This is a prospective observational study of a group of patients consecutively admitted to intensive care unit, University Hospital Virgen del Rocío Seville, with a diagnosis of AAS from October 2017 to January 2020. The study was authorized by the Andalusian Biomedical Research Ethics Platform (PEIBA) (GEN-AOR Internal Code 2436-N-20, Clinical trials [NCT04751058]). Informed consent for the inclusion of the genetic study and for research purposes was collected from all patients or legalguardians when appropriate.

The clinical manifestations considered to be suggestive signs of AAS during initial patient evaluation were the following: signs of sudden, severe chest pain with typical irradiation, hypertension, pulse deficit, difference in blood pressure readings between arms, semiology of aortic insufficiency, signs of ventricular failure, nonspecific electrocardiogram (normal or with left ventricular hypertrophy), normal or with mediastinal widening radiograph and other physical findings that could alert the clinician to the presence of an underlying Marfan syndrome using the Ghent scale. Confirmation of AAS was performed using multislice computed tomography (CT) angiography (angio-CT) (1mm slices), with and without iodinated contrast, for subsequent reconstruction. After confirmation and image analysis, patients were treated according to the recommendations of the European Society of Cardiology Guidelines15 for the affected segment and angiographic conditions.

After clinical evaluation and imaging tests, patients meeting the following criteria were enrolled in the genetic study: a) AAS confirmed by angio-CT; b) aged >16 years and c) absence of signs consistent with syndromic conditions. On the other hand, lifeless entry patients were excluded. Once the patient cohort was selected, the electronic health record was used to collect information on their personal and family history, clinical presentation, and disease course. In this light, the family history was only considered positive if other relatives with aortopathies and/or sudden death were found.

A family segregation analysis was recommended for all cases in which a genetic variant was detected, including likely pathogenic and variants of uncertain significance. After genetic counseling, family members that agreed to undergo clinical examination after a positive or uncertain result in the genetic test were evaluated by echocardiography. Angio-CT was applied during the clinical assessment only after an altered echocardiogram. All positive relatives were referred to the departments of comprehensive care for familial heart diseases and rare diseases of our center.

Automatic genomic deoxyribonucleic acid (DNA) extraction was performed using a Chemagic 360 instrument (Perkin Elmer, United States). DNA integrity and purity were checked by gel electrophoresis and using both fluorometric and spectrophotometric methods. A detailed description of the complete process can be found in the methods of the supplementary data.

A capture panel (Roche, USA) including 42 genes (table 1 of the supplementary data) related to different aortic diseases was used. The selection of captured genes was based on current guidelines, public databases, and expert consensus. Further details are shown in the methods of the supplementary data.

One microgram of genomic DNA was used for each library preparation using the SeqCap EZ Library SR version 5.1 (Roche, USA) according to the manufacturer's protocol with minor modifications (methods of the supplementary data). Sequencing was performed in the Illumina NextSeq500 platform (Illumina, USA).

Data analysis was performed using a previously validated pipeline16 with some modifications (methods of the supplementary data).

Variants were prioritized based on established clinical and genetic criteria (methods of the supplementary data). Finally, only variants classified as pathogenic, likely pathogenic or variants of unknown significance were reported.

Continuous variables are expressed as mean±standard deviation, while categorical variables are expressed as absolute numbers and percentages. All cases were classified in 2 groups: a) individuals with clearly pathogenic or likely pathogenic variants, and b) individuals with no variants or with clinically uncertain variants. Differences between both groups for categorical variables were estimated by Fisheŕs exact test, except for contingency tables larger than 2x2, for which the Fisher-Freeman-Halton Exact test was used. The obtained results were considered statistically significant for 2-sided P-values <.05. All statistical analyses were performed using the IBM SPSS Statistics for Windows, Version 25.0 (IBM Corp., USA).

To evaluate the diagnostic accuracy of significant clinical variables and their combination to be used as heritability-prediction tools, the following attributes were calculated17,18: true positive rate (or sensitivity), true negative rate (or specificity), false positive rate, false negative rate, and the odds ratio (OR), representing the predicted probability that a case with a given clinical variable is genetically positive against the probability that a case without that clinical variable has a positive genetic test. An OR greater than 1 implies a fold-increased odd that a positive genetic case manifests a given clinical variable vs it does not. Of note, cases classified as uncertain were not included in these calculations.

From October 2017 to January 2020, 83 AAS patients admitted to the intensive care unit met the inclusion criteria for the genetic study. However, 10 of them died in the first hours without the possibility of obtaining DNA samples, leaving a total of 73 patients for the study. Table 1 shows the general clinical characteristics of this group of patients, while detailed information for each of the cases is shown in table 2 of the supplementary data. The median age was 57 years with an approximate male/female ratio of 3:1. Based on Standford classification, 63% of patients were Stanford A and 37%, Stanford B cases. Dissection (74%) was the most frequent clinical form. Hypertension and smoking were the most common risk factors, being even more represented in patients with aortic dissection type B (70.3% and 81.4%, respectively). In 57 patients, the first symptom was pain, accompanied by a hypertensive crisis in 27 patients. Sixty patients (82.2%) underwent surgical treatment, 46.2% on an emergency and/or urgent basis. Of note, 47.9% of the patients had an aortic diameter less than 5cm.

Our NGS application allowed us to cover 97.1% of target bases, resulting in a mean coverage of 357.9x and a percentage of reads mapped on target of 83.4%. All coding regions corresponding to genes associated with aortic diseases (table 1 of the supplementary data), were covered> 20x. NGS data analysis led to the detection of 34 heterozygous candidate genetic variants in a total of 32 patients (table 2 and figure 1), including 24 missense, 4 splicing, 3 frameshift, 2 nonframeshift deletions, and 1 copy number variation (figure 2A).

Figure 1.

Central illustration. Flowchart of patients, family screening, and genetic testing outcomes. Orange boxes reflect the principal steps performed in our workflow, while the main outcomes are shown in red boxes. The ongoing actions derived as a consequence of this study are indicated in green boxes. AAS, acute aortic syndrome; Angio-CT, multislice computed tomography angiography.

(0.27MB).

Figure 2.

Classification of the 34 identified variants associated with aortic disease in our cohort. A: type of variants according to consequence type. B: distribution of variant types, including missense, frameshift, nonframeshift, splicing and gross duplication in genes associated with aortic disease in our cohort.

(0.19MB).

These variants were found in 14 different genes, with MYH11 being the most prevalent mutated gene followed by MYLK and FBN1 (figure 2B). Although these genes are involved in a wide spectrum of molecular pathways (table 1 of the supplementary data), most of them (22 cases, 69%) were associated with aortic dissection in our cohort (figure 3A). In all cases, our results were compatible with an autosomal dominant inheritance pattern except for family 10, for which compound heterozygosis of 2 changes in GAA was considered the most likely disease cause.

Figure 3.

Genotype-phenotype correlations. A: representation of the type of aortic disease associated with the genes identified in this study. B: number of cases harboring variants in genes associated with syndromic, nonsyndromic or to both forms of aortic disease. The asterisk denotes a variant in the TGFBR1 gene present in 1 of the cases harboring a concomitant variant in TGFBR2.

(0.25MB).

Approximately, 40.7% of patients (n=13) harbored genetic variants in genes previously associated with familial, nonsyndromic forms, 21.8% (n=7) with syndromic conditions, and 37.5% (n=12) carried variants in genes associated with both syndromic and nonsyndromic forms (figure 3B). Interestingly, 14 a priori nonsyndromic cases harbored pathogenic or likely pathogenic variants in genes previously associated with a potential syndromic aortic disease. Consequently, patients 10, 53, 64, and 69 were diagnosed with Pompe disease, arterial tortuosity syndrome, neurofibromatosis, and Marfan syndrome, respectively, and patients 28 and 65 were diagnosed with Ehler-Danlos syndrome. This allowed us to reclassify 18.7% of cases to a syndromic aortopathy, which has important implications for patient and family management.

Ten of the 34 identified variants had previously been reported as pathogenic or likely pathogenic in public databases and/or the literature, whereas 8 were previously described as variants of unknown significance, and 16 were novel without any clinical association so far (table 2 and figure 4). A re-evaluation of variants was performed according to the previous clinical associations, the allele count in gnomAD, the in silico pathogenicity predictions and the American College of Medical Genetics and Genomics criteria, resulting in the reclassification of 19 variants (figure 4). This resulted in the identification of 11 variants of unknown significance in 11 cases (15.07%) and 23 pathogenic or likely pathogenic variants in 21 patients, although only 20 cases were considered solved (27.39%), since a unique heterozygous variant in the autosomal recessive gene GAA was found in patient 49. If a clinical association is confirmed for these variants, the diagnostic rate of this study could reach up to 42.46%.

Figure 4.

Workflow of variant re-evaluation, classification, and clinical interpretation. ACMG, variant classification based on American College of Medical Genetics and Genomics criteria; B, benign; LB, likely benign; LP, likely pathogenic; P, pathogenic; VUS, variant of unknown significance.

(0.25MB).

Family members belonging to 9 families with detected candidate variants were willing to be studied (table 3 and figure 1). We confirmed the presence of the detected variants in 13 out of 31 relatives analyzed by Sanger sequencing. Three (I:4, I:5 and I:17) of the 13 relatives were carriers of genetic variants associated with an autosomal recessive trait and hence their genotypes were not compatible with a AAS phenotype. The remaining 10 individuals harbored genetic variants compatible with an autosomal dominant inheritance and therefore their genotypes were consistent with the occurrence of AAS.

A clinical reassessment was offered to the 10 relatives with a positive result, but only 8 have attended to date. Among them, 4 patients already showed cardiovascular alterations: a) individual I:16 (16 years) showed a resting cardiac electrical activity with QTc in upper limits (464ms); b) individual I:20 (67 years) showed mild dilatation of the root (37mm) and ascending aorta (40mm), and a grade II aortic insufficiency; c) individual I:21 (63 years) showed mild dilatation of the root (43mm) and ascending aorta (38mm), and a mild diffuse calcified atheromatosis; and d) individual I:31 (15 years) showed a mild dilatation of pulmonary artery. Remarkably, with the exception of individual I:7, all relatives with positive genetic results and no current signs of aortopathies were younger than the index patient of their family. Indeed, 11 out of 13 relatives with a finding of familial variants are currently of reproductive age (< 50 years).

Statistical tests showed significant differences between patients with a clear genetic diagnosis and those with no causative or uncertain variants, for the following clinical variables: hypertension, patient age, family history, and pain (table 1 and figure 5). The absence of hypertension, an age <55 years old, having a family history of AAS and without pain as a first symptom, significantly correlated with the detection of pathogenic or likely pathogenic genetic variants and, therefore, with hereditary cases.

Figure 5.

Association between hereditary acute aortic syndrome cases and significant clinical features. Each circle corresponds to 1 of the families under study, labeled with its identification number. The color of the circles indicates whether pathogenic or likely pathogenic variants (red), variants of unknown significance (blue), or no candidate variants (grey), were found.

(0.25MB).

All solved cases were included in at least 1 of these 4 statistically significant groups whereas the combination of the 4 significant variables encompassed only 3 of these cases (figure 5). Indeed, table 4 shows that, while sensitivity was highest for age and hypertension (a true positive rate of 70% and 80%, respectively), family history and pain were the most specific (a true negative rate of 90% and 85%, respectively). The combination of these clinical variables allowed us to obtain a balance between sensitivity and specificity, as well as to improve their predictive ability according to the calculation of OR. If we analyze the highest values of this attribute, the combination of age and pain (OR=21.54) and the combination of hypertension and family history (OR=13.33) would be the sets with a greatest predictive strength, although their sensitivities were decreased (35% and 25%, respectively). However, the joint use of these 2 combinations allowed us to maintain a high predictive value (OR=15.96) without excessively reducing sensitivity (45%) and ensuring specificity (95%).