Unclassified white matter disorders: A diagnostic journey requiring close collaboration between clinical and laboratory services

Abstract

Background:

Next generation sequencing studies have revealed an ever-increasing number of causes for genetic disorders of central nervous system white matter. A substantial number of disorders are identifiable from their specific pattern of biochemical and/or imaging findings for which single gene testing may be indicated. Beyond this group, the causes of genetic white matter disorders are unclear and a broader approach to genomic testing is recommended.

Aim:

This study aimed to identify the genetic causes for a group of individuals with unclassified white matter disorders with suspected genetic aetiology and highlight the investigations required when the initial testing is non-diagnostic.

Methods:

Twenty-six individuals from 22 families with unclassified white matter disorders underwent deep phenotyping and genome sequencing performed on trio, or larger, family groups. Functional studies and transcriptomics were used to resolve variants of uncertain significance with potential clinical relevance.

Results:

Causative or candidate variants were identified in 15/22 (68.2%) families. Six of the 15 implicated genes had been previously associated with white matter disease (COL4A1, NDUFV1, SLC17A5, TUBB4A, BOLA3, DARS2). Patients with variants in the latter two presented with an atypical phenotype. The other nine genes had not been specifically associated with white matter disease at the time of diagnosis and included genes associated with monogenic syndromes, developmental disorders, and developmental and epileptic encephalopathies (STAG2, LSS, FIG4, GLS, PMPCA, SPTBN1, AGO2, SCN2A, SCN8A). Consequently, only 46% of the diagnoses would have been made via a current leukodystrophy gene panel test.

Discussion:

These results confirm the importance of broad genomic testing for patients with white matter disorders. The high diagnostic yield reflects the integration of deep phenotyping, whole genome sequencing, trio analysis, functional studies, and transcriptomic analyses.

Conclusions:

Genetic white matter disorders are genetically and phenotypically heterogeneous. Deep phenotyping together with a range of genomic technologies underpin the identification of causes of unclassified white matter disease. A molecular diagnosis is essential for prognostication, appropriate management, and accurate reproductive counseling.

1. Introduction

Genetic disorders of white matter include the leukodystrophies, which are primary disorders of white matter due to glial cell pathology, and the genetic leukoencephalopathies in which the CNS white matter is affected as a secondary phenomenon with the primary disorder being neuronal, vascular or systemic in origin.(Parikh et al., 2015; van der Knaap and Bugiani, 2017; Vanderver et al., 2020). Unclassified white matter disorders are those conditions for which a specific cause cannot be recognized by the pattern of white matter abnormalities on magnetic resonance imaging (MRI) or a pathognomonic biochemical or molecular abnormality. Owing to the genetic heterogeneity of this group, broad genetic testing is recommended early in the diagnostic work up (van der Knaap et al., 2019; Vanderver et al., 2020) rather than single gene sequencing or panel testing.

The reported diagnostic yield from genomic sequencing for individuals with presumed genetic white matter disorders ranges from 80% with exome sequencing (ES)(Vanderver et al., 2016) to 85% with genome sequencing (GS)(Helman et al., 2020; Vanderver et al., 2020). MRI pattern recognition and biochemical tests play an important role in diagnosing certain white matter disorders early but should otherwise not delay a genomic sequencing test (Vanderver et al., 2020). Clinical phenotyping is important for identifying this group of disorders, prioritising and interpreting candidate variants, and in guiding the next steps if genomic testing is not informative (Stark et al., 2017).

Collaboration between neurologists, neuroradiologists, clinical and laboratory genetic services as well as researchers contributes to the diagnostic process for individuals with white matter disease via the facilitation of timely genomic testing, phenotyping of patients with atypical or complex phenotypes, and ongoing investigation when initial genomic testing is non-diagnostic. Pathways to diagnosis for patients with unclassified disorders of white matter when initial testing is uninformative include deep phenotyping to re-prioritise variants for curation, extended bioinformatic analysis, transcriptomic analysis, functional studies, data-sharing to resolve variants of uncertain significance, and reanalysis of genomic data over time (Azzariti and Hamosh, 2020; Cummings et al., 2017; Eratne et al., 2021; Tan et al., 2020). These processes and emerging technologies, supported by international research collaboration, play a critical role in improving diagnostic rates for patients with genetic white matter disorders. A timely diagnosis is essential for accurate prognostication, diagnosis and management of comorbidities, psychological adjustment to a condition, reproductive counseling, and delivery of precision therapies.

2. Methods

2.1. Patient phenotyping

Children and adults with white matter disease of presumed genetic aetiology (living and deceased) were referred by Australian and New Zealand neurology and genetic specialists to a leukodystrophy gene discovery project at the Royal Children’s Hospital (RCH), Melbourne. Affected individuals or their parents/guardians provided informed consent and the project was approved by the RCH Human Research Ethics Committee (HREC Project Number 28097). Individuals attended RCH for clinical review where possible. The following test results were reviewed before enrollment to exclude diagnoses not requiring genomic testing: very long chain fatty acids, lysosomal enzymes, urine organic acids, chromosome microarray and, if relevant to the individual’s phenotype, cytomegalovirus PCR from a neonatal dried blood sample. Some individuals had more extensive investigation prior to referral including tissue biopsies and single gene testing. Two individuals had previous singleton exome sequencing (ES) which was uninformative. Clinical details and MRI images were reviewed for each patient by at least two neurologists (RJL, AV) and a geneticist (CAS). MRI brain and spine images were reviewed for diagnostic patterns using established pattern recognition guidelines (Schiffmann and van der Knaap, 2009; van der Knaap et al., 2019). 36 individuals were initially referred for assessment. Those with a recognizable pattern of MRI abnormalities specific for a monogenic disorder were included if the suspected disorder had been excluded with single gene or genomic testing, as were those with clinical features of a leukodystrophy or genetic leukoencephalopathy and non-specific white matter abnormalities. Ten individuals were excluded as their clinical presentation was consistent with an acquired or genetic static developmental disorder, one of whom was subsequently diagnosed with HNRNPU-related developmental and epileptic encephalopathy from ES and one tested positive for congenital cytomegalovirus from their newborn screening card. A total of 26 individuals from 22 families were enrolled in the study. A case summary for each of these families is provided in Supplementary file 1.

2.2. Genome sequencing (GS)

GS analysis was performed for the proband±affected sibling and both parents as a trio or quad analysis to identify the causative variant (s). For two individuals, only one parent was available, and the analysis was performed as a duo. GS was performed using 2 × 150 nt paired end reads on an Illumina × (Illumina Cambridge Ltd, Little Chesterford UK). Read alignment was performed using BWA-mem; variant calling of the nuclear genome was performed using GATK HaplotypeCaller v3.7, BCFtools was used to call mtDNA variants (Li et al., 2009; McKenna et al., 2010). SnpEff was used for variant annotation, and a custom script was utilized for variant filtration and prioritization (Cingolani et al., 2012). Prioritized variants were curated using the ACMG/AMP guidelines (Richards et al., 2015a,b) and variants of uncertain significance (VUS) were subclassified according to their clinical relevance into ‘uncertain significance: favor benign’ (VUS:FB) or ‘uncertain significance: favor pathogenic’ (VUS:FP) (McLaughlin et al., 2014a,b). Only VUS:FP were included as candidate variants for further study.

3. Results

3.1. Genome sequencing

Causative or candidate variants were identified in 15/22 (68.2%) families. Of the 15 genes implicated, all have a Mendelian disease association, seven associated with autosomal recessive inheritance (LSS, SLC17A5, BOLA3, DARS2, PMPCA, FIG4, NDUFV1); seven with autosomal dominant inheritance (COL4A1, TUBB4A, AGO2, SCN2A, SCN8A, GLS, SPTBN1) and one with X-linked dominant inheritance (STAG2). Six of the eight dominantly inherited variants were confirmed to have occurred de novo but, for two, only one parent was available for testing. For 11/15 families, the phenotype was within the spectrum previously associated with the gene; four of these were genes previously associated with leukodystrophy (COL4A1, NDUFV1, SLC17A5 and TUBB4A) and seven were genes associated with monogenic syndromes and developmental and epileptic encephalopathies (STAG2, LSS, SCN2A, SCN8A, AGO2, SPTBN1, GLS). For 4/15 families, the clinical presentation expanded the phenotypic spectrum associated with the gene, two of which were leukodystrophy-associated genes: BOLA3 (MIM# 614299) (Nikam et al., 2018) and DARS2 (MIM# 611105)(Scheper et al., 2007), and two were associated with other neurological disease at the time of variant identification: FIG4 (MIM# 611228)(Zhang et al., 2008) and PMPCA (MIM# 213200)(Choquet et al., 2016). A summary of the genotypes and phenotypes for the diagnosed individuals is provided in Table 1.

Table 1.

Molecular diagnoses.

Family ID	Sexa	MRI abnormalities	Additional clinical features	Gene	Genetic variant(s)	Zygosity	Variant classification (ACMG)^	OMIM phenotype for gene	Previously published
RL_0006	M (M)	Hypomyelination	Peripheral neuropathy, myopia	FIG4	NM_014845.5(FIG4) c.2459+1G > A ClinVar: VCV000246120.2	Hom	P	Charcot Marie Tooth, type 4J (CMT4J) MIM#611228	Lenk et al. (2019)
RL_0007	F	Subcortical hypomyelination and white matter atrophy	Syndromic intellectual disability	STAG2	NM_001042750.2 (STAG2):c.3097C > T (p.Arg1033*) ClinVar: VCV000996008.2	Het De novo	P	Mullegama-Klein-Martinez syndrome MIM#301022
RL_0008	M	Diffuse dysmyelination with incomplete resolution	Acute regression with recovery	BOLA3	NM_212552.3 (BOLA3):c.176G >A (p.Cys59Tyr) ClinVar: VCV001189446.1	Het Maternal	LP	Multiple mitochondrial dysfunctions syndrome 2 (MMDS2) MIM#614299	Stutterd et al. (2019)
					NM_212552.3 (BOLA3):c.136C > T (p.Arg46*) ClinVar: VCV000224514.5	Het Paternal	P
RL_0010	M	Diffuse dysmyelination with cystic degeneration and late spinal cord involvement and normal MRS.	Slowly progressive spasticity	DARS2	NM_018122.5 (DARS2):c.562C > T (p.Arg188*) ClinVar: VCV001256059.1	Het Maternal	P	Leukoencephalopathy with brain stem and spinal cord involvement and lactate elevation (LBSL) MIM#611105	Stellingwerff et al. (2021)
					NM_018122.5 (DARS2):c.1762C > G (Leu588Val) ClinVar: VCV000427120.2	Het Paternal	LP
Rl_0012	M	Multifocal dystrophic white matter with calcifications and porencephaly	Diplegia	COL4A1	NM_001845.6 (COL4A1):c.2494G > A (p.Gly832Arg) ClinVar: VCV000208663.1	Het De novo	P	Brain small vessel disease with or without ocular anomalies MIM#175780
RL_0013	F	Diffuse dysmyelination	GDD	NDUFV1	NM_007103.4 (NDUFV1):c.1156C > T (p.Arg386) ClinVar: VCV000419230.2	Hom	LP	Mitochondrial complex 1 deficiency, nuclear type 4 MIM#618225
RL_0016	F	Hypomyelination	GDD	TUBB4A	NM_006087.4 (TUBB4A):c.796T > A (p.Phe266Ile) ClinVar: VCV001256060.1	Het De novo	LP	Leukodystrophy, hypomyelinating, 6 (HLD6) MIM#612438
RL_0017	F	Multifocal, slowly progressive dysmyelination	Mild ID, hypotonia and lower limb hyporeflexia	AGO2	NM_012154.5 (AGO2) c.1070C > T (p.Thr357Met) ClinVar: VCV000995792.2	Het Unknown	LP	Lessel-Kreienkamp syndrome MIM#619149
RL_0019	F	Dysmyelination	Severe ID, quadriplegia, epilepsy, skin lesions	GLS	NM_014905.5(GLS): c.866A > T (p. Lys289Ile) ClinVar: VCV001256052.1	Het De novo	LP	Infantile cataract, skin abnormalities, glutamate excess, and impaired intellectual development MIM#618339
RL_0020	M	Hypomyelination	Severe ID, quadriplegia	PMPCA	chr9:139316429G >A NM_015160.1 (PMPCA): c.1408+1G > A ClinVar: VCV001256053.1	Het Maternal	P	Spinocerebellar ataxia, autosomal recessive 2 MIM#213200
					chr9:139311137T >G NM_015160.1 (PMPCA): c.634-266T > G ClinVar: VCV001256054.1	Het Paternal	LP
RL_0021	M	Cerebral atrophy with abnormal signal of affected white matter	Severe ID, quadriplegia, dysmorphic features	SPTBN1	NM_003128.2 (SPTBN1):c.469T > G (p.Phe157Val) ClinVar: VCV001256058.1	Het Unknown	VUS:LP	Developmental delay, impaired speech, and behavioural abnormalities MIM#619475
RL_0024	F	Diffuse T2 hyperintensity with restricted diffusion suggestive of myelin oedema	Severe GDD, epilepsy	SCN8A	NM_001330260.2 (SCN8A):c.2549G > A (p.Arg850Gln) ClinVar: VCV000135651.4	Het De novo	P	SCN8A-related developmental and epileptic encephalopathy MIM#614558
RL_0025	F	Diffuse dysmyelination, cortical atrophy	Severe GDD, epilepsy	SCN2A	NM_001040142.2 (SCN2A): c.2588C > A (p.Ser863Tyr) ClinVar: VCV001256055.1	Het De novo	P	Epileptic encephalopathy, early infantile, 11 MIM#613721
RL_0026	F (F)	Hypomyelination	Mod-severe ID, quadriplegia,	SLC17A5	NM_012434.5 (SLC17A5):c.918T > G (p.Tyr306*) ClinVar: VCV000440272.3	Het Maternal	P	Sialic acid storage disorder, infantile MIM#269920
					NM_012434.5 (SLC17A5):c.115C > T (p.Arg39Cys) ClinVar: VCV000005615.5	Het Paternal	P
RL_0029	F	Diffuse dysmyelination, cortical atrophy	Severe GDD, epilepsy, hypotrichosis	LSS	NM_002340.6(LSS): c.857A > G (p. Tyr286Cys) Clinvar: VCV001256056.1	Het Maternal	LP	Hypotrichosis 14 MIM#818275
					NM_002340.6(LSS): c.647G > A; p. (Trp216*) Clinvar: VCV001256057.1	Het Paternal	P

^{^}(Richards et al., 2015).

M = Male; F = Female; Hom = Homozygous; Het = Heterozygous; P=Pathogenic; LP = Likely pathogenic; VUS:LP = Variant of uncertain significance:Favours pathogenic(McLaughlin et al., 2014).

^aSex of proband (affected sibling).

3.2. Additional investigations

For seven of the 15 families with causative or candidate variants, the diagnosis was identified from the initial trio or quad GS analysis. For the remaining eight families, additional clinical or genomic analysis was required to either confirm the pathogenicity of a candidate variant or identify the causative or candidate variant. This included research collaboration to identify a cohort of patients with a new gene-disease association (LSS) (Besnard et al., 2019), functional studies (BOLA3) (Stutterd et al., 2019a,b), RNA sequencing (PMPCA)(Takahashi et al., 2021), literature review of recently described novel gene-disease association (GLS)(Rumping et al., 2019), data reanalysis including recently discovered genes (AGO2, SPTBN1)(Cousin et al., 2021; Lessel et al., 2020), repeated clinical assessments for deeper phenotyping (FIG4) (Lenk et al., 2019a,b), and repeated MRI brain (DARS2)(Stellingwerff et al., 2021a,b). For seven families, trio GS reanalysis has been uninformative and they remain undiagnosed.

4. Discussion

Genetic disorders of the CNS white matter are a genetically and phenotypically heterogeneous group of diseases. In this study, which applied deep phenotyping, whole genome sequencing and functional studies to the diagnosis of unclassified CNS white matter disorders, a causative or candidate variant was identified in 15/22 (68.2%) families. All causative or candidate variants were absent from the population dataset gnomAD (Lek et al., 2016) or present at a low frequency if recessive, reflecting the typically rare nature and complete penetrance of these severe disorders. Previously published diagnostic yields from genomic studies of white matter disorders have varied, depending on the phenotypic homogeneity of the patient cohorts and the extent of pre-genomic investigations. A comparable study to this, by Vanderver et al., utilized trio ES to investigate a cohort of 71 patients with unclassified white matter disease and identified clinical diagnoses in 30 (42%) (Vanderver et al., 2016). Studies that have included patients diagnosed by non-genomic tests (‘standard of care’) and genomic tests (ES or GS) provide a more accurate reflection of the current clinical diagnostic yield for white matter disorders which has ranged from 75% (Mahdieh et al., 2021) to 76.5% (Vanderver et al., 2020) and 85% (Helman et al., 2020).

4.1. Utility of genome sequencing

All but one of the causative or candidate variants were within coding regions or canonical splice sites of genes with a reported disease association and therefore potentially detectable via ES, which is a more cost-effective genomic test. However, for one patient, the causative variant in SCN8A was in a coding region but was not identified by ES prior to this study due to the low coverage of the gene. Detection of this variant by GS highlights one of the advantages of GS over ES, being the uniformity of sequence coverage, which is especially helpful for G-C rich regions (Helman et al., 2020; Posey, 2019). GS is ideal for investigating novel intronic variants, though it is a significant challenge to recognise a pathogenic non-coding variant as compared to a coding variant. In this study, a deep intronic variant in PMPCA was identified as likely pathogenic, but recognized only by an RNAseq analysis looking for a second hit in trans with a pathogenic PMPCA variant. Therefore, the GS data, in this case, was more effectively interpreted following RNAseq. A potential advantage of GS over ES is the detection of copy number variants (CNV), though, in this study, SNP microarray was performed prior to recruitment, excluding causative variants above 10 kb in size. GS used in a study of 41 patients with unclassified white matter disease attributed five of the 14 diagnoses to the superiority of GS in detecting CNVs and variants in deep intronic and technically difficult regions (Helman et al., 2020). GS also includes analysis of mitochondrial DNA, though no causative variants were identified in mtDNA in this study.

4.2. Diagnosis of classical leukodystrophies

MRI pattern recognition is a powerful clinical tool to inform appropriate investigation of white matter disease, which, in a proportion of cases, can lead to a confirmed diagnosis via a specific biochemical test or focused genetic test (Schiffmann and van der Knaap, 2009). The purpose of expert MRI review in this study was to identify patients who would benefit from broader agnostic genomic testing due to the absence of a recognizable MRI pattern. Six of the 15 diagnosed cases implicated genes known to cause leukodystrophy, though two had atypical phenotypes (discussed below). The four cases with ‘classical leukodystrophies or genetic leukoencephalopathies’ were caused by pathogenic variants in SLC17A5, TUBB4A, NDUFV1 and COL4A1. Hypomyelinating leukodystrophies, such as those associated with TUBB4A and SLC17A5, are radiologically recognizable as a group but genetically heterogeneous (Wolf, Ffrench-Constant and van der Knaap, 2021). There may be clinical and radiological clues to the molecular basis (Barkovich and Deon, 2016; Steenweg et al., 2010), though genomic testing provides the highest diagnostic yield (Vanderver et al., 2020). Likewise for white matter disorders associated with mitochondrial dysfunction, as in the case associated with NDUFV1, radiological features can distinguish this group, but genomic testing is indicated to confirm the molecular cause (Vanderver et al., 2020).

4.3. Expanded phenotypes for leukodystrophy-associated genes

Two patients had causative variants in leukodystrophy-associated genes, DARS2 (MIM# 611105)(Scheper et al., 2007) and BOLA3 (MIM# 614299)(Nikam et al., 2018), and presented with novel, atypical phenotypes, thereby expanding the phenotypic spectrum for these genes. Patient RL_0010 has bi-allelic DARS2 variants causative for leukoencephalopathy with brainstem and spinal cord involvement and elevated lactate (LBSL) but presented with cerebral white matter abnormalities without the long tract involvement that is typical for this condition. The more typical MRI features evolved over four years and were evident on repeat MRI. Subsequently, additional patients with atypical forms of LBSL have been identified and a more severe form of DARS2-associated disease has been described (Stellingwerff et al., 2021a,b). The second novel phenotype was associated with a mitochondrial leukoencephalopathy in an eight-year-old boy with compound heterozygous variants in BOLA3. RL_0008 presented with acute regression followed by near complete neurological and radiological recovery, in contrast to the expected progressive and life-limiting course associated with BOLA3-related disease. Mitochondrial functional analysis in cultured fibroblasts confirmed the pathogenicity of candidate variants in BOLA3 (Stutterd et al., 2019a,b). These cases demonstrate the need for collaborative studies of atypical phenotypes and functional studies to confirm pathogenicity of candidate variants.

4.4. White matter abnormalities in monogenic syndromes and developmental and epileptic encephalopathies

For six individuals, the cause for their white matter disease was not a primary leukodystrophy or leukoencephalopathy. These causes included genes associated with developmental and epileptic encephalopathies (SCN2A, SCN8A), an X-linked dominant cohesinopathy (STAG2), a recessive cholesterol biosynthesis disorder (LSS) and two recently described monogenic syndromes (AGO2, SPTBN1). Static white matter abnormalities are a recognized feature of several monogenic and chromosomal syndromes (Shukla et al., 2020; Vanderver et al., 2014; Vigdorovich et al., 2020). Clinically, they may be distinguishable by a static disease course and the presence of abnormal growth, dysmorphism or congenital malformation. MRI abnormalities characteristic of such developmental disorders are static, multifocal, asymmetric MRI T₂ hyperintensities and non-specific abnormalities such as enlarged perivascular spaces (Vanderver et al., 2014; Vigdorovich et al., 2020). Trio analysis provides an advantage in the diagnosis as the majority have de novo dominant cause (Tan et al., 2019). A genetic diagnosis is key to informing the prognosis and management of such conditions which may differ significantly from a progressive leukodystrophy and may avoid ongoing investigation such as repeated MRI studies. Accurate clinical phenotyping can also inform the prioritization of variants for curation which increases the diagnostic yield from genomic testing (Stark et al., 2017).

4.5. Genetic causes not previously associated with white matter disease

For three families, the candidate gene had not been previously associated with white matter disease:

4.5.1. A homozygous splice variant in FIG4

A sibling pair with peripheral neuropathy and central hypomyelination were homozygous for a canonical splice variant in FIG4. Pathogenicity of the splice site variant was demonstrated with RNA analysis and subsequently, the association of bi-allelic FIG4 variants with central hypomyelination was identified in two other families and this cohort of three families was published (Lenk et al., 2019a,b). Deep phenotyping of our siblings identified skeletal abnormalities, such as thin shafts of the tubular bones with over-tubulation and thin metacarpals, overlapping the Yunis Varón syndrome which is the consequence of two FIG4 null alleles. This observation suggested a continuum of disease between Charcot-Marie-Tooth disease type 4J (MIM#611228) and Yunis Varón syndrome (MIM#216340), and additional phenotypes on this continuum have since been reported to include Parkinsonism (Zimmermann et al., 2020). The identification of FIG4 as the cause for our patients’ phenotype was facilitated by data sharing and research collaboration. Such developments in gene-disease associations necessitate the use of dynamic gene lists, such as those provided by PanelApp, to support phenotype-driven analyses (Martin et al., 2019).

4.5.2. De novo missense variant in GLS

A de novo missense variant in GLS was identified in individual RL_0019, who had died at seven years of age with severe, progressive spastic quadriplegia, vasculitic skin rash since infancy and a heterogeneous white matter signal abnormality with diffuse atrophy on MRI. The variant NM_014905(GLS):c.866A > T (p.Lys289Ile) alters a highly conserved residue in GLS, the gene encoding glutaminase which catalyzes the conversion of glutamine into glutamate. Bi-allelic variants in GLS are associated with developmental delay, developmental and epileptic encephalopathy and elevated glutamine. However, one individual with CNS disease and vasculitic rash has been reported with a de novo heterozygous missense variant in GLS (p.Ser482Cys) (Rumping et al., 2019). This individual had normal plasma and CSF levels of glutamine and glutamate, but abnormal levels subsequently detected in urine and on magnetic resonance spectroscopic (MRS) imaging in the cortex and white matter. Functional analysis of this variant in a cell model demonstrated gain-of-function of the protein (Rumping et al., 2019). Our patient similarly had normal plasma and CSF glutamine and glutamate levels. Unfortunately, a urine sample and brain MRS imaging are not available for assessment. Given the location of this variant affecting a functional residue with multiple lines of computational evidence supporting a deleterious effect, it’s absence from controls, and de novo status, it is considered to be likely pathogenic with a suspected gain-of-function mechanism.

4.5.3. Compound heterozygous variants in PMPCA

Compound heterozygous variants in PMPCA were identified in RL_0020 who died at 17 years with progressive quadriplegia and diffuse CNS hypomyelination, cerebellar atrophy, thin corpus callosum and progressive global atrophy. PMPCA encodes one subunit of a heterodimeric protein, alpha mitochondrial processing peptidase (α-MPP), responsible for maturation of nuclear-encoded mitochondrial proteins (Geli and Glick, 1990). Bi-allelic variants cause a spinocerebellar ataxia associated with degeneration of the cerebellum with variable involvement of the brainstem and spinal cord (MIM#213200). The phenotypic spectrum has recently been expanded in association with novel mutations in two families. Two related individuals had profound global developmental delay and multisystem disease with functional studies demonstrating defective α-MPP protein (Joshi et al., 2016). A third individual had infantile-onset, severe and progressive developmental delay, cerebellar ataxia, and extrapyramidal symptoms with cerebellar atrophy and excessive brain iron accumulation in the bilateral globus pallidi and substantia nigra (Takahashi et al., 2021). For RL_0020, the initial analysis of genomic variants identified a maternally inherited heterozygous +1 splice donor variant in exon 12 of PMPCA. However, this variant was not prioritized as no paternally inherited variant was identified. This case was included in a pilot study whereby RNA sequencing was performed on RNA extracted from patient fibroblasts and analyzed to identify aberrant splicing events. The RNAseq data confirmed the exon 12 splice donor variant results in skipping of PMPCA exon 12 resulting in a loss of reading frame. Careful inspection of the RNAseq data in PMPCA resulted in the identification of a cryptic exon in intron 6 with an in-frame termination codon. The cryptic exon results from a paternally inherited variant NM_015160.1:c.634-266T>G that generates a canonical splice donor site used by the cryptic exon. Overall PMPCA expression in this patient was significantly reduced compared to a panel of similarly sequenced controls (log2 fold change: 2.15, Padj <7.8×10⁻⁸) (Wai et al., 2020).

4.6. Further investigation of undiagnosed patients

Seven families remain without a diagnosis or candidate variant following family-based GS. Emerging technologies may assist in identifying variants not tractable by GS. These include RNA sequencing to detect variants affecting splicing that may be synonymous and therefore filtered out by the GS bioinformatic pipeline (Cummings et al., 2017), long-read sequencing to achieve more accurate sequencing data and improve the detection of structural variants (Amarasinghe et al., 2020), and proteomic analysis to identify cellular protein deficiencies which may be particularly useful in diagnosing mitochondrial disorders(Helman et al., 2021). Ongoing review of unclassified phenotypes and novel genotypes, in collaboration with other research groups and the use of phenotype/genotype data sharing platforms such as Matchmaker Exchange(Azzariti and Hamosh, 2020), may contribute to gene discovery, and investigation of affected sibships provide a unique opportunity in this regard (Posey et al., 2019). A schematic of this diagnostic pathway is present in Fig. 1. Regular reanalysis of GS data is important due to the discovery of new gene-disease associations and improved bioinformatic pipelines (Tan et al., 2020). For white matter diseases with suspected genetic aetiology, ongoing investigation for a precise diagnosis is indicated as it can provide significant benefits to an individual and their family (Adang et al., 2017; Splinter et al., 2018; Stavropoulos et al., 2016; Wright et al., 2018). As noted in the clinical summaries in Supplementary file 1, the differential diagnosis in some cases includes an acquired cause. These patients were included in this study as testing for a potential genetic cause was clinically indicated. In such cases however, continuation of the genetic studies may not be indicated and carries the risk of a misdiagnosis if irrelevant variants are wrongly assigned to the cause.

Fig. 1.

Approach to diagnosis of patients with suspected genetic disorder of white matter.

5. Conclusions

Genetic disorders of CNS white matter are phenotypically and genetically heterogeneous, requiring multidisciplinary collaboration for diagnosis. Genomic testing is indicated as a first-tier diagnostic test following phenotyping, including MRI pattern recognition, to assist with variant prioritization and improve diagnostic yield. Further investigation of undiagnosed cases may include the use of advanced bioinformatics, data sharing, and functional studies. The role of emerging technologies such as RNAseq, long-read sequencing and proteomics is yet to be established. In addition to the clinical benefits of a precise diagnosis, identifying the molecular basis of white matter diseases is key to understanding their pathogenesis and identifying targets for therapeutic intervention.