Clinical genetics evaluation and testing of connective tissue disorders: a cross-sectional study

Abstract

Background: Heritable connective tissue disorders (HCTDs) consist of heterogeneous syndromes. The diagnosis of HCTDs is aided by genomic biotechnologies (e.g., next-generation sequencing panels) facilitating the discovery of novel variants causing disease.

Methods: Detailed clinical exam data and CLIA-approved genetic testing results from next generation sequencing of 74 genes known to play a role in HCTDs were manually reviewed and analyzed in one hundred consecutive, unrelated patients with phenotypic features indicative of a HCTD referred over a 3.5-year period (2016-2020) to a specialized academic genetics clinic. The prevalence of symptoms was evaluated in the context of genetic variants. We also determined if symptoms among different organ systems were related and performed latent class analysis to identify distinct groups of patients based on symptomatology.

Results: In the cohort of 100 consecutive, unrelated individuals there were four pathogenic, six likely pathogenic and 35 classified potentially pathogenic variants of unknown clinical significance. Patients with potentially pathogenic variants exhibited similar symptom profiles when compared to patients with pathogenic/likely pathogenic variants in the same genes. Although results did not meet a multiple testing corrected threshold, patients with connective tissue symptoms had suggestive evidence of increased odds of having skin (odds ratio 2.18, 95% confidence interval 1.12 to 4.24) and eye symptoms (odds ratio 1.89, 95% confidence interval 0.98 to 3.66) requiring further studies. The best performing latent class analysis results were identified when dividing the dataset into three distinct groups based on age, gender and presence or absence of symptoms in the skeletal, connective tissue, nervous, gastrointestinal and cardiovascular systems. These distinct classes of patients included individuals with: (1) minimal skeletal symptoms, (2) more skeletal but fewer connective tissue, nervous or gastrointestinal symptoms and (3) more nervous system symptoms.

Conclusions: We used novel approaches to characterize phenotype-genotype relationships, including pinpointing potentially pathogenic variants, and detecting unique symptom profiles in patients with features of HCTDs. This study may guide future diagnosis and disease/organ system monitoring with continued improvement and surveillance by clinicians for patients and their families.

Keywords: Ehlers–Danlos syndrome; Gene variants; Heritable connective tissue disorders; Latent class analysis; Next-generation sequencing; Phenotype-genotype relationships.

Fig. 1

Genetic and phenotypic profiles of patients presenting for heritable connective tissue disorders (HCTDs). Flow chart with details for patients who presented to the genetics clinic over a 3.5 year period. Presence of symptoms observed across seven biological systems and next-generation sequencing results for 74 genes included on commercially available connective tissue disorder testing panels were evaluated for 100 unrelated patients. Variants reported as unknown clinical significance (VUS) according to ACMG classifications were further evaluated for potential pathogenicity based on allele frequency, biological conservation, Grantham distance and damaging in silico predictions. Shown are system symptom comparisons for patients with pathogenic (in green) or likely pathogenic (in blue) variants in the same genes as those with potentially pathogenic VUSs (in gray). Reported inheritance patterns are also noted; R = autosomal or X-linked recessive, D = autosomal or X-linked dominant, R/D indicates both autosomal or X-linked recessive and dominant have been reported

Fig. 2

Detection of unique classes of symptom profiles among patients. Shown are statistics on the y-axis calculated for latent class analysis of the possibility of distinct patient groups based on four models of symptom and demographic profiles. For each model, the number of evaluated groups are on the x-axis and ranged from the undivided dataset (1 group) to seven groups. Models 1 and 2 evaluated patient classes based on symptoms in skeletal, connective tissue, nervous, gastrointestinal, and cardiovascular systems; Model 2 included age and gender. Models 3 and 4 evaluated skeletal, skin, eye, nervous, gastrointestinal, and cardiovascular system symptoms with age and gender in Model 4. The y-axis was split to visualize statistics measured on different scales. Akaike (AIC; blue circles) and Bayesian information criterion (BIC; red triangles) values ranged from 611 to 960. Pearson’s chi-square goodness of fit (Chisq; yellow squares) and likelihood ratio chi-square (Gsq; gray diamonds) values ranged from 2 to 48. The optimal result was achieved using Model 2 to group patients into three classes which reflected the lowest AIC value and maintained smaller values for other statistics