Use of whole genome sequencing to determine genetic basis of suspected mitochondrial disorders: cohort study

Abstract

Objective: To determine whether whole genome sequencing can be used to define the molecular basis of suspected mitochondrial disease.

Design: Cohort study.

Setting: National Health Service, England, including secondary and tertiary care.

Participants: 345 patients with suspected mitochondrial disorders recruited to the 100 000 Genomes Project in England between 2015 and 2018.

Intervention: Short read whole genome sequencing was performed. Nuclear variants were prioritised on the basis of gene panels chosen according to phenotypes, ClinVar pathogenic/likely pathogenic variants, and the top 10 prioritised variants from Exomiser. Mitochondrial DNA variants were called using an in-house pipeline and compared with a list of pathogenic variants. Copy number variants and short tandem repeats for 13 neurological disorders were also analysed. American College of Medical Genetics guidelines were followed for classification of variants.

Main outcome measure: Definite or probable genetic diagnosis.

Results: A definite or probable genetic diagnosis was identified in 98/319 (31%) families, with an additional 6 (2%) possible diagnoses. Fourteen of the diagnoses (4% of the 319 families) explained only part of the clinical features. A total of 95 different genes were implicated. Of 104 families given a diagnosis, 39 (38%) had a mitochondrial diagnosis and 65 (63%) had a non-mitochondrial diagnosis.

Conclusion: Whole genome sequencing is a useful diagnostic test in patients with suspected mitochondrial disorders, yielding a diagnosis in a further 31% after exclusion of common causes. Most diagnoses were non-mitochondrial disorders and included developmental disorders with intellectual disability, epileptic encephalopathies, other metabolic disorders, cardiomyopathies, and leukodystrophies. These would have been missed if a targeted approach was taken, and some have specific treatments.

Fig 1

Overview of analyses and sources of diagnoses. Variants in nuclear genes were analysed using Genomics England tiering system. All tier 1 and tier 2 variants were reviewed, and these provided 66% of diagnoses. Another 20% of diagnoses were based on feedback from Genomic Medicine Centre (GMC) laboratories, comparison with Clinvar pathogenic and likely pathogenic variants, and review of top 10 Exomiser prioritised variants. Mitochondrial DNA (mtDNA) variants were analysed separately using in-house pipeline and comparison against list of 89 pathogenic variants, yielding another 5 (5%) diagnoses. Copy number variants (CNVs) accounted for 8% of diagnoses and short tandem repeat (STR) expansions for 3%. WGS=whole genome sequencing

Fig 2

Demographics for participants recruited. Top: ethnicities recorded in participants reflected ethnicity of overall population in England. Middle: most commonly recruited family structures were trios with both parents and singletons. Bottom: participants were recruited from Genomic Medicine Centres across England

Fig 3

Human Phenotype Ontology (HPO) terms for participants recruited. Top: most commonly recorded clinical HPO terms, including delayed gross motor development, intellectual disability, and myopathy. Middle: most commonly recorded investigation results HPO terms, including decreased activity of mitochondrial complex IV, lactic acidosis, and decreased activity of mitochondrial complex I. Bottom: total number of HPO terms recorded for 345 participants according to ancestor HPO system (some HPO terms have more than one ancestor HPO system)

Fig 4

Overview of proportion of families with and without diagnosis

Fig 5

Age distribution of participants at time of enrolment and type of diagnoses made. Diagnostic yield was higher in younger participants, but diagnoses were still being made in patients enrolled in their 70s and 80s

Fig 6

Types of nuclear genetic disorder identified. Inheritance patterns in nuclear mitochondrial disorders and different types of non-mitochondrial disorders. Most families with nuclear mitochondrial disorders showed autosomal recessive inheritance. De novo dominant pathogenic variants were common in families with developmental disorders causing intellectual disability and in epileptic encephalopathies

Fig 7

Human Phenotype Ontology (HPO) terms in patients with mitochondrial diagnoses. HPO ancestor systems for HPO terms recorded in participants with definite mitochondrial diagnoses (excluding partial diagnoses). Each column represents one participant (family number in brackets). Each row represents a different HPO ancestor system, listed in order of how frequently they were affected, with nervous system at top. Numbers indicate how many of participant’s HPO terms related to HPO ancestor system (eg, nervous system or musculoskeletal system). Colours go from green through to red as number of terms related to the HPO ancestor system increases

Fig 8

Comparison of Human Phenotype Ontology (HPO) terms in patients with non-mitochondrial diagnoses. HPO ancestor systems for HPO terms recorded in participants with definite non-mitochondrial diagnoses (excluding partial diagnoses). Each column represents one participant (family number in brackets). Each row represents a different HPO ancestor system, listed in order of how frequently they were affected, with nervous system at top. Numbers indicate how many of participant’s HPO terms related to HPO ancestor system (e.g., nervous system or musculoskeletal system). Colours go from green through to red as number of terms related to the HPO ancestor system increases

Fig 9

Human Phenotype Ontology modified Nijmegen Mitochondrial Diagnostic Criteria (MDC) scores in participants with confirmed genetic diagnoses of mitochondrial and non-mitochondrial disorders and in participants with no diagnosis. Participants with “probable,” “possible,” and “partial” diagnoses were excluded from this analysis. MDC scores were higher in patients with mitochondrial diagnoses than non-mitochondrial diagnoses or no diagnosis (P<0.05)