If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaRare Diseases Functional Genomics, Kids Research, Sydney Children’s Hospital Network and Children’s Medical Research Institute, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaDepartment of Medicine, University of Melbourne, Parkville, Victoria, AustraliaDepartment of Pathology, University of Melbourne, Parkville, Victoria, AustraliaDepartment of Genomic Medicine, The Royal Melbourne Hospital, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, AustraliaPeter MacCallum Cancer Centre, Melbourne, Victoria, Australia
Genetic Health Queensland, Royal Brisbane and Women’s Hospital, Herston, Queensland, AustraliaThe University of Queensland, Herston, Queensland, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Pediatric Nephrology, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Genetic Metabolic Disorders Service, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaWestern Sydney Genetics Program, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaSpecialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Department of Clinical Genetics, Royal North Shore Hospital, St Leonards, New South Wales, AustraliaNorthern Clinical School, Royal North Shore Hospital, St Leonards, New South Wales, Australia
Department of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaAustralian Genomics Health Alliance, Parkville, Victoria, AustraliaCentre for Clinical Genetics, Sydney Children’s Hospital Randwick, Randwick, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaThe Royal Melbourne Hospital, Parkville, Victoria, AustraliaMurdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Medical Genomics, Royal Prince Alfred Hospital, Camperdown, New South Wales, AustraliaCentral Clinical School, The University of Sydney, Camperdown, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Genetics of Learning Disability Service, Hunter Genetics, Waratah, New South Wales, AustraliaThe University of Newcastle, Callaghan, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Molecular Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaMonash Genetics, Monash Health, Clayton, Victoria, AustraliaMonash Heart and Monash Children’s Hospital, Monash Health, Clayton, Victoria, AustraliaMonash Cardiovascular Research Centre, Clayton, Victoria, Australia
Department of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Genomic Medicine, Westmead Hospital, Westmead, New South Wales, Australia
NSW Health Pathology, Randwick, New South Wales, AustraliaCenter for Clinical Genetics, Sydney Children’s Hospital, Randwick, New South Wales, Australia
Department of Neurogenetics, Kolling Institute, Faculty of Medicine and Health, University of Sydney, Royal North Shore Hospital, St Leonards, New South Wales, AustraliaTranslational Genomics, Kinghorn Centre for Clinical Genomics, Garvan Institute for Medical Research, Darlinghurst, New South Wales, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaMurdoch Children’s Research Institute, Parkville, Victoria, AustraliaDepartment of Neurology, The Royal Children’s Hospital, Parkville, Victoria, Australia
Department of Clinical Genetics, Liverpool Hospital, Liverpool, New South Wales, AustraliaSchool of Women’s and Children’s Health, Faculty of Medicine and Health, University of New South Wales, Kensington, NSW, Australia
Department of Pathology, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Medical Genomics, Royal Prince Alfred Hospital, Camperdown, New South Wales, AustraliaDivision of Genomics and Epigenetics, Garvan Institute of Medical Research, Darlinghurst, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Pediatric Nephrology, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaThe TY Nelson Department of Neurology and Neurosurgery, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaThe TY Nelson Department of Neurology and Neurosurgery, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Center for Clinical Genetics, Sydney Children’s Hospital, Randwick, New South Wales, AustraliaSchool of Women’s and Children’s Health, Faculty of Medicine and Health, University of New South Wales, Kensington, NSW, Australia
Rare Diseases Functional Genomics, Kids Research, Sydney Children’s Hospital Network and Children’s Medical Research Institute, Westmead, New South Wales, Australia
Center for Clinical Genetics, Sydney Children’s Hospital, Randwick, New South Wales, AustraliaSchool of Women’s and Children’s Health, Faculty of Medicine and Health, University of New South Wales, Kensington, NSW, Australia
Center for Clinical Genetics, Sydney Children’s Hospital, Randwick, New South Wales, AustraliaSchool of Women’s and Children’s Health, Faculty of Medicine and Health, University of New South Wales, Kensington, NSW, Australia
NSW Health Pathology, Randwick, New South Wales, AustraliaCenter for Clinical Genetics, Sydney Children’s Hospital, Randwick, New South Wales, AustraliaNeuroscience Research Australia, University of New South Wales, Randwick, New South Wales, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
School of Women’s and Children’s Health, Faculty of Medicine and Health, University of New South Wales, Kensington, NSW, AustraliaNewborn Care, Royal Hospital for Women, Randwick, New South Wales, Australia
Specialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Clinical Genetics, Nepean Hospital, Kingswood, New South Wales, Australia
Department of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaSpecialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Neurogenetics, Kolling Institute, Faculty of Medicine and Health, University of Sydney, Royal North Shore Hospital, St Leonards, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaThe TY Nelson Department of Neurology and Neurosurgery, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Specialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Genomic Medicine, Westmead Hospital, Westmead, New South Wales, Australia
Department of Pathology, The Royal Melbourne Hospital, Parkville, Victoria, AustraliaDepartment of Clinical Pathology, University of Melbourne, Parkville, Victoria, Australia
Department of Medicine, University of Melbourne, Parkville, Victoria, AustraliaDepartment of Genomic Medicine, The Royal Melbourne Hospital, Parkville, Victoria, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Department of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaSpecialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaThe TY Nelson Department of Neurology and Neurosurgery, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaSpecialty of Genomic Medicine, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Cytogenetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Paediatrics, University of Melbourne, Parkville, Victoria, AustraliaVictorian Clinical Genetics Services, Murdoch Children’s Research Institute, Parkville, Victoria, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Clinical Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Department of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaDepartment of Molecular Genetics, The Children’s Hospital at Westmead, Westmead, New South Wales, Australia
Correspondence and requests for materials should be addressed to Sandra T. Cooper, Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead; Children’s Medical Research Institute; Discipline of Child and Adolescent Health, Sydney Medical School, The University of Sydney. Locked Bag 4001, Westmead, New South Wales 2145, Australia
Kids Neuroscience Centre, Kids Research, The Children’s Hospital at Westmead, Westmead, New South Wales, AustraliaDepartment of Child and Adolescent Health, Faculty of Medicine and Health, The University of Sydney, Westmead, New South Wales, AustraliaThe Children’s Medical Research Institute, Westmead, New South Wales, Australia
Genetic variants causing aberrant premessenger RNA splicing are increasingly being recognized as causal variants in genetic disorders. In this study, we devise standardized practices for polymerase chain reaction (PCR)-based RNA diagnostics using clinically accessible specimens (blood, fibroblasts, urothelia, biopsy).
Methods
A total of 74 families with diverse monogenic conditions (31% prenatal-congenital onset, 47% early childhood, and 22% teenage-adult onset) were triaged into PCR-based RNA testing, with comparative RNA sequencing for 19 cases.
Results
Informative RNA assay data were obtained for 96% of cases, enabling variant reclassification for 75% variants that can be used for genetic counseling (71%), to inform clinical care (32%) and prenatal counseling (41%). Variant-associated mis-splicing was highly reproducible for 28 cases with samples from ≥2 affected individuals or heterozygotes and 10 cases with ≥2 biospecimens. PCR amplicons encompassing another segregated heterozygous variant was vital for clinical interpretation of 22 of 79 variants to phase RNA splicing events and discern complete from partial mis-splicing.
Conclusion
RNA diagnostics enabled provision of a genetic diagnosis for 64% of recruited cases. PCR-based RNA diagnostics has capacity to analyze 81.3% of clinically significant genes, with long amplicons providing an advantage over RNA sequencing to phase RNA splicing events. The Australasian Consortium for RNA Diagnostics (SpliceACORD) provide clinically-endorsed, standardized protocols and recommendations for interpreting RNA assay data.
However, the vast majority of splicing variants outside the conserved GT-AG essential splice site will be classified as variants of unknown significance (VUS) according to the existing American College of Medical Genetics and Genomics and the Association of Molecular Pathology (ACMG-AMP) guidelines.
Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
It is often not possible to confidently predict if and how a genetic variant will disrupt splicing. The only way to know with certainty is through functional testing of the spliced mRNA to define consequences for the encoded protein, enabling ACMG-AMP–guided variant reclassification for a definitive molecular diagnosis.
Technical platforms devised to sequence DNA are imminently transferable to RNA. However, RNA is vastly more complicated than DNA. The central difference is that genomic DNA has 1 reference sequence and all sequencing reads are aligned back to this reference sequence. The challenge with RNA arises from alternative splicing, which leads to multiple reference mRNA isoforms for each gene.
The use of short-read RNA sequencing (RNA-seq) was previously investigated using muscle specimens, diagnosing 35% of 50 exome-negative families with neuromuscular disorders.
Despite these successes, short-read RNA-seq showed significant diagnostic limitations. Short reads of ≤150 nucleotides (nt) regularly do not span multiple exons to unambiguously identify which isoform is affected by any identified aberrant splicing. Furthermore, mis-spliced reads often do not match the reference genome and can be filtered out, mis-aligned, and/or present at comparatively low levels because of nonsense-mediated decay (NMD),
A further challenge arises for heterozygous variants for which normally spliced mRNA is transcribed from the allele in trans. Unless a single read contains a segregated heterozygous variant to phase splicing events (ie, discern from which allele observed mis-splicing and normal splicing is arising), it is impossible to know whether a variant induces complete or partial mis-splicing.
For many genetic disorders, tissues from affected vital organs are rarely available for RNA studies. RNA-seq using RNA from whole blood, fibroblasts, or Epstein-Barr virus transformed lymphocytes (EBV-LCLs) is increasingly being used to improve diagnostic yield.
However, many OMIM genes are expressed at too low levels in blood, EBV-LCLs, or fibroblasts for diagnostic confidence of splicing outcomes via RNA-seq.
(Supplemental Table 1) at levels our research indicates is sufficient for diagnostically informative results by RT-PCR (>0.5 transcripts per million [TPM]) (Figure 1A).
Figure 1Overview of the cohort of 74 families triaged in real time from clinical genomics into RNA diagnostics. A. Informatics analyses of RNA sequencing data from whole blood, EBV-LCLs, skin fibroblasts, or urothelial cells shows that 81.3% of OMIM genes linked to clinically relevant Mendelian disorders are expressed at TPM >0.5 levels feasible for analysis by RT-PCR. B. Pie chart depicting age of disease onset for affected individuals from 74 families subject to RNA diagnostics. C. Venn diagram summarizing the biospecimens used as a source of RNA for this study. D. Position of putative splicing variants analyzed in this study relative to the donor splice site or acceptor splice site. Red: variants shown to induce mis-splicing. Black: variants maintaining normal splicing. Variant classification of putative splicing variants (E) before and (F) after splicing studies. G, H. Summary of mis-splicing events detected and their effect on protein reading frame. I. Summary of clinical impact metrics returned from referring clinicians by survey. J. Schematic illustrating the range of theoretical mis-splicing outcomes that may arise from a splicing variant. K. Schematic design of a junctional PCR primer bridging 2 exon junctions. CSS, cryptic splice site; EBV, Epstein-Barr virus; LCLs, lymphoblastoid cell lines; PGD, preimplantation genetic diagnosis; PGS, preimplantation genetic screening; RT-PCR, reverse transcription polymerase chain reaction; TPM, transcripts per million; TSS, transcription start site.
In this study, the Australasian Consortium for RNA Diagnostics (SpliceACORD) devise and evaluate standardized practices for PCR-based RNA diagnostics using RNA from clinically accessible specimens. Our specific goals were to (1) establish diagnostic criteria for clinical recommendation of RNA testing with high diagnostic return, (2) triage families in real time to undergo RNA testing and determine reproducibility of variant-associated (mis)splicing between multiple affected individuals or heterozygotes and multiple biospecimens (blood, skin fibroblasts, urothelial cells, available biopsies), (3) devise standard operational procedures and provide evidence deemed to be of sufficient rigor by pathologists and diagnostic scientists for variant classification, (4) perform microcosting analyses, (5) develop recommendations for clinical interpretation of splicing assay data, and (6) collate the diagnostic and health impacts of RNA testing.
Materials and Methods
Patient recruitment
A total of 74 families were recruited from local area health districts across Australia and New Zealand, fulfilling the devised inclusion criteria that are described in the Results section, because devising consensus ascertainment criteria was part of the consultation process for clinical codesign of standard operating procedures.
Sample collection and RNA extraction
Whole blood was collected in a PAXgene (PreAnalytiX) blood RNA tube and RNA was isolated using the PAXgene blood RNA kit according to kit instructions. Peripheral blood mononuclear cells (PBMCs) were isolated using SepMate-15 tubes (StemCell Technologies) and Ficoll Paque Plus (GE Healthcare). Urothelial cell culture was performed as described in steps 1 to 16 of Zhou et al.
RNA from PBMCs, transformed lymphocytes, urothelial, and cultured skin-derived fibroblasts was isolated using the standard TRIzol (Invitrogen) procedure followed by the RNase-free DNase set (Qiagen) and RNeasy (Qiagen) mini kit cleanup protocols. PBMCs, transformed lymphocytes, primary fibroblasts, and urothelia were treated with dimethyl sulfoxide (Sigma-Aldrich) or 100 μg/mL cycloheximide (Sigma-Aldrich) for 6 hours before harvesting in TRIzol reagent. Detailed protocols can be found in the Supplemental Methods.
RT-PCR and Sanger sequencing
SuperScript IV (Invitrogen) first-strand synthesis system was used to make complementary DNA (cDNA) from 500 ng of RNA according to kit instructions. Recombinant Taq DNA polymerase (Invitrogen) and MasterAmp 2X PCR PreMix D (Epicentre Biotechnologies) or LongAmp Taq DNA Polymerase (New England BioLabs) were used for the PCRs. Control cDNA was obtained from healthy individuals (when available), family members, or affected individuals from cases with genetic variants in an unrelated gene. All PCR products were analyzed on a 1.2% agarose gel. Bands were manually excised from an agarose gel with a scalpel and the cDNA was purified using the GeneJET gel extraction kit (Thermo Scientific) according to the manufacturer’s instructions. Purified cDNA was subject to Sanger sequencing at the Australian Genomics Research Facility.
Replicate Libraries were prepared from 1 μg of RNA using Illumina TruSeq Stranded mRNA (using poly-A selection) or Illumina TruSeq Stranded Total RNA Kit (using RiboZero beads to deplete ribosomal RNA) (Illumina Inc) per the manufacturer’s instructions. Sequencing was performed on the NextSeq 500/550 sequencing platform using the High Output Reagent Cartridge v2, 300 cycles, 150 bp paired-end reads (Illumina Inc). Reads were aligned to GRCh37/hg19 reference genome using STAR aligner
were utilized for splicing predictions, using default thresholds (Alamut) or a Δ score threshold of 0.2 (SpliceAI).
Alternative splicing of the gene was scrutinized using in-house RNA-seq data or RNA-seq data derived from organs and tissues (fetal, child, and adult) obtained from the Genotype Tissue Expression (GTEx) project
was used to manage case ascertainment and data capture via online surveys (Supplemental Figures 1-3).
Results
Our devised inclusion criteria to ascertain variants with high clinical suspicion of causality were (1) a high likelihood of a monogenic Mendelian disorder, (2) variant allele frequency consistent with disease incidence, (3) putative splicing variant in a clinician-defined, phenotypically concordant gene, and (4) preferably the variant segregates with disease (some cases studied in parallel with segregation because of clinical urgency). A total of 74 families with diverse Mendelian conditions were recruited with 6% with prenatal, 25% with severe-congenital, 47% with early childhood, and 22% with teenage-adult onset (Figure 1B; Supplemental Table 4). RNA was derived from blood (65/74), skin fibroblasts (13/74), urothelial cells (7/74), and/or an available biopsy specimen (3/74) (Figure 1C). RT-PCR and Sanger sequencing were performed for 129 individuals (2 quads, 18 trios, 13 duos, and 41 singletons) encompassing 79 splicing variants (Figure 1D and E). Of note, 19% of the variants affect the essential GT-AG splice sites, 71% affect the extended donor (5ʹ) splice site or acceptor (3ʹ) splice-site regions, 27% were exonic variants (Figure 1D), and 2 were structural/copy number variants.
Informative PCR data establishing normal splicing or mis-splicing of the target mRNA were obtained for 96% of the cases (71/74), enabling variant reclassification for 75% of variants (58 cases) (Figure 1F and H). We were unable to study cases A005-CFTR and A103-PLP1 using blood RNA (<0.1 TPM) and these analyses are being repeated using urothelial cells (CFTR) or skin fibroblasts (PLP1). Case A158-EDN3 could not be studied using RNA from blood or urothelia. Splicing studies that confidently established no evidence for variant-induced mis-splicing enabled classification of an alternative putative causal variant in 2 additional cases (3%). Importantly, 87% of the results were reported within a clinically relevant timeframe, including rapid turnaround of 10 to 21 days for 4 neonates in intensive care who had undergone ultrarapid genomic sequencing.
Australian Genomics Health Alliance Acute Care Flagship
Lunke S.
Eggers S.
et al.
Feasibility of ultra-rapid exome sequencing in critically ill infants and children with suspected monogenic conditions in the Australian public health care system.
Of note, 71% of the diagnoses were used for genetic counseling, including additional diagnoses of 9 similarly affected family members; 41% of diagnoses were used for prenatal counseling with half of these cases with intended use for preimplantation genetic diagnosis or screening; 32% of diagnoses informed clinical care; 75% of diagnoses were clinician-reported to have a positive impact for the family; and for 24% of cases, the classification remain unchanged, having no or neutral impact (Figure 1I). For 2 cases, provision of a molecular diagnosis enabled eligibility for a clinical trial.
Importantly, an identical pattern of variant-induced mis-splicing was observed for all (28) cases with samples from at least 2 affected individuals or heterozygotes and all (10) cases with RNA studies of 2 or more biospecimens (including a manifesting tissue for 3 cases). Critical review of the available RNA-seq data (see Materials and Methods) established that there was no significant alternative splicing in the region of the gene containing the variant between the manifesting tissues and clinically accessible specimens, with case A134-CDH23 being a notable exception (Supplemental Figure 4). Although our findings provide important validation for the reproducibility of variant-associated mis-splicing between biospecimens, it is not possible to be certain of the pattern of splicing in any tissue without testing the RNA.
Standard operating procedures for informed design of PCR-based RNA diagnostics
Procedural guidelines for RT-PCR of cDNA are presented in Supplemental Figure 5. Technical aspects deemed essential by >90% of clinical variant curator (CVC) respondents (N = 18) were (1) critical scrutiny of tissue-specific or developmentally regulated alternative splicing of the target gene between manifesting tissue(s) and clinically accessible biospecimens through mining of RNA-seq data; (2) strategic design of primers to specifically interrogate all theoretically possible (mis)splicing outcomes (Figure 1J and K) and to mitigate technical caveats of PCR; (3) 2 methods confirming each splicing outcome (see Figure 2, Figure 3, Figure 4), ideally via 2 separate primer pairs or, if this is not possible, a repeat experiment with the same primer pair; (4) use of at least 2, and preferably 3, age-, sex- and tissue-matched controls; (5) gel extraction and Sanger sequencing of each PCR amplicon from the 2 methods (or experimental repeat). The following were deemed nonessential but highly desirable: (6) when possible for heterozygous splicing variants, use of a long amplicon encompassing a segregated heterozygous variant to phase splicing events from each allele to discern whether a variant induces complete or partial mis-splicing; and (7) test multiple affected individuals or heterozygotes to increase confidence in the reproducibility of variant-induced mis-splicing, although actionable results are possible with testing of a single proband.
Figure 2RT-PCR to interrogate for multiple mis-splicing events. A. Schematic of detected mis-splicing for case A022-TAZ. The NM_000116.3:c.238G>C variant (red asterisk) led to exon 2 skipping (red splice junction and arrows) and use of a cryptic donor (green splice junction and arrows). B. Left: RT-PCR of blood-derived cDNA from the proband (P: male, 7 months), his mother (M: female, 35 years), and controls (C1: male, 1 year; C2: male, 2 years; C3: female, 31 years; C4 female, 36 years). Right: cDNA derived from myocardium from the proband (P: male, 10 months old) and available myocardium controls (C5: female, 8 months; C6: female, 10 years). C. Western blot of 10 g of myocardial protein lysate shows marked reduction of encoded tafazzin protein. Detected tafazzin protein showed similar molecular weight and may represent the insertion of 12 amino acids, which leads to only a subtle increase in size. Alpha-cardiac actin, GAPDH, and Coomassie stained membrane are shown as loading controls. D. Schematic of detected mis-splicing for case A040-PIGN. The NM_176787.4:c.923-6T>G variant (red asterisk) led to exon 11 skipping (red splice junction and arrows), exon 11 and 12 skipping (green splice junction and arrows), and use of a cryptic acceptor (orange splice junction and arrow). E. RT-PCR of blood-derived cDNA from the parents (M: female, 36 years; F: male, 39 years) and controls (left) or cDNA from fibroblasts (middle) or liver (right) from the affected proband and disease controls. C1 blood cDNA: female, 35 years; C2 blood cDNA: male, 38 years; C3 fibroblast cDNA: male, 2 months; C4 fibroblast cDNA male, 8 months; C5 liver cDNA: male, 5 months; C6 liver cDNA female, 2 months. Diagnostic utility of heterozygous coding SNVs to discern complete from partial mis-splicing. F. Schematic of detected mis-splicing for case A206-FANCA. The NM_000135.2:c.1715+3_1715+13del variant (red asterisk) with NM_000135.2:c.1307A>G (green asterisk) in trans. G. FANCA mRNA studies using fibroblasts and bone marrow showed 2 abnormal splicing events in the proband (P: male, 28 years) that were absent in controls (C1: male, 28 years; C2: male, 48 years), namely exon 18 skipping (green splice junction and arrows) and use of a cryptic donor (red box and arrow). H. Sanger sequencing of cDNA with normal splicing (forward primer in exon 13 and reverse primer annealing to the exon 18/19 splice junction) shows apparent hemizygosity of the missense variant c.1307A>G in trans, establishing undetectable levels of normal splicing arising from the c.1715+3_1715+13del allele. Conversely, Sanger sequencing of smaller band corresponding to exon 18 skipping shows absence of the missense variant c.1307A>G in trans (thus confirming that detected transcripts with exon 18 skipping are arising from the c.1715+3_1715+13del allele). CD, cryptic donor; cDNA, complementary DNA; RT-PCR, reverse transcription polymerase chain reaction.
Figure 3Diagnostic utility of nonsense-mediated decay (NMD) inhibition (A-F): intron 1 variants causing abnormal initiation of transcription (G-O). Isolated peripheral blood mononuclear cells and fibroblasts were treated with CHX or DMSO before RNA extraction. A. Schematic of detected mis-splicing for case A054-UBE3A. Maternal NM_130838.2:c.1900G>C variant (red asterisk) in a male proband with Angelman syndrome (MIM#105830) led to use of a cryptic acceptor (red splice junction and arrow). B. CHX treatment increased relative abundance of the higher band (top red arrow) corresponding to use of the cryptic acceptor, inducing a frameshift. Proband (P: male, 7 years); mother (M: female, 31 years); controls (C1: male, 5 years, C2: male, 39 years). C. All detected UBE3A transcripts with normal splicing arise from the c.1900G paternal allele. D. Schematic of detected mis-splicing for case A064-GLI3. The NM_000168.5:c.473+5G>A variant (red asterisk) identified in a male proband with polydactyly (MIM#174700) led to exon 4 skipping (red splice junction and arrow), inducing a frameshift. E, F. There was little difference in the intensity of amplicons with and without CHX treatment using flanking primers in exon 2 and exon 5. Proband (P: male, 28 years); controls (C1: male, 40 years, C2: male, 48 years). G. A088-PRPH2 is a female proband with macular dystrophy (MIM#169150) associated with NM_000322.4:c.581+5G>A (red asterisk). H, I. Sanger sequencing trace files of an exon 1-2 amplicon (not shown) or exon 1-3 cDNA amplicon shows apparent hemizygosity of benign variant NM_000322.4:c.910C>G (green asterisk) in trans, whereas an exon 2 to 3 amplicon shows heterozygosity of c.910C>G variant (green asterisk). Acknowledging that PCR is not quantitative, peak height of the c.910C allele in cis with c.581+5G>A was consistently lower––suggestive either of NMD degradation of transcripts from this allele and/or inefficient transcription initiation at an ectopic start site. Proband (P: female, 64 years); controls (C1: female, 43 years, C2: female, 71 years). J. A094-NF1 is a male proband with neurofibromatosis (MIM#162200) associated with a missense variant in exon 1 NM_000267.3:c.59A>C (red asterisk). K, L. Sanger sequencing showed absence of the c.59A>C allele in amplicons spanning exons 1-8 (red asterisk) (or exons 1-3 or exons 1-4, not shown). Concordantly, NM_000267.3:c.702G>A in cis with c.59A>C was absent in an exon 1-8 amplicon (green asterisk) although it appeared heterozygous in amplicons spanning exons 2 to 8, 3 to 8, or 4 to 8 (green asterisks). Evidence therefore suggests that NF1 transcription initiates after exon 1. Proband (P: male, 34 years); controls (C1: male 31 years, C2: male 35 years). CHX, cycloheximide; DMSO, dimethyl sulfoxide.
Figure 4Utility of in silico predictive algorithms and comparative diagnostic utility of RNA sequencing (RNA-seq). A, B. Histograms showing the predictive accuracy of splicing prediction algorithms. A. Author-defined confidence thresholds (see Materials and Methods) were used to assign predictions for normal splicing or mis-splicing, or VUS if predictive scores fell outside of confidence thresholds. Green = correct prediction. Red = incorrect prediction. Yellow = VUS. Gray = could not identify the authentic splice site. B. Detection of cryptic splice sites. Gray = cryptic splice site used by the spliceosome not recognized by the algorithm. Green = cryptic splice site score higher than the resultant variant splice site. Yellow = VUS; cryptic splice-site score lower than resultant variant splice site. C. Concordance matrix of splicing prediction algorithms for donor splice-site variants (left) and acceptor splice-site variants (right). D-O. Diagnostic utility of reverse transcription polymerase chain reaction (RT-PCR) vs RNA-seq. D-F. Overview showing detected mis-splicing for case A014-SPG11. The NM_025137:c.2317-13C>G variant (red asterisk) identified in a female proband with spastic paraplegia (MIM#604360) was shown by RT-PCR to cause exon 13 skipping (red splice junction and arrow) and use of a cryptic acceptor (green splice junction and arrow). Proband (P: female 43); controls (C1: female, 36 years; C2: female, 37 years). G, H. RNA-seq (CHX untreated sample) confidently identifies exon 13 skipping and cryptic-acceptor use, as well as allele bias over the NM_025137.3:c.5392G>A missense variant (green asterisk) in trans. I, J. Overview showing detected mis-splicing for case A089-TRPM6, a male with hypomagnesaemia, seizures, and developmental delay (MIM#602014). The NM_017662.4:c.1308+7T>G variant (red asterisk), inherited in trans with NM_017662.4:c.4710G>A (green asterisk), was shown by RT-PCR to induce exon 11 skipping (red splice junction and arrow). Proband (P: male, 8 months); mother (M: female, 24 years); father (F: male, 31 years); controls (C1: male, 2 years; C2: male, 3 years). Splicing algorithms predict negligible impact of c.1308+7T>G variant. K. RNA-seq sashimi plots of the entire TRPM6 gene showing 3ʹ bias because of polyA capture and/or 5ʹ decay of transcripts with zoom-up showing absence of reads mapping to the exon skipping event. L, M. Overview showing detected mis-splicing for A031-PGAP1, a male with mental retardation (MIM#615802). RT-PCR showed the homozygous NM_024989.3:c.1221-3A>G variant (red asterisk) caused use of a cryptic acceptor (red box and arrow). N, O. RNA-seq showed low read depth for PGAP1 relative to disease controls analyzed in the same run and failed to correctly align reads corresponding to use of the cryptic acceptor with soft clipped reads (orange box) revealed to highlight the alignment error. CA, cryptic-acceptor; CHX, cycloheximide; GS, GeneSplicer; HSF, Human Splicing Finder; MES, MaxEntScan; NNS, NNSplice; SAI, SpliceAI; SSF, SpliceSiteFinder-like; VUS, variants of unknown significance.
Our rigorous RT-PCR approach required an average of 8 primer pairs per case and an average costs of A$1823 per singleton and A$2563 per trio per research diagnostic report (Supplemental Table 5), which required subsequent evaluation by a genetic pathologist. Cost of testing of A$2500 per trio was deemed acceptable by 56% of survey respondents (clinicians, pathologists, and scientists), unacceptable by 3%, and 41% were unsure and made optional comments declaring that (1) costs of $2500 are insignificant relative to other clinical expenditure and (2) unanimous agreement that despite clear clinical utility of RNA diagnostics, a funding model to support RNA testing does not yet exist within the Australian public health system.
Reproducibility of variant-induced mis-splicing in multiple affected individuals or heterozygotes and different biospecimens
A022-TAZ is a male neonate admitted to intensive care with suspected Barth syndrome (MIM#302060) who presented with cardiac and metabolic abnormalities. Genomic sequencing
Australian Genomics Health Alliance Acute Care Flagship
Lunke S.
Eggers S.
et al.
Feasibility of ultra-rapid exome sequencing in critically ill infants and children with suspected monogenic conditions in the Australian public health care system.
identified a TAZ missense variant affecting the last nucleotide of exon 2, which was classified as a VUS (ChrX[GRCh37]:g.153640551G>C; NM_000116.3:c.238G>C; p.[Gly80Arg]). RNA studies using blood-derived cDNA from the proband and his mother (segregation not available at time of this testing) established that the hemizygous TAZ c.238G>C variant induces 2 in-frame splicing defects (Figure 2A), namely (1) use of a cryptic-donor splice site in intron 2 (r.238_239ins[238+1_238+36]; p.Trp79_Gly80insArgThrArgAlaSerValLeuGlyArgGlyArgLys) and (2) exon 2 skipping (r.110_238del; p.Lys37_Gly80delinsArg). Strategic use of primers bridging splice junctions confirmed undetectable levels of normally spliced TAZ mRNA in the hemizygous proband (exon 1/2 junctional primer) (Figure 2B) and, conversely, that abnormal use of the cryptic-donor is detected only in the proband and is absent in controls (cryptic-donor/exon 3 junctional primer) (Figure 2B). Subsequent to reclassification of c.238G>C as likely pathogenic, the proband required surgical intervention for heart complications and a cardiac specimen was available. RNA studies confirmed identical mis-splicing events in the cardiac tissue with a western blot showing marked reduction of tafazzin protein (Figure 2C), enabling reclassification of c.238G>C as pathogenic. An affected younger male sibling established the mother to be germline mosaic. Case A079-LAMP2 also involved maternal germline mosaicism (Supplemental Figure 6).
Family A040-PIGN underwent termination of pregnancy because of multiple congenital anomalies (diaphragmatic defect, pulmonary hypoplasia, cardiovascular malformation, genital malformation, absent olfactory bulbs, and absent 12th ribs) suggestive of Fryn’s syndrome
(MIM#614080). Genomic testing in the affected fetus identified a homozygous variant in PIGN (Chr18[GRCh37]:g.59810585A>C; NM_176787.4:c.923-6T>G) classified as a VUS. RT-PCR using blood-derived mRNA from the proband’s heterozygous parents showed the c.923-6T>G variant-induced exon 11 skipping (r.c.923_963del; p.Glu308Glyfs∗2) or skipping of exons 11 and 12 (r.923_1023del; p.Glu308Glyfs∗9 (Figure 2D). Studies of mRNA derived from skin fibroblasts and liver specimen from the proband confirmed c.923-6T>G–induced complete mis-splicing with no detectable normal splicing of PIGN (Figure 2E).
Use of heterozygous variants to phase events and discern complete from partial mis-splicing
Heterozygous coding variants were crucial for clinical interpretation of splicing assay data for 22 of 66 variants with autosomal dominant or compound heterozygous recessive disease to distinguish complete from partial mis-splicing. For example, A206-FANCA is a male proband with acute myeloid leukemia and suspected Fanconi anemia (MIM#227650) with a paternal pathogenic missense variant (Chr16[GRCh37]:g.89857863T>C; NM_000135.2:c.1307A>G; p.[Gln436Arg]) and maternal VUS in trans (Chr16[GRCh37]:g.89846264_89846274del; NM_000135.2:c.1715+3_1715+13del) (Figure 2F). FANCA mRNA studies using fibroblasts and bone marrow showed 2 mis-splicing events in the proband that were absent in controls (Figure 2G), namely exon 18 skipping (r.1627_1715del; p.Pro543Hisfs∗26) and use of a cryptic-donor (r.1715_1716ins[1715+1_1715+258]; p.Ser572Argfs∗73). Use of a forward primer upstream of the missense c.1307A>G variant (Figure 2F and H) and a reverse primer annealing to the exon 18/19 splice junction enabled specific amplification of transcripts with normal splicing. This PCR data established that all correctly spliced FANCA transcripts (exons 13-18) bear the paternal missense variant c.1307A>G. These data infer that c.1715+3_1715+13del induces (near) complete mis-splicing of all detected FANCA transcripts with both mis-splicing outcomes encoding a premature termination codon, enabling variant reclassification as likely pathogenic.
Cycloheximide treatment recommended as a second investigation
We explored the diagnostic utility conferred by an additional RNA sample preparation step of cycloheximide (CHX) inhibition of NMD for 25 cases (Figure 3A-F, Supplemental Figure 7). Of 23 cases, 15 cases were subsequently shown to produce at least 1 NMD-compliant outcome
that is an encoded premature termination codon >50 nt upstream of the last exon-exon junction (see Supplemental Table 6, CHX sensitivity). CHX rescue of NMD-compliant events was evident with primers in flanking exons for 10 of 15 cases (eg, case A054-UBE3A in Figure 3A-C). However, CHX effects were not readily apparent for 5 of 15 cases using flanking primers that amplify multiple events, with rescue by CHX evident only when using a primer pair specific for the NMD-targeted event (eg, case A064-GLI3 in Figure 3D-F) likely because of competition inherent with multitemplate PCRs. CHX treatment strengthened evidence in several cases by showing mis-splicing was not a rare event and rather that NMD was effective.
Because cycloheximide treatment doubles the costs and time required for RNA testing, >90% of CVC respondents (n = 18) agreed that CHX treatment should be used as a second investigation for cases in which there is clear diagnostic utility for protecting an NMD-compliant mis-splicing outcome. The educational needs of genetic pathologists and diagnostic scientists were reflected by our respondents who declared that they were not aware (45%) or only somewhat aware (30%) that (1) only spliced transcripts successfully transported out of the nucleus for a pilot round of translation in the cytoplasm can activate NMD and (2) there are innate protective mechanisms to prevent mis-spliced mRNA with atypical features (eg, retained intron) from exiting the nucleus. Thus, a proportion of mis-spliced transcripts are retained in the nucleus and are incapable of activating NMD but are also unable to be translated.
Importance of strategic consideration of abnormal initiation or termination of transcription
Although transcription by RNA polymerase and pre-mRNA splicing by the spliceosome are separate processes, there is complex interplay between these 2 key events underpinning gene regulation. Variants affecting promoter regions or untranslated regions or splice-site motifs of intron 1 or a terminal intron can lead to abnormal initiation or termination of transcription.
Our study identified 4 cases with an intron 1 splice-site variant shown to induce abnormal initiation of transcription (A001-CLN5, A088-PRPH2, A094-NF1, A100-TUBA1A) and 1 case with pathogenic abnormal termination of transcription (A113-KCNH2). Because of the complexity of the mechanisms involved, deeper mechanistic investigations of cases A001-CLN5, A100-TUBA1A, and A113-KCNH2 will be submitted for publication separately. However, evidence for the activation of an alternative transcription start site for variants affecting the donor splice site of intron 1 in cases A088-PRPH2 and A094-NF1 are shown in Figure 3G-L. In both cases, PCR encompassing a distal benign single-nucleotide variation (SNV) was crucial to show loss of correctly spliced transcripts containing exon 1 (containing the start AUG). This was achieved by showing that a heterozygous coding SNV appears hemizygous (if in trans) or absent (if in cis) by Sanger sequencing of amplicons derived using an exon 1 forward primer, but heterozygous using a forward primer in exon 2, 3, 4, and so on. These data infer abnormal initiation of transcription downstream of exon 1 and upstream of exon 2. Both PRPH2 and NF1 have the start AUG encoded by exon 1; when transcription does not initiate at exon 1, it is impossible to predict the translational start site and any AUG codon within a reasonable Kozak sequence may be used,
(Figure 4A-C). We used thresholds defined by the algorithm’s authors to assign predictions of normal splicing or mis-splicing or, if predictive scores fell within an author-defined “grey zone” (outside of confidence thresholds for normal splicing or mis-splicing), we assigned VUS (Figure 4A, yellow segment). SpliceAI was the most accurate (84%) followed by NNSplice (60%). NNSplice and GeneSplicer were often unable to recognize the authentic human splice site to offer a prediction of 15% (NNSplice) or 18% (GeneSplicer) variants (Figure 4A, gray segment). A majority of this difficult cohort of splicing variants were “VUS” (yellow segment), reaching 58% VUS for Human Splicing Finder and 48% VUS for MaxEntScan. Many activated cryptic splice sites were not recognized by the prediction algorithms (Figure 4B, gray segment). Overall, there was significant discordance and inaccuracy in the predicted outcomes among algorithms, especially for donor splice-site variants (Figure 4C), highlighting the significant challenge with clinical interpretation of in silico splicing predictive tools.
Comparative evaluation of diagnostic utility of RNA-seq
RNA-seq (150 bp paired-end reads) was performed subsequently for 19 cases studied by RT-PCR. Diagnostically informative RNA-seq data were obtained for 40% (6/19) of the cases, identifying all splicing events detected by RT-PCR and detectable allele bias of coding SNVs reflecting active NMD (eg, case A014-SPG11 in Figure 4D-H). RNA-seq was nondiagnostic for 60% (12/19) of cases because of low read depth (eg, case A089-TRPM6 in Figure 4I-K) and exacerbated by NMD (eg, case A031-PGAP1 in Figure 4L-O and case A066-VPS13D in Supplemental Figure 8). Notably, 4 samples failed library preparation primarily because of low RNA concentration (RT-PCR informed diagnoses were secured before sending residual RNA for RNA-seq). Retrospective analyses indicate a read depth of approximately 5 TPM is required for diagnostically informative RNA-seq, whereas genes in this cohort with TPM values >0.5 could be reliably studied by RT-PCR (35 cycles) (Supplemental Figure 9). Extrapolating these thresholds to our entire cohort infers 42% had a read depth too low for diagnostically informative results by transcriptomic RNA-seq (50M paired-end reads).
Discussion
Our study demonstrates the significant diagnostic and health benefits of RNA diagnostics as adjunct testing to extend diagnostic yield from genomic testing. Although blood, skin fibroblasts, and EBV-LCLs are being used widely for RNA studies,
we demonstrate the unrealized diagnostic utility of urothelial cells. Urine collection is an attractive biospecimen because of ease of sampling, particularly for young children. Importantly, urothelial cells express a different repertoire of genes than skin fibroblasts or blood cells, increasing the breadth of genes able to be studied via RNA diagnostics.
Health economics analyses performed for A058-ASNS demonstrated that an early diagnosis, enabled through rapid RNA diagnostics, reduced hospitalization costs by A$117,800.
Although not yet measured formally, there are significant cost benefits of this study related to additional diagnoses of 9 similarly affected family members and facilitation of preventative medicine. Of note, 41% of diagnoses were used for prenatal counseling with half of those cases intending to use the molecular diagnosis for preimplantation genetic diagnosis or screening, reducing both the significant emotional anxiety for parents related to recurrence risk and the lifetime financial cost of health services caring for children with severe genetic disorders.
Approximately 39% of our CVC respondents (n = 18) agree on a requirement for pre-mRNA testing to be performed within an accredited laboratory, whereas 39% of respondents were satisfied with clinical actioning of splicing data from a reputable research laboratory with expertise in splicing studies (with an appropriate ethical and governance framework). The vast majority of SpliceACORD members support a hybrid model of research-pathology laboratory collaboration during this transitional period of RNA diagnostics. All CVC respondents endorsed the following consensus:
(1)
An accredited regulatory framework is favored although it does not yet exist.
(2)
PCR studies of cDNA are necessarily bespoke. Although protocols used for confirmatory Sanger sequencing of gDNA amplicons are readily adaptable to cDNA, it would be inefficient to perform multiple PCRs in this manner; however, it could be conducted as an accredited test to confirm 1 or more key (mis)splicing outcomes.
(3)
The most important factor is the expertise of the testing center in the complexities of pre-mRNA splicing and rigor of the scientific methods used.
(4)
There is an urgent need to establish quality standards for RNA diagnostics and ACMG-AMP–aligned interpretation rubrics for complex mRNA assay data (in frame, out of frame, multiple events, abnormal initiation, or termination of transcription).
Therefore, informed by study outcomes, Figure 5 details SpliceACORD consensus recommendations for ACMG-AMP–aligned interpretation of mRNA assay data for variant classification. We consider the PS3/BS3 criterion most appropriate for pre-mRNA splicing assay data using RNA isolated from patient biospecimens. We define recommendations for very strong, strong, and moderate evidence levels––to provide, when appropriate, the same level of evidence afforded by PVS1.
We emphasize that PS3 should not be used in combination with the PVS1 predicted loss-of-function criterion to avoid double counting of evidence. We recommend application of BS3 with robust evidence supporting maintenance of a normal splicing pattern (Figure 5) only when an effect on transcription or pre-mRNA splicing is the sole concerning possible effect of a variant. Protein biochemistry assays must be used to accurately determine the consequences of a synonymous or noncoding variant upon translation. Overall, complex RNA assay data do not retrofit well into existing ACMG-AMP guidelines. Adaptation of the functional evidence criterion to a points system may provide a more viable solution in the longer term to enable collective weighting of functional genomics evidence derived from patient specimens or in vitro assays (eg, RNA, protein, epigenetics).
Figure 5SpliceACORD recommendations for interpretation of RNA functional testing data aligning with ACMG/AMP evidence criteria of PS3 and BS3. PS3 (experimental evidence of mis-splicing outcomes) may not be used together with PVS1 (null variant). BS3 can be applied with robust evidence supporting maintenance of a normal splicing pattern only when an effect on transcription or pre-mRNA splicing is the sole, concerning possible effect of a variant. Protein biochemistry assays must be used to establish the potential impact of a synonymous or noncoding variant upon translation. We concur that a testing laboratory should establish the reproducibility and reliability of their RNA testing assays for at least 11 validation controls including a mix of positive and negative controls.
However, it is not possible in the RNA Diagnostics standard operating procedures we propose herein to include a mix of 11 benign and pathogenic variant controls, especially for novel variants, within a single experiment that deploys multiple PCRs to interrogate for each conceivable mis-splicing event. Therefore, our recommendation for each individual tested is comparative analysis with 2 to 4 age-, gender- and specimen-matched controls or disease controls (ie, individuals with a genetic variant in a different gene in an unrelated pathway) and wherever possible, testing of multiple affected individuals or heterozygotes to confirm reproducibility of variant-associated (mis)splicing outcomes. cDNA, complementary DNA; NMD, nonsense-mediated decay; PCR, polymerase chain reaction; pre-mRNA, premessenger RNA; RNA-seq, RNA sequencing; RT-PCR, reverse transcription polymerase chain reaction; SNV, single-nucleotide variation.
RT-PCR and RNA-seq have different strengths and weaknesses. Although high sensitivity of PCR is a diagnostic strength, a strong caveat is reliance upon the expertise of the testing center and their strategic positioning of primers––you detect only what your primers are capable of amplifying under the PCR conditions used. RNA-seq is the test of choice for cases with multiple putative splicing variants or for exome-negative cases with strong phenotypic concordance to known associated genes, if there is sufficient read depth for diagnostic confidence. Our experience with short-read RNA-seq highlights several caveats of diagnostic importance, namely (1) mis-spliced reads are regularly mis-aligned or filtered out (eg, A031-PGAP1, A058-ASNS
); (2) short reads are regularly mis-mapped between homologous gene paralogues in contig (eg, hemoglobin, tubulin, myosin heavy chain gene clusters); (3) pathogenic mis-splicing events confirmed by RT-PCR can be missed by RNA-seq because of low read depth as a result of low gene expression in the available biospecimen, NMD, 3ʹ bias because of polyA selection, or natural RNA decay of the 5ʹ end of long transcripts (A089-TRPM6; A066-VPS13D); (4) reads are regularly too short to encompass a heterozygous SNV to phase mis-splicing events and discern complete from partial mis-splicing. SpliceACORD’s recommendation therefore is to perform RT-PCR of cDNA to confirm pathogenic mis-splicing detected by RNA-seq until we have greater depth of experience to establish standard operating procedures that minimize the risk of false negatives and false positives.
In the future, clinical RNA diagnostics for Mendelian conditions is likely to require multiple technical platforms tailored to the genomic context and expression levels of the target gene in available specimens, including genome-wide transcriptomics, targeted transcriptomics of genes or gene panels, and RT-PCR for genes with low expression in available specimens. Service delivery of RNA diagnostics must consider the requirement for a biobank and ethical framework of informed consent for diagnostic use of previously tested age- and gender-matched biospecimens as controls for prospective cases. In conclusion, SpliceACORD leverages the collective expertise of approximately 80 multidisciplinary members with diverse expertise across all stages of a patient’s journey through genomic diagnosis to propose recommended triage criteria, standard operating procedures, and interpretation rubrics for PCR-based RNA diagnostics using clinically accessible specimens.
Data Availability
Access to data not provided herein may be requested via the corresponding author.
Conflict of Interest
Sandra T. Cooper is director of Frontier Genomics Pty Ltd (Australia). Sandra T. Cooper currently receives no consultancy fees or other remuneration for this role. Frontier Genomics Pty Ltd (Australia) has no existing financial relationships that will benefit from publication of these data. Samuel P. Strom is an employee and shareholder of Fulgent Genetics. Michael F. Buckley is an employee and shareholder of Invitae. The remaining coauthors declare no conflicts of interest.
Acknowledgments
We thank the families for their participation and invaluable contributions to this research. We also thank the clinicians and health care workers involved in their assessment and management. Special thanks to Naomi L. Baker, Bethany Buckley, Tenielle Clinch, Fiona Cunningham, Ryan L. Davis, Elizabeth Farnsworth, Tegan French, Janette Hayward, Katherine Holman, Cass Hoskins, Anna Jarmolowicz, McKenna Kyriss, Crystle Lee, Sarah Pantaleo, and Emma Wright who were either involved in patient care, counseling, ascertainment, diagnostic laboratory analysis, variant curation, and/or completed surveys. The data sets used for the analyses described in this manuscript were obtained from dbGaP at http://www.ncbi.nlm.nih.gov/gap through dbGaP accession number phs000424.v7.p2. We downloaded the call sets from ENCODE (https://www.encodeproject.org/) with the following identifiers: (Cerebellum) ENCFF602BYA, ENCFF113PDT and (Camera type eye) ENCFF980GGP, ENCFF883SDA.
Funding
Sandra T. Cooper is supported by a National Health and Medical Research Council of Australia Senior Research Fellowship under grant APP1136197. This project received funding through the Medical Research Future Fund Rapid Applied Research Translation Program grant awarded to Sydney Health Partners. Adam M. Bournazos is supported by a University of Sydney Research Training Scholarship. Part of this work was supported by Luminesce Alliance––Innovation for Children’s Health, a not for profit cooperative joint venture between the Sydney Children’s Hospitals Network, the Children’s Medical Research Institute, and the Children’s Cancer Institute. It has been established with the support of the New South Wales Government to coordinate and integrate pediatric research. Luminesce Alliance is also affiliated with the University of Sydney and the University of New South Wales, Sydney. Carolyn M. Sue is a National Health and Medical Research Council Practitioner Fellow under APP1136800.
Author Information
Conceptualization: A.M.B., L.G.R., S.T.C.; Data Curation and Formal Analysis: A.M.B., R.D., H.J.; Investigation: all authors; Resources: all authors; Validation: all authors; Visualization: A.M.B., L.G.R., S.T.C.; Funding Acquisition: S.T.C.; Methodology: A.M.B., L.G.R., S.B., A.N.; Project Administration: S.T.C.; Writing-original draft: A.M.B., L.G.R., S.T.C.; Writing-review and editing: all authors; Supervision: K.J.J., B.B., S.T.C.
Ethics Declaration
Consent for diagnostic genomic testing was supported by governance infrastructure of the relevant local ethics committees of the participating Australian Public Health Local Area Health Districts. Kids Neuroscience Centre’s biobanking and functional genomics human ethics protocol was approved by the Sydney Children’s Hospitals Network Human Research Ethics Committee (protocol 10/CHW/45 renewed with protocol 2019/ETH11736 [July 2019-2024]) with informed, written consent for all participants.
Australasian Consortium for RNA Diagnostics (SpliceACORD)
Ghusoon Abdulrasool, Lauren S. Akesson, Ghamdan Al Eryani, Mohammad Al-Shinnag, Peer Arts, Richard Bagnall, Naomi L. Baker, Christopher Barnett, Sarah Beecroft, Bruce Bennetts, Marina Berbic, Victoria Beshay, Michael Black, Jim Blackburn, Piers Blombery, Kirsten Boggs, Adam M. Bournazos, Susan Branford, Jimmy Breen, Natasha J. Brown, Samantha J. Bryen, Leslie Burnett, Daffodil Canson, Pak Cheong, Edward Chew, Belinda Chong, John Christodoulou, Seo-Kyung Chung, Mike Clark, Corrina Cliffe, Melissa Cole, Felicity Collins, Alison Compton, Antony Cooper, Sandra T. Cooper, Mark Corbett, Mark Cowley, Mark R. Davis, Martin Delatycki, Tracy Dudding, Matthew Edwards, Stefanie Eggers, Lisa J. Ewans, Eduardo Eyras, Fathimath Faiz, Miriam Fanjul Fernandez, Andrew Fellowes, Andrew Fennell, Michael Field, Ron Fleischer, Chiara Folland, Lucy Fox, Mary-Louise Freckmann, Clara Gaff, Melanie Galea, Roula Ghaoui, Himanshu Goel, Ilias Gornanitis, Thuong Ha, Bernadette Hanna, James Harraway, Rippei Hayashi, Ian Hayes, Alex Henderson, Luke Hesson, Erin Heyer, Michael Hildebrand, Michael Hipwell, Gladys Ho, Ari E. Horton, Cass Hoskins, Matthew F. Hunter, Matilda Jackson, Paul James, Kristi J. Jones, Justin Jong-Leong Wong, Sarah Josephi-Taylor, Himanshu Joshi, Karin Kassahn, Peter Kaub, Lucy Kevin, Edwin Kirk, Emma Krzesinski, Smitha Kumble, Sarah Kummerfeld, Nigel Laing, Chiyan Lau, Eric Lee, Sarah Leighton, Ben Lundie, Sebastian Lunke, Amali Mallawaarachchi, Chelsea Mayoh, Julie McGaughran, Alison McLean, Mary McPhillips, Cliff Meldrum, Edwina Middleton, Di Milnes, Kym Mina, David Mowat, Amy Nisselle, Emily Oates, Alicia Oshlack, Elizabeth E. Palmer, Gayathri Parasivam, Michael Parsons, Chirag Patel, Jason R. Pinner, Michael Quinn, John Rasko, Gina Ravenscroft, Anja Ravine, Krista Recsei, Matthew Regan, Jacqueline Rehn, Lisa G. Riley, Stephen Robertson, Anne Ronan, Tony Roscioli, Georgina Ryland, Simon Sadedin, Sarah A. Sandaradura, Andreas Schreiber, Hamish Scott, Rodney Scott, Christopher Semsarian, Cas Simons, Emma Singer, Janine M. Smith, Renee Smyth, Amanda Spurdle, Zornitza Stark, Patricia Sullivan, Samantha Sundercombe, Tiong Y. Tan, Michel C. Tchan, Bryony A. Thompson, David Thorburn, John Toubia, Ronald Trent, Emma Tudini, Irina Voneague, Leigh Waddell, Logan Walker, Mathew Wallis, Nick Warnock, Robert Weatheritt, Deborah White, Susan M. White, Mark G. Williams, Meredith J. Wilson, Ingrid Winship, Lisa Worgan, Dale C. Wright, Kathy Wu, Alison Yeung, Andrew Ziolowski.
Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.
Australian Genomics Health Alliance Acute Care Flagship
Lunke S.
Eggers S.
et al.
Feasibility of ultra-rapid exome sequencing in critically ill infants and children with suspected monogenic conditions in the Australian public health care system.
04128-9&id=gr1.jpg){kind=link}
04128-9&id=gr2.jpg){kind=link}
04128-9&id=gr3.jpg){kind=link}
04128-9&id=gr4.jpg){kind=link}
04128-9&id=gr5.jpg){kind=link}