Optimising the mutation screening strategy in Marfan syndrome and identifying genotypes with more severe aortic involvement

Marfan syndrome (MFS) is a systemic connective tissue disorder with a prevalence of 1:3000–1:5000 [1]. The main clinical manifestations typically involve the cardiovascular (CV), musculoskeletal and ocular systems [2].

The disease is inherited in an autosomal dominant manner and caused by mutations of the FBN1 gene, which is located on the long arm of chromosome 15 (15q21.1) and consists of 65 coding exons [3]. It encodes the fibrillin-1 protein, which is secreted into the extracellular matrix and cooperates with elastin to build up the connective tissue through the formation of elastic fibres. It also has structural roles even independently from elastin, for example building up ciliary zonules in the eye [4]. Its regulatory function is to keep transforming growth factor-β (TGF-β) in an inactive form [5]. A particularly important amino acid in the structure of fibrillin-1 is cysteine, of which more than 360 can be found in the protein. Cysteine plays a critical role in the stability of fibrillin-1 due to the formation of disulphide bridges [6]. Mutations that eliminate this amino acid have been proved to result in more severe CV involvement than the ones introducing new cysteine [7, 8], emphasising the particular role of this disulphide-bonding amino acid.

To date, more than 3000 genetic variants of the FBN1 gene have been reported in the Human Gene Mutation Database [9]. They spread throughout the gene, affecting all exons [3]. Around half of them are missense mutations, the others are nonsense, splice-site mutations, and small in-frame or, frameshift insertions and deletions (indels, ≤ 50 bp) or copy number variations (CNVs, > 50 bp) [9, 10]. CNVs are deletions and duplications affecting more than 50 bp and they account for around 10% of disease-causing genetic variants in Mendelian diseases [11]. FBN1 mutations can be classified into haploinsufficient (HI) and dominant negative (DN) groups based on their effect on the encoded protein. HI mutations result in the reduction of protein quantity, therefore in this case, only/mainly the normal protein can be found in the connective tissue [12, 13]. As opposed to that, DN mutations lead to abnormal protein structure, so the connective tissue contains both the normal and abnormal fibrillin-1 [14].

Genetic testing of the FBN1 gene has been receiving growing attention in the past few years and besides clinical features, it has become one of the key criteria of the diagnosis of MFS in the revised Ghent nosology [2].

To date, only a few well-established connections between genetic background and phenotype have been described [15], e.g. FBN1 mutations affecting a cysteine amino acid are more likely to lead to ectopia lentis [16]. There is also a strong relationship between the severity of MFS and a specific part of the gene: if the so-called neonatal region, which is spread throughout the exons 24–32, is affected, then there is a significantly increased chance for the occurrence of neonatal MFS, which is the most severe form of the disease [16]. Mutations in this region were found to lead to a higher probability of ascending aortic dilation, aortic surgery, mitral valve abnormalities, ectopia lentis, scoliosis and shorter survival even when neonatal MFS was excluded. Therefore, a genetic variant in the region of exons 24–32 can result in a more severe phenotype and can be an indicator of early onset aortic risk even in the absence of neonatal MFS [16].

Establishing the differential diagnosis is an important aspect as many related disorders of MFS lead to similar clinical appearance, but require different therapeutic approach [2]. Comparing to MFS, the Marfan-related Loeys-Dietz syndrome (LDS) may present with more aggressive clinical presentation, characterised by rapidly growing aneurysms, and aortic dissections occurring at a younger age and with smaller diameters. Furthermore, LDS dissections can take place in the peripheral arteries too. These indicate the need for more frequent and extended surveillance and lower aortic diameter as the indication criterion for a prophylactic surgery [17]. Therefore, differentiating between MFS and LDS carries huge clinical importance. In LDS, the TGFBR1, TGFBR2, SMAD3, TGFB2, TGFB3 and SMAD2 genes are affected [18]. On the other hand, some of the diseases that result in a Marfan-like appearance may not lead to severe CV complications. One example of that is congenital contractural arachnodactyly (Beals syndrome), which is caused by a mutation in the FBN2 gene [19]. Therefore, it is pivotal to identify the actual syndrome the patient has, so the proper management can be carried out.

Some of the characteristic features of MFS are aortic dilation and dissection, mitral valve prolapse, chest deformities, dolichostenomelia, scoliosis, skin striae, myopia and ectopia lentis [2].

The most dangerous, life-threatening complication of MFS and other genetic aortopathies is aortic dissection, which occurs at the average age of 63 years in the general population, ~ 38 years in MFS [20] and ~ 27 years in LDS patients [21]. Approximately two-thirds of aortic dissections belong to group type A in the Stanford classification system, meaning that they involve the ascending aorta. Without surgical intervention, an acute type A aortic dissection has a mortality rate of 20% after 24 h and it increases to 30% after 48 h [22]. The mortality of the operation of an acute type A aortic dissection can reach even 20%, while it is only around 1.5% in case of prophylactic surgery [23].

The indication for prophylactic surgery is based on aortic diameter. The threshold is 50 mm for MFS patients, which goes down to 45 mm in the presence of any of the risk factors stated in the European Society of Cardiology (ESC) guidelines [24]. The main concern is that a relevant number of aortic dissections occur at smaller diameters. In a study by Neri and his colleagues, one-third of aortic dissections took place at normal or mildly enlarged diameters [25], and in another article, Kim and his colleagues reported a smaller than 45 mm aortic diameter for 26% of dissections [26]. However, too early, and unnecessary operations should be also avoided. As aortic diameter alone is not appropriate to predict aortic dissection, a model that could precisely determine the risk and the possible onset should be created [27‐29].

Therefore, our research aimed to identify the pathogenic genetic variants of clinically diagnosed MFS patients with the highest achievable detection rate. We also aimed to examine the relevance of the use of a multi-gene panel in the investigation of patients with a Marfanoid habitus.

Our further objective was to identify genotype–phenotype correlations that could improve the risk stratification for severe aortic events and could be applied within clinical settings.

Figure 1 shows the distribution of genetic screening steps applied for the two sets of patients and the results of the two phases of the study.

Currently, the most widely applied algorithm for genetic testing in MFS patients is the sequence analysis of the FBN1 gene followed by CNV screening, and in negative cases the investigation of the TGFBR1 and TGFBR2 genes [41].

In our two-phase genetic testing study, we included 136 patients with either a clinical diagnosis of MFS or Marfanoid habitus, the first thorough genetic screening study of the Hungarian Marfan population. Due to the overlapping features of MFS and its related disorders, we used a multi-gene panel to investigate the relevant genes.

The mutation detection rate of 62% in our study is in the range found in the literature. The success of FBN1 mutation identification can be as high as 93% in individuals with clinically diagnosed MFS. However, this rate shows a huge variability. Arnaud and his colleagues had a 56% detection rate in their study [42] while Baetens reached a 92% success rate [31]. The previously mentioned studies only investigated the FBN1 gene, but examples of multigene testing are also available. Yang and his colleagues used a 15-gene aortopathy panel for 248 patients and they identified 92 (37.1%) (likely) pathogenic mutations, 89% of them affecting the FBN1 gene [43]. In the study of Lerner-Ellis, 594 individuals with the suspicion of Marfan syndrome, Loeys-Dietz syndrome and Thoracic Aortic Aneurysms and Dissections were tested, and 112 individuals had a positive result (19%). The authors state that the reason behind the low detection rate is the errors made during the patient selection process as they were not referred for testing by individuals with expertise in the field [44]. In a study by Wooderchak-Donahue, a 10-gene panel was applied, and 10.3% of the 175 tested individuals appeared to be positive, while 18.3% had a VUS [41].

When we compared the general traits of patients with a detected mutation to the individuals without a detected mutation, only their Body Mass Index (BMI) differed significantly, the characteristic Marfanoid features and their systemic score were similar. This could suggest that errors in our patient selection were not the main reason for not reaching the maximal detection rate.

We investigated 9 genes, so one explanation for the lower success rate can be the involvement of other genes that were not covered by our gene panel. Mutations in the intronic regions have been reported [33], which could also contribute to our number, as we have not sequenced deep/non-canonical intronic regions.

Yang and his colleagues applied MLPA for 115 samples that came back negative after a 15-gene panel. As a result, they found 5 large deletions in the FBN1 gene (4.3%) [45]. Consistently with our previous study, these findings highlight the importance of CNV screening in point-mutation negative cases, to increase the detection rate of disease-causing genetic variants [11].

Increasing the mutation detection rate is highly important as identifying the patients’ pathogenic mutation carries several benefits.

First, having an identified mutation helps to confirm the diagnosis of the disease. Therefore, it verifies the need for the management and treatment of the affected individuals. Furthermore, having a definite diagnosis makes it easier for patients to accept the fact that they must live with such a serious condition.

A detected pathogenic mutation can help to identify affected family members with targeted screening for the known genetic variant. This method is quick and inexpensive. When the result is negative, the disease can be excluded and unnecessary follow-ups can be avoided, no further management is required. Ruling out the possibility of the disease has a great influence on people’s anxiety level and therefore the quality of life, as a significant difference was reported on the scores of trait anxiety between the Marfan and the normal population [46]. In the case of a positive test, adequate treatment and patient management need to be initiated. It is most useful in younger patients, as in most families phenotype only becomes apparent with increasing age [1], which makes the early diagnosis uncertain and delays the definite one. The clinical importance of early clinical diagnosis is the possibility of having a greater effect of early treatment to prevent the serious complications of the disease.

Patients can also benefit from knowing their disease-causing mutation in the process of family planning. The parents could make their decision of having children with the information on their risk of passing on the disease to their offspring. A heterozygous MFS patient and a healthy individual have a 50% chance of passing on the mutation to their child.

A dynamically evolving field, where the identified mutation could be useful is preimplantation genetic diagnostics [47].

Seven percent of the identified (likely) pathogenic variants affected some of the genes associated with LDS, which is a related disorder of MFS with a potentially more aggressive clinical presentation. Therefore, it requires a different therapeutic approach, including follow-up with extended imaging and a lower aortic diameter threshold for prophylactic aortic root replacement surgery [17]. Our patients with LDS had a significantly lower systemic score and they showed a tendency to be younger than people with MFS. However, they both resulted in severe CV manifestations. In Hungary, our research group was the first that carried out genetic testing for LDS patients. The significance of making the right diagnosis can be clearly seen, and due to the overlapping clinical features, in certain cases, it can be achieved by using gene-panel testing.

The clinically most relevant use of genotype–phenotype correlations is identifying genetic variants which are associated with more severe CV involvement.

We found that the combined group of HI and DN Cys mutations led to aortic involvement (aortic dissection, aortic dilation) significantly more frequently than DN non-Cys genetic variants. Furthermore, aortic surgery was significantly more common among patients with DN Cys mutations than in the other two groups. Our results show a strong correlation between the type of mutation and the severity of CV manifestations, which has the potential to improve the risk stratification of the patients in order to optimise the decision making on the necessity and timing of prophylactic aortic root surgeries.

Some articles focusing on the connections between the genotype and CV involvement have been published. Becerra-Muñoz and his colleagues carried out genetic testing for 108 individuals with suspected MFS and they detected 90 FBN1 mutations. They found that patients with MFS and truncating variants had a higher percentage of aortic events than patients with a missense mutation [15]. Similarly, after examining 179 probands with pathogenic or likely pathogenic mutations, Baudhuin and colleagues reported a higher frequency of truncating and splicing variants in patients with an aortic event. They also found that patients with these mutations had an aortic event at a younger age than people with a missense variant [48]. The results of Franken et al. also support these findings by stating that HI patients have a 2.5-fold higher risk of CV death, a 2.4-fold increased risk of reaching the combined endpoint of dissection or death and a 1.6-fold higher risk of any aortic complication compared to DN mutations [49]. These articles treat the missense mutations as a homogenous group, and they all consider this mutation type as lower risk for serious CV manifestations. However, we further differentiated the DN group and we identified a subgroup that seems to be even more dangerous than the HI genetic variants.

Our findings support the results published by Takeda et al. on this topic. They also identified a subgroup within the DN mutations with a higher CV risk. As they reported, DN variants affecting or creating cysteine residues and in-frame deletions in exons 25–36 and 43–49 (DN-CD) had a 6.3-fold higher risk for aortic events than the other DN mutations (DN-non CD), which is comparable with or more deleterious than HI variants. The growth of the aortic diameter also appeared to be larger in the DN-CD + HI group than in the DN-non CD one [40]. Faivre et al. reported a significantly higher probability of ascending aortic dilation in patients with a mutation eliminating a cysteine than in patients with a variant resulting in cysteine creation [16]. These results also confirm the relevance of differentiating among DN mutations; however, we propose that DN Cys genetic variants could be considered higher CV risk than HI ones as they reached the need for aortic surgery significantly more frequently. There was no difference in their age at the last follow-up. More investigations need to be carried out to support this idea. Until then DN Cys and HI variants should be treated equally as high risk in the clinical practice. The location of the mutation can also have an impact on aortic involvement. In the above mentioned study by Faivre et al., mutations in exons 24–32 led to ascending aortic dilation significantly more frequently than variants found in other exons [16]. In contrast, our results did not show significant difference in terms of aortic involvement in patients with variants in exons 24–32 compared to individuals with mutations in other parts of the gene. However, the effect of the location of mutations is worth further investigating in future studies. The relevance of these studies is also highlighted by the finding that some patients with a DN non-Cys variant underwent aortic surgery at younger age than patients with DN Cys and HI mutations, even if no statistical difference could be observed in terms of age at the time of surgery among the mutation types. This could indicate that certain factors like location of mutation or affected domains may also influence CV severity. We are going to continue our study by involving further patients for gene panel testing. We also need to consider whether adding more genes to our panel would be a cost-effective way of increasing our detection rate.

We acknowledge the limitations of our study, including the small sample size. However, it is comparable to the sample sizes of previous studies on this topic and we were able to draw significant conclusions on the studied cohort.

We analysed a limited number of genes, therefore our multi-gene panel did not cover all the known disease-causing genes. In addition, the applied targeted screening method has its limitations, such as incomplete coverage and the need for regularly updating the gene panel due to novel gene-disease associations [10]. Despite these, our study successfully demonstrated the importance of using a multi-gene panel for patients with Marfanoid habitus.

Deep/non-canonical intronic regions were not sequenced, which could be a further limitation of our study [50].

Most of the patients in our study were started on antihypertensive drugs at the time of diagnosis of MFS or when significant and/or progressive aortic dilation was observed. Therefore, we cannot evaluate the contribution of the specific mutation and of the medication to the severity of aortic phenotype. However, as most MFS patients take blood pressure medications to prevent aortic complications, our study still provides relevant findings on the different effect of specific mutations on the severity of CV manifestations. It would be highly beneficial to carry out prospective studies to investigate the response of specific mutations to cardiac medications.

Prospective studies are needed to produce results that could be used in clinical practice in terms of genotype and CV manifestations. Despite the retrospective design of our study, it still provides significant findings on the correlations between genetic background and aortic involvement, and it could provide a base for further research.

Based on our results, we propose that the optimal way of genetic testing of MFS is the use of a gene panel, including FBN1 and MFS-related genes, combined with CNV analysis with MLPA, if necessary. A summary of the recommended algorithm can be found in Fig. 4. This method can help to make the exact diagnosis and increase the detection rate.

We emphasize the need for genetic testing for patients who show Marfanoid features but their systemic score is below 7, as LDS patients may have lower scores, but they are likely to have severe CV manifestations.

MFS patients with DN Cys and HI mutations are at increased risk for aortic involvement, and DN Cys mutations seem to lead to the most severe aortic involvement. This finding could have an impact on patient management; however, further research is required to enable genetic information to be included in the risk stratification.