Diagnostic implications of pitfalls in causal variant identification based on 4577 molecularly characterized families

Abstract

Despite large sequencing and data sharing efforts, previously characterized pathogenic variants only account for a fraction of Mendelian disease patients, which highlights the need for accurate identification and interpretation of novel variants. In a large Mendelian cohort of 4577 molecularly characterized families, numerous scenarios in which variant identification and interpretation can be challenging are encountered. We describe categories of challenges that cover the phenotype (e.g. novel allelic disorders), pedigree structure (e.g. imprinting disorders masquerading as autosomal recessive phenotypes), positional mapping (e.g. double recombination events abrogating candidate autozygous intervals), gene (e.g. novel gene-disease assertion) and variant (e.g. complex compound inheritance). Overall, we estimate a probability of 34.3% for encountering at least one of these challenges. Importantly, our data show that by only addressing non-sequencing-based challenges, around 71% increase in the diagnostic yield can be expected. Indeed, by applying these lessons to a cohort of 314 cases with negative clinical exome or genome reports, we could identify the likely causal variant in 54.5%. Our work highlights the need to have a thorough approach to undiagnosed diseases by considering a wide range of challenges rather than a narrow focus on sequencing technologies. It is hoped that by sharing this experience, the yield of undiagnosed disease programs globally can be improved.

Fig. 1. Overview of the challenges encountered based on a cohort of 4577 molecularly characterized families.

In a cohort of 4577 molecularly characterized families, we encountered 5 main scenarios we found to be challenging in 1570 families. First, there are phenotype-related challenges that can be further sub-categorized into phenotypic heterogeneity, phenotypic expansion, novel allelic disorders, blended phenotype, and misleading diagnoses. Gene-related challenges comprise the discovery of novel disease genes, novel mutation mechanisms, and cases where the animal model did not corroborate the phenotype observed in human patients. Pedigree-related challenges comprise gonadal mosaicism, allelic and genetic heterogeneity, and pseudo-dominance. Variant-related challenges comprise interpretation and technical level challenges. Pitfalls of autozygosity include the lack of detectable ROH at the disease locus and apparent sharing of the candidate ROH with an unaffected member of the family. We then dissect the prevalence of these challenges in 314 families referred to us with negative clinical exome or genome sequencing. We observed that most of the challenges encountered are either because the causal gene was novel at the time of analysis or because of non-technical variant related challenges. Logos for Gene-related, phenotype-related, and variant-related challenges are created using BioRender.com.

Fig. 2. Clinical images of select challenges.

a–g Phenotype-related challenges: a Pedigree of family F2640 with intrafamilial phenotypic heterogeneity molecularly diagnosed with Temple and Kagami-Ogata syndrome. b Chest X-ray of deceased individual IV:4 displaying “coat hanger” appearance. c Chest X-ray of individual IV:1 with the same pathogenic deletion on Chr 14 showing normal chest shape. d Phenotypic expansion: ultrasound imaging of fetus from family F7829 showing bilateral renal agenesis caused by a homozygous variant in CD151. e, f Novel allelic disorder: ultrasound imaging of fetuses from families F6581 and F6582, respectively, highlighting edema and COL25A1-related fetal akinesia. g Blended phenotype: ultrasound imaging of the right kidney of a child from family F6917 showing polycystic kidneys caused by two heterozygous variants in HNF1B and PKD1. h–j Gene-related challenges: h MRI images of a brain of an affected individual from family F4159 molecularly diagnosed with biallelic variants in SLC20A2 showing calcification. i Clinical images of an affected member of family F732 with a homozygous variant in TCOF1. j Clinical image of an affected child from family F3151 showing typical Cohen syndrome facies with the typical “grimace” upon smiling, molecularly diagnosed with a missense (rather than LOF) variant in VPS13B. k–m Variant-related challenge: k Pedigree of family F6211 with one child affected with Microcephaly, short stature, and limb abnormalities (l, m) caused by a compound heterozygous variant in DONSON. Logos for Gene-related, phenotype-related, and variant-related challenges are created using BioRender.com.

Fig. 3. A novel allelic disorder caused by biallelic LOF in ABL1.

a Pedigree of family F6666 with two similarly affected children and one child affected with Down syndrome. b Schematic representation contrasting the phenotypic differences between monoallelic gain of function and biallelic loss of function variants in ABL1. c–e Clinical photographs showcasing facial dysmorphia in affected individual IV:1. f Photograph showing short fingers in individual IV:1. g Clinical photograph of the similarly affected sister IV:3. h Sanger sequencing showing the homozygous 46 bp duplication in ABL1 in the two affected siblings, which is heterozygous in the parents and brother affected with Down syndrome. Panel (b) was created with BioRender.com.

Fig. 4. Identification of a homozygous deletion affecting the regulatory elements of DKK1 in four siblings with complex syndactyly.

a Clinical image of index individual of family F3029 showing syndactyly in feet. b X-ray images of the feet. c Photograph of the hands showing syndactyly. d X-ray of the hands. e Clinical image of the hands of a similarly affected sister with syndactyly of the middle finger. f Clinical image of the feet of the affected sister with syndactyly. g Representative illustration of DKK1 gene in mouse and human genomes. In mouse, the highlighted regions A, B, C, and D correspond to the four conserved non-coding elements (CNE25, CNE114, CNE190, and CNE195, respectively) identified downstream of DKK1 and are shown to drive its expression. In humans, peaks corresponding to H3K27Ac and DNase-seq experiments are also highlighted in the deleted region identified in the affected members of this family. h Schematic representation of WT and Dkk1 knockout mice showing syndactyly in hands and feet. i RT-qPCR experiment measuring the transcript levels of DKK1 in the index individual and his affected sister compared to the control. Data show 60–80% reduction in DKK1 transcript levels. Data are presented as mean values +/− standard deviation (SD). Error bars represent the SD of three experiments. Panels (h) and (g) are created with BioRender.com.

Fig. 5. Gonadal mosaicism and intrafamilial genetic heterogeneity are examples of pedigree-related challenges.

a Pedigree of family F4923 with 3 half siblings affected with Split hand/foot malformation syndrome. b Clinical image of the three affected siblings highlighting the hand malformations. c Screenshot of Bionano analysis output showing the identified heterozygous duplication on Chr10 previously associated with split hand/foot malformation syndrome. d Overview of structural variants identified in Chr10 by optical  genome mapping. e Table summarizing the identified disease-causing lesion in the family. f Pedigree of family F5162 with three siblings affected with variable degrees of intellectual disability, abnormal facial shape, and inability to completely open their jaws. Molecular analysis revealed that two siblings were homozygous for a LOF variant in THUMPD1 while the younger affected sibling had a de novo variant in HIST1H1E. g, h clinical images of affected individual IV:7 highlighting his inability to completely open his mouth. i, j Facial images of affected individual IV:8 demonstrating his incapacity to open his jaw.