In this study, we examined the ability of three 4D Flow CMR derived methods of estimating FL pressurization to predict growth in patients with chronic dissection of the descending thoracic aorta: FLEF, MSDR, and vWERP-derived FL maximum relative pressure (FL ΔPmax). We have shown that FL hemodynamics are significantly altered in patients with aortic growth: FLEF was significantly increased and FL ΔPmax was significantly decreased. Neither MSDR, nor any of the anatomical metrics (such as baseline maximum aortic diameter) clearly differentiated enlarging from stable patients. TL and transseptal hemodynamic parameters did not differ between groups, which is unsurprising considering aortic growth in TBAD is the direct consequence of FL enlargement. Furthermore, FLEF and FL ΔPmax demonstrated moderate-strong correlation with each other, and both were found to be independent predictors of aortic growth after adjusting for baseline aortic diameter. This work is the first application of vWERP to provide a direct pressure assessment of the FL in TBAD patients, and to understand the relation between direct and indirect techniques for assessing FL pressure using 4D Flow CMR. These results lend further credence to the importance of FL pressurization in promoting growth in TBAD, and support further investigation of FLEF and FL ΔPmax as hemodynamic biomarkers of risk in TBAD patients.
Current approaches and mechanisms of aortic growth
Current treatment protocols and surgical criteria are largely based on anatomic variables, the most important of which is maximum aortic diameter [7]. Aortic diameter is a simple metric, and has a direct relationship with tensile wall stress (i.e., Law of Laplace). However, in a recent systematic review of growth in TBAD, maximal aortic diameter was associated with growth in only ~ 50% of studies, highlighting the need for better risk stratification tools [7]. While our results support the association between baseline diameter and aortic growth we also found that FLEF and FL ΔPmax provided additional predictive value over baseline diameter alone. This is likely due to the fact that anatomic variables such as aortic diameter are a consequence of aortic wall pathology rather than a direct cause of it. While aortic growth is the result of a complex set of factors (e.g., hemodynamic stress, mechanobiological responses, aortic tissue strength), most of these factors are challenging to non-invasively measure in patients, highlighting the unique value of hemodynamic assessment by 4D Flow CMR.
Metrics of false lumen pressurization
In this study, we chose to assess the predictive value of three different methods of assessing FL hemodynamic stress: FLEF, MSDR, and FL ΔPmax. These three metrics represent different ways of—either directly or indirectly—describing the relationship between flow and pressure in the FL, and the three also represent different ways of interrogating the acquired flow field: utilizing bulk flow, acceleration rate, or relative pressure, respectively.
FLEF has been previously described in TBAD [18], with this metric describing the ratio between retrograde and antegrade flow through the dominant entry tear. FLEF is a regional and indirect measure of assessing FL pressurization. An increasing proportion of FL inflow relative to outflow will result in increased FL pressure (particularly diastolic pressure [12]) and increased resistance to forward flow. During diastole, flow reversal occurs when FL diastolic pressure supersedes diastolic pressure in the TL, and this reversed pressure gradient drives blood from the FL into the TL across entry tears. This ratio of retrograde and antegrade flow at the entry tear, mediated by diastolic pressure gradients, can thus be posed as a surrogate measure of the pressure difference between FL and TL (a conceptual illustration of the relationship between antegrade/retrograde and aortic growth is show in Fig. 4). Interestingly, the relationship between retrograde flow and pressure overload has been described not only in the FL of TBAD patients, but also as a maker of pulmonary hypertension severity [31]. Advantages to the FLEF approach include: bulk flow rate measurements in TBAD with 4D Flow CMR have been shown to be highly reproducible [32] and are easily performed using a variety of commercially available software, flow measurements are performed at a discreet anatomic location (i.e., entry tear) and are thus fairly robust to variations in dissection anatomy, and this approach avoids technical difficulty and potential inaccuracy related to conversion of spatiotemporal flow gradients in its computation. However, FLEF does not directly measure pressure, does not take into account the hemodynamics at distal re-entry tears, and the definition of a 2-dimensional flow analysis plane can be difficult if entry tear anatomy is complex. Nevertheless, the moderate-strong correlation between FLEF and aortic growth in our data underlines this metrics potential clinical utility.
Alternatively, MSDR is a recently proposed semi-regional, indirect method of assessing FL pressurization [24], representing the average peak systolic deceleration rate in a proximal sub-section of the FL. As with FLEF, the coupling to FL pressurization is intuitive: with increasing FL pressures, resistance to forward flow will increase, antegrade systolic flow will decelerate more rapidly, and the MSDR will consequently increase (see Fig. 4 for a conceptual depiction of changes in MSDR with increasing aortic growth). MSDR was recently introduced and studied in a cohort of 29 patients (combined repaired type A dissection and TBAD), where a weak-moderate, positive correlation with aortic growth rate (r = 0.48) was reported [24]. While we identified a similarly weak-moderate positive correlation between MSDR and aortic growth rate (r = 0.40), this correlation did not reach statistical significance, possibly owing to the smaller size of our cohort. However, the weaker correlation between MSRD and aortic growth compared to FLEF and FL ΔPmax may be a result of the MSDR metric itself, given that the MSDR measurement relies on SNR and the temporal sampling rate of the acquired flow field, which can be variable in TBAD. Furthermore, the presence of significant secondary flow features (e.g., helices and vortices) may affect MSDR measurements, and appropriate definition of the analysis subsection becomes less clear when the entry tear is located more distally along the descending aorta.
While FLEF and MSDR are proposed as indirect measures of FL pressure, recent developments in physics-based image analysis now enable the accurate extraction of relative pressure over vascular sections. vWERP is a validated method for relative pressure measurement that utilizes the concept of virtual work-energy, which has shown particular promise in assessing complex vasculatures and has been explicitly tested in TBAD anatomy [25]. As such, the FL ΔPmax represents a global and direct measure of change in pressure from the ascending aorta to the distal thoracic FL. FL ΔPmax should thus decrease with increasing FL pressure (and constant ascending pressure) and growth. This is concordant with our observed moderate-strong negative correlation between FL ΔPmax and aortic growth rate (see Fig. 4 for a conceptual depiction of changes in FL relative pressure with increasing aortic growth). Similar to FLEF and MSDR, vWERP can be derived from 4D Flow data alone, however, the method does require the creation of an auxiliary virtual field in order to compute relative pressure. Although not an overly time-consuming process (~ 5 min on a local desktop computer), vWERP computation requires aortic segmentation, which can be time-consuming in TBAD, and separate computational implementation, making it a more complex analysis than other measures such as FLEF. Lastly, an advantage of the vWERP approach is that relative pressure is derived over the entire thoracic aorta, and thus the global hemodynamic state of the FL is accounted for in its computation.
When considering the differences between FLEF and FL ΔPmax, we found a slightly stronger correlation of FLEF with aortic growth compared to FL ΔPmax on both adjusted and unadjusted analyses (r = 0.78 vs. r = − 0.64, Fig. 3), although it's unclear if such small differences in strength of correlation are meaningful or simply related to statistical noise. The simplicity in derivation makes FLEF an attractive metric in clinical instances where the FL has a single dominant tear in the thoracic aorta, whereas vWERP may be better suited in scenarios with multiple or complex flap fenestrations/tears (as shown in previous in-silico work [25]. Regardless of the metric, our analysis underlines the pathophysiological importance of FL pressurization in TBAD growth, adding to the increasing number of studies highlighting its diagnostic role [10,11,12,13, 18, 24, 33]. In reality, since both FLEF and FL ΔPmax can be measured from the same 4D Flow acquisition, measurement of multiple parameters may be a complementary approach that lends additional diagnostic certainty in cases where pressurization assessments are concordant.
Lastly, it is important to acknowledge potential sources of measurement variability in these hemodynamic metrics: manual plane placement at the entry tear for FLEF, and aortic segmentation for MSDR and FL ΔPmax. The results from our reproducibility analyses address this in detail (Additional file 1: Appendix B). In brief, for the three derived hemodynamic metrics we identified no significant inter-rater bias, although agreement was highest for FL ΔPmax and lowest for FLEF. However, the degree of variability for FLEF and FL ΔPmax were substantially lower than the mean differences in these metrics between stable and enlarging groups, suggesting that this degree of variability is still acceptable for separating patients at high and lower risk of growth. Additionally, FL ΔPmax demonstrated higher variability related to aortic mask dilation rather than erosion into the lumen, a behavior that is consistent with prior vWERP results [34]. Along the same lines, modest interrater differences were noted with measurement of aortic growth rate, an observation which is not surprising given the known variability of diameter measurement in dissected aortas; however, growth categorizations (stable vs. enlarging) remained concordant between readers and for all cases despite this measurement variability. Nevertheless, care must be taken when deriving any of the aforementioned metrics in future scientific or clinical studies, and raters with experience in accurately delineating dissection anatomy are key for reliable anatomic or hemodynamic assessment.
Limitations
Our study has several limitations. First, our study cohort was relatively small in size, and thus these findings should be interpreted as preliminary and hypothesis generating, although efforts to study these metrics in larger cohorts are ongoing. In part our small sample size was due to the fact that we excluded 3 patients for non-contrast exams (inability to accurately segment the TL and FL) and another 3 patients for arrhythmia-related artifacts. However, we believe that these results can be viewed as a representative example of how advanced flow imaging can provide unique insights into the complex hemodynamic mechanism of TBAD.
Second, exclusively chronic dissections were analyzed in our study, and as such it is impossible to infer causal relationships between aortic growth and our three investigational metrics. However, given that abnormal FL blood flow been linked to aneurysmal growth in previous imaging studies of acute TBAD patients [8] and in computational studies [35], we believe it is reasonable to assume that such abnormal hemodynamics are a precursor of progressive FL growth. Efforts are ongoing to validate these findings in acute/subacute TBAD patients.
Third, although not unique to our study, the acquisition of 4D Flow CMR data is not available or part of routine clinical CMR protocols at most centers, although such data can be acquired on almost all modern clinical CMR systems. Further investigation into the specific clinical utility of these metrics of FL pressurization will be needed to promote more wide-spread clinical translation. Lastly, aortic growth was determined by measuring diameter changes between different modalities (e.g., CT at baseline and MRA at follow-up). However, a recent study comparing inter-modality differences in aortic measurements indicate only small and non-significant differences when a consistent measurement technique is used [28], as was done in this study.