The pathogenesis of MFS

Mutations in FBN1 can cause (1) reduction in the fibrillin produced in the cell, that is, quantitative defects, (2) change the structure or stability of the protein, and/or (3) alter the ability of fibrillin to be exported to the ECM (figure 1). Mutations in FBN1 that result in haploinsufficiency (HI) lead to reduced amounts of fibrillin, fewer microfibrils and increased activated TGFβ levels due to decreased TGFβ sequestration. In contrast, dominant negative (DN) mutations in FBN1 result in qualitative defects of the protein that may affect function such as folding or protein–protein interactions leading to disorganisation of the ECM, altered ECM strength and increased TGFβ. The understanding of disease pathogenesis has been facilitated by mouse models of MFS that demonstrate the importance of HI.2 Conceptually, DN mutations should result in more phenotypic variability because mutations at different region of the protein will have distinct effects, whereas HI mutations should have more consistent phenotypes. Only patients for whom a DN or HI mutation type could be assigned were studied by Franken et al. Along with software predictions, the authors also used existing literature to classify the functional effects of several specific variants in FBN1 that were previously studied in detail in vitro. This interesting work presents evidence that patients with HI mutations have a greater aortic risk as compared with DN mutation carriers. Specifically, those patients with HI mutations had aortic surgery at a younger age (but similar overall rate of surgery), had a larger aortic root at the time of initial referral to their centres and showed more rapid dilation of the aortic root over the course of clinical follow-up.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Genetics and pathogenesis in Marfan syndrome. A schematic overview of the effect of haploinsufficiency (HI) or dominant negative (DN) mutations is shown. The most common type of mutations that lead to HI or DN effect are listed. However, these classifications are not absolute, and the precise effect of a specific DNA mutation must be determined at the protein level. Mutations that cause HI result in less protein because only the normal, non-mutated FBN1 allele generates protein products. In contrast, a combination of normal protein products (fibrillin-1) generated by the non-mutated FBN1 allele and stable abnormal protein products generated by the DN allele are seen with DN mutations. The green circle in the fibrillin-1 protein represents the altered amino acid residue. Normal microfibrils are associated with latency associated peptides (LAP; red circles), which bind TGFβ in a biologically inactive complex in combination with latent TGFβ binding protein (LTBP1; blue oval). When microfibrils are reduced due to HI, interaction with matrix fibres is disrupted, and increases in activated TGFβ are seen due to a reduced number of microfibrils. DN mutations can have a range of effects depending on the specific mutation. Shown here, a DN mutation results in abnormalities in fibrillin-1 protein folding and increased activated TGFβ. TGFβ, transforming growth factor beta.

Genetics and mutation type in MFS

Previous studies have also examined mutation type to determine whether genotype–phenotype correlations exist in MFS, dividing mutations into protein truncating or in-frame mutation types.3 4 These studies fail to completely account for the effect of a mutation on the protein product. For example, a protein truncating mutation may result in HI or DN effects depending on the stability of the protein. Interestingly, there was no difference noted in aortic dilation when classifying mutations in this manner, but comparison within the in-frame group based on whether a cysteine residue was involved did show a significantly higher probability of aortic dilation and mitral valve prolapse.4 It would be of interest to determine whether the subset of DN mutations that alter cysteine residues show similar results in the current study. The current study’s approach to mutation classification is more sophisticated and is supported by prior evidence that the software accurately predicts presence or absence of the mutated transcript.5 Nevertheless, examples in the literature demonstrate that in order to assign HI versus DN effects with confidence, one must understand the mutation’s consequences at the protein level.

A limitation for this retrospective study is that patients with the most severe disease who had aortic dissection or aortic replacement surgery (ARS) prior to referral to the centre were not eligible for longitudinal study. Mutations in exons 24–32 are particularly associated with severe, early aortic disease. The majority of mutations in this region are missense or in-frame insertions/deletions, which manifest a more severe phenotype than the minority of truncating variants within this region,4 and a DN effect was observed in fibroblasts derived from patients with mutations in this region.6 The severe nature of these DN mutations seems to contradict the study’s major conclusion of HI mutations’ having more severe effects. One potential explanation would be that patients carrying in-frame variants in this region died at a young age or underwent ARS and were thus ineligible for longitudinal study. Consistent with prior literature, patients with DN mutations in the study cohort were more likely to have ectopia lentis. This may have introduced an important ascertainment bias if ectopia lentis led to earlier diagnosis of MFS at referring centres and facilitated earlier initiation of medical therapies.

The wide phenotypic variability among patients with MFS includes variability even between relatives who carry the same mutation. Intrafamilial variability is a strong counterargument for classifying patients according to FBN1 genotype. Related individuals were included in this study for the majority of the analyses between HI and DN groups. There were 169 probands among the 290 patients studied, with the remaining cohort comprised of relatives. It is interesting that when only one patient from each family was included, the SD of the rate of aortic root dilation decreased, in spite of an overall decrease in number tested. This indirectly suggests that including family members resulted in greater variability in aortic dilation. However, this could have been an artefact of how the patients were selected. Whether certain genotypes are more susceptible to specific genetic or environmental modifiers is an important question for future research.

Cardiac phenotyping of patients with MFS

In the longitudinal analyses of the current study, patients with HI mutations were found to have more rapid dilation than DN mutation carriers at the levels of the aortic root and the ascending aorta by approximately 0.3 mm/year. As the authors discuss, despite having a relatively large cohort, this study was still underpowered to conclusively evaluate differential risk of aortic dissection or need for ARS between HI and DN. Requirement for elective ARS is inherently vulnerable to variations in clinical practice, and institutional biases may have amplified the observations documented here. Given these limitations, the authors used absolute aortic size as another marker of aortic disease severity. When using absolute aortic root size or dilation rate as an endpoint, it is important to appraise the specifics of centre practice variations, individual technical limitations, interval duration between measurements and, importantly, differences in overall body size. Though we would advocate adjusting aortic size relative to body surface area, we recognise important limitations to this approach. First, there are numerous published nomograms that substantially differ from one another due to inconsistent factors including different methods of diameter measurement (eg, in systole or diastole), distribution of ages, genders and race/ethnicity within the reference population, formula for body surface area calculation and whether to include age and gender in Z-score calculations.7 8 Guidelines for timing of elective ARS are mostly based on absolute diameter, but the revised Ghent nosology uses a Z-score threshold as a criterion to diagnose MFS.9 10 This lack of a consensus approach for how to index aorta to body size hinders consistent clinical care and creates research challenges. Given the important genotype–phenotype correlations identified in the current study, the design and interpretation of future validation studies will need to consider the method for determining aortic size. Based on the data provided for the current study, it appears that longitudinal analyses included children as young as age 5 years and included around 50 who were age 18 or younger at the start of this study. This means that some of the observed growth in aortic size was physiological.11 Whether using Z-scores would alter the findings of this study is undetermined, but at baseline there were no differences between age, gender or body surface between mutation type groups. Finally, morphological and haemodynamic factors, such as bicuspid aortic valve and aortic regurgitation, should be considered, if available, when assessing factors that control TAA progression.

Precision diagnostics

Altogether, the current study makes an important stride towards improving cardiovascular risk classification of patients with MFS and supports the authors’ previous findings in the Dutch CONgenital CORvitia registry.12 The study detects an aortic phenotype difference based on FBN1 mutation classifications and stands as an example of the increasingly powerful capability to stratify patients based on individualised genetic testing results. Thus, the study shows the potential value of genotyping all patients with MFS and consideration of the role of allelic heterogeneity in management. In addition, the study highlights the value of large volume centres specialised in genetic and cardiac management of aortopathy to analyse robustly phenotyped cohorts and propel clinical practice forward. Whether and how the study’s main finding, that is, that HI mutations confer worse prognosis than DN mutations, should be incorporated into routine clinical management practices remains to be determined and will require follow-up in larger cohorts. There is prior evidence that mutation type may alter the patient’s response to losartan, providing promise for increasingly tailored patient-specific therapy.5 Additional mechanistic studies to determine how FBN1 mutation type alters aortic phenotype severity or response to medical therapies are warranted. Genotype–phenotype correlations promise to improve clinical management decisions to prevent adverse outcomes such as aortic dissection in higher risk patients and also to avoid unnecessary interventions in patients at lower risk.