The revised role of TGF-β in aortic aneurysms in Marfan syndrome

For this sub-study we measured baseline TGF-β levels of 99 MFS patients, all monitored by the Academic Medical Centre Amsterdam, of whom 55 were randomised to 100 mg losartan and 44 to no losartan. In 42 patients on losartan therapy, plasma TGF-β levels were also assessed after 1 month of treatment. We recruited 22 ‘healthy controls’ in whom MFS was definitely ruled out (Fig. 1a). In order to determine AoDR, MFS patients underwent magnetic resonance imaging of the aorta at baseline and after 3 years of follow-up.

Despite inter-laboratory variations in TGF-β measurements throughout the medical world, our TGF-β measurements show that MFS patients had significantly higher TGF-β levels compared with healthy controls (121 pg/ml versus 54 pg/mL, p = 0.006), which is in line with our previous results [2]. Surprisingly, 1 month of losartan therapy did not reduce circulating TGF-β levels (101 pg/mL; 95% CI: −27:229 pg/mL; p = 0.12) (Fig. 1b). TGF-β levels were only normalised to the control level in 15 of the 42 patients (p = 0.26). In the remaining 27 patients TGF-β did not decrease after one-month losartan therapy, and showed a significantly higher TGF-β compared with the healthy controls (292 pg/ml versus 54 pg/mL, p = 0.028). Unexpectedly, after 3 years of losartan therapy, patients with a decrease in TGF-β had a significantly higher AoDR compared with patients without reduced TGF-β levels (1.5 ± 0.8 mm/3 years versus 0.5 ± 1.2 mm/3 years, respectively; p = 0.04). In addition, 91% of the patients with a decrease in TGF-β showed a significant increase in AoDR despite losartan therapy, compared with 33% of the patients with increased/stable TGF-β levels after losartan (p = 0.013) (Fig. 1a,b). In order to explain these unexpected results, we compared baseline characteristics between these groups. Patient groups were comparable, with the exception of baseline TGF-β levels (Table 1). Patients showing a decrease in TGF-β after losartan therapy had significantly elevated baseline TGF-β levels compared with patients who did not show this decrease (189 ± 166 pg/ml versus 94 ± 113 pg/ml, p = 0.05) (Fig. 1b). Interestingly, we found a linear association between the change in TGF-β level and the increase in AoDR in patients using losartan therapy (r = 0.47, p = 0.02) (Fig.1c). This effect could not be ascribed to β-blocker therapy, because β-blocker therapy +/− losartan therapy was not different at any level (data not shown).

This sub-study highlights a paradoxical topic of TGF-β in the pathogenesis of MFS. Although 3 years of losartan therapy reduces the overall AoDR in MFS patients [9], only one-third of MFS patients responded by a reduction of plasma TGF-β after 1 month of losartan therapy. These responders had elevated baseline TGF-β levels and an increase in AoDR after 3 years, despite losartan therapy. Considering treatment with 100 mg losartan is a sufficient dose to reduce AoDR in MFS [9], and assuming that one-month treatment is sufficient to initiate a TGF-β response, we have three possible explanations for our findings.

The first explanation for the fact that losartan did not reduce overall TGF-β levels is that TGF-β is a readout of the diseased state of the aorta. Losartan did reduce TGF-β levels in a subgroup of MFS patients, yet these patients revealed a higher AoDR after 3 years of therapy. This increase in AoDR may be explained by the elevated baseline TGF-β levels and the slightly, but non-significant larger aortic root dimension at baseline (45 ± 4 mm versus 43 ± 6 mm, p = 0.215). Both factors are associated with an increase in AoDR. [5] These results suggest that elevated plasma TGF-β is a marker for aortic damage, such as fibrosis, rather than the initial cause of aortic dilatation.

A second explanation for the variable response of TGF-β to losartan therapy comprises the multitude of FBN1 mutations. At present, more than 2900 different mutations have been described in the Universal Mutation Database [13]. Most FBN1 mutations result in expression of mutated fibrillin-1 proteins, which are improperly folded. The abnormal fibrillin-1 protein may have a dominant-negative effect on the structure of the extracellular fibrillin network when it interacts with normal fibrillin-1 protein of the non-mutated allele and other extracellular matrix proteins. Both the strength of the fibrillin-1 matrix may be changed as well as the release of TGF-β that is captured in the fibrillin-1 network. Mutations in one of the seven TGF-β binding protein-like (TB) domains of the FBN1 gene may especially alter TGF-β levels.

Other FBN1 mutations will lead to reduced fibrillin-1 protein levels as a result of deletion of the entire FBN1 gene on one allele [14] or for example upon deletion of the first exon of FBN1 (such that the mRNA is not translated), causing ‘haploinsufficiency’. The reduced level of normal fibrillin-1 protein presumably results in a thinner fibrillin-1 matrix in the vasculature and thus in reduced aortic wall strength. In such patients, angiotensin II (AngII) activation may be increased to maintain normal blood pressure. The AngII-mediated signalling cascade is a common inducer of TGF-β production in the vessel wall and thus involved in the increased plasma TGF-β levels in these patients. Blocking the AngII receptor-1 (AT1) with losartan will diminish TGF-β production as well as other AngII-mediated detrimental processes in the vessel wall such as blood pressure increase, enhanced pro-inflammatory responses, myofibroblast differentiation and reactive oxygen species (ROS) generation. Therefore, we hypothesise that in patients with an FBN1 mutation leading to haploinsufficiency, the effect of losartan on aortic dilatation is better compared with patients with a dominant-negative mutation. We anticipate that genetics plays a critical role in the TGF-β response to losartan [15].

A third explanation for the variable TGF-β response to losartan therapy comprises variation in the abundance of AT1 expression or activity, or by polymorphisms in the renin-angiotensin-aldosterone system (RAAS). Increased AngII-mediated signalling as a result of this type of polymorphisms will coincide with increased TGF-β production. A number of polymorphisms have been identified in RAAS that are associated with aneurysms. For example, the ACE deletion/insertion polymorphism is significantly associated with an increased risk for thoracic aortic aneurysm formation [16].

In MFS mice, direct inhibition of TGF-β was effective against AoDR. [8] However, in most studies TGF-β signalling has been extrapolated from the abundance of Smad2 activation (pSmad2) in the dilated aortic tissue [17]. Smad signalling is best known from its role in the TGF-β-induced signalling cascade, where it transfers the extracellular TGF-β signal to the nucleus to act as a transcription factor and to regulate gene expression [18]. Before birth, TGF-β is essential for the development of the cardiovascular system [19]. In later life, TGF-β is expressed in reaction to injury, mediating a fibrotic response for repair. This accentuates the possibility that TGF-β is a result and not the cause of aortic damage. Interestingly, AngII can induce Smad2 activation directly through its receptor AT1 [20, 21], as well as indirectly by enhancing TGF-β expression (Fig. 2). Thus, increased levels of TGF-β and pSmad2 levels in MFS may both result from increased AngII-mediated signalling. When AngII is considered a primary cause of aortic disease in MFS, other detrimental AngII-mediated pathways may be responsible for initiation of arterial damage (Fig. 2). Excessive TGF-β signalling subsequently leads to secondary disease progression.

In mice, chronic infusion of AngII is known to affect the integrity of the vasculature resulting in aneurysms in the ascending and descending aorta [22, 23]. We propose that the beneficial effect of losartan on AoDR in MFS mice and patients [5, 6, 8] is obtained through inhibition of the unfavourable AngII-mediated signalling cascades, involving TGF-β synthesis and pSmad2 signalling, blood pressure increase, myofibroblast differentiation, ROS generation and pro-inflammatory responses [22, 24].

We wish to emphasise that the power of our study was limited, therefore a prospective trial with a larger patient cohort is needed to confirm our results. Furthermore, TGF-β levels are known to vary between different laboratories. In order to prevent these variations, we collected plasma samples and performed all TGF-β measurements in the same laboratory at the same time. Finally, it would have been interesting to have the follow-up TGF-β measurements after 3 years of losartan therapy. Despite these study limitations, our results are in line with our previous results [5].