of 33. Wahlström R, Hummers-Pradier E, anti-inflammatory treatment in children with Lundborg CS, Muskova M, Lagerløv P, Denig asthma: rationale and design of the PACMAN P et al. Variations in asthma treatment in five cohort. Pharmacogenomics. 2009;10(8): European countries--judgement analysis of 1351–61. case simulations. Fam Pract. 2002;19(5): 452–60. 24. Turner SW, Ayres JG, Macfarlane T V, Mehta A, Mehta G, Palmer CN et al. A 34. Szentpetery S.E., Forno E, Canino G, methodology to establish a database to study Celedón J.C. Asthma in Puerto Ricans: gene environment interactions for childhood Lessons from a high-risk population. J Allergy asthma. BMC Med Res Methodol. 2010; Clin Immunol. 2016;138(6):1556–8. 10:107. 35. Bouzigon E, Corda E, Aschard H, Dizier 25. Hawcutt DB, Jorgensen AL, Wallin N, M-H, Boland A, Bousquet J et al. Effect of Thompson B, Peak M, Lacy D et al. Adrenal 17q21 Variants and Smoking Exposure in responses to a low-dose short synacthen test Early-Onset Asthma. N Engl J Med. 2008;359 in children with asthma. Clin Endocrinol (Oxf). (19):1985–94. 2015;82(5):648–56.
163
36. Hirota T, Harada M, Sakashita M, Doi 44. Miller M, Tam AB, Cho JY, Doherty TA, S, Miyatake A, Fujita K et al. Genetic poly- Pham A, Khorram N et al. ORMDL3 is an morphism regulating ORM1-like 3 (Sac- inducible lung epithelial gene regulating charomyces cerevisiae) expression is metalloproteases, chemokines, OAS, and associated with childhood atopic asthma in ATF6. Proc Natl Acad Sci. 2012;109(41): a Japanese population. J Allergy Clin 16648–53. Immunol. 2008;121(3):769–70. 45. Ha S.G., Ge X.N., Bahaie NS, Kang BN, 37. Sleiman P.M.A, Annaiah K, Imielinski Rao A, Rao SP, et al. ORMDL3 promotes M, Bradfield JP, Kim CE, Frackelton EC et eosinophil trafficking and activation via al. ORMDL3 variants associated with regulation of integrins and CD48. Nat asthma susceptibility in North Americans of Commun. 2013;4:2479. European ancestry. J Allergy Clin Immunol. 2008;122(6):1225–7. 46. Schedel M, Michel S, Gaertner VD, Toncheva AA, Depner M, Binia A et al. 38. Wu H, Romieu I, Sienra-Monge J-J, Li Polymorphisms related to ORMDL3 are H, del Rio-Navarro BE, London SJ. Genetic associated with asthma susceptibility, variation in ORM1-like 3 (ORMDL3) and alterations in transcriptional regulation of gasdermin-like (GSDML) and childhood ORMDL3, and changes in TH2 cytokine asthma. Allergy. 2009;64(4):629–35. levels. J Allergy Clin Immunol. 2015;136 (4):893–903.e14. 39. Breslow D.K., Collins S.R., Bodenmiller B, Aebersold R, Simons K, Shevchenko A 47. Schmiedel BJ, Seumois G, Samaniego- et al. Orm family proteins mediate sphingo- Castruita D, Cayford J, Schulten V, Chavez lipid homeostasis. Nature. 2010; 463 L et al. 17q21 asthma-risk variants switch (7284):1048–53. CTCF binding and regulate IL-2 production by T cells. Nat Commun. 2016;7:13426. 40. Miller M, Rosenthal P, Beppu A, Gordillo R, Broide DH. Oroscomucoid like 48. Das S, Miller M, Beppu A.K., Mueller J, protein 3 (ORMDL3) transgenic mice have McGeough M.D., Vuong C et al. GSDMB reduced levels of sphingolipids including induces an asthma phenotype charac- sphingosine-1-phosphate and ceramide. J terized by increased airway respon- Allergy Clin Immunol. 2017;139(4):1373– siveness and remodeling without lung 1376.e4. inflammation. Proc Natl Acad Sci. 2016;113 (46):13132–7. 41. Paulenda T, Draber P. The role of ORMDL proteins, guardians of cellular 49. Panganiban RA, Sun M, Dahlin A, Park sphingolipids, in asthma. Allergy. 2016;71 H-R, Kan M, Himes BE et al. A functional (7):918–30. splice variant associated with decreased asthma risk abolishes the ability of 42. Chen J, Miller M, Unno H, Rosenthal P, gasdermin B to induce epithelial cell Sanderson MJ, Broide DH. Orosomucoid- pyroptosis. J Allergy Clin Immunol. 2018; like 3 (ORMDL3) upregulates airway pii: S0091. smooth muscle proliferation, contraction, and Ca2+ oscillations in asthma. J Allergy 50. Reddel HK, Taylor DR, Bateman ED, Clin Immunol. 2017;pii: S0091. Boulet L-P, Boushey HA, Busse WW et al. An official American Thoracic Society/ 43. Miller M, Rosenthal P, Beppu A, Mueller European Respiratory Society statement: JL, Hoffman HM, Tam AB et al. ORMDL3 asthma control and exacerbations: stan- Transgenic Mice Have Increased Airway dardizing endpoints for clinical asthma Remodeling and Airway Responsiveness trials and clinical practice. Am J Respir Crit Characteristic of Asthma. J Immunol. 2014; Care Med. 2009;180(1):59–99. 192(8):3475–87.
164
Online supplementary
17q21 variant increases risk of exacerbations in asthmatic children despite inhaled corticosteroids use
165
Figure S1. Principal components for PACMAN individuals classified by being dutch/non-dutch. A) PC1 versus PC2; B) PC2 versus PC3.
Figure S2. Quantile-quantile plot of GWAS of ICS response not adjusted by principal components.
166
Figure S3. Forest Plot of the association between rs7216389 and asthma-related hospitalizations/ER visits in patients 5 years and older. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals with rs7216389, controlling for age, gender, and BTS treatment step.
Figure S4. Forest Plot of the association between rs7216389 and OCS use in patients 5 years and older. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals with rs7216389, controlling for age, gender, and BTS treatment step.
167
Figure S5. Forest Plot of the association between rs7216389 and asthma-related hospitalizations/ER visits in patients younger than 5. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals carrying the risk allele (T), controlling for age, gender, and BTS treatment steps.
Figure S6. Forest Plot of the association between rs7216389 and OCS/high dose ICS in patients younger than 5. Odds Ratios (OR) and corresponding 95% Confidence Intervals (95% CI) for individuals carrying the risk allele (T), controlling for age, gender, and BTS treatment steps.
168
169
170
CHAPTER 3.3
Genome-wide association study of inhaled corticosteroid response in African-admixed children with asthma
Hernandez-Pacheco N, Farzan N, Francis B, Karimi L, Repnik K, Vijverberg S.J, Soares P, Schieck M, Gorenjak M, Forno E, Eng C, Oh SS, Pérez- Méndez L, Berce V, Tavendale R, Samedy L.A, Hunstman S, Hu D, Meade K, Farber H.J, Avila P.C, Serebrisky D, Thyne S.M, Brigino-Buenaventura E, Rodriguez-Cintron W, Sen S, Kumar R, Lenoir M, Borrell L.N, Rodriguez- Santana J.R, Celedón J.C, Mukhopadhyay S, Potočnik U, Pirmohamed M, Verhamme KM, Kabesch M, Palmer CN, Hawcutt D.B, Flores C, Maitland-van der Zee A.H, Burchard EG, Pino-Yanes M
Accepted for publication in Clinical and Experimental Allergy (CEA)
171
Abstract
Background: Inhaled glucocorticosteroids (ICS) are the most widely prescribed and effective medication to control asthma symptoms and exacerbations. However, this treatment does not prevent asthma exacerbations in many children, particularly in African-admixed individuals. Although the importance of the genetic component in ICS response has been demonstrated, a few genome-wide association studies (GWAS) have been performed in European and Asian populations.
Objective: We aimed to identify genetic loci associated with asthma exacerbations in African-admixed children taking ICS and to validate previous associations identified in different populations.
Methods: A meta-analysis of two GWAS of asthma exacerbations was performed in 1,347 African-admixed children, analyzing 8.7 million genetic variants. Those with p≤5x10-6 were followed up for replication in 1,699 asthmatic patients from six European cohorts. Associations of ICS response described in published GWAS were followed up for replication in the African-admixed cohorts. Results: A total of 15 independent variants were suggestively associated with asthma exacerbations in African-admixed populations (p≤5x10-6). One of them, located in the intergenic region of APOBEC3B and APOBEC3C, showed evidence of replication in Europeans (rs5995653, p = 0.01) and was also associated with change in lung function after treatment with ICS (p = 7.54x10-4). Additionally, a previously reported association of the L3MBTL4-ARHGAP28 locus was confirmed in Hispanics/Latinos and African Americans, although a different variant was identified.
Conclusion: This study revealed the association of APOBEC3B and APOBEC3C as a novel locus with asthma exacerbations in children despite the use of ICS and validated known loci.
172
Introduction
Asthma is considered the most common chronic condition in children and young adults (1). A significant global burden has been attributed to this disorder, which is driven by economic costs on health care systems and substantial productivity loss related to high work and/or school absenteeism rates (2) associated with the occurrence of severe exacerbations (3,4). Attempts to reduce the burden of asthma should move towards better management of asthma.
Inhaled glucocorticosteroids (ICS) are the most effective and commonly prescribed medication to control symptoms and prevent severe exacerbations in asthma patients (2). Albeit most of the children taking ICS experience a notable decrease in asthma symptoms, it has been found that this medication does not prevent the occurrence of asthma exacerbations in approximately 30-40% of the children, and 10-15% of non- responders to ICS could even suffer severe exacerbations (5-10). High variability in ICS response has not only been described among individuals, but also among populations (11). In addition to the high asthma morbidity, exacerbations rates and mortality (12-14), African-admixed populations have been shown to have low ICS response due to poor effectiveness or adverse effects (15). These strong racial differences have suggested an important contribution of the genetic component in the response to treatment with ICS (16). In fact, heritability estimates have evidenced that approximately 40-60% of the variation in ICS response may be due to genetic factors (17-20).
During the last several decades, pharmacogenomic studies have mainly used candidate-gene approaches (21-24), which have only evaluated a small portion of the genomic variation. More recently, asthma pharmacogenetic studies of ICS response have evolved towards hypothesis-free approaches by implementing genome-wide association studies (GWAS) (23,25). Eight GWAS of ICS response have been performed to date (26-36), revealing the association of only 14 loci with this trait. Despite the large advantages of GWAS compared to candidate-gene association studies, the genes identified to date represent a small proportion of the total estimated variability of ICS response and still have low accuracy in predicting an individual’s responsiveness to this asthma treatment (23,37,38). The design of the GWAS performed to date may be the main reason, where analyses are statistically underpowered to detect genetic associations (25). In fact, most GWAS of ICS response have included a relatively small number of individuals (N<1,000) of primarily
173
European and, to a lesser extent, Asian ancestry, with poor representation of African- admixed populations (15-39), which include Hispanics/Latinos and African Americans. However, the increased asthma prevalence among African-admixed individuals and the greater genetic diversity and specific genetic architecture of these populations present a unique opportunity to study the response to asthma treatment with ICS (11, 15, 40-43).
We hypothesized that a large pharmacogenomic study of ICS response in African- admixed individuals with asthma that exhaustively explores the association of genetic variants across the whole genome could reveal novel genes associated with this trait. We also attempted to evaluate whether the associations described in GWAS performed in European and Asian populations are relevant for African-admixed populations.
Methods
Study Populations
A total of eight independent studies participating in the Pharmacogenomics in Childhood of Asthma (PiCA) consortium (44) were analyzed as part of discovery and replication phases of this meta-GWAS. Individuals from two African-admixed cohorts were included in the discovery phase: the Genes-environments & Admixture in Latino Americans Study (GALA II) and the Study of African Americans, Asthma, Genes and Environments (SAGE). Samples from six European PiCA cohorts were used for replication. All the studies have been approved by their local institutional review boards and all participants/parents provided written assent/consent, respectively (45- 54).
African-admixed populations analyzed in the discovery phase
Patients from the GALA II and SAGE studies with a physician diagnosis of asthma who reported having active symptoms and asthma medication use within the last 2 years were analyzed in the discovery phase. These are two independent studies focused on two different racial/ethnic groups based on the self-identified ethnicity of the four grandparents of each subject: Hispanics/Latinos (GALA II) and African
174
Americans (SAGE). Both studies recruited unrelated children and young adults, aged 8 to 21 years old, using the same protocol and questionnaires from different areas in the United States (51, 55).
Analyses were performed for a subset of 854 subjects from GALA II and 493 individuals from SAGE for whom complete data were available. Specifically, we assessed self-reported ICS use, genome-wide genotyping data (52,54), and information regarding presence/absence of severe asthma exacerbations, as defined by the European Respiratory Society (ERS) and the American Thoracic Society (ATS) (56). We examined exacerbations that occurred during the 12 months preceding the study enrollment (need to seek emergency asthma care, hospitalizations or the administration of oral corticosteroids).
European populations analyzed in the replication phase
Validation was carried out in European individuals from six independent studies participating in the PiCA consortium: the follow-up stage of the Multicenter Asthma Genetics in Childhood Study (followMAGICS); the Pharmacogenetics of Adrenal Suppression study (PASS); Pharmacogenetics of Asthma Medication in Children: Medication with Anti-inflammatory effects (PACMAN); Effectiveness and Safety of Treatment with Asthma Therapy in Children (ESTATe); BREATHE and SLOVENIA studies. Detailed methods about each validation cohort are described in the Supplementary Material.
The use of ICS and availability of data related to the presence/absence of asthma exacerbations during the previous 12 or 6 months were also applied as inclusion criteria in the validation cohorts, whereas non-availability of data related to ICS use, asthma exacerbations, age, gender and genotype data were considered as exclusion criteria. For those studies without data related to the events included in the ATS/ERS definition of asthma exacerbations, information regarding school absences, unscheduled general practitioner (GP) or respiratory system specialist visits was also considered.
175
Genome-wide genotyping and imputation in the discovery and replication studies
Both GALA II and SAGE samples were genotyped using the Axiom® LAT1 array (Affymetrix Inc.), and quality control (QC) procedures were performed as described elsewhere (52,54). Genotyping of the subjects included in the replication phase was performed on different genotyping platforms, as described in previous publications (45-50,53) (see Table E1 in the Online Repository). In addition, four of the studies were genotyped for the purposes of the PiCA consortium and their QC is described in the Online Repository Text.
In all the studies, imputation was carried out by means of the Michigan Imputation Server (57) using the second release of the Haplotype Reference Consortium (HRC) (r1.1 2016) as reference panel (58). Haplotype reconstruction and imputation were performed with SHAPEIT (59) and Minimac3 (60), respectively.
Association testing and meta-analysis in the discovery phase
GWA analyses were carried out separately in GALA II and SAGE. Logistic regressions were used to evaluate the association of genetic variants with ICS response in GALA II and SAGE by means of the binary Wald test implemented in the software EPACTS 3.2.6 (61). The presence or absence of any asthma exacerbations during the last 12 or 6 months in patients treated with ICS was considered as a measure of ICS response, which was evaluated as a binary outcome variable. Age, gender and the first two principal components (PCs), obtained with EIGENSOFT (62), were included as covariates in the regression models.
Single nucleotide polymorphisms (SNPs) with a minor allele frequency (MAF)≥1% and also with high imputation quality score (Rsq≥0.3) in GALA II and SAGE that were shared by both populations were meta-analyzed using METASOFT (63). Fixed-effects or random-effects models were selected for each variant depending on absence or presence of heterogeneity, respectively, which was assessed by means of the Cochran Q-test. A suggestive genome-wide threshold of p-value≤5x10-6 was applied to select variants that were suggestively associated with asthma exacerbations in African-admixed populations. Among those variants, independent associations were detected by means of logistic regression analyses conditioned on the most significant SNP of each loci, which were performed using R 3.4.3 (64). This analysis provided a list of independent variants that were followed up for replication.
176
Association testing and meta-analysis in the replication phase
Statistical analyses were performed following the same methodology as in the discovery phase, with the exception of the definition of asthma exacerbations that was available in each study and the number of PCs included as covariates in the association analyses (Table S1). Evidence of replication was considered for those SNPs that showed a combined p-value ≤ 0.05 in a meta-analysis of all the European studies and consistent directions of effects in both discovery and validation cohorts.
Association with ICS response measured as change in FEV1
SNPs significantly associated with asthma exacerbations in both African-admixed and European populations, were evaluated for association with the change in the forced expiratory volume in 1 second (FEV1) after 6 weeks of treatment with ICS in 184 ICS users from the SLOVENIA study, the only cohort with this outcome measured (65).
This variable was dichotomized with a threshold of ≥12% FEV1 change to define responders and non-responders to ICS treatment, and it was used in logistic regression models that included age, gender, and the first two PCs as covariates.
Validation of previous associations in African-admixed populations
Since previous GWAS of ICS response have focused on European and Asian populations (26-31,35-36), we attempted to validate their results in African-admixed populations. A total of 25 SNPs near or within 14 loci declared as associated with ICS response (26-30,35) were followed up for replication in the GALA II and SAGE studies. Replication was attempted at the SNP and gene levels, the latter considering variants located within 100 kilobases (kb) upstream or downstream from the genes. Evidence of replication was considered for SNPs nominally associated with ICS response (p≤0.05) that had the same direction of the effect as the published GWAS. For gene- level replication, a Bonferroni-like correction was applied to account for the number of independents variants tested within each locus, as estimated with empirical autocorrelations assessed with the R package coda (64). According to this, a Bonferroni-corrected significance threshold was estimated for each locus as: α=0.05/ number of independent variants.
177
Functional evaluation of variants associated with ICS response
Functional annotation and search of evidence of significant expression quantitative trait loci (eQTL) for the associated variants SNPs and those in high linkage disequilibrium (LD) (r2>0.9) were performed using HaploReg v4.1 (66), based on data provided by the Encyclopedia of DNA Elements (ENCODE) project (67). Gene expression was inspected using the Portal for the Genotype-Tissue Expression (GTEx) (68) and the Gene Expression Atlas (69). Moreover, evidence of association with enhancers was searched using GeneCards: The Human Gene Database (70).
Results
Characteristics of the study populations
The characteristics of the 1,347 African-admixed asthmatic children and young adults from the GALA II and SAGE studies analyzed in the discovery phase and the 1,699 Europeans subjects included in the replication are shown in Table 1 and Table E1, respectively. Hispanics/Latinos reported a higher proportion of asthma exacerbations in the 12 months preceding study enrollment (66.4%) than African Americans (51.9%). Although asthma exacerbations were differentially defined in the validation cohorts, similar proportions were found across the discovery and replication cohorts, with the exception of PACMAN and SLOVENIA, with values of 11.0 % and 37.5%, respectively (Table S1).
178
Table 1. Clinical and demographic characteristics of the African-admixed populations analyzed in the discovery phase.
GALA II SAGE (n = 854) (n = 493)
Gender (% male) 57.3 54.2
Mean age ± SD (years) 12.1 ± 3.2 13.5 ± 3.4
Ethnicity Hispanic/Latino African American
Mean genetic ancestry (%)
African 13.6 79.4 European 51.5 20.6 Native-American 34.9 NA Asthma exacerbations in the last 12 66.4 51.9 months (%) Emergency asthma care (%) 56.6 43.2
OCS use (%) 40.2 29.4
Hospitalizations (%) 12.6 5.7
SD: standard deviation; OCS: systemic corticosteroids; NA: not available.
Discovery phase in African-admixed populations
The meta-analysis of the GALA II and SAGE GWAS included 8.7 million genotyped and imputed SNPs that were shared among Hispanics/Latinos and African Americans and had MAF≥1% and Rsq≥0.3. The Q-Q plot of the association results after combining both African-admixed cohorts did not reveal major genomic inflation due to population stratification (λGC = 1.04, see Figure S1 in the Online Repository). Although the genome- wide significant threshold (p-value ≤ 5x10-8) was not reached by any of the variants in the discovery phase, 27 SNPs were suggestively associated with asthma exacerbations despite the use of ICS (p-value ≤ 5x10-6) in African-admixed children and young adults (Figure 1).
179
Figure 1. Manhattan plot of association results of ICS response in African-admixed populations.
Association results are represented as -log10 p-value on the y-axis along the chromosomes (x-axis). The suggestive significance threshold for replication is indicated by the black line (p ≤ 5x10-6).
After performing pairwise regression models conditioned on the most significant variant for each gene region with at least two suggestive associations, one independent variant was detected per locus, except for APOBEC3B-APOBEC3C and ANKRD30B, where two SNPs remained significant after conditioning on each gene’s most significant variant (see Table S2 in the Online Repository). As a result, 15 SNPs were identified as independently associated with ICS response in African-admixed populations (Table S3) and were followed up for replication.
Validation phase in European populations
Of the 15 SNPs selected for replication in Europeans, 11 SNPs had a MAF≥1% in Europeans and were forwarded for replication. Of those, the SNP rs5995653, located within the intergenic region of APOBEC3B and APOBEC3C (Figure 2), showed evidence of nominal replication after combining the European cohorts. The direction of effect for this SNP was the same in Europeans (OR for A allele: 0.77, 95% CI: 0.63- 0.94, p = 0.010) as in the African-admixed samples from discovery (OR for A allele = 0.66, 95% CI: 0.56-0.79, p = 4.80 x 10-6) (Table 2). A meta-analysis of this SNP across the two African-admixed discovery and six European replication cohorts, resulted in a suggestive genome-wide significant association (OR for A allele = 0.71, 95% CI: 0.62- 0.81, p = 2.66 x 10-7, Figure 3).
180
Figure 2. Regional plot of association results in the discovery phase for the APOBEC3B- APOBEC3C intergenic region, which represents a novel locus of ICS response. Statistical significance of association results (-log10 p-value) (y-axis) is represented by chromosome position (x-axis) for each SNP as a dot. A diamond represents the independent association signal with evidence of replication in Europeans (rs5995653) and the remaining SNPs are color-coded based on their LD with this SNP, indicated by pairwise r2 values for American populations from the 1KGP.
181
182
Table 2. Association results for the suggestive SNPs followed up for replication in European populations.
African-admixed populations European populations
(n = 1,347) (n = 1,699)
SNP Chr.a Position Nearest gene(s) A1/A2 OR (95% CI) b p-value OR (95% CI) b p-value
rs11121611 1 6367219 ACOT7 G/T 0.55 (0.43-0.70) 1.65 x 10-6 0.96 (0.61-1.51) 0.248 c
rs35514893 6 15909525 DTNBP1-MYLIP T/C 0.36 (0.23-0.55) 2.86 x 10-6 0.73 (0.22-2.45) 0.611
rs4897302 6 123886231 TRDN T/C 1.58 (1.31-1.91) 1.75 x 10-6 0.98 (0.84-1.16) 0.831
rs61585310 7 104006510 LHFPL3 G/T 0.61 (0.49-0.75) 2.85 x 10-6 0.90 (0.74-1.11) 0.329
rs7851998 9 126828514 LHX2 A/G 0.56 (0.44-0.72) 3.97 x 10-6 0.88 (0.70-1.12) 0.291
rs2125362 11 86167136 ME3 A/G 1.31 (0.68-2.56) 3.53 x 10-6 c 0.95 (0.80-1.13) 0.569
rs450789 13 33578233 KL G/A 0.64 (0.53-0.77) 3.33 x 10-6 0.95 (0.81-1.12) 0.561
rs12959468 18 15182381 ANKRD30B A/G 0.39 (0.26-0.58) 2.99 x 10-6 1.40 (0.75-2.60) 0.289
rs2278992 19 18095769 KCNN1 C/T 0.59 (0.47-0.74) 3.76 x 10-6 1.02 (0.82-1.26) 0.870
rs6001366 22 39399941 APOBEC3B-APOBEC3C T/C 0.47 (0.35-0.65) 2.53 x 10-6 0.99 (0.72-1.36) 0.940
rs5995653 22 39404249 APOBEC3B-APOBEC3C A/G 0.66 (0.56-0.79) 4.80 x 10-6 0.77 (0.63-0.94) 0.010 aChromosome; b OR: Odds ratio for the effect alleles (additive model); c Random-effect model was applied since heterogeneity was found between African- admixed/European populations. A1: Effect allele; A2: Non-effect allele; CI: Confidence Interval.
182
Figure 3. Forest plot of association effect of rs5995653 in African-admixed and European populations. Odds ratio (OR) for the effect allele (A) is shown for each analyzed cohort and after combining them by black boxes and a blue diamond. Gray lines indicate the corresponding 95% Confidence Intervals (95% CI) for each individual cohort
Association of rs5995653 with ICS response measured as change in FEV1
The SNP rs5995653 was significantly associated with a positive response to the treatment of ICS in SLOVENIA, measured as an increase of FEV1 (OR for A allele = 2.59, 95% CI: 1.49-4.51, p = 7.54x10-4), which is concordant with the protective effect of this SNP with asthma exacerbations in both discovery and validation cohorts.
In silico functional role of the novel association detected
The experimental data provided by the ENCODE project shows that the SNP rs5995653 is located within an enhancer histone mark and a DNAse hypersensitivity site in blood cells. This is concordant with the fact that APOBEC3B has been associated with enhancers that regulate multiple transcription factor binding sites, indicating its involvement in the regulation of gene expression in different cell types, including lung fibroblasts. Moreover, this variant is also in high LD with several SNPs associated with the expression of the genes APOBEC3A (rs9607601: p=1.80x10-13 and rs5995654: p=9.10x10-14), APOBEC3G (rs9607601: p=0.003), and CBX6 (rs9607601: p=3.94x10-4 and rs5995654: p=4.00x10-4), acting as eQTL in blood cells
183
(65-66,70-73). Moreover, previous functional studies have evidenced high levels of gene expression of both APOBEC3B and APOBEC3C in pulmonary cells (GTEx) (68,69).
Validation of previous associations of ICS response
None of the 25 SNPs previously associated with ICS response in published GWAS was consistently associated with asthma exacerbations in GALA II and SAGE studies (see Table S4 in the Online Repository). To assess whether the lack of replication of previous GWAS hits could be due to the association of alternative genetic variants among different populations, gene-level replication was performed. A total of 36,261 variants located within 100 kb upstream and downstream from 14 loci previously associated with ICS response were evaluated, which represented 2,916 independent variants. After applying a Bonferroni-like correction for the number of variants analyzed within each locus, suggestive associations were observed for nine SNPs near three loci: ALLC (min p-value = 4.69 x 10-4 for the SNP rs113903375), L3MBTL4-ARHGAP28 (min p-value = 1.57 x 10-5 for the SNP rs62081416), and ELMO2-ZNF334 (min p-value = 3.56x10-4 for the SNP rs2425845) (Table S5). However, applying a more restrictive correction for the total number of variants across all loci (p ≤ 1.71x10-5 for 2,916 independent variants tested), only the association of the SNP rs62081416, located within the intergenic region of L3MBTL4 and ARHGAP28, was significantly associated with ICS response in African-admixed individuals (OR for A allele = 2.44, 95% CI: 1.63- 3.65, p = 1.57x10-5).
Discussion
In this study, we carried out the first GWAS of ICS response in Hispanic/Latino and African American children and young adults with asthma performed to date. After combining the association results from these two African-admixed populations, 15 independent suggestive association signals were associated with asthma exacerbations despite the use of ICS, and one of them showed evidence of nominal replication in European populations. This SNP was also significantly associated with
ICS response defined as an increase in FEV1 after 6 weeks of treatment with ICS in one of the European cohorts where this outcome was measured. These results revealed the APOBEC3B and APOBEC3C genes as novel loci of ICS response in asthmatic children and young adults. Additionally, we validated the association of the
184
L3MBTL4-ARHGAP28 locus in African-admixed populations, which was previously described in a GWAS of ICS response in European descent individuals.
The APOBEC3B and APOBEC3C genes encode two members of the apolipoprotein B mRNA-editing catalytic polypeptide 3 (APOBEC3) family. APOBEC3 proteins are involved in RNA editing through the deamination of cytidine to uracil (75-79). Their main function is related to the innate immunity to endogenous retroelements and exogenous viruses and are considered important factors against viral and retroviral infections (75-76,80). In fact, APOBEC3B has been proposed as a restriction factor against a broad range of viruses (32,81-88). However, APOBEC3 proteins are also involved in a number of cellular processes related to mutagenic activity (72,89-96), which has been suggested to play a key role in several types of cancer (93-94). In fact, APOBEC3B has been associated with an increased risk of lung cancer (97-98).
We found that the A allele of rs5995653, located 5.8 kb from the 3’UTR of APOBEC3C, showed a protective effect against asthma exacerbations and was associated with improvement on FEV1 in patients treated with ICS. While no asthma-related functions have been attributed to any of the APOBEC3 flanking genes, evidence of high levels of RNA expression has been found in pulmonary fibroblasts for both genes (68-69). Furthermore, the functional evidence found for rs5995653 suggests that this SNP plays a key role in regulating the expression of genes involved in several cellular processes in the lung. Interestingly, respiratory viral infections are important risk factors for exacerbations in asthmatic children (99-101). This fact is concordant with the consistent function of APOBEC3B and APOBEC3C as restrictors of viral infections, suggesting that the expression of these genes in pulmonary tissues could be involved in fighting against viral infections and the risk of asthma exacerbations despite the patients are treated with ICS.
Interestingly, the protective effect of the rs5995653 A allele was stronger in African- admixed populations than in Europeans. This is concordant with the data provided by the 1000 Genome Project (1KGP) Phase III data (102), showing that the effect allele of rs5995653 (A) is more frequent in populations with African ancestry (MAF of 55% in African populations, 48% in African Americans and 34% in Puerto Ricans) compared to European populations (ranging from 26 to 31%) (Figure 4). These differences could explain why these genes had not been detected in previous GWAS of ICS response, which had been mainly focused on asthma patients of Asian and European ancestry (26-31,35-36). This fact highlights the scientific importance and benefits of studying
185
African-admixed populations, whose inclusion improves the statistical power to detect novel associations with complex traits like ICS response (11,15,43,103).
Figure 4. Allele frequency map for rs5995653, a novel protective variant against asthma exacerbations in response to ICS in African-admixed and European populations. Frequency proportions for the effect (A) and non-effect (G) alleles are represented in blue and yellow, respectively. Modified from the Geography of Genetic Variants Browser (http://www.popgen.uchicago.edu/ggv)106.
Our study has several strengths that should be highlighted. First, this is the largest meta-GWAS of ICS response with a discovery phase specifically focused on Hispanic/Latino and African- American asthma patients published to date, which are the minority ethnic groups most affected by asthma in the United States (5,12,15,104- 106). This is especially relevant given that African-admixed populations have been underrepresented in the asthma pharmacogenomic studies of ICS response (15,107). Second, we demonstrated the existence of genes involved in ICS response shared among different populations by identifying a novel locus shared among African- admixed and European populations. These results suggest that the locus identified could be also influential in other populations. Third, we validated the association of three loci previously described in GWAS of ICS response in European and Asian populations (30-31), even though one of them was associated with a different measurement of ICS response from asthma exacerbations (improvement in FEV1 after treatment with ICS in adults) (30). This evidence reinforces the validity of asthma exacerbations as a good measure of response to the asthma treatment with ICS. Additionally, the fact that this gene has been previously identified in adults could suggest the existence of common genetic markers of ICS response among adulthood
186 and childhood asthma. Finally, this is the first meta-GWAS of response to asthma treatment that used the extensive catalogue of variants from the whole genome provided by the HRC reference panel (58).
Despite the strengths of our study, there are also some limitations that should be considered. First, the most significant variant associated with ICS response in African- admixed and European populations did not reach genome-wide significance. However, the results found suggested that this variant could have an important implication in preventing asthma exacerbations in response to treatment with ICS, since the result was replicated in independent samples. Second, asthma exacerbations were differentially defined in the European validation cohorts included in the replication phase of this meta-GWAS. Nevertheless, this outcome was homogeneously defined in the cohorts included in the discovery phase of this meta-GWAS, suggesting that the identified locus is robustly associated with asthma exacerbation across a range of definitions. Third, a quantitative measurement of ICS response was only available in one of the European cohorts. Additional studies should seek to validate the association signal when using change in FEV1 after the treatment with ICS as the response variable. Finally, functional evidence relating the intergenic region of APOBEC3B and APOBEC3C with ICS response in asthma patients was not directly assessed in this current study. Therefore, in vitro experiments are needed to evaluate the functional roles of these loci in order confirm their implication in the response to the asthma treatment with ICS.
In summary, our meta-GWAS in African-admixed children and young adults identified the intergenic region of APOBEC3B and APOBEC3C as a novel locus for ICS asthma treatment response. We also validated the association of one locus previously in a previous GWAS of ICS response. Our study demonstrates the advantages of including African-admixed populations in asthma pharmacogenomic studies of ICS response.
187
References
1. 10 Facts on Asthma. World Health et al. Lower bronchodilator responsiveness Organization;2011. Available at: http:// in Puerto Rican than in Mexican subjects www.who.int/features/factfiles/asthma/en/. with asthma. Am J Respir Crit Care Med 2004; 169:386-92. 2.Global strategy for asthma management 13. Centers for Disease Control and and prevention. Global Initiative for Prevention. CDC health disparities and Asthma (GINA); 2017. Available at: inequalities report. United States. Morbidity http://ginasthma.org/. and Mortality Weekly Report 2013; 62 (suppl 3. Levy ML, Winter R. Asthma deaths: 3):85-179. what now? Thorax. 2015;70(3):209-10 14. Akinbami LJ, Moorman JE, Simon AE, 4. Puranik S, Forno E, Bush A, Celedon Schoendorf KC. Trends in racial disparities JC. Predicting Severe Asthma Exacer- for asthma outcomes among children 0 to 17 bations in Children. Am J Respir Crit Care years, 2001-2010. J Allergy Clin Immunol Med 2017; 195:854-9. 2014; 134:547-53 e5. 5. Chan MT, Leung DY, Szefler SJ, Spahn 15. Ortega V.E., Meyers D.A. Pharmaco- JD. Difficult-to-control asthma: clinical genetics: implications of race and ethnicity characteristics of steroid insensitive on defining genetic profiles for personalized asthma. J Allergy Clin Immunol 1998; 101: medicine. J Allergy Clin Immunol 2014; 594-601. 1331:16-26. 6. Mannino DM, Homa DM, Akinbami LJ, 16. Weiss ST. New approaches to Moorman JE, Gwynn C, Redd SC. personalized medicine for asthma: where Surveillance for asthma--United States, are we? J Allergy Clin Immunol 2012; 1980-1999. MMWR Surveill Summ 2002; 129:327-34. 51:1-13. 17. Lemiere C, Bai T, Balter M, Bayliff C, 7. Speer CP, Silverman M. Issues relating Becker A, Boulet LP, et al. Adult Asthma to children born prematurely. Eur Respir J Consensus Guidelines update 2003. Can Suppl 1998; 27:13s-6s. Respir J 2004;11 Suppl A:9A-18A. 8. Tough SC, Hessel PA, Ruff M, Green 18. Lester L.A., Rich S.S., Blumenthal M.N., FH, Mitchell I, Butt JC. Features that Togias A, Murphy S, Malveaux F, et al. distinguish those who die from asthma Ethnic differences in asthma and associated from community controls with asthma. J phenotypes: collaborative study on the Asthma 1998; 35(8):657-65. genetics of asthma. J Allergy Clin Immunol 9. Szefler SJ, Martin RJ, King TS, 2001;108:357-62. Boushey HA, Cherniack RM, Chinchilli 19. Newman D.L., Abney M, McPeek M.S., VM, et al. Significant variability in Ober C, Cox N.J. The importance of response to inhaled corticosteroids for genealogy in determining genetic associa- persistent asthma. J Allergy Clin Immunol tions with complex traits. Am J Hum Genet 2002; 109:410-8. 2001;69:1146-8. 10. Szefler SJ, Phillips BR, Martinez FD, 20. Ober C, Abney M, McPeek MS. The Chinchilli VM, Lemanske RF, Strunk RC, genetic dissection of complex traits in a et al. Characterization of within-subject founder population. Am J Hum Genet. 2001; responses to fluticasone and montelukast 69(5):1068-79. in childhood asthma. J Allergy Clin 21. Vijverberg SJ, Raaijmakers JA, Immunol 2005; 115:233-42. Maitland-van der Zee AH. ADRB2 Arg16 11. Szefler SJ, Phillips BR, Martinez FD, and the need for collaboration in childhood Chinchilli VM, Lemanske RF, Strunk RC, asthma pharmacogenomics. Pharmaco- et al. Characterization of within-subject genomics 2013;14:1937-9. responses to fluticasone and montelukast 22. Vijverberg SJ, Turner SW, Palmer CN, in childhood asthma. J Allergy Clin Tantisira KG, Maitland-van der Zee AH. Immunol 2005; 115:233-42. Realizing Personalized Medicine in 12. Burchard EG, Avila PC, Nazario S, Asthmatic Children Requires Large-Scale Casal J, Torres A, Rodriguez-Santana JR, Collaboration. Pediat Therapeut 2015;5: e127.
188
23. Park HW, Tantisira KG, Weiss ST. Biochem Biophys Res Commun 2015; 457: Pharmacogenomics in asthma therapy: 295-9. where are we and where do we go? Annu 33. Bin B.H, Bhin J, Yang SH, Shin M, Nam Rev Pharmacol Toxicol 2015; 55:129-47. YJ, Choi DH, et al. Membrane-Associated 24. Farzan N, Vijverberg SJ, Arets HG, Transporter Protein (MATP) Regulates Raaijmakers JA, Maitland-van der Zee Melanosomal pH and Influences Tyrosinase AH. Pharmacogenomics of inhaled Activity. PLoS One 2015; 10:e0129273. corticosteroids and leukotriene modifiers: 34. Paternoster L, Standl M, Waage J, a systematic review. Clin Exp Allergy Baurecht H, Hotze M, Strachan DP, et al. 2017;47:271-93. Multi-ancestry genome-wide association 25. Vijverberg SJH, Farzan N, Slob EMA, study of 21,000 cases and 95,000 controls Neerincx AH, Maitland-van der Zee AH. identifies new risk loci for atopic dermatitis. Treatment response heterogeneity in Nat Genet 2015; 47:1449-56. asthma: the role of genetic variation. 35. Wang Y, Tong C, Wang Z, Mauger D, Expert Rev Respir Med 2018; 12:55-65. Tantisira KG, Israel E, et al. Pharmaco- 26. Tantisira KG, Lasky-Su J, Harada M, dynamic genome-wide association study Murphy A, Litonjua AA, Himes BE, et al. identifies new responsive loci for gluco- Genomewide association between corticoid intervention in asthma. Pharmaco- GLCCI1 and response to glucocorticoid genomics J 2015; 15:422-9. therapy in asthma. N Engl J Med 2011; 36. Mosteller M, Hosking L, Murphy K, Shen 365:1173-83. J, Song K, Nelson M, et al. No evidence of 27. Tantisira KG, Damask A, Szefler SJ, large genetic effects on steroid response in Schuemann B, Markezich A, Su J, et al. asthma patients. J Allergy Clin Immunol Genome-wide association identifies the T 2017; 139:797-803 e7. gene as a novel asthma pharmacogenetic 37. Acosta-Herrera M, Pino-Yanes M, locus. Am J Respir Crit Care Med 2012; Blanco J, Ballesteros JC, Ambros A, 185:1286-91. Corrales A, et al. Common variants of 28. Wu A.C., Himes B.E., Lasky-Su J, NFE2L2 gene predisposes to acute Litonjua A, Peters S.P., Lima J, et al. respiratory distress syndrome in patients Inhaled corticosteroid treatment modu- with severe sepsis. Crit Care 2015; 19:256. lates ZNF432 gene variant's effect on 38. Villar J, Blanco J, Kacmarek RM. Current bronchodilator response in asthmatics. J incidence and outcome of the acute res- Allergy Clin Immunol 2014; 133:723-8 e3. piratory distress syndrome. Curr Opin Crit 29. Park HW, Dahlin A, Tse S, Duan QL, Care 2016;22:1-6. Schuemann B, Martinez FD, et al. Genetic 39. Wells KE, Cajigal S, Peterson EL, predictors associated with improvement of Ahmedani BK, Kumar R, Lanfear DE, et al. asthma symptoms in response to inhaled Assessing differences in inhaled cortico- corticosteroids. J Allergy Clin Immunol steroid response by self-reported race - 2014;133:664-9 e5. ethnicity and genetic ancestry among 30. Park TJ, Park JS, Cheong HS, Park asthmatic subjects. J Allergy Clin Immunol BL, Kim LH, Heo JS, et al. Genome-wide 2016; 137:1364-9e2. association study identifies ALLC poly- 40. Gonzalez Burchard E, Borrell LN, morphisms correlated with FEV(1) change Choudhry S, Naqvi M, Tsai HJ, Rodriguez- by corticosteroid. Clin Chim Acta 2014; Santana JR, et al. Latino populations: a 436:20-6. unique opportunity for the study of race, 31. Dahlin A, Denny J, Roden DM, Brilliant genetics, and social environment in epi- MH, Ingram C, Kitchner TE, et al. CMTR1 demiological research. Am J Public Health. is associated with increased asthma 2005; 95:2161-8. exacerbations in patients taking inhaled 41. Choudhry S, Burchard EG, Borrell LN, corticosteroids. Immun Inflamm Dis 2015; Tang H, Gomez I, Naqvi M, et al. Ancestry- 3:350-9. environment interactions and asthma risk 32. Ahasan M.M., Wakae K, Wang Z, among Puerto Ricans. Am J Respir Crit Kitamura K, Liu G, Koura M, et al. Care Med 2006; 174:1088-93. APOBEC3A and 3C decrease human 42. Shriner D, Adeyemo A, Ramos E, papillomavirus 16 pseudovirion infectivity.
189
Chen G, Rotimi CN. Mapping of disease- air pollution and asthma risk in minority associated variants in admixed pop- children. The GALA II and SAGE II studies. ulations. Genome Biol. 2011; 12:223. Am J Respir Crit Care Med 2013; 188:309- 43. Hernandez-Pacheco N, Flores C, Oh 18. SS, Burchard EG, Pino-Yanes M. What 52. Pino-Yanes M, Thakur N, Gignoux CR, Ancestry Can Tell Us About the Genetic Galanter JM, Roth LA, Eng C, et al. Genetic Origins of Inter-Ethnic Differences in ancestry influences asthma susceptibility Asthma Expression. Curr Allergy Asthma and lung function among Latinos. J Allergy Rep 2016; 16:53. Clin Immunol 2015; 135:228-35. 44. Farzan N, Vijverberg S.J., Andiappan 53. Hawcutt DB, Jorgensen AL, Wallin N, A.K., Arianto L, Berce V, Blanca-Lopez N, Thompson B, Peak M, Lacy D, et al. Adrenal et al. Rationale and design of the multi- responses to a low-dose short synacthen ethnic Pharmacogenomics in Childhood test in children with asthma. Clin Endocrinol Asthma consortium. Pharmacogenomics (Oxf) 2015; 82:648-56. 2017;18:931-43. 54. White MJ, Risse-Adams O, Goddard P, 45. Palmer CN, Doney AS, Lee SP, Murrie Contreras MG, Adams J, Hu D, et al. Novel I, Ismail T, Macgregor DF, et al. Glu- genetic risk factors for asthma in African tathione S-transferase M1 and P1 American children: Precision Medicine and genotype, passive smoking, and peak the SAGE II Study. Immunogenetics 2016; expiratory flow in asthma. Pediatrics 2006; 68:391-400. 118:710-6. 55. Thakur N, Oh SS, Nguyen EA, Martin M, 46. Palmer CN, Lipworth BJ, Lee S, Ismail Roth LA, Galanter J, et al. Socioeconomic T, Macgregor DF, Mukhopadhyay S. status and childhood asthma in urban Arginine-16 beta2 adrenoceptor genotype minority youths. The GALA II and SAGE II predisposes to exacerbations in young studies. Am J Respir Crit Care Med 2013; asthmatics taking regular salmeterol. 188:1202-9. Thorax 2006; 61:940-4. 56. Reddel H.K., Taylor D.R., Bateman ED, 47. Moffatt MF, Kabesch M, Liang L, Dixon Boulet L.P., Boushey H.A., Busse WW, et al. AL, Strachan D, Heath S, et al. Genetic An official American Thoracic Society/ variants regulating ORMDL3 expression European Respiratory Society statement: contribute to the risk of childhood asthma. asthma control and exacerbations: standar- Nature 2007; 448:470-3. dizing endpoints for clinical asthma trials 48. Tavendale R, Macgregor D.F., and clinical practice. Am J Respir Crit Care Mukhopadhyay S, Palmer C.N. A poly- Med 2009; 180:59-99. morphism controlling ORMDL3 ex- 57. Das S, Forer L, Schonherr S, Sidore C, pression is associated with asthma that is Locke AE, Kwong A, et al. Next-generation poorly controlled by current medications. J genotype imputation service and methods. Allergy Clin Immunol 2008; 121:860-3. Nat Genet 2016; 48:1284-7. 49. Koster ES, Raaijmakers JA, Koppelman 58. McCarthy S, Das S, Kretzschmar W, G.H., Postma D.S., van der Ent C.K., Delaneau O, Wood AR, Teumer A, et al. A Koenderman L, et al. Pharmacogenetics of reference panel of 64,976 haplotypes for anti-inflammatory treatment in children with genotype imputation. Nat Genet 2016; asthma: rationale and design of the 48:1279-83. PACMAN cohort. Pharmacogenomics 59. Delaneau O, Coulonges C, Zagury JF. 2009;10:1351-61. Shape-IT: new rapid and accurate algorithm for haplotype inference. BMC Bioinformatics 50. Berce V, Kozmus CE, Potocnik U. 2008; 9:540. Association among ORMDL3 gene expression, 17q21 polymorphism and 60. Fuchsberger C, Abecasis GR, Hinds DA. response to treatment with inhaled minimac2: faster genotype imputation. Bio- corticosteroids in children with asthma. informatics 2015; 31:782-4. Pharmacogenomics J 2013;13:523-9. 61. Kang H.M. EPACTS (Efficient and 51. Nishimura KK, Galanter JM, Roth LA, Parallelizable Association Container Tool- Oh SS, Thakur N, Nguyen EA, et al. Early- box).http://genome.sph.umich.edu/wiki/EPA life CTS (2016).
190
62. Price AL, Patterson NJ, Plenge RM, 73. Lazaridis I, Nadel D, Rollefson G, Weinblatt M.E., Shadick N.A., Reich D. Merrett D.C, Rohland N, Mallick S, et al. Principal components analysis corrects for Genomic insights into the origin of farming in stratification in genome-wide association the ancient Near East. Nature 2016; 536: studies. Nat Genet 2006; 38:904-9. 419-24. 63. Han B, Eskin E. Random-effects 74. Mathias RA, Taub MA, Gignoux CR, Fu model aimed at discovering associations W, Musharoff S, O'Connor TD, et al. A in meta-analysis of genome-wide continuum of admixture in the Western association studies. Am J Hum Genet Hemisphere revealed by the African 2011; 88:586-98. Diaspora genome. Nat Commun 2016;7: 64. R Development Core Team. R: A 12522. language and environment for statistical 75. Navaratnam N, Morrison JR, computing. R Foundation for Statistical Bhattacharya S, Patel D, Funahashi T, Computing, Vienna, Austria. http://www. Giannoni F, et al. The p27 catalytic subunit R-project.org/ (2013). of the apolipoprotein B mRNA editing 65. Berce V, Kozmus C.E., Potocnik U. enzyme is a cytidine deaminase. J Biol Association among ORMDL3 gene ex- Chem 1993; 268:20709-12. pression, 17q21 polymorphism and 76. Smith HC, Bennett RP, Kizilyer A, response to treatment with inhaled McDougall WM, Prohaska KM. Functions corticosteroids in children with asthma. and regulation of the APOBEC family of Pharmacogenomics J 2013;13:523-9. proteins. Semin Cell Dev Biol 2012; 23:258- 66. Ward LD, Kellis M. HaploReg v4: 68. systematic mining of putative causal 77. Vieira VC, Soares MA. The role of variants, cell types, regulators and target cytidine deaminases on innate immune genes for human complex traits and responses against human viral infections. disease. Nucleic Acids Res 2016; 44: Biomed Res Int 2013; 2013:683095. D877-81. 78. Deng Y, Du Y, Zhang Q, Han X, Cao G. 67. The ENCODE Project Consortium. An Human cytidine deaminases facilitate integrated encyclopedia of DNA elements hepatitis B virus evolution and link inflamma- in the human genome. Nature 2012; tion and hepatocellular carcinoma. Cancer 489:57-74. Lett 2014; 343:161-71. 68. GTEx Consortium. The Genotype- 79. Desimmie BA, Delviks-Frankenberrry Tissue Expression (GTEx) project. Nat KA, Burdick RC, Qi D, Izumi T, Pathak VK. Genet 2013; 45:580-5. Multiple APOBEC3 restriction factors for 69. Kapushesky M, Emam I, Holloway E, HIV-1 and one Vif to rule them all. J Mol Biol. Kurnosov P, Zorin A, Malone J, et al. Gene 2014; 426:1220-45. expression atlas at the European bio- 80. Janahi EM, McGarvey MJ. The inhibition informatics institute. Nucleic Acids Res of hepatitis B virus by APOBEC cytidine 2010; 38:D690-8. deaminases. J Viral Hepat 2013; 20:821-8. 70. Fishilevich S, Nudel R, Rappaport N, 81. Doehle BP, Schafer A, Cullen BR. Hadar R, Plaschkes I, Iny Stein T, et al. Human APOBEC3B is a potent inhibitor of GeneHancer: genome-wide integration of HIV-1 infectivity and is resistant to HIV-1 Vif. enhancers and target genes in Gene- Virology. 2005; 339:281-8. Cards. 2017;2017. 82. Bogerd H.P., Wiegand H.L., Doehle B.P, 71. Elias PM, Williams ML. Basis for the Cullen B.R. The intrinsic antiretroviral factor gain and subsequent dilution of epidermal APOBEC3B contains two enzymatically pigmentation during human evolution: The active cytidine deaminase domains. barrier and metabolic conservation hypo- Virology 2007; 364:486-93. theses revisited. Am J Phys Anthropol 83. Berger G, Durand S, Fargier G, Nguyen 2016; 161:189-207. XN, Cordeil S, Bouaziz S, et al. APOBEC3A 72. Kanu N, Cerone MA, Goh G, Zalmas is a specific inhibitor of the early phases of LP, Bartkova J, Dietzen M, et al. DNA HIV-1 infection in myeloid cells. PLoS replication stress mediates APOBEC3 Pathog. 2011; 7:e1002221. family mutagenesis in breast cancer. 84. Minkah N, Chavez K, Shah P, Maccarthy Genome Biol 2016; 17:185. T, Chen H, Landau N, et al. Host restriction of murine gammaherpesvirus 68 replication
191
By human APOBEC3 cytidine de- 95. Long J, Delahanty RJ, Li G, Gao YT, Lu aminases but not murine APOBEC3. W, Cai Q, et al. A common deletion in the Virology 2014;454-455:215-26. APOBEC3 genes and breast cancer risk. J 85. Lucifora J, Xia Y, Reisinger F, Zhang Natl Cancer Inst 2013;105:573-9. K, Stadler D, Cheng X, et al. Specific and 96. Nik-Zainal S, Wedge DC, Alexandrov nonhepatotoxic degradation of nuclear LB, Petljak M, Butler AP, Bolli N, et al. hepatitis B virus cccDNA. Science 2014; Association of a germline copy number 343:1221-8. polymorphism of APOBEC3A and APO- 86. Warren CJ, Xu T, Guo K, Griffin LM, BEC3B with burden of putative APOBEC- Westrich JA, Lee D, et al. APOBEC3A dependent mutations in breast cancer. Nat functions as a restriction factor of human Genet 2014; 46:487-91. papillomavirus. J Virol 2015; 89:688-702. 97. Yan S, He F, Gao B, Wu H, Li M, Huang 87. Nakaya Y, Stavrou S, Blouch K, L, et al. Increased APOBEC3B Predicts Tattersall P, Ross SR. In Vivo Worse Outcomes in Lung Cancer: A Examination of Mouse APOBEC3- and Comprehensive Retrospective Study. J Human APOBEC3A- and APOBEC3G- Cancer 2016; 7:618-25. Mediated Restriction of Parvovirus and 98. Gansmo LB, Romundstad P, Hveem K, Herpesvirus Infection in Mouse Models. J Vatten L, Nik-Zainal S, Lonning PE, et al. Virol 2016;90:8005-12. APOBEC3A/B deletion polymorphism and 88. Chen Y, Hu J, Cai X, Huang Y, Zhou cancer risk. Carcinogenesis 2018; 39:118- X, Tu Z, et al. APOBEC3B edits HBV DNA 24. and inhibits HBV replication during 99. Saraya T, Kurai D, Ishii H, Ito A, Sasaki reverse transcription. Antiviral Res 2018; Y, Niwa S, et al. Epidemiology of virus- 149:16-25. induced asthma exacerbations: with special 89. Kostrzak A, Henry M, Demoyen P.L., reference to the role of human rhinovirus. Wain-Hobson S, Vartanian JP. APO- Front Microbiol 2014; 5:226. BEC3A catabolism of electroporated 100. Iwasaki J, Smith W.A., Khoo S.K., plasmid DNA in mouse muscle. Gene Bizzintino J, Zhang G, Cox D.W., et al. Ther 2015; 22:96-103. Comparison of rhinovirus antibody titers in 90. Sharma S, Patnaik S.K., Taggart R.T., children with asthma exacerbations and Kannisto E.D., Enriquez S.M., Gollnick P, species-specific rhinovirus infection. J et al. APOBEC3A cytidine deaminase Allergy Clin Immunol 2014; 134:25-32 induces RNA editing in monocytes and 101. Duenas Meza E, Jaramillo C.A., Correa macrophages. Nat Commun 2015; 6: E, Torres-Duque C.A., Garcia C, Gonzalez 6881. M, et al. Virus and Mycoplasma pneumoniae 91. Niavarani A, Currie E, Reyal Y, Anjos- prevalence in a selected pediatric pop- Afonso F, Horswell S, Griessinger E, et al. ulation with acute asthma exacerbation. J APOBEC3A is implicated in a novel class Asthma 2016; 53:253-60. of G-to-A mRNA editing in WT1 102. Abecasis GR, Auton A, Brooks LD, transcripts.PLoS One 2015;10: e0120089. DePristo MA, Durbin RM, Handsaker RE, et 92. Niavarani A, Currie E, Reyal Y, Anjos- al. An integrated map of genetic variation Afonso F, Horswell S, Griessinger E, et al. from 1,092 human genomes. Nature. 2012; APOBEC3A is implicated in a novel class 491:56-65. of G-to-A mRNA editing in WT1 103. Mensah-Ablorh A, Lindstrom S, transcripts.PLoS One 2015; 10:e0120089. Haiman CA, Henderson BE, Marchand LL, 93. Burns MB, Lackey L, Carpenter MA, Lee S, et al. Meta-Analysis of Rare Variant Rathore A, Land AM, Leonard B, et al. Association Tests in Multiethnic Pop- APOBEC3B is an enzymatic source of ulations. Genet Epidemiol. 2016; 40:57-65. mutation in breast cancer. Nature 2013; 104. Homa DM, Mannino DM, Lara M. 494:366-70. Asthma mortality in U.S. Hispanics of 94. Burns MB, Temiz NA, Harris RS. Mexican, Puerto Rican, and Cuban heritage, Evidence for APOBEC3B mutagenesis in 1990-1995. Am J Respir Crit Care Med multiple human cancers. Nat Genet 2013; 2000; 161:504-9. 45:977-83.
192
105. Moorman JE, Zahran H, Truman BI, 107. Bustamante CD, Burchard EG, De la Molla MT. Current asthma prevalence - Vega FM. Genomics for the world. Nature United States, 2006-2008. Morbidity and 2011; 475:163-5. Mortality Weekly Report 2011; 60:84-6. 108. Marcus J.H., Novembre J. Visualizing 106. Barr RG, Aviles-Santa L, Davis SM, the Geography of Genetic Variants. Bio- Aldrich TK, Gonzalez F, 2nd, Henderson informatics 2016; 33:594–595. AG, et al. Pulmonary Disease and Age at Immigration among Hispanics. Results from the Hispanic Community Health Study/Study of Latinos. Am J Respir Crit Care Med; 193:386-95.
193
Online Supplementary
Genome-wide association study of inhaled corticosteroid response in African-admixed children with asthma
194
European populations included in the replication phase followMAGICS (n = 147)
FollowMAGICS is the follow-up phase of the observational Multicenter Asthma Genetics in Childhood Study (MAGICS). This includes physician diagnosed and self- identified asthmatic children and young adults aged 7 to 25 years old recruited at secondary and tertiary centers from Austria and Germany (1). A description of the genome-wide genotyping with the Illumina Sentrix HumanHap300 BeadChip array (Illumina) and quality control (QC) procedures is provided elsewhere (2-4).
PASS (n = 402)
The Pharmacogenetics of Adrenal Suppression study (PASS) is a multicenter cohort that was initially conceived to explore the clinical and pharmacogenomic associations between the use of corticosteroids and the adrenal suppression. Children and young adults aged 5 to 18 years old with clinical diagnosis of asthma, inhaled glucoco- rticosteroids (ICS) therapy under pediatric supervision, and clinical concern about adrenal suppression were recruited from the United Kingdom. Detailed description about the study design, data collection, characteristics of participants, genotyping with the Illumina Omni Express 8v1 array (Illumina) and QC procedures is described in previous publications (5,6).
PACMAN (n = 654)
The Pharmacogenetics of Asthma Medication in Children: Medication with Anti- inflammatory effects (PACMAN) study is an observational cohort that includes children aged 4 to 12 years old with a self-reported use of any asthma medication recruited through records of community pharmacies in the Netherlands. Detailed information on asthma symptoms, exacerbations and medication over the last 12 months was collected during visits to community pharmacies is available elsewhere (7).
ESTATe (n = 102)
The Effectiveness and Safety of Treatment with Asthma Therapy in children (ESTATe) is a case-control study that includes children and young adults (4-19 years) with a physician diagnosis of asthma recruited from primary care units in the Netherlands. Patients were selected from either Interdisciplinary Processing of Clinical Information (IPCI) database or the PHARMO Database Network. Both databases contain the
195
electronic medical records of more than one million patients throughout the Netherlands with detailed information on patient diagnosis, patient prescription (IPCI) or patient dispensing (PHARMO). During the study period (2000 - 2012) all children with asthma, aged 5 years and older and treated with asthma controller therapy were selected. Within this cohort, cases with asthma exacerbations based on use of systemic corticosteroids, emergency room (ER) visits or hospitalizations because of asthma were selected. Each case was matched to four controls based on similarity in age, gender, general practitioner (GP) and type of asthma controller therapy. Next, all potential cases and controls were invited to participate via their respective GP. If patients agreed to participate, they provided written consent, completed a research questionnaire including questions on asthma control and provided a saliva sample (for DNA extraction).
BREATHE (n = 210)
BREATHE is a study that includes children and young adults aged 3 to 22 years old with physician diagnosis of asthma recruited at primary and secondary care units from the United Kingdom. A detailed clinical history and, demographic and anthropometric information was obtained from all participants (8-10).
SLOVENIA (n = 184)
SLOVENIA is a case-control cohort including children and young adults with mild and moderate persistent asthma aged 5 to 18 years old of Slovenian origin recruited from tertiary health centers. Asthma was defined by physician diagnosis and hospital records. ICS was regularly administered to part of the asthmatic patients included in the study. Patients with other chronic inflammatory diseases excepting for those with asthma and atopic diseases and asthmatics treated with other asthma medications were excluded from the study (11).
Quality control analyses in the validation cohorts genotyped as part of the current study
QC analyses were performed in four of the validation cohorts (PACMAN, ESTATe, BREATHE and SLOVENIA) using PLINK 1.09 (12), which were genotyped for the purposes of the Pharmacogenomics in Childhood of Asthma (PiCA) consortium. Several QC steps were performed at individual level. Firstly, concordance between the reported and the genetic gender assessed by means of the genotype data from the X chromosome was inspected and individuals with discordances in gender information were discarded from further analyses. Secondly, subjects with a genotyping
196 completion rate (CR)<95% were discarded, as well as those with heterozygosity rates higher or lower than 4 standard deviations of the population mean. Thirdly, cryptic relatedness of individuals and population stratification were assessed. For that, single nucleotide polymorphisms (SNPs) and regions of extended linkage disequilibrium were pruned out keeping approximately 100,000 SNPs for each study. An identity-by- descent (IBD) matrix was estimated to remove those duplicated or related individuals. Evidence of relatedness was considered for second-degree relatives or higher evidenced by values of IBD ≥0.2. Then, a Principal Component (PC) analysis was performed with EIGENSOFT (13) in order to detect population stratification due to existence of individuals with large differences in ancestry. Additionally, this analysis provided PC estimations that were included as covariates in the association testing. Finally, those individuals with a reported use of ICS and available information regarding the presence or absence of asthma exacerbations were selected for association analyses.
Moreover, genetic markers were filtered in order to exclude those with >5% missing genotypes. However, deviations from Hardy-Weinberg Equilibrium were not inspected since the datasets analyzed only included patients with asthma.
For PACMAN, a total of 893 samples were genome-wide genotyped with the Illumina Infinium CoreExome-24 BeadChip (Illumina). From these, 23 individuals with CR<95% in addition to 20 subjects with excessive or reduced heterozygosity rates were discarded. Furthermore, ten individuals with discordance in gender information were discarded from further analyses. Fifty-three pairs of related subjects were detected and one individual from each pair was selected based on availability of information related to the presence/absence of severe asthma exacerbations and medication use. After QC, a total of 487,050 autosomal markers and 654 asthma patients treated with ICS were selected for the analyses.
Genotyping of 111 samples from ESTATe was performed with the Illumina Infinium CoreExome-24 BeadChip (Illumina), but only those subjects with a reported use of ICS were selected. From the remaining samples, those with a CR<99% and excessive autosomal heterozygosity were excluded. Furthermore, three pairs of related individuals were identified and one subject from each pair was excluded. A total of 526,121 SNPs located at autosomal chromosomes remained after QC analyses.
In the BREATHE study 288 samples were genotyped with the Illumina Infinium CoreExome-24 BeadChip (Illumina). During QC procedures, five individuals were
197
discarded due to excessive or reduced heterozygosity rates, in addition to three subjects that showed large differences in ancestry. Moreover, eight pairs of related individuals were detected, and only one participant was selected from each pair based on availability of clinical information. Furthermore, a total of 176,412 SNPs accomplished the QC criteria.
In the SLOVENIA study, genotyping of 336 samples was performed with the Illumina Global Screening Array-24 v1.0 BeadChip (Illumina). Ten subjects with discordances in gender information were removed. Moreover, 13 subjects with a genotyping completion rate <95% and two with an excessive or reduced proportion of heterozygote genotypes were discarded for association analyses. After QC analyses, 184 individuals with a reported use of ICS and availability of data related to the presence/absence of asthma exacerbations during the previous 12 months were kept for the analyses. The number of autosomal genetic variants that passed the QC was 560,996.
198
Online Supplementary References
1. Moffatt MF, Kabesch M, Liang L, Dixon 9. Palmer CN, Doney AS, Lee SP, Murrie I, AL, Strachan D, Heath S, et al. Genetic Ismail T, Macgregor DF, et al. Glutathione variants regulating ORMDL3 expression S-transferase M1 and P1 genotype, contribute to the risk of childhood asthma. passive smoking, and peak expiratory flow Nature 2007;448:470-3. in asthma. Pediatrics 2006;118:710-6. 2. Pandey RC, Michel S, Schieck M, Binia 10. Tavendale R, Macgregor DF, A, Liang L, Klopp N, et al. Polymorphisms Mukhopadhyay S, Palmer CN. A in extracellular signal-regulated kinase polymorphism controlling ORMDL3 family influence genetic susceptibility to expression is associated with asthma that asthma. J Allergy Clin Immunol 2013;131: is poorly controlled by current medications. 1245-7. J Allergy Clin Immunol 2008;121:860-3. 3. Schieck M, Michel S, Suttner K, Illig T, 11. Berce V, Kozmus CE, Potocnik U. Zeilinger S, Franke A, et al. Genetic Association among ORMDL3 gene variation in TH17 pathway genes, child- expression, 17q21 polymorphism and hood asthma, and total serum IgE levels. J response to treatment with inhaled cortico- Allergy Clin Immunol 2014;133: 888-91. steroids in children with asthma. Phar- 4. Nieuwenhuis MA, Siedlinski M, van den macogenomics J 2013;13:523-9. Berge M, Granell R, Li X, Niens M, et al. 12. Purcell S, Neale B, Todd-Brown K, Combining genomewide association study Thomas L, Ferreira MA, Bender D, et al. and lung eQTL analysis provides evidence PLINK: a tool set for whole-genome for novel genes associated with asthma. association and population-based linkage Allergy 2016;71:1712-20. analyses. Am J Hum Genet 2007;81:559- 5. Hawcutt DB, Jorgensen AL, Wallin N, 75. Thompson B, Peak M, Lacy D, et al. 13. Price AL, Patterson NJ, Plenge RM, Adrenal responses to a low-dose short Weinblatt ME, Shadick NA, Reich D. synacthen test in children with asthma. Clin Principal components analysis corrects for Endocrinol 2015;82:648-56 stratification in genome-wide association 6. Hawcutt D.B., Francis B, Carr D.F., studies. Nat Genet 2006;38:904-9. Jorgensen A.L., Yin P, Wallin N, et al. Susceptibility to corticosteroid induced adrenal suppression: a genome-wide association study. Lancet Respir Med 2018 (in press). 7. Koster ES, Raaijmakers JA, Koppelman G.H., Postma D.S., van der Ent C.K., Koenderman L, et al. Pharmacogenetics of anti-inflammatory treatment in children with asthma: rationale and design of the PACMAN cohort. Pharmacogenomics 2009 10:1351-61. 8. Palmer CN, Lipworth BJ, Lee S, Ismail T, Macgregor DF, Mukhopadhyay S. Arginine- 16 beta2 adrenoceptor genotype pre- disposes to exacerbations in young asthmatics taking regular salmeterol. Thorax 2006;61:940-4.
199
Figure S1. Quantile-quantile plot of association results of ICS response in African- admixed populations. Observed and expected association results are represented as -log10 p-value on the y-axis and x-axis, respectively.
200
Table S1. Clinical and demographic characteristics of the European populations analyzed in the replication phase.
followMAGICS PASS PACMAN ESTATe BREATHE SLOVENIA (n = 147) (n = 402) (n = 654) (n = 102) (n = 210) (n = 184)
Gender (% male) 59.9 55.0 61.6 58.8 60.5 43.5
Mean age ± SD 17.2 ± 3.0 12.0 ± 2.0 8.7 ± 2.3 10.6 ± 4.2 9.1 ± 4.0 10.8 ± 3.4 (years)
Recruitment country Germany/Austria United Kingdom Netherlands Netherlands United Kingdom Slovenia
Ethnicity European European European European European European
Asthma exacerbations in the 53.1 51.7 a 11.0 48 51.4 a 37.5 last 12 months (%)
ER visits/ ER visits/ OCS use/ ER visits/ hospitalizations/ ER visits/ Definition OCS use hospitalization/ hospit alizations/ hospitalizations/ GP visits/specialist OCS use OCS use school absences OCS use visits
ER visits (%) 7.5 NA 6.1 NA NA 27.7
OCS use (%) NA 51.7 6.7 35.3 47.6 10.3
Hospitalizations (%) 3.4 NA NA 12.7 b 46.2 9.8
201
201
202
GP visits (%) 49.0 NA NA NA NA NA
Specialist visits (%) 21.8 NA NA NA NA NA
School absences (%) NA NA NA NA 46.2 NA
Illumina Infinium Illumina Global Illumina Sentrix Illumina Omni Illumina Infinium Illumina Infinium CoreExome-24 Screening Array- Genotyping platform HumanHap300 Express 8v1 CoreExome-24 CoreExome-24 BeadChip 24 v1.0 BeadChip BeadChip (Illumina) (Illumina) BeadChip (Illumina) BeadChip (Illumina) (Illumina) (Illumina)
a Asthma exacerbations-related data was available for the 6 precedent months of the study enrollment; b ER visits and hospitalizations were considered as a single variable. SD: standard deviation; ER: emergency room; OCS: systemic corticosteroids; GP: general practitioner; NA: not available.
202
Table S2. Summary of the conditional regression models for each gene region with suggestive associations in African-admixed populations.
Meta-analysis Conditional regression model
Nearest gene(s) SNP Chr. a Position OR (95% CI) b p-value Conditioned on p-value
rs11121611 1 6367219 0.55 (0.43-0.70) 1.65 x 10-6 NA ACOT7 rs11121611 rs3789494 1 6370476 0.55 (0.43-0.70) 2.14 x 10-6 0.279
HLX rs116561422 1 221136237 0.31 (0.19-0.50) 2.14 x 10-6 NA NA
ZP4 rs606572 1 238746080 0.58 (0.46-0.72) 1.80 x 10-6 NA NA
DTNBP1-MYLIP rs35514893 6 15909525 0.36 (0.23-0.55) 2.86 x 10-6 NA NA
TRDN rs4897302 6 123886231 1.58 (1.31-1.91) 1.75 x 10-6 NA NA
LHFPL3 rs61585310 7 104006510 0.61 (0.49-0.75) 2.85 x 10-6 NA NA
LHX2 rs7851998 9 126828514 0.56 (0.44-0.72) 3.97 x 10-6 NA NA
rs7122239 11 86165109 1.40 (0.89-2.21) 4.27 x 10-6 c 0.315
ME3 rs2125362 11 86167136 1.31 (0.68-2.56) 3.53 x 10-6 c rs2125362 NA
rs2125363 11 86167202 1.31 (0.68-2.56) 3.53 x 10-6 c 0.278
KL rs450789 13 33578233 0.64 (0.53-0.77) 3.33 x 10-6 NA NA
rs140275688 18 15096270 0.42 (0.30-0.61) 2.26 x 10-6 0.263 rs540731596, ANKRD30B rs540731596 18 15097277 0.41 (0.29-0.59) 1.34 x 10-6 NA rs12959468 -6
203 rs142954031 18 15112933 0.41 (0.29-0.60) 2.05 x 10 0.485
203
204
rs147911586 18 15115442 0.41 (0.29-0.59) 1.34 x 10-6 0.939
rs141514992 18 15116537 0.41 (0.29-0.59) 1.34 x 10-6 0.904
rs570126373 18 15151837 0.43 (0.30-0.62) 3.84 x 10-6 0.927
rs12959468 18 15182381 0.39 (0.26-0.58) 2.99 x 10-6 NA
KCNN1 rs2278992 19 18095769 0.59 (0.47-0.74) 3.76 x 10-6 NA
rs2278993 19 18096073 0.59 (0.47-0.74) 3.76 x 10-6 rs2278992 0.589
rs76657538 19 18098215 0.59 (0.47-0.74) 3.76 x 10-6 0.336
SNX21 rs113480515 20 44461764 0.28 (0.16-0.47) 1.86 x 10-6 NA NA
rs6001366 22 39399941 0.47 (0.35-0.65) 2.53 x 10-6 NA
rs5995653 22 39404249 0.66 (0.56-0.79) 4.80 x 10-6 NA APOBEC3B- rs6001366, APOBEC3C rs5995653 rs6001375 22 39407116 0.64 (0.53-0.77) 4.36 x 10-6 0.361
rs4299420 22 39407685 0.64 (0.53-0.78) 4.81 x 10-6 0.335 a Chromosome; b OR: Odds ratio for the effect alleles (additive model); c Random-effect model was applied since heterogeneity was found between Latinos/Hispanics and African Americans. Independent SNPs of each gene region are in boldface.
204
Table S3. Summary of the independent SNPs suggestively associated with asthma exacerbations in African-admixed individuals treated with ICS. GALA II (n=854) SAGE (n=493) Meta-analysis (n=1,347) Nearest Freq. OR Freq. OR OR SNP Chr.a A1/A2 p-value p-value p-value gene(s) b (95% CI) c b (95% CI) c (95% CI) c
0.58 0.42 0.55 rs11121611 1 ACOT7 G/T 0.201 6.29 x 10-5 0.062 5.08 x 10-3 1.65 x 10-6 (0.44-0.75) (0.23-0.77) (0.43-0.70)
0.31 0.31 0.31 rs116561422 1 HLX T/G 0.011 6.76 x 10-3 0.059 1.00 x 10-4 2.14 x 10-6 (0.13-0.72) (0.17-0.56) (0.19-0.50)
0.51 0.62 0.58 rs606572 1 ZP4 G/A 0.884 8.75 x 10-4 0.644 4.44 x 10-4 1.80 x 10-6 (0.34-0.76) (0.47-0.81) (0.46-0.72)
DTNBP1- 0.51 0.29 0.36 rs35514893 6 T/C 0.020 6.67 x 10-2 0.082 7.46 x 10-6 2.86 x 10-6 MYLIP (0.25-1.05) (0.17-0.50) (0.23-0.55)
1.58 1.59 1.58 rs4897302 6 TRDN T/C 0.505 1.28 x 10-4 0.221 4.26 x 10-3 1.75 x 10-6 (1.25-2.00) (1.16-2.18) (1.31-1.91)
0.59 0.63 0.61 rs61585310 7 LHFPL3 G/T 0.796 3.27 x 10-4 0.763 2.54 x 10-3 2.85 x 10-6 (0.44-0.78) (0.46-0.85) (0.49-0.75)
0.52 0.78 0.56 rs7851998 9 LHX2 A/G 0.191 2.68 x 10-6 0.046 0.421 3.97 x 10-6 (0.40-0.69) (0.43-1.43) (0.44-0.72)
1.84 0.93 1.31 rs2125362 11 ME3 A/G 0.684 1.57 x 10-7 0.750 0.624 3.53 x 10-6 c
(1.46-2.30) (0.70-1.24) (0.68-2.56)
205
205
206
0.64 0.64 0.64 rs450789 13 KL G/A 0.334 2.82 x 10-4 0.271 3.69 x 10-3 3.33 x 10-6 (0.50-0.81) (0.47-0.86) (0.53-0.77)
0.34 0.44 0.41 rs540731596 18 ANKRD30B T/A 0.021 3.19 x 10-3 0.119 1.07 x 10-4 1.34 x 10-6 (0.17-0.70) (0.29-0.67) (0.29-0.59)
0.48 0.33 0.39 rs12959468 18 ANKRD30B A/G 0.039 1.46 x 10-2 0.077 4.28 x 10-5 2.99 x 10-6 (0.27-0.86) (0.19-0.56) (0.26-0.58)
0.59 0.59 0.59 rs2278992 19 KCNN1 C/T 0.176 2.92 x 10-4 0.151 4.04 x 10-3 3.76 x 10-6 (0.44-0.79) (0.41-0.85) (0.47-0.74)
0.15 0.37 0.28 rs113480515 20 SNX21 G/C 0.010 6.64 x 10-5 0.046 2.46 x 10-3 1.86 x 10-6 (0.06-0.38) (0.19-0.70) (0.16-0.47)
APOBEC3B- 0.50 0.42 0.47 rs6001366 22 T/C 0.079 2.27 x 10-4 0.064 3.04 x 10-3 2.53 x 10-6 APOBEC3C (0.34-0.72) (0.24-0.75) (0.35-0.65)
APOBEC3B- 0.59 0.75 0.66 rs5995653 22 A/G 0.285 2.82 x 10-5 0.508 0.024 4.80 x 10-6 APOBEC3C (0.46-0.76) (0.58-0.96) (0.56-0.79)
a Chromosome; b Frequency of the effect allele; c OR: Odds ratio for the effect alleles (additive model); d Random-effect model was applied since heterogeneity was found between Latinos/Hispanics and African Americans. A1: Effect allele; A2: Non-effect allele; CI: Confidence Interval.
206
Table S4. Summary results of replication at SNP level in African-admixed individuals of associations reported in published GWAS of ICS response. Meta-analysis GALA II (n=854) SAGE (n=493) (n=1,347) Freq. OR Freq. OR OR p- Nearest gene(s) SNP Chr.a A1/A2 p-value p-value Citation b (95% CI) c b (95% CI) c (95% CI) c value ALLC 1.10 0.96 1.07 rs17445240 2 G/A 0.067 0.687 0.020 0.936 0.746 (0.70-1.73) (0.39-2.38) (0.71-1.60)
0.98 0.95 0.96 rs13418767 2 T/G 0.111 0.915 0.226 0.728 0.739 (0.69-1.4) (0.70-1.29) (0.76-1.21)
1.06 0.79 0.94 rs6754459 2 T/C 0.298 0.605 0.675 0.108 0.528 (0.84-1.35) (0.59-1.05) (0.78-1.13) 1 1.05 1.15 1.07 rs17017879 2 C/G 0.038 0.867 0.017 0.795 0.789 (0.61-1.79) (0.40-3.27) (0.66-1.72)
1.28 0.88 1.03 rs7558370 2 C/A 0.066 0.293 0.107 0.526 0.842 (0.81-2.01) (0.60-1.30) (0.77-1.38)
0.84 1.02 0.92 rs11123610 2 A/G 0.704 0.164 0.487 0.895 0.348 (0.66-1.07) (0.78-1.33) (0.77-1.10)
FBXL7 1.01 1.31 1.14 rs10044254 5 G/A 0.240 0.920 0.349 0.057 0.170 2 (0.78-1.32) (0.99-1.73) (0.94-1.38)
1.16 0.65 1.04 FTSJD2 rs2395672 6 A/G 0.170 0.342 0.045 0.175 0.781 3 (0.86-1.56) (0.34-1.21) (0.79-1.36)
1.00 1.06 1.02 MMS22L-FBXL4 rs6924808 6 A/G 0.530 0.974 0.574 0.643 0.792 4
(0.81-1.23) (0.82-1.39) (0.87-1.21)
207
207
208 0.94 0.99 0.97
rs6456042 6 C/A 0.746 0.657 0.639 0.964 0.725 PDE10A-T (0.73-1.22) (0.76-1.29) (0.81-1.16) 5 0.94 0.99 0.97 rs3127412 6 T/C 0.746 0.657 0.639 0.964 0.725 (0.73-1.22) (0.76-1.29) (0.81-1.16)
0.83 1.34 1.04 rs1134481 6 G/T 0.686 0.119 0.786 0.067 0.233 (0.66-1.05) (0.98-1.82) (0.66-1.66)
0.87 1.17 0.97 rs2305089 6 T/C 0.495 0.226 0.263 0.295 0.718 5 (0.70-1.09) (0.87-1.58) (0.81-1.15)
0.88 1.33 1.07 rs3099266 6 C/T 0.658 0.276 0.802 0.075 0.435 (0.70-1.11) (0.97-1.83) (0.71-1.6)
1.20 1.10 1.17 UMAD1-GLCCI1 rs37972 7 C/T 0.613 0.095 0.788 0.523 0.084 6 (0.97-1.49) (0.81-1.50) (0.98-1.39)
1.07 1.16 1.11 MAGI2 rs2691529 7 T/C 0.743 0.562 0.734 0.302 0.271 3 (0.85-1.36) (0.87-1.55) (0.92-1.33)
1.06 0.98 1.03 TRIM24 rs6467778 7 G/A 0.759 0.651 0.831 0.890 0.770 3 (0.83-1.35) (0.69-1.39) (0.84-1.26)
0.85 1.00 0.92 SHB-ALDH1B1 rs4271056 9 C/T 0.139 0.295 0.200 0.988 0.446 3 (0.62-1.15) (0.73-1.37) (0.73-1.15)
0.98 1.14 1.05 NAV2-HTATIP2 rs1353649 11 G/A 0.635 0.851 0.577 0.290 0.572 4 (0.79-1.22) (0.89-1.45) (0.89-1.23)
L3MBTL4- 0.99 1.13 1.04 ARHGAP28 rs9303988 18 C/T 0.617 0.910 0.585 0.372 0.631 3 (0.80-1.22) (0.87-1.46) (0.88-1.23)
208
1.19 0.96 1.08 HRH4-ZNF521 rs9955411 18 T/A 0.238 0.175 0.273 0.792 0.409 5 (0.93-1.53) (0.73-1.27) (0.90-1.30)
0.85 1.37 0.92 ZNF432-ZNF841 rs3752120 19 T/C 0.200 0.196 0.055 0.262 0.483 (0.66-1.09) (0.79-2.36) (0.73-1.16) 7 0.86 1.25 0.97 rs3450 19 C/T 0.232 0.227 0.142 0.232 0.739 (0.68-1.10) (0.87-1.78) (0.79-1.18)
0.86 1.23 0.91 rs12460587 19 G/T 0.203 0.224 0.060 0.447 0.443 7 (0.66-1.10) (0.73-2.07) (0.73-1.15)
0.65 0.92 0.77 ELMO2-ZNF334 rs279728 20 T/C 0.104 0.013 0.214 0.613 0.037 3 (0.46-0.91) (0.66-1.28) (0.61-0.98)
a Chromosome; b Freq.: Frequency of the effect allele; c OR: Odds ratio for the effect alleles. A1: Effect allele; A2: Non-effect allele; CI: Confidence Interval. Citations: 1. Park TJ, Park JS, Cheong HS, Park BL, Kim LH, Heo JS et al. Genome-wide association study identifies ALLC polymorphisms correlated with FEV1 change by corticosteroid. Clin Chim Acta 2014; 436:20-26. 2. Park HW, Dahlin A, Tse S, Duan QL, Schuemann B, Martinez FD et al. Genetic predictors associated with improvement of asthma symptoms in response to inhaled corticosteroids. J Allergy Clin Immunol 2014; 133:664-9 e5. 3. Dahlin A, Denny J, Roden DM, Brilliant MH, Ingram C, Kitchner TE et al. CMTR1 is associated with increased asthma exacerbations in patients taking inhaled corticosteroids. Immun Inflamm Dis 2015; 3:350-359. 4. Wang Y, Tong C, Wang Z, Mauger D, Tantisira KG, Israel E et al. Pharmacodynamic genome-wide association study identifies new responsive loci for glucocorticoid intervention in asthma. Pharmacogenomics J 2015; 15:422-429. 5. Tantisira KG, Damask A, Szefler SJ, Schuemann B, Markezich A, Su J et al. Genome-wide association identifies the T gene as a novel asthma pharmacogenetic locus. Am J Respir Crit Care Med 2012; 185:1286-1291. 6. Tantisira KG, Lasky-Su J, Harada M, Murphy A, Litonjua AA, Himes BE et al. Genomewide association between GLCCI1 and response to glucocorticoid therapy in
asthma. N Engl J Med 2011; 365:1173-1183.
209
209
210 Table S5. Top signals of replication at gene level in African-admixed populations within 100 kb of previously reported genes.
Significant Bonferroni #SNPs #Independent SNPs after SNP min Gene p-value A1/A2 OR (95% CI) a p-value tested signals Bonferroni-like p-value threshold correction
rs113903375 ALLC 1197 61 8.23 x 10-4 rs113903375 G/A 2.55 (1.51-4.31) 4.69 x 10-4 rs113903375
FBXL7 1406 53 9.52 x 10-4 NA rs80016637 G/A 2.13 (1.34-3.37) 1.33 x 10-3 FTSJD2 819 156 3.20 x 10-4 NA rs72855423 A/G 0.63 (0.47-0.83) 1.12 x 10-3 MMS22L-FBXL4 4069 158 3.17 x 10-4 NA rs77248643 A/G 2.10 (1.39-3.17) 4.28 x 10-4
PDE10A-T 4173 265 1.88 x 10-4 NA rs519368 C/A 0.67 (0.54-0.84) 5.16 x 10-4 UMAD1-GLCCI1 2941 292 1.71 x 10-4 NA rs11978146 C/T 0.73 (0.60-0.88) 1.40 x 10-3 MAGI2 6171 196 2.55 x 10-4 NA rs75174008 T/C 0.53 (0.37-0.78) 1.07 x 10-3 TRIM24 891 479 1.04 x 10-4 NA rs79076168 G/A 1.88 (1.15-3.10) 0.013 SHB-ALDH1B1 2205 254 1.97 x 10-4 NA rs113593997 C/T 0.43 (0.25-0.74) 2.31 x 10-3 NAV2-HTATIP2 3088 291 1.72 x 10-4 NA rs7126277 G/A 1.37 (1.14-1.64) 9.09 x 10-4 rs62081416 rs61481914 L3MBTL4-ARHGAP28 4181 150 3.33 x 10-4 rs9789132 rs62081416 C/T 2.44 (1.63-3.65) 1.57 x 10-5 rs4337383 rs12604117
HRH4-ZNF521 3301 404 1.24 x 10-4 NA rs8094894 T/A 1.73 (1.30-2.29) 1.77 x 10-4
ZNF432-ZNF841 993 129 3.86 x 10-4 NA rs8107315 T/C 0.80 (0.68-0.95) 0.011
210
rs2425845 ELMO2-ZNF334 826 28 1.79 x 10-3 rs2425845 T/C 1.08 (0.34-3.44) 3.56 x 10-4 rs2425846 a OR: Odds ratio for the effect alleles. A1: Effect allele; A2: Non-effect allele; CI: Confidence Interval; NA: not available. Significant p-values after multiple comparison adjustment are in boldface.
2
11
211
212
Risk factors associated CHAPTER 4 with asthma exacerbations despite ICS use
213
214
CHAPTER 4.1
Risk factors of asthma exacerbations in asthmatic children treated with ICS: is there an added value of genetic risk factors?
Farzan N
Vijverberg SJ
Maitland-van der Zee AH
Manuscript in preparation
215
Abstract
Introduction: Preventing asthma exacerbations is one of the main goals of asthma management guidelines. However, 30-40% of the children continue to suffer from exacerbations despite treatment with inhaled corticosteroids.
Objectives: 1) to identify clinical, environmental and demographic factors independently associated with an increased risk of asthma exacerbations in children who use ICS, 2) to assess whether previously identified genetic or pharmacogenetic markers were associated with an increased risk of asthma exacerbations in children treated with ICS, and 3) to assess the added value of the genetic and pharmacogenetic variants in identifying children with an increased risk of asthma exacerbations compared to non-genetic risk factors solely.
Methods: Asthmatic children with a regular use of ICS (n=645) were selected from the pharmacy-based PACMAN study. Asthma exacerbations were defined as ≥ 1 short course of OCS use and/or emergency room visit (ER) in the past 12 months. The association between asthma exacerbations and non-genetic/genetic variants was assessed using logistic regression models. All genetic markers were analysed assuming additive genetic models.
Results: Of the 645 patients who used ICS, 67 (10.7%) reported at least one exacerbation in the past 12 months. In total seven non-genomics risk factors were assessed, of which two (age and food allergy) were independently associated with an increased risk of asthma exacerbations in children treated with ICS. In total, 49.3% of the children had a parent-reported food allergy. Patients with food allergy had approximately 2-fold higher risk of exacerbations compared to the children without food allergy (adjusted Odds Ratio (adjOR) = 2.05, 95% confidence interval (CI): 1.20-3.49= 0.008). A weak but significant association was found between age and the risk of exacerbations. Each additional year of age was associated with 14% decrease in the risk of asthma exacerbations (adjOR =0.86, 95%CI: 0.76-0.97, p= 0.01). None of the previously published genetic (seven variants) or pharmacogenetic (eight variants) variants were associated with asthma exacerbations in our study population.
Conclusion: A diagnosis of food allergy and younger age were associated with an increased risk of exacerbations in asthmatic children using ICS. None of the studied (pharmaco)genetic factors were independently associated with an increased risk of
216 exacerbations. Asthmatic children who also suffer from food allergy might benefit from a more intensive disease monitoring program to decrease the risk of exacerbations.
217
Introduction
Inhaled corticosteroids (ICS) are the most efficacious and commonly prescribed medications in persistent childhood asthma (1). ICS reduce airway inflammation by 1) suppressing transcription of the genes encoding for pro-inflammatory cytokines, chemokines, and enzymes and 2) increasing transcription of the anti-inflammatory genes (2). Glucocorticoid molecules exert these actions by binding to the glucocorticoid receptors (GR) in the cytosol. This complex subsequently translocates to the nucleus of the cell (2). In addition to these genomic mechanisms, ICS reduce inflammation through non-genomic mechanisms which have more rapid effects compared to the genomic mechanisms (3). In vitro studies have shown that ICS interact with endothelium smooth muscle cell membranes which subsequently changes the physicochemical properties of the membrane (4). This interaction inhibits the uptake of norepinephrine in smooth muscle cells (5) which subsequently results in vasoconstriction and a decrease in the infiltration of peripheral inflammatory cells into the lungs (6).
Despite these broad effects, 30-40% of the children continue to experience asthma exacerbations despite using ICS (7,8). According to the Global Initiative for Asthma (GINA) guideline, asthma exacerbations are a progressive worsening of asthma symptoms (i.e. chest tightness, cough, and shortness of breath) that require an urgent change in treatment to prevent a severe event such as hospitalisation, mechanical ventilation or in rare cases death (1). Exacerbations are one of the main causes of emotional stress in asthmatic children and their parents/caregivers, and they are the major cause of asthma associated morbidity and mortality (9). Various factors can increase the risk of exacerbations; the GINA 2017 guideline has identified 17 independent risk factors of asthma exacerbation for both children and adults (1) (Table 1). Furthermore, recent scientific evidence suggests that there might be a genetic susceptibility to asthma exacerbations (10–12). In addition, pharmacogenetic markers that decrease ICS efficacy might increase the risk of asthma exacerbations.
Considering the complex nature of asthma exacerbations and the corticosteroid signalling pathway, one could speculate that a single marker either genetic or non- genetic can explain only a small part of the inter-individual variability in treatment response. Recent studies show that interactions between different risk factors result in these exaggerated symptoms. Therefore, assessment of an individual’s exacerbation risk based solely on a single factor could disregard the influence of remaining
218 components on this phenotype and prevent capturing the complex relationship between genetic/non-genetic factors and exacerbations in ICS users.
Genetic risk scores are commonly used to capture the joint effect of genetic variations on different diseases and treatment responses (13). Indeed, integrating these scores with the clinical data may help physicians to identify patients at higher risks of asthma exacerbations and opt for the most appropriate interventions for different subgroups of patients. Therefore, in this study, we aimed to 1) assess the association between clinical, environmental and demographic factors and asthma exacerbation in children treated with ICS 2) assess the single and joint effect of the previously identified genetic markers on the risk of asthma exacerbations in children treated with ICS and 3) assess the added value of genetic variants to identify children with an increased risk of asthma exacerbations compared to non-genetic risk factors solely.
Methods
Study population
In total, 645 asthmatic children with regular use of ICS were selected from the PACMAN study (Figure E2). PACMAN is an observational cohort of 995 children (4- 12 years of age) who are regular users of asthma medications (ATC code R03) (≥3 asthma medication prescriptions within the last two years and ≥1 prescription in the last 6 months). Children were selected from community pharmacies in the Netherlands that belonged to the Utrecht Pharmacy Practice Network for Education and Research (UPPER) (14). During a study visit in a community pharmacy, parents/caregivers filled in questionnaires to collect information regarding demographic characteristics, general health, asthma control and severity, exacerbations in the past year, medication use, environmental and sociodemographic factors (15).
Outcome definition
Asthma exacerbations were defined as ≥ 1 short course of OCS use and/or emergency room visit (ER) due to asthma in the past 12 months based on parental-reported questionnaires.
219
Risk factors associated with asthma exacerbations
Non-genetic risk factors for asthma exacerbations were identified from the GINA 2017 report (16). From 17 independent risk factors in GINA 2017, seven were available in our dataset (Table 1). In addition to the risk factors in GINA 2017, several studies have reported ‘age’ to be a risk factor specific for asthma exacerbations in children (17–20). Therefore, we also assessed whether ‘age’ was a risk factor for exacerbations in our study population. In total, the following risk factors were assessed in this study: age, exposure to tobacco smoke, food allergy, higher Body Mass Index (BMI), poor inhalation technique, poor adherence to medication, and lower parents’ education.
Table 1. Non-genetic risk factors of asthma exacerbations identified by literature and assessed in our study
GINA guideline 2017 Data available in our study population
1 Exposure to tobacco smoke Self-reported current smoking status of parents/caregivers (dichotomous: yes/no)
2 Confirmed food allergy Parental/self-reported allergy to food (dichotomous: yes/no) 3 Obesity Body Mass index (continuous) 4 Incorrect inhalation technique inhalation control checklist score <80 (dichotomous: poor vs good)
5 Poor adherence to medication MARS-5 questionnaire score <21 (dichotomous: poor vs good) 6 Socio-demographic problems Paternal education level (dichotomous: low vs high) maternal education level (dichotomous: low vs high) 7 Uncontrolled asthma symptoms Not available 8 High SABA use Not available
9 In adequate ICS: not prescribed Not relevant
10 Low FEV1 (<60% predicted) Not available
11 Major psychological problem Not available
12 allergen exposure if sensitized Not available 13 Rhinosinusitis Not available
14 Pregnancy Not relevant
15 Sputum/blood eosinophilia Not available
16 Ever intubated or intensive care Not available
17 ≥1 severe exacerbation in last 12 Not available months FEV1, forced expiratory volume-one second; MARS, Medication Adherence Report Scale; SABA, short acting b2 agonists.
220
Selection of genetic risk factors of exacerbations
We performed a literature review to search for scientific evidence regarding the influence of genetic variants on asthma exacerbations. Articles published from 1998 until the end of June 2018 were searched in PubMed using specified search terms (TableS1). We included single nucleotide polymorphisms (SNPs) from candidate gene studies that were successfully replicated at least once in an independent population.
SNPs from GWAS were selected using the online catalog of the National Human Genome Research Institute (NHGRI catalog) (21). In brief, the catalog is a regularly updated list of all published associations between SNPs and human diseases and phenotypes which examined at least 100,000 SNPs. In this study, we included SNPs that had reached genome-wide significance (≤ 10-8).
Selection of pharmacogenetic markers for poor ICS response
Pharmacogenomics markers were selected from previously published candidate gene and GWA studies of ICS response. To select SNPs from candidate gene studies, we used the results of a systematic review that was published by our group (22). The systematic review reports the results of published pharmacogenomics studies of ICS until June 2015. Furthermore, using the same research keywords, we searched for publication between June 2015 and June 2018. We selected the SNPs that were successfully replicated at least once in an independent population. SNPs from the GWAS were selected using the online catalog of the National Human Genome Research Institute (NHGRI catalog) (21).
Imputation of missing non-genetic variables
We performed imputation using iterative chained equations to model missing values of risk factors (23). The imputation was performed in R version 3.4.3 using “mice package”. Details of the imputation are explained in the online repository.
221
Genotyping
DNA was collected using Oragene Saliva kits. Genotyping in PACMAN was performed using Human Core-24 BeadChip Marker information. The SNPs were imputed from the Haplotype Reference Consortium using Minimac3, with a prior phasing step using SHAPEIT, running the two steps with the Michigan Imputation Server available at https://imputationserver.sph.umich.edu/.
Genetic risk score for risk of asthma exacerbations
We aimed to construct three GRS based on the treatment outcomes and study populations (children or adult asthmatics) of the previously published studies (Figure S1); 1) SNPs that were associated with asthma exacerbations and/or asthma symptoms in asthmatic children, 2) SNPs from the first model plus SNPs that were found to be associated with changes in lung function in children, 3) SNPs from the second model plus SNPs that were associated with any of the three outcomes in adult asthmatics.
Statistical analysis
For descriptive statistics, the chi-squared or Student t-test was used to compare demographic and clinical characteristics of the children with and without exacerbations. Logistic regression models were used to assess the association between genetic/non- genetic variables and asthma exacerbations. Odds ratios (OR) and their corresponding 95% confidence intervals (CI) were calculated. Univariate analysis was performed to investigate the association between each variable and the outcome. The significant threshold for univariate analyses was set at p<0.1. Multiple regression was performed to assess the independence of associated factors. All variables were forced into the model and the analysis was adjusted for gender and disease severity. For the multiple regression model, the significance threshold was set at P<0.05. All statistical analysis was performed using SPSS v 24.0. We used a modified form of the British Thoracic Society (BTS) treatment steps as a surrogate for asthma severity (Table S2). All genetic markers were analyzed assuming an additive genetic model. To check the correlations between variables (aka multicollinearity), the variable inflation factor (VIF)
222 was calculated using “car package” in R version 3.4.3. The VIF value higher than 10 indicated multicollinearity.
Table 2. Study population characteristics.
All Patients with Patients without n=627 exacerbations exacerbations n=67 n=560
Age at study visit, mean (±SD)* 8.85 (±2.3) 8.31 (±2.33) 8.91 (±2.28)
Gender, % (n) 61.6% (397) 61.2% (41) 61.6% (345)
BMI, mean (±SD) 16.9 (±2.9) 17.2 (±3.02) 16.91 (±2.87)
Food allergy, % (n)* 49.3% (318) 64.2% (43) 47.5% (266)
Exposure to tobacco smoke, % (n) 8.6% (54) 13.4% (9) 8% (45)
Low dose ICS, % (n) 70.02 (439) 70.1%(47) 70% (392)
ICS plus LABA or LTM, % (n) 24.2 (152) 22.4% (15) 24%(137)
ICS plus LABA plus LTM, % (n) 5.74 (36) 7.5% (5) 5.5 (31)
Poor Inhalation technique, % (n) 14.8% (93) 9% (6) 15.5% (87)
Poor therapy adherence#, % (n) 42.3% (265) 38.8% 42.7%
Low maternal education, % (n) 19.5% (122) 28.4%(19) 18.4% (103)
Low paternal education, % (n) 18.7% (117) 23.9%(16) 18%(101)
Asthma exacerbations, % (n) 10.7% (67)
#Therapy adherence was assessed by the MARS questionnaire. Body Mass Index, BMI; Inhaled corticosteroids, ICS; Long-acting beta2 agonists, LABA; Leukotriene modifier, LTM.
Results
From 645 patients, 67 (10.7%) had at least one exacerbation in the past 12 months. The mean age of the patients was 8.9 (±2.3) years, and 61.6% of the children were boys. (Table 2). Most of the patients were treated with low doses of ICS (BTS treatment step 2).
223
Association between exacerbations and non-genetic variants
From seven non-genetic risk factors we assessed, three (age, food allergy, and maternal education level) were significantly associated with asthma exacerbations in univariate logistic regression analyses (p-value < 0.1) (Table 3). When adding all seven non-genetic variables together with gender and treatment steps as co-variables to the multiple logistic regression models, two out of three variables (age and food allergy) remained significantly associated with asthma exacerbations in children treated with ICS (p-value < 0.05) (Table 2). In total, 49.3% of the children had parent-reported food allergy (Table 3). Patients with food allergy had approximately 2-fold higher risk of exacerbations compared to the children without food allergy (adjusted Odds Ratio (adjOR)= 2.05, 95% confidence interval (CI): 1.20-3.49, p = 0.008). A weak but statistically significant association was found between age and the risk of exacerbations. Each additional year of age was associated with 14% decrease in the risk of asthma exacerbations (adjOR =0.86, 95%CI: 0.76-0.97, p= 0.01).
Table 3. Non-genetic variants and risk of exacerbations
Univariate p-value Multiple p-value analysis regression model OR (95% CI) OR (95% CI)
Age 0.89 (0.80-0.99) 0.04 0.86 (0.76-0.97) 0.01
BMI 1.03 (0.95-1.13) 0.43 1.06 (0.97-1.17) 0.19
Food allergy 1.98 (1.17-3.35) 0.01 2.05 (1.20-3.49) 0.008
Exposure to tobacco 1.78 (0.83-3.18) 0.14 1.5 (0.68-3.59) 0.29 smoke Poor therapy adherence 0.85 (0.51-1.43) 0.54 0.85 (0.50-1.46) 0.56
Poor inhalation 0.53 (0.22-1.27) 0.15 0.52 (0.21-1.27) 0.15 technique Low maternal education 1.76 (0.99-3.11) 0.05 1.63 (0.80-3.30) 0.17 level Low paternal education 1.43 (0.80-2.60) 0.25 1.08 (0.53-2.20) 0.84 level BMI, body mass index; OR, odds ratio.
224
Association between previously identified genetic variants and risk of exacerbations
In total, seven SNPs were extracted from previously published articles; five from GWAS (rs928413 (IL33), rs6871536 (RAD50), rs1558641 (IL1RL1), and rs6967330 (CDHR3) and two from candidate gene studies (rs1805011 (IL4R) and rs4950928 (CH13L1))) (Table 4). The summary statistics of the genetic variations can be found in the supplementary table S3. In the univariate analysis, rs1558641 (within the IL1RL1 gene), was associated with an increased risk of asthma exacerbations (OR= 1.70, 95%CI: 0.93-3.11). However, the SNP did not remain statistically significantly associated with the outcome in the multivariate logistic regression. None of the remaining SNPs was independently associated with asthma exacerbations; therefore, we could not build the GRSs.
225
226
Table 4. Genetic risk factors of exacerbations.
Gene, SNPs Reference Number of patients Outcomes that the SNP was associated
SNPs from candidate gene studies
Cunningham et al. (2011) Caucasian children and Asthma exacerbations: adolescents. School absences, usage of OCS and hospital (BREATHE, n =1,071) admissions over the previous 6 months. CH13L1 rs4950928 Ortega H et al. (2012) African-American adolescents Asthma exacerbations: and adults (n = 322) Hospitalization, OCS, unscheduled urgent care, ≥30% decrease in forced expiratory volume in 1 second (FEV1) from baseline, or ≥30% decrease in morning peak expiratory flow (AM PEF) on any two consecutive days.
Wenzel et al. (2006) White and African-American A history of intubation or stay intensive care unit adults from 2 cohorts: due to asthma exacerbation.
National Jewish (NJ)(n=140) A history of an ER visit or hospitalisation due to IL4R and SARP (n=423) asthma exacerbation. rs1805011
Anderson et al. (2013) African-American, Asian and Acute asthma symptoms that required treatment white adults (n=701) with an oral or parenteral corticosteroid, an ER visit, or hospitalisation
226
SNPs from GWAS
GSDMB Bønnelykke et al. (2014) Caucasian children Asthma exacerbations: rs2305480 (COPSAC , n=1,173) Cumulative risk of asthma exacerbations during the exacerbation IL-33 first 6 years of life defined as: rs928413 OCS or high-dose ICS use for wheezy symptoms RAD50 or acute hospitalization rs6871536
IL1RL1 rs1558641
CDHR3 rs6967330
FEV1; forced expiratory volume, ER; emergency room, ICS; Inhaled corticosteroids. OCS; Oral corticosteroids. PEF; peak expiratory flow.
227
227
Association between exacerbations and pharmacogenetics markers
In total, eight SNPs were identified from previously published studies; four from candidate gene studies (rs37973 (GLCCI1), rs28364072 (FCER2), rs242941 (CRHR1), and rs1876828 (CRHR1)) and four from GWAS (rs10044254 (FBXL7), rs2388639 (LOC728792), rs6924808 (intergenic), and rs1353649 (intergenic) (Table 5). None of these genetic variants was significantly associated with asthma exacerbations in univariate logistic regression analysis when assuming an additive genetic model. Hence, we could not build GRS using these SNPs (Table S4).
228
Table 5. Pharmacogenomics variants included in the analysis.
Gene, SNPs Reference Number of patients Outcomes that the SNP was associated
SNPs from candidate gene studies
Tantisira et al. (2007) African-American and Caucasian Severe exacerbations: children (CAMP, n = 311) Emergency Room visits and/or hospitalizations.
FCER2 Rogers et al. (2009) African-American and Caucasian Lung function measurements: rs28364072 children (CAMP, n = 311) change in FEV1% pred: ≤ 7.5% considered as poor responders
Koster et al. (2011) Caucasian children Exacerbations: Emergency department visits and PACMAN (n = 386) Hospitalization BREATHE (n = 939) CAMP (n = 311) Asthma symptoms: ACQ- scores Respiratory symptoms (wheeze, shortness of breath and cough) Asthma-related sleep disturbances Asthma-related limitations in daily activities Additional (airway) medication use during
the preceding 12 months
229
229
230
Tantisira et al. (2004) Caucasian children and adults: Changes in FEV1% from baseline
CRHR1 Adult study (Adults, n=415) rs242941 CAMP (Children, n=201) rs1876828 ACRN (Adults, n=224)
Mougey et al. (2013) Caucasian children, adolescents, and Lung function: adults (n=65) Changes in FEV1
CRHR1 Rogers et al. (2009) African-American and Caucasian Lung function: rs242941 children (CAMP, n=311) Changes in FEV1%
GLCCI1 Izuhara et al. (2014) Adult, Japanese (n=224) Lung function: rs37973 Annual decline in FEV1 30ml/year or more
Xu et al. (2017) Chinese (n=418) Lung function: Changes in FEV1
230
SNPs from GWAS
FBXL7, H. Park et al. Caucasian Self -reported asthma symptoms based on diary cards. rs10044254 (2014) children (CAMP, n=124)
LOC728792, rs2388639
rs6924808, Wang Y et al. Asthmatic adults (n=120) Changes in FEV1 intergenic (2015)
rs1353649, intergenic
ACQ, asthma control questionnaire; FEV1, forced expiratory volume in one second.
231
231
Discussion
In this study, we aimed to identify non-genetic and genetic risk factors associated with asthma exacerbations in children who use ICS in an observational pharmacy-based cohort. Our study suggests that younger age and the presence of food allergy are two independent risk factors for asthma exacerbations in children treated with ICS.
Several studies have shown that younger children are at higher risk of asthma exacerbations (17–20). In the most recent study including 3776 children (5-12 years of age) from the UK real-world primary care data, increase in age (per year) was associated with 7% decrease in the risk of asthma exacerbations (OR=0.93 per increase in life year, 95%CI: 0.89, 0.97, p=0.001) (18). Compared to our study, exacerbation rates in this population were slightly higher (14% versus 10%). Like in our study, the effect size of the association analysis was small.
The association between asthma exacerbations and food allergy has been shown frequently in previous studies (24–26). In our study, the presence of parent-reported food allergy resulted in a 2-fold increase in the risk of exacerbation. PACMAN included mostly children treated with low dose ICS (step 2). Therefore, our study population mainly consisted of children with mild asthma. In a small matched case-control study (n=57) by Roberts et al., the presence of food allergy was compared between children (1-16 years of age) with and without life-threatening asthma exacerbation. The result of the regression analysis showed that asthmatic children with food allergy had an approximately 6-fold higher risk of ventilation due to life-threatening asthma exacerbations compared to patients without food allergies (24). The results of the study by Roberts et al. suggest that food allergy could also be associated with more severe forms of asthma exacerbations.
232
Co-existence of food allergy and asthma seem to adversely influence the severity of both disorders (27,28). Therefore, one possible explanation for higher rates of asthma exacerbations in children with food allergy could be attributed to the severity of asthma. Furthermore, ingested or inhaled food particles in sensitised patients can result in respiratory symptoms (29). One of the possible mechanisms by which food allergens cause asthma exacerbations is the exposure of mast cells located in the lungs to inhaled food particles (30–32). It has been shown that in spite of dietary avoidance, children with confirmed food allergy experience chronic asthma symptoms when exposed to environmental food allergens (33). Therefore, it is of high importance for parents/caregivers of asthmatic children with food allergies to limit the exposure of their children to aerosolised food allergens to maintain asthma control.
On the other hand, it is well recognised by now that in most cases asthma exacerbations are caused by several risk factors which act in an additive/synergic manner (34). Approximately 85% of asthma exacerbations in asthmatic children are caused by respiratory viral infections of which rhinovirus infection accounts for almost 70% of these episodes (35–37). Recent studies have investigated the potential interaction between viral and allergen triggered immune response to explain severe exacerbations in children (34,38,39). It has been shown that cross-linking between allergen and high-affinity IgE receptor could impair virus-induced type 1 and 3 interferon production in peripheral blood cells (40). Furthermore, IgE sensitisation to food allergens is associated with local and systemic type 2 inflammatory (Th2) markers (41). There is also evidence that Th2 cytokines (IL-4 and IL-13), impair RV-induced interferons in epithelial cells (42). Therefore, allergic sensitisation could result in deficient anti-viral responses in immune and airway epithelial cells (40,42). It seems that food allergens in sensitised children not only can increase the risk of exacerbations but might also increase the risk of respiratory viral infections.
Although the association between parent/caregiver’s education and asthma exacerbations was not significant after adjusting for other factors, there was a trend towards a higher risk of exacerbations in children who had parents (especially mothers) with a lower education level. Previous studies have underlined the importance of maternal education in asthma control; poor knowledge of asthma medication and environmental risk factors results in uncontrolled asthma (43,44). Therefore, an increase in parent/caregivers’ perception of asthma and its symptoms and more
233
effective communication with healthcare professionals can improve disease control. We did not find an association between other previously identified risk factors.
The number of children with poor inhalation technique was relatively low (14.8%). Interestingly, the number of patients with poor inhalation technique and poor therapy adherence was slightly higher in the patients that did not have an exacerbation in the previous year (Table 3). The higher percentage of these variables is presumably due to the time of their evaluation which was during the study visit. It is possible that patients who had experienced an exacerbation became more compliant with their asthma medications.
A relatively low number of the children (8%) was exposed to tobacco smoke and the number of children who were exposed to tobacco smoke was slightly higher in the exacerbation group (13.4%) compared to exacerbation-free group (8%). However, this difference was not statistically significant. Although not statistically significant, the mean BMI was slightly higher in patients with exacerbations (17.2 (±3.02)) compared to the exacerbation-free group (16.9 (±2.87)).
From the (pharmaco)genetic markers that we assessed, none was significantly associated with asthma exacerbations in the multiple regression analysis. However, one SNP within the IL1RL1 gene was associated with the outcome in the univariate analysis. IL1RL1, a gene consistently found to be associated with asthma onset in GWA studies (45,46), is a member of the interleukin 1 receptor family and it is expressed by inflammatory cells located in the lung (47).
One of the reasons that could explain the lack of association between genetic variations and asthma exacerbations is the low number of patients with exacerbations (n=67, 10.7%) in the PACMAN cohort. Also, considering the low minor allele frequency of the most SNPs (ranged between 12%-27%), lack of power could be the reason why we did not find any significant associations in our study. Furthermore, seven out of eight pharmacogenomics SNPs in our study were associated with treatment responses (asthma symptoms or lung function measurements) other than exacerbations (Table 4). These outcomes might reflect distinct underlying molecular/cellular mechanisms of the disease (17); therefore, a significant association with one treatment outcome does not guarantee an association with other outcomes. However, most pharmacogenomic studies have focused on outcomes other than exacerbations and thereby we could include only one SNP (rs28364072 within FCER2) that was found to be associated
234 with asthma exacerbations (22). In addition to the treatment outcome, several reasons might explain why there were no significant associations between the genetic markers and exacerbations: First, two SNPs (rs1353649 and rs6924808) from the GWAS of response to ICS and one SNP (rs1805011) from the candidate gene studies of asthma exacerbations were identified in adult asthmatics with no further investigation in children. The differences between the pathophysiology of childhood asthma and adult- onset asthma have been emphasised previously (48). Therefore, the difference in the study population could be the reason why these SNPs were not associated with asthma exacerbations in our study.
On the other hand, except the GWAS of asthma exacerbations by Bønnelykke et al. (n=1,173) (12), GWAS of asthma treatment response had relatively small sample sizes (<150 patients) (49,50) which suggests the overestimation of the findings by these studies. In addition, unlike SNPs from the candidate-gene studies, which are selected based on prior biological knowledge, the function of SNPs identified in GWAS is usually unknown. Experimental in vitro or in-silico studies after GWAS can generate insights into the function of the identified SNPs.
The PACMAN cohort includes children with mild to severe asthma and therefore, it reflects the childhood asthma population in the Netherlands. Observational studies such as PACMAN generate important insights into ‘real-world’ settings. For example, most studies of asthma exacerbations mainly include patients with severe asthma. However, as we showed in this study, patients on lower treatment steps can experience occasional sudden exacerbations as well. Therefore, it is of utmost importance to assess the risk of this life-threatening event in patients with all forms of asthma. Our study has several limitations. First, we did not have available information regarding previous exacerbation, as they are the strongest risk factor of future exacerbations. The negative influence of previous exacerbations has been consistently reported in studies investigating the risk of exacerbations. Second, the estimated prevalence of challenged-proved food allergy in children ranges between 1% and 10.8% (51). Using self/parent-reported data, we might have overestimated the prevalence of food allergy.
In our study, none of the genetic and pharmacogenomics markers was associated with asthma exacerbations. However, there might be a more complex relationship between genetic variants and asthma exacerbations. It has been shown that interaction between genetic markers and environmental factors could result in asthma
235
exacerbations. For example, in a study by Sharma et al. (52), researchers identified several SNPs in the TGFB1 gene to be associated with asthma exacerbations only in children exposed to house dust mite. Interestingly, the researchers in this study, could not find any associations between these SNPs and exacerbations in children who were not exposed to house dust mite. Therefore, integration of the data produced by genomics technologies with clinical, environmental and demographic factors has the potential to improve diagnosis and disease management (53,54). Hopefully, these efforts will lead to new biomarker discoveries that can help researchers and clinicians select the most appropriate therapy for different subgroups of patients and thereby, bring precision medicine to daily clinical practice.
In conclusion, we identified young age and presence of food allergy to be the two clinical/demographic risk factors of asthma exacerbations in asthmatic children despite treatment with ICS. Considering the strong association between the food allergy and exacerbations, we hypothesize that asthmatic children with a food allergy might benefit from multidisciplinary care with allergologists and respiratory physicians for better management of both asthma and food allergy.
236
References
1. Global Initiative for Asthma. Global Risk in Asthma Management. Allergy, Strategy for Asthma Management and Asthma Immunol Res. 2011;3(4):217. Prevention 2018. www.ginasthma.org. 10. Anderson WH, Koshy BT, Huang L, 2. Hossny E, Rosario N, Lee BW, Singh M, Mosteller M, Stinnett SW, Condreay LD, et El-Ghoneimy D, SOH JY, et al. The use of al. Genetic analysis of asthma exacer- inhaled corticosteroids in pediatric asthma: bations. Ann Allergy, Asthma Immunol. update. World Allergy Organ J. 2016;9(1): 2013;110(6):416–422.e2. 26. 11. Rumpel JA, Ahmedani BK, Peterson 3. Borski RJ. Nongenomic membrane EL, Wells KE, Yang M, Levin AM, et al. actions of glucocorticoids in vertebrates. Genetic ancestry and its association with Trends Endocrinol Metab. 2000;11(10): asthma exacerbations among African 427–36. American subjects with asthma. J Allergy Clin Immunol. 2012;130(6):1302–6. 4. Horvath G, Wanner A. Inhaled cortico- steroids: effects on the airway vasculature 12. Bønnelykke K, Sleiman P, Nielsen K, in bronchial asthma. Eur Respir J. 2006;27 Kreiner-Møller E, Mercader JM, Belgrave (1):172–87. D, et al. A genome-wide association study identifies CDHR3 as a susceptibility locus 5. Horvath G, Lieb T, Conner GE, Salathe for early childhood asthma with severe M, Wanner A. Steroid sensitivity of nor- exacerbations. Nat Genet. 2014;46(1):51– epinephrine uptake by human bronchial 5. arterial and rabbit aortic smooth muscle cells. Am J Respir Cell Mol Biol. 2001;25(4) 13. Cooke Bailey JN, Igo RP. Genetic Risk 500–6. Scores. In: Current Protocols in Human Genetics. Hoboken, NJ, USA: John Wiley & 6. Kumar SD, Brieva JL, Danta I, Wanner Sons, Inc. 2016. p. 1.29.1-1.29.9. A. Transient effect of inhaled fluticasone on airway mucosal blood flow in subjects with 14. Koster ES, Blom L, Philbert D, Rump W, and without asthma. Am J Respir Crit Care Bouvy ML. The Utrecht Pharmacy Practice Med. 2000;161(3 Pt 1):918–21. network for Education and Research: a network of community and hospital 7. Szefler SJ, Martin RJ, King TS, Boushey pharmacies in the Netherlands. Int J Clin HA, Cherniack RM, Chinchilli VM, et al. Pharm. 2014;36(4):669–74. Significant variability in response to inhaled corticosteroids for persistent asthma. J 15. Koster ES, Raaijmakers JA, Allergy Clin Immunol. 2002;109(3):410–8. Koppelman GH, Postma DS, van der Ent CK, Koenderman L et al. Pharmaco- 8. Szefler SJ, Phillips BR, Martinez FD, genetics of anti-inflammatory treatment in Chinchilli V.M., Lemanske R.F., Strunk children with asthma: rationale and design R.C., et al. Characterization of within- of the PACMAN cohort. Pharmaco- subject responses to fluticasone and genomics. 2009;10(8):1351–61. montelukast in childhood asthma. J Allergy Clin Immunol. 2005;115(2):233–42. 16. Global strategy for asthma manage- ment and prevention. Global Initiative for 9. Sims EJ, Price D, Haughney J, Ryan D, Asthma (GINA); 2017. Available at: Thomas M. Current Control and Future http://ginasthma.org/.
237
17. Wu AC, Tantisira K, Li L, Schuemann 25. Mitchell I, Tough SC, Semple LK, Green B, Weiss ST, Fuhlbrigge AL. Predictors of FH, Hessel PA. Near-fatal asthma: a symptoms are different from predictors of population-based study of risk factors. severe exacerbations from asthma in Chest. 2002;121(5):1407–13. children. Chest. 2011;140(1):100–7. 26. Vogel NM, Katz HT, Lopez R, Lang DM. 18. Turner SW, Murray C, Thomas M, Food Allergy Is Associated With Potentially Burden A, Price DB. Applying UK real-world Fatal Childhood Asthma. J Asthma. 2008; primary care data to predict asthma attacks 45(10):862–6. in 3776 well-characterised children: a retro- spective cohort study. NPJ Prim care 27. Friedlander JL, Sheehan WJ, Baxi SN, Respir Med. 2018;28(1):28. Kopel LS, Gaffin JM, Ozonoff A, et al. Food Allergy and Increased Asthma Morbidity in 19. Heymann PW, Carper HT, Murphy DD, a School-Based Inner-City Asthma Study. J Platts-Mills TA., Patrie J, McLaughlin AP, et Allergy Clin Immunol Pract. 2013;1(5):479– al. Viral infections in relation to age, atopy, 84. and season of admission among children hospitalized for wheezing. J Allergy Clin 28. Arabkhazaeli A, Vijverberg SJH, van Immunol. 2004;114(2):239–47. Erp FC, Raaijmakers JAM, van der Ent CK, Maitland van der Zee AH. Characteristics and severity of asthma in children with and 20. Silverman RA, Stevenson L, Hastings without atopic conditions: a cross-sectional HM. Age-related seasonal patterns of study. BMC Pediatr. 2015;15(1):172. emergency department visits for acute asthma in an urban environment. Ann 29. James JM. Respiratory manifestations Emerg Med. 2003;42(4):577–86. of food allergy. Pediatrics. 2003;111(6 Pt 3):1625–30. 21. MacArthur J, Bowler E, Cerezo M, Gil L, 30. Caffarelli C, Garrubba M, Greco C, Hall P, Hastings E, et al. The new NHGRI- Mastrorilli C, Povesi Dascola C. Asthma EBI Catalog of published genome-wide and Food Allergy in Children: Is There a association studies (GWAS Catalog). Connection or Interaction? Front Pediatr. Nucleic Acids Res. Oxford University Press 2016;4:34. 2017;45(D1):D896–901. 31. Roberts G, Lack G. Food allergy and 22. Farzan N, Vijverberg SJH, Arets HG, asthma--what is the link? Paediatr Respir Rev. 2003;4(3):205–12. Raaijmakers JAM, Maitland-van der Zee
AH. Pharmacogenomics of inhaled cortico- 32. Leonardi S, Pecoraro R, Filippelli M, steroids and leukotriene modifiers: A Miraglia del Giudice M, Marseglia G, systematic review. Clin Exp Allergy J. 2017; Salpietro C, et al. Allergic reactions to foods 47(2):271–93. by inhalation in children. Allergy Asthma Proc. 2014;35(4):288–94.
23. White I.R., Royston P, Wood A.M.. 33. Roberts G, Golder N, Lack G. Bronchial Multiple imputation using chained equ- challenges with aerosolized food in ations: Issues and guidance for practice. asthmatic, food-allergic children. Allergy. Stat Med. 2011;30(4):377–99. 2002;57(8):713–7.
24. Roberts G, Patel N, Levi-Schaffer F, 34. Green RM, Custovic A, Sanderson G, Habibi P, Lack G. Food allergy as a risk Hunter J, Johnston SL, Woodcock A. Synergism between allergens and viruses factor for life-threatening asthma in child- and risk of hospital admission with asthma: hood: a case-controlled study. J Allergy case-control study. BMJ. 2002;324(7340): Clin Immunol. 2003;112(1):168–74. 763.
238
35. Papadopoulos NG, Christodoulou I, 44. Peterson-Sweeney K, McMullen A, Rohde G, Agache I, Almqvist C, Bruno A, et Yoos HL, Kitzman H. Parental perceptions al. Viruses and bacteria in acute asthma of their child’s asthma: Management and exacerbations - A GA2LEN-DARE* sys- medication use. J Pediatr Heal Care. 2003; tematic review. Allergy. 2011; 66(4): 458– 17(3):118–25. 68. 45. Moffatt MF, Gut IG, Demenais F, 36. Friedlander SL, Busse WW. The role of Strachan DP, Bouzigon E, Heath S et al. A rhinovirus in asthma exacerbations. J large scale, consortium-based genome Allergy Clin Immunol. 2005;116(2):267–73. wide association study of asthma. N Engl J Med. 2010;363(13):1211–21. 37. Johnston SL, Pattemore PK, Sanderson G, Smith S, Lampe F, Josephs 46. Torgerson DG, Ampleford EJ, Chiu GY, L, et al. Community study of role of viral Gauderman W.J., Gignoux C.R., Graves infections in exacerbations of asthma in 9- P.E., et al. Meta-analysis of genome-wide 11 year old children. BMJ. 1995;310 association studies of asthma in ethnically (6989):1225–9. diverse North American populations. Nat Genet. 2011;43(9):887–92. 38. Murray CS, Poletti G, Kebadze T, Morris J, Woodcock A, Johnston SL, et al. 47. Coyle AJ, Lloyd C, Tian J, Nguyen T, Study of modifiable risk factors for asthma Erikkson C, Wang L, et al. Crucial role of exacerbations: virus infection and allergen the interleukin 1 receptor family member exposure increase the risk of asthma T1/ST2 in T helper cell type 2-mediated hospital admissions in children. Thorax. lung mucosal immune responses. J Exp 2006;61(5):376–82. Med. 1999;190(7):895–902.
39. Holt P.G., Sly P.D. Viral infections and 48. Larsen G.L. Differences between adult atopy in asthma pathogenesis: new and childhood asthma. J Allergy Clin rationales for asthma prevention and Immunol. 2000;106(3 Suppl):S153-7. treatment. Nat Med. 2012;18(5):726–35. 49. Wang Y, Tong C, Wang Z, Wang Z, 40. Durrani SR, Montville DJ, Pratt AS, Mauger D, Tantisira K.G., et al. Pharmaco- Sahu S, DeVries MK, Rajamanickam V, et dynamic genome-wide association study al. Innate immune responses to rhinovirus identifies new responsive loci for gluco- are reduced by the high-affinity IgE corticoid intervention in asthma. Pharmaco- receptor in allergic asthmatic children. J genomics J. 2015;15(5):422–9. Allergy Clin Immunol. 2012;130(2):489–95. 50. Park H-W, Dahlin A, Tse S, Duan Q., 41. Patelis A, Alving K, Middelveld R, Schuemann B, Martinez FD, et al. Genetic James A, Ono J, Ohta S, et al. IgE predictors associated with improvement of sensitization to food allergens and airborne asthma symptoms in response to inhaled allergens in relation to biomarkers of type 2 corticosteroids. J Allergy Clin Immunol. inflammation in asthma. Clin Exp Allergy. 2014;133(3):664–669.e5. 2018;48(9):1147–54. 51. Rona RJ, Keil T, Summers C, Gislason 42. Contoli M, Ito K, Padovani A, Poletti D, D, Zuidmeer L, Sodergren E, et al. The Marku B, Edwards MR, et al. Th2 cytokines prevalence of food allergy: A meta- impair innate immune responses to analysis. J Allergy Clin Immunol. 2007; 120 rhinovirus in respiratory epithelial cells. (3):638–46. Allergy. 2015;70(8):910–20. 52. Sharma S, Raby BA, Hunninghake GM, 43. Koster ES, Wijga AH, Koppelman GH, Soto-Quirós M, Avila L, Murphy AJ, et al. Postma DS, Brunekreef B, De Jongste JC, Variants in TGFB1, dust mite exposure, et al. Uncontrolled asthma at age 8: the and disease severity in children with importance of parental perception towards asthma. Am J Respir Crit Care Med. 2009; medication. Pediatr Allergy Immunol. 2011; 179(5):356–62 22(5):462–8.
239
53. Farzan N, Vijverberg SJ, Kabesch M, 54. Wang B, Mezlini AM, Demir F, Fiume M, Sterk PJ, Maitland-van der Zee AH. The Tu Z, Brudno M, et al. Similarity network use of pharmacogenomics, epigenomics, fusion for aggregating data types on a and transcriptomics to improve childhood genomic scale. Nat Methods. 2014;11(3): asthma management: Where do we stand? 333–7 Pediatr Pulmonol. 2018; 53(6):836-845.
240
Online Supplementary
Risk factors of asthma exacerbations in asthmatic children treated with ICS: is there an added value of genetic risk factors?
241
Definition of non-genetic Variables
Therapy adherence
Parental reported therapy adherence was collected using the 5-item Medication Adherence Report Scale (MARS) questionnaire (1). The MARS contains questions on forgetting to take a dose, altering the dose, deciding to miss a dose, taking less medication then instructed by the physician and deciding not to take medication for a while. Previous research showed that this questionnaire was also a satisfactory screening tool to identify non-adherent ICS users within Dutch community pharmacies. In this study, we dichotomized patients using the cut-off point of 21 points. Patients with a MARS score ≥21 were considered to be highly adherent.
Inhalation technique
Patient’s inhalation technique was evaluated and scored with an inhaler specific inhalation control checklist that was developed by van der Palen et al (2). Patient’s inhalation technique was considered to be good if the checklist score was ≥80%.
Parental educational level
Education level was categorized as high/intermediate (higher vocational or university education, higher secondary education) versus low (no formal education, lower secondary education or intermediate secondary education) (3).
242
Figure S1. Genetic variant inclusion. We aimed to develop three GRS based on the treatment outcome and study populations of the previously published studies (figure S1): 1. The first model included SNPs that were associated with asthma exacerbations and/or asthma symptoms in asthmatic children 2. The second model included SNPs from the first model and SNPs that were associated with changes in lung function in children. 3. The third model included SNPs from the second model and SNPs that were associated with these three outcomes in adult asthmatics.
Table S1. PubMed search terms.
(("asthma"[MeSH Terms] OR "asthma"[All Fields])) AND ((emergenc[All Fields] OR emergency[All Fields] OR emergency'[All Fields] OR emergencyacceptance[All Fields] OR emergencyadmissions[All Fields] OR emergencyand[All Fields] OR emergencydepartment[All Fields] OR emergencymedicine[All Fields] OR emergencyroom[All Fields]) OR (("hospitals"[MeSH Terms] OR "hospitals"[All Fields] OR "hospital"[All Fields]) AND admission[All Fields]) OR (hospitalization[All Fields] OR hospitalization'[All Fields] OR hospitalization's[All Fields] OR hospitalizationall[All Fields] OR hospitalizationin[All Fields] OR hospitalizationis[All Fields] OR hospitalizations[All Fields] OR hospitalizations'[All Fields]) OR (attack[All Fields] OR attack'[All Fields] OR attack''[All Fields] OR attack's[All Fields] OR attackes[All Fields] OR attackfree[All Fields] OR attackof[All Fields] OR attacks[All Fields] OR attacks'[All Fields] OR attacks''[All Fields]) OR (exacerbat[All Fields] OR exacerbatated[All Fields] OR exacerbate[All Fields] OR exacerbated[All Fields] OR exacerbated'[All Fields] OR exacerbater[All Fields] OR exacerbates[All Fields] OR exacerbatin[All Fields] OR exacerbating[All Fields] OR exacerbatioja[All Fields] OR exacerbation[All Fields] OR exacerbation'[All Fields] OR exacerbation's[All Fields] OR exacerbational[All Fields] OR
243
exacerbationof[All Fields] OR exacerbationrelated[All Fields] OR exacerbations[All Fields] OR exacerbations'[All Fields] OR exacerbative[All Fields] OR exacerbator[All Fields] OR exacerbator'[All Fields] OR exacerbators[All Fields] OR exacerbators'[All Fields] OR exacerbatory[All Fields]))
AND
(("polymorphism, genetic"[MeSH Terms] OR ("polymorphism"[All Fields] AND "genetic"[All Fields]) OR "genetic polymorphism"[All Fields] OR "polymorphism"[All Fields]))
Statistical analysis
Disease severity
British Treatment Society (BTS) guidelines (4) were used to define different treatment steps as a surrogate of disease severity: step 0: no use of inhaled short-acting beta2 agonist (SABA) as needed in the past month, step 1: SABA as needed, step 2: step 1 plus regular ICS, step 3: step 2 plus regular long-acting inhaled beta-2 agonists (LABA) and, step4: step 3 plus oral leukotriene receptor antagonists (Table S2).
Table S2. BTS treatment steps Medication step No SABA as needed 0
SABA as needed 1
ICS low dose + SABA as needed 2
ICS plus LABA or ICS plus LTM + SABA as needed 3
ICS plus LABA plus LTM + SABA as needed 4
BTS, British Thoracic Society; ICS, Inhaled corticosteroids; LABA, long acting beta 2 agonist; LTM, Leukotriene Modifier; SABA, short acting beta 2 agonists
244
Table S3. Percentage of missing values.
Total n = 627 Missing, % (n)
Exposure to tobacco smoke 2.1% (21)
Food allergy 2.8% (28)
Inhalation technique 7.5% (75)
Therapy adherence 10.2% (101)
Paternal education 7.2%(72)
Maternal education 3.6% (36)
BMI 34.8% (364)
BMI, body mass index
Handling of missing data
From all variables, only age and gender did not have any missing values. After excluding 114 non-Dutch patients, imputation was performed on the remaining 881 children (BTS 0 to 4) using iterative chained equations using all non-genetic variables and gender to replace missing values. The percentage of missing values for each variable is shown in the table. Imputation was performed in R version 3.4.3 using ‘mice package’ (5).
After the imputation, patients on < BTS step 2, patients without genotyping data and patients with missing exacerbation data were excluded. In total, 627 asthmatic children were included in the final model.
245
Figure S2. Flowchart of the study population
Association between single nucleotide polymorphisms (SNPs) and asthma exacerbations:
Table S4. SNPs and risk of exacerbations
Genetic markers
rsID Gene MAF, allele OR (95%CI) P value
rs4950928 CH13L1 0.20, G 1.15 (0.72-1.84) 0.6
rs1805011 IL4R 0.12, C 0.84 (0.48-1.48) 0.5 rs2305480 GSDMB 0.34, A 0.75 (0.51-1.12) 0.16
rs928413 IL-33 0.22, A 1.02 (0.61-1.7) 0.9
rs6871536 RAD50 0.27, C 1.14 (0.76-1.72) 0.5
rs1558641 IL1RL1 0.15, A 0.59 (0.32-1.07) 0.08
rs6967330 CDHR3 0.19, A 0.94 (0.59-1.51) 0.8
Pharmacogenetic markers rs28364072 FCER2 0.29, G 1.31 (0.91-1.88) 0.15
rs10044254 FBXL7 0.19, G 0.88 (0.59-1.40) 0.6
rs2388639 LOC728792 0.19, A 1.02 (0.64-1.63) 0.9
246 rs242941 CRHR1 0.32, A 0.91 (0.62-1.34) 0.6 rs1876828 CRHR1 0.20, T 0.94 (0.59-1.5) 0.8 rs37973 GLCCI1 0.44, G 0.75 (0.52-1.07) 0.11 rs6924808 Intergenic 0.46, G 1.10(0.77-1.59) 0.6 region rs1353649 Intergenic 0.24, A 1.15(0.75-1.77) 0.5 region MAF, minor allele frequency; OR, odds ratio; CI, confidence interval.
247
References
1. Horne R, Weinman J. Patients’ beliefs 4. Levy ML, Thomas M, Small I, Pearce L, about prescribed medicines and their role in Pinnock H, Stephenson P. Summary of the adherence to treatment in chronic physical 2008 BTS/SIGN British Guideline on the illness. J Psychosom Res. 1999;47(6):555– management of asthma. Prim Care Respir J. 67. 2009;18 Suppl 1:S1-16
2. van der Palen J, Klein JJ, Kerkhoff AH, 5. Buuren S van, Karin G.O. mice: Multi- van Herwaarden CL, Seydel ER. variate Imputation by Chained Equations in Evaluation of the long-term effectiveness of R. J Stat Softw. 2010;pp. 1-68. three instruction modes for inhaling medicines. Patient Educ Couns. 1997;32(1 Suppl):S87-95.
3. Koster ES, Wijga AH, Koppelman GH, Postma DS, Brunekreef B, De Jongste JC, et al. Uncontrolled asthma at age 8: the importance of parental perception towards medication. Pediatr Allergy Immunol. 2011; 22(5):462–8.
248
249
250
Cost-effectiveness of CHAPTER 5 pharmacogenetic-guided treatment
251
252
CHAPTER 5.1
Cost-effectiveness of genotyping before starting LABA therapy in asthmatic children and young adults
Farzan N van den Heuvel J.M
Vijverberg S.J
Maitland van der Zee AH
Hövels A
Manuscript in preparation
253
Abstract
Background: Current evidence shows that a genetic variant in the gene encoding the beta-2 adrenergic receptor (ADRB2 rs1042713), is associated with poor treatment response to long-acting beta2 agonists in childhood asthma populations. Children with this variant might benefit from alternative treatment. Therefore, genotype-guided LABA treatment could lead to a better management of asthma in children and increase the quality of life of the patients and their caregivers. Before the implementation of a new technique or test in clinical practice, its cost-effectiveness should be assessed.
Objective: Objective of this study is to perform an economic evaluation of initiating a genetic screening program for ADRB2-rs1042713 in asthmatic children before start with LABA (‘ADRB2-guided treatment’) as compared to the standard treatment involving no genetic testing.
Methods: Using a decision-analytic Markov model, cost-effectiveness of ADRB2- guided therapy versus standard care was evaluated. The target population was children and adolescents (6-18 years of age) with uncontrolled asthma despite low dose ICS use (step 2 treatment) who needed a step up in their current treatment. The effect measure in the analysis was avoided asthma exacerbations. The economic evaluation was performed from the societal perspective in the Netherlands and a time horizon of two years was considered. Furthermore, to assess the combined impact of uncertainties about input parameters on the estimated cost-effectiveness of genotyping a Probabilistic sensitivity analysis (PSA) using Monte Carlo simulation was perfumed.
Results: ADRB2-guided treatment is associated with an expected cost per patient of 1782 Euros, while the current standard of care costs 1934 Euros. Genotyping results in 0.35 expected exacerbations per patient over 2 years, while the standard care results in 0.68 expected exacerbations per patient over 2 years. Since genotyping results in cost-savings and additional avoided exacerbations, genotyping dominates the current standard of care. Probabilistic sensitivity analysis confirmed this by showing a proportion of 60.7% in the southwest quadrant indicating a high probability of both cost-savings and patient benefit in terms of avoided exacerbations.
254
Conclusion: This cost-effectiveness study showed that genotype-guided LABA treatment is an economically attractive strategy in children with asthma since a large group of patients might benefit from it while saving costs.
255
Introduction
Asthma is the most common chronic disease in children that requires a long-term and sometimes life-long treatment with asthma medications (1). Childhood asthma not only has a substantial impact on the health and quality life of the patients, but it also imposes large healthcare expenditures for society (2). In 2005, the societal economic burden of childhood asthma in 25 countries of the European Union was estimated to be three billion euros with the Netherlands contributing approximately 106 million euros (3). Of these costs, approximately half of the direct cost of asthma was related to asthma exacerbations leading to emergency room visits and hospitalizations (4–6). Asthma exacerbations are worsening of symptoms that require urgent actions such as an increase in rescue medications, and they can be life-threatening if remain untreated (7,8). Exacerbations impose a significant financial burden on both caregivers and society (4) and are one of the main causes of emotional stress in asthmatic children and their caregivers (7,8). Therefore, disease control and subsequent prevention of asthma exacerbations provide a significant scope for cost reduction. Clinical asthma guidelines suggest a step-wise treatment approach to identify the most appropriate treatment option for each patient based on their disease severity and future risk of exacerbations (7). A low dosage of inhaled corticosteroids (ICS) is the first line therapy in children with persistent asthma that suppresses inflammation and prevents exacerbations (9,10). Guidelines recommend three step-up options when a child’s asthma remains uncontrolled despite low dose ICS: 1) increasing the dose of ICS, 2) adding a long-acting beta2 agonist (LABA) or 3) adding a leukotriene modifier (LTM) to the existing ICS. Despite these effective treatments, there is an inter-individual variability in response to treatment (11). The first step to achieving asthma control is to improve treatment adherence and inhalation technique (1). However, even in clinical trials with closely monitored treatment adherence, some patients remain uncontrolled despite regular use of medication (11). Approximately 60-80% of the inter-individual variability in treatment response has been attributed to genetic variations (12). To date, more than 80 pharmacogenomics studies have investigated the influence of genetic variations on response to three different asthma medications (ICS, LTM, SABA and LABA) (13). Pharmacogenomics in childhood asthma has not reached clinical practice yet mainly due to inconsistent results between studies and limitations in the study designs (13–15). To date, the most promising results have been shown for pharmacogenomics of LABA (16,17). A genetic variant in the ADRB2 gene, rs1042713, encoding the beta-2 adrenergic receptor, has been positively associated with different
256 measures of poor response in childhood asthma populations (18–20). This variant results in an amino acid substitution at the position 16 (Gly to Arg) in ADRB2 (15).
The largest study assessing the influence of this genetic variant on asthma exacerbations was published in 2016 by Turner et al (15). In this study, a meta-analysis of more than 4000 asthmatic children/adolescents participating in the multi-ethnic Pharmacogenomics in Childhood Asthma (PiCA) consortium showed that patients treated with ICS plus LABA have a 1.5-fold increase in the risk of asthma exacerbations for addition of each copy of the ADRB2 risk allele, while this increased risk was not observed in children treated with ICS solely or ICS plus LTM (15). These findings suggest that asthmatic children carrying this allele could benefit from alternative treatment options such as high dose of ICS or LTM in addition to the current low dose ICS when their asthma is not sufficiently controlled with low dosages of ICS (step 2). Therefore, genotype-guided LABA treatment could lead to a better management of asthma in children and increase the quality of life of the patients and their caregivers.
However, before implementation of a new technique or test in clinical practice, its cost- effectiveness should be assessed. Therefore, the objective of this study is to perform an economic evaluation of initiating a genetic screening program for ADRB2- rs1042713 in asthmatic children as compared to the standard treatment involving no genetic testing to allow for a more personalized treatment approach.
Material and Methods
Overview
A Markov model was developed using Microsoft Excel to investigate the costs and effectiveness of one-time genotyping prior to LABA start versus standard treatment. Standard treatment included the prescription of either higher dose of ICS or low dose ICS plus LABA or LTM without genotyping. The target population was children and adolescents (6-18 years of age) with uncontrolled asthma despite low dose ICS use (step 2 treatment) who needed a step up in their current treatment (21). The effect measure in the analysis was avoided asthma exacerbations. The economic evaluation was performed from the societal perspective in the Netherlands and a time horizon of two years was considered. The analysis is performed according to Dutch guidelines and ISPOR good modeling practices (22).
257
Model structure
The model consisted of a decision tree and five identical Markov traces (Figure 1). The decision tree included two arms: Genotyped group (test group) and Standard treatment group (ST group). In the test arm, patients were stratified by genotype into two different groups. The first group consisted of heterozygotes and homozygotes for the risk allele (A, genotypes: Arg16Arg and Arg16Gly) and received an increased dose of ICS for step 3 treatment (Markov trace 1). The second group consisted of patients homozygous for the protective allele (G, genotype: Gly16Gly) and based on the physician’s preferences, they received either an increased dose of ICS, ICS plus LABA or ICS plus LTM on step 3 (Markov trace 2). We assumed that patients who do not have the risk allele would benefit equally from the three treatment options.
The ST arm subsequently divided into three arms with different treatments options; increased dose of ICS (Markov trace 3), ICS plus LABA (Markov trace 4) or ICS plus LTM (Markov trace 5).
To reflect the dynamic nature of asthma, each Markov trace consisted of five mutually exclusive health states (Figure 2). The model consisted of three exacerbation free health states. These states differed in the severity of asthma, and therefore medical treatments; 1) a health state with patients on treatment step 3, 2) a health state with patients on treatment step 4, and 3) a state with patients on treatment step 5. Patients’ asthma was assumed to be well-controlled while being on either of these health states. In the coming paragraphs, we will refer to the exacerbation-free states by their treatment steps. Depending on the severity of the exacerbations, patients might require urgent actions such as a change in the medication, unscheduled GP or emergency room visit or hospitalization. Therefore, in the model three different exacerbation states were considered; exacerbations that led to 1) hospitalization, 2) emergency room visit (without hospitalization) and 3) exacerbations that led to an unscheduled GP/ pediatrician visit.
The model cycle length for the following was set at one week. At the end of each exacerbation state, patients were assumed to be fully recovered and transitioned to step 4. Furthermore, to avoid unnecessary complexity, we decided not to separate different exacerbation states for patients transitioning from step 4 to step 5. On the other hand, patients who had inadequate asthma control despite step 3 treatment could transition directly from step 3 to step 4 through three monthly scheduled GP
258 visits. Clinical guidelines recommend stepping down in treatment if an adequate asthma control has been maintained for two to three months. Therefore, in the model, patients could have a step down in their treatment, from step 5 to step 4, and from step 4 to step 3, only through three monthly scheduled visits.
Data sources and extraction
Model inputs were based primarily on the data obtained from the Pharmacogenetics of Asthma medication in Children: Medication with Anti-inflammatory effects (PACMAN) cohort (23). PACMAN was initiated in April 2009 as an observational pharmacy-based study. Children (4-12 years of age) with regular use of asthma medications (≥3 prescriptions within the last two years and ≥1 prescription in the last 6 months) were selected from community pharmacies in the Netherlands that were part of the Utrecht Pharmacy Practice Network for Education and Research (UPPER) (23). In the Netherlands, individuals are often registered at one pharmacy, which provides a full record of a patients’ medication use. Records of dispensed asthma medication between birth date and date of extraction (until 2014) were extracted from the computerized pharmacy dispensing systems. In total, 61,127 asthma prescriptions were available for 3573 children from birth to date of pharmacy data extraction. Furthermore, for 995 children, data on demographic characteristics, general health, asthma control and severity, exacerbations in the past year, medication use, environmental and sociodemographic factors were collected using questionnaires. Since children with mild to severe asthma are included in PACMAN, it is a good representative of the Dutch childhood asthma population. The PACMAN study has been approved by the Medical Review Ethics Committee of the University Medical Centre Utrecht.
Transition probabilities
Transition probabilities for one week were calculated using the following formula: -ln (1-probability in one year)/52. Ln refers to the natural logarithm and 52 is the number of weeks in a year. All transition probabilities including exacerbation rates were assumed to be stable for both test and no test group throughout the 2-year period. ADRB2 genotype data were available for 842/995 (84%) patients. From 842, 562
259
(67%) were heterozygote (Arg16Gly) or homozygous (Arg16Arg) for the risk allele / and 280 (33%) were homozygous for the protective allele (Gly16Gly).
Data regarding the emergency room visits and unscheduled physician visits in the past 12 months were extracted from the PACMAN dataset (Table 1), based on parental- reported questionnaires. Since hospitalization data were not available in PACMAN, we used hospital records from the Amsterdam University Medical Center to gather information regarding the hospitalization rates in one year. Approximately, one-third of the asthmatic children who visited the outpatient clinic were hospitalized. Therefore, we assumed that one-third of the PACMAN patients who visited the emergency room were hospitalized.
To keep the model simple, the same transition probabilities were considered for patients moving from step 4 to step 5 in all Markov traces. Therefore, the most determinative part of the model was assumed to be the transitions from step 3 to step 4.
Resource use
Transition probabilities of step up or step down through scheduled visits were extracted from PACMAN prescription data. Any step up or down with a time gap of at least 2 months was assumed as a scheduled visit to the physician’s office (Table 1).
In PACMAN, the risk allele was significantly associated with asthma exacerbations in ICS plus LABA used. However, the same association was not found for asthma symptoms assessed by the Asthma Control Questionnaire (ACQ) (16,17). Therefore, for all Markov traces, we assigned the same transitions probabilities from step 3 to step 4 through three monthly scheduled visits.
Although all three options for step 3 treatment (ICS plus LABA, increased dose of ICS and ICS plus LTMs) are suggested by the guidelines, the prescription pattern of these medications might differ. To obtain this information we used data from the NControl database. NControl dataset contains data of 554 pharmacies, spread across the Netherlands. The total number of public pharmacies in the Netherlands is approximately 1,900. Since 2011, the NControl database contains data related to over 600 million prescriptions and 7.5 million patients. The database contains (not
260 exhaustive) information about the prescriptions, dispensed medication and quantity, dispensing date, prescribed daily dosage, prescriber type and the patient’s age and gender. Patients in the database cannot be identified but can be tracked over time across pharmacies in the database. Prescribers are anonymized and cannot be identified nor tracked over time. NControl is allowed to use these prescription data for research purposes (24). We extracted the type and number of step 3 medications dispensed between 2016-2018 in the Netherlands for children and adolescents (6-18 years of age) who had their first step up from low dose ICS from NControl. We used this time interval to capture the most current asthma prescription patterns in the Netherlands. From 706 patients treated with ICS low dose, 437 (~62%) had switched to ICS plus LABA, 209 (~30%) had an increase in the dose of ICS and 60 patients (~8%) had switched to ICS plus LTM. We assigned these numbers to the no-test arm of the decision tree and to the test arm of the Gly16Gly patients.
Costs
All direct costs within health care were estimated as well as direct costs outside health care. All costs are presented in 2017 Euros.
Medication costs
Medication costs were extracted from the Drug Information System and The Pharmacy Purchase Price database of the Dutch National Health Care Institute (25,26). For exacerbation free states and exacerbations leading to unscheduled GP visits and emergency room visits, medication costs were calculated per cycle length (one week). Using the data from the NControl, we extracted the brands that were dispensed between 2016-2018 for children and adolescents. For each medication group, we calculated a weighted mean of the prices based on the prevalence of different brands.
For exacerbations that led to three days of hospitalizations, medication costs of step 4 treatment were projected for the remaining 4 days of the state. Medication costs at the emergency room and during three days of hospital stay were projected in the emergency room visit costs and hospitalization costs respectively.
261
Asthma Guidelines have recommended 3-5 days of oral corticosteroids use (OCS) for exacerbations. We projected the costs of five-milligram prednisolone tablets twice daily for an average of 4 days for all exacerbation states.
Cost of genotyping
Base on expert opinion, the cost of a point-of-care genotyping test for ADRB2 rs1042713 was estimated to be 80 euros per test.
Cost of health care utilization
Exacerbation free states
For exacerbation free states, only the costs of medical treatment related to each step were projected. Cost of the scheduled physician visits was added to the medical costs of the exacerbation free states every 12 weeks.
Exacerbations states
A telephone call to the GP’s office for consultation was considered for all children/adolescents experiencing an asthma exacerbation. After the consultation, patients could travel either to the GPs office or to the emergency room. Patients who had an unscheduled GP visit were assumed to have a milder exacerbation and patients who traveled to the emergency department were assumed to have a more severe form of asthma exacerbation. Patients who visited the emergency room could either receive treatment at the department or could be hospitalized. An average of three days was assumed for hospitalizations. For all exacerbation states, four days of OCS use and three days of missed work days for one of the parents was assumed. For exacerbations leading to the emergency room and unscheduled physician visits, seven days of step 4 treatment and for exacerbations leading to hospitalization, four days of step 4 treatment was considered.
262
Direct costs of medical health care utilization such as telephone call to the physician, unscheduled and three monthly scheduled physician visits, emergency room visits, and hospitalization as well as indirect costs such as travel to the GPs office and emergency room, and loss of productivity from missed were retrieved from reference prices published by the Dutch Manual for Costing in Economic evaluations (27).
Analysis
The main outcome was the incremental cost-effectiveness ratio (ICER). For this purpose, costs and effectiveness of genotype-guided treatment were compared to the total cost and effectiveness of the three standard treatment arms. ICER was calculated as the following: ICER = (costs of the genotyped arm–costs of ST arm)/(exacerbations in the genotyped arm–exacerbations in ST arm).
Sensitivity analysis
A Probabilistic sensitivity analysis (PSA) using Monte Carlo simulation was perfumed to assess the combined impact of uncertainties about input parameters on the estimated cost-effectiveness of genotyping. Distributions were assigned to all variables, for probabilities Beta distributions were used and for costs gamma distributions. The model picked a random value for all parameters from these distributions and the results were recalculated. This was repeated 1,000 times and the results were depicted in a scatterplot.
263
Figure 1. Decision tree. Schematic representation of the decision tree. The decision tree starts with a decision node, patients who need a step up from low dose ICS, could either be assigned to the test arm (genotype-guided treatment) or to the ST arm (standard treatment). The circle node indicates that genotyping will divide patients into two groups based on the presence of the risk allele. In the ST arm, the decision node indicates that based on the physician’s decision patients will receive either ICS plus LABA (ICS+LABA), high dose ICS or ICS plus LTM (ICS+LTM). All arms lead to five identical Markov Traces. MT, Markov Trace; ST, Standard Treatment.
Figure 2. Schematic representation of the Markov trace. The Markov trace consists of six health states: three exacerbations-free and three exacerbation states. The figure also depicts transition possibilities between the health states.
264
Table 1. model input parameters
Parameters Base case value Standard Error Reference
Markov trace 1 (test group-high dose ICS) Exacerbation free (step 3) to hospitalization, (risks per week) 0.00046 0.00058 Hospital records Exacerbation free (step 3) to Emergency room visit (risks per week) 0.00094 0.00012 PACMAN Exacerbation free (step 3) to unscheduled physician visit (risks per week) 0.002 0.00025 PACMAN Exacerbation free (step 3) to exacerbation free step 4 through a scheduled visit 0.043 0.0055 PACMAN Exacerbation free step 4 to exacerbation free step 5 (risks per week) 0.0034 0.00043 PACMAN
Markov trace 2 (test group, high dose of ICS, ICS/LABA or ICS/LTM) Exacerbation free (step 3) to hospitalization (risks per week) 0.00024 0.0017 Hospital records Exacerbation free (step 3) to Emergency room visit (risks per week) 0.00072 0.003 PACMAN Exacerbation free (step 3) to unscheduled physician visit (risks per week) 0.0071 0.0093 PACMAN Exacerbation free (step 3) to exacerbation free step 4 through a scheduled visit 0.057 0.0047 PACMAN Exacerbation free step 4 to exacerbation free step 5 (risks per week) 0.008 0.01 PACMAN Markov trace 3 (ST, ICS plus LABA) Exacerbation free (step 3) to hospitalization (risks per week) 0.00027 0.0014 Hospital records
Exacerbation free (step 3) to Emergency room visit (risks per week) 0.00068 0.0022 PACMAN Exacerbation free (step 3) to unscheduled physician visit (risks per week) 0.016 0.01 PACMAN Exacerbation free (step 3) to exacerbation free step 4 through a scheduled visit 0.040 0.0047 PACMAN Exacerbation free step 4 to exacerbation free step 5 (risks per week) 0.017 0.01 PACMAN
Markov trace 4 (ST, high dose of ICS) Exacerbation free (step 3) to hospitalization (risks per week) 0.00032 0.0023 Hospital records
265 Exacerbation free (step 3) to Emergency room visit (risks per week) 0.00098 0.004 PACMAN
265
266
Exacerbation free (step 3) to unscheduled physician visit (risks per week) 0.0013 0.0047 PACMAN
Exacerbation free (step 3) to exacerbation free step 4 through a scheduled visit 0.043 0.0047 PACMAN Exacerbation free step 4 to exacerbation free step 5 (risks per week) 0.003 0.01 PACMAN
Markov trace 5 (ST, ICS plus LTM)
Exacerbation free (step 3) to hospitalization (risks per week) 0.00032 0.0023 Hospital records Exacerbation free (step 3) to Emergency room visit (risks per week) 0.00094 0.0067 PACMAN Exacerbation free (step 3) to unscheduled physician visit (risks per week) 0.021 0.032 PACMAN Exacerbation free (step 3) to exacerbation free step 4 through a scheduled visit 0.074 0.0047 PACMAN Exacerbation free step 4 to exacerbation free step 5 (risks per week) 0.022 0.01 PACMAN
From step 5 to step 4 (three monthly) 0.23 0.044 PACMAN From step 4 to step 3 (three monthly) 0.13 0.013 PACMAN
Treatment (total n=706) Ncontrol Higher dose of ICS, n (%) 209 (30) 0.000000001 ICS plus LABA, n (%) 437 (62) 0.000000001 ICS plus LTM, n (%) 60 (8) 0.000000001
Genotypes (total n = 842) PACMAN
Risk allele carriers, AA + AG, n (%) 562 (66.7) 0.000000001 Homozygous for the protective allele, GG, n (%) 280 (33.3) 0.000000001 ICS, inhaled corticosteroids; LABA, long acting beta2 agonist; LTM, leukotriene modifier. To define the range, we used 95% confidence intervals. For all probabilities, Beta distributions were used.
266
Table 2. Costs Direct non-medical costs per week Euros Range (base price ±20%) Distribution Direct costs within the healthcare system: Telephone call to the GP 17 13.6-20.4 Gamma GP visit 33 0.26.4-39.6 Gamma Emergency room visit 259 207.2-310.8 Gamma Hospitalization (average 3 days) 1884 1507.2-2260.8 Gamma Three monthly visits to the pediatrician/lung specialist. 101 80.8-121.2 Gamma
(OCS, average 4 days). 0.28 0.22-0.34 Gamma Standard therapy costs per week Step 3: High dose ICS 4.6 4.55-4.65 Gamma Combination of ICS plus LABA 5.07 4.97-5.16 Gamma ICS plus LTM 3.84 3.46-4.22 Gamma Step 4 6.07 6.02-6.12 Gamma Step 5 11.07 10.95-11.18 Gamma Direct Costs Outside the Healthcare System (per week) Travel expenses to ER/hospital/ pediatrician 2.66 2.13-4.8 Gamma Travel expenses to GP/ pediatrician 0.418 0.33-0.50 Gamma Indirect Costs Outside the Healthcare System: Loss of working days of the parents (average 3 days) 758.4 606.7-910.08 Gamma Costs of genotyping 80 64-96 Gamma
GP, general practitioner; ICS, inhaled corticosteroids; LABA, long-acting beta2 agonists; LTM, leukotriene modifier; OCS, oral corticosteroids.
267
267
Results
Results show that the ADRB2 genotyping strategy before start LABA is associated with an expected total cost per patient of 1782 Euros, while the current standard of care costs 1934 Euros per patient. ADRB2-guided treatment results in 0.35 expected exacerbations per patient over 2 years, while the standard care results in 0.68 expected exacerbations per patient over 2 years (Table 3).
Since genotyping results in cost-savings and additional avoided exacerbations, ADRB2-guided treatment dominates the current standard of care.
Probabilistic sensitivity analysis confirms this by showing a proportion of 60.7% in the southwest quadrant indicating a high probability of both cost-savings and patient benefit in terms of avoided exacerbations (Figure 3).
Table 3. outcome table for the base case results MT1 MT 2 MT 3 MT 4 MT 5 Total costs (Euros) 1610.67 1768.91 2103.58 1515.66 2127.92 Total exacerbations 0.26 0.52 0.86 0.21 0.92 Incremental costs - 151 Euros Prevented exacerbations 0.33 ICER The genotyping strategy dominates standard therapy ICER, incremental cost-effectiveness ratio; MT, Markov Trace.
268
1500
1000
500
0 -1 -0.5 0 0.5 1 1.5
-500
INCREMENTAL COSTS INCREMENTAL -1000
-1500 AVOIDED EXACERBATIONS
Figure 3. Probabilistic sensitivity analysis. This scatter plot reflects the results of the probabilistic sensitivity analysis with 1000 simulations. A total of 86.2% of simulations is associated with avoided exacerbations and 60.7% of all simulations is associated with avoided exacerbations and cost-savings.
Discussion
Results of this cost-effectiveness study show that ADRB2-guided LABA treatment has the potential to decrease the exacerbation risks while incurring cost-savings in health care. To the best of our knowledge, this is the first study to attempt to investigate the cost-effectiveness of genotyping in asthma treatment.
As we mentioned, for children whose asthma is not adequately controlled with low doses of ICS (treatment step 2), current asthma guidelines suggest three options: increasing the dose of ICS or addition of either LABA or low LTM to low dose ICS. While preferences of guidelines differ in their first choice for this treatment step, studies comparing the efficacy and effectiveness of these three options continue to deliver conflicting results. In 2015, a Cochrane systematic review and meta-analysis comparing the efficacy of LABA plus ICS versus high dose of ICS in asthmatic children, overall showed no significant differences between the two groups in terms of asthma exacerbation rates (28). Recently, a large matched cohort study in the UK showed that
269
both increased ICS and addition of LTM are as effective as ICS/LABA combination in reducing asthma exacerbations (29). On the other hand, a large randomized crossover trial of 182 children (6-17 years of age) has shown differential responses to three step- up treatments (30). Children with uncontrolled asthma received each of the three treatment options for 16 weeks. A differential response was defined as at least one treatment period that was ranked better than another based on the number of days with a good asthma control or the need for an OCS to treat acute asthma exacerbations. The proportion of children responding to ICS plus LABA was highest compared to the other two treatment options, however, there was substantial inter- individual variability in each treatment group. As we showed in this study, a part of this heterogeneity in the LABA treated group could be attributed to genetic variations in the ADRB2 gene.
One of the most important factors that support the potential clinical value of ADRB2- guided treatment in decreasing asthma exacerbations is the high frequency of the risk allele (A) in the asthma population. In PACMAN, the risk allele was present in 41% of the study population which is comparable to the allele frequency in healthy Europeans (39%) reported by the 1000 Genome project (31). In the largest meta-analysis of five asthma cohorts which included PACMAN as well, per copy of the risk allele, the risk of asthma exacerbations showed an increase of 52% in children/young adults exposed to LABA (17). Therefore, both heterogeneous and homogenous patients for the risk allele could benefit from the ADRB2-guided treatment.
The clinical benefits of ADRB2-guided treatment in children have been shown previously in a small clinical trial as well. In a study performed by Lipworth et al. in the UK, children homozygous for the risk allele (Arg16Arg, n=62) were randomized to treatment with ICS plus LTM or to treatment with ICS plus LABA (20). This one-year trial showed that children treated with LTM had better asthma control in terms of fewer school absences, fewer symptoms, less rescue medication and a better quality of life compared to the LABA treated children. In this study, LTM was used as an alternative to LABA treatment. However, prescription patterns might differ from country to country. For example, as we showed, compared with the other treatment options ICS plus LTMs is less commonly used in the Netherlands. In the results obtained from the NControl dataset, only 8% of the patients had received an LTM for step 3. In the ST arm of our study, we found the highest and the lowest exacerbation rates for ICS plus LTM (Markov Trace 5) and higher dose of ICS, respectively. Therefore, at least in the
270
Netherlands, higher dose of ICS could be the most appropriate option for children who carry the risk allele.
For this study, we used the data of the observational PACMAN cohort. This cohort included children with a broad range of asthma severity recruited through community pharmacies based on the use of regular asthma medications; therefore, we believe that PACMAN is a good representative of the Dutch pediatric asthma population. To avoid unnecessary complexity, we applied fixed transition probabilities for asthma exacerbation during the two-year time horizon of the study. However, it is well recognized that previous asthma exacerbations are the most important risk factors for future exacerbations (1). Additionally, patients on different treatment steps might have different exacerbation rates as well. This means that in real-life settings, patients on ICS/LABA would probably experience more exacerbations.
Although clinical trials are still considered the gold standard for drug efficacy assessment, observational studies provide valuable real-world information. However, clinical implementation of genotype-guided LABA treatment will only occur if randomized clinical trials can show that this approach is more cost-effective compared to the current standard treatment. The multi-center double-blind Pharmacogenetics Use For Further treatment Improvement in childreN (PUFFIN) trial started in 2017 to assess both efficacy and cost-effectiveness of ADRB2-genotype guided treatment in childhood asthma (32). In this trial, children (6 to 17 years of age) with a physician diagnosed of asthma and uncontrolled asthma symptoms despite adequate adherence and regular use of ICS for at least three months are recruited by medical centers in the Netherlands (32). Asthmatic children will be randomized to a precision medicine, genotype-guided treatment arm or to a usual care, non-genotype guided, control arm. Patient centered outcomes such as asthma control have been selected as the primary outcome to understand the impact of treatment on the patient. Additionally, asthma- related school absences, exacerbation rates, change in therapy, change in lung function and change in the fraction of exhaled Nitric Oxide will be analyzed as secondary outcomes (32).
In conclusion, there is a large variability in asthma treatment response to asthma medication and a one-size fits all approach based on current guidelines might not be optimal for all patients with a heterogeneous disease such as asthma. However, one of the main unmet clinical needs for children with persistent asthma is the lack of clinically available biomarkers to guide treatment to gain asthma control. This cost-
271
effectiveness study showed that genotype-guided LABA treatment is an economically attractive strategy in children with asthma since a large group of patients might benefit and the strategy will help saving costs. The PUFFIN trial will provide final answers to the question whether the ADRB2-genotype-guided treatment is cost-effective in daily clinical setting and could further pave the way for precision medicine as part of clinical practice in childhood asthma.
272
References:
1. Global Initiative for Asthma. Global asthma: update. World Allergy Organ J. Strategy for Asthma Management and 2016;9(1):26. Prevention 2018. Available from: www.ginasthma.org. 11. Szefler SJ, Phillips BR, Martinez FD, Chinchilli VM, Lemanske RF, Strunk RC, et 2. Fuhlbrigge AL, Adams RJ, Guilbert TW, al. Characterization of within-subject res- Grant E, Lozano P, Janson SL et al. The ponses to fluticasone and montelukast in burden of asthma in the United States: level childhood asthma. J Allergy Clin Immunol. and distribution are dependent on inter- 2005;115(2):233–42. pretation of the national asthma education and prevention program guidelines. Am J 12. Drazen JM, Silverman EK, Lee TH. Respir Crit Care Med. 2002;166(8):1044–9. Heterogeneity of therapeutic responses in asthma. Br Med Bull. 2000;56(4):1054–70. 3. van den Akker-van Marle ME, Bruil J, Detmar SB. Evaluation of cost of disease: 13. Farzan N, Vijverberg S.J.H, Arets H.G., Assessing the burden to society of asthma Raaijmakers J.A.M., Maitland-van der Zee in children in the European Union. Allergy. A.H. Pharmacogenomics of inhaled cortico- 2005;60(2):140–9. steroids and leukotriene modifiers: A sys- tematic review. Clin Exp Allergy J. 2017;47 4. Williams AE, Lloyd AC, Watson L, Rabe (2):271–93. KF. Cost of scheduled and unscheduled asthma management in seven European 14. Kersten E.T.G., Koppelman G.H. Union countries. Eur Respir Rev. 2006; 15 Pharmacogenetics of asthma. Curr Opin (98):4–9. Pulm Med. 2017;23(1):12–20.
5. Serra-Batlles J, Plaza V, Morejón E, 15. Turner S, Francis B, Vijverberg S, Pino- Comella A, Brugués J. Costs of asthma Yanes M, Maitland-van der Zee AH, Basu according to the degree of severity. Eur K, et al. Childhood asthma exacerbations Respir J. 1998;12(6):1322–6. and the Arg16 β2-receptor polymorphism: A meta-analysis stratified by treatment. J 6. Nurmagambetov T, Kuwahara R, Garbe Allergy Clin Immunol. 2016;138(1):107- P. The Economic Burden of Asthma in the 113.e5. United States, 2008–2013. Ann Am Thorac Soc. 2018;15(3):348–56. 16. Zuurhout M.J.L., Vijverberg S.J.H., Raaijmakers JAM, Koenderman L, Postma 7. From the Global Strategy for Asthma DS, Koppelman GH, et al. Arg16 ADRB2 Management and Prevention, Global genotype increases the risk of asthma Initiative for Asthma (GINA) 2015. exacerbation in children with a reported use http://www.ginasthma.org/. of long-acting β2-agonists: results of the PACMAN cohort. Pharmacogenomics. 8. Sims EJ, Price D, Haughney J, Ryan D, 2013;14(16):1965–71. Thomas M. Current Control and Future Risk in Asthma Management. Allergy, 17. Turner S, Francis B, Vijverberg S, Pino- Asthma Immunol Res. 2011;3(4):217. Yanes M, Maitland-van der Zee AH, Basu K, et al. Childhood asthma exacerbations 9. De Bosscher K, Vanden Berghe W, and the Arg16 β2-receptor polymorphism: Haegeman G. The Interplay between the A meta-analysis stratified by treatment. J Glucocorticoid Receptor and Nuclear Allergy Clin Immunol. 2016;138(1):107– Factor-κB or Activator Protein-1: Molecular 113.e5. Mechanisms for Gene Repression. Endocr Rev. 2003;24(4):488–522. 18. Slob EMA, Vijverberg SJH, Palmer CNA, Zazuli Z, Farzan N, Oliveri NMB, et 10. Hossny E, Rosario N, Lee BW, Singh al. Pharmacogenetics of inhaled long-- M, El-Ghoneimy D, SOH JY, et al. The use beta2-agonists in asthma: a systematic of inhaled corticosteroids in pediatric
273
review. Pediatr Allergy Immunol. 2018 28. Chauhan BF, Chartrand C, Ni Chroinin M, Milan SJ, Ducharme FM. Addition of 19. Zuurhout M.J. Vijverberg S.J., long-acting beta2-agonists to inhaled co- Raaijmakers JA, Koenderman L, Postma rticosteroids for chronic asthma in children. DS, Koppelman GH, et al. Arg16 ADRB2 Cochrane Database Syst Rev. 2015;(11): genotype increases the risk of asthma CD007949 exacerbation in children with a reported use of long-acting 2-agonists: Results of the 29. Murray CS, Thomas M, Richardson K, pacman cohort. Pharmacogenomics. 2013; Price D.B., Turner S.W. Comparative 14(16):1965–71 Effectiveness of Step-up Therapies in Children with Asthma Prescribed Inhaled 20. Lipworth BJ, Basu K, Donald HP, Corticosteroids: A Historical Cohort Study. Tavendale R, Macgregor DF, Ogston SA, et J Allergy Clin Immunol Pract. 2017;5(4): al. Tailored second-line therapy in asthma- 1082–1090.e7. tic children with the Arg(16) geno-type. Clin Sci (Lond). 2013;124(8): 521–8. 30. Lemanske RF, Mauger DT, Sorkness CA, Jackson DJ, Boehmer SJ, Martinez 21. Briggs A, Sculpher M. An introduction to FD, et al. Step-up therapy for children with Markov modelling for economic evaluation. uncontrolled asthma receiving inhaled Pharmacoeconomics. 1998;13(4): corticosteroids. N Engl J Med. 2010;362 397–409. (11):975–85.
22. McGhan WF, Al M, Doshi JA, Kamae I, 31. 1000 Genomes Project Consortium, Marx SE, Rindress D. The ISPOR Good Abecasis GR, Auton A, Brooks LD, Practices for Quality Improvement of Cost- DePristo MA, Durbin RM, et al. An Effectiveness Research Task Force integrated map of genetic variation from Report. Value Heal. 2009;12(8):1086–99. 1,092 human genomes. Nature. 2012;491 (7422):56–65 23. Koster ES, Raaijmakers JA, Koppelman G.H., Postma D.S., van der Ent C.K., 32. Vijverberg SJ, Pijnenburg MW, Hovels Koenderman L et al. Pharmacogenetics of AM, Koppelman GH, Maitland-van der Zee anti-inflammatory treatment in children with A-H. The need for precision medicine asthma: rationale and design of the PAC- clinical trials in childhood asthma: rationale MAN cohort. Pharmacogenomics. 2009 and design of the PUFFIN trial. Pharmaco- Aug;10(8):1351–61. genomics. 2017;18(4):393–401.
24. Buurma H, Bouvy M.L., De Smet P.A.G.M, Floor-Schreudering A, Leufkens HGM, Egberts ACG. Prevalence and deter- minants of pharmacy shopping behaviour. J Clin Pharm Ther. 2008;33(1):17–23.
25. www.medicijnkosten.nl/english. asp. National Health Care Institute (Zorginstituut Nederland). English Information.
26. National Health Care Institute (Zorg- instituut Nederland). GIP databank. The Drug Information System of National Health Care institute. www.gipdatabank.nl/ info Pagina.asp.
27. Tan SS, Bouwmans CAM, Rutten FFH, Hakkaart-van Roijen L. Update of the Dutch Manual for Costing in Economic Eva- luations. Int J Technol Assess Health Care. 2012;28(2):152–8.
274
275
276
CHAPTER 6 General Discussion & Summary
277
General Discussion
There is a great interest in identifying biomarkers to enable precision medicine in different disease areas. Biomarker discovery in complex diseases such as asthma could help to subgroup patients based on their molecular profiles, which could eventually help to predict treatment outcomes and tailor asthma management. Pharmacogenomics aims to discover genetic variations that could influence drug efficacy or safety. These studies can help to identify genetic markers that can select the most appropriate drug and dosage for each patient. Furthermore, pharmaco- genomics studies may reveal novel underlying pathways related to the disease or drug which can subsequently lead to the design of new drugs directed against targets in these pathways. However, before a pharmacogenomic asthma biomarker can reach clinical practice, it should pass several stages and overcome various challenges. The main aim of this thesis was to identify and validate genetic variants that could help to tailor asthma management in children. For this purpose, pharmacogenomics, epidemiological and cost-effectiveness approaches were combined and a multi- disciplinary approach was applied. In this chapter, the main findings of the thesis will be discussed and placed in a broader perspective, linking the main findings to the stages and challenges of biomarker discovery and implementation in patient care as depicted in Figure 1.
278
279
Figure 1. From pharmacogenomics marker discovery to clinical implementation. GWAS, genome- wide association study; LD, linkage dis-equilibrium; PGx, pharmacogenomics; RCT, Randomized Clinical Trials; WES, whole-exome sequencing; WGS, whole-genome sequencing.
From pharmacology to differences in treatment response
Like most scientific research incentives in medicine, the drive of any pharmaco- genomics study is a clinical problem. In pharmacogenomics, the main clinical problem is often an observed intra-individual variability in treatment response, resulting in suboptimal disease control or an increased frequency of adverse drug reactions. Generally, the rationale behind pharmacogenomics studies is that genetic variations can influence response to treatment by altering the drug-body interactions. In pharmacology, the drug-body interactions are divided into two classes; pharmaco- kinetics, the study of the influence of body on drug which consists of absorption, distribution, metabolism or excretion (ADME) of the drug, and pharmacodynamics, the study of the effect of the drug on body, especially the interaction between the drug and its target (i.e. receptors) (1). These processes are mainly orchestrated by different enzymes and proteins. Therefore, any alterations in genes encoding these molecules could subsequently alter the drug concentration at the site of action or change the affinity of the drug. These alterations can influence the efficacy of the drug and thereby treatment response (2,3). Depending on the physicochemical characteristics, and administration route of the drug that is being studied, different molecular components can be involved in the drug-body interactions (1). Gathering information regarding these properties can help to ask a relevant research question and choose an appropriate methodological approach (selection of response outcome, study design, and duration, study population, genotyping methods) for a pharmacogenomics study. Furthermore, this knowledge can help to interpret the results as well.
In this thesis, we mainly focus on the pharmacogenomics of inhaled corticosteroids (ICS), and whether genetic variants can explain why there is a large variability in reaction to this drug. Asthmatic children treated with ICS show differences in how well they respond to these drugs (4,5). Corticosteroids exert their effects primarily by binding to the glucocorticoid receptors (GR) in the cytosol (6,7). GRs are expressed in different cells including airway epithelial cells and immuno-inflammatory cells (8). The GR complex subsequently translocates to the nucleus of the cell and this results in suppression of the genes encoding for pro-inflammatory cytokines, chemokines, and enzymes and an increase in transcription of the anti-inflammatory genes (9). In addition
280 to these genomic mechanisms, ICS reduce inflammation through non-genomic mechanisms which have more rapid effects compared to the previously mentioned genomic mechanisms (10). In vitro studies have shown that ICS can interact with endothelium smooth muscle cell membranes (11). This interaction inhibits the uptake of norepinephrine in smooth muscle cells (12) which subsequently results in vasoconstriction and a decrease in the infiltration of peripheral inflammatory cells into the lungs (13). These non-genomic actions might explain the rapid effect of corticosteroids such as changes in bronchial blood flow within minutes after ICS use (13). Considering the broad anti-inflammatory effects of ICS, any changes in the genes encoding for proteins involved in ICS signalling pathway could modify the efficacy of this medication. As observed in chapter 2.2, over the past two decades, more than thirty pharmacogenomics studies have investigated the potential role of more than 120 genes (≥500 Single Nucleotide Polymorphisms (SNPs)) in ICS response hetero- geneity. Despite these high numbers so far none of the genetic variations has reached daily clinical practice. In the following paragraphs, we will address the potential reasons behind the lack of a clinically available genetic marker for asthma management as well as possible improvements in the field of pharmacogenomics of asthma.
The optimal design of a pharmacogenomic study vs reality
In order to design a pharmacogenomics study decisions have to be made regarding the appropriate study population, study duration and setting, type and dosage of the drug, treatment outcome, genotyping methods, type and time of data collection, potential covariates, genotyping methods, and statistical analysis plan (14). These factors altogether influence the validity and precision of a pharmacogenomics study. In chapter 2.2, we showed that pharmacogenomics studies are often post-hoc analyses of clinical studies of drug efficacy that are not specifically designed as pharmacogenomics studies. Other commonly applied study designs for pharmaco- genomics studies of asthma include observational population-based cohort and (nested) case-control studies (chapter 3.1).
Different study designs provide different levels of evidence. In general, RCTs are considered to provide a higher level of evidence than observational studies (15). However, due to the pre-selection and strict monitoring, an effect estimate of a genetic marker on a certain outcome might be overestimated in a trial setting compared to the
281
‘real world’ asthma population. On the other hand, observational studies are more prone to unpredictable confounding factors, but they provide insights into real-life situations. Pharmacogenomics studies generally require a large sample size (especially in the discovery phase) to have enough power to detect a modest effect size (14). Since observational studies are less costly than RCTs, they can more easily include higher numbers of individuals. Therefore, the value of the observational studies in the discovery phase of the pharmacogenomics studies should not be underestimated.
Defining asthma treatment response
A clear, preferably objective, outcome definition for the efficacy of therapy is of utmost importance in pharmacogenomics studies (14). Not only in the discovery phase, but also for later steps, for example to replicate the findings in an independent population or to pool the results of different studies for a meta-analysis.
Choosing a good proxy for drug efficacy can be challenging, especially for medications with a broad therapeutic effect such as ICS. These drugs can affect different features of asthma; such as; lung function, asthma symptoms and exacerbations (7,16).
Corticosteroids can reverse and improve lung function; therefore, this outcome is a suitable candidate for the assessment of treatment response (17). Lung function measurements are a fundamental part of asthma control and severity assessment in clinical practice (16). As observed in chapter 2.2, the most commonly used measure of lung function in pharmacogenomics studies of ICS is forced expiratory volume in 1 second (FEV1). In most studies, pre-bronchodilator changes in FEV1% (a ratio of FEV1 to forced vital capacity (FVC)) from baseline after starting the treatment was considered as the primary outcome. This outcome was considered as a continuous variable and calculated as ΔFEV1 percent predicted = FEV1% after treatment – FEV1% before treatment.
Several factors, however, are crucial to consider in the studies with lung function measurements as the main outcome of treatment response; 1) Age of the asthmatic children: since children younger than 5 years of age cannot perform spirometry; this outcome might not be suitable for preschool children. 2) The baseline value of the FEV1 before starting ICS: This is particularly important if the study aims to define
282 responders and non-responders based on FEV1 improvement. Studies show that even in children with life-threatening asthma exacerbations, FEV1 might be higher than 80% of predicted values on hospital admission (18). Furthermore, asthmatic children might have normal FEV1 measures between life-threatening asthma exacerbations (18). Therefore, patients might mistakenly be classified as non-responders since their FEV1 does do not show a significant improvement.
Unlike lung function, which is an objective outcome, asthma symptoms and exacerbations are more subjective, and from the patients’ perspective, they have the most impact on quality of life. Asthma symptoms are the second mostly assessed outcome in pharmacogenomics studies of asthma (chapter 2.2). Wheezing, coughing, daytime and nighttime symptoms, and shortness of breath influence the daily activity and thereby quality of life of the asthmatic children. Therefore, decreasing asthma symptoms is one of the main goals of asthma management programs (16). Asthma symptom control is usually assessed in clinical studies using standardised questionnaires such as the Asthma Control Questionnaire (ACQ) (19) and the Asthma Control Test (ACT) (20). In pharmacogenomics studies, asthma symptom control is either analysed as a dichotomous (controlled versus uncontrolled patients) or a quantitative variable (increase or decrease in symptom scores) (chapter 2.2). Quantitative variables are preferred in genetic association studies since they generate more power to detect a true association, and the results of different studies can be easily combined to in meta-analyses. However, often different methods and questionnaires are used by different study groups. Several studies have shown that there is low to moderate agreement between results of different questionnaires (21,22). Therefore, data pooling becomes challenging in collaborations like the Pharmaco- genomics in Childhood Asthma (PiCA) consortium (chapter 3.1). One possible solution is to use extreme phenotypes by categorising patients into well-controlled and uncontrolled asthma. However, it is important to consider that this approach might result in a loss of statistical power and loss of important genetic information from patients with partially controlled disease.
The third and least studied asthma treatment outcome in asthma pharmacogenomics studies is asthma exacerbations. Asthma exacerbations are often defined as “the worsening of symptoms that require urgent actions” (16). Untreated asthma exacerbations could lead to severe outcomes such as emergency room (ER) visits, hospitalisations, intensive care treatments, or in rare cases death. Exacerbations are
283
one of the main causes of stress in asthmatic children and their parents/caregivers and are significantly associated with morbidity and mortality (23). Despite the high importance of this outcome, only in the past few years, they have been used as a primary outcome variable in pharmacogenomics studies of ICS response (chapter 2.2). ICS decrease the frequency and intensity of the exacerbations by decreasing the underlying inflammation of the lungs. However, since there is no direct and standardised measure of asthma exacerbations yet, surrogate measures such as the hospitalisations, ER visits or short-course oral corticosteroid (OCS) use are commonly considered in the genetic association studies of asthma (24–26). In 2009, a statement by the European Respiratory Society and the American Thoracic Society was published in an attempt to standardise the definition of the exacerbations in asthma studies (27). In this report, an exacerbation was defined as an event that leads to hospitalisation or ER visits due to asthma or an event that requires a short course of systemic corticosteroids administration to prevent a more serious outcome such as intubation or death. Use of these definitions as proxies for asthma exacerbations makes the data pooling across studies possible. However, it is important to realize that the prevalence of asthma exacerbations might differ between studies due to factors such as cultural differences. For example, as we showed in chapter 3.2, the prevalence of OCS use ranged between 6.5% (PACMAN) and 77.2% (HPR) in different studies included in our meta-analysis. A part of this heterogeneity could be explained by cultural differences in the willingness of the physicians to prescribe OCS in different countries (28).
Furthermore, hospitalisation/ER visit rates ranged between 6% (PACMAN) and 58% (GALA II and HPR). Recruitment of the patients into different healthcare settings (e.g. primary/secondary/tertiary care, community pharmacies) could be the reason of differences in exacerbation rates. In our meta-analyses in chapter 3.2, we used two statistical tests of heterogeneity, I2 and Cochran’s Q-test (29,30), to measure the heterogeneity between studies. The results of the tests showed low to moderate heterogeneity between studies. Therefore, within-study differences did not have a significant impact on the results of the association analyses.
Previous studies have shown that outcome measures such as exacerbations, lung function and symptoms correlate with each other, at least partially (16). For example, both low FEV1 and persistent symptoms are associated with an increased risk of asthma exacerbations (16). However, neither low FEV1 nor persistent symptoms are
284 the strongest predictors of asthma exacerbations (16). Additionally, epidemiological studies have shown that predictors of poor asthma symptoms are different from the predictors of asthma exacerbations (31). In a study by Wu et al., researchers showed that both demographic characteristics and inflammatory profile of children with severe exacerbations were different from children with persistent symptoms (31). In their study, younger age and high levels of eosinophils were predictors of severe exacerbations but not asthma symptoms. In addition, in a study by Rogers et al. (32) including children treated with ICS, genetic predictors of recurrent asthma exacerbations and poor lung function were found to be different.
These findings suggest that despite the correlation between the three outcomes, their underlying molecular mechanisms might differ. Therefore, different measures of asthma control might reflect distinct dimensions of the disease. In chapter 3.1, we showed that different treatment outcomes consisted of different patient populations. By calculating κ values for three different PiCA studies, we showed only minimal-to- moderate agreements between exacerbations and asthma symptoms. Therefore, it is important to study all different outcomes in the pharmacogenomics of asthma.
Another important challenge when studying asthma treatment outcomes is that it is complicated to distinguish between severe asthma and poor treatment response, especially in observational studies. One possible approach to solve this problem (at least partly) in pharmacogenomics studies is to adjust the association analysis for disease severity. For this purpose, we used British Thoracic Society (BTS) treatment steps as a surrogate of asthma severity in chapter 3.2, while focusing on asthma exacerbations as a proxy for poor treatment outcome. However, it is important to realize that patients with severe symptoms might be undertreated and might show a better response to higher dosages of asthma medication.
Homogeneous vs heterogeneous study populations
Structure of the study population is one of the major factors in pharmacogenomics studies that could directly influence the generalizability of the results of the study (14). Combining genetically different subgroups of patient populations that demonstrate the same phenotype could confound the results of the association analysis by generating
285
false positive or false negative results (33). In the coming paragraphs, we will address two important factors influencing the homogeneity of the asthma study populations:
Dealing with diverse ethnic backgrounds:
There seem to be significant disparities in genetic structure and allele frequencies of different ethnicities (34,35). Because of the strong correlation among genetic variants (in other words linkage disequilibrium (LD)) in European populations, more genetic variations are inherited together. In contrast, a greater diversity of genetic variations is observed in populations with primarily African ancestry due to the higher number of recombination events that resulted in shorter regions of LD (34,35).
Asthma prevalence differs among distinct ethnic groups. In the United States, the highest childhood asthma prevalence is found for Puerto Ricans (21.9 %) followed by African Americans (14.6 %), European Americans (8.2 %) with the lowest prevalence in Mexican Americans (3.1 %) (36,37). Furthermore, compared to non-Hispanic white patients, African-Americans and Hispanics have a greater asthma burden in terms of severity, morbidity, and mortality (38). In addition, African-admixed populations have been shown to have low ICS response due to poor effectiveness or more adverse effects (38). The inter-ethnic differences in disease burden and response might be partially explained by differences in environmental exposures such as stress, violence, tobacco smoke, allergens and socioeconomic factors such as access to health care or low adherence due to concerns about medical costs (39–41). However, studies including African ancestry and non-Hispanic white patients with similar age, height, and gender, have shown significant differences in baseline lung function and asthma exacerbations between the two groups (42,43). These differences can partly be attributed to the disparities in the genetic structure of the different ethnic backgrounds. For example, while the association between SNPs in the 17q21 locus and asthma- related phenotypes has been consistently replicated in non-Hispanic white populations and Asians, studies of African-ancestry have delivered inconsistent results. As we showed (chapter 3.2), in the subgroup analyses, 17q21 rs7216389 was significantly associated with asthma exacerbations despite ICS in non-Hispanic European patients. However, subgroup analysis of African-admixed ethnicities delivered inconsistent results. While the SNP was significantly associated with asthma ER visit and hospitalisations in African-admixed populations, it was not associated with OCS use.
286
Additionally, in our GWAS of African-admixed populations (chapter 3.3), we did not detect any significant signals related to the 17q21 locus. The inclusion of African- Americans in the GWAS could be one of the main reasons for not detecting these variants. Furthermore, as we showed in chapter 3.3, the protective effect of the A allele at rs5995653 was stronger in African-admixed populations compared to non-Hispanic white patients. The allele was more frequent in African-admixed populations compared to non-Hispanic white patients as well.
These results highlight two important factors in pharmacogenomics studies of asthma. First, a study population with multiple ethnicities results in a heterogeneous population that could confound the results of the study. Population stratification should be considered when designing a genetic association study. The best approach to prevent population stratification is to exclude patients with a different ethnic background than the target population. Both self-reported ethnicity data or principal component analysis using genome-wide genotyping data are commonly used when selecting the participants (44,45). However, if for some reasons (e.g. low sample size), it is not possible to exclude patients, ethnicity data can be used as a covariate in association analysis. In chapter 3.3, in the discovery phase of the GWAS, we used principal components scores as covariates in the regression models.
Second, these results underline the importance of multi-ethnic studies in pharmacogenomics studies since they enable identifying ethnicity-specific and shared pharmacogenomics markers among different ethnicities. As we showed in chapters 2.2, GWAS of ICS response has primarily included non-Hispanic white and, to a lesser extent, East-Asian ancestry. Therefore, African-admixed populations are under- represented in pharmacogenomics studies of asthma. Compared to populations of European and Asian origin, patients with African ancestry seem to have a three-fold higher number of rare variants (34,46). In fact, part of the inter-ethnic differences of complex diseases can be attributed to the difference in the frequency of rare variants (47,48). Therefore, the inclusion of African-admixed populations in genetic association studies can increase statistical power to detect less frequent variants in complex diseases such as asthma. Furthermore, as we showed, there might be pharmaco- genomics factors specific to each ethnic group as well. Therefore, subgroup analysis can be particularly important to detect genetic variations that are specific to a certain ethnic group.
287
Asthma in children versus adults
Age of asthma onset has been considered as one of the factors separating phenotypes of asthma (16,49). Studies show that childhood onset and adult-onset asthma differ considerably concerning its risk factors and triggers, type of airway inflammation, natural disease course and response to asthma therapy (50). Therefore, it is of utmost importance to study children and adults separately.
Risk factors such as psychological/socioeconomics stress and exposure to pollutants, airway infections can trigger asthma in both children and adults (50). However, adult- onset asthma is mostly associated with active smoking and exposure to irritants (occupational asthma) (51–53). On the other hand, childhood asthma is mostly associated with allergic sensitisation and a family history of asthma or allergic diseases such as eczema, food/drug allergy (54). Allergic asthma in adults is mainly due to the persistence of childhood asthma to adulthood (55).
Molecular phenotypes between adults with childhood-onset and adult-onset asthma seem to differ as well (55). A recent study that investigated the gene networks underlying severe asthma in adults with childhood onset and adult-onset asthma (n=421) showed differential gene enrichment patterns in severe asthma (56). In this study, researchers showed that childhood-onset adult asthma was mostly associated with induced lung injury and/or fibrosis associated gene signature. In adult-onset, however, mostly gene signatures of mast cells, eosinophils and ILC3 pathways were present (56).
Taken together, these differences reflect the distinct cellular and molecular mechanisms between childhood and adult-onset asthma. In addition, it is well recognized by now that childhood asthma itself is phenotypically heterogeneous (49,57). Recent (semi)unbiased cluster analyses in children have identified different subgroups of childhood asthma (49,58,59). One of these studies was published by Fitzpatrick et al. in 2010 (49). Analysis of 161 children with mild to severe asthma who participated in the Severe Asthma Research Program (SARP) children (6–17 years of age) identified four clusters of patients: ‘late-onset symptomatic asthma’, ‘early-onset atopic asthma and normal lung function’, ‘early-onset atopic asthma with mild airflow limitation and comorbidities’, and ‘early-onset atopic asthma with advanced airflow limitation’. Asthma duration, baseline lung function and number of used controlled medications were the major discriminators of the clusters (49). These clusters were
288 further replicated in 611 children and young adults (6 to 18 years) participated in Childhood Asthma Research and Education (CARE) Network clinical trials (58). In another study that included The Childhood Asthma Management Program (CAMP) children, five clusters of patients were identified (59). These clusters were distinguished based on the degree of airway obstruction, atopic burden, and exacerbation rates. The results of the SARP and the CAMP children were similar concerning the airway obstruction, atopic burden, and exacerbation rates present within each of the clusters.
On the other hand, for decades, childhood asthma has been mainly associated with the allergic-Th2 inflammatory profile. However, recent studies show that Th2 inflammation does not cover the full spectrum of asthma pathophysiology in children (49). Linking phenotypes with gene expression analysis, studies have found Th17 and neutrophil-predominant asthma to be present in childhood asthma as well (57,60). Studies in adult asthmatics have linked neutrophilic asthma with poor corticosteroid response (61). Although corticosteroids reduce eosinophilic inflammation by inducing apoptosis, they seem to increase the number and survival of neutrophils (62,63). However, results in children have been conflicting. In a study by Su et al. (57) based on the gene expression profiles of peripheral blood mononuclear cells (PBMC), neutrophilic asthma was present in children with severe asthma and showed unique gene expression profiles enriched for corticosteroid-related pathways. This cluster had the highest number of non-responders to ICS. In another study (64), Andersson et al. found increased airway Th17 related cytokines expression in children with severe therapy-resistant asthma (STRA). In this study, researchers identified two subgroups within the STRA phenotype (neutrophil-high and neutrophil-low) indicating that different underlying molecular mechanisms could lead to the same phenotypes. Interestingly, increased intraepithelial airway neutrophilia in STRA was associated with better lung function. In contrast to the study by Su et al. (57), this study suggested a beneficial role for neutrophils in children with severe asthma (64). However, both studies underline the importance of subphenotyping to move towards precision medicine. Sub-phenotypes of childhood asthma reflect distinct cellular and molecular mechanisms. These findings show that subphenotyping will indeed increase the homogeneity of the study population in pharmacogenomics studies. However, subphenotyping could also result in small sample sizes and underpowered studies. Therefore, in future studies, based on the specific aims of the studies, decisions need to be made regarding the characteristics of the study participants to have a balance
289
between selecting a homogenous population with probably low sample size and having a larger sample size with less well-characterised cases. In collaborations where data regarding immune-inflammatory profiles of patients is available, one of the possible solutions could be to perform subgroup analysis to identify subphenotype-specific biomarkers. On the other hand, a less stringent characterisation of study subjects could help to identify shared genetic variants between different subgroups.
Genotyping phase
For the genetic part of the pharmacogenomics study, decisions have to be made regarding the number of genetic variations to be studied, the type of genotyping array, and source of genetic material (e.g. blood, saliva) (14). Candidate gene and genome- wide association (GWA) studies are the two most commonly applied approaches in pharmacogenomics studies in pediatric asthma (chapter 2.2). Both approaches investigate the association between common genetic variations (>5% allele frequency) and the trait of interest (14). However, these two methods differ mainly in the number of the variants.
In a candidate gene study (applied in chapter 3.2), typically one to several hundred variants are pre-selected based on prior knowledge of the biological mechanisms related to the disease or drug signalling pathway (14). Therefore, candidate gene studies are hypothesis-driven approaches. This approach can be applied in a discovery and replication phase.
On the contrary, GWAS (applied in chapter 3.3) is a hypothesis-generating approach meaning that thousands or even millions of genetic variants are studied simultaneously without any pre-selection (14). Using genotype imputation, the remaining variants that are not assayed on the genotyping arrays can be estimated with great precision using reference panels based on the knowledge about the correlation among variants (LD) (65). This approach is often applied in a discovery phase.
Genetic structure of complex traits includes a large number of variants with low to moderate frequency and effect sizes (65). Therefore, the unbiased nature of GWAS can provide valuable insights into the underlying complex biological pathways. In fact, “common variant common disease” is the main hypothesis behind GWAS as well as
290 candidate gene studies (66). Although GWAS is considered a robust technique to investigate common genetic variants, it has several important limitations.
First, because of the small effect sizes, to detect true association signals with enough statistical power, a large sample size is required. Although sample sizes of the pharmacogenomics studies of asthma are increasing, the largest sample sizes analyzed especially in GWAS (67) are still relatively low compared to GWAS of asthma (68,69) and asthma-related phenotypes (70) where healthy individuals are compared to asthmatic patients (chapter 2.2). The study population of pharmacogenomics studies consists of participants (mainly patients) treated with a specific medication. Finding large numbers of patients treated with the same medication is more challenging. A low sample size is one of the most important limiting factors in performing a good pharmacogenomics study since it can result in both overestimation of the effect size or a false negative finding due to low statistical power (14). Previous modelling studies have shown that to detect an OR of 1.3, for a SNP with an allele frequency of 0.4, at a genome-wide level (1x10-8), a sample size larger than 2,500 is required (14). Second, despite the considerable decreases in the cost of the GWAS arrays, compared to candidate gene studies, GWAS costs more because of the higher number of variants and the higher required sample size. The third limitation of GWAS is that because of the high number of genetic variants in the analysis, a stringent threshold (p ≤ 1x10-8) is applied to prevent multiple testing errors that could lead to false positive results. Furthermore, GWAS in different disease areas shows that even the strongest and most significant hits have small effect sizes and can only explain a small proportion of the estimated genetic variation (so-called missing heritability)(71).
A part of the missing heritability could be explained by the role of rare variants in complex diseases (72). Due to the approximately 3-fold increase in the growth of human populations over the past 400 years, the amount of the rare variations in the human genome has increased (73,74). In the 1000 genome project in 2000, the total number of rare variants was higher than common variants in the study population (46).
Unlike monogenic Mendelian diseases, the effect sizes of rare variants are low in common complex diseases making them rather impossible to detect in GWAS (75). Furthermore, imputation of rare variants is difficult and often imprecise. However, the emergence of next-generation sequencing (NGS) methods such as whole exome or whole genome sequencing, has made it possible to investigate the role of rare variants in complex diseases (76,77) Advances in NSG make the study of a broader range of
291
genetic variations (e.g. copy number variations, insertions/deletions, and trans- locations) feasible as well. These techniques are commonly used in detecting Mendelian disorders, especially in children since large effects of the rare variants are seen at young ages (78). Whole exome sequencing allows for detection of the variants in the protein-coding regions (exons), while whole genome sequencing reports all nucleotide bases across the entire genome (76,77). Whole-exome sequencing can help to identify functional variants by investigating both rare and common variants in exons. However, a considerable number of genetic variants associated with common diseases are located outside of exons which exome-sequencing would miss (79).
Furthermore, over the past years costs of both exome and whole genome sequencing have dropped substantially, however, compared to genotyping methods, genome sequencing is more expensive. According to the data collected from NHGRI-funded genome-sequencing groups, costs of whole-genome sequencing was below $1,121 (but still ~100 times higher than GWAS cost) per sample by July 2017 (80). Recently, the first whole-genome sequencing of asthma pharmacogenomics was performed by Mak et al. (48). The study identified population-specific and shared genetic variants associated with short-acting beta2 agonist response in a meta-analysis of 1,441 asthmatic children from three ethnically diverse populations. Identified variants were near five genes previously associated with immunity (NFKB1 and PLCB1), beta- adrenergic signalling (ADAMTS3 and COX18) and lung capacity (DNAH5).
Like GWAS, both whole exome and whole genome sequencing approaches provide an unbiased study of the genome. However, they require even larger sample sizes to detect true associations as well. One of the options to increase the overall sample size is to pool summary data across several studies in a meta-analysis as we have performed in chapter 3.2. International collaborative efforts and consortia can help achieve this goal by creating a platform to perform large-scale pharmacogenomics studies. There are several successful international consortia in different disease areas, and PiCA is the first international consortium to study the pharmacogenomics of childhood asthma (69,81). These consortia not only enable performing large-scale meta-analyses, but they can also harmonise the pharmacogenomics studies by standardisation of the methods and outcome definitions. Since meta-analysis approaches use summary data, they overcome limitations such as sharing privacy- sensitive genotype and clinical data on an individual level between different centres and/or countries.
292
NGS is the next big step in pharmacogenomics. However, since these approaches are relatively new in pharmacogenomics studies, successful replication of the findings face major drawbacks. Future pharmacogenomics research of NGS approaches should focus on international collaborations to include high numbers of participants with more ethnically diverse populations. Furthermore, collaborations can decrease expenses of sequencing for each research group by decreasing the number of participants required for the analysis. However, it is warranted to reach consensus on the methodology to enable more efficient and transparent scientific communications.
Replicating previously identified pharmacogenetic variants
Once a genetic variant is identified in a genetic association study, replication in independent populations is required to validate the association signals. Replication is one of the important steps that could determine the potential clinical value of a genetic variant (33). An ideal replication study investigates the influence of the same variant (or a variant in a very strong LD with the original variant) on the same treatment outcome in an independent population with the same LD pattern as the original study (33). Replication is considered successful when the effect estimate is statistically significant and in the same direction as the original discovery study. Although candidate gene approaches are used more commonly to perform replication analysis, hypothesis-free approaches can also be used for replication of the previous findings. As mentioned earlier, genetic components of a complex trait could be different among different populations and ethnicities because of the differences in their genetic structure. This means that findings from one population cannot be and should not be directly generalised to other populations. Therefore, trans-ethnic studies are required to assess the effect of the genetic variant in populations with different ethnic backgrounds. International consortia such as PiCA enable replication across different ethnic populations. In chapter 3.3, by pooling the summary data of five PiCA studies including non-Hispanic white asthmatic children, we could positively replicate the association between rs5995653 and asthma exacerbations despite the use of ICS found in the GWAS of the African-admixed population.
293
Being cautious when interpretation replication results
As mentioned earlier, the first factor that influences the fate of a genetic variant is the quality of the discovery study. As we showed in chapter 2.2, most discovery studies of asthma pharmacogenomics suffer from limitations such as low sample size that could lead to overestimation of the association signals and effect sizes. Therefore, many replication attempts fail because of false positive results of the discovery study.
Nonetheless, in some cases, replication fails because of the limitations in study designs of the replication studies. For example, studies that use different outcome definitions than the original studies might lose statistical power to replicate the findings. On the other hand, as mentioned in the previous sections, there are differences in the genetic structure of the populations with different ethnic backgrounds. Therefore, there might be population-specific genetic variants contributing to the phenotype of interest. This means that a failed replication in a population with a LD pattern other than the original study should not question the validity of the findings of the original study.
Over the last decade, by the emergence of relatively cheap genotyping arrays, the number of pharmacogenomics studies have increased tremendously (chapter 2.2). The consequence of the increased number of publications is an accumulation of studies with conflicting results (chapter 2.2). Additionally, heterogeneity between the studies has complicated the comparison of results and limited data pooling in a meta- analysis. On the other hand, negative results have lower publication rates leading to positive publication bias. In some cases, especially for genetic variants identified in hypothesis-generating approaches (such as GWAS) further functional studies are needed to generate more insight into the biological relevance of the association signals. If the biological mechanism is clear this will provide more confidence that there is a true association. However, if a genetic variant is a good predictor (replicated in different study populations) for a certain clinical outcome, biological mechanisms may not always need to be elucidated to be a clinically valuable marker.
From genetic variant to biological pathway
Since hypothesis-free approaches interrogate the entire genome, the potential role of the detected signals and their relationship with the trait of interest might not be clear. The following factors are the main reasons that challenge the translation of the
294 association signals into the disease-related functionality of the genetic variations. First, there is a very high chance that the lead SNP from the GWAS is not a causal variant but in very high LD with a causal variant (79). The main reason is that genetic variations on the microarrays are a selection of all SNPs present on the genome. These SNPs are in high LD with neighbouring SNPs. In fact, these ‘tag SNPs’ are proxies for a large part of the remaining unmeasured variants in the region (65). Different methods can be used to reveal the potential role of the genetic variants and their relation with the trait of interest. For example, immortalised peripheral blood B lymphocytes by Epstein– Barr virus transfection, have been used in several pharmacogenomics studies of asthma to evaluate the influence of the identified SNP on the expression of nearby genes (82,83) The same in-vitro method can be used to further evaluate the gene-drug interactions, by assessing differential expression rates of the genes in response to drugs. For example, GLCCI1 was identified by Tantisira et al. in a GWAS to be associated with response to ICS in asthmatic children (82). To validate the functional role of the GLCCI1 gene and its SNP in asthma control, Tantisira et al. performed an in-vitro study using lymphoblastoid B cells derived from children who participated in the study. Results showed that administration of dexamethasone significantly increased the expression of GLCCI1 (82). Specifically, gene expression in the presence of dexamethasone was significantly lower in patients homozygous for the mutant allele compared to the homozygotes for the wild-type allele. The results indicated that GLCCI1 could be one of the target genes of glucocorticoids. This method was applied by the same research group to reveal the function of the FBXL7 gene in modifying asthma symptom control in ICS users as well (83). One of the limitations of using immortalised B cell lines is that they probably do not reflect gene expression patterns of the other cells involved in asthma pathogenesis. Gene expression patterns are tissue-specific, and they are highly influenced by environmental triggers (e.g. allergens, infections). Recently, generation of disease-specific cell types from induced pluripotent cells obtained from human peripheral blood cells or fibroblasts has opened a new avenue to explore functional mechanisms of pharmacogenomics variants (84). Further in-vitro or in-vivo studies can also reveal the function of the identified genes and their relation to the trait of interest.
Recently, in-silico studies are commonly used for functional annotation and to discover the causal genetic variant(s). Fine-mapping is an in-silico approach that seeks the causal genetic variant(s) associated with the trait of interest (79). The statistical methods used in fine-mapping explore the genomic regions of the lead SNPs based
295
on the LD between measured SNPs and causal variants. Among the most important factors influencing the quality of fine-mapping are the number of causal variants and the magnitude of their association with the trait of interest, as well as SNP density, sample size, the local LD structure, and allele frequency. From these factors, only sample size and SNP density can be altered by the investigators (79). As mentioned in the previous sections, one of the most effective methods to increase sample size is performing a meta-analysis. Another factor to improve the quality of fine-mapping is to increase SNP density. Different methods can increase SNPs density: DNA sequencing, genotype imputation and additional genotyping. The latter two are more cost-efficient. Therefore, large sample size and imputation not only increase the quality of the discovery cohort, but they also influence the quality of fine-mapping to detect the causal variant. Using genomic annotation tools, possible biological function of SNPs selected by fine-mapping can further be investigated. It has been shown that approximately 90% of the identified SNPs in the GWAS map to the non-coding regions of the genome (85). Non-coding regions include promoters, enhancers, transcription sites, transcription factor binding sites, and long non-coding RNA loci. Therefore, protein-coding genetic variants account for only a small portion of the GWAS hits. Interestingly, most genetic variants identified by GWAS seem to be expression quantitative loci (eQTL), meaning that they modify the expression of nearby genes (86). Altered gene expression ultimately influences the trait/phenotype. Numerous public resources facilitate the discovery of gene expression patterns in different tissues, such as the Genotype-Tissue Expression (GTEx) project (87). GTEx includes genotypes, gene expression, histological and clinical data of 42 distinct tissues from 449 human donors. In chapter 3.3, we used GTEx to explore gene expression of APOBEC3B and APOBEC3C in asthma-related cells. High gene expression levels were shown for both APOBEC3B and APOBEC3C in pulmonary cells. Furthermore, disease-specific online tools such as AsthmaMap are becoming available as well (88) and might help to interpret pharmacogenomics findings. AsthmaMap aims to facilitate detailed representation of asthma mechanisms by collecting relevant information from literature search, accessible databases and input from experts in the respiratory field. As stated by Mazin et al. (88) AsthmaMap aims to: 1) develop a pathway-based network representation of the asthma mechanisms 2) Map fingerprints (using one type of 'omics) and handprints data (multilayer 'omics) to the network. 3) Perform network- based analysis, interpretation and hypothesis generation; 4) and to propose disease endotypes.
296
In-silico tools are becoming more popular day by day since laboratory-based functional studies are costly and time-consuming. However, it is important to consider that these online tools cannot replace traditional in-vitro or in-vivo methods. But in combination with the laboratory-based studies, they can provide valuable insights into the molecular function of the identified SNPs and genes.
The potential to reach clinical translation
One of the important steps for any genetic variant to move towards clinical implementation is the assessment of its clinical utility. If evidence from discovery and exploratory studies is confirmed by replication studies, a decision must be made regarding the introduction of the genetic marker into clinical practice at this stage or need for further confirmatory studies such as RCTs. As we stated in the previous sections, the required evidence for applying the pharmacogenetic test in clinical settings can be obtained from both RCTs and observational studies. However, often, for clinical implementation of pharmacogenetic testing, evidence gathered in RCTs are required. While no genetic marker of ICS response seems to be ready for a RCT (see chapter 2.2), the use of a pharmacogenomics marker to guide LABA treatment might be closer to clinical practice. ADRB2-genotype-guided LABA treatment RCTs have been conducted and are ongoing (89,90). The first RCT of ADRB2-genotype-guided LABA treatment included 62 asthmatic children. In this small study from the UK, Lipworth et al. (90) showed that children homozygous for the Arg16 allele within the ADRB2 gene would benefit more from a leukotriene modifier than from a LABA as an add-on treatment to ICS. Although this genotype-stratified RCT provided valuable information regarding the potential clinical utility of pharmacogenomic testing, the generalizability of the results was limited because only children homozygote for the variant were included. Another approach which provides the best level of evidence is to compare the genotype-guided treatment to the standard treatment. One of these so- called Precision Medicine Trials is the multi-center double-blind Pharmacogenetics Use For Further treatment Improvement in childreN (PUFFIN) trial (89) that has been started in 2017 to assess both efficacy and cost-effectiveness of ADRB2-genotype- guided treatment in childhood asthma. In this trial, children (6 to 17 years of age) with physician-diagnosed asthma and uncontrolled asthma symptoms despite adequate adherence and regular use of ICS for at least three months are recruited by 20 medical centers in the Netherlands. Asthmatic children will be randomised to a precision
297
medicine, genotype-guided treatment arm or to a usual care, non-genotype guided, control arm.
Implementation in clinical practice
As proposed by Hollander et al. (91), a biomarker should be ‘Superior’ (outperform current practice), ‘Actionable’ (change patient management), ‘Valuable’ (improve patient outcomes), ‘Economical’ (cost-saving or cost-effective) and ‘clinical Deployable’ (analysis technology available in clinical laboratory) (so-called the SAVED model) in order to move from observational evidence to clinical implementation. Therefore, even obtaining promising results from the RCTs or observational studies does not guarantee clinical implementation of the pharmacogenomics testing. First, in order to implement a new technique and device into routine clinical practice, consideration of costs is warranted. In fact, health insurance companies often require information about the cost-effectiveness of a new test before reimbursement. In chapter 5.1, we performed a cost-effectiveness study using a Markov model on ADRB2-guided LABA treatment and showed that compared to the standard care, pharmacogenomics-guided LABA use has the potential to decrease the exacerbation risks while incurring cost-savings in health care. A cost-effectiveness study using the ongoing PUFFIN trial data might provide final answers to the question whether the ADRB2-genotype-guided treatment is cost-effective in daily clinical setting in the Netherland. Furthermore, patients may not be willing to or unable to undergo this test, if not reimbursed. In order to implement a cost-effective test in clinical practice, clear reimbursement policies are needed (92).
Second, not only patients and policymakers, pharmacists’ and physicians’ awareness and acceptance of pharmacogenomics testing are important factors. When a pharmacogenomics test is accepted by policymakers, patients and health care providers, a workflow should be designed (93). This means that decisions must be made regarding which healthcare provider should order the test and for which patient. Also, the timing of genotyping (preemptive or point-of-care genotyping) should be determined (94). Ideally, information regarding clinically actionable pharmaco- genomics markers should be available at the point-of-care, through electronic medical records (EMR) (94). Only if a marker passes all these barriers, it will be implemented in clinical practice. Of the genetic markers studied in this thesis, the Arg16 variant in
298
ADRB2 might be most close to clinical implementation in pediatric asthma management (chapter 5.1).
Clinical implication and future directions
Additive effect of pharmacogenetic variants and respiratory viruses requires further attention
The main aim of most studies in this thesis was to identify markers that could explain heterogeneity in response to ICS. We mainly focused on asthma exacerbations as the treatment response. Evidence shows that approximately 85% of the exacerbations in children are caused by viral infections of which rhinovirus (RV) infection accounts for almost 70% of these episodes (95,96). However, it is likely that exacerbations are the result of complex interactions between different risk factors that might have an additive/synergic effect (97–99). Respiratory viruses replicate in epithelial cells of the airways (100). In addition to our study in chapter 3.2, several studies have found significant associations between the T allele at rs7216389 and asthma exacerbations in children (70,101,102). Rs7216389 seems to be an eQTL for several co-regulated genes (i.e. GSDMB, ORMDL3) in the 17q21 locus (103,104). ORMDL3 has been shown to be expressed in airway epithelial cells. ORMDL3 inhibits serine palmitoyl- CoA transferase, the rate-limiting enzyme for sphingolipid biosynthesis (105,106). Furthermore, ORMDL3 modifies Endoplasmic Reticulum (ER)-mediated Ca2+ homeostasis by inhibiting the sarco/endoplasmic reticulum Ca2+ ATPase (SERCA2b) pump. By altering Ca2+ amount in ER, ORMDL3 facilitates the unfolded protein response (UPR) (105,106). UPR is one of the key mechanisms by which viruses protect host cells from ER stress-mediated death (107,108). It has been shown that viral infections increase the expression of ORMDL3 especially in the presence of the TT genotype. Therefore, increased ORMDL3 expression in cells infected by RV may increase the efficiency of virus replication (109). On the other hand, viral infections seem to increase the expression of GSDMB as well. Recent studies are shedding more light on the function of the genes in the GSDM family (110). Gasdermins seem to mediate an inflammatory cell death called pyroptosis. These proteins participate in the cellular defence mechanisms in inflammatory conditions such as viral infections by forming pores in the cell membranes (110). GSDMB is expressed in airway epithelial cells and in a recent study a splice variant rs11078928, an SNP in relatively high LD
299
with rs7216389 (r2 = 0.75), was shown to cause deletion of the 6th exon of the GSDMB (111). Deletion of this exon subsequently resulted in the diminished pyroptotic activity of the GSDMB protein. Taken together, these results suggest that children and young adults who carry the T allele at rs7216389 are at higher risk of asthma exacerbations probably because of the higher replication rates of the respiratory viruses as well as diminished antiviral activities in epithelial cells. Furthermore, in-silico analyses of the APOBEC3 flanking genes identified in chapter 3.3 provided evidence of high levels of APOBEC3 RNA expression in pulmonary fibroblasts. In addition, based on the functional evidence that we obtained from HaploReg v4.1 for rs5995653, it seems that this SNP plays a key role in regulating the expression of genes involved in several cellular processes in the lung. The main function of these genes is related to the innate immunity against endogenous retro elements and exogenous viruses. As restrictors of viral infections, the expression of APOBEC3B and APOBEC3C in pulmonary tissues could be involved in fighting against viral infections. Therefore, alterations in the expression of these genes could influence the risk of asthma exacerbations despite the patients are treated with ICS as well. It has been shown that asthma exacerbations are more difficult to control in younger children (5). On the other hand, epidemiologic studies show that more than 80% of the asthmatic children are sensitised to environmental allergens (112,113). Recent studies have investigated the potential interaction between viral and allergen triggered immune response to explain severe exacerbations in children (97–99). It has been shown that cross-linking between allergen and high-affinity IgE receptor could impair virus-induced type 1 and 3 interferon production in peripheral blood cells. There is also evidence that Th2 cytokines (IL-4 and IL-13) impair RV-induced interferons in epithelial cells (114). Therefore, allergic sensitisation could result in deficient anti-viral responses in immune and airway epithelial cells. This means that allergens in sensitised children not only increase the risk of exacerbations but also increase the risk of respiratory viral infections. Overall, the result of the studies in this thesis support the hypothesis that an impaired defense system against respiratory viral infections could be one of the reasons of high exacerbation rates in some children. Genetic variations studied in this thesis (chapter 3.2 and chapter 3.3) seem to predispose a subset of children for an asthma phenotype that is less responsive to ICS. Therefore, asthma exacerbations might not be prevented adequately with the treatment regimens currently recommended by the asthma guidelines. Higher doses of ICS or additional treatment options might be required to achieve asthma control.
300
The need for large collaboration in precision medicine
Pharmacogenomics can provide important information for both basic scientists and the practicing clinician. However, most of the previous pharmacogenomics of ICS suffered from limitations such as low sample sizes. In this thesis we showed that large international collaborations such as PiCA can help to overcome some of these limitations while enabling sensitivity and subgroup analysis. In other disease areas, international attempts such as “Ubiquitous Pharmacogenomics (U-PGx)” project “An Horizon2020 Program to Drive Pharmacogenomics into Clinical Practice” are ongoing to bring pharmacogenomics testing into clinical practice (81). Especially by moving towards NGS studies, international collaborations are inevitable to overcome small sample sizes. Furthermore, as we showed in chapter 2.1, in addition to genomics studies, other ‘omics fields (epigenomics, transcriptomics etc) can provide important insights into mechanisms underlying asthma onset and progression. However, one of the problems of current ‘omics studies is the large variability between study designs, analytical techniques, and methods which makes the comparison of the results challenging. Therefore, international collaborative efforts can help to standardize the methods and techniques used for these studies.
Integrating multi-omics
Pediatric asthma once identified as a single allergic disorder is a disease with many phenotypes and endotypes. Therefore, ‘one size fits all’ approach in childhood asthma treatment will not lead to optimal disease control for everyone. It is well recognised by now that complex interactions between environmental, genetic, and epigenetic factors result in asthma development and progress. Therefore, to capture the complexity of the disease, the integration of different layers of information is needed. Large international collaborations such as U-BIOPRED (115) and the SysPharmPediA project that is ongoing in our department have already started to improve our understanding of severe asthma mechanisms by integrating high-dimensional molecular, physiological, and clinical data using systems biology approaches. A deeper understanding of these mechanisms can help to disentangle the complexity of asthma, identify phenotypes that are predictive of therapy response and discovery of new drug targets. This approach could ultimately lead to the development of clinically
301
applicable algorithms consisted of several markers to subphenotype asthma and guide asthma management.
In conclusion
In this thesis, we aimed to identify and validate pharmacogenomic variants that could help to tailor asthma management in children. Systematic literature review showed that most of the previously performed pharmacogenomics studies of asthma suffered from limitations, such as small sample sizes. These limitations prevented further replication and clinical implementation of the pharmacogenomics asthma markers. Our studies in this thesis highlighted the importance of international collaborations to perform large- scale pharmacogenomics studies. Using large-scale meta-analysis made it possible to identify novel pharmacogenomics variants, as well as validate previously identified variants such as 17q21. With the emergence of NGS studies to identify the potential role of both rare and common variants in complex traits, international collaborations remain inevitable to achieve enough statistical power and sample size. For any clinically validated pharmacogenomics marker, it is crucial to assess cost- effectiveness. None of the genetic variants identified in previous pharmacogenomics studies of ICS was ready for this step. However, we showed that ADRB2-genotype- guided treatment could be more effective and cost-saving compared to standard treatment. The PUFFIN trial designed to assess the efficacy and costs of ADRB2- genotype-guided treatment will provide final answers to the question whether the ADRB2-genotype-guided treatment is cost-effective in the daily clinical setting. It is well recognised by now that a single biomarker approach cannot capture the complex nature of asthma and its related phenotypes. Studies integrating multiple layers of information from molecular, biological, clinical, to environmental factors are becoming increasingly popular, and might be the next big step in asthma diagnosis and management.
302
References
1. B.G. Katzung and A. J. Trevor: Basic and 11. Horvath G, Wanner A. Inhaled cortico- clinical pharmacology. 13th edition, Mc steroids: effects on the airway vasculature Graw Hill, 2015. Accessible online via in bronchial asthma. Eur Respir J. 2006; accessmedicine.mhmedical.com. 27(1):172–87.
2. Burton PR, Tobin MD, Hopper JL. Key 12. Horvath G, Lieb T, Conner GE, Salathe concepts in genetic epidemiology. Lancet. M, Wanner A. Steroid sensitivity of nor- 2005;366(9489):941–51. epinephrine uptake by human bronchial arterial and rabbit aortic smooth muscle 3. Li H, Achour I, Bastarache L, Berghout J, cells. Am J Respir Cell Mol Biol. 2001;25 Gardeux V, Li J, et al. Integrative genomics (4):500–6. analyses unveil downstream biological effectors of disease-specific poly- 13. Kumar SD, Brieva JL, Danta I, Wanner morphisms buried in intergenic regions. A. Transient effect of inhaled fluticasone on NPJ genomic Med. 2016;1(1):16006. airway mucosal blood flow in subjects with and without asthma. Am J Respir Crit Care 4. Drazen JM, Silverman EK, Lee TH. Med. 2000;161(3 Pt 1):918–21. Heterogeneity of therapeutic responses in asthma. Br Med Bull. 2000;56(4):1054–70. 14. Hattersley A.T., McCarthy M.I. What makes a good genetic association study? 5. Bisgaard H, Gillies J, Groenewald M, Lancet. 2005;366(9493):1315–23. Maden C. The effect of inhaled fluticasone propionate in the treatment of young 15. Sørensen H.T., Lash T.L., Rothman asthmatic children: a dose comparison K.J. Beyond randomized controlled trials: a study. Am J Respir Crit Care Med. 1999; critical comparison of trials with non- 160(1):126–31. randomized studies. Hepatology. 2006;44 (5):1075–82. 6. Yudt MR, Cidlowski JA. The gluco- corticoid receptor: coding a diversity of 16. Global Initiative for Asthma. Global proteins and responses through a single Strategy for Asthma Management and gene. Mol Endocrinol. 2002;16(8):1719– Prevention 2018: www.ginasthma.org. 26. 17. Szefler SJ, Boushey HA, Pearlman DS, 7. Barnes PJ, Adcock IM. How do Togias A, Liddle R, Furlong A, et al. Time corticosteroids work in asthma? Ann Intern to onset of effect of fluticasone propionate Med. 2003;139(5 Pt 1):359–70. in patients with asthma. J Allergy Clin Immunol. 1999;103(5 Pt 1):780–8. 8. Pujols L, Mullol J, Roca-Ferrer J, Torrego A, Xaubet A, Cidlowski JA, et al. Exp- 18. Spahn JD, Cherniack R, Paull K, ression of glucocorticoid receptor α- and β- Gelfand EW. Is forced expiratory volume in isoforms in human cells and tissues. Am J one second the best measure of severity in Physiol Physiol. 2002;283(4):C1324–31. childhood asthma? Am J Respir Crit Care Med. 2004;169(7):784–6. 9. Hossny E, Rosario N, Lee BW, Singh M, El-Ghoneimy D, SOH JY, et al. The use of 19. Juniper E.F., Gruffydd-Jones K, Ward inhaled corticosteroids in pediatric asthma: S, Svensson K. Asthma Control Question- update. World Allergy Organ J. 2016;9(1): naire in children: validation, measurement 26. properties, interpretation. Eur Respir J. 2010;36(6):1410–6. 10. Borski RJ. Nongenomic membrane actions of glucocorticoids in vertebrates. 20. Nathan R.A., Sorkness C.A., Kosinski Trends Endocrinol Metab. 2000;11(10): M, Schatz M, Li J.T., Marcus P, et al. 427–36.
303
Development of the asthma control test. A 28. Wahlström R, Hummers-Pradier E, survey for assessing asthma control. J Lundborg C.S., Muskova M, Lagerløv P, Allergy Clin Immunol. 2004;113(1):59–65. Denig P et al. Variations in asthma treatment in five European countries- 21. Schuler M, Faller H, Wittmann M, judgement analysis of case simulations. Schultz K. Asthma Control Test and Fam Pract. 2002;19(5):452–60. Asthma Control Questionnaire: factorial validity, reliability and correspondence in 29. Higgins J.P.T., Thompson S.G. assessing status and change in asthma Quantifying heterogeneity in a meta-- control. J Asthma. 2016;53(4):438–45. analysis. Stat Med. 2002;21(11):1539–58.
22. O’Byrne PM, Reddel HK, Eriksson G, 30. Higgins JPT, Thompson SG, Deeks JJ, Ostlund O, Peterson S, Sears MR, et al. Altman DG. Measuring inconsistency in Measuring asthma control: a comparison of meta-analyses. BMJ. 2003;327(7414): 557 three classification systems. Eur Respir J. –60. 2010;36(2):269–76. 31. Wu A.C., Tantisira K, Li L, Schuemann 23. Fleming L, Murray C, Bansal AT, B, Weiss S.T., Fuhlbrigge A.L. Predictors of Hashimoto S, Bisgaard H, Bush A, et al. symptoms are different from predictors of The burden of severe asthma in childhood severe exacerbations from asthma in and adolescence: results from the paedi- children. Chest. 2011;140(1):100–7. atric U-BIOPRED cohorts. Eur Respir J. 2015;46(5):1322–33. 32. Rogers AJ, Tantisira K., Fuhlbrigge AL, Litonjua A., Lasky-Su JA, Szefler SJ, et al. 24. Koster ES, Maitland-van der Zee A H, Predictors of poor response during asthma Tavendale R, Mukhopadhyay S, Vijverberg therapy differ with definition of outcome. SJH, Raaijmakers JAM, et al. FCER2 Pharmacogenomics. 2009;10(8):1231–42. T2206C variant associated with chronic symptoms and exacerbations in steroid- 33. Ioannidis J.P.A., Thomas G, Daly M.J. treated asthmatic children. Allergy. 2011; Validating, augmenting and refining 66(12):1546–52. genome - wide association signals. Nat Rev Genet. 2009;10(5):318–29. 25. Turner S, Francis B, Vijverberg S, Pino- Yanes M, Maitland-van der Zee AH, Basu 34. Marth G, Schuler G, Yeh R, Davenport K et al. Childhood asthma exacerbations R, Agarwala R, Church D, et al. Sequence and the Arg16 β2-receptor polymorphism: variations in the public human genome data A meta-analysis stratified by treatment. J reflect a bottlenecked population history. Allergy Clin Immunol. 2016;138(1):107- Proc Natl Acad Sci U S A. 2003;100(1): 113.e5. 376–81.
26. Vijverberg SJH, Tavendale R, Leusink 35. Harpending H, Rogers A. Genetic M, Koenderman L, Raaijmakers JAM, perspectives on human origins and dif- Koppelman GH, et al. Pharmacogenetic ferentiation. Annu Rev Genomics Hum analysis of GLCCI1 in three north European Genet. 2000;1(1):361–85. pediatric asthma populations with a reported use of inhaled corticosteroids. 36. Moorman JE, Zahran H, Truman BI, Pharmacogenomics. 2014;15(6):799–806. Molla MT, Centers for Disease Control and Prevention (CDC). Current asthma pre- 27. Reddel H.K., Taylor D.R., Bateman valence-United States, 2006-2008. MMWR E.D., Boulet L.P., Boushey H.A., Busse Suppl. 2011;60(1):84–6. W.W. et al. An official American Thoracic Society/European Respiratory Society 37. Moorman JE, Rudd RA, Johnson CA, statement: asthma control and exacer- King M, Minor P, Bailey C, et al. National bations: standardizing endpoints for clinical surveillance for asthma--United States, asthma trials and clinical practice. Am J 1980-2004. MMWR Surveill Summ. 2007; Respir Crit Care Med. 2009;180(1):59–99. 56(8):1–54.
304
38. Ortega V.E., Meyers D.A. Pharmaco- 47. Cohen J.C., Kiss R.S., Pertsemlidis A, genetics: implications of race and ethnicity Marcel Y.L., McPherson R, Hobbs H.H. on defining genetic profiles for personalized Multiple rare alleles contribute to low medicine. J Allergy Clin Immunol. 2014; plasma levels of HDL cholesterol. Science. 133(1):16–26. 2004;305(5685):869–72.
39. Grant E.N., Lyttle C.S., Weiss K.B. The 48. Mak ACY, White MJ, Eckalbar WL, relation of socioeconomic factors and Szpiech ZA, Oh SS, Pino-Yanes M, et al. racial/ethnic differences in US asthma Whole-Genome Sequencing of Pharmaco- mortality. Am J Public Health. 2000;90(12): genetic Drug Response in Racially Diverse 1923–5. Children with Asthma. Am J Respir Crit Care Med. 2018;197(12):1552–64. 40. Nishimura K.K., Galanter J.M., Roth L.A., Oh S.S., Thakur N, Nguyen E.A. et al. 49. Fitzpatrick A.M., Teague W.G., Meyers Early-life air pollution and asthma risk in D.A., Peters S.P., Li X, Li H, et al. minority children. The GALA II and SAGE II Heterogeneity of severe asthma in studies. Am J Respir Crit Care Med. 2013; childhood: confirmation by cluster analysis 188(3): 309–18. of children in the National Institutes of Health/National Heart, Lung, and Blood 41. Akinbami L.J., Moorman J.E., Simon Institute Severe Asthma Research Pro- AE, Schoendorf KC. Trends in racial dis- gram. J Allergy Clin Immunol. NIH Public parities for asthma outcomes among Access; 2011;127(2):382-389. e1-13. children 0 to 17 years, 2001-2010. J Allergy Clin Immunol. 2014;134(3):547–553.e5. 50. de Nijs S.B., Venekamp L.N., Bel E.H. Adult-onset asthma: is it really different? 42. Hankinson JL, Odencrantz JR, Fedan Eur Respir Rev. 2013;22(127):44–52. KB. Spirometric Reference Values from a Sample of the General U.S. Population. Am 51. Polosa R, Knoke JD, Russo C, Piccillo J Respir Crit Care Med. 1999;159(1):179– G, Caponnetto P, Sarvà M, et al. Cigarette 87. smoking is associated with a greater risk of incident asthma in allergic rhinitis. J Allergy 43. Brehm JM, Acosta-Pérez E, Klei L, Clin Immunol. 2008;121(6):1428–34. Roeder K, Barmada MM, Boutaoui N, et al. African ancestry and lung function in Puerto 52. Dykewicz MS. Occupational asthma: Rican children. J Allergy Clin Immunol. Current concepts in pathogenesis, dia- 2012;129(6):1484–1490.e6. gnosis, and management. J Allergy Clin Immunol. 2009;123(3):519–28. 44. Tang H, Quertermous T, Rodriguez B, Kardia SLR, Zhu X, Brown A, et al. Genetic 53. Jamrozik E, Knuiman MW, James A, structure, self-identified race/ethnicity, and Divitini M, Musk AWB. Risk factors for confounding in case-control association adult-onset asthma: a 14-year longitudinal studies. Am J Hum Genet. 2005;76(2):268– study. Respirology. 2009;14(6):814–21. 75. 54. Bisgaard H, Bønnelykke K. Long-term 45. Price AL, Patterson NJ, Plenge RM, studies of the natural history of asthma in Weinblatt ME, Shadick NA, Reich D. childhood. J Allergy Clin Immunol. 2010; Principal components analysis corrects for 126(2):187–97. stratification in genome-wide association studies. Nat Genet. 2006;38(8):904–9. 55. Wenzel SE. Asthma phenotypes: the evolution from clinical to molecular app- 46. 1000 Genomes Project Consortium, roaches. Nat Med. 2012;18(5):716–25. Abecasis G.R., Auton A, Brooks L.D., DePristo M.A., Durbin R.M., et al. An 56. Hekking P-P, Loza MJ, Pavlidis S, de integrated map of genetic variation from Meulder B, Lefaudeux D, Baribaud F, et al. 1,092 human genomes. Nature. 2012;491 Pathway discovery using transcriptomic (7422):56–65. profiles in adult-onset severe asthma. J
305
Allergy Clin Immunol. 2018;141(4):1280– F, Kann M, editors. PLoS Comput Biol. 90. 2012;8(12):e1002822.
57. Su M.W., Lin W.C., Tsai C.H., Chiang 66. Sunyaev SR, Lathe WC, Ramensky VE, B.L., Yang Y.H., Lin Y.T., et al. Childhood Bork P. SNP frequencies in human genes asthma clusters reveal neutrophil pre- an excess of rare alleles and differing dominant phenotype with distinct gene modes of selection. Trends Genet. 2000; expression. Allergy. 2018;73(10):2024–32. 16(8):335–7.
58. Chang TS, Lemanske RF, Mauger DT, 67. Mosteller M, Hosking L, Murphy K, Fitzpatrick A.M., Sorkness C.A., Szefler Shen J, Song K, Nelson M, et al. No S.J., et al. Childhood asthma clusters and evidence of large genetic effects on steroid response to therapy in clinical trials. J response in asthma patients. J Allergy Clin Allergy Clin Immunol. 2014;133(2):363– Immunol. 2016; 139(3):797-803.e7. 369.e3. 68. Yan Q, Brehm J, Pino-Yanes M, Forno 59. Howrylak J.A., Fuhlbrigge A.L., Strunk E, Lin J, Oh S.S., et al. A meta-analysis of R.C., Zeiger R.S., Weiss S.T., Raby B.A., genome wide association studies of et al. Classification of childhood asthma asthma in Puerto Ricans. Eur Respir J. phenotypes and long-term clinical res- 2017;49(5):1601505. ponses to inhaled anti-inflammatory medications. J Allergy Clin Immunol. 2014; 69. Moffatt MF, Gut IG, Demenais F, 133(5):1289–300, 1300.e1-12. Strachan DP, Bouzigon E, Heath S et al. A large-scale, consortium-based genome- 60. Yeh Y.L., Su M.W., Chiang B.L., Yang wide association study of asthma. N Engl J Y.H., Tsai C.H., Lee Y.L. Genetic profiles of Med. 2010;363(13):1211–21. transcriptomic clusters of childhood asthma determine specific severe subtype. Clin 70. Bønnelykke K, Sleiman P, Nielsen K, Exp Allergy. 2018;48(9):1164–72. Kreiner-Møller E, Mercader JM, Belgrave D, et al. A genome-wide association study 61. Green RH, Brightling CE, Woltmann G, identifies CDHR3 as a susceptibility locus Parker D, Wardlaw AJ, Pavord ID. Analysis for early childhood asthma with severe of induced sputum in adults with asthma: exacerbations. Nat Genet. 2014;46(1):51– identification of subgroup with isolated 5. sputum neutrophilia and poor response to inhaled corticosteroids. Thorax. 2002;57 71. Maher B. Personal genomes: The case (10):875–9. of the missing heritability. Nature. 2008;456 (7218):18–21. 62. Meagher L.C., Cousin J.M., Seckl J.R., Haslett C. Opposing effects of gluco- 72. Manolio TA, Collins FS, Cox NJ, corticoids on the rate of apoptosis in Goldstein DB, Hindorff LA, Hunter DJ, et al. neutrophilic and eosinophilic granulocytes. Finding the missing heritability of complex J Immunol. 1996;156(11):4422–8. diseases. Nature. 2009;461(7265):747–53.
63. Cox G. Glucocorticoid treatment inhibits 73. Roberts L. 9 billion? Science. 2011;333 apoptosis in human neutrophils. Separation (6042):540–3. of survival and activation outcomes. J Immunol. 1995;154(9):4719–25. 74. Keinan A, Clark AG. Recent explosive human population growth has resulted in an 64. Andersson C.K., Adams A, Nagakumar excess of rare genetic variants. Science. P, Bossley C, Gupta A, De Vries D, et al. 2012;336(6082):740–3. Intraepithelial neutrophils in pediatric severe asthma are associated with better 75. Bomba L, Walter K, Soranzo N. The lung function. J Allergy Clin Immunol. 2017; impact of rare and low-frequency genetic 139(6):1819–1829.e11. variants in common disease. Genome Biol. 2017;18(1):77. 65. Bush WS, Moore JH. Chapter 11: GenomeWide Association Studies. Lewitter 76. Metzker ML. Sequencing technologies
306
- the next generation. Nat Rev Genet. eQTLs: annotation to enhance discovery 2010;11(1):31–46. from GWAS. Gibson G, editor. PLoS Genet. 2010;6(4):e1000888. 77. Tennessen JA, Bigham AW, O’Connor TD, Fu W, Kenny EE, Gravel S, et al. 87. Lonsdale J, Thomas J, Salvatore M, Evolution and functional impact of rare Phillips R, Lo E, Shad S, et al. The coding variation from deep sequencing of Genotype Tissue Expression (GTEx) human exomes. Science. 2012;337(6090): project. Nat Genet. 2013;45(6):580–5. 64–9. 88. Mazein A, Knowles RG, Adcock I, 78. Jamuar S.S., Tan E.C. Clinical Chung KF, Wheelock CE, Maitland-van der application of next-generation sequencing Zee AH, et al. AsthmaMap: An expert- for Mendelian diseases. Hum Genomics. driven computational representation of 2015;9(1):10. disease mechanisms. Clin Exp Allergy. 2018;48(8):916–8. 79. Schaid DJ, Chen W, Larson NB. From genome-wide associations to candidate 89. Vijverberg SJ, Pijnenburg MW, Hovels causal variants by statistical fine-mapping. AM, Koppelman GH, Maitland-van der Zee Nat Rev Genet. 2018;19(8):491–504. A-H. The need for precision medicine clinical trials in childhood asthma: rationale 80.From:https://www.genome.gov/275651 and design of the PUFFIN trial. Pharmaco- 09/the-cost-of-sequencing-a-human- genomics. 2017;18(4):393–401. genome/. 90. Lipworth BJ, Basu K, Donald HP, 81. Manson L.E., van der Wouden C.H., Tavendale R, Macgregor DF, Ogston SA, et Swen J.J., Guchelaar H.J.. The Ubiquitous al. Tailored second-line therapy in Pharmacogenomics consortium: making asthmatic children with the Arg(16) effective treatment optimization accessible genotype. Clin Sci (Lond). 2013;124(8): to every European citizen. Pharmaco- 521–8. genomics. 2017;18(11):1041–5. 91. Hollander Z, DeMarco ML, Sadatsafavi 82. Tantisira KG, Lasky-Su J, Harada M, M, McManus BM, Ng RT, Sin DD. Murphy A, Litonjua AA, Himes BE, et al. Biomarker Development in COPD: Moving Genomewide association between GLCCI1 From P Values to Products to Impact and response to glucocorticoid therapy in Patient Care. Chest. 2017;151(2):455–67. asthma. N Engl J Med. 2011;365(13): 1173–83. 92. Burke W. Genetic Tests: Clinical Validity and Clinical Utility. In: Current 83. Park H-W, Dahlin A, Tse S, Duan QL, Protocols in Human Genetics. Hoboken, Schuemann B, Martinez FD, et al. NJ, USA: John Wiley & Sons, Inc.; 2014. Genome-wide association study of leuko- 9.15.1-9.15.8. triene modifier response in asthma. J Allergy Clin Immunol. 2014; 133(3):664– 93. Alfirevic A, Pirmohamed M. Genomics 9.e5. of Adverse Drug Reactions. Trends Phar- macol Sci. 2017;38(1):100–9. 84. Loh Y-H, Agarwal S, Park I-H, Urbach A, Huo H, Heffner GC, et al. Generation of 94. van der Wouden C, Cambon-Thomsen induced pluripotent stem cells from human A, Cecchin E, Cheung K, Dávila-Fajardo C, blood. Blood. 2009;113(22):5476–9. Deneer V, et al. Implementing Pharmaco- genomics in Europe: Design and Im- 85. Maurano MT, Humbert R, Rynes E, plementation Strategy of the Ubiquitous Thurman RE, Haugen E, Wang H, et al. Pharmacogenomics Consortium. Clin Phar- Systematic localization of common disease macol Ther. 2017;101(3):341–58. associated variation in regulatory DNA. Science. 2012;337(6099):1190–5. 95. Papadopoulos NG, Christodoulou I, Rohde G, Agache I, Almqvist C, Bruno A, et 86. Nicolae DL, Gamazon E, Zhang W, al. Viruses and bacteria in acute asthma Duan S, Dolan M.E., Cox N.J. Trait- exacerbations - A GA2LEN- DARE* associated SNPs are more likely to be
307
review. Allergy. 2011;66(4):458–68. Dixon AL et al. Genetic variants regulating ORMDL3 expression contribute to the risk 96. Edwards MR, Saglani S, Schwarze J, of childhood asthma. Nature. 2007;448 Skevaki C, Smith JA, Ainsworth B, et al. (7152):470–3. Addressing unmet needs in understanding asthma mechanisms. Eur Respir J. 2017; 105. Breslow D.K., Collins SR, Bodenmiller 49(5):1602448. B, Aebersold R, Simons K, Shevchenko A et al. Orm family proteins mediate 97. Murray CS, Poletti G, Kebadze T, sphingolipid homeostasis. Nature. 2010; Morris J, Woodcock A, Johnston SL, et al. 463(7284):1048–53. Study of modifiable risk factors for asthma exacerbations: virus infection and allergen 106. Cantero-Recasens G, Fandos C, exposure increase the risk of asthma Rubio-Moscardo F, Valverde MA, Vicente hospital admissions in children. Thorax. R. The asthma-associated ORMDL3 gene 2006;61(5):376–82. product regulates endoplasmic reticulum- mediated calcium signaling and cellular 98. Holt P.G., Sly P.D. Viral infections and stress. Hum Mol Genet. 2010;19(1):111– atopy in asthma pathogenesis: new 21. rationales for asthma prevention and treatment. Nat Med. 2012;18(5):726–35. 107. Smith J.A., Turner M.J., DeLay ML, Klenk E.I., Sowders D.P., Colbert R.A. 99. Green R.M., Custovic A, Sanderson G, Endoplasmic reticulum stress and the Hunter J, Johnston SL, Woodcock A. unfolded protein response are linked to Synergism between allergens and viruses synergistic IFN-beta induction via X-box and risk of hospital admission with asthma: binding protein 1. Eur J Immunol. 2008; case-control study. BMJ. 2002;324(7340): 38(5):1194–203. 763. 108. Trujillo-Alonso V, Maruri-Avidal L, 100. Mosser A.G., Vrtis R, Burchell L, Lee Arias CF, Lopez S. Rotavirus Infection W.M., Dick C.R., Weisshaar E, et al. Induces the Unfolded Protein Response of Quantitative and Qualitative Analysis of the Cell and Controls It through the Rhinovirus Infection in Bronchial Tissues. Nonstructural Protein NSP3. J Virol. 2011; Am J Respir Crit Care Med. 2005;171(6): 85(23):12594–604. 645–51. 109. Calışkan M, Bochkov YA, Kreiner- 101. Bisgaard H, Bønnelykke K, Sleiman Møller E, Bønnelykke K, Stein MM, Du G, PMA, Brasholt M, Chawes B, Kreiner- et al. Rhinovirus wheezing illness and Møller E et al. Chromosome 17q21 gene genetic risk of childhood-onset asthma. N variants are associated with asthma and Engl J Med. NIH Public Access; 2013;368 exacerbations but not atopy in early (15):1398–407. childhood. Am J Respir Crit Care Med. 2009;179(3):179–85. 110. Ruan J, Xia S, Liu X, Lieberman J, Wu H. Cryo-EM structure of the gasdermin A3 102. Tavendale R, Macgregor DF, Mukho- membrane pore. Nature. 2018;557(7703): padhyay S, Palmer CNA. A polymorphism 62–7. controlling ORMDL3 expression is associated with asthma that is poorly 111. Panganiban RA, Sun M, Dahlin A, controlled by current medications. J Allergy Park H-R, Kan M, Himes BE, et al. A Clin Immunol. 2008;121(4):860–3. functional splice variant associated with decreased asthma risk abolishes the ability 103. Dixon AL, Liang L, Moffatt MF, Chen of gasdermin B to induce epithelial cell W, Heath S, Wong KCC, et al. A genome- pyroptosis. J Allergy Clin Immunol. 2018; wide association study of global gene 142(5):1469-1478.e2. expression. Nat Genet. 2007;39(10):1202– 7. 112. Sporik R, Holgate ST, Platts-Mills TAE, Cogswell JJ. Exposure to House-Dust 104. Moffatt MF, Kabesch M, Liang L, Mite Allergen (Der p I) and the
308
Development of Asthma in Childhood. N 115. Fleming L, Murray C, Bansal AT, Engl J Med. 1990;323(8):502–7. Hashimoto S, Bisgaard H, Bush A, et al. The burden of severe asthma in childhood 113. Permaul P, Hoffman E, Fu C, Sheehan and adolescence: results from the W, Baxi S, Gaffin J, et al.Allergens in urban paediatric U-BIOPRED cohorts. Eur Respir schools and homes of children with asthma. J. 2015;46(5):1322–33. Pediatr Allergy Immunol. 2012;23(6):543– 9.
114. Durrani SR, Montville DJ, Pratt AS, Sahu S, DeVries MK, Rajamanickam V, et al. Innate immune responses to rhinovirus are reduced by the high-affinity IgE receptor in allergic asthmatic children. J Allergy Clin Immunol. 2012;130(2):489–95.
309
310
Scientific Summary
311
Scientific Summary
Inhaled corticosteroids (ICS) have been the mainstay treatment for persistent asthma in children for decades (1). However, despite their high efficacy, there is substantial inter- and intra-individual variability in treatment response to ICS (2). Suboptimal asthma control might lead to persistent asthma symptoms and/or asthma exacerbations (1). Asthma exacerbations reduce the quality of life and are the main source of stress in asthmatic children and their parents/caregivers. If severe, they can lead to hospitalizations or even death (3). Given the high influence of asthma exacerbations on morbidity, mortality and healthcare costs, there is an urgent need for identification of (bio)markers that could help to identify patients at higher risk of asthma exacerbation, optimize treatment and identify new treatment targets. Environmental factors (i.e. exposure to allergens, tobacco smoke etc.), clinical/demographic factors (i.e. high BMI, young age), as well as poor treatment adherence and incorrect inhalation technique, are among the known risk factors of poor asthma control in children (1). Furthermore, there has been a considerable interest in the role of genomic variations in treatment response heterogeneity (4–8). While rapid advances in genome technologies and reduced genotyping costs over the past decades have resulted in a tremendous rise in the number of pharmacogenomics studies of asthma (9), most studies have suffered from limitations such as low statistical power and methodological issues. There is still no genetic marker ready for clinical implementation in asthma management.
There is a need for international collaborative efforts in the field of pharmacogenomics of asthma to obtain large sample sizes of well-characterized asthmatic children in order to perform large-scale meta-analyses to 1) assess the clinical value of genetic markers for asthma management, 2) identify markers that can guide asthma treatment and 3) to identify novel drug targets.
The main focus of this thesis was to identify genetic variations that could explain the heterogeneity in response to ICS predominantly in the concept of asthma exacerbations. For this purpose, we have used candidate gene and Genome Wide Association (GWA) study approaches, using data from clinical trials and cohorts that participated in the Pharmacogenomics in Childhood Asthma (PiCA) consortium. Furthermore, we investigated which clinical/environmental factors and (pharmaco) genetic markers were associated with an increased risk of exacerbations in children with regular use of ICS using data of the PACMAN cohort. Finally, we carried out an
312 economic evaluation to examine if a pharmacogenomic-guided strategy to prevent exacerbations is cost-effective for asthmatic children who need a step up from low dose ICS. For this purpose, we performed a cost-effectiveness study by comparing ADRB2-guided treatment (rs1042713) to the standard treatment before starting long- acting beta2 agonists (LABA). This genetic variant is the most consistently replicated variation in pharmacogenomics studies of LABA (10). Overall, we applied a multi- disciplinary approach by combining pharmacogenomics, epidemiology and cost- effectiveness studies.
In the first part of this thesis (chapter 2), we conducted one narrative and one systematic literature review to search genomics markers (genetic and epigenetic) with potential clinical value.
In chapter 2.1 we provide an overview on the latest insights of pharmacogenomic, epigenomic and transcriptomic studies related to childhood asthma management. This narrative review showed that most of the epigenomic and transcriptomic studies in children have focused on identifying factors associated with asthma susceptibility and asthma-phenotypes. Although the heterogeneity between study designs, patient selection, clinical outcomes, and sampling tissues made the comparison of findings difficult, most epigenomic and transcriptomic studies highlighted the importance of exposure to different environmental factors (i.e. tobacco smoke exposure, farm animals, viral infections, house dust mite) in asthma development, specifically during the first years of life. Furthermore, despite the high number of pharmacogenomic studies, only one variant currently, shows the promise to be a clinically relevant pharmacogenetic marker. Based on the review we can conclude a Systems Medicine approach is needed to deepen our understanding of different omics layers and their interaction with each other and environment. Further understanding of asthma and its phenotypes can help to identify new drug targets and biomarkers that are predictive of therapy response and can ultimately lead to better management of the disease.
In chapter 2.2 we present the results of a systematic review that we conducted to gain more insight into the pharmacogenomics studies of ICS and LTMs, two anti- inflammatory medications of asthma. We showed that more than 700 SNPs have been studied in pharmacogenomics studies, but few have been robustly replicated. Moreover, despite successful replications in pharmacogenomics of LTMs (MRP1 and ALOX5) and of ICS (CRHR1, FCER2, and GLCCI1), there were inconsistencies in the results of these studies. In addition, none of the previously studied SNPs in candidate
313
gene studies were identified in GWAS. Furthermore, only one SNP (rs37973, GLCCI1) identified in a GWAS was later replicated in independent populations. The most consistent results were shown in candidate gene studies of ICS for an SNP (rs28364072) within the FCER2 gene in different paediatric populations.
In chapter 3.1, we described the characteristics of the clinical trial/cohorts participating in the PiCA consortium. PiCA brings together more than 21 asthma and birth cohorts and more than 14,000 asthmatic children/young adults from different ethnic backgrounds. The consortium includes a broad spectrum of children/young adults with mild-to-severe asthma. We showed that currently more than 60% of the children/young adults in PiCA are treated with ICS. From those, approximately one third have reported severe exacerbations despite ICS treatment in the past 6 or 12 months or during the first year of the trial.
In chapter 3.2, we performed large-scale meta-analyses to investigate the association between a genetic variant within the 17q21 locus (rs7216389), and asthma exacerbations despite ICS use in 13 PiCA studies (n=4,529 patients). Rs7216389 is a well-replicated childhood asthma susceptibility locus that was initially identified in a large-scale GWAS and we were interested in whether this variant is also associated with exacerbations despite ICS treatment. We found a significant association between the T allele (allele frequency ranged between 0.54-0.81) at rs7216389 SNP and risk of asthma exacerbations in children/young adults despite the use of ICS. The T allele increased the risk of asthma-related hospitalization/ED (adjusted OR per increase in the T allele: 1.32, 95%CI: 1.17-1.49, p<0.0001, n=4,454) and short-term Oral Corticosteroid use (OCS) (adjusted OR per increase in the T allele: 1.19, 95%CI: 1.04- 1.36, p=0.01, n=4,050). Furthermore, we performed a sensitivity analysis by categorizing the patients based on their age (2-4 years of age versus ≥5) at the time of the outcome measurement. The results of our sensitivity analysis showed significant associations between the SNP and asthma exacerbations in children ≥5 years of age remained significant. However, the results for pre-school children were not significant, this might be due to low patient numbers. In a subgroup analysis, we found the same association for hospitalizations/ER visits in both non-Hispanic whites (adjusted OR: 1.33, 95%CI: 1.10-1.61, p=0.004, n= 2,888) and Hispanics (adjusted OR: 1.31, 95%CI: 1.06-1.63, p=0.01, n= 916). However, for OCS use, the result of the meta-analysis of Hispanics was not statistically significant (adjusted OR: 0.96, 95%CI: 0.76-1.22, p=0.7, n= 916, adjusted OR: 1.26, 95%CI: 1.09-1.45, p=0.002, n=2,492 in non-Hispanic white
314 patients). It seems that asthma exacerbations in children and young adults approximately increases by 30% per copy of the T allele at rs7216389. Therefore, asthma exacerbations in patients carrying the T allele are not adequately prevented with the current recommendations of the guidelines. Higher doses of ICS or additional treatment options might be required to achieve asthma control in these children.
In chapter 3.3, we performed a GWAS meta-analysis in African-admixed children and young adults from two PiCA studies (GALA II and SAGE II) to identify (novel) genetic markers associated with asthma exacerbations despite ICS use. We found that a total of 15 independent variants were suggestively associated with asthma exacerbations in African-admixed children despite ICS use (n= 1347, aged 8 to 21 years old in both studies, p≤5x10-6). Rs5995653, located in the intergenic region of APOBEC3B and APOBEC3C, was positively replicated in the pooled analysis of five PiCA studies (ESTATE, followMAGICS, PACMAN, PASS, and SLOVENIA) including non-Hispanic white asthmatics populations (n=1,699) (p = 0.01). Compared to non-Hispanic white asthmatics, the protective effect of the A allele at rs5995653 was stronger in African- admixed populations (OR: 0.66, 95%CI: 0.56-0.79 in African-admixed versus OR: 0.77, 95%CI: 0.63-0.94 in non-Hispanic white patients). In addition to asthma exacerbations, the SNP was also associated with changes in lung function after 6-8 weeks of treatment with ICS in the SLOVENIA cohort including non-Hispanic white asthmatic patients (p = 7.54x10-4). To date, no asthma-related functions have been associated with any of the APOBEC3 flanking genes. Based on the functional evidence that we obtained from HaploReg v4.1 for rs5995653, it seems that this SNP plays a key role in regulating the expression of genes involved in several cellular processes in the lung. The main function of these genes is related to the innate immunity against endogenous retro-elements and exogenous viruses. Viral infections are responsible for most of the exacerbations in asthmatic children. As restrictors of viral infections, the expression of APOBEC3B and APOBEC3C in pulmonary tissues could be involved in fighting against viral infections. Therefore, alterations in the expression of these genes could influence the risk of asthma exacerbations despite the patients are treated with ICS. Furthermore, a previously reported association of the L3MBTL4-ARHGAP28 locus in non-Hispanic white and Asian patients was confirmed in our African-admixed population as well. This is the first GWAS of ICS response in African-Admixed children and young adults with asthma.
315
In chapter 4.1, we aimed to identify genetic and non-genetic risk factors associated with an increased risk of asthma exacerbations in children who use ICS in a real-life pharmacy-based cohort (PACMAN cohort study). We investigated the association between asthma exacerbations and clinical, demographic, and environmental risk factors in the PACMAN children treated with ICS. One of the aims of the study was to build a genetic risk score to assess the added value of the genetic and pharmacogenetic variants in identifying children with an increased risk of asthma exacerbations compared to non-genetic risk factors solely. However, none of the genetic factors was statistically significantly associated with asthma exacerbations despite ICS use. For the non-genetic factors, we found an association with food allergy and aged. Patients with food allergy had an approximately 2-fold higher risk of exacerbations compared to the children without food allergy (adjusted Odds Ratio (adjOR) = 2.05, 95% confidence interval (CI): 1.20-3.49, p= 0.008). A weak but significant association was found between age and the risk of exacerbations. Each additional year of age was associated with 14% decrease in the risk of asthma exacerbations (adjOR =0.86, 95%CI: 0.76-0.97, p= 0.01). We suggest that asthmatic children who also suffer from food allergy might benefit from a more intensive disease monitoring program to decrease the risk of exacerbations.
Finally, in chapter 5.1, we carried out an economic evaluation to examine if a pharmacogenomic-guided strategy to prevent exacerbations is cost-effective for asthmatic children who need a step up from low dose ICS. For this purpose, we performed a cost-effectiveness study by comparing the ADRB2-guided treatment (rs1042713) to the standard treatment before starting long-acting beta2 agonists (LABA). The largest study assessing the influence of this genetic variant on asthma exacerbations was published in 2016 by Turner et al. In this study, a meta-analysis of more than 4000 PiCA patients showed that children/young adults treated with ICS plus LABA have a 1.5-fold increase in the risk of asthma exacerbations for addition of each copy of the ADRB2 risk allele, while this increased risk was not observed in children treated with ICS solely or ICS plus LTM. By developing a probabilistic Markov model, we investigated the costs and effectiveness of one-time genotyping prior to LABA start versus standard treatment. Our target population was children and adolescents (6-18 years of age) with uncontrolled asthma despite low dose ICS use who needed a step up in their current treatment. We defined treatment effectiveness as avoided asthma exacerbations. The economic evaluation was performed from the societal perspective in the Netherlands and a time horizon of two years was considered. Our results showed
316 that ADRB2-guided LABA use has the potential to decrease the exacerbation risks while incurring cost-savings in health care. The genotyping strategy was associated with expected costs per patient of 1782 Euros, while the current standard of care was 1932 Euros. Therefore, based on our modeling, the total costs for genotype-guided treatment would be 151 euros less than standard treatment. Furthermore, probabilistic sensitivity analysis confirms this by showing a proportion of 60.7% in the southwest quadrant indicating a high probability of both cost-savings and patient benefit in terms of avoided exacerbations. Therefore, compared with standard care, ADRB2-guided LABA treatment is an economically attractive strategy in children with asthma since a large group of patients might benefit from while saving costs.
Chapter 6 provides a general discussion of the findings and recommendations for clinical practice and future research.
In conclusion, there is a need for shifting from a single biomarker approach to multidimensional approaches in which the clinical value of a combination of markers is investigated. This approach could ultimately lead to the development of clinically applicable algorithms consisting of several markers to subphenotype asthma and guide asthma management.
317
References
1. Global Initiative for Asthma. Global 7. Drazen JM, Yandava CN, Dubé L, Strategy for Asthma Management and Szczerback N, Hippensteel R, Pillari a, et al. Prevention 2018: www.ginasthma.org. Pharmacogenetic association between ALOX5 promoter genotype and the response 2. Drazen JM, Silverman EK, Lee TH. to anti-asthma treatment. Nat Genet. 1999; Heterogeneity of therapeutic responses in 168–70. asthma. Br Med Bull. 2000;56(4):1054–70. 8. Tantisira KG, Silverman ES, Mariani TJ, Xu
J, Richter BG, Klanderman BJ, et al. FCER2: 3. Sims EJ, Price D, Haughney J, Ryan D, a pharmacogenetic basis for severe exacer- Thomas M. Current Control and Future bations in children with asthma. J Allergy Clin Risk in Asthma Management. Allergy, Immunol. 2007;120(6):1285–91. Asthma Immunol Res. 2011;3(4):217.
9. Davis J., Weiss S, Tantisira K. Asthma 4. Park H.W., Dahlin A, Tse S, Duan Q.L., Pharmacogenomics: 2015 Update. Curr Schuemann B, Martinez FD, et al. Genome Allergy Asthma Rep. 2015;15(7). wide association study of leukotriene
modifier response in asthma. J Allergy Clin 10. Slob EMA, Vijverberg S.J.H, Palmer CNA, Immunol. 2014;133(3):664–9.e5. Zazuli Z, Farzan N, Oliveri NMB, et al.
Pharmacogenetics of inhaled long-acting 5. Tantisira KG, Lasky-Su J, Harada M, beta 2 agonists in asthma: a systematic Murphy A, Litonjua AA, Himes BE, et al. review. Pediatr Allergy Immunol. 2018. Genomewide association between GLCCI1
and response to glucocorticoid therapy in asthma. N Engl J Med. 2011;365(13): 1173–83.
6. Chung L.P., Waterer G, Thompson P.J. Pharmacogenetics of β2 adrenergic re- ceptor gene polymorphisms, long-acting β- agonists and asthma. Clin Exp Allergy. 2011;41(3):312–26.
318
Wetenschappelijke samenvatting
319
Wetenschappelijke samenvatting
Inhalatiecorticosteroïden (ICS) zijn al decennialang de hoeksteen in de behandeling van aanhoudende astma in kinderen. Echter, ondanks hun grote effectiviteit, is er aanzienlijke inter- en intra individuele variabiliteit in respons op de behandeling. Suboptimale controle van astma kan leiden tot aanhoudende astmasymptomen en/of astma exacerbaties. Astma exacerbaties verminderen de kwaliteit van leven en zijn de voornaamste oorzaak van stress in astmatische kinderen en hun ouders/verzorgers. Ernstige exacerbaties kunnen tot ziekenhuisopname of zelfs sterven leiden. Gezien de grote impact van astma exacerbaties op morbiditeit, mortaliteit en gezondheidskosten, is er een dringende behoefte aan het identificeren van (bio)markers. Met name (bio)markers welke patiënten kunnen helpen te identificeren, die een hoger risico hebben op astma exacerbaties, behandelingen kunnen optimaliseren en nieuwe targets.
Omgevingsfactoren (bijvoorbeeld blootstelling aan allergenen, tabaksrook etc.) en klinische/demografische factoren (bijvoorbeeld hoog BMI, jonge leeftijd etc.), alsmede slechte therapietrouw en incorrecte inhalatietechniek, zijn bekende risicofactoren voor astma exacerbaties in kinderen. Bovendien, is er substantiële interesse in de rol van genetische variatie in de heterogeniteit van de behandelrespons. Snelle ontwikkelingen in genetische technologieën en een afname van de kosten voor genotypering in de laatste decennia hebben geleid tot een enorme toename in farmacogenetische studies in astma. Echter, de meeste studies waren beperkt door een lage statistische power en methodologische problemen. Dit heeft geleid tot het feit dat er nog geen genetische marker is welke klaar is voor klinische implementatie.
Omgevingsfactoren (bijvoorbeeld blootstelling aan allergenen, tabaksrook etc.) en klinische/demografische factoren (bijvoorbeeld hoog BMI, jonge leeftijd etc.), alsmede slechte therapietrouw en incorrecte inhalatietechniek, zijn bekende risicofactoren voor astma exacerbaties in kinderen. Bovendien, is er substantiële interesse in de rol van genetische variatie in de heterogeniteit van de behandelrespons. Snelle ontwikkelingen in genetische technologieën en een afname van de kosten voor genotypering in de laatste decennia hebben geleid tot een enorme toename in farmacogenetische studies in astma. Echter, de meeste studies waren beperkt door een lage statistische power en methodologische problemen. Dit heeft geleid tot het feit dat er nog geen genetische marker is welke klaar is voor klinische implementatie.
320
In het eerste deel van dit proefschrift (hoofdstuk 2) hebben we een beschrijvende en een systematische literatuurreview uitgevoerd. Met als doel het identificeren van genomische markers (genetische en epigenetisch) met potentiele klinische waarde.
In hoofdstuk 2.1 leveren we een overzicht van farmacogenetische, epigenetische en transcriptomische studies welke gerelateerd zijn aan kinderastma controle. Dit review heeft laten zien dat de meeste epigenetische en transcriptomische studies in kinderen gefocust zijn geweest op identificerende factoren geassocieerd met astma gevoeligheid en astma fenotypes
De heterogeniteit tussen studie ontwerp, patiënten selectie, klinische uitkomsten en weefsels maakte de vergelijking van de bevindingen complex. Echter, de meeste epigenetische en transcriptomische studies benadrukten het belang van blootstelling aan omgevingsfactoren (bijvoorbeeld tabaksrook, boerderijdieren, virale infecties, huisstofmijt) in astma ontwikkeling met name gedurende de eerste jaren van het leven. Naast de focus op astma gevoeligheid hebben vele van de genetische associaties studies van astma zich gefocust op de invloed van genetische variaties op de respons op de meest gebruikte astma medicaties: ICS, LABA en leukotriënnantagonisten (LTMs).
De meest klinisch relevante genetische marker is geïdentificeerd in kandidaat gen studies voor LABA. Deze variant is een substitutie van het aminozuur Glycine door Arginine op positie 16 van de beta-2 receptor. De invloed van deze mutatie is bevestigd in pediatrische astma populaties. De grootste kandidaat gen studie voor LABA-respons tot nog toe is uitgevoerd in PiCA en bevatte meer dan 4000 astmatische kinderen. Deze meta-analyse liet een toegenomen kans op astma exacerbaties zien voor elk extra risico allel (Odds Ratio (OR) 1.52, 95% betrouwbaarheidsinterval (BI) 1.17-1.99; P=0.002).
In hoofdstuk 2.2 hebben we de resultaten laten zien van een systematisch overzichtsartikel gericht op het verkrijgen van meer inzicht in de farmacogenetica studies naar ICS en LTMs, beide anti-inflammatoire behandelingen voor astma. We hebben laten zien dat meer dan 700 SNPs bestudeerd zijn in farmacogenetische studies maar dat slechts enkele gerepliceerd konden worden. Ondanks succesvolle replicatie in de farmacogenetica van LTMs (MRP1 en ALOX5) en van ICS (CRHR1, FCER2 en GLCC11) waren en inconsistenties in de resultaten van deze studies. Daarnaast werden geen van de SNPS gevonden in kandidaat-gen studies
321
geïdentificeerd in GWAS. Tevens was er slechts één SNP (rs37973, GLCC11) geïdentificeerd in een GWAS en later gerepliceerd in onafhankelijke populaties. Het meest consistente resultaat voor een farmacogenetische interactie voor behandeling met ICS werd gevonden in kandidaat gen studies in verschillende pediatrische populaties voor een SNP (rs28364072) in het FCER2 gen.
In hoofdstuk 3.1 hebben we de karakteristieken van de klinische trials/cohorten die deelnemen aan het PiCA consortium beschreven. PiCA brengt meer dan 21 astma en geboorte cohorten en meer dan 14,000 astmatische kinderen/jongvolwassen van verschillende etnische achtergronden samen. Het consortium omvat een breed spectrum van kinderen/jong volwassenen met mild tot ernstige astma. We hebben laten zien dat op dit moment meer dan 60% van de kinderen/jong volwassenen in PiCA behandeld worden met ICS. Van hen heeft ongeveer een derde gerapporteerd ernstige exacerbaties te hebben in de laatste 6 of 12 maanden of tijdens het eerste jaar van de studie. Dit ondanks het gebruik van ICS-therapie.
In hoofdstuk 3.2 beschrijven wij een grootschalige meta-analyse gericht op het onderzoeken van de associatie tussen de genetische variant rs7216389, gelokaliseerd in de 17q21 locus, en astma exacerbaties ondanks ICS gebruik. Hiervoor hebben we 13 studies uit het PiCA consortium gebruikt. Het effect van rs7216389 is een goed gerepliceerde bevinding geassocieerd met de gevoeligheid voor astma in kinderen. Oorspronkelijk is deze variant geïdentificeerd in een grootschalige GWAS. Wij hebben een significante associatie laten zien tussen het T-allel in de rs7216389 SNP en het risico op astma exacerbaties in kinderen/jong volwassenen, ondanks het gebruik van ICS. Het T-allel veroorzaakte een toename in astma gerelateerde ziekenhuisopnames/ eerste hulp bezoeken (adjusted OR per extra T-allel: 1.32, 95%BI: 1.17-1.49, p<0.0001, n=4,454) en kortdurende orale corticosteroïdentherapie (OCS) (adjusted OR per extra T-allel: 1.19, 95%BI: 1.04-1.36, p=0.01, n=4,050).
Daarnaast hebben wij een sensitiviteitsanalyse uitgevoerd door de patiënten te categoriseren op basis van hun leeftijd (2-4 jaar versus ≥5 jaar) op het moment van de uitkomst (ziekenhuisopname of OCS gebruik). De resultaten van onze sensitiviteitsanalyse lieten een significante associatie zien tussen de SNP en astma exacerbaties in kinderen ≥ 5 jaar. Echter, de resultaten voor jongere kinderen (1-4 jaar) waren niet significant. Wij concluderen dat de kleine groepsgrootte (ongeveer een twintigste van de gehele studie populatie) de reden kan zijn dat wij geen significante associatie hebben gevonden voor deze jongste groep.
322
In een subgroep analyse hebben wij dezelfde associatie gevonden voor ziekenhuisopname/ eerste hulp bezoeken in niet-Latijns Amerikaanse blanken (adjusted OR: 1.33, 95%BI: 1.10-1.61 p=0.004, n=2,888) en Latijns-Amerikanen (adjusted OR: 1.31, 95%BI: 1.06-1.63, p=0.01, n=916). Echter, wij hebben niet dezelfde trend kunnen vinden voor OCS gebruik, sinds de resultaten van de meta- analyse niet statistisch significant waren (adjusted OR: 0.96, 95%BI: 0.76-1.22, p=0.07, n=916 in Latijns-Amerikanen versus adjusted OR: 1.26, 95%BI:1.09-1.45, p=0.002, n=2,492 in niet-Latijns Amerikaanse blanken).
Het lijkt erop dat astma exacerbaties in kinderen en jongvolwassenen die drager zijn van een T-allel op rs7216389 niet adequaat voorkomen worden met de huidige aanbevelingen in de richtlijnen. Derhalve, hogere doseringen van ICS of additionele behandel opties zijn mogelijk nodig om de astma voldoende onder controle te krijgen.
In hoofdstuk 3.3 hebben wij een GWAS meta-analyse om (nieuwe) genetische markers geassocieerd met astma exacerbaties ondanks ICS gebruik. Kinderen en jongvolwassenen in deze analyse waren van gemengde Afrikaanse afkomst en verkregen uit twee PiCA studies (GALA II en SAGE II).
Wij hebben gevonden dat 15 onafhankelijke varianten suggestief geassocieerd waren met astma exacerbaties ondanks ICS gebruik in Afrikaanse kinderen van gemengde afkomst (n=1347, leeftijden van 8 tot 21 jaar in beide studies, p<5x10-6). Rs5995653, gelokaliseerd in de intergenic regio van APOBEC3B en APOBEC3C, was positief gerepliceerd in astmatische niet-Latijns Amerikaanse blanken in het PiCA cohort (n=1,699, p=0.01). Vergeleken met astmatische niet-Latijns Amerikaanse blanken was het beschermende effect van het A-allel in rs5995653 sterker in gemengde Afrikaanse populaties (OR: 0.66, 95%BI:0.56-0.79 in gemengd Afrikaans versus OR: 0.77, 95%BI: 0.63-0.94 in Europeanen). Naast astma exacerbaties was de SNP ook geassocieerd met veranderingen in longfunctie na 6-8 van een behandeling met ICS in het SLOVENIA-cohort, welke ook astmatische niet-Latijns Amerikaanse blanken bevatte (p=7.54x10-4). Tot op heden zijn er geen astma gerelateerde functies geassocieerd met enige van de APOBEC2 flankerende genen. Gebaseerd op het functionele bewijs dat we hebben verkregen van HaploReg v4.1 voor rs5995653, lijkt het dat deze SNP een vitale rol speelt in het reguleren van expressie van genen die betrokken zijn in verschillende cellulaire processen in de longen. De voornaamste functie van deze genen is gerelateerd aan de aangeboren immuniteit tegen endogene retro-elementen en exogene virussen. Virale infecties zijn verantwoordelijk voor de meerderheid van
323
de exacerbaties in astmatische kinderen. Als remmers van virale infecties kan de expressie van APOBEC3B en APOBEC3C in longweefsel betrokken zijn in het vechten tegen deze virale infecties. Daarom kunnen veranderingen in de expressie van deze genen invloed hebben op het risico op astma exacerbaties ondanks dat patiënten behandeld worden met ICS.
Daarnaast, een eerder gerapporteerde associatie van de L3MBTL4-ARHGAP28 locus in niet-Spaanse blanken en Aziatische patiënten werd bevestigd in ons gemengd- Afrikaanse cohort. Dit is de eerste GWAS van ICS-respons in LatijnsAmerikaanse en Afrikaans-Amerikaanse kinderen en jongvolwassenen met astma.
In hoofdstuk 4.1 wilden we de genetische en niet-genetische factoren identificeren die geassocieerd zijn met een verhoogd risico op astma exacerbaties in kinderen die ICS gebruiken. We hebben deze studie uitgevoerd in een cohort van kinderen die volgens de apotheekgegevens astma medicatie gebruiken (PACMAN studie). Een van de doelen van de studie was het maken van een genetische risicoscore om de toegevoegde waarde van (farmaco-) genetische varianten te bepalen ten opzichte van alleen niet-genetische factoren. Alleen geen van de genetische factoren was significant geassocieerd met astma exacerbaties ondanks het gebruik van ICS. Patienten met voedselallergie hadden een ongeveer twee keer verhoogd risico op exacerbaties vergeleken met kinderen zonder voedselallergie. (adjOR 2,05, 95% BI: 1.20-3.49, p=0,008) . Een zwakke maar significante relate werd gevonden tussen leeftijd en het risico op exacerbaties. Elk toegevoegd levensjaar was geassocieerd met een 14% toename in het risico op astma exacerbaties (adjOR=0.86, 95% BI: 0.76- 0.97, p=0.01). Onze advies was dat kinderen met astma, die ook last hebben van voedselallergie baat zouden kunnen hebben bij een meer intensieve monitoring om het risico op exacerbaties te verkleinen.
Ten slotte, in hoofdstuk 5.1 hebben wij een economische evaluatie uitgevoerd om te onderzoeken of een farmacogenetica gestuurde behandelstrategie kosteneffectief is voor kinderen die een intensievere behandeling nodig hebben dan lage dosering ICS. Voor dit doel hebben wij een kosteneffectiviteitsstudie uitgevoerd waarbij wij ADRB2- gestuurde behandeling (rs1042713) hebben vergeleken met de standaardbehandeling, voor het starten met lang werkende beta2 agonisten (LABA). Tot op heden is de rs1042713 de meest consistent gerepliceerde variant in farmacogenetica studies van astma. Wij hebben laten zien dat genotyperen voor het starten van LABA-therapie het aantal exacerbaties significant kan laten verminderen.
324
Tevens waren de totale kosten voor de genotype-gestuurde behandeling 151 euro minder dan de standaardbehandeling. Vergeleken met de standaardtherapie was een farmacogenetisch gestuurde therapie dan ook kostenbesparend.
In hoofdstuk 6 staat de discussie van dit proefschrift waarin we onze bevindingen bediscussiëren en aanbevelingen doen voor toekomstig onderzoek en waarin we de betekenis van onze bevindingen voor de klinische praktijk beschrijven.
Onze belangrijkste conclusies zijn dat het belangrijk is om niet langer naar een enkele biomarker te kijken, maar om een multidimensionale aanpak te kiezen waarbij een combinatie van biomarkers wordt bestudeerd. Deze aanpak kan uiteindelijk leiden tot het ontwikkelen van klinisch bruikbare algoritmes die het mogelijk maken om sub fenotypes van astma te diagnosticeren en de behandeling van astma te optimaliseren.
325
Appendices
326
Acknowledgements
327
Undertaking this PhD has been a truly life-changing experience for me. My heart is filled with wonderful memories, and this experience would not have been possible without the support and guidance of my supervisors, colleagues, friends and family. Hence, I would like to take this opportunity to show my gratitude to those who have assisted me in a myriad of ways.
I would like to express my deepest gratitude to my promotors and co-promotor for guiding me through this amazing experience. Dear Anke-Hilse, thank you for giving me the opportunity to pursue the dream of becoming a researcher. I am grateful for your constant support, guidance and constructive criticism. Thank you for your patience, you know how much I grew during the past four years. I would not be here today if it was not for your support and patience. Thank you for meeting me every week despite your busy schedule and for inspiring me with your amazing ideas. I always enjoyed our weekly meetings and I will truly miss them.
Dear Susanne, thank you for guiding me, often with big doses of patience, through the subtleties of scientific writing. I appreciate your attention to detail and your willingness to help. Thank you for your efficient and quick feedback on my draft manuscripts and thank you for helping me become a better researcher.
Dear Jan, although you are no longer with us, you continue to inspire me and I feel very lucky and grateful to have met you and to have you as my promotor. Thank you for teaching me the most valuable lesson in life: No matter what never forget to smile. I will always remember you.
I would like to say a very big thank you to Dr. Hadi Hamishekar for all the support and encouragement he gave me to start this journey. Your contagious and motivational enthusiasm for science and more importantly your passion for helping others encouraged me to push my boundaries to become a researcher and a better person.
My sincere appreciation goes to the members of my reading committee for the time they have spent reading my thesis: prof. dr. Bel, prof. dr. Sterk, dr. Terheggen, prof. dr. Bartlett-Esquilant and prof. dr. Wilffert.
I would like to thank all my co-authors and all the PiCA members for their valuable contributions to the work in this thesis. It would not have been possible without your valuable suggestions, comments and ideas.
A very big thanks to Maria, Natalia, Leila and Patricia for their help and friendship during these four years. Dear Maurik, thank you for your help on chapter 5. I always enjoyed our meetings and our brainstorming sessions. Dear Anke, it was great to be able to work together and I am deeply grateful for your guidance, patience and help on chapter 5.
328
My immense gratitude goes to Marije, Yared and Maaike for being my paranimfen, thank you for organizing my promotion and for your support and amazing ideas.
A special thanks to all my friends and colleagues at Utrecht University for the good moments that we had together: Alfi, Hedy, Mohammed, Ruben, Heleen, Heshu, Turki, YuMao, Jamal, Hamid, Sulmaz, Claudia, Sander, Gert-Jan, Yaser, Sanni, Svetlana, Paul, Niels, Maarten, Fawaz, Lydia, Jarno, Mariette, Joost, Laurens, Renate, Teresa, Rianne, Ali, Ineke, Anja, Suzanne, Fariba, Neda, Negar, Marzi, Maripaz, Amr, Lucia and my dear Sjaak.
I would like to express my heartfelt gratitude to my friends and colleagues at the AMC: Job, Reim, Levi, Olga, Paul, Guus, Peter-Paul, Anirban, Rianne, Zulfan, Simone, Hanneke, Julia, Lizzy, Marianne, Katerine, Mahmoud, Elise, Anne, Jacquelien, Henny, Pearl, Cristina and Luca. Thank you all for your support, help and friendship and thank you for the lovely chats and discussions during our short (wink wink) coffee breaks.
Dear Armina, Delphi, Kat & Marcel, Yared & Artem, thank you for all those unforgettable and delightful moments, fun discussions, laughs and mostly your friendship.
There are moments in our lives that we feel lost and vulnerable, especially when we lose our loved ones. I cannot thank enough my dear friends Naghme, Maaike, Yared, Marije and Annika; I will never forget your emotional support during the most difficult times of my life. Thank you for cheering me up and for being there for me and for tolerating me on my worst moments. Thank you for your kind hearts and generosity. Also, thank you for the fun girls’ night (and day) outs :D.
My acknowledgement would be incomplete without thanking the biggest source of my strength, my family. My dear baba, how I wish you were still here. Thank you for always believing in me and for encouraging me to learn, thank you for raising me with a love of science and supporting me in all my pursuits. I could not have been here without your guidance, courage and unconditional love. Thank you Maman, for your unselfish love and support and for putting up with me in difficult moments where I felt confused and thank you for guiding me through life and helping me follow my dreams. You have been an outstanding inspiration to me. Getting to this stage in my life has taken a lot of work, but it is nothing compared to how you worked and sacrificed for me. Without your support, I would not have been where I am today.
329
CURRICULUM VITAE
Niloufar Farzan
Date of birth: January 13, 1990
Place of birth: Tabriz, Iran
Education:
2014 –2016 PhD candidate at division of Pharmacoepidemiology & Clinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, Netherlands. 2008 –2014 Faculty of Pharmacy, Tabriz University of Medical sciences, Tabriz, Iran.
2001- 2008 Farzanegan (National Organization for Development of Exceptional Talents (NODET)), Tabriz, Iran.
330
List of publications
1. Pharmacogenomics of inhaled corticosteroids and leukotriene modifiers: A systematic review. Farzan N, Vijverberg SJ, Arets HG, Raaijmakers JAM & Maitland-van der Zee AH. Clinical and Experimental Allergy (CEA), 2017;47(2):271-293.
2. Rationale and design of the multiethnic Pharmacogenomics in Childhood Asthma consortium. Farzan N, Vijverberg SJ, Andiappan AK, Arianto L, Berce V, Blanca-López N, Bisgaard H, Bønnelykke K, Burchard EG, Campo P, Canino G, Carleton B, Celedón JC, Chew FT, Chiang WC, Cloutier MM, Daley D, Den Dekker HT, Dijk FN, Duijts L, Flores C, Forno E, Hawcutt DB, Hernandez-Pacheco N, de Jongste JC, Kabesch M, Koppelman GH, Manolopoulos VG, Melén E, Mukhopadhyay S, Nilsson S, Palmer CN, Pino-Yanes M, Pirmohamed M, Potočnik U, Raaijmakers JA, Repnik K, Schieck M, Sio YY, Smyth RL, Szalai C, Tantisira KG, Turner S, van der Schee MP, Verhamme KM, Maitland-van der Zee AH. Pharmacogenomics, 2017;18(10):931-943
3. The use of pharmacogenomics, epigenomics and transcriptomics to improve childhood asthma management: where do we stand? Farzan N, Vijverberg SJ, Kabesch M, Sterk PJ, & Maitland van der Zee AH. Pediatric Pulmonology, 2018;53(6):836-845.
4. 17q21 variant increases the risk of exacerbations in asthmatic children despite inhaled corticosteroids use. Farzan N, Vijverberg S.J, Hernandez-Pacheco N, Bel E.H.D, Berce V, Bønnelykke K, Bisgaard H, Burchard E.G, Canino G, Celedón J.C, Chew F.T, Chiang WC, Cloutier MM, Forno E, Francis B, Hawcutt DB, Kabesch M, Karimi L, Melén E, Mukhopadhyay S, Merid S.K, Palmer CN, Pino-Yanes M, Pirmohamed M, Potočnik U, Repnik K, Schieck M, Sevelsted A, Sio YY, Smyth RL, Soares P, Söderhäll C, Tantisira KG, Tavendale R, Tse SM, Turner S, Verhamme K.M, Maitland-van der Zee A.H. Allergy, 2018;73(10):2083-2088.
5. Treatment response heterogeneity in asthma: the role of genetic variation. Vijverberg SJ, Farzan N, Slob EMA, Neerincx AH & Maitland-van der Zee AH. Expert Review of Respiratory Medicine, 2018;12(1):55-65
6. Pharmacogenetics of inhaled long-acting beta2-agonists in asthma: a systematic review. Slob EMA, Vijverberg SJ, Palmer CN, Zazuli Z, Farzan N, Oliveri NMB, Pijnenburg MW, Koppelman GH, Maitland-van der Zee AH. Pediatr Allergy Immunol. 2018.
331
PhD training Year ECTs
General courses
Introduction to epidemiology 2015 1.5
Introduction to statistics 2015 1.5
Introduction to R 2015 1.5
Pharmacoeconomics 2015 1.5
Genetic epidemiology 2016 1.1
Systems Medicine 2017 1.4
The national lung course 2017 5
Total 13.5
International conferences International Society for Pharmacoepidemiology (ISPE) congress 2017, 2018
EAACI Congress 2016, 2017
ERS international congress 2017
4th Pediatric allergy and asthma meeting (PAAM) 2015
Awards EAACI Congress Scholarship 2017
The best abstract prize at the EAACI congress 2017
The ICPE 34 attendance scholarship 2018
332
333
334