Clinical and genetic differences between pustular psoriasis subtypes Sophie Twelves, MRes,a Alshimaa Mostafa, MD,b,c Nick Dand, PhD,a Elias Burri, MBBS,b Katalin Farkas, PhD,d Rosemary Wilson, BA,e Hywel L. Cooper, BM,f Alan D. Irvine, MD, DSc,g Hazel H. Oon, MD,h Ku € lli Kingo, MD, PhD,i Sulev Ko € ks, PhD,j Ulrich Mrowietz, MD,k Luis Puig, MD, PhD,l Nick Reynolds, MD,m Eugene Sern-Ting Tan, MBBS,h Adrian Tanew, MD,n Kaspar Torz, MD,k Hannes Trattner, MD,n Mark Valentine, MD,o Shyamal Wahie, MD,p Richard B. Warren, MB ChB, PhD,q Andrew Wright, MB ChB,r Zsuzsa Bata-Cso € rgo } , MD,s Marta Szell, DSc,t Christopher E. M. Griffiths, MD, A. David Burden, MD, Siew-Eng Choon, MBBS,v Catherine H. Smith, MD,e q u Jonathan N. Barker, MD,e* Alexander A. Navarini, MD, PhD,b* and Francesca Capon, PhDa* London, Portsmouth, Newcastle upon Tyne, Durham, Darlington, Manchester, Bradford, and Glasgow, United Kingdom; Zurich, Switzerland; Beni Suef, Egypt; Szeged, Hungary; Dublin, Ireland; Singapore; Tartu, Estonia; Kiel, Germany; Barcelona, Spain; Vienna, Austria; Seattle, Wash; and Johor Bahru, Malaysia GRAPHICAL ABSTRACT Clinical and genetic differences between pustular psoriasis subtypes 863 pustular psoriasis patients Generalised Pustular Psoriasis, n=251 Acrodermatitis Continua of Hallopeau, n=28 PalmoPlantar Pustulosis, n=560 Multiple diagnoses, n=24 Clinical features Genetic features whole cohort, n = 863 cohort subset, n = 475 **** *** **** 100 0.4 *** ** *** c.115+6T>C **** Age of ons e t (yrs) 100 100 No PV c.115+6T>C; Allele frequency **** ** Perc ent age of cases 80 0.3 p.Pro76Leu PV Mean age (yrs) 80 80 p.Ser113Leu 60 Other 0.2 60 60 mutations 40 40 40 0.1 20 20 20 0 0.0 0 1 2 an So M ian ro ian Eu A s y IL36RN h la pe 0 0 ut a s A disease alleles P PP P st H H PP PP PP Ea C C G G A A ACH: acrodermatitis continua of Hallopeau; GPP: generalised pustular psoriasis; PPP: palmoplantar pustulosis; PV: psoriasis vulgaris From athe Department of Medical and Molecular Genetics, School of Basic and Medical NHS Foundation Trust, Newcastle upon Tyne; nthe Department of Dermatology, Med- Biosciences, King’s College London; bthe Department of Dermatology, University ical University of Vienna; othe Division of Dermatology, University of Washington Hospital Zurich; cthe Department of Dermatology, Beni Suef University; dthe Depart- School of Medicine, Seattle; pUniversity Hospital of North Durham and Darlington ment of Medical Genetics, University of Szeged; eSt John’s Institute of Dermatology, Memorial Hospital; qthe Dermatology Centre, Salford Royal Hospital, University of School of Basic and Medical Biosciences, King’s College London; fthe Portsmouth Manchester and the Academic Health Science Centre, Manchester; rSt Lukes Hospital, Dermatology Unit, Portsmouth Hospitals Trust; gPaediatric Dermatology, Our Lady’s Bradford, and the Centre for Skin Science, University of Bradford; sthe MTA-SZTE Children’s Hospital Crumlin, and Clinical Medicine, Trinity College Dublin; hthe Dermatological Research Group, Szeged, and the Department of Dermatology and Al- Department of Dermatology, National Skin Centre, Singapore; ithe Department of lergology, University of Szeged; tthe MTA-SZTE Dermatological Research Group, Dermatology, University of Tartu, and the Clinic of Dermatology, Tartu University Szeged, and the Department of Medical Genetics, University of Szeged; uthe Institute Hospital; jthe Department of Pathophysiology, University of Tartu; kthe Psoriasis Cen- of Infection, Inflammation and Immunity, University of Glasgow; and vthe Department ter at the Department of Dermatology, University Medical Center, Schleswig-Holstein, of Dermatology, Hospital Sultanah Aminah, Jeffrey Cheah School of Medicine and Campus Kiel; lthe Department of Dermatology, Hospital de la Santa Creu i Sant Pau, Health Sciences, Monash University Malaysia, Johor Bahru. Barcelona; mthe Institute of Cellular Medicine, Medical School, Newcastle University *These authors contributed equally to this work. and the Department of Dermatology, Royal Victoria Infirmary, Newcastle Hospitals 1021 1022 TWELVES ET AL J ALLERGY CLIN IMMUNOL MARCH 2019 Background: The term pustular psoriasis indicates a group of severe skin disorders characterized by eruptions of neutrophil- Abbreviations used filled pustules. The disease, which often manifests with ACH: Acrodermatitis continua of Hallopeau concurrent psoriasis vulgaris, can have an acute systemic ERASPEN: European Rare and Severe Psoriasis Expert Network (generalized pustular psoriasis [GPP]) or chronic localized GPP: Generalized pustular psoriasis PPP: Palmoplantar pustulosis (palmoplantar pustulosis [PPP] and acrodermatitis continua of PV: Psoriasis vulgaris Hallopeau [ACH]) presentation. Although mutations have been uncovered in IL36RN and AP1S3, the rarity of the disease has hindered the study of genotype-phenotype correlations. Objective: We sought to characterize the clinical and genetic features of pustular psoriasis through the analysis of an Although these defects are observed mostly in the homozygous or extended patient cohort. compound heterozygous state, a number of patients carrying single Methods: We ascertained a data set of unprecedented size, heterozygous changes have also been reported.4 including 863 unrelated patients (251 with GPP, 560 with PPP, Disease alleles associated with GPP have been identified 28 with ACH, and 24 with multiple diagnoses). We undertook subsequently in AP1S3 (encoding a subunit of the adaptor protein mutation screening in 473 cases. 1 complex)5 and CARD14 (encoding a keratinocyte nuclear factor Results: Psoriasis vulgaris concurrence was lowest in PPP (15.8% kB adaptor protein).6 Of note, IL36RN, CARD14, and AP1S3 vs 54.4% in GPP and 46.2% in ACH, P < .0005 for both), whereas mutations have also been described in patients with PPP and those the mean age of onset was earliest in GPP (31.0 vs 43.7 years in PPP with ACH, demonstrating a shared genetic basis for pustular forms and 51.8 years in ACH, P <.0001 for both). The percentage of female of psoriasis.5,7,8 Patients harboring disease alleles at 2 distinct loci patients was greater in PPP (77.0%) than in GPP (62.5%; (IL36RN and AP1S3; IL36RN and CARD14) have also been P 5 5.8 3 1025). The same applied to the prevalence of smokers reported.9,10 Thus an increasingly complex picture is emerging (79.8% vs 28.3%, P < 10215). Although AP1S3 alleles had similar with evidence of substantial genetic heterogeneity, pleiotropy (the frequency (0.03-0.05) across disease subtypes, IL36RN mutations phenomenon whereby a single gene can influence more than 1 trait), were less common in patients with PPP (0.03) than in those with and digenic inheritance. GPP (0.19) and ACH (0.16; P 5 1.9 3 10214 and .002, respectively). In this context analysis of genotype-phenotype correlations Importantly, IL36RN disease alleles had a dose-dependent effect on would facilitate stratification of patient cohorts and streamline the age of onset in all forms of pustular psoriasis (P 5 .003). genetic diagnosis of disease subtypes. However, rigorous studies Conclusions: The analysis of an unparalleled resource revealed have been hindered by the rarity of pustular psoriasis, which has key clinical and genetic differences between patients with PPP and prevented the ascertainment and standardized phenotyping of those with GPP. (J Allergy Clin Immunol 2019;143:1021-6.) sizeable patient resources. Here we sought to address this issue through formation of a Key words: Generalized pustular psoriasis, palmoplantar pustulo- multicenter consortium. We brought together 8 tightly pheno- sis, acrodermatitis continua of Hallopeau, IL36RN, AP1S3, geno- typed patient cohorts through a collaboration with the European type-phenotype correlation Rare and Severe Psoriasis Expert Network (ERASPEN). This enabled us to ascertain a unique clinical resource, including 863 The term pustular psoriasis refers to a group of severe unrelated cases and exceeding by nearly 3-fold the size of any inflammatory skin disorders manifesting with repeated eruptions published data set. Analysis of this extended cohort revealed very of painful neutrophil-filled pustules. These conditions can present significant differences in the clinical and genetic features of with acute episodes of skin pustulation and systemic upset pustular psoriasis subtypes. Specifically, it demonstrated that PPP (generalized pustular psoriasis [GPP]) or chronic pustular differs from ACH and GPP in terms of patients’ demographics, eruptions that affect the palms and soles (palmoplantar pustulosis disease presentation, and underlying genetic abnormalities. [PPP]) or the tips of fingers and toes (acrodermatitis continua of Hallopeau [ACH]). Of note, all forms of the disease can be METHODS complicated by concurrent psoriasis vulgaris (PV).1 Patient ascertainment We and others have shown that mutations of the gene encoding This research was carried out in accordance with the principles of the the IL-36 receptor antagonist (IL36RN) are associated with GPP.2,3 Declaration of Helsinki and was approved by the ethics committees of Supported by the Department of Health through the National Institute for Health Received for publication February 2, 2018; revised May 14, 2018; accepted for publica- Research (NIHR) BioResource Clinical Research Facility and comprehensive tion June 15, 2018. Biomedical Research Centre awards to Guy’s and St Thomas’ NHS Foundation Trust Available online July 21, 2018. in partnership with King’s College London and King’s College Hospital NHS Corresponding author: Jonathan N. Barker, MD, St John’s Institute of Dermatology, Foundation Trust (guysbrc-2012-1) and to the NIHR-Newcastle Biomedical Research School of Basic and Medical Biosciences, King’s College London, 9th floor Tower Centre. This work was funded by a Medical Research Council (MRC) Stratified Wing Guy’s Hospital, London SE1 9RT, United Kingdom. E-mail: jonathan. Medicine award (MR/L011808/1; to J.N.B., F.C., and C.H.S.) and by the Efficacy and barker@kcl.ac.uk. Mechanism Evaluation (EME) Programme, an MRC and NIHR partnership (grant The CrossMark symbol notifies online readers when updates have been made to the EME 13/50/17 to C.H.S., F.C., J.N.B., C.E.M.G., and N.R.). N.R. is also supported by article such as errata or minor corrections the Newcastle MRC/EPSRC Molecular Pathology Node. S.T. is supported by the 0091-6749 King’s Bioscience Institute and the Guy’s and St Thomas’ Charity Prize PhD Ó 2018 The Authors. Published by Elsevier Inc. on behalf of the American Academy of Programme in Biomedical and Translational Science. The European Rare and Severe Allergy, Asthma & Immunology. This is an open access article under the CC BY li- Psoriasis Expert Network is funded by a PPRC grant from the European Association of cense (http://creativecommons.org/licenses/by/4.0/). Dermatology and Venereology (EADV; to A.A.N. and J.N.B.). The views expressed in https://doi.org/10.1016/j.jaci.2018.06.038 this publication are those of the authors and not necessarily those of the MRC, NHS, NIHR, or Department of Health. J ALLERGY CLIN IMMUNOL TWELVES ET AL 1023 VOLUME 143, NUMBER 3 TABLE I. Summary description of the patient cohort Ethnicity Sex Clinical diagnosis European Asian African Other* Female Male Unknown ACH PPP GPP ACH 1 GPP ACH 1 PPP GPP 1 PPP Total Total 591 161 78 33 620 233 10 28 560 251 9 4 11 863 *Includes unknown ethnicity (n 5 19), mixed ethnicity (n 5 4), and Middle Eastern (n 5 4), Finnish (n 5 2), Filipino (n 5 1), Hispanic (n 5 1), Jamaican (n 5 1), and Romani (n 5 1) ethnicity. participating institutions. Written informed consent was also obtained from all features, as observed in clinical practice. We analyzed 863 participants. The study aligned 8 patient cohorts (n 5 863) recruited in the unrelated patients, the majority of whom (823/863 [95.4%]) reference centers listed in Table E1 in this article’s Online Repository at were recruited through 6 European, 1 North African, and 1 Asian www.jacionline.org. The largest resource (n 5 255 British and Irish cases) reference center (Table I and see Table E1). Of note, key patients’ was provided by St John’s Institute of Dermatology (London, United Kingdom) demographics (male/female ratios and mean age of onset for and combined a historical data set (n 5 177) with patients ascertained prospectively (n 5 78) through the Anakinra in Pustular Psoriasis, various disease types) were comparable across these Response in a Controlled Trial (APRICOT) clinical trial (EudraCT no. cohorts (see Table E2 in this article’s Online Repository at 2015-003600-23) and its sister mechanistic study, Pustular Psoriasis, www.jacionline.org). Elucidating Underlying Mechanisms (PLUM). An additional 40 affected While patients with GPP (251/863 [29.1%]) and PPP (560/863 subjects (listed as ‘‘others’’ in Table E1) were recruited outside the main [64.9%]) accounted for most of the data set, the ACH sample was reference centers by clinicians who sent individual samples to the ERASPEN substantially smaller (28/863 [3.2%]), reflecting the extreme Consortium or St John’s Institute of Dermatology. rarity of this condition. Of note, the concurrence of multiple Pustular psoriasis was diagnosed by expert dermatologists based on direct disease forms (most notably GPP with ACH and GPP with PPP) clinical examination, with the ERASPEN consensus criteria1 used in at least was reported in a small percentage of affected patients (24/863 506 cases. The observation of primary, sterile, macroscopically visible [2.8%]). pustules affecting nonacral skin (GPP), palms/soles (PPP), or the nail apparatus (ACH) was the main inclusion criterion. Conversely, the occurrence A number of comorbidities were observed, with diabetes and of pustules restricted to the edges of psoriatic plaques represented an exclusion hypertension figuring most prominently, regardless of the criterion. patient’s ethnicity (see Table E3 in this article’s Online Repository at www.jacionline.org). In keeping with published associations,15 we also found that 11 (3.9%) of 281 European Mutation screening IL36RN, AP1S3, and CARD14 mutations were screened by using Sanger patients with PPP had autoimmune thyroid disease. sequencing in 473 patients for whom DNA was available. Primer sequences and Mean age of onset differed considerably across disease types cycling conditions have been described elsewhere.3,5,6 Nucleotide substitutions and was lower in patients with GPP (31.0 6 19.7 years) than in were identified by using Sequencher 4.9 (Gene Codes, Ann Arbor, Mich). The those with PPP (43.7 6 14.4, P 5 9.3 3 10219) and those deleterious effect of the newly identified c.11515G>A mutation was confirmed with ACH (51.8 6 20.4, P 5 1.2 3 1027; Fig 1, A, and see by using Spliceman and MaxEntScan,11,12 whereas the pathogenic potential of Table E2). Despite these marked differences, there was CARD14 alleles was assessed with Combined Annotation Dependent Depletion substantial heterogeneity within the individual disease cohorts, (CADD).13 with very early-onset (<10 years) and very late-onset (>70 years) cases observed in all forms of pustular psoriasis. Statistics Although the prevalence of PV in the overall data set (29.1%) The clinical and demographic characteristics of study participants were was much greater than that reported for the general population analyzed by using a binomial test (to establish the presence of a sex bias (2% to 3%), concurrence rates varied among disease variants. In among patients with pustular psoriasis), the x2 test with the Yates correction particular, the frequency of PV among patients affected by PPP (to analyze differences in the prevalence of PV and proportion of affected (15.8%) was significantly lower than that seen in the ACH female subjects across disease types), and a Kruskal-Wallis test followed by (46.2%, P 5 .0004) and GPP (54.4%, P 5 2.2 3 10216) groups the Dunn multiple comparison test (to analyze differences in age of onset (Fig 1, B). Although the latter result was driven in part by a between PPP, ACH, and GPP cases). Analysis of genetic data was based on a x2 test with the Yates correction (to compare the frequency of disease alleles very high prevalence of PV among Malaysian patients with in PPP, ACH, and GPP cases and the combined prevalence of IL36RN GPP (see Table E2), the difference remained significant mutations across ethnic groups) and a 1-tailed Fisher exact test (for association (P 5 .01) when the sizeable Malaysian cohort (n 5 138) was between the IL36RN p.Ser113Leu allele and PPP). Genotype-phenotype removed from the analysis. Thus our investigations have correlations were investigated by implementing logistic (for PV concurrence demonstrated key differences between disease subtypes, and sex ratios) and linear (for age of onset) regression analysis with disease highlighting PPP as a late-onset condition with low PV subtype as a covariate. All tests were implemented in R software.14 concurrence. Patients with multiple diagnoses were excluded from all statistical analyses because they could not be assigned to a single disease group. PPP is the form of pustular psoriasis most RESULTS influenced by sex and smoking status Age of onset and PV concurrence rates vary It has been reported that female patients and smokers are at significantly among disease subtypes greater risk of PPP than male patients and nonsmokers.16 Here we As members of the ERASPEN network, we previously defined observed a degree of sex bias in all forms of pustular psoriasis as consensus criteria for the diagnosis of pustular psoriasis.1 Here we the female/male ratio was 1.5 in patients with ACH, 1.7 in patients build on this work to describe the presentation of key disease with GPP, and 3.5 in patients with PPP. The distortion in sex ratios 1024 TWELVES ET AL J ALLERGY CLIN IMMUNOL MARCH 2019 A B **** did not (12.4% vs 1.6%, P 5 .009), suggesting that cigarette **** *** smoking can modulate disease manifestations. In fact, smoking 100 **** 100 No PV has a well-documented effect on aryl hydrocarbon **** PV receptor signaling,17 a pathway that modulates the severity of Percentage of cases 80 80 Mean age (yrs) inflammation in psoriatic skin.18 60 60 Although the ACH sample was too small for analysis, the 40 40 percentage of smokers in the GPP data set (26/96 [28.3%]) was significantly less than that observed in patients with PPP 20 20 (P < 10215), indicating that the adverse effect of cigarette 0 0 smoking is specific to the latter condition. P PP H P H PP PP PP C C G G A A Definition of a patient subset for genetic analysis C D Having investigated the key clinical manifestations of pustular psoriasis, we sought to define their relationship with the patient’s **** *** 100 Male 0.4 *** c.115+6T>C genotype. For this purpose, we examined the mutation status Female c.115+6T>C; of 473 affected subjects for whom DNA was available (see Percentage of cases 80 ** Allele frequency 0.3 p.Pro76Leu Table E1). We collated genetic data previously generated by our p.Ser113Leu 60 Other mutations group (n 5 358)4-6,9 while also examining 115 newly recruited 0.2 40 cases. Importantly, Table E4 in this article’s Online Repository at www.jacionline.org shows that the patient subset screened for 0.1 20 mutations is representative of the broader data set, suggesting that the findings obtained in this sample can be generalized to 0 0.0 the whole resource. PP H P an y n n ut la sia Eu sia PP C pe G So Ma A A A ro st h Ea E Frequency of IL36RN mutations differentiates PPP 100 from ACH and GPP ** The IL36RN coding sequence and exon/intron junctions were Age of onset (yrs) 80 screened in the entire patient resource, uncovering 66 patients 60 (4 with ACH, 45 with GPP, 12 with PPP, and 5 with multiple diagnoses) with disease alleles (Table II and see Table E5 40 in this article’s Online Repository at www.jacionline.org). 20 Thirty-six of these subjects harbored biallelic (homozygous/ compound heterozygous) changes, with the remaining 30 0 carrying monoallelic (single heterozygous) variants. All the 0 1 2 IL36RN disease alleles observed mutations had been described previously, except for a c.11515G>A splicing variant uncovered in a North American FIG 1. Features of pustular psoriasis observed in the disease cohort. patient with GPP (see Table E5). A, Mean age of onset was compared across disease groups by using a Kruskal-Wallis test followed by the Dunn multiple comparison test. IL36RN disease alleles were present in a variety of ethnic B, Differences in PV concurrence were analyzed with a x2 test. C, Differences groups, with the greatest prevalence observed among patients of in the proportion of affected female subjects were assessed by using a x2 European (34.7%) and East Asian (28.8%) descent (Fig 1, D). test. The dashed line indicates the percentage of female subjects in the Although we did not detect any rare changes in the 21 South Asian general population. D, Differences in combined frequency of IL36RN cases we examined, a homozygous p.Leu21Pro mutation has been mutations were assessed across ethnic groups by using a x2 test. Pairwise comparisons were undertaken with the Fisher exact test. The analysis was described in a Pakistani GPP pedigree,19 suggesting that restricted to patients with GPP because this is the only group for which data deleterious IL36RN alleles can also be found within the Indian were available for multiple ethnicities. Other mutations indicates alleles subcontinent. seen only once in the cohort. The notation c.11516T>C; p.Pro76Leu refers The proportion of subjects harboring IL36RN disease alleles to patients carrying the 2 variants on the same haplotype. E, Effects of IL36RN mutations on age of onset were assessed by using linear was greater in GPP and ACH (23.7% and 18.2%, respectively) regression. **P < .01, ***P < .001, and ****P < .0001. compared to PPP (5.2%). Patients with GPP and those with ACH were also more likely to carry biallelic mutations compared to individuals affected by PPP (see Table E5). As a result, the observed in GPP and PPP was statistically significant (P < 1025 prevalence of IL36RN mutations was significantly increased in and P < 10215, respectively) and readily recognizable in patients with GPP (0.19) and ACH (0.16) compared with that in individual cohorts (see Table E2). Of note, the difference between patients with PPP (0.03; P 5 1.9 3 10214 and .0018, respectively; the PPP and GPP female/male ratios was also significant Table II). Nonetheless, the association between IL36RN (P 5 5.8 3 1025), highlighting PPP as the condition most mutations and PPP, which has been recently questioned,10 was influenced by sex-related factors (Fig 1, C). statistically significant. In fact, an analysis of the recurrent In our data set 79.8% (249/312) of patients with PPP for whom p.Ser113Leu variant showed that its frequency in British data were available were current or past smokers. Of interest, the patients was almost 10 times greater than that observed in rate of PV concurrence was much greater in patients with PPP population-matched control subjects (P 5 9.3 3 1028; odds ratio, who smoked (or had done so in the past) compared with those who 10.8; 95% CI, 5.3-22.0; Table III). J ALLERGY CLIN IMMUNOL TWELVES ET AL 1025 VOLUME 143, NUMBER 3 TABLE II. IL36RN and AP1S3 mutation frequencies across disease types ACH GPP PPP Multiple diagnoses No. of IL36RN-positive patients* 4/23 (17.4%) 45/190 (23.7%) 12/234 (5.1%) 5/18 (27.8%) IL36RN mutation count (frequency) 7/46 (0.15) 72/380 (0.19) 15/468 (0.03) 8/36 (0.22) No. of AP1S3-positive patients* 2/19 (10.5%) 4/37 (10.8%) 14/212 (6.6%) 4/11 (36.4%) AP1S3 mutation count (frequency) 2/38 (0.05) 4/74 (0.05) 14/424 (0.03) 4/22 (0.18) *Patients were classified as ‘‘positive’’ if they were carrying at least 1 mutation at the examined locus. p.Phe4Cys and p.Arg33Trp mutations have no frequency in East Asian populations and therefore were not screened in patients from this ethnic group. TABLE III. Association between IL36RN p.Ser113Leu and PPP Repository at www.jacionline.org). No additional mutations were p.Ser113Leu WT observed in the subjects who were screened for the entire coding Cases* 11 (3.6%) 291 (96.4%) region. Of note, 3 patients (2 with GPP and 1 with PPP) carried Control subjects 26 (0.4%) 7402 (99.6%) both AP1S3 and IL36RN disease alleles (see Table E9 in this WT, Wild-type. article’s Online Repository at www.jacionline.org). *British patients only. The prevalence of AP1S3 mutations was not significantly Control subjects from publicly accessible cohorts (TWINSUK and ALSPAC). different across disease types (Table II) and did not seem to influence PV concurrence or age of onset (see Table E10 in this We next sought to determine whether IL36RN alleles were article’s Online Repository at www.jacionline.org). However, it associated with key features of pustular psoriasis across was noteworthy that almost all patients with AP1S3 disease disease subtypes. Therefore we implemented a regression alleles (23/24 [95.8%]) were female. Although this observation analysis using clinical diagnosis as a covariate. Although we was not statistically significant (P 5 .06), a trend toward female did not observe a consistent effect of IL36RN mutations on overrepresentation was apparent in all clinical variants PV concurrence (see Table E6 in this article’s Online (see Table E10), suggesting that the penetrance of AP1S3 Repository at www.jacionline.org), we found a significant mutations might be modified by sex-specific factors, such as association with early age of onset (P 5 .003; Fig 1, E), which hormone levels or X-linked modifiers. was observed in all 3 forms of the disease (see Table E6). Thus IL36RN alleles have shared genetic effects across pustular psoriasis subtypes but occur at a very low frequency among DISCUSSION patients with PPP. The purpose of our study was to robustly define clinical and genetic features of pustular psoriasis by investigating a patient cohort of unprecedented size. CARD14 mutations are observed in only a small Initially, we sought to define the presentation of the various minority of cases clinical variants through a rigorous statistical analysis of key Although a sizeable patient subset (n 5 106/473) was phenotypic features. This work, which builds on the definition of sequenced for the entire CARD14 coding region, a targeted consensus diagnostic criteria by ERASPEN,1 underscores the screening of exons 3 and 4 was undertaken in the rest of the importance of collaborative efforts in the analysis of rare diseases. sample, given that the only disease alleles associated with Here a common case report form was used in all prospectively pustular (p.Asp176His) or plaque (p.Gly117Ser) psoriasis map recruited cases, enabling standardized patient phenotyping and to this mutation hotspot.6,20,21 robust data collection. The participation of multiple centers also We found 3 previously described6 GPP patients of Chinese allowed us to monitor the effects of ascertainment bias and descent bearing the p.Asp176His variant. We did not detect any show that key patients’ demographics were comparable across CARD14 substitutions among European patients with GPP the various data sets. but observed 5 British patients with PPP harboring rare Our analysis demonstrated novel and significant differences nonsynonymous changes with deleterious potential (see Table between disease subtypes. Specifically, it showed that PPP is E7 in this article’s Online Repository at www.jacionline.org). associated with patients’ demographics (very high prevalence of Although most of the above subjects (6/8 [75%]) had concurrent female subjects and smokers), clinical (low rates of PV) and PV, the small size of the data set prevented us from establishing genetic features (low prevalence of IL36RN mutations) that are genotype-phenotype correlations. clearly distinct from those observed in ACH and GPP. Given that abnormal IL-36 signaling has now been implicated in the pathogenesis of plaque psoriasis,22 it is tempting to speculate AP1S3 mutations occur with comparable frequency that these observations might be correlated with each other and across disease types that the decreased prevalence of PV in PPP might be linked to Although a substantial patient subset (n 5 249) was screened for the low frequency of deleterious IL36RN alleles in this patient the entire coding region, the rest were sequenced only for exon 2, group. given that the only known AP1S3 mutations (p.Phe4Cys, We also found that IL36RN mutations are associated with an p.Arg33Trp) map to this genomic segment 2.5,9 This revealed 24 earlier age of onset across all variants of pustular psoriasis. This European cases (2 patients with ACH, 4 with GPP, 14 with PPP, validates the results we obtained originally in patients with and 4 with multiple diagnoses) bearing the p.Phe4Cys or GPP4 and indicates that IL36RN should be prioritized for p.Arg33Trp changes (see Table E8 in this article’s Online mutation screening when patients have disease symptoms before 1026 TWELVES ET AL J ALLERGY CLIN IMMUNOL MARCH 2019 the age of 30 years (40 years in the case of ACH/PPP). Given that REFERENCES biologics that counter the effect of IL36RN mutations by blocking 1. Navarini AA, Burden AD, Capon F, Mrowietz U, Puig L, Koks S, et al. European consensus statement on phenotypes of pustular psoriasis. J Eur Acad Dermatol Ve- IL-36 signaling are now under development,23 such targeted nereol 2017;31:1792-9. screening could have important implications for patient 2. Marrakchi S, Guigue P, Renshaw BR, Puel A, Pei XY, Fraitag S, et al. Interleukin- management. 36-receptor antagonist deficiency and generalized pustular psoriasis. N Engl J Med Our study showed that IL36RN mutations are the most frequent 2011;365:620-8. 3. Onoufriadis A, Simpson MA, Pink AE, Di Meglio P, Smith CH, Pullabhatla V, genetic abnormality observed in pustular psoriasis. In fact, et al. Mutations in IL36RN/IL1F5 are associated with the severe episodic inflam- deleterious AP1S3 alleles were found in only 7% to 10% of matory skin disease known as generalized pustular psoriasis. Am J Hum Genet patients, and CARD14 variants were observed in a very small 2011;89:432-7. number of affected subjects. Importantly, our analysis 4. Hussain S, Berki DM, Choon SE, Burden AD, Allen MH, Arostegui JI, et al. IL36RN mutations define a severe auto-inflammatory phenotype of generalized demonstrated that known genes account only for a minority of pustular psoriasis. J Allergy Clin Immunol 2015;135:1067-70. disease cases. This is especially the case in patients with PPP, 5. Setta-Kaffetzi N, Simpson MA, Navarini AA, Patel VM, Lu HC, Allen MH, et al. in whom the combined frequency of AP1S3 and IL36RN AP1S3 mutations are associated with pustular psoriasis and impaired Toll-like re- mutations is less than 10%. Therefore additional studies will be ceptor 3 trafficking. Am J Hum Genet 2014;94:790-7. needed to illuminate the genetic landscape of this condition, 6. Berki DM, Liu L, Choon SE, David Burden A, Griffiths CE, Navarini AA, et al. Activating CARD14 mutations are associated with generalized pustular psoriasis facilitate its diagnosis, and better understand the correlation but rarely account for familial recurrence in psoriasis vulgaris. J Invest Dermatol between genotype and clinical phenotype. Although the 2015;135:2964-70. discovery of novel genetic determinants has thus far been 7. Mossner R, Frambach Y, Wilsmann-Theis D, Lohr S, Jacobi A, Weyergraf A, et al. hindered by the rarity and heterogeneous nature of the disease, Palmoplantar pustular psoriasis is associated with missense variants in CARD14, but not with loss-of-function mutations in IL36RN in European patients. J Invest the ascertainment and rigorous phenotyping of our clinical Dermatol 2015;135:2538-41. resource lays a robust foundation for future gene identification 8. Setta-Kaffetzi N, Navarini AA, Patel VM, Pullabhatla V, Pink AE, Choon SE, et al. studies. Rare pathogenic variants in IL36RN underlie a spectrum of psoriasis-associated pustular phenotypes. J Invest Dermatol 2013;133:1366-9. 9. Mahil SK, Twelves S, Farkas K, Setta-Kaffetzi N, Burden AD, Gach JE, et al. We thank the Psoriasis Association for their continued support with AP1S3 mutations cause skin autoinflammation by disrupting keratinocyte auto- patient recruitment. We also thank the following PLUM and APRICOT phagy and up-regulating IL-36 production. J Invest Dermatol 2016;136:2251-9. investigators for their contribution to patient ascertainment: Mahmud Ali 10. M€ossner R, Wilsmann-Theis D, Oji V, Gkogkolou P, L€ohr S, Schulz P, et al. The (Worthing Hospital, Worthing, United Kingdom), Suzannah August (Poole genetic basis for most patients with pustular skin disease remains elusive. Br J Der- Hospital, Poole, United Kingdom), Herve Bachelez (AP-HP Saint-Louis matol 2018;178:740-8. Hospital, Paris, France), Anthony Bewley (Whipps Cross University 11. Lim KH, Fairbrother WG. Spliceman—a computational web server that predicts Hospital, London, United Kingdom), John Ingram (Cardiff University, sequence variations in pre-mRNA splicing. Bioinformatics 2012;28:1031-2. 12. Yeo G, Burge CB. Maximum entropy modeling of short sequence motifs with ap- Cardiff, United Kingdom), Susan Kelly (The Royal Shrewsbury Hospital, plications to RNA splicing signals. J Comput Biol 2004;11:377-94. Shrewsbury, United Kingdom), Mohsen Korshid (Basildon Hospital, 13. Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J. A general Basildon, United Kingdom), Effie Ladoyanni (Russell’s Hall Hospital, framework for estimating the relative pathogenicity of human genetic variants. Dudley, United Kingdom), and John McKenna (Leicester Royal Infirmary, Nat Genet 2014;46:310-5. Leicester, United Kingdom). We also thank the following external 14. R Core Team. R: a language and environment for statistical computing. Vienna: R collaborators, who contributed to patient recruitment outside of the main Foundation for Statistical Computing; 2017. reference centers: Ivona Aksentijevich (National Institutes of Health, 15. Mrowietz U, van de Kerkhof PC. Management of palmoplantar pustulosis: do we Bethesda, Md), Sibel Dogan (Hacettepe University, Ankara, Turkey), need to change? Br J Dermatol 2011;164:942-6. Carlos Ferrandiz (Germans Trias i Pujol Hospital, Barcelona, Spain), 16. Burden AD, Kirby B. Psoriasis and related disorders. In: Griffiths CEM, Barker JN, Eduardo Fonseca (Complejo Hospitalario Juan Canalejo, Badalona, Spain), Bleiker T, Chalmers RJ, Creamer D, editors. Rook’s textbook of dermatology. Chi- chester: Wiley-Blackwell; 2016. Joanna E. Gach (University Hospitals Coventry and Warwickshire, United 17. Chen Y, Widschwendter M, Teschendorff AE. Systems-epigenomics inference of Kingdom), Maja Mockenhaupt (University of Freiburg, Freiburg, transcription factor activity implicates aryl-hydrocarbon-receptor inactivation as Germany), Jason Pinner (Royal Prince Alfred Hospital, Sidney, Australia), a key event in lung cancer development. Genome Biol 2017;18:236. Christa Prins (Geneva University Hospital, Switzerland), Annamarie Ranki 18. Di Meglio P, Duarte JH, Ahlfors H, Owens ND, Li Y, Villanova F, et al. Activation (Helsinki University Central Hospital, Helsinki, Finland), Raquel Rivera of the aryl hydrocarbon receptor dampens the severity of inflammatory skin condi- (Hospital Universitario 12 de Octubre, Madrid, Spain), Marieke M Seyger tions. Immunity 2014;40:989-1001. (Radboud University Medical Center, Nijmegen, The Netherlands), 19. Ellingford JM, Black GC, Clayton TH, Judge M, Griffiths CE, Warren RB. A novel Pere Soler-Palacin (Vall d’Hebron Research Institute, Barcelona, Spain), mutation in IL36RN underpins childhood pustular dermatosis. J Eur Acad Derma- tol Venereol 2016;30:302-5. Eoin R. Storan (Paediatric Dermatology, Our Lady’s Children’s Hospital 20. Jordan CT, Cao L, Roberson ED, Pierson KC, Yang CF, Joyce CE, et al. PSORS2 is Crumlin, Dublin, Ireland), Virginia Sybert (University of Washington, due to mutations in CARD14. Am J Hum Genet 2012;90:784-95. Seattle, Wash), Ra€ ul Tortosa (Vall d’Hebron Hospital Research Institute, 21. Sugiura K, Muto M, Akiyama M. CARD14 c.526G>C (p.Asp176His) is a signif- Barcelona, Spain), and Helen S. Young (the University of Manchester, icant risk factor for generalized pustular psoriasis with psoriasis vulgaris in the Jap- Manchester, United Kingdom). anese cohort. J Invest Dermatol 2014;134:1755-7. 22. Mahil SK, Catapano M, Di Meglio P, Dand N, Ahlfors H, Carr IM, et al. An anal- ysis of IL-36 signature genes and individuals with IL1RL2 knockout mutations val- Clinical implications: The association between IL36RN idates IL-36 as a psoriasis therapeutic target. Sci Transl Med 2017;9:eaan2514. mutations and early-onset pustular psoriasis defines a patient 23. Ganesan R, Raymond EL, Mennerich D, Woska JR, Caviness G, Grimaldi C, et al. group that should be prioritized for IL36RN screening and Generation and functional characterization of anti-human and anti-mouse IL-36R might benefit from the development of IL-36 inhibitors. antagonist monoclonal antibodies. MAbs 2017;9:1143-54.