You are here

Article Text

Article menu
Miscellaneous
Ultra-rare genetic variation in relapsing polychondritis: a whole-exome sequencing study
Free
  1. Yiming Luo1,2,
  2. Marcela A Ferrada2,
  3. Keith A Sikora3,
  4. Cameron Rankin2,
  5. Hugh D Alessi2,
  6. Daniel L Kastner4,
  7. Zuoming Deng5,
  8. Mengqi Zhang6,
  9. Peter A Merkel7,
  10. Virginia B Kraus8,
  11. Andrew S Allen9,
  12. Peter C Grayson2
  1. 1 Division of Rheumatology, Department of Medicine, Columbia University Irving Medical Center, New York, New York, USA
  2. 2 Vasculitis Translational Research Program, Systemic Autoimmunity Branch, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Maryland, USA
  3. 3 Pediatric Translational Research Branch, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Maryland, USA
  4. 4 Metabolic, Cardiovascular and Inflammatory Disease Genomics Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland, USA
  5. 5 Biodata Mining and Discovery Section, Office of Science and Technology, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health, Bethesda, Maryland, USA
  6. 6 Biostatistics and Research Decision Sciences, Merck & Co., Inc, Rahway, New Jersey, USA
  7. 7 Division of Rheumatology, Department of Medicine, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, USA
  8. 8 Division of Rheumatology, Department of Medicine, Duke University School of Medicine, Durham, North Carolina, USA
  9. 9 Division of Integrative Genomics, Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, North Carolina, USA
  1. Correspondence to Dr Peter C Grayson, National Institutes of Health, Bethesda, MD 20892, USA; peter.grayson{at}nih.gov

Abstract

Objective Relapsing polychondritis (RP) is a systemic inflammatory disease of unknown aetiology. The objective of this study was to examine the contribution of rare genetic variations to RP.

Methods We performed a case–control exome-wide rare variant association analysis that included 66 unrelated European American cases with RP and 2923 healthy controls (HC). Gene-level collapsing analysis was performed using Firth’s logistics regression. Exploratory pathway analysis was performed using three different methods: Gene Set Enrichment Analysis, sequence kernel association test and higher criticism test. Plasma DCBLD2 levels were measured in patients with RP and HC using ELISA.

Results In the collapsing analysis, RP was associated with a significantly higher burden of ultra-rare damaging variants in the DCBLD2 gene (7.6% vs 0.1%, unadjusted OR=79.8, p=2.93×10−7). Plasma DCBLD2 protein levels were significantly higher in RP than in HC (median 4.06 ng/µL vs 0.05 ng/µL, p<0.001). The pathway analysis revealed a statistically significant enrichment of genes in the tumour necrosis factor signalling pathway driven by rare damaging variants in RELB, RELA and REL using higher criticism test weighted by eigenvector centrality.

Conclusions This study identified specific rare variants in the DCBLD2 gene as a putative genetic risk factor for RP. These findings should be validated in additional patients with RP and supported by future functional experiments.

  • Systemic vasculitis
  • Epidemiology
  • Exome Sequencing

Data availability statement

Data are available upon reasonable request.

https://doi.org/10.1136/ard-2023-224732

Statistics from Altmetric.com

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Relapsing polychondritis (RP) is a rare disease whose mechanism is poorly understood.

    • Genetic studies in RP have been limited, including associations with germline variants in HLA and somatic variants in UBA1.

    • Genome- or exome-wide genetic association studies in RP have not been performed.

    WHAT THIS STUDY ADDS

    • Our exome-wide association analyses show an association between RP and rare variants in the DCBLD2 gene.

    • Plasma DCBLD2 protein level is potentially a novel biomarker in patients with RP.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • Understanding the genetic risk factors of RP paves the way for further mechanistic studies and the development of targeted therapies.

    Introduction

    Relapsing polychondritis (RP) is a systemic inflammatory disease of unknown aetiology that affects cartilage and other organs. RP is a rare disease with an estimated incidence of 3.5 cases per million.1 Patients with RP often have substantial diagnostic delays and have significant tissue damage from uncontrolled inflammation. To date, there are no approved therapies for RP. Thus, there is an unmet need to understand causal factors and identify potential therapeutic targets for this condition.

    Genetic association studies hold the potential to play a pivotal role in understanding the mechanisms of complex diseases and aid in the discovery of effective therapeutics.2 Previous studies using targeted human leukocyte antigens (HLA) genotyping have identified specific HLA associations with RP, including HLA-DRB1*16:02, HLA-DQB1*05:02 and HLA-B*67:01. Recently, somatic mutations in UBA1 have been associated with VEXAS (vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome, a monogenic form of RP with a unique phenotype that includes chondritis, haematological abnormalities, poor response to immunosuppressive therapy and high mortality rate.3 However, genome- or exome-wide genetic association studies focusing on germline variants in RP have not been performed.

    In this study, we performed a case–control genetic association study based on whole-exome sequencing (WES) data in RP. The objective of this study was to identify the genetic risk of rare coding variants with predicted deleterious effects in RP at both the gene and pathway levels.

    Materials and methods

    Study cohort

    Patients with RP were enrolled into an ongoing prospective observational cohort study at the National Institutes of Health. Patients who met the following criteria were included in this study: (1) Clinically diagnosed RP with objective evidence of chondritis defined as physician-observed erythema and swelling of the nose or ear consistent with nasal or auricular chondritis; upper airway inflammation confirmed by laryngoscopy; or damaging sequelae of chondritis such as cauliflower ear, saddle nose deformity, subglottic stenosis or tracheomalacia. (2) Self-reported ancestry being non-Hispanic European American. (3) WES data available by January 2021. There were no age restrictions. If two or more biologically related patients with RP were eligible for inclusion, we randomly included only one family member. The details of the clinical assessment in our cohort can be found in online supplemental material.

    Supplemental material

    Control individuals were included from the Atherosclerosis Risk in Communities (ARIC) cohort study.4 The ARIC cohort consists of individuals randomly sampled from four US communities for research in cardiovascular disease prevention. The phenotype and genotype data of the ARIC cohort were obtained from the Database of Genotypes and Phenotypes (dbGaP, accession number: phs000280.v8.p2). Three thousand sex-matched, non-Hispanic European Americans with WES data were selected as controls for this study. Individuals with a high coverage depth of WES data were prioritised.

    WES and bioinformatic processing

    WES was performed on peripheral leucocyte DNA from patients with RP using Agilent 51 Mb Human Exome V5 (Agilent Technologies) capture and PE100-125 Illumina HiSeq2500 sequencing at Otogenetics Corporation (Atlanta, Georgia, USA) with a 100× average read coverage. For control individuals, WES was performed using NimbleGen SeqCap EZ VCRome (Roche, Basel, Switzerland) capture and Illumina HiSeq sequencing at Human Genome Sequencing Center at Baylor College of Medicine (Texas, Houston, USA) with a 92× average read coverage.5

    The raw sequencing data (fastq format) from patients with RP and controls were processed by the same bioinformatic pipeline following the best practice guideline by Genome Analysis Toolkit (GATK) (Broad Institute, Cambridge, Massachusetts, USA). Subsequent processing, including additional quality controls, coverage harmonisation, ancestral homogeneity filtering with principal component analysis (PCA) and identity-by-descent (IBD) analysis, was described in online supplemental material. A flow chart of the detailed bioinformatic and statistical methods is provided in online supplemental figure 1).

    Sanger sequencing

    Sanger sequencing was performed to validate the discovered ultra-rare damaging variants in DCBLD2 in patients with RP. More details can be found in online supplemental material.

    Enzyme-linked immunosorbent assay (ELISA)

    Plasma expression levels of DCBLD2 were quantified by ELISA (R&D Systems, Minneapolis, Minnesota, USA) in 30 patients with RP (5 with DCBLD2 ultra-rare qualifying variants (QVs)) and 32 healthy controls (HC) from the National Institutes of Health Blood Bank in accordance with the manufacturer’s instructions. Patients with RP with and without DCBLD2 ultra-rare QVs were also selected for comparison. Patients were selected based on comparable physician global assessment (PhGA) and immunosuppressive regimens.

    Statistical analysis

    For the collapsing analysis, we constructed five models based on varying QVs criteria (online supplemental table 1). Our primary model, the ultra-rare damaging model, was defined as non-synonymous variants not present in the Genome Aggregation Database (gnomAD),6 and having a Combined Annotation Dependent Depletion (V.1.6) score7 of at least 20 or being protein-truncating variants (PTVs). PTVs were included only if they were annotated as ‘high confidence’ without any flag by the loss-of-function transcript effect estimator.6

    Firth’s logistic regression8 was used to compare the burden of QV in each collapsing model between cases with RP and controls, adjusting for sex and the first three principal components. Study-wide significance level was set as 6.7×10−7 (0.05/ (20 000×4)). QQ plot was used to examine whether there is inflation of results from systematic bias. More details can be found in online supplemental material.

    To further evaluate the calibration of our results, we performed variant-level association analyses in all the common and rare variants that met our quality control criteria. Because of the high computational intensity, we used an approximate Firth’s logistic regression, adjusting for the leave-one-chromosome-out predictions using the REGENIE V.3.0.3.9 This approximation approach has been shown to give very similar results to the exact Firth’s test used in our gene-level analyses.9 Exome-wide significance level was set at 1×10−7 (0.05/500 000). More details can be found in online supplemental material.

    We also performed exploratory analysis to examine whether RP was associated with rare genetic variants at the pathway level. Three pathway-analysis approaches were used including Gene Set Enrichment Analysis (‘GSEA preranked’),10 Sequence kernel association test (‘SKAT_robust’)11 and higher criticism test.12 The input of GSEA and higher criticism test was from genes ranked by p values from the gene-level collapsing analysis. For the higher criticism test, both unweighted and weighted higher criticism tests were performed. The weight was based on a gene intolerance index that was based on the Residual Variation Intolerance Score,13 degree centrality,14 eigenvector centrality15 and RankPage centrality.16 (More details in the higher criticism test can be found in online supplemental material).

    The gene sets studied included all 50 gene sets of the ‘Hallmark Gene Sets’ from the Molecular Signatures Database.17 An additional three gene sets related to cartilage (‘GOBP_REGULATION_OF_CARTILAGE_DEVELOPMENT’, ‘GOBP_CARTILAGE_DEVELOPMENT’ and ‘HP_ABNORMAL_CARTILAGE_MORPHOLOGY’) were included in the analysis. Test-level multiple comparison adjustment was applied for each collapsing model in each method in the pathway-level analysis. For GSEA, family-wise error rate provided by the GSEA software18 was used to adjust for multiplicity. For SKAT, Bonferroni correction for 53 gene sets was used to adjust for multiplicity. For the higher criticism test, the minP algorithm19 was used to obtain multiplicity-adjusted p values across all the gene sets analysed.

    R (V.4.1.2) was used for statistical analyses, including R packages ‘logistf’, ‘QQperm’, ‘SKAT’ and ‘wHC’. This study was approved by the National institutes of Health Intramural Research Program Institutional Review Board (IRB Number: 14AR0200).

    Results

    Study cohorts

    There were 89 cases of RP with WES data available for this study. Among them, there were five pairs with parent-offspring relationship, and one of each pair was randomly excluded. Sixteen additional cases were excluded due to self-identified ancestry not being non-Hispanic European American. After unbiased filtering methods, including PCA (online supplemental figure 2) and IBD analysis, a total of 2989 genetically homogeneous and unrelated individuals were included for association analysis. This group comprised 66 cases with RP and 2923 controls (figure 1). On average, 90.4% of the exonic sites were with well and balanced coverage and included (online supplemental table 2).

    Supplemental material

    Figure 1
    Figure 1

    Flow chart of study population. Sixty-six cases of relapsing polychondritis and 2923 healthy controls were ultimately included in this exome-wide rare variant genetic association study. GATK, Genome Analysis Toolkit; PCA, principal component analysis; RP, relapsing polychondritis; WES, whole-exome sequencing.

    The clinical characteristics of the included cases with RP are summarised in table 1. Among the 66 included cases of RP, 77% were female sex. Mean age at symptom onset was 32 years, and mean age at diagnosis was 38 years. Auricular chondritis (64%), nasal chondritis (76%), costochondritis (42%), ocular inflammation (24%) and cutaneous inflammation (24%) were prevalent. Use of prednisone at doses ≥60 mg daily (52% of patients) and treatment with biological or targeted synthetic immunosuppressants (65% of patients) was common. Over the course of disease, 9% had been hospitalised in an intensive care unit, and 3% died over the course of this study.

    View this table:
    Table 1

    Clinical features of 66 cases with RP included for rare variant association analysis

    Gene-level analysis

    In the ultra-rare damaging model, only the DCBLD2 gene reached both exome-wide and study-wide statistical significance: 5 out of 66 cases with RP and 3 out of 2923 controls carried a DCBLD2 QV in the ultra-rare damaging model (7.6% vs 0.1%, unadjusted OR=79.8, p=2.93×10−7). The QQ plot showed no evidence of bias due to the population genetic structure (Lambda=1.02, figure 2). DCBLD2 remained the top ranked gene for the rare damaging model (7.6% vs 0.5%, p=3.91×10−5) and the rare PTV model (3% vs 0%, p=5.59× 10−5). The sites of ultra-rare damaging QVs within the DCBLD2 gene in patients with RP were located in regions of acceptable coverage depth in gnomAD (online supplemental table 3). The amino acid position changes of the DCBLD2 ultra-rare QVs and the affected domains are shown in online supplemental figure 3. The three-dimensional structure of the DCBLD2 protein with the four discovered ultra-rare damaging variants predicted by AlphaFold2 is shown in online supplemental figure 4.

    Figure 2
    Figure 2

    Manhattan plot and QQ plot of genetic associations for the ultra-rare damaging model. DCBLD2 was associated with RP (7.6% vs 0.1%, unadjusted OR=79.8, p=2.93×10−7) in the ultra-rare damaging model (primary model). QQ plot shows that the distribution of observed p values was consistent with permutation-based null distribution except for DCBLD2. This indicates that there is no evidence of systematic confounding bias due to population genetic structures.

    Rare somatic variants in UBA1 were previously demonstrated to be pathogenic for VEXAS syndrome, a monogenic disease with a large degree of clinical overlap with RP.3 In this study, rare germline variants in UBA1 were not increased in patients with RP versus controls (0/66 vs 4/2923 for the ultra-rare damaging model, 0/66 vs 23/2923 for the rare damaging model, 0/66 vs 0/2923 for the PTV model and 0/66 vs 1/2923 for the recessive model).

    For the remaining collapsing models, no genes reached exome-wide or study-wide statistical significance (online supplemental figure 5, online supplemental table 4). Ultra-rare DCBLD2 QVs were not found in the 16 non-European American patients with RP who were excluded from the association analysis due to not meeting ancestry eligibility. There were no significant differences in the allele frequency of total missense variants (10.6% vs 12.4%, p=0.85) and rare missense variants (variant allele frequency<0.001 regardless of the CADD score, 4.5% vs 2.5%, p=0.24) in DCBLD2 between cases and controls. The information about all DCBLD2 variants identified in this study, irrespective of allele frequency and predicted impact, is shown in online supplemental table 5.

    Supplemental material

    Supplemental material

    Familial clustering and clinical phenotype of patients with RP with DCBLD2 ultra-rare QVs

    At the variant level, there were four specific DCBLD2 ultra-rare QVs identified among patients with RP: two QVs were missense variants (G261V and I514F); one QV was a stop-gain variant (L250X) and one QV was a frameshift variant (Q435fs). We next examined the DCBLD2 ultra-rare damaging QVs in all 89 patients with RP with WES available. Two additional patients with RP, both first-degree relatives of patients with DCBLD2 ultra-rare damaging QVs (G261V and Q435fs), had the same DCBLD2 variant as their familial counterpart. For one patient with RP who carried an ultra-rare damaging DCBLD2 QV (L250X), the WES data for two first-degree healthy relatives were also available, and both had the same DCBLD2 variant without clinical symptoms suggestive of RP. Thus, there were 7 patients with RP, among a total of 89 patients, with DCBLD2 ultra-rare damaging QVs, as summarised in table 2. The Pedigree plot of the three families with carriers of DCBLD2 ultra-rare damaging QVs is shown in figure 3. Sanger sequencing confirmed the validity of all the four discovered DCBLD2 ultra-rare QVs in the seven patients (online supplemental figure 6).

    Figure 3
    Figure 3

    Pedigree plot of patients with RP with DCBLD2 ultra-rare damaging QVs and genotype data from family members available. In the five unrelated patients with RP and DCBLD2 ultra-rare damaging QVs, three of them have genotype data of family members available. In two families, there is another first-degree family member who has RP and carries the DCBLD2 ultra-rare damaging QV. In the third family, there are two first-degree family members who carry the DCBLD2 ultra-rare damaging QV but do not have RP. QV, qualifying variants; RP, relapsing polychondritis.

    View this table:
    Table 2

    DCBLD2 ultra-rare damaging QVs identified in patients with RP

    Among the 89 cases of RP, the clinical characteristics of patients with and without DCBLD2 ultra-rare damaging QVs are summarised in online supplemental table 6. Compared with patients without DCBLD2 ultra-rare damaging QVs, there was a numerically higher proportion of patients carrying DCBLD2 ultra-rare damaging QVs with oral ulcerations (57% vs 21%), genital ulcerations (29% vs 15%) and cardiovascular manifestations, including venous thromboembolism (14% vs 5%), Raynaud’s phenomenon (29% vs 15%), pericarditis (14% vs 2%) and aortitis (14% vs 1%), as well as those who were clinically classified as mouth and genital ulcers with inflamed cartilage (MAGIC) syndrome (29% vs 13%); however, statistical significance was not reached. Regarding additional cardiac manifestations in RP with DCBLD2 ultra-rare damaging QVs, one patient had restrictive cardiomyopathy, and another patient had severe valvular heart disease in both mitral and tricuspid values and required mitral valvuloplasty. The clinical manifestations of patients with RP and DCBLD2 ultra-rare damaging QVs are summarised in online supplemental table 7.

    Among the 12 cases of familial RP (comprising 6 families, with each family having 2 first-degree relatives having RP) cohort, aside from the 4 cases having DCBLD2 ultra-rare damaging variants reported in the manuscript, no other non-synonymous variants in DCBLD2 were identified.

    Variant-level analyses

    Among a total of 529 675 common and rare variants tested, no variant reached exome-wide significance. Variants with p values <0.01 are described in online supplemental table 8. There was no evidence of systematic bias or poor calibration. (Lambda=1.07, more details can be found in online supplemental material and online supplemental figure 7.)

    Supplemental material

    Pathway-level analysis

    As expected, pathways containing DCBLD2 genes, including ‘HALLMARK_APICAL_SURFACE’, ‘HALLMARK_INFLAMMATORY_RESPONSE’ and ‘HALLMARK_KRAS_SIGNALING_UP’ were statistically significant or top ranked with multiple collapsing models across the three statistical methods. However, after removing the DCBLD2 gene, these pathways were no longer significantly associated with a diagnosis of RP.

    ‘HALLMARK_TNFA_SIGNALING_VIA_NFKB’ (not containing DCBLD2) reached test-level statistical significance using higher criticism test weighted by eigenvector centrality in the rare damaging model (p=0.013). The genes favourably weighted by eigenvector centrality (defined as the weighted p values being smaller than the unweighted p values) and with at least one QV in RP in the model were: RELB, RELA, REL, ABCA1, NFE2L2, TRAF1, BIRC2, TNIP1, NR4A1, TNFAIP3 (online supplemental table 9). After removing the top three genes RELB, RELA and REL, the statistical significance was lost (p=1) (online supplemental table 10).

    Supplemental material

    Supplemental material

    ‘HALLMARK_TNFA_SIGNALING_VIA_NFKB’ was also top ranked by GSEA, unweighted higher criticism test and higher criticism test weighted by degree and PageRank centrality in the rare damaging model, but did not reach multiplicity-adjusted statistical significance. The protein interaction network of RELB, RELA and REL is shown in online supplemental figures 8–10).

    The results of the pathway analysis with each method are included in online supplemental tables 10–12. The top ranked three pathways for each collapsing model with the three statistical methods are summarised in online supplemental table 13.

    Supplemental material

    Supplemental material

    Supplemental material

    DCBLD2 plasma protein levels

    DCBLD2 plasma protein levels were measured using ELISA in 30 patients with RP, 5 of them had DCBLD2 ultra-rare QV, and in 32 HC. The median PhGA among the patients with RP at the time of DCBLD2 level measurement was 3 (IQR 1–3). Notably, 29 out of 30 patients (97%) had a PhGA above 0. DCBLD2 levels were significantly higher in patients with RP compared with those in HC (median 4.06 ng/µL vs 0.05 ng/µL, p<0.001, figure 4). In subgroups of RP, plasma DCBLD2 levels were numerically lower in patients with DCBLD2 ultra-rare damaging QVs and numerically higher in patients with PhGA ≥3 (online supplemental figure 11).

    Figure 4
    Figure 4

    Plasma DCBLD2 protein levels in patients with RP and HC. DCBLD2 levels were significantly higher in patients with RP compared with those in HC (median 4.06 ng/µL vs 0.05 ng/µL, p<0.001). HC, healthy controls; RP, relapsing polychondritis.

    Therapeutic response to tumour necrosis factor inhibitor (TNFi)

    Among the total 89 patients with RP, 48 (54%) were treated with TNFi. Six (13%) patients had at least one rare damaging QV in the ‘HALLMARK_TNFA_SIGNALING_VIA_NFKB’ pathway (‘TNF pathway’). One patient with a rare damaging QV in the TNF pathway who was treated with a TNFi was excluded due to loss of follow-up before therapeutic response could be assessed. For the remaining 47 patients who were treated with a TNFi, all 5 (100%) patients with RP with a TNF pathway rare damaging QVs and 23 of 42 (55%) patients with RP without a TNF pathway rare damaging QVs responded to TNFi treatment (p=0.072).

    Discussion

    This is the first exome-wide rare variant association study performed in patients with RP. The results demonstrate an association between RP and certain ultra-rare damaging variants in the DCBLD2 gene. Further examination of these ultra-rare damaging QVs in DCBLD2 revealed familial clustering in two mother–daughter pairs with RP, providing additional evidence supporting a genetic link between DCBLD2 and RP. Taken together, DCBLD2 represents the first putative non-HLA germline genetic variation associated with RP.

    DCBLD2, whose full name is discoidin, CUB and LCCL domain containing 2, is a type I transmembrane protein primarily located in the plasma membrane.20 The extracellular region of DCBLD2 contains a signal sequence, followed by CUB, LCCL and coagulation factor V/VIII type-C (discoidin) domains. All the ultra-rare damaging QVs of DCBLD2 identified in cases with RP were located in this extracellular region. The L250X and G261V variants are in the LCCL domain, Q435fs is in the FV/FVIII domain, and I514F is located in the extracellular region close to the transmembrane region.

    The genetic variants in DCBLD2 identified in this study are likely risk factors contributing to a complex disease process, rather than cause a monogenic form of RP. The ultra-rare damaging QVs of DCBLD2, which contain both missense and PTV variants, suggest that the functional impact of these variants is likely a loss of function (LoF). The ‘probability of being LoF intolerant score’ of the DCBLD2 gene is 0.1, arguing against its capacity to cause haploinsufficiency with heterozygous LoF variation.21 The loss-of-function observed/expected upper bound fraction (LOEUF) is another metric to assess each gene’s tolerance to inactivation along a continuous spectrum.6 The LOEUF for DCBLD2 is 0.446, suggesting a moderate evolutionary constraint against LoF variants. Additionally, the segregation pattern observed in this study among family members of patients with RP, who also have ultra-rare damaging QVs in DCBLD2, argues against Mendelian inheritance; two healthy relatives harboured the same ultra-rare damaging QVs in DCBLD2 as their proband case with RP. However, given its large effect size, the evident deviation against permutation and the observed familial clustering in this study, the likelihood of the association between ultra-rare damaging variants in DCBLD2 and RP occurring merely by chance is expected to be extremely low.

    The role of DCBLD2 ultra-rare damaging variants in the development of RP is unclear. MAGIC syndrome is a subset of RP characterised by overlapping features of Behcet’s disease and higher prevalence of aortitis and Raynaud’s phenomenon.22 Our study observed a numerically higher proportion of clinical features of MAGIC syndrome, including mucosal ulcerations and cardiovascular manifestations, in patients with DCBLD2 ultra-rare damaging QVs, but the sample size in this study was insufficient to definitively conclude whether these ultra-rare variants lead to a distinctive phenotype of RP. Animal and in vitro studies have suggested that DCBLD2 is involved in cardiac function, vascular repair and thrombosis.23–27 Recently, a child with a homozygous DCBLD2 stop-gained variant (W27*) was reported.28 This patient had severe restrictive cardiomyopathy, neurovascular malformation and skeletal abnormalities, and died at the age of 5. However, no autoimmune or inflammatory features, including chondritis, were reported, and the functional implications of DCBLD2 loss-of-function were not investigated. Wang et al performed exome-wide rare variant association analyses on over 280 000 individuals from the UK Biobank.29 There was no significant association between rare variants in DCBLD2 and any included phenotypes, including common inflammatory and cardiovascular diseases. It is worth noting that the sample size of cases with RP in the UK Biobank was fewer than 30, and thus, they were not included in their analyses. Thus, it is unclear whether the DCBLD2 ultra-rare damaging variants cause chondritis or modify RP phenotypes in a disease-specific manner.

    Genetic association studies are a powerful tool for generating mechanistic hypotheses applicable to the overall disease population.2 This study supports a possible common causal mechanism of RP, in which a critical biological pathway involving DCBLD2 is disrupted. This is also supported by the significantly elevated plasma DCBLD2 protein levels in patients with RP compared with those in HC. DCBLD2 has an overall low tissue specificity, including low expression in immune cells and moderate expression in chondrocytes.30 31 Given that DCBLD2 is known to be involved in vascular damage repair,23 it is possible that plasma DCBLD2 levels in RP are elevated secondary to tissue damage, while the ultra-rare damaging variants in DCBLD2 may render RP target tissues, particularly cartilage, more susceptible to inflammatory insult. Interestingly, a recent analysis revealed that upregulation of DCBLD2 is associated with tumour immunosuppression across multiple cancer types.32

    This study performed exploratory pathway analysis and demonstrated that RP is associated with genetic variations in the TNF signalling pathway. This association was largely driven by rare genetic variants in three genes: RELB, RELA and REL. These genes are all members of the second class of Rel/nuclear factor-κB(NF-κB) transcription factors, which are involved in various human immune and inflammatory responses.33 This study reports that 60% of patients with RP responded to TNFi treatment, a finding consistent with the efficacy results from a recent systematic review of RP.34 This study also found a numerically higher proportion of TNFi response in patients with RP who carry rare-damaging QVs in the TNF pathway. Overall, these clinical observations support the hypothesis that dysregulation of the TNF pathway plays a pivotal role in the pathogenesis of RP, though potential role of genetic sequencing in treatment selection merits further investigation. This study also highlights the advantage of incorporating external information for signal detection, such as centrality measures in the protein interaction networks, into pathway-level rare variant association studies to identify potential therapeutic targets in rare diseases.

    There are several limitations of this study. Foremost, the findings were not validated in an independent cohort, which due to the rarity of this disease, is challenging to recruit. With a relatively small sample size, this study was primarily designed to evaluate ultra-rare variants with a very large effect size, thus potentially missing rare variants with smaller effects. The analysis was restricted to cases with RP with non-Hispanic European ancestry due to an insufficient number of cases from other ancestries. Although patients with RP with ultra-rare variants in DCBLD2 had numerically higher prevalence of cardiovascular manifestations, the sample size was too small to provide sufficient statistical power to confirm these associations. Finally, the biological plausibility of the association between rare variants in DCBLD2 and RP etiopathogenesis remains unclear.

    In conclusion, in this exome-wide rare variant association study, we identified rare variants in DCBLD2 as putative genetic risk factors for RP. In the exploratory analysis, rare damaging variants in the TNF pathway may be associated with RP, and the potential role of these variants in therapeutic responsiveness to TNFi warrants further investigation.

    Data availability statement

    Data are available upon reasonable request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    This study involves human participants and was approved by IRB Number: 14AR0200 Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    The work utilized the computational resources of the NIH high-performance computing cluster Biowulf (http://hpc.nih.gov)

    The work was previously presented at the ACR Convergence 2022: Luo Y, Ferrada M, Sikora K, Kastner D, Deng Z, Zhang M, Alessi H, Kraus V, Allen A, Grayson P. Ultra-Rare Genetic Variation in Relapsing Polychondritis: A Whole-Exome Sequencing Study [abstract]. Arthritis Rheumatol. 2022; 74 (suppl 9)

    References

    1. ↵
      2. Kent PD ,
      3. Michet CJ ,
      4. Luthra HS
      . Relapsing polychondritis. Curr Opin Rheumatol 2004;16:56–61. doi:10.1097/00002281-200401000-00011
      OpenUrlCrossRefPubMed
    2. ↵
      2. Tam V ,
      3. Patel N ,
      4. Turcotte M , et al
      . Benefits and limitations of genome-wide association studies. Nat Rev Genet 2019;20:467–84. doi:10.1038/s41576-019-0127-1
      OpenUrlCrossRefPubMed
    3. ↵
      2. Beck DB ,
      3. Ferrada MA ,
      4. Sikora KA , et al
      . Somatic mutations in UBA1 and severe adult-onset autoinflammatory disease. N Engl J Med 2020;383:2628–38. doi:10.1056/NEJMoa2026834
      OpenUrlCrossRefPubMed
    4. ↵
      The Atherosclerosis risk in communities (ARIC) study: design and objectives. The ARIC investigators. Am J Epidemiol 1989;129:687–702. doi:10.1093/oxfordjournals.aje.a115184
      OpenUrlCrossRefPubMed
    5. ↵
      2. Lubitz SA ,
      3. Brody JA ,
      4. Bihlmeyer NA , et al
      . Whole exome sequencing in atrial fibrillation. PLoS Genet 2016;12:e1006284. doi:10.1371/journal.pgen.1006284
    6. ↵
      2. Karczewski KJ ,
      3. Francioli LC ,
      4. Tiao G , et al
      . The mutational constraint spectrum quantified from variation in 141,456 humans. Nature 2020;581:434–43. doi:10.1038/s41586-020-2308-7
      OpenUrlCrossRefPubMed
    7. ↵
      2. Rentzsch P ,
      3. Witten D ,
      4. Cooper GM , et al
      . CADD: predicting the deleteriousness of variants throughout the human genome. Nucleic Acids Res 2019;47:D886–94. doi:10.1093/nar/gky1016
      OpenUrlCrossRefPubMed
    8. ↵
      2. Wang X
      . Firth logistic regression for rare variant association tests. Front Genet 2014;5:187. doi:10.3389/fgene.2014.00187
    9. ↵
      2. Mbatchou J ,
      3. Barnard L ,
      4. Backman J , et al
      . Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet 2021;53:1097–103. doi:10.1038/s41588-021-00870-7
      OpenUrlCrossRefPubMed
    10. ↵
      2. Subramanian A ,
      3. Tamayo P ,
      4. Mootha VK , et al
      . Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545–50. doi:10.1073/pnas.0506580102
      OpenUrlAbstract/FREE Full Text
    11. ↵
      2. Zhao Z ,
      3. Bi W ,
      4. Zhou W , et al
      . UK Biobank whole-exome sequence binary phenome analysis with robust region-based rare-variant test. Am J Hum Genet 2020;106:3–12. doi:10.1016/j.ajhg.2019.11.012
      OpenUrlCrossRef
    12. ↵
      2. Zhang M ,
      3. Gelfman S ,
      4. McCarthy J , et al
      . Incorporating external information to improve sparse signal detection in rare-variant gene-set-based analyses. Genet Epidemiol 2020;44:330–8. doi:10.1002/gepi.22283
      OpenUrl
    13. ↵
      2. Petrovski S ,
      3. Wang Q ,
      4. Heinzen EL , et al
      . Genic intolerance to functional variation and the interpretation of personal genomes. PLoS Genet 2013;9:e1003709. doi:10.1371/journal.pgen.1003709
    14. ↵
      2. Borgatti SP
      . Centrality and network flow. Soc Netw 2005;27:55–71. doi:10.1016/j.socnet.2004.11.008
      OpenUrlCrossRef
    15. ↵
      2. Stephenson K ,
      3. Zelen M
      . Rethinking centrality: methods and examples. Soc Netw 1989;11:1–37. doi:10.1016/0378-8733(89)90016-6
      OpenUrlCrossRef
    16. ↵
      2. White S ,
      3. Smyth P
      . Algorithms for estimating relative importance in networks. Proceedings of the ninth ACM SIGKDD international conference on Knowledge discovery and data mining; 2003:266–75 doi:10.1145/956750.956782
    17. ↵
      2. Liberzon A ,
      3. Birger C ,
      4. Thorvaldsdóttir H , et al
      . The molecular signatures database (MSigDB) hallmark gene set collection. Cell Syst 2015;1:417–25. doi:10.1016/j.cels.2015.12.004
      OpenUrl
    18. ↵
      GSEA software download, Available: http://www.gsea-msigdb.org/gsea/downloads.jsp
    19. ↵
      2. Ge Y ,
      3. Dudoit S ,
      4. Speed TP
      . Resampling-based multiple testing for Microarray data analysis. Test 2003;12:1–77. doi:10.1007/BF02595811
      OpenUrl
    20. ↵
      2. Schmoker AM ,
      3. Ebert AM ,
      4. Ballif BA
      . The DCBLD receptor family: emerging signaling roles in development, homeostasis and disease. Biochem J 2019;476:931–50. doi:10.1042/BCJ20190022
      OpenUrlAbstract/FREE Full Text
    21. ↵
      2. Lek M ,
      3. Karczewski KJ ,
      4. Minikel EV , et al
      . Analysis of protein-coding genetic variation in 60,706 humans. Nature 2016;536:285–91. doi:10.1038/nature19057
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      2. Luo Y ,
      3. Bolek EC ,
      4. Quinn KA , et al
      . A prospective observational cohort study and systematic review of 40 patients with mouth and genital ulcers with inflamed cartilage (MAGIC) syndrome. Semin Arthritis Rheum 2022;52:151924. doi:10.1016/j.semarthrit.2021.10.007
      OpenUrl
    23. ↵
      2. Nie L ,
      3. Guo X ,
      4. Esmailzadeh L , et al
      . Transmembrane protein ESDN promotes endothelial VEGF signaling and regulates angiogenesis. J Clin Invest 2013;123:5082–97. doi:10.1172/JCI67752
      OpenUrlCrossRefPubMed
    24. ↵
      2. Wang S ,
      3. Liu X ,
      4. Meng Z , et al
      . DCBLD2 regulates vascular hyperplasia by modulating the platelet derived growth factor receptor-beta endocytosis through caveolin-1 in vascular smooth muscle cells. FASEB J 2022;36:e22488. doi:10.1096/fj.202200156RR
    25. ↵
      2. Smith CL ,
      3. Blake JA ,
      4. Kadin JA , et al
      . Mouse genome database (MGD)-2018: knowledgebase for the laboratory mouse. Nucleic Acids Res 2018;46:D836–42. doi:10.1093/nar/gkx1006
      OpenUrlCrossRefPubMed
    26. ↵
      2. Jung J-J ,
      3. Ahmad AA ,
      4. Rajendran S , et al
      . Differential BMP signaling mediates the interplay between genetics and leaflet numbers in aortic valve calcification. JACC Basic Transl Sci 2022;7:333–45. doi:10.1016/j.jacbts.2021.12.006
      OpenUrl
    27. ↵
      2. Vermeersch E ,
      3. Nuyttens BP ,
      4. Tersteeg C , et al
      . Functional genomics for the identification of modulators of platelet-dependent thrombus formation. TH Open 2018;2:e272–9. doi:10.1055/s-0038-1670630
      OpenUrl
    28. ↵
      2. Alhamoudi KM ,
      3. Barhoumi T ,
      4. Al-Eidi H , et al
      . A homozygous nonsense mutation in DCBLD2 is a candidate cause of developmental delay, dysmorphic features and restrictive cardiomyopathy. Sci Rep 2021;11:12861. doi:10.1038/s41598-021-92026-0
    29. ↵
      2. Wang Q ,
      3. Dhindsa RS ,
      4. Carss K , et al
      . Rare variant contribution to human disease in 281,104 UK biobank exomes. Nature 2021;597:527–32. doi:10.1038/s41586-021-03855-y
      OpenUrl
    30. ↵
      2. Sjöstedt E ,
      3. Zhong W ,
      4. Fagerberg L , et al
      . An Atlas of the protein-coding genes in the human, pig, and mouse brain. Science 2020;367:eaay5947. doi:10.1126/science.aay5947
    31. ↵
      DCBLD2 expression. n.d. Available: https://www.proteinatlas.org/ENSG00000057019-DCBLD2
    32. ↵
      2. Xie P ,
      3. Liu J-Y ,
      4. Yan H , et al
      . Pan-cancer analyses identify DCBLD2 as an oncogenic, immunological, and prognostic biomarker. Front Pharmacol 2022;13:950831. doi:10.3389/fphar.2022.950831
    33. ↵
      2. Gilmore TD
      . The REL/NF-kappaB signal transduction pathway: introduction. Oncogene 1999;18:6842–4. doi:10.1038/sj.onc.1203237
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      2. Biya J ,
      3. Dury S ,
      4. Perotin J-M , et al
      . Assessment of TNF-Α inhibitors in airway involvement of relapsing polychondritis: a systematic review. Medicine (Baltimore) 2019;98:e17768. doi:10.1097/MD.0000000000017768

    Supplementary materials

    Footnotes

    • Handling editor Josef S Smolen

    • Contributors All authors meet the criteria for authorship and more specifically for contributorship statement: YL and PCG were involved in the conception and design of the study; acquisition of data and/or analysis and interpretation of data; drafting of manuscript and revising it critically for important intellectual content; final approval of the version to be published. MF, KAS, VBK and ASA were involved in the conception and design of the study; interpretation of data; revising the manuscript critically for important intellectual content; final approval of the version to be published. CR, HA, ZD and MZ were involved in the acquisition of data and/or analysis and interpretation of data; revising the manuscript critically for important intellectual content; final approval of the version to be published. DLK and PAM were involved in the interpretation of data; revising the manuscript critically for important intellectual content; final approval of the version to be published. YL and PCG is responsible for the overall content as guarantor.

    • Funding YL’s work was supported by the Rheumatology Research Foundation Scientist Development Award. This study was also supported by the Intramural Research Program at the National Institute of Arthritis and Musculoskeletal and Skin Diseases.

    • Competing interests MZ is an employee of Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, New Jersey, USA.

    • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.