Evaluating the strength of evidence for genes implicated in peroxisomal disorders using the ClinGen clinical validity framework and providing updates to the peroxisomal disease nomenclature

Abstract

Peroxisomal disorders are heterogeneous in nature, with phenotypic overlap that is indistinguishable without molecular testing. Newborn screening and gene sequencing for a panel of genes implicated in peroxisomal diseases are critical tools for the early and accurate detection of these disorders. It is therefore essential to evaluate the clinical validity of the genes included in sequencing panels for peroxisomal disorders.

The Peroxisomal Gene Curation Expert Panel (GCEP) assessed genes frequently included on clinical peroxisomal testing panels using the Clinical Genome Resource (ClinGen) gene–disease validity curation framework and classified gene-disease relationships as Definitive, Strong, Moderate, Limited, Disputed, Refuted, or No Known Disease Relationship. Subsequent to gene curation, the GCEP made recommendations to update the disease nomenclature and ontology in the Monarch Disease Ontology (Mondo) database.

Thirty-six genes were assessed for the strength of evidence supporting their role in peroxisomal disease, leading to 36 gene-disease relationships, after two genes were removed for their lack of a role in peroxisomal disease and two genes were curated for two different disease entities each. Of these, 23 were classified as Definitive (64%), one as Strong (3%), eight as Moderate (23%), two as Limited (5%), and two as No known disease relationship (5%). No contradictory evidence was found to classify any relationships as Disputed or Refuted. The gene-disease relationship curations are publicly available on the ClinGen website (https://clinicalgenome.org/affiliation/40049/). The changes to peroxisomal disease nomenclature are displayed on the Mondo website (http://purl.obolibrary.org/obo/MONDO_0019053).

The Peroxisomal GCEP-curated gene-disease relationships will inform clinical and laboratory diagnostics and enhance molecular testing and reporting. As new data will emerge, the gene-disease classifications asserted by the Peroxisomal GCEP will be re-evaluated periodically.

1. Introduction

1.1. Peroxisomal Disorders

Peroxisomal disorders are grouped into two broad categories - the peroxisome biogenesis disorders (PBDs) and single peroxisomal protein/enzyme defects (1). There are 16 mammalian PEX genes, which encode the PEX proteins (also termed peroxins) required to assemble functional peroxisomes. Autosomal recessive defects in any one of the 13 PEX genes cause a specific PBD, Zellweger spectrum disorder (ZSD), a term which replaces older names designated before a common peroxisome etiology was known (Zellweger syndrome, neonatal adrenoleukodystrophy, infantile Refsum disease) and other descriptive terms (Heimler syndrome). Defects in PEX7 and the PEX5 domain that binds to PEX7 cause the PBD, Rhizomelic Chondrodysplasia Punctata (RCDP), characterized by a phenotype distinct from ZSD (2). Recently, it was reported that specific pathogenic variants in PEX6 could function as partial autosomal dominant alleles (3). No human disease has yet been reported with defects in PEX11A or PEX11G.

Defects in ABCD1 cause the most common single peroxisome protein deficiency, X-linked adrenoleukodystrophy (ALD). The ABCD1 protein is an integral peroxisome membrane protein in the class of ABC half-transporters and is required for the transport of very long chain fatty acids (VLCFA) into the peroxisomal matrix where they can be broken down (4). The remaining single enzyme/protein defects are much rarer and feature peroxisomal enzymes/proteins involved in transport or oxidation of very long chain and branched chain fatty acids, and plasmalogen biosynthesis. Defects in these genes result in phenotypes that overlap ZSD and RCDP.

The estimated incidence of ZSD has been reported as 1/50,000 (1), and a report using carrier statistics from public databases estimated an incidence of about 1/80,000 (5). The incidence of ALD is estimated to be 1/15,000 (6). However, the true incidence will be determined only over the next several years with the continued implementation of newborn screening for peroxisomal disorders in the USA (7). Diagnosis has traditionally been from clinical suspicion to biochemical testing for peroxisome dysfunction, followed by the assessment of enzyme function and/or molecular confirmation of the gene defect. Recently, whole exome sequencing (WES) and gene panels have superseded many of the functional biochemical studies, although these studies still provide valuable evidence for the accurate classification of the genetic variants identified and are particularly critical when molecular variants of uncertain significance are identified (8).

Clinically, PBD and related single enzyme/protein defects are progressive neurodegenerative and multi-system disorders without effective therapies. The exception is cerebral ALD, in which allogeneic stem cell transplant can arrest inflammatory leukodystrophy if performed at an early stage of the demyelination process. ABCD1 stem cell gene therapy has also proven efficacious to treat childhood cerebral ALD at the early stages. Newborn screening for elevated very long chain fatty acids (VLCFAs) was implemented to detect ALD but will secondarily detect all PBD/single protein/enzyme defects with elevated VLCFAs. This will expedite the early diagnosis of these conditions and contribute to existing and novel variant classifications.

1.2. Need for gene curation

Newborn screening identifying elevated C26:0-lysophosphatidyl choline (C26:0-LPC) and molecular genetic testing are critical tools for the early detection of peroxisomal disorders, which aid in the initiation of personalized management and treatment strategies for affected individuals as well as appropriate counseling for carriers. Thus, it is crucial to evaluate the clinical validity of genes and associated variants implicated in peroxisomal disorders. The ClinGen peroxisomal disorders gene curation expert panel (Peroxisomal GCEP) was convened in 2019, under the umbrella of the Inborn Errors of Metabolism clinical domain working group, and tasked with assessing the strength of evidence supporting peroxisomal gene-disease relationships using the ClinGen gene-disease clinical validity framework (9).

In addition to curating the genes involved in peroxisomal disorders, the Peroxisomal GCEP also focused on updating the peroxisomal disease nomenclature. As a result of the gene curation process, the Peroxisomal GCEP identified inconsistencies in the peroxisomal disease entity nomenclature, including (a) the use of old classifications that do not reflect the current understanding of the disease spectrum and (b) inconsistent terminology for disease entities that varies between clinical phenotype and disease mechanism-based classifications. The Peroxisomal GCEP proposed changes to revise and update the terminology within the Monarch Disease Ontology (Mondo, URL: https://www.ebi.ac.uk/ols/ontologies/mondo) by curating the peroxisomal genes for appropriate disease entities. Here, we discuss the results of the gene curation process and the updates to Mondo.

2. Materials and Methods

2.1. Expert Panel Composition

The membership of the ClinGen Peroxisomal GCEP consists of experts in the clinical (NB, SS, AM, HB), biochemical and/or molecular genetics (IDB, RM, SS, TY, AP) fields, biocurators (SM, MM, JG, JM) and a coordinator (MW) with expertise in the ClinGen gene-disease validity assessment process. The Peroxisomal GCEP webpage can be accessed at https://www.clinicalgenome.org/affiliation/40049/.

2.2. Gene selection for curation

Discussion within the expert panel led to the selection of thirty six genes for analysis by the Peroxisomal GCEP. The genes were categorized under three tiers: the first tier or primary gene list (15 genes) consisted of ABCD1, which causes ALD, and all the PEX genes known to cause PBDs; the second tier included 11 genes that cause single protein/enzyme defects; and the third tier comprised 10 genes that were included on at least one of 26 peroxisomal panels (last accessed February, 2022) reported in the National Institutes of Health Genetic Testing Registry (NIH GTR, http://www.ncbi.nlm.nih.gov/gtr/). As expected, most of the tier 1 (PEX) genes were covered in a majority of the panels, while fewer panels included the tier 2 and 3 genes. Of note, ABCD1 was included in only over half of the peroxisomal panels, but was present in a number of leukodystrophy and intellectual disability/autism panels, while PEX11A and PEX11G were found only on two panels from a research laboratory.

2.3. Curating a gene for the most appropriate disease entity

The ClinGen curation process includes a precuration step that helps the expert panel determine the most appropriate disease entity to assign to genes associated with multiple disease and/or phenotypic entities. This involves applying the lumping and splitting guidelines (10), which take into consideration the phenotypic variability, inheritance patterns, molecular mechanisms and assertions in the literature and by other nosological authorities for the given gene. The biocurator reviewed the primary literature as well as assertions in Online Mendelian Inheritance in Man (OMIM), Orphanet, Mondo, and GeneReviews (when available) for each gene and presented the evidence and a recommendation on whether multiple entities must be lumped into one entity for curation or split into separate gene-disease relationship curations. In several instances, the Peroxisomal GCEP recommended new peroxisomal disease terminology to be used in Mondo (See Section 2.5). All precuration information and decisions on the lumping and splitting are published and available on the ClinGen website.

2.4. Curating gene-disease relationships

The Peroxisomal GCEP employed the ClinGen gene-disease validity curation process (9) versions 7–9 to evaluate 36 gene-disease relationships (GDAP1 and DYM were excluded from the gene list after precuration; FAR1 and ACOX1 were each curated for two different disease entities, ACOX2 was previously curated by the ClinGen Inborn Errors of Metabolism (IEM) GCEP). The biocurator reviewed primary evidence supporting each gene-disease relationship and assigned preliminary classifications in the ClinGen gene curation interface (GCI) using the point system as follows: genetic evidence contributed a maximum of 12 points while experimental evidence was scored a maximum of 6 points. Based on the total number of points for a gene-disease relationship, one of the following classifications was assigned: No known disease relationship (0 points for genetic evidence), Limited (0.1–6 points), Moderate (7–11 points), Strong (12–18 points), Definitive (12–18 points with replication over time). There is no difference in the accumulated points between the Strong and Definitive classifications; however, at least 2 independent publications over 3 years’ time should be available for definitive classifications. The calculated classifications can be manually modified if the expert panel deemed such a change appropriate and when there is data to support the change. The Peroxisomal GCEP modified the classification of the ACOX1 – ACOX1 Upregulation relationship (See section 3.2) Any modifications made are published to the website along with the rationale behind the changes.

Well-known gene-disease relationships that a biocurator could classify as Definitive were submitted to at least two experts by email for review. If the two experts agreed with the evidence and classification, the curation was reviewed and approved offline. For gene-disease relationships that did not reach a Definitive classification or when there were questions regarding the evidence, the biocurator presented the curation for discussion on a conference call. To avoid conflicts of interest, experts who were involved in the discovery of a particular gene recused themselves from approving its curation classification following ClinGen policies. Following expert approval, all gene-disease relationship curations and precurations from the Peroxisomal GCEP were made publicly available on the ClinGen website (https://clinicalgenome.org/affiliation/40049/).

2.5. Mondo disease nomenclature revisions

ClinGen recommends the use of Mondo identifiers (IDs) for gene-disease relationship curations in the gene curation interface (GCI). Mondo IDs represent unique disease entities in the Mondo database. Upon completing the evaluation of clinical validity of all the genes on the gene list, the Peroxisomal GCEP worked closely with the curators at Mondo to revise the Peroxisomal disease nomenclature and reorganize the ontology by introducing and/or modifying several superclasses and subclasses. A Google document for collaborative work was created, which listed all the proposed terminologies and revised ontology (Supplemental file). Suggestions to Mondo disease nomenclature are made by raising issue tickets on GitHub (https://github.com/monarch-initiative/mondo/issues), which allows other involved members of the community to view and comment on the issues. The Peroxisomal GCEP followed this routine method to make a bulk edit and held several conference calls, including members from the Mondo team, to discuss and clarify the proposed changes and reach a consensus.

3. Results

The ClinGen Peroxisomal GCEP evaluated 36 gene-disease relationships. Of these, 23 were classified as “Definitive” (66%), one as Strong (3%), eight as “Moderate” (23%), one as “Limited” (3%), and two as “No known disease relationship” (5%) (Fig.1). With the exception of the ACOX1-Mitchell syndrome curation, there were no gene-disease relationships leading to a “Strong” classification, as all other curations reached a score >12 and also had evidence that was replicated over time. Moreover, no contradictory evidence was identified to classify any relationships as Disputed or Refuted.

Fig 1.

Gene-Disease classification distribution and proportion of gene-disease relationships at each classification.

3.1. Tier 1 Genes

The tier 1 genes (ABCD1 and all the PEX genes, except PEX11A and PEX11G) all reached a Definitive classification with an overall score ≥12 points and replication over time (Table 1). PEX11B and PEX14 received a total score of 11.5 and 11.25 points, respectively, which were automatically rounded up to 12 points as the Peroxisomal GCEP was in agreement that these were Definitive relationships.

Table 1.

Curation summary of the 34 genes evaluated by the Peroxisomal GCEP

HGNC Gene Symbol	OMIM associated disease entities; Inheritance (MIM #) (Inheritance pattern)	Associated MonDO disease entity (Mondo ID)	Peroxisomal GCEP recommended Mondo disease entity (Mondo ID)	Classification (points)	# Panels (Total = 26)	Notes on Proposed Mondo terminology
ABCD1	ALD (MIM: 300100) (X-linked)	ALD (MONDO:0018544)	ABCD1 deficiency (MONDO:0018544)	Definitive 12 GE, 6 EE, r/t	15	SpC of cerebral ALD (MONDO:0010247), AMN (MONDO:0015339) and isolated adrenal insufficiency (MONDO:0100315)
PEX1	Heimler syndrome 1 (MIM: 234580); PBD 1A (Zellweger) (MIM: 214100); PBD 1B (NALD/IRD) (MIM: 601539) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX1 defect (MONDO:0100259)	Definitive 12 GE, 6 EE, r/t	26	SpC of classic PBD 1A (MONDO:0008953) and non-classic PBD 1B (MONDO:0011101)
PEX2	PBD 5A (Zellweger) (MIM: 614866); PBD 5B (MIM: 614867) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX2 defect (MONDO:0100260)	Definitive 10.5 GE, 6 EE, r/t	25	SpC of classic PBD 5A (MONDO:0013932) and non-classic PBD 5B (MONDO:0013933)
PEX3	PBD 10B (MIM: 617370); PBD 10A (Zellweger) (MIM: 614882) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX3 defect (MONDO:0100261)	Definitive 12 GE, 4.5 EE, r/t	25	SpC of classic PBD 10A (MONDO:0013948) and non-classic PBD 10B (MONDO:0054549)
PEX5	PBD 2A (Zellweger) (MIM: 214110); PBD 2B (MIM: 202370); RCDP, type 5 (MIM: 616716) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX5 defect (MONDO:0100262) PBD due to PEX5 defect in the PEX7-binding domain (MONDO:0100265)	Definitive 7.5 GE, 5 EE, r/t	25	SpC of classic PBD 2A (MONDO:0008954) and non-classic PBD 2B (MONDO:0008736). Also includes RCDP, type 5 (MONDO:0014743) under non-Zellweger spectrum disorders
PEX6	Heimler syndrome 2 (MIM: 616617); PBD 4A (Zellweger) (MIM: 614862); PBD 4B (MIM: 614863) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX6 defect (MONDO:0100263)	Definitive 12 GE, 3 EE, r/t	26	SpC of classic PBD 4A (MONDO:0013930) and non-classic PBD 4B (MONDO:0013931)
PEX7	PBD 9B (MIM: 614879); RCDP, type 1 (MIM: 215100) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX7 defect (MONDO:0100272)	Definitive 12 GE, 5 EE, r/t	23	SpC of RCDP, type 1 (MONDO:0008972) and adult refsum disease due to PEX7 defect (MONDO:0100307) under non-Zellweger spectrum disorders
PEX10	PBD 6A (Zellweger) (MIM: 614870); PBD 6B (MIM: 614871) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX10 defect (MONDO:0100264)	Definitive 12 GE, 5.5 EE, r/t	26	SpC of classic PBD 6A (MONDO:0013936) and non-classic PBD 6B (MONDO:0013937)
PEX11A	No associated entity	PBD (MONDO:0019234)	N/A	No known disease relationship 0 GE, 1 EE	2	No Mondo term proposed
PEX11B	PBD 14B (MIM: 614920) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX11B defect (MONDO:0100279)	Definitive 7 GE, 4.5 EE, r/t	23	SpC of non-classic PBD 14B (MONDO:0013967)
PEX11G	No associated entity	PBD (MONDO:0019234)	N/A	No known disease relationship 0 GE, 0 EE	2	No Mondo term proposed
PEX12	PBD 3A (Zellweger) (MIM: 614859); PBD 3B (MIM: 266510) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX12 defect (MONDO:0100266)	Definitive 12 GE, 4.5 EE, r/t	26	SpC of classic PBD 3A (MONDO:0013927) and non-classic PBD 3B (MONDO:0009959)
PEX13	PBD 11A (Zellweger) (MIM: 614883); PBD 11B (MIM: 614885) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX13 defect (MONDO:0100267)	Definitive 11 GE, 5.5 EE, r/t	25	SpC of classic PBD 11A (MONDO:0013949) and non-classic PBD 11B (MONDO:0013950)
PEX14	PBD 13A (Zellweger) (MIM: 614887) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX14 defect (MONDO:0100268)	Definitive 6.25 GE, 5 EE, r/t	25	SpC of classic PBD 13A (MONDO:0013952)
PEX16	PBD 8A (Zellweger) (MIM: 614876); PBD 8B (MIM: 614877) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX16 defect (MONDO:0100269)	Definitive 12 GE, 3.5 EE, r/t	25	SpC of classic PBD 8A (MONDO:0013942) and non-classic PBD 8B (MONDO:0013943)
PEX19	PBD 12A (Zellweger) (MIM: 614886) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX19 defect (MONDO:0100270)	Definitive 7.75 GE, 4 EE, r/t	25	SpC of classic PBD 12A (MONDO:0013951)
PEX26	PBD 7A (Zellweger) (MIM: 614872); PBD 7B (MIM: 614873) (Autosomal recessive)	PBD (MONDO:0019234)	PBD due to PEX26 defect (MONDO:0100271)	Definitive 12 GE, 2.5 EE, r/t	26	SpC of classic PBD 7A and non-classic PBD 7B (MONDO:0013939)
ABCD3	CBASD 5 (MIM: 616278) (Autosomal recessive)	CBASD (MONDO:0018841)	ABCD3 CBASD (MONDO:0014564)	Limited 1 GE, 4 EE	10	SbC of DPT (MONDO:0100372)
ACBD5	Retinal dystrophy with leukodystrophy (MIM: 618863) (Autosomal recessive)	ACBD5 deficiency (MONDO:0100112)	ACBD5 deficiency (MONDO:0100112)	Moderate 6 GE, 3.5 EE	3	SbC of DPT (MONDO:0100372) and DPB (MONDO:0017986)
ACOX1	Peroxisomal ACOX deficiency (MIM: 264470) (Autosomal recessive)	Peroxisomal ACOX deficiency (MONDO:0009919)	ACOX1 deficiency (MONDO:0009919)	Definitive 12 GE, 5.5 EE, r/t	18	SbC of disorder of peroxisomal beta-oxidation (MONDO:0019233)
	Mitchell syndrome (MIM: 618960) (Autosomal dominant)	Mitchell syndrome (MONDO_0030073)	ACOX1 upregulation (MONDO_0030073)	Strong*5.5 GE, 2 EE	7.5	Sbc of disorder of defective peroxisome oxidative status. “Upregulation” to indicate gain-of-function mechanism
AGPS	RCDP, type 3 (MIM: 600121) (Autosomal recessive)	RCDP MONDO:0015776	AGPS deficiency (MONDO:0100274)	Definitive 7.5 GE, 6 EE, r/t	14	SbC of DPB (MONDO:0017986)
AMACR	AMACR deficiency (MIM: 614307); CBASD 4 (MIM: 214950) (Autosomal recessive)	CBASD (MONDO:0018841)	AMACR deficiency (MONDO:0013681)	Moderate 8.1 GE, 2 EE	17	CBASD 4 (MONDO:0008967) is a SbC
EHHADH	Fanconi renotubular syndrome 3 (MIM: 615605) (Autosomal dominant)	Fanconi renotubular syndrome 3 (MONDO_0014275)	N/A	Limited 2.5 GE, 1 EE	2	No Mondo term proposed
FAR1	Peroxisomal FAR1 disorder (MIM:616154) Cataracts, spastic paraparesis, and speech delay (MIM:619338) (Autosomal recessive)	severe intellectual disability-epilepsy-cataract syndrome due to peroxisomal disorder (MONDO:0014510)	FAR1 deficiency (MONDO:0014510)	Moderate 5.5 GE, 1 EE	8	SbC of FAR1 defects (MONDO:0100275) under DPB (MONDO:0017986). “Deficiency” to indicate loss-of-function mechanism
	Peroxisomal FAR1 disorder (MIM:616154) (Autosomal dominant)	N/A	FAR1 upregulation (MONDO:0100230)	Moderate 6 GE, 1 EE	8	SbC of FAR1 defects (MONDO:0100275) under DPB (MONDO:0017986). “Upregulation” to indicate gain-of-function mechanism
GNPAT	RCDP, type 2 (MIM: 222765) (Autosomal recessive)	RCDP (MONDO:0015776)	GNPAT deficiency (MONDO:0100273)	Definitive 12 GE, 3.5 EE, r/t	12	RCDP type 2 (MONDO:0009112) retained as synonym
HSD17B4	D-bifunctional protein deficiency (MIM:261515); Perrault syndrome 1 (MIM: 233400) (Autosomal recessive)	D-bifunctional protein deficiency (MONDO:0009855)	D-bifunctional protein deficiency (MONDO:0009855)	Definitive 12 GE, 2.5 EE, r/t	18	SbC of disorder of peroxisomal beta-oxidation (MONDO:0019233)
PHYH	Refsum disease (MIM:266500) (Autosomal recessive)	Adult Refsum disease (MONDO:0009958)	phytanoyl-CoA hydroxylase deficiency (MONDO:0100258)	Definitive 12 GE, 2.5 EE, r/t	18	SbC of peroxisomal alpha oxidation (MONDO:0100277)
SCP2	Leukoencephalopathy with dystonia and motor neuropathy (MIM: 613724) (Autosomal recessive)	leukoencephalopathy-dystonia-motor neuropathy syndrome (MONDO:0013391)	Sterol carrier protein 2 deficiency (MONDO:0013391)	Moderate 6 GE, 4 EE	14	SbC of peroxisomal beta oxidation (MONDO:0019233). Term replaces descriptive phenotype nomenclature
AGXT	Hyperoxaluria, primary, type 1 (MIM:259900) (Autosomal recessive)	primary hyperoxaluria type 1 (MONDO:0009823)	AGXT deficiency (MONDO:0100278)	Definitive 12 GE, 4.5 EE, r/t	9	SbC of disorder of glyoxylate metabolism (MONDO_0100278)
BAAT	Hypercholanemia, familial (MIM: 607748) (Autosomal recessive)	Familial hypercholanemia (MONDO:0011905)	BAAT deficiency (MONDO:0100305)	Moderate 5.1 GE, 2 EE	1	SbC of disorder of bile acid aminotransferase (MONDO:0100304)
CAT	Acatalasemia (MIM:614097) (Autosomal recessive)	Acatalasia (MONDO:0013571)	Acatalasia (MONDO:0013571)	Moderate 5.5 GE, 3 EE	7	SbC of disorder of defective peroxisome oxidative status (MONDO:0100306)
DNM1L	Encephalopathy, lethal, due to defective mitochondrial peroxisomal fission 1 (MIM:614388), Optic atrophy 5 (MIM:610708) (Autosomal dominant)	lethal encephalopathy due to mitochondrial and peroxisomal fission defect (MONDO:0013726)	DNM1L-associated encephalopathy due to peroxisomal and mitochondrial fission defect (MONDO:0013726)	Definitive 12 GE, 4 EE, r/t	13	SbC of disorder of defective peroxisomal and mitochondrial fission (MONDO_0100276)
MFF	Encephalopathy due to defective mitochondrial and peroxisomal fission 2 (MIM: 617086) (Autosomal recessive)	encephalopathy due to defective mitochondrial and peroxisomal fission 2 (MONDO:0014905)	MFF-associated encephalopathy due to peroxisomal and mitochondrial fission defect (MONDO:0014905)	Moderate 7 GE, 0.5 EE	3	SbC of disorder of defective peroxisomal and mitochondrial fission (MONDO_0100276)
TRIM37	Mulibrey nanism (MIM: 253250) (Autosomal recessive)	Mulibrey nanism (MONDO:0009664)	Mulibrey nanism (MONDO:0009664)	Definitive 12 GE, 2 EE, r/t	10	SbC of disorder of defective peroxisome oxidative status (MONDO:0100306)

^*ACOX1- ACOX1 upregulation classification was upgraded from Moderate to Strong based on unpublished case-level evidence.

Abbreviations: SpC: Superclass; SbC: Subclass; GE: Genetic evidence; EE: Experimental evidence; ALD: adrenoleukodystrophy; AMN: adrenomyeloneuropathy; PBD: Peroxisome biogenesis disorder; DPT: disorder of peroxisomal transporter; DPB: disorder of plasmalogens biosynthesis; RCDP: rhizomelic chondrodysplasia punctata; CBASD: Congenital bile acid synthesis defect; ACOX: Acyl-CoA Oxidase; AGPS: alkylglycerone-phosphate synthase; FAR1: Fatty Acyl-CoA Reductase 1; AGXT: alanine glyoxylate aminotransferase deficiency; BAAT: bile acid CoA:amino acid N-acyltransferase

All the PEX genes in this tier were curated in relation to PBD, which encompasses the severe Zellweger spectrum disorders and the milder subtypes including Heimler syndrome, infantile Refsum disease and rhizomelic chondrodysplasia punctata (1). Since the clinical presentation associated with PBD caused by any of the PEX genes is overlapping without a genetic test (except for PEX7 for which the clinical picture is of rhizomelic chondrodysplasia punctata), the Peroxisomal GCEP decided that it was appropriate to curate all genes in relation to the lumped disease entity of PBD (MONDO:0019234). Pathogenic variants in the ABCD1 gene result in three related phenotypes, X-linked cerebral adrenoleukodystrophy, adrenomyeloneuropathy and isolated adrenal insufficiency, which were lumped together in the curation. This curation was performed in collaboration with the Intellectual Disability/Autism GCEP. The Peroxisomal GCEP proposed the term, ABCD1 deficiency (MONDO:0018544), to include all three phenotypes.

Genetic evidence scores among the tier 1 curations ranged from 6.25 pts to 12 pts (Fig. 2), with PEX14 receiving the lowest score, with evidence from only 4 published probands. The second lowest score was for PEX11B, which received 7 points based on 5 published probands. Eight of the 14 PEX genes received a full score of 12 points for genetic evidence.

Fig 2.

Gene curation scores and classifications of 36 peroxisomal gene-disease relationships. Genetic and experimental scores are indicated by pink and blue bars, respectively. Score range for each classification - No known disease relationships (N): 0 genetic evidence points, Limited (light grey): 0.1 – 6 points, Moderate (medium grey): 7–11, Strong: 12–18 & <3 years since first report, Definitive (dark grey): 12–18 & replication over >3 years (r/t).

Experimental evidence, including in vitro functional assays and animal models that recapitulate disease presentation, was not always available to reach the maximum of 6 points in this category. Only ABCD1, PEX1 and PEX2 received all 6 points, with the availability of mouse and/or Drosophila models (11–15), in addition to in vitro assays. For all PEX genes in this tier, increased points were awarded for the combined functional role of the peroxins in peroxisome biogenesis as well as their interactions with each other. Complementation assays in patient fibroblasts, which showed restoration of peroxisome assembly with the introduction of the deficient PEX gene cDNA, were awarded the default score of 1 point in the rescue in patient cells category (16). ABCD1 was scored for rescue in humans based on the STARBEAM study, which involved the use of hematopoietic stem cells to deliver ABCD1 gene therapy to affected individuals (17).

3.2. Tier 2 Genes

The genes causing single peroxisomal enzyme/protein defects constituted the tier 2 list (Table 1). Of the 13 gene-disease relationships curated in this tier, five were classified as Definitive, six were Moderate and two were Limited (Fig. 2). GNPAT, HSD17B4 and PHYH reached a Definitive classification with 12 points each for genetic evidence. Patients with causative HSD17B4 variants lacking biochemical evidence of d-bifunctional protein deficiency but with consistent clinical symptoms have been identified; for this reason, cases both with and without biochemical features were included in the curation. AGPS reached a Definitive classification without maxing out the genetic evidence score but with sufficient experimental evidence points. The AGPS curation also included one individual with a 128-bp deletion in the AGPS cDNA; however, this variant could not be registered in the GCI as the breakpoints for causative genomic variant were unidentified (18). ACOX1 was initially curated for the loss of function phenotype, ACOX1 deficiency (MONDO:0009919), but was subsequently also curated separately in relation to the recently reported gain of function phenotype, Mitchell syndrome (19). The Peroxisomal GCEP introduced the term ACOX1 upregulation (MONDO_0030073) to indicate the gain of function phenotype, steering away from the eponymous disease name and being consistent with other mechanism-based nomenclature that the GCEP has proposed. All individuals reported with ACOX1 upregulation showed a progressive peripheral neuropathy and significant deterioration in motor functions and carried a de novo heterozygous missense variant, p.Asn237Ser, that was initially returned as a VUS in whole genome/exome studies, but determined to be causative upon reanalysis and was also supported by experimental studies (19). The curation reached a moderate classification with a total of 7.5 points; however, based on evidence of at least 10 additional unpublished probands with the same variant and similar phenotype (H. Chung, personal communication), the GCEP upgraded this classification to Strong. Since three years have not elapsed since the first case report of ACOX1 upregulation (Mitchell syndrome in paper), the curation could not reach a definitive classification.

AMACR, ACBD5 and FAR1 were classified as Moderate in relation to their protein deficiencies. AMACR deficiency (MONDO:0013681) was reported in at least 12 patients with missense variants, of which eight were observed with the recurrent p.Ser52Pro variant (20) that is shown to result in an inactive protein when expressed as a fusion to maltose-binding protein in E.coli lysates. The disease entities associated with AMACR gene defects in OMIM are congenital bile acid synthesis defect 4 and AMACR deficiency. The Peroxisomal GCEP curated AMACR for the lumped entity of AMACR deficiency, which encompasses congenital bile acid synthesis defect as well as the neurological phenotype thought to be consequent to elevated pristanic acid levels (21). ACBD5 deficiency (MONDO:0100112) has been reported in at least 3 individuals including one adult patient with retinal dystrophy and leukodystrophy (22), which increased the classification to Moderate. At the time of curation, OMIM did not note a disease associated with ACBD5; however, it currently displays retinal dystrophy and leukodystrophy as the associated disease entity. Reports of retinal dystrophy and ACBD5 deficiency were found in the literature (23, 24). Additionally, there was one report of a missense variant in ACBD5 in a family with autosomal dominant thrombocytopenia (25), but this was later disputed (26). The Peroxisomal GCEP thus decided to create a new term in Mondo and curate the gene in relation to ACBD5 deficiency. FAR1 was initially curated for the loss of function phenotype and a Moderate classification was reached with three patients reported in two publications (27, 28). The disease entity was originally reported as severe intellectual disability-epilepsy-cataract syndrome due to fatty acyl-CoA reductase 1 deficiency in Mondo, which the Peroxisomal GCEP revised as FAR1 deficiency (MONDO:0014510). As the Peroxisomal GCEP was nearing the conclusion of gene curations in April 2021, Ferdinandusse et al reported an autosomal dominant neurological disorder associated with three de novo gain of function variants in FAR1, all located at the same amino acid residue, p.Arg480, in 12 unrelated individuals (29). Consistent with other disease naming revisions proposed by the Peroxisomal GCEP, the term, FAR1 upregulation (MONDO:0014510), was introduced. Twelve de novo cases were scored, gathering a total of 6 points, and with minimal experimental evidence, a Moderate classification was reached. Loss of function variants in another tier 2 gene, SCP2, have been reported in two individuals (30–32), and at least three published mouse models recapitulate the biochemical and some clinical phenotypes seen in patients (33–35). With an overall score of 10 points, this gene-disease relationship earned a Moderate classification.

The remaining genes in tier 2, ABCD3 and EHHADH, were classified as Limited. ABCD3 was reported in relation to congenital bile acid synthesis defect 5 (MONDO:0014564) in only one publication (36). The variant was a homozygous 1785-bp deletion of the last exon and the subsequent 3’ UTR, with rescue evidence in patient cells. Earlier papers associated ABCD3 with Zellweger syndrome; however, these patients were later found to have PEX1 defects (36). Additional mouse model evidence (37) and protein interaction with ABCD1 and PEX19 (38, 39) garnered a total of 4 points for experimental evidence and a Limited classification for this gene. A heterozygous missense variant in EHHADH gene has been reported in a single family in the literature to cause Fanconi renotubular syndrome 3 (40). The mechanism of disease in this family was expected to be dominant negative and the pathology resulting from the mistargeting of the protein to the mitochondria and interfering with the function of the mitochondrial trifunctional protein. Scores awarded for segregation in the family of nine affected individuals and a proband and functional studies in renal tubular cell line accumulated a total of 3.5 points for this curation. Should additional case-level data emerge in the future, these curations could be upgraded to a Moderate classification.

3.3. Tier 3 Genes

The genes on this tier were found on one or more peroxisome gene sequencing panels (Table 1, Fig. 3). Two of the 10 genes in this tier (GDAP1 and DYM) were removed from the list as they were considered not to be within the purview of the Peroxisomal GCEP (See section 4.2).

Fig 3:

Distribution of genes on the NIH Genetic Testing Registry sequencing panels that were curated by the Peroxisomal GCEP, along with their classifications indicated by the colored background.

Among the eight curated genes, three were classified as Definitive (AGXT, DNM1L and TRIM37), three as Moderate (BAAT, CAT and MFF) and two as No known disease relationship (PEX11A and PEX11G).

Several patients reported in the literature with AGXT deficiency (primary hyperoxaluria type 1; MONDO:0100278) and mouse models with rescue evidence provided abundant case-level and experimental evidence, respectively, for the AGXT gene curation. DNM1L and MFF were both curated in relation to encephalopathy due to defective mitochondrial and peroxisomal fission (MONDO:0054865). Most variants in DNM1L reported in the literature are de novo heterozygous missense variants, which are thought to act in a dominant negative manner and interfere with protein assembly (41). A similar phenotype has been reported in a few patients with homozygous or compound heterozygous null variants, wherein a loss of function mechanism is expected. Although the inheritance pattern and molecular mechanism were different, the Peroxisomal GCEP decided not to perform split curations as the phenotypes observed in patients constituted the same disease entity. The autosomal recessive cases were noted in the curation but not scored. MFF defects have been reported in five patients, including a recent report (42), and combined with some experimental evidence, the curation reached a moderate classification. Publications reporting DNM1L and MFF variants reported mitochondrial abnormalities more often than peroxisomal abnormalities in patients (4/10 for DNM1L and 3/5 for MFF). These genes have also been independently assessed in relation to Leigh syndrome by the ClinGen Mitochondrial GCEP.

Based on abundant case-level evidence, TRIM37- autosomal recessive mulibrey nanism (MONDO:0009664) relationship reached a definitive classification. TRIM37 is shown to partially localize to the peroxisome and regulates the ubiquitination of PEX5. In human cell models, depletion of TRIM37 may result in peroxisomal matrix protein import defect (43). Mouse models of TRIM37 recapitulate the human disease; however no overt peroxisome pathology is reported (44). Also, normal peroxisomal immunofluorescence staining was observed in fibroblasts from patients with TRIM37 variants, and no metabolite abnormalities reminiscent of PBD were reported (45). The disease entity associated with BAAT in OMIM at the time of curation was familial hypercholanemia. One publication reported this association, but familial hypercholanemia is a heterogeneous entity (46) and forms just one of the phenotypes of BAAT protein deficiency. Therefore, the Peroxisomal GCEP curated this gene for the more inclusive term, BAAT deficiency (MONDO:0100305). Variants in the CAT gene result in reduced or absent catalase activity. Although patients with acatalasia (a.k.a acatalasemia) do not display a clinical phenotype, oxidative stress resulting from acatalasia may be a risk factor for many diseases including progressive oral gangrene, type II diabetes mellitus, schizophrenia, cancer, arthritis and microcytic anemia (47). The curation considered acatalasia (MONDO:0013571) as the disease entity and scored for the biochemical phenotype in patients.

Finally, PEX11A and PEX11G were curated because they are protein isoforms of PEX11B and were seen on two panels from a research laboratory in the GTR (Fig. 3). However, these two genes were not reported in humans in relation to PBD at the time of curation, and thus were classified as No known disease relationship. PEX11A received 1 point for experimental evidence from two Pex11a knock-out mouse models. While one of them reported no external phenotypes or peroxisome defects (48), the other suggested that Pex11a deficiency may lead to alterations in peroxisome abundance, function and variability in expression of genes involved in fatty acid oxidation (49, 50). Of note, the latter mouse model was characterized by steatosis, hypertension and albuminuria; however, the Peroxisomal GCEP decided to award it a minimum score of 1 point for the functional impact of Pex11a deficiency. The PEX11G curation received no points as there has been only one report of an individual with mevalonate kinase deficiency who had a homozygous PEX11G variant (p.Leu216Phe) in addition to the causative variant in MVK (51). Authors suggested that the variant might be a modifier; however, this variant is quite common in the general population and is reported at the highest MAF of 0.03814 in the non-Finnish European population with 62 homozygotes in gnomAD v2.1.1. So, this variant is unlikely to be causative of any disease phenotype.

3.4. Recuration

The ClinGen-recommended timeline for recuration of gene-disease relationships is two years since the last approval date for Moderate classifications and three years since the last approval date for Limited classifications. Strong classifications will be upgraded to Definitive if three years have elapsed since the first report and no contradictory evidence has emerged. At this time, there is no specified ClinGen-recommended timeline within which Definitive relationships and No known disease relationships have to be recurated. These will have to be evaluated on a case-by-case basis when any new information is published, such as contradictory evidence for Definitive relationships or the report of a disease association in humans for No known disease relationship classifications. In addition, to consider additional genes to curate, the Peroxisomal GCEP will review the literature every 2 years, when the moderate classifications are recurated, to evaluate if novel genes have been identified in which defects cause disorders of peroxisome function.

3.5. Mondo Revisions

Following the gene curation process, the Peroxisomal GCEP worked with curators at the Monarch Disease Ontology to revise and update the Mondo disease ontology and introduce ClinGen-preferred terminology and classifications. The rationale behind these changes is discussed in section 4.1. The Peroxisomal GCEP-updated Mondo page can be accessed at http://purl.obolibrary.org/obo/MONDO_0019053.

4. Discussion

The ClinGen Peroxisomal GCEP aimed to provide the community with quality assessments of the level of evidence supporting the role of the many peroxisomal genes in disease using the ClinGen gene-disease validity framework. Such information is crucial in determining the inclusion or exclusion of genes in molecular testing panels. The majority of the genes on our list have a moderate, strong or definitive classification (30/34) warranting the inclusion of these genes on sequencing panels testing for peroxisomal disorders.

The GTR contains about 26 panels that are labeled for ‘Peroxisome disorder’. While the PEX genes are appropriately represented in these panels, other definitive peroxisomal genes are seen in only a fraction of panels (Fig. 3). For instance, ABCD1, which is found in a number of panels for adrenoleukodystrophy and intellectual disability, is seen only on 15 out of 26 peroxisomal panels (58%). Other peroxisomal single enzymes such as ACOX1, AMACR, HSD17B4 and PHYH are included in more than 17 panels, whereas AGPS, GNPAT and FAR1 are found in 14 panels or less. Counterintuitively, some of the genes with moderate classifications were seen on fewer panels than the one gene that reached a limited classification. ACBD5 and BAAT were included only on 3 and 1 panels, respectively; whereas ABCD3 was on at least 10 panels. While the evidence supporting the role of ABCD3 in disease is limited at this time, it is likely that publication of additional patients with ABCD3-congenital bile acid synthesis defect will upgrade its classification to moderate, warranting its place on sequencing panels.

The American College of Medical Genetics (ACMG) released technical guidelines to design molecular testing panels in 2019, according to which genes with a gene-disease validity classification below moderate would be considered genes of uncertain significance (GUS), and variants in these genes should not be returned as causative for disease upon testing (52). This will help avoid reporting uncertain results to patients. In our evaluation, only ABCD3 reached a limited classification at this time; however, the availability of more case-level data might push this gene-disease relationship to the moderate classification. PEX11B and PEX11G have both not been reported in relation to PBDs, and case reports published in the future may change the classification upon recuration. Some of the genes curated only had one or a few causative variants reported. For example, the recurrent variant, p.Ser52Pro in AMACR has been reported in several individuals with AMACR deficiency, while gain of function variants in ACOX1 and FAR1 have been reported in only one codon each, Asn237 and Arg480, respectively.

The ClinGen framework uses rigorous metrics to evaluate various pieces of evidence that support or refute a gene’s involvement in disease. Applying the framework to genes causing autosomal recessive biochemical disorders is generally straightforward, but also poses a few challenges. Since most biochemical disorders can be diagnosed with measurable analytes, it provides ample evidence for a biochemical defect as well as functional data for the experimental evidence categories. However, this can also be a challenge in some curations since the biochemical abnormality may be the only phenotype that can be evaluated as patients may clinically be completely asymptomatic. For instance, acatalasia caused by variants in CAT is not characterized by any clinical symptoms in patients. Another technical challenge in curating genes for biochemical disorders is scoring compound heterozygous cases. As most biochemical disorders are inherited in an autosomal recessive manner, evidence suggesting that the two variants are in trans is essential to score cases, but this information is not always provided in publications. The Peroxisomal GCEP reached out to authors in several instances and successfully obtained parental segregation data or other methods of in trans confirmation to score additional cases, which have been noted in the relevant curations.

4.1. Revisions to the Peroxisomal Disease ontology in Mondo

In order to harmonize the peroxisomal disease terminology based on curated evidence, the Peroxisomal GCEP made extensive revisions to the disease ontology in Mondo and introduced several updated nomenclatures. The peroxisomal disease ontology within Mondo was initially organized under the classes of (a) disorder of alpha, beta, omega oxidation, (b) disorder of plasmalogen synthesis, and (c) PBD. The proposed and now adopted architecture follows disease classification based on mechanism, with (a) PBD, (b) peroxisomal single enzyme/protein defect, and (c) disorder of defective peroxisomal and mitochondrial fission.

Within the class of PBD, our recommendation is to classify the disorder by gene, for instance, ‘PBD due to PEX10 defect’ would encompass all the entities attributed to PEX10 variation in the past, all of which form a spectrum: PBD type 6A (the severe Zellweger phenotype that we termed ‘classic’ Zellweger spectrum disorders), type 6B (the intermediate phenotype that we termed ‘non-classic’ Zellweger spectrum disorders). This also addressed the confusing numbering system based on complementation groups that the different PEX genes belonged to, for instance, PEX10 belongs to complementation group 7. The rationale for these changes is three-fold: (A) these conditions fall on a spectrum with overlapping phenotypes, although the severe form is usually clinically distinguishable and thus given the term ‘classic Zellweger spectrum’; (B) the nomenclature currently used with A and B subtypes are confusing since it does not denote the severity; (C) several milder phenotypes are being further classified into distinct disease entities (such as Heimler syndrome), when in fact they are a part of the continuous clinical spectrum. Further, in order to discourage the use of historical and/or eponymous disease naming and to indicate the common etiology of global peroxisome deficiency, the Peroxisomal GCEP recommended that the terms, ‘Zellweger syndrome’, ‘neonatal adrenoleukodystrophy’, ‘infantile Refsum disease’ and ‘Heimler syndrome’ be removed from the peroxisomal disease ontology in Mondo.

The peroxisomal single enzyme/protein defect included several subclasses to incorporate the different mechanisms by which they cause disease. Our approach to naming entities under this subclass was based on the protein deficiency. For instance, the disorder associated with SCP2 variants was termed SCP2 deficiency to replace the phenotype-based nomenclature, leukoencephalopathy-dystonia-motor neuropathy syndrome. Our proposed terminology is simple, more inclusive and allows for expansion of the phenotype in the future. The Peroxisomal GCEP’s approach to disease nomenclature is in alignment with that of the International Classification of Inherited Metabolic Disorders (53).

We also introduced novel disease entity nomenclature such as ACBD5 deficiency, FAR1 upregulation (gain of function), FAR1 deficiency (loss of function), ACOX1 upregulation (gain of function) and ACOX1 deficiency (loss of function), based on curated evidence from the published literature. ABCD1, ABCD3 and ACBD5 were incorporated under the new subclass of ‘disorder of peroxisomal transporter’. The additional superclass of ‘disorder of defective peroxisomal and mitochondrial fission’ was introduced to contain the encephalopathy entities associated with DNM1L and MFF defects. These genes would overlap with the mitochondrial disease tree as well.

Lastly, we recommended the removal of ‘mevalonic aciduria’, ‘Refsum disease with increased pipecolic acidemia’ and ‘glutaric acidemia type 3’ as these disease entities did not directly involve the peroxisome.

4.2. Note on GDAP1, DYM and ACOX2

Two genes, GDAP1 and DYM, were removed from the Peroxisomal GCEP gene list when the panel designated them more appropriate for curation by another relevant GCEP. GDAP1 is a mitochondrial fission factor, also expressed in peroxisomes as a tail-anchored protein, where it is thought to function in peroxisomal fission and in regulating peroxisome morphology (54). However, there have not been any reported patients with GDAP1 defects and defects in peroxisome morphology or functions. DYM was not seen on panels specific for PBDs, but was included in combined panels for mitochondrial and peroxisomal disorders and was part of the differential diagnosis in skeletal dysplasia panels (along with PEX5 and PEX7). We precurated this gene and the only evidence found was nonspecific elevation of pipecolic acid in plasma and urine in a child with Dyggve-Melchior-Clausen syndrome; other parameters of peroxisomal function were normal (55). Thus, we decided that GDAP1 was more appropriate to curate in the Charcot-Marie-Tooth GCEP and DYM in the Skeletal Disorders GCEP.

ACOX2 encodes a peroxisomal branched-chain acyl-CoA oxidase involved in bile acid synthesis. This gene was previously curated by the ClinGen IEM GCEP and reached a moderate classification.

5. Conclusion

The ClinGen Peroxisomal GCEP assessed the clinical validity and classified 34 genes associated with peroxisomal disorders. A majority of the genes that were found frequently on sequencing panels reached a definitive classification. Genes with moderate classifications were found on fewer panels than those with a limited classification. Curation of evidence relating gene-disease pairs also led to the revision of disease nomenclature and ontology within Mondo. Our work will serve as a valuable resource to clinical, laboratory and biochemical geneticists and diagnosticians with an interest in peroxisomal disorders.