Discussion
The evolution of IMHA is very dynamic from complete resolution to aortic dissection [10][16][17][27]. However, for patients with uncomplicated IMHA, which is significantly more common in Asian countries, the “wait-and-watch strategy” is the first-line treatment because of the lower mortality with early medical therapy than with type A aortic dissection [18-20] . However, the development of classic aortic dissection (19%) and retrograde type A aortic dissection (27%) are very common fatal evolutions after medical treatment, and the need for delayed further interventions rises up to 30% within the first 6 months [18][28] . In our study, the hospital aorta-related mortality in both groups (Table 2 ) was similar to that reported by Song et al. [18] (7.9%) and lower than that reported by Sandhu et al [28] (12%).
Newly diagnosed type 2 DM patients with IMHA probably benefit from antidiabetic treatments and DM is possibly a protective factor for IMHA during the chronic phase (>90 days) whereas tight glycemic control may influence the evolution of IMHA at the acute phase (<14 days) (Figure 4D ). The potentially protective value of DM has been well described [ 4-7], and the possible explanations included the increasing matrix of the aortic wall (suppression of plasmin and decreased levels/activity of MMP-2 and MMP-9), and decreased aortic mural macrophage infiltration and neovascularization [12] . MMP-9 is involved in tissue degradation and remodeling in aortic dissection and is significantly increased in aortic dissection patients, and a higher level of MMP-9 can weaken the aortic media by degrading multiple extracellular components; DM patients have a 2-fold decreased level of MMP-9, which may restrict the degradation of the aortic wall [12]. Tan et al[29] demonstrated that MMP-9 may be a useful biomarker for aortic dissection. In our study, the MMP-9 level in the DM group was lower than that in the no DM group (P< 0. 001,Figure 2 ) especially one year after the onset of IMHA. However, all three deaths during the acute phase occurred in the DM group (although there were no significant differences) even though the MMP-9 level was obviously lower (Figure 2 , Figure 4D ) than that in the no DM group. The probable explanation is that insulin treatment may diminish this protective effect of hyperglycemia that prevent the aortic aneurysm development process (under laboratory conditions) [14]; after receiving tight glycemic control recommended by guidelines [1][11][22], theDM group had an MMP-9 level that was dramatically increased (reached the highest value) during the acute phase, which probably indicated the potentially decreased protective effect of hyperglycemia.
Rupture of retrograde type A aortic dissection and ascending aortic pseudoaneurysm after receiving TEVAR were the main causes of death in our study. The explanation for these complications is the choice of TEVAR devices, stent graft landing zones and commitment arch reconstruction surgery which can probably influence the outcomes after TEVAR. Although we avoided balloon dilatation because of the potential risk of retrograde type A aortic dissection [25] , we did have one patient who probably died of retrograde type A aortic dissection after balloon dilation (Patient 4, Supplement 2 ). Additionally, we employed only stent grafts with a proximal bare spring design. Although non-proximal bare-spring stent grafts yield a similar incidence of retrograde type A aortic dissection [25],proximal bare-spring stent grafts have long been regarded as a risk factor and may injure the fragile aortic wall, resulting in the development of retrograde type A aortic dissection [26] . Moreover, in our study, intramural hematomas that affected the aortic arch without obvious entry tears characterized as aortic dissection always required arch revascularization. However, the partial occlusion clamp may injure the fragile aortic wall (during the debranching procedure) and cause further damage that results in lethal complications[29] ( Patient 16, Supplement 2). In addition, a dilated ascending aorta (> 4 cm) (Patients 6,8,9 and 12,Supplement 2 ) and TEVAR in the chronic phase are potential risk factors for fatal complications after TEVAR [30][31].Additionally, five patients died of aorta-related complications complaining of refractory pain after TEVAR (Patients 5, 6, 8, 9 and 10,Supplement 2 ). Juvonen, et al [32] reported that chest or back pain is predictive of aortic rupture, and that patients with uncharacteristic or atypical pain have a higher risk of rupture over time, and that medical management is unwarranted for these patients[33] . In sum, for IMHA, it seems logical to recommend prophylactic replacement of the aortic wall that presented the intramural hemorrhage, because of the risk of fatal complications especially for those with a dilated aorta, intramural hematomas that affect the aortic arch, and refractory pain after TEVAR.
In the “wait-and-watch strategy” the interval of the imaging evaluation is also important. Kitai, et al recommended careful serial imaging because conventional 5-mm axial images may not completely identify ulcer-like projections (ULPs) smaller than 5 mm[34] . Thus, with low-resolution computed tomography scan results, simply equating the lack of a ULP with a favorable prognosis is probably unjustified, and a more precise CTA scan and closer monitoring for the development of a ULP are necessary. Moreover, intravascular ultrasonography [16] is another meaningful examination method that permits a dynamic real-time evaluation of the aorta and can detect the origin of side branches, evaluate adequate expansion after the deployment of the stent graft and exclude potential complications (such as retrograde type A aortic dissection occurring intraoperatively).
There are several limitations of this study. First, future longitudinal prospective investigations with a multicenter cooperation focusing on more patients are necessary. Second, techniques (such as TEVAR devices and arch reconstruction methods) probably affect patient outcomes, and further studies are required for a more standardized and uniform management strategy. Third, the anti-inflammatory effect of oral antidiabetic medication drugs could lower the risk of fatal progression[13], and the enrollment of more patients with different medical treatment strategies (insulin with/without oral antidiabetic medication drugs) may provide more meaningful insight into this question.