Specialist multidisciplinary input maximises rare disease diagnoses from whole genome sequencing

Abstract

Diagnostic whole genome sequencing (WGS) is increasingly used in rare diseases. However, standard, semi-automated WGS analysis may overlook diagnoses in complex disorders. Here, we show that specialist multidisciplinary analysis of WGS, following an initial 'no primary findings' (NPF) report, improves diagnostic rates and alters management. We undertook WGS in 102 adults with diagnostically challenging primary mitochondrial disease phenotypes. NPF cases were reviewed by a genomic medicine team, thus enabling bespoke informatic approaches, co-ordinated phenotypic validation, and functional work. We enhanced the diagnostic rate from 16.7% to 31.4%, with management implications for all new diagnoses, and detected strong candidate disease-causing variants in a further 3.9% of patients. This approach presents a standardised model of care that supports mainstream clinicians and enhances diagnostic equity for complex disorders, thereby facilitating access to the potential benefits of genomic healthcare. This research was made possible through access to the data and findings generated by the 100,000 Genomes Project: http://www.genomicsengland.co.uk .

Fig. 1. Methodology for data analysis and themes identified in additional diagnoses.

a Methodology adopted in study. Green panels = routine analysis in 100,000 Genomes Project clinical arm with interpretation undertaken by clinical scientist; Blue panels = enhanced clinician and bioinformatician involvement. b Venn diagram representing the factors contributing to new findings (red genes are strong VUSs). MOI, mode of inheritance; SNV, single nucleotide variant.

Fig. 2. Factors contributing to additional diagnoses, part 1.

a A novel non-coding MYH2 variant, c.4188-23T>A, with elevated splicing prediction scores was detected in trans with a loss-of-function variant (c.30del). (i) MYH2 transcripts were reduced (>99%) in the muscle tissue of Patient (Pt) E compared with controls (CTRs). This loss of MYH2 was supported by reverse phenotyping undertaken by pathology; MYH2 is expressed in 2A fast fibres. (ii) Left image—immunostaining for myosin heavy chains showed marked slow fibre predominance. Right image—labelling for 2A fibre specific antibody ‘7.5.2B’ was negative suggesting complete loss of 2A fibres. Each staining was performed in two serial sections. b Increased expression of COX7B transcripts in Pt Q (c.40 + 5G>A) fibroblasts; we consider this a suspicious VUS—see Supplementary Data for further functional work. c Pedigree for Pt C and D (Pink = cardiomyopathy, Grey = myopathy, Blue = spastic gait, Yellow = sensory neuropathy). These twin females initially presented with myopathy and a paternal history of cardiomyopathy, raising the possibility of dominant disease. Pt C had normal genetic testing for congenital myopathy and myasthenic syndromes but had abnormal respiratory chain enzyme activities suggestive of mitochondrial dysfunction (reduced complex I, II, III and IV activity). Pt C developed a spastic paraparesis and white matter disease in later adulthood, while Pt D developed a sensory neuropathy. However, the siblings’ phenotypes have become more similar over time, so were re-evaluated as a recessive neuropathy leading to the diagnosis of compound heterozygous POLR3A-related disease (including a heterozygous intronic mutation). MRI brain from Pt C (right panel) demonstrated symmetric signal increase within the mid brain, superior cerebellar peduncles, and dentate nuclei (highly suggestive of a POLR3A disorder). d Reinterpretation of Pt H’s pedigree (Blue = deafness, Purple = cardiomyopathy, Orange = axonal sensory polyneuropathy) suggested there are multiple disorders in this family, and a novel variant in KCNQ4 p.(Tyr101_His102insLeuValTyr) was confirmed to segregate with the deafness phenotype. This underlines the importance of considering ‘double trouble’ especially for common genetic diseases, e.g. non-syndromic hearing loss.

Fig. 3. Factors contributing to additional diagnoses, part 2.

a Reinterpretation of this pedigree suggested multiple conditions present in Patient (Pt) O (Grey = reversible COX deficiency, Teal = dysmorphism and intellectual disability, Yellow = myalgia and proximal weakness). Both siblings have a homozygous variant in CAPN3, which can cause a late-onset muscular dystrophy, whereas only Pt O had a de novo variant in MYCN that explained her dysmorphism, microcephaly, cardiac disease, and developmental delay. We suspect a cryptic third mutation may exist in this family to account for the COX deficiency. b MRI brain in Pt G demonstrated occipitoparietal white matter changes in keeping with the recently identified phenotypic spectrum of COL4A2, now known to cause seizures and exhibit variable penetrance. c A KIF22 variant was identified in Pt R whose phenotype included midface flattening, velvety skin, and unusual hands. Skeletal survey suggested a mild version of spondyloepimetaphysial dysplasia with joint laxity (right radiograph demonstrates elongated femoral necks, left radiograph shows long and tapered fingers) and review of the history revealed recurrent joint subluxations. d Improved data filtering enabled identification of a de novo mosaic mutation in the myopathy/mtDNA maintenance gene DNM2. (i) Muscle biopsy supported this diagnosis, showing fibre size disproportion with mild overall fast fibre predominance (left = fast fibre staining, right = slow fibre staining). Each staining was performed in two serial sections. (ii) Black arrow highlights the mosaic nucleotide in the Sanger sequencing read out. e MCOLN1 variants (one coding, one intronic) were identified in two affected non-dysmorphic siblings. As shown, the male sibling developed an unusual, large ulcerated gastric tumour. Given this disorder leads to achlorhydria, elevated gastrin, and implicates the same protein targeted by H. pylori’s virulence factor, we postulate it increases risk of gastric neoplasia.

Fig. 4. Model for a specialist genomic MDT.

We suggest that an evolved specialist genomic MDT model is needed for complex cases. After initial analysis of variants (1), a genomic medicine clinician should re-evaluate the case (2), and data should be updated to address the nuances of the patient presentation (3). The clinician can then review new variants (4), and feed promising variants back to the diagnostic laboratory (5) who, where necessary, would work with a translational scientist and clinician (6a and 6b) to ensure maximum evidence is gathered to confirm the pathogenicity of variants (7), ultimately resulting in improved patient management and counselling (8).