Abstract

   Hallux valgus is a common form of foot deformity, and genetic factors
   contribute substantially to the pathogenesis of hallux valgus
   deformity. We conducted a genetic study on the structural variants
   underlying familial hallux valgus using whole exome sequencing
   approach. Twenty individuals from five hallux valgus families and two
   sporadic cases were included in this study. A total of 372 copy number
   variations were found and passed quality control filtering. Among them,
   43 were only present in cases but not in controls or healthy
   individuals in the database of genomic variants. The genes covered by
   these copy number variations were enriched in gene sets related to
   immune signaling pathway, and cytochrome P450 metabolism. The
   hereditary CNVs demonstrate a dominant inheritance pattern. Two
   candidate pathogenic CNVs were further validated by quantitative-PCR.
   This study suggests that hallux valgus is a degenerative joint disease
   involving the dysregulation of immune and metabolism signaling
   pathways.

   Keywords: hallux valgus, whole exome sequencing, copy number variation,
   immune dysregulation, cytochrome p450 metabolism

Introduction

   Hallux valgus (HV) deformity refers to the lateral deflection of the
   great toe at the first metatarsophalangeal joint ([52]Perera et al.,
   2011; [53]Hecht and Lin, 2014), which is the most common forefoot
   deformity often requires surgery with a prevalence rate of 23% in
   people aged 18–65 (CI: 16.3%–29.6%) ([54]Nix et al., 2010). African
   Americans are more likely to have HV than people of European ancestry
   [adjusted odds ratio (aOR) = 2.01, 95% confidence interval [CI] =
   1.39–2.92] ([55]Golightly et al., 2012), and no significant difference
   has been found in the incidence of HV in other populations.
   Conservative treatment is feasible for patients with deformity but
   without symptoms or mild symptoms ([56]Bayar et al., 2011). Although
   the symptoms can be alleviated, they can not be completely reversed. If
   the patient’s pain persists, surgery is necessary. However, many
   surgical complications and poor prognosis have brought great burden to
   individuals and their families ([57]Sammarco and Idusuyi, 2001).
   Prevention and early intervention are therefore important.

   The pathogenesis of HV deformity is complex ([58]Mann and Coughlin,
   1981). HV may be associated with inappropriate footwear. HV is 15 times
   more common in people who wear shoes than in those who don’t
   ([59]Perera et al., 2011), and shoes that tighten the front foot appear
   to be one of the leading causes of HV. Heredity is another important
   risk factor for HV development, especially in adolescents. A positive
   family history of HV has been reported in many studies ([60]Lieberson
   and Mendes, 1991; [61]Pique-Vidal et al., 2007), and study has found
   that HV susceptibility is related to genetic polymorphisms associated
   with arthritis ([62]Perera et al., 2011; [63]Hecht and Lin, 2014).
   Linear bone arrangement or static stabilizer relaxation due to heredity
   may also lead to HV deformity ([64]Perera et al., 2011).

   Genetic factors make substantial contribution to the pathogenesis of HV
   deformity. In a study of 350 patients with three generations family
   trees, 90% had at least one affected relative, which is consistent with
   autosomal dominant inheritance pattern ([65]Pique-Vidal et al., 2007).
   The heritability of HV in European ancestry populations is between 0.29
   and 0.89 ([66]Hannan et al., 2013), while the rate of HV in Korea is
   ∼0.51 ([67]Lee et al., 2014). In a genome-wide association study (GWAS)
   on European ancestry population including 1786 cases of HV deformity
   and 2623 controls, genome-wide SNPs accounted for 50% of the phenotypic
   variance in males and 48% of the phenotypic variance in females
   ([68]Hsu et al., 2015). The missing heritability may lie in the
   contribution of rare variants and structural variants, which have been
   underexplored in previous studies.

   Structural variation is generally defined as a region of DNA of
   approximately 1 kb or larger that includes changes in copy number,
   chromosomal position, or orientation between individuals ([69]Escaramis
   et al., 2015). A major class of genomic structural variation is copy
   number variation (CNV), which includes deletion and duplication of
   sequences ([70]Ionita-Laza et al., 2009). In addition to single
   nucleotide polymorphisms (SNP) ([71]Sachidanandam et al., 2001), CNVs
   are a major source of variation in the human genome, with significant
   effects on evolution and disease susceptibility ([72]Conrad and Hurles,
   2007). For example, at least 15% of neurodevelopmental diseases are
   caused by local dose imbalances in dozens of genes due to CNVs
   ([73]Girirajan et al., 2011). Furthermore, CNVs make significant
   contributions to development of bone disorders and CNV analysis
   increases the diagnostic yield for these diseases. CNVs are
   significantly associated with osteoporosis ([74]Yang et al., 2008;
   [75]Costantini et al., 2018) and CNV is an important genetic factor for
   the etiology of fetal skeletal dysplasia ([76]Wit et al., 2014; [77]Bai
   et al., 2022). In addition, studies have shown that many CNVs confer
   greater disease risk than SNPs ([78]Ionita-Laza et al., 2009). However,
   the contribution of CNVs to the pathogenesis and development of HV has
   not been investigated.

   Aiming to explore the contribution of potential pathogenic CNVs to the
   development of HV deformity, we conducted the first study on the
   structural variants underlying familial HV using whole exome sequencing
   (WES) approach, which may be helpful for future risk prediction of HV.

Materials and methods

Samples and ethics statement

   In this study, we recruited a total of 22 Chinese participants from 5
   families (including 17 cases and 3 controls without any foot deformity)
   and two sporadic cases ([79]Figure 1). The recruitment and the WES
   study were approved by Tianjin Hospital, and all participants provided
   written informed consent. The diagnosis of HV was made by clinical
   experts of the Foot and Ankle Surgery Group of Orthopaedic Branch of
   Chinese Medical Association and Foot and Ankle Surgery Professional
   Committee of Orthopaedic Physician Branch of Chinese Medical
   Association, according to the expert consensus ([80]Foot and Ankle
   Working Committee et al., 2015). HV is diagnosed by combining the
   evaluation of clinical presentations, physical examination, auxiliary
   imaging examination and medical history. The severity of HV was
   determined by hallux valgus angle (HVA) and intermetatarsal angle (IMA)
   (normal: HVA < 16°, IMA < 10°; mild: HVA < 20°, IMA < 13°; moderate:
   20° < HVA ≤ 40°, 13° < IMA ≤ 16°; severe: HVA > 40°,IMA > 16°). The
   single nucleotide variant analysis of the WES data of three families
   has been reported in our previous publication ([81]Jia et al., 2021).
   With the addition of data from HV families and sporadic cases, we
   performed the current CNV study. The genetic genealogy of the 5
   families shows that each proband has at least one first-degree relative
   as a HV patient.

FIGURE 1.

   [82]FIGURE 1
   [83]Open in a new tab

   The family pedigree diagrams for all the subjects in the study. S-1 and
   S-2 are two sporadic cases. Families F1, F3 and F5 were included in our
   previous study of single nucleotide variants ([84]Jia et al., 2021).The
   color indicates the severity of the patient’s illness from mild to
   severe.

Genomic DNA extraction and whole-exome sequencing

   Genomic DNA was extracted from peripheral blood sample of each subject
   following standard procedures. The TargetSeqTM Enrichment Kit
   (iGeneTechTM) Human Exome Capture Kit was used for library
   construction. The Illumina sequencing platform was used for paired
   next-generation sequencing.

Quality control of sequencing result files

   The Trim Galore software ([85]Martin, 2011)
   ([86]https://github.com/FelixKrueger/TrimGalore) was used to remove
   low-quality base (−q 25), limit maximum allowable error rate (default
   −e 0.1), remove reads < 36 nt (-length 36), remove double-ended
   overlap>3, and remove reads as a unit (-paired) from the raw fastQ
   files. Burrows-Wheeler Aligner (BWA)-MEM (version 0.7.17) ([87]Li,
   2013) was used to align the reads with the reference genome to obtain
   the SAM file.

   After reordering SAM files were converted to BAM (Raw BAM) files and
   polymerase chain reaction (PCR) duplicates were marked using Picard
   (v1.91). SamTools (version 1.58) ([88]Li et al., 2009) was used for
   quality control of the binary alignment graph files generated between
   them. Then, the Genome Analysis ToolKit (GATK) (version 3.8)
   ([89]McKenna et al., 2010) was used to re-align indel regions and
   correct the base mass fraction. And after Base quality score
   recalibrations (BQSRs), we used verifyBamID (version 1.1.3) (Jun et
   al., 2012) to confirm that there was no contamination of the cross
   samples. The resulting BAM files were used for CNV calling.

CNVs calling by GATK and XHMM

   The eXome-Hidden Markov Model (XHMM) ([90]Fromer and Purcell, 2014) was
   used to call CNVs from filtered normalized targeted exome sequence
   data. The XHMM divided the chromosomal regions into three types:
   diploid, deletion, and duplication. The entire pipeline is plotted as a
   flowchart and shown in [91]Figure 2.

FIGURE 2.

   [92]FIGURE 2
   [93]Open in a new tab

   The pipeline of copy number variation analysis based on whole exome
   sequencing data.

CNVs annotation

   We used “scan_regoin.pl” program in GenGen([94]Wang et al., 2007) to
   scan genomic features and to find the CNVs that overlapped with those
   in the database of genomic variants (DGV) ([95]MacDonald et al., 2014)
   and filtered them out. The DGV database contains genomic structural
   variants over 50 bp in healthy individuals.

   We used “scan_regoin.pl” program in GenGen ([96]Wang et al., 2007) with
   the hg19_refGene and hg19_refLink files to annotate the CNV regions
   against the RefGene annotation to find the genes overlapped the CNV
   regions.

   Then, we filtered out CNVs present among controls, and retained
   case-only CNVs that were repeated among the HV cases but not carried by
   any of the controls.

Quality control filtering of CNVs

   We performed quality control filtering based on “Q_EXACT” and “Q_SOME”
   scores in the XCNV file according to the instruction of XHMM software.
   “Q_EXACT” and “Q_SOME” represent the phred-scaled quality of a CNV
   event along the entire interval and the same CNV event in the interval
   respectively.

Pathway enrichment of CNVs

   The 67 genes overlapped with the 43 CNVs were used as input to the
   web-portal of OBAS ([97]http://kobas.cbi.pku.edu.cn/) ([98]Bu et al.,
   2021) and STRING (version 11.0) ([99]https://string-db.org/)
   ([100]Szklarczyk et al., 2021) for pathway enrichment and
   protein-protein interaction (PPI) analysis. Kyoto Encyclopedia of Genes
   and Genomes (KEGG) pathway database and Search Tool for the Retrieval
   of Interacting Genes/Proteins database were used as the reference
   databases.

Examination of CNVs using integrative genomics viewer

   For candidate CNVs of interest, we further conducted visual inspection
   using the Integrative Genomics Viewer, based on the aligned bam files
   of the WES data of each CNV carrier and non-carriers within the same
   family.

qPCR validation of CNVs

   For the potential pathogenic CNVs, we chose a ∼100 bp fragment in each
   CNV for qPCR validation. The qPCR experiment was carried out with the
   Sybr Green I system. Each reaction in a 10 µL system contains 4 ng of
   genomic DNA, 10 µm per-primer, 2× ChamQ universal SYBR qPCR Master Mix,
   and DD-H[2]O. Each sample was repeated for three times. The geometric
   mean values of the CT values of the control sequence GAPDH and the
   sample sequence were calculated to obtain △CT values for each sample.
   We calculated the 2^−△△CT of each sample by taking the 2^−△ CT of
   members without CNV in the family as the relative reference value.
   Finally, the existence of CNV was examined by the relative value of
   2^−△△CT of each sample in the pedigree.

Results

The identification of 43 CNVs related to HV

   To assess the potential contribution of structural variants to HV, we
   carried out a CNV study based on the WES data of five HV families
   including 17 cases and 3 family members without HV and two sporadic
   cases ([101]Supplementary Table S1). A total of 965 CNVs were detected
   based on the WES data and 372 passed quality control filtering
   ([102]Supplementary Table S2).

   Among these 372 CNVs, 325 CNVs were present only in cases, including
   142 deletions and 183 duplications. They were not detected in the any
   of the three control samples. Furthermore, 51 CNVs (26 deletions and 25
   duplications) were carried by more than one samples. In order to screen
   for potential pathogenic CNVs, we further filtered out any CNVs with
   70% overlap with those carried by healthy individuals in the DGV. Then,
   43 CNVs were retained for further analysis, covering the exons of the
   67 genes, with 1 CNV co-occurring in four cases, 5 CNVs co-occurring in
   three cases, and 37 CNVs co-occurring in two cases ([103]Supplementary
   Table S3).

The involvement of immune and metabolism genes

   In order to further explore how these CNVs may contribute to the
   pathogenesis of HV, we carried out pathway enrichment analysis among
   the 67 genes covered by case-only CNVs through KEGG PATHWAY database.
   Ten pathways were statistically significantly enriched (p < 0.001) for
   cytokine genes and genes involved in cytochrome P450 related metabolism
   and immunity/inflammation ([104]Figure 3A; [105]Supplementary Table
   S4), such as KEGG pathways “Drug metabolism—cytochrome P450”,
   “Toll-like receptor signaling pathway” and “Cytokine-cytokine receptor
   interaction”, suggesting their contribution to the pathogenesis of HV.

FIGURE 3.

   [106]FIGURE 3
   [107]Open in a new tab

   Pathway and protein-protein interaction analyses on the genes covered
   by 43 case-only CNVs (A) KEGG pathway enrichment analysis on the genes
   covered by the 43 case-only CNVs. The color of the dots represents the
   size of the corrected p-value, and the size of the dots represents the
   number of input genes contained in the corresponding pathway. (B)
   Protein-protein interaction analysis on the genes covered by 43
   case-only CNVs. The nodes represent the proteins, and the lines between
   the nodes indicate the interactions between the proteins.

   We then conducted the PPI analysis on these genes ([108]Figure 3B). The
   PPI network also supports the consistent involvement of a group of
   immune genes and genes functioning in drug metabolism. In the PPI
   network, we observed the cluster formed by proteins involved in drug
   metabolism (UGT2B28, CYP2D6, CYP2A6, GSTM1, SULT1A1). In addition to
   the well-known involvement of cytokine genes CCL4L1 and CCL3L3 in
   immunity, several other genes also function in regulation of the immune
   system. HLA-H (HFE) is an important link between iron homeostasis and
   immune regulation ([109]Bahram et al., 1999; [110]Lio et al., 2002);
   DYRK1A gene plays a key role in regulating the differentiation of Th17
   and regulatory T cells ([111]Khor et al., 2015); and evidence suggests
   that genetic polymorphisms of CYP2D6 is associated with autoimmune
   bullous diseases induction ([112]Rychlik-Sych et al., 2013).

Inheritance pattern of HV-related CNVs

   To identify the potential disease-contributing CNVs, we examined the
   inheritance pattern of the CNVs which passed quality control filtering.
   We did not find any CNV that fits the recessive pattern. Multiple
   case-only CNVs consistent with dominant inheritance pattern were found
   in each family. For family 1, 3 and 5, dominant CNVs were referred to
   those present in cases (child and parent with HV) but absent in
   controls (parent without HV). For family 2 and family 4 in which all
   family members were affected individuals with HV, the CNVs consistent
   with dominant inheritance were those carried by the child and one of
   the parents ([113]Supplementary Table S5). Our results are consistent
   with the hypothesis that HV has a dominant inheritance
   ([114]Pique-Vidal et al., 2007), but certainly we can not exclude the
   possibility of recessive inheritance model as a large proportion of the
   genome were not assayed in our analysis by WES.

Validation of disease-contributing CNVs

   Two inherited CNVs were of particular interest and they have the
   highest quality (Phred-scaled quality of Non-Diploidy = 99) in our
   analyses ([115]Table 1). Chr22: 42522498–42536739 deletion occurs to
   all patients in family 1 and absent in the healthy member of the
   family. This segment of CNV covered genes CYP2D7 and CYP2D6 which are
   involved in bone excitation effects ([116]Jayaraman et al., 2021) and
   metabolic pathways related to human immunity ([117]Effner et al.,
   2017). Chr6: 29855550–29895036 deletion appears to F2-I-1 and F2-II-1
   the two individuals with moderate HV but not in family members F2-I-2
   and F2-II-2 who have only mild HV condition. HCG4B and HLA-H genes are
   in this segment of CNV. HCG4B is a proinflammatory gene whose
   expression is positively correlated with the expression of HLA-A and
   may regulate the expression of HLA-A ([118]Chen et al., 2017). HLA-A
   and HLA-H function in immune homeostasis ([119]Jordier et al., 2019).
   Therefore evidence from literature suggests that these two CNVs may
   contribute to the pathogenesis of HV.

TABLE 1.

   Candidate pathogenic CNVs.
   SAMPLE CNV INTERVAL Q_NON_DIPLOID Q_SOME GENE
   F1-I-1,F1-II-1,F1-II-2 DEL chr22: 42522498–42536739 99 99 CYP2D6,
   CYP2D7
   F2-I-1,F2-II-1 DEL chr6: 29855550–29895036 99 93 HCG4B, HLA-H
   [120]Open in a new tab

   SAMPLE, sample ID; CNV, type of copy number variation (DEL or DUP);
   INTERVAL, genomic range of the called CNV; Q_NON_DIPLOID, Phred-scaled
   quality of not being diploid, i.e., DEL or DUP event in the interval;
   Q_SOME, Phred-scaled quality of some CNV event in the interval; GENE,
   gene name.

   We performed qPCR ([121]Supplementary Table S6) as an independent
   experimental approach to validate these two CNVs. The presence of these
   two CNVs in family 1 and family 2 respectively was consistent with the
   results from the CNV calling of WES data ([122]Figure 4).

FIGURE 4.

   [123]FIGURE 4
   [124]Open in a new tab

   qPCR validation of two CNVs. (A) Copy number detected by qPCR at chr22:
   42522498–42536739 for F1-I-1, F1-II-1, F1-II-2 and F1-I-2. (B) Copy
   number detected by qPCR at chr6: 29855550–29895036 for F2-I-1, F2-II-1
   and F2-I-2. (C) Copy number detected by qPCR at chr6: 29855550–29895036
   for F2-I-1, F2-II-1 and F2-II-2.

Discussion

   This is the first study that identified the structural variants
   underlying familial HV using WES approach, The identification provided
   us new knowledge about the genetics basis of HV, which highlighted the
   potential contribution of immunity/inflammation, and cytochrome P450
   metabolism to the etiology of HV.

   Previous studies indicated that HV malformations are likely to be
   caused by a variety of contributing factors, including genetics,
   dorsiflexion of the first metatarsal, gastrocnemius equinus, abnormal
   foot mechanics, and joint hypermobility ([125]Coughlin and Jones,
   2007). Interestingly, patients with autoimmune arthritis conditions
   such as rheumatoid arthritis and psoriatic arthritis are more likely to
   develop HV malformations ([126]Rojas-Villarraga et al., 2009;
   [127]Hyslop et al., 2010).

   The CNV chr6: 29855550–29895036 observed in both F2-I-1 and F2-II-1
   covers the HCG4B and HLA-H genes. It has been found that HCG4B is a
   pro-inflammatory factor and may regulate the expression of HLA-A by
   acting on competing endogenous RNAs spongingmiR-122 and miR-1352
   ([128]Chen et al., 2017). HLA-A and HLA-H play essential roles in the
   immune system, including regulation of the innate immune system, class
   I MHC-mediated antigen processing and presentation, interferon gamma
   signaling, and antigen processing-cross-presentation ([129]Allan et
   al., 2002; [130]Nitschke et al., 2016; [131]Jordier et al., 2019). In
   psoriatic arthritis and rheumatoid arthritis, HLA alleles have been
   shown to influence susceptibility and severity of these autoimmune
   joint disease ([132]Mc et al., 2015).

   The CNV chr 22: 42529569–42529669 seen in F1-I-1, F1-II-1 and F1-II-2
   covers CYP2D6 and CYP2D7.These genes have roles in the cytochrome P450
   related xenobiotic metabolic process and oxidation-reduction process
   ([133]Lessard et al., 1997). Cytochrome P450 changes the structure of
   cell antigens by metabolizing foreign organisms to produce reactive
   oxygen species, and then initiates and/or amplifies the autoimmune
   phenomenon through molecular simulation of the autoimmune response to
   the original antigen ([134]Namazi, 2009). In addition, studies have
   shown that secondary metabolites produced by CYP2D6-dependent
   biotransformation have bone excitation effects ([135]Jayaraman et al.,
   2021).

   These two CNVs may have pathological effects and lead to bone structure
   deformities through the dysregulation of the immune system and the
   process of reactive oxygen species production.

   Previous genetic studies on HV, including our WES study ([136]Jia et
   al., 2021), revealed the association of a few genetic variants with HV.
   Genome-wide association study in population of European ancestry found
   the sex-specific association of SNVs close to genes AXIN2, ESD, ANXA1
   and MRGPRX3 at the marginal significance level ([137]Hsu et al., 2015).
   A later GWAS with increased sample size identified the genome-wide
   significant association with SNP in gene CLCA2, being an expression
   quantitative trait locus for COL24A1 ([138]Arbeeva et al., 2020).
   Candidate gene study in Chinese population suggested variants in VDR
   gene and SNP rs1800629 at the 5’ of TNF gene are related to HV
   ([139]Tao et al., 2018; [140]Arbeeva et al., 2020). In our study, we
   did not find outstanding CNVs in these genes, which could be due to
   differences in study population, variant types. We also focused on
   different regions of the human genome from the previous GWAS (the
   coding exons versus the non-coding regions). However, similar
   underlying genetic mechanisms were revealed. As discussed by Hsu and
   colleagues, AXIN2 and ANXA1 have important functions in regulation of
   innate and adaptive immunity, suggesting the potential role of
   inflammation in HV ([141]Hsu et al., 2015). Hsu and colleagues also
   discussed the involvement of the serine hydrolase gene ESD gene in the
   recycling of sialic acids which is related to decreased antioxidant
   levels. Interestingly, CYP2D6 and CYP2D7 also have roles in
   oxidation-reduction process. The SNP rs1800629 associated with HV in
   Chinese population is an eQTL SNP for genes including complement gene
   C4A and CYP21A1P a member of the Cytochrome P450 Family. Therefore,
   these identifications and results from our study converge on the
   similar immune and drug metabolism signaling.

   Because of the importance of CNV in the etiology of complex human
   diseases, various CNV detection methods and computational algorithms
   have been developed. The traditional array-based approach including
   array-based comparative hybridization and SNP-array approaches can
   efficiently detect large CNVs with a relatively accurate rate
   ([142]Gabrielaite et al., 2021). CNV calling from high-throughput
   sequencing data has become more and more commonly utilized in research
   and clinical diagnosis of complex human disorders. Though still being
   limited by short read length, high-throughput sequencing offers the
   potential to identify novel CNVs in the genomic regions with sparse
   coverage or lack of coverage by array probes. As the most frequently
   used sequencing approach, WES focuses on the exomes, the copy number
   alterations of which are more likely to have a pathogenic effect than
   the other regions of the human genome. Calling CNVs from WES data also
   has the advantages of detecting small CNVs ranging from 1–100 kb
   ([143]Gordeeva et al., 2021). This approach has the drawback due to
   non-uniform read depth distribution between exons which makes it prone
   to false positive detection ([144]Ozden et al., 2022). It is also
   subjected to missing CNVs in regions with few exons. As another
   sequencing-based CNV detection approach, CNV-seq technology can
   potentially provide a unbiased coverage of the human genome and more
   precise estimation of CNV breakpoints ([145]Liang et al., 2014),
   however it also has the limitation of relative low read depth and
   resolution of 0.1 Mb in size which render it more suitable to detect
   large CNVs. Owing to the low cost, popularity in research use, abundant
   existing data, simultaneous usage for single nucleotide variant
   detection, and a number of well-tested CNV-calling algorithms, WES data
   are still a common choice of CNV analysis. With the gradually reduced
   sequencing cost and further improved CNV-calling algorithms, WES and
   CNV-seq are complimentary to each other and be combined in research and
   clinical diagnosis to improve clinical efficiency and diagnostic yield
   of multisystem anomalies ([146]Chen et al., 2022).

   There are a few limitations in our study. There are certain technical
   drawbacks for CNV calling based on WES as aforementioned. Small sample
   size is another major limitation of our study. To optimize study power
   and increase the possibility of identifying genetic components of HV,
   we adopted a family-based approach ([147]Borecki and Province, 2008)
   focusing on familial Hallux Valgus. To find the potentially pathogenic
   CNVs related to HV, we also filtered the case-only CNVs against all the
   CNVs carried by healthy individuals in the DGV database. Larger sample
   size will be needed to further investigate the substantial contribution
   of various CNV regions in the human genome ([148]Zarrei et al., 2015)
   to the etiology of HV, and particularly will gain statistical evidence
   for CNVs of uncertain clinical significance.

   In summary, from an unbiased genomic approach, this first study on the
   structural variants in familial HV gained us new insights into the HV
   pathogenesis, mediated by genetic variation of immunity and
   inflammation, and abnormal cytochrome P450 related metabolism. It
   provides evidence supporting the dominant inheritance mode of HV. Our
   results suggest that HV is a degenerative joint disease involving the
   dysregulation of immune system and metabolism system.

Acknowledgments