Abstract

Background

   Episodic angioedema with eosinophilia (Gleich syndrome) is a rare
   disease marked by periodic angioedema, fever, and severe eosinophilia,
   with limited understanding of its pathogenesis.

Objective

   We sought to identify pathogenic factors contributing to severe Gleich
   syndrome through a comprehensive multiomics approach, using
   whole-genome sequencing (WGS) and RNA sequencing (RNA-seq).

Methods

   A multiomics analysis was conducted on a 16- to 20-year-old female
   patient with severe Gleich syndrome, presenting with periodic high
   fever, extensive urticaria/eczema, and marked eosinophilia. The
   analysis included WGS and RNA-seq of blood samples.

Results

   WGS revealed high-impact pathogenic mutations that have the potential
   to significantly alter gene function in 16 genes, including PR domain
   containing 16 (gene involved in transcriptional regulation). RNA-seq
   identified differentially expressed genes linked to immune response
   regulation and viral defense. Combined z-score analysis of WGS and
   RNA-seq highlighted angiotensin-converting enzyme as a key gene, with
   significant downregulation during disease progression that normalized
   with treatment. IFNG was also implicated.

Conclusions

   The findings suggest that decreased angiotensin-converting enzyme
   expression, driven by PR domain containing 16 (gene involved in
   transcriptional regulation) mutations and altered IFNG expression, may
   contribute to increased bradykinin levels and activation of the
   arachidonic acid cascade, leading to the severe inflammation and
   angioedema characteristic of Gleich syndrome. This study underscores
   the utility of integrating WGS and RNA-seq data in elucidating the
   molecular basis of rare diseases and offers a foundation for developing
   therapeutic strategies for hypereosinophilic syndromes.

   Key words: Eosinophilia, Gleich syndrome, hypereosinophilic syndrome,
   episodic angioedema with eosinophilia, whole-genome sequencing, RNA
   sequencing, multiomics analysis
     __________________________________________________________________

   “Eosinophilia” is defined as a condition characterized by an absolute
   eosinophil count in the peripheral blood exceeding 500
   cells/μL.[27]^1^,[28]^2 Hypereosinophilia (HE) is defined as (1) an
   absolute eosinophil count in the peripheral blood exceeding 1500
   cells/μL in 2 blood samples taken at least 4 weeks apart; and/or (2)
   evidence of HE in tissues, which includes 1 or more of the following
   criteria: (a) eosinophils constituting more than 20% of the total
   nucleated cells in the bone marrow; (b) an abnormally intense
   eosinophilic infiltration in tissues as assessed by a pathologist; or
   (c) extensive deposition of eosinophil granule proteins identified by
   special staining techniques.[29]^3 Hypereosinophilic syndrome (HES) is
   defined by the following criteria: (1) the criteria for HE are met, (2)
   there is organ damage caused by peripheral tissue HE, and (3) other
   conditions that could cause organ damage have been excluded.[30]^3
   Another syndrome associated with HE is angioedema with eosinophilia,
   also known as Gleich syndrome. This rare clinical entity was first
   described by Gerald Gleich in 1984 and is characterized by angioedema
   with marked eosinophilia, accompanied by fever, periodic weight gain,
   and urticaria.[31]^4 HES was first proposed as a disease concept by
   Hardy and Anderson in 1968, with diagnostic criteria later developed by
   Chusid et al in 1975. Initially, it was considered a syndrome
   encompassing many conditions diagnosed primarily by exclusion. Since
   then, the causes of some conditions previously classified as HES have
   been clarified,[32]^5^,[33]^6 but many remain unexplained. The same
   applies to HE, including Gleich syndrome. The low prevalence of HE and
   HES, combined with the need to initiate treatment before confirming
   strict diagnostic criteria, has left many aspects of these diseases
   medically unexplored.

   In recent years, genomic analysis technologies, including whole-genome
   sequencing (WGS), whole-exome sequencing, and targeted sequencing, have
   gained attention as research tools for elucidating the causes and
   mechanisms of diseases.[34]^7 By comprehensively analyzing the entire
   spectrum of genetic variation, this technology can identify etiological
   genes and associated factors that are challenging to detect using
   conventional methods. WGS is particularly effective in the study of
   rare diseases, where many pathologies are strongly influenced by
   genetic factors.[35]8, [36]9, [37]10 In HE and HES, somatic mutations
   in hematopoietic cells have been investigated using targeted
   sequencing, as well as the relationship between clonal proliferation of
   eosinophils and novel disease-related mutations, using whole-exome
   sequencing.[38]^11^,[39]^12 However, knowledge about the genetic
   background and mechanisms underlying the pathogenesis of HE and HES
   remains incompletely understood because few cases have been analyzed
   using WGS. The aim of this study was to identify pathological factors
   in a patient diagnosed with HE/Gleich syndrome of unknown etiology. We
   investigated the pathological factors of idiopathic hypereosinophilia
   by analyzing WGS and RNA-sequencing (RNA-seq) data obtained from this
   patient.

Methods

   The materials and methods are described in this article’s Methods
   section in the Online Repository at [40]www.jaci-global.org.

Results

Clinical findings

   A 19-year-old woman experienced periodic episodes of fever and myalgia,
   with temperatures ranging from 38°C to 40°C, accompanied by chills and
   shivering, occurring 1 or 2 times a month since April 2022. Around
   October of the same year, she developed edema, urticaria, and a
   generalized skin rash that left pigmentation. Blood tests from this
   period revealed markedly increased eosinophilia. During febrile
   episodes, the patient exhibited transient neutrophil predominance and
   significantly elevated C-reactive protein levels. Thorough examinations
   were conducted at the CLINIC FOR Tamachi and Gunma University Hospital.
   A repeated clinical examination revealed a total white blood cell count
   of 19,200/μL, with an eosinophil count of 16.4% (3148/μL). Hemoglobin
   and platelet counts were within normal limits.

   Total IgE level was elevated at 504 IU/mL, but specific IgE levels for
   common grass and tree pollens, mites, animal dander, molds, and foods
   were within normal limits or negative. The patient, having been exposed
   to formalin, underwent a skin test for formalin sensitivity, the result
   for which was negative. Screening for parasitic antibodies was also
   negative as listed in [41]Table E5 (in the Online Repository available
   at [42]www.jaci-global.org).

   Rheumatoid factor, anti–nuclear antibodies, anti–ARS antibodies,
   myeloperoxidase-antineutrophil cytoplasmic antibodies, PR3-ANCA, IgG,
   IgA, IgM, C3, C4, and total hemolytic complement were all within normal
   limits. A peripheral blood smear examination revealed no morphological
   abnormalities. Lymphocyte surface marker tests showed the following:
   CD7 at 56.9%, CD3 at 44.7%, CD2 at 59.3%, CD4(−)CD8(+) at 9.3%,
   CD4(−)CD8(−) at 48.8%, CD4(+)CD8(−) at 41.2%, CD4(+)CD8(+) at 0.7%, and
   a CD4/CD8 ratio of 4.19.

   Various imaging studies, including computed tomography and magnetic
   resonance imaging, did not reveal any tumorous lesions. However,
   imaging showed enlarged axillary lymph nodes, splenomegaly, myositis
   centered around both shoulders, and bilateral pleural effusions. The
   soluble IL-2 receptor level was elevated at 1668 U/mL, but the
   pathology of the enlarged lymph nodes indicated dermatopathic
   lymphadenopathy, without evidence of lymphoma or malignancy, including
   immunostaining results. Pathological examination of the skin tissue
   raised a suspicion of vasculitis, though this was not definitive.

   Genetic testing was performed to exclude familial Mediterranean fever
   and hematolymphoid neoplasms, including chronic eosinophilic leukemia.
   Results were negative for the FIP1L1-PDGFRα fusion gene, PDGFB split
   signal, FGFR1 split signal, and MEFV gene mutation, with no clonal gene
   rearrangement detected in the T-cell receptor beta chain Cβ1. A summary
   of the clinical and laboratory information is provided below ([43]Table
   I), and full test results are available in this article’s Online
   Repository at [44]www.jaci-global.org ([45]Supplementary Information
   2).

Table I.

   Characteristics of the studied patient
   Clinical parameter Patient’s data Normal range
   Sex Female —
   Age of onset (y) 16-20 —
   Age at diagnosis (y) 20 —
   Symptoms Angioedema —
   Urticaria
   Pruritus
   Periodic fever
   Skin rash that leaves hyperpigmentation
   Muscle pain
   Splenomegaly
   Lymphadenopathy
   Serum total IgM (mg/dL) 136 46-260
   Serum total IgG (mg/dL) 1384 870-1700
   Serum total IgA (mg/dL) 106 110-410
   Serum total IgE (IU/mL) 504 ≤170
   Aberrant T-cell population No —
   C-reactive protein (mg/dL) 17.2 ≤0.14
   sIL-2R (U/mL) 1668 157-474
   Antinuclear antibody (titer) <1:40 <1:40
   MPO-ANCA (U/mL) <1.0 <3.5
   PR3-ANCA (U/mL) <1.0 <3.5
   C3 (mg/dL) 114 86-160
   C4 (mg/dL) 22 17-45
   Anti–ARS antibodies (index) <5.0 <25
   EBV DNA Not detected —
   Trunk/maxillofacial CT Splenomegaly —
   Enlarged axillary lymph nodes
   Trunk/maxillofacial MRI Suspected myositis around both shoulders —
   Reactive swelling of cervical and submandibular lymph nodes
   Small amount of pleural effusion
   Left axillary lymph node biopsy Findings suggestive of dermatopathic
   lymphadenopathy —
   No lymphoma or malignant findings
   Skin biopsy Possible vasculitis (unclear) —
   FIP1L1-PDGFRα fusion gene Negative —
   PDGFB split signal Negative —
   FGFR1 split signal Negative —
   MEFV gene mutations Negative —
   [46]Open in a new tab

   ARS, Aminoacyl-tRNA synthetase; CT, computed tomography; MPO-ANCA,
   myeloperoxidase-antineutrophil cytoplasmic antibodies; MRI, magnetic
   resonance imaging; PR3, proteinase 3; sIL-2R, soluble IL-2 receptor.

   On the basis of these results and the exclusion of autoimmune and
   hematological disorders and secondary eosinophilia, the patient was
   diagnosed with episodic angioedema with hypereosinophilia (Gleich
   syndrome). Treatment was initiated in July 2023 with oral
   glucocorticoids (prednisolone 30 mg/d), with subsequent tapering
   scheduled to the current maintenance dose of 1 mg/d. The patient
   achieved clinical and biological remission, generally by 30 days after
   treatment initiation and almost complete remission by 90 days.

WGS-based identification of patient-specific genetic variants and associated
mutations

   Patient-specific genetic variants were analyzed using WGS for
   diagnostic purposes. Three variant analysis tools—AnnotSV, SNPnexus,
   and wANNOVAR—were used. The AnnotSV analysis was filtered to include
   only those variants that were classified with The American College of
   Medical Genetics and Genomics (ACMG) class of 3 or higher ([47]Table
   II). As a result, 26 variants with an ACMG class of 3 were identified,
   with no variants classified as more severe. Further filtering using a
   LOEUF_bin score of 0 identified deletion mutations in the FOXO1, PR
   domain containing 16 (gene involved in transcriptional regulation)
   (PRDM16), SPAST, ITSN1, and DLGAP2 genes. Mutations in the FOXO1 gene
   have been previously associated with rhabdomyosarcoma, a type of
   cancer. Variants in the PRDM16 gene have been linked to various
   cancers. Mutations in the SPAST gene are known to cause neurological
   disorders, particularly hereditary spastic paraplegia. However, no
   disease associations were identified for mutations in the ITSN1 and
   DLGAP2 genes.

Table II.

   Analysis of SVs using AnnotSV
   Gene Chr Range Loc MOI GenCC_disease OMIM_phenotype
   FOXO1 13 40661542-40661624 CDS — Alveolar rhabdomyosarcoma
   PRDM16 1 3299300-3299421 CDS AD Dilated cardiomyopathy, familial
   isolated dilated cardiomyopathy,
   left ventricular noncompaction,
   left ventricular noncompaction 8 Dilated cardiomyopathy, left
   ventricular noncompaction 8
   PRDM16 1 3325997-3326054 CDS AD Dilated cardiomyopathy, familial
   isolated dilated cardiomyopathy, left ventricular noncompaction, left
   ventricular noncompaction 8 Dilated cardiomyopathy
   Left ventricular noncompaction 8
   SPAST 2 32111248-32111315 CDS AD;AR Charlevoix-Saguenay spastic ataxia,
   hereditary spastic paraplegia 4 Spastic paraplegia 4
   ITSN1 21 33876194-33876246 CDS — —
   DLGAP2 8 999891-999994 UTR AD Complex neurodevelopmental disorder —
   [48]Open in a new tab

   Chr, Chromosome; Loc, location; CDS, coding sequence; UTR, untranslated
   region; MOI, mode of inheritance; AD, autosomal dominant; AR, autosomal
   reccesive; FOXO, forkhead box O; SPAST, Spastin; ITSN, intersectin;
   DLGAP, DLG associated protein.

   In the SNPnexus-based analysis, the genes were listed as pathogenic
   according to ClinVar annotations ([49]Table III). The analysis
   identified the following relevant mutations and corresponding genes:
   PTPRJ (rs1566734), SLC4A1 (rs121912749), FGFR4 (rs351855), and
   C1GALT1C1 (rs17261572). In PTPRJ, a nonsynonymous single nucleotide
   variant (SNV) (A/C) was found at position 44,260,501 on chromosome 17,
   causing an amino acid change from Q276P and associated with carcinoma
   of the colon. In SLC4A1, a nonsynonymous SNV (G/A) was observed in the
   intron region at position 177,093,242 on chromosome 5, resulting in an
   amino acid change to G388R, suggesting an association with cancer
   progression and tumor cell proliferation. A nonsynonymous SNV (A/T) was
   identified in C1GALT1C1 at position 120,626,774 on the X chromosome,
   causing an amino acid change to D131E, and was associated with
   polyagglutinable erythrocyte syndrome.

Table III.

   Analysis of SNVs using SNPnexus
   Gene Chr Region Position SNV AA change ClinVar
   PTPRJ 11 Coding 48123823 A/C Q276P Carcinoma of colon
   SLC4A1 17 Coding 44260501 C/T G130R Spherocytosis type 4
   FGFR4 5 Intronic 177093242 G/A G388R Cancer progression and tumor cell
   motility
   C1GALT1C1 X Coding 120626774 A/T D131E Polyagglutinable erythrocyte
   syndrome
   [50]Open in a new tab

   AA, Amino acid.

   In wANNOVAR, we identified a series of mutations across diverse genes,
   focusing on exonic regions ([51]Table IV). As a result, we detected 2
   distinct frameshift mutations across multiple exons (14, 16, and 17) in
   the APC gene. These mutations were identified as c.5400delT
   (p.N1800Kfs45), c.5454delT (p.N1818Kfs45), c.5399_5400del
   (p.N1800Kfs7), and c.5453_5454del (p.N1818Kfs7), all of which were
   heterozygous. In OR13C8, a c.242delT mutation in exon 1 (p.A83Pfs6) was
   observed in a homozygous form, and in PCDH17, a c.237_238del mutation
   (p.D80Qfs78) was found in exon 1, also in a homozygous form. For GSPT1,
   multiple frameshift mutations were identified in exon 15, including
   c.1902_1909del (p.K635Sfs63), c.1491_1498del (p.K498Sfs63), and
   c.1905_1912del (p.K636Sfs∗63), all of which were homozygous. In MED1, 2
   heterozygous frameshift mutations were identified in exon 17,
   c.3210delA (p.T1071Qfs2) and c.3210_3211del (p.T1071Sfs28). In SLC35G4,
   synonymous SNVs c.C570T and c.C570C were detected in exon 1, both
   heterozygous, indicating no change in the amino acid sequence of the
   protein (p.G190G). In exon 4 of MADCAM1, a nonsynonymous SNV, c.T718C
   (p.S240P), was identified in a heterozygous form.

Table IV.

   Analysis of SNVs using wANNOVAR
   Chr Position Exon SNV Gene Exonic Func Zygosity HGVS
   chr5 112176745 Exon 14, exon 16,
   exon 17 T/- APC Frameshift deletion het c.5400delT(p.N1800Kfs∗45),
   c.5399_5400del(p.N1800Kfs∗7),
   c.5454delT(p.N1818Kfs∗45),
   c.5453_5454del(p.N1818Kfs∗7)
   chr9 107331690 Exon 1 A/- OR13C8 Frameshift deletion hom
   c.242delT(p.A83Pfs∗6)
   chr13 58206917 Exon 1 T/- PCDH17 Frameshift deletion hom
   c.237_238del(p.D80Qfs∗78)
   chr16 11966978 Exon 15 A/- GSPT1 Frameshift deletion hom
   c.1902_1909del(p.K635Sfs∗63),
   c.1491_1498del(p.K498Sfs∗63),
   c.1905_1912del(p.K636Sfs∗63)
   chr17 37565264 Exon 17 A/- MED1 Frameshift deletion het
   c.3210delA(p.T1071Qfs∗2),
   c.3210_3211del(p.T1071Sfs∗28)
   chr18 11610164 Exon 1 G/T SLC35G4 Synonymous SNV het c.C570T(p.G190G)
   chr18 11610164 Exon 1 G/C SLC35G4 Synonymous SNV het c.C570C(p.G190G)
   chr19 501719 Exon 4 T/C MADCAM1 Nonsynonymous SNV het c.T718C(p.S240P)
   [52]Open in a new tab

   HGVS, Human Genome Variation Society.

Enrichment analysis and protein-protein interaction network analysis using
RNA-seq differentially expressed genes

   The differentially expressed genes (DEGs) identified from the RNA-seq
   data in healthy subjects and the patient were further analyzed using
   gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG)
   pathway analyses. Upregulated genes were significantly enriched in the
   biological process categories of “regulation of response to biotic
   stimulus” and “response to virus,” as well as in the molecular function
   categories of “DNA-binding transcription factor binding” and
   “carbohydrate binding.” These genes were also significantly enriched in
   the cellular component categories of “endocytic vesicle” and “specific
   granule” ([53]Fig 1, A). Conversely, downregulated genes were
   significantly enriched in the biological process categories of
   “leukocyte-mediated immunity” and “cell killing,” in the molecular
   function categories of “carbohydrate binding” and “immune receptor
   activity,” and in the cellular component categories of “external side
   of plasma membrane” and “endoplasmic reticulum lumen” ([54]Fig 1, B).
   In addition, KEGG pathway analysis revealed that the most significantly
   enriched gene pathways were those associated with the “PI3K-Akt
   signaling pathway” and “human papillomavirus infection” ([55]Fig 1, C).

Fig 1.

   [56]Fig 1
   [57]Open in a new tab

   Pathway enrichment analysis using GO and KEGG, and PPI network
   analysis. Pathway enrichment analysis using GO and KEGG terms, and PPI
   network analysis based on patient RNA-seq data. (A) Pathway enrichment
   analysis using GO terms for DEGs of patient-derived upregulated genes
   in biological process (BP), molecular function (MF), and cellular
   component (CC) categories. The top 10 terms in each category are shown.
   (B) Pathway enrichment analysis using GO terms for DEGs of
   patient-derived downregulated genes. (C) Pathway enrichment analysis
   using KEGG terms for DEGs of patient-derived genes with variable
   expression. (D) Pathway enrichment analysis using GO terms for DEGs of
   patient-derived genes with variable expression. The vertical axis
   represents the percentage of DEGs in each term. PPI network analysis
   using DEGs of patient-derived genes with variable expression. Red
   circles represent nodes, and blue lines represent edges.

   Next, a protein-protein interaction (PPI) network analysis was
   performed using the DEGs. In the PPI network, which consisted of 185
   DEGs, 214 nodes and 676 edges were mapped on the basis of
   protein-protein interactions, with interaction scores greater than 0.7
   ([58]Fig 1, D). Using an interaction score greater than 0.95 to
   identify hub genes, the most highly connected hub genes were RSAD2
   (degree = 9), MX1 (degree = 8), ISG15 (degree = 8), and IFI44L
   (degree = 7). A heatmap of the DEGs and a volcano plot showing the
   time-series changes in each gene expression level are provided in
   [59]Fig E2 (in the Online Repository available at
   [60]www.jaci-global.org).

Identification of key genes using Z-score analysis

   Variants from WGS data labeled as HIGH in annotation impact, along with
   gene expression levels from RNA-seq data and centrality scores from PPI
   network analysis, were evaluated using z-score analysis ([61]Table V).
   Among the Z scores, IFI27 had the highest value (Z score: 4.47619923).
   OAS2 (degree: 19) and angiotensin-converting enzyme (ACE) (ABS_log2FC:
   9.10262032) exhibited the highest centrality score and absolute log[2]
   fold change, respectively. The smallest adjusted P value was observed
   for IFI27 (P[adj]: 4.18 × 10^−65). The effects of the mutations were as
   follows: splice acceptor variant and intron variant in 3 cases, start
   lost and conservative inframe deletion in 1 case, frameshift variant in
   3 cases, stop lost in 1 case, stop gained in 2 cases, and splice donor
   variant and intron variant in 1 case. Z-score values were used for GO
   term and KEGG pathway enrichment analysis as well as PPI network
   analysis (see [62]Fig E3 in this article’s Online Repository at
   [63]www.jaci-global.org). The plots of Effect Type count values for
   genomic variants in this patient are provided in [64]Fig E4 (in the
   Online Repository available at [65]www.jaci-global.org).

Table V.

   Identification of key disease-associated genes based on Z-score values
   Gene POS Variant Effect Degree ABS_log[2]FC P[adj] Z_score
   IFI27 14:94105764 G/T splice_acceptor_variant &
   intron_variant 14 7.22991464 4.18 × 10^−65 4.47619973
   IFI27 14:94115784 TGGCCATGGC/T start_lost &
   conservative_inframe_deletion 14 7.22991464 4.18 × 10^−65 4.47619973
   IFI44 1:78662885 G/GT frameshift_variant 18 3.62858894 1.20 × 10^−30
   3.63485564
   OAS2 12:113010483 A/G stop_lost 19 2.51335138 7.25 × 10^−23 3.33028482
   EPSTI1 13:42888298 T/TTCAGG frameshift_variant 17 3.22174861
   6.81 × 10^−18 3.27200626
   ACE 17:63486300 C/A stop_gained 4 9.10262032 0.00107677 3.11768804
   MSR1 8:16186158 C/T splice_donor_variant &
   intron_variant 4 3.28217749 5.35 × 10^−04 0.73409481
   COL6A2 21:46125455 A/AC splice_acceptor_variant &
   intron_variant 1 3.28721175 7.06 × 10^−18 0.34946627
   SIGLEC10 19:51417359 C/T splice_acceptor_variant &
   intron_variant 2 1.88542826 1.09 × 10^−08 −0.0807563
   RETSAT 2:85350974 TA/T frameshift_variant 3 1.48768267 2.26 × 10^−04
   −0.1410527
   CC2D2A 4:15480736 C/T stop_gained 1 1.84782388 4.07 × 10^−04 −0.4083765
   [66]Open in a new tab

   ABS, Absolute value; FC, fold change; POS, position.

Dynamics of gene expression in pretreatment and posttreatment time series

   The temporal changes in expression levels for selected genes were
   assessed. For each gene group, log[2] fold changes were calculated at 3
   time points: pretreatment, 1 month, and 3 months posttreatment.
   Significant changes in expression levels were observed when comparing
   pretreatment samples to those from healthy subjects, particularly for
   the genes ACE, CC2D2A, COL6A2, EPST11, IFI27, IFI44, MSR1, OAS2,
   RETSAT, and SIGLEC10 ([67]Fig 2, A). Overall, many genes exhibited
   significant temporal changes in expression levels compared with
   pretreatment. For example, EPST11 was strongly upregulated before
   treatment, with this upregulation alleviated after 1 month and further
   reduced after 3 months. A similar pattern was observed for IFI27,
   IFI44, and CC2D2A. Conversely, ACE and RETSAT exhibited a reverse
   trend, with marked suppression of expression before and 1 month
   posttreatment, followed by recovery after 3 months. PTPRJ and SIGLEC10
   showed reduced expression at 1 and 3 months posttreatment compared with
   healthy subjects, whereas OR13C8 and SLC35G had a read count of 0 in
   both the patient and healthy subjects. Enrichment analysis revealed
   that GO terms related to the regulation of immune response,
   leukocyte-mediated immunity, and cell killing were enriched in
   pretreatment samples but were no longer enriched at 1 and 3 months
   posttreatment ([68]Fig 2, B). In contrast, GO terms involved in
   chemotaxis, cell migration, and cell differentiation were newly
   enriched at 1 and 3 months posttreatment. The pretreatment samples were
   plotted in the third quadrant, whereas the 1- and 3-month posttreatment
   samples were plotted in the first quadrant ([69]Fig 2, C). A heatmap
   identified genes with high loading on the negative axis of PC1,
   including BFAR, XG, RCE1, and COL18A1 ([70]Fig 2, D). For each gene
   showing a recovery trend after 3 months, the OMIM database was used to
   search for causative genes with phenotypes similar to those observed in
   the patient. The results indicated that a decrease in ACE led to an
   increase in bradykinin, contributing to the development of
   angioedema.[71]^13^,[72]^14

Fig 2.

   [73]Fig 2
   [74]Open in a new tab

   Analysis of time-series changes in gene expression levels and pathway
   enrichment. Analysis of changes in gene expression levels and enriched
   pathways using patient samples collected at pretreatment, 1-month, and
   3-month time points. (A) Comparison of expression levels of candidate
   target genes at each time point. Three asterisks indicate an adjusted P
   value of less than .01, representing a significant difference from the
   healthy group. (B) Transition of enriched GO terms at each time point,
   representing biological process (BP), molecular function (MF), and
   cellular component (CC). (C) PCA plot using RNA-seq data from healthy
   subjects and patient at each time point. (D) PCA for each gene. Heatmap
   displaying loading values, showing PC1 to PC5, clustered in rows. PCA,
   Principal-component analysis.

Exploration of upstream and master regulators of the ACE gene by Ingenuity
Pathway Analysis

   Analysis of the dynamics of gene expression in pretreatment and
   posttreatment time series suggested a relationship between the ACE gene
   and the patient with the disease. To further explore this, we
   investigated the relationship between the ACE gene and the disease by
   visualizing upstream regulators and causal networks using Ingenuity
   Pathway Analysis (IPA) for the ACE gene. When we searched for upstream
   regulators of ACE using pretreatment RNA-seq data, we found that the
   IFNG gene had the highest Activation Z score of 4.403 (see [75]Fig E5,
   A, in this article’s Online Repository at [76]www.jaci-global.org).
   Similarly, in the causal network’s Activation Z scores, the IFNG gene
   also had the highest Activation Z score of 4.217 ([77]Fig E5, B). The
   network visualization showed an interaction from IFNG in the
   extracellular space to ACE in the plasma membrane ([78]Fig E5, C). To
   further investigate the association between IFNG and this patient, a
   drill-down analysis using IPA for the IFNG gene identified an
   association with idiopathic eosinophilia.[79]^6 Pathways enriched in
   IPA for this disease, including pathway diagrams for RNA virus
   infection and network diagrams for angioedema, which are suggested to
   be relevant to the disease pathogenesis, are provided in [80]Figs E6
   and [81]E7 (in the Online Repository available at
   [82]www.jaci-global.org).

Discussion

   The aim of this study was to identify the pathological factors in a
   patient diagnosed with HE/Gleich syndrome of unknown etiology using WGS
   and RNA-seq analysis.

   The patient presented with clinical symptoms consistent with Gleich
   syndrome, including marked eosinophilia exceeding 5000/μL at its peak,
   periodic fever, and recurrent edema and urticaria. However, the patient
   also exhibited a generalized skin rash with pathological findings of
   vasculitis and pigmentation, transient neutrophil-predominant leukemoid
   reactions associated with fever, markedly elevated C-reactive protein
   levels, and splenomegaly. These findings are not commonly observed in
   Gleich syndrome, but similar cases have been reported.[83]15, [84]16,
   [85]17 Although this patient did not meet the strict diagnostic
   criteria for HES, because eosinophilic infiltration was not clearly
   demonstrated in various organ assessments, some consider Gleich
   syndrome itself to be a broader category of HES.[86]^3 This reflects
   the fact that the disease concepts of HE and HES are not fully defined,
   and their classification has evolved over time. In any case, the
   patient exhibited a very advanced and dramatic clinical presentation,
   with no spontaneous remission for more than 6 months, leading to a
   diagnosis of severe Gleich syndrome after all other differential
   diagnoses were ruled out.

   Gleich syndrome is a rare disease with no currently approved or clearly
   defined diagnostic criteria, and to date, fewer than 100 adult cases
   have been reported.[87]^16 Nevertheless, patients with Gleich syndrome
   typically present with homogeneous clinical symptoms and
   well-characterized laboratory findings, suggesting that the disease
   should be considered a distinct eosinophilic disorder, likely caused by
   a common pathogenic mechanism.[88]^17 In cases with clear and intense
   symptoms, as in this case, even a small number of disease samples are
   likely to lead to the identification of some pathological factors by
   multiomics analysis, which is very significant for the purpose of
   elucidating the pathology of rare diseases. This is very significant
   for the purpose of elucidating the pathology of rare diseases.

   In addition, within the molecular function category, several GO terms
   were enriched for carbohydrate binding, monosaccharide binding, glucose
   binding, proteoglycan binding, and glycan-related genes, such as
   “pentosyltransferase activity.” Glycosylation is a common
   posttranslational modification of proteins and lipids and serves as a
   critical recognition determinant in cell-immune system interactions.
   Immune and stromal cells are equipped with glycan-binding proteins that
   sense and decode diverse glycans.[89]^18^,[90]^19 This network of
   glycans and glycan-binding proteins is crucial for the recognition of
   pathogens and the regulation of inflammatory and autoimmune responses,
   with alterations in the cellular glycome ultimately leading to
   pathological phenotypes.[91]20, [92]21, [93]22 C1GALT1C1 (rs17261572),
   also known as Cosmc, identified in variant analysis, galactosylates Tn
   antigen (GalNAcα1-Ser/Thr-R) during the biosynthesis of mucin-type
   O-glycans to core 1 Galβ1-3GalNAcα1-Ser/Thr (T antigen)[94]^23 (see
   [95]Fig E8 in this article’s Online Repository at
   [96]www.jaci-global.org). Given that glycans function as drug
   recognition markers, the potential involvement of glycan-related genes
   in drug regulation has been investigated in public database
   analyses.[97]^24 The relationship between rs17261572 on C1GALT1C1 and
   HE/Gleich syndrome, as well as its connection to other glycan-related
   genes, warrants further investigation.

   The Z-score values derived from WGS annotation impact, along with the
   ranking of DEGs and PPI network analysis from RNA-seq data, identified
   IFI27, OAS2, and ACE as key candidate genes. In this case, the
   expression levels of these genes were variable compared with healthy
   subjects and showed a recovery trend after treatment. It is well
   established that a reduction in ACE leads to an increase in bradykinin,
   which contributes to the development of angioedema.[98]^13^,[99]^14
   Bradykinin itself induces angioedema[100]^25 and further increases
   phospholipase A2 activity, enhances arachidonic acid metabolism, and
   triggers a wide range of inflammatory responses.[101]^26 In addition,
   large amounts of bradykinin have been reported to recruit
   eosinophils.[102]^26^,[103]^27 In this patient, who presented with
   severe systemic inflammation in addition to edema, ACE was found to
   play a major role as a pathological factor. The patient achieved a good
   state of remission with oral prednisolone treatment, not only due to
   the general anti-inflammatory effects of corticosteroids but also
   because corticosteroids have been suggested to directly restore ACE
   expression,[104]^28 which may have contributed to the improvement in
   the condition.

   Notably, there is a case report of a patient with congenital ACE
   deficiency presenting with recurrent angioedema.[105]^29 This finding
   further supports the involvement of ACE as a pathological factor in
   diseases characterized by angioedema. However, in the present case, no
   mutation in the ACE gene itself critically defines a loss of
   expression, suggesting that an alteration in an upstream factor leading
   to reduced ACE expression may be involved.

   PRDM16, a gene with a confirmed structural abnormality in this case,
   has been shown to reduce ACE mRNA expression in the systemic RAAS when
   conditionally knocked out in the heart, although the detailed mechanism
   remains unclear.[106]^30 Although ACE expression in the intrarenal RAAS
   is compensated for in this conditionally knocked out model, it is
   plausible that the PRDM16 mutation observed in this patient results in
   a systemic reduction in ACE expression.

   Although the direct mechanism by which PRDM16 mutations lead to reduced
   ACE expression is not fully understood and warrants further
   investigation, the 2-hit model will be proposed ([107]Fig 3). Namely,
   based on the PRDM16 mutation, some acquired event that causes an INF-γ
   “storm” is the trigger, leading to abnormal activation of the
   bradykinin pathway through reduced ACE. In this case, hyperactive
   eosinophil biology was not observed, which is probably because
   eosinophilia was caused mainly by the ACE-bradykinin pathway. This may
   explain the difference in clinical presentation from eosinophilia
   caused by strictly immunologic mechanisms (ie, secretion of
   eosinophil-stimulating signals such as IL-5 from T[H]2 and/or ICL2
   cells).

Fig 3.

   [108]Fig 3
   [109]Open in a new tab

   The role of ACE as a pathogenic factor in a type of HE/Gleich syndrome.
   This figure illustrates the pathway by which a PRDM16 mutation leads to
   pathogenic outcomes in HE/Gleich syndrome. The mutation downregulates
   ACE expression, resulting in elevated bradykinin levels. This triggers
   the arachidonic acid cascade, which contributes to both angioedema and
   severe inflammation. In addition, IFN-γ levels influence ACE
   expression, further exacerbating these processes.

   This study has certain limitations. Gleich syndrome does not behave as
   a genetic disorder, and the pathological model is highly speculative.
   Although it is certainly significant that newly identified factors such
   as ACE and PRDM16 can be involved in the pathogenesis, these results
   would need to be confirmed in a larger cohort. Although multiomics
   analyses using WGS and RNA-seq have advanced our understanding of the
   pathogenesis of HE/Gleich syndrome at the genetic and gene expression
   levels, these approaches must be supplemented with protein-level
   analyses and in vivo model-based studies. Confirming blood levels of
   bradykinin and ACE would further substantiate the pathogenesis of this
   condition. Moreover, other factors influencing ACE expression need to
   be investigated, and detailed mechanisms must be elucidated. Gene
   expression data alone do not adequately capture posttranslational
   modifications, protein interactions, or functional dynamics within
   cells and tissues. In vivo models can provide a more comprehensive
   understanding of disease mechanisms within a physiological context.
   Therefore, future research should incorporate proteomic analyses, gene
   editing techniques, and animal models.

Key messages.

     * •
       PRDM16 mutations and decreased ACE expression likely contribute to
       severe inflammation and angioedema in Gleich syndrome.
     * •
       Integration of WGS and RNA-seq data reveals molecular insights that
       could aid in developing therapies for hypereosinophilic syndromes.

Disclosure statement

   There are no sources of funding to declare for the research reported in
   this manuscript.

   Disclosure of potential conflict of interest: The authors declare that
   they have no relevant conflicts of interest.

Acknowledgments