The hemodynamic effects of acute aortic regurgitation into a stiffened left ventricle resulting from chronic aortic stenosis

Abstract

Acute aortic regurgitation (AR) post-chronic aortic stenosis is a prevalent phenomenon occurring in patients who undergo transcatheter aortic valve replacement (TAVR) surgery. The objective of this work was to characterize the effects of left ventricular diastolic stiffness (LVDS) and AR severity on LV performance. Three LVDS models were inserted into a physiological left heart simulator. AR severity was parametrically varied through four levels (ranging from trace to moderate) and compared with a competent aortic valve. Hemodynamic metrics such as average diastolic pressures (DP) and reduction in transmitral flow were measured. AR index was calculated as a function of AR severity and LVDS, and the work required to make up for lost volume due to AR was estimated. In the presence of trace AR, higher LVDS had up to a threefold reduction in transmitral flow (13% compared with 3.5%) and a significant increase in DP (2-fold). The AR index ranged from ∼42 to 16 (no AR to moderate AR), with stiffer LVs having lower values. To compensate for lost volume due to AR, the low, medium, and high LVDS models were found to require 5.1, 5.5, and 6.6 times more work, respectively. This work shows that the LVDS has a significant effect on the LV performance in the presence of AR. Therefore, the LVDS of potential TAVR patients should be assessed to gain an initial indication of their ability to tolerate post-procedural AR.

NEW & NOTEWORTHY

This study has demonstrated the hemodynamic effects of acute aortic regurgitation (AR) on a left ventricle (LV) that has experienced chronic aortic stenosis (AS). The LV diastolic stiffness resulting from chronic AS dramatically alters the LV's ability to compensate for the lost volume due to AR.

the purpose of the aortic valve (AV) is to prevent the backflow of blood into the left ventricle (LV) from the aorta. Impairment of this function results in retrograde flow from a high-pressure aorta into a lower pressure LV. This complication can result from a multitude of pathological events that can affect the valve primarily e.g., endocarditis, rheumatic disease, or affect the ascending aorta primarily, e.g., ankylosing spondylitis, idiopathic dilation of the aorta, Marfan syndrome, etc. (3, 7, 12). However, a new mechanism for aortic regurgitation (AR), commonly known as paravalvular leak, stems from a relatively recent AV replacement technique, transcatheter aortic valve replacement (TAVR). Although this minimally invasive surgical technique has many advantages for the treatment of high-risk and inoperable aortic stenosis (AS) patients, it usually results in trace to mild, and in some cases moderate, paravalvular AR (16, 21, 23, 29). This complication (paravalvular AR) has significant correlation to increased patient mortality (16, 21, 23, 29, 30).

When compared with patients who have isolated AR, the 1-yr mortality rate of mild AR patients with preexisting AS is about three to four times higher (21, 30). The preexisting isolated AS in TAVR patients has most likely resulted in an increase in the transvalvular pressure gradient across the AV, eventually leading to myocardial hypertrophy. Myocardial stiffening impairs LV relaxation during the diastolic phase of the cardiac cycle (17, 18). Compounding the pathological changes due to isolated chronic AS with the effects of AR compromises the LV's ability to increase its end diastolic volume (EDV) effectively; hence, the total stroke volume (SV) and forward SV decrease (6, 10, 17). It is therefore suspected that this sudden change from pressure overload due to AS, to volume overload resulting from acute AR, affects patient survival.

To gain deeper insight on the severity of AR post-chronic AS, Sinning et al. (28) proposed a new metric to supplement what is clinically used for AR classification, the AR index. This novel metric utilizes the ratio of the pressure gradient across the AV during diastole to the peak systolic aortic pressure. Sinning demonstrated that this new index is correlated to a 1-yr mortality risk. Nevertheless, the AR index only provides information regarding the severity of acute AR post-chronic AS and offers insight to the chances of survival of an individual. It is also necessary to understand the hemodynamic and survival consequences of inducing AR in an individual who has already experienced LV remodeling due to chronic AS. In other words, how well can a LV, which has undergone remodeling resulting from chronic AS, tolerate acute AR?

We hypothesize that the increased LV diastolic stiffness (LVDS) resulting from chronic AS decreases the efficiency of the LV while concurrently increasing the stroke work (SW) required to maintain a healthy forward SV. Decoupling AR and LV function could generate a more cohesive understanding of AR in the presence of AS and could lead to better patient-specific therapeutic strategies.

In this current study, the isolated effects of LVDS and AR on LV performance were evaluated. We demonstrated that the LVDS of a patient should be taken into consideration to gain insight on the patient's ability to tolerate AR post-TAVR. In this study, AR severity and LVDS were varied in an extensively validated, physiological left heart simulator (26, 27). The use of this fully controllable experimental environment allowed for the study to be performed in the absence of other confounding factors that exist in vivo.

MATERIALS AND METHODS

Left ventricle.

The LV model geometry (Fig. 1A) was designed based on the end systolic cardiac phase of a healthy human subject [please see our previous work for more details (26)]. A 1-mm thick, isotropic, silicone polymer was used to model the LV wall. The LVDS was increased by varying the composition of the polymer, thereby modifying its ability to stretch under an applied pressure.

Fig. 1.

Left ventricle (LV) model (A) and increase in LV pressure (B) as a function of LV volume from peak systolic state. The y-axis is a change in volume from the end systolic volume of the LV models. The diastolic stiffness values for the low, medium, and high LV models were 1.24, 1.72, and 2.72 mmHg/ml, respectively.

To measure the LVDS of each of the models, the inner LV chamber was filled with fluid (to the end systolic volume-nonstretched state of the LV model) and then sealed. Fluid was then incrementally added, stretching the LV models (please see Fig. 1B). At each volume increment, the absolute pressure within the LV was measured. The stiffness of each of the LV models was then approximated as the slope of a linear fit to its corresponding measurements. For this study, three LVDS values (1.24, 1.72, and 2.72 mmHg/ml) were chosen such that they varied within the range of patients with chronic AS (15, 18). For the purpose of this work, we refer to the above LVDS values as low, medium, and high stiffness, respectively. The LV models were custom manufactured by Venair (Terrassa, Spain) via a casting process.

Aortic valve.

The AV used in this work was constructed from an excised 23-mm porcine aortic valve. The commissures of the valve leaflets were sutured to the posts of a three-pillared stent (Fig. 2A). The sutured aortic valve was then lightly fixed with 0.1% glutaraldehyde solution for 6 h. AR was induced via a second stent (AR stent), which was inserted into the coaptation zone of the AV such that the leaflets coapted along the walls of a central, hollow AR channel (Fig. 2, B and C). The walls of the AR channel were 0.2 mm thick, too thin to significantly affect the pressure drop across the AV (Fig. 2D and Table 1). The geometry of the AR stent regurgitant orifice area (ROA) was taken from computed tomographic images of patients with central AR as reported by Calleja et al. (9). To induce multiple levels of AR, the extracted ROA was scaled by a constant. In this work, five levels of AR conditions (4 plus control) were investigated, with ROAs ranging from 0 to 0.16 cm² (Table 1). This ROA range led to AR fractions of up to ∼45% [mild to moderate according to American Heart Association guidelines (6)].

Fig. 2.

A: excised porcine aortic valve sutured to the 3-pillared stent. B: aortic regurgitation (AR) stent model. C: aortic valve leaflets coapting along the sides of the AR stent model walls. D: open aortic valve during systolic flow with AR stent present.

Table 1.

Hemodynamic parameters and classification of AR according to AHA (6)

AR Stent	ROA, cm2	Mean TVPG, mmHg	Mean AR Fraction, %	AHA Classification
Control	0	8.13 ± 0.35	2.20 ± 1.64	None (closing volume)
1	0.0121	9.98 ± 0.56	9.20 ± 1.77	Trace
2	0.0331	10.48 ± 0.49	15.7 ± 0.25	Mild
3	0.0866	11.71 ± 0.12	32.0 ± 1.24	Mild/moderate
4	0.1598	12.45 ± 1.22	43.7 ± 4.58	Moderate

The term “mean” refers to an average across all left ventricular models tested. AR, aortic regurgitation; AHA, American Heart Association; ROA, regurgitant orifice area; TVPG, transvalvular pressure gradient.

Left heart simulator.

The LV models and AV were integrated into a validated physiologic left heart simulator (LHS) (26, 27). In summary the LHS consists of a rigid atrium, a compliant LV, and a rigid aorta. The LHS was designed to be modular such that the valves and the LV could be easily exchanged. A programmable linear piston pump controlled the motion of the LV (Fig. 3). The blood analog used was a mixture of 36 vol/vol% glycerin-water solution, which mimicked the viscosity of blood. A 33-mm Hancock II valve was used at the mitral valve position.

Fig. 3.

Schematic of left heart simulator. The dashed arrows indicate the direction of flow. PPP stands for programmable piston pump. F1 and F2 are flow probes.

Lumped systemic resistance and compliance elements were used to tune the simulator to physiologic conditions. Flow rates into and out of the LV were measured using ultrasonic flow probes (TS 410; Transonic Systems, Ithaca, NY). Pressures were measured in the LV and aorta using strain-gage pressure transducers (model 6199; Utah Medical Products, Midvale, UT). An in-house LabVIEW (National Instruments, Austin, TX) program was used to record 200 cardiac cycles of hemodynamic data at a frequency of 2 kHz. Due to the robustness and high repeatability of the system, the cycle-to-cycle variations were insignificant. The standard deviations in the global hemodynamic parameters collected were found to be <1% of their respective mean values.

Hemodynamics conditions.

Five levels (including a competent AV-control) of ROAs were parametrically varied along with three levels of LVDS. For each of the LV models, baseline hemodynamic conditions were initially established in the LHS utilizing a competent AV, after which the AR stents were sequentially introduced to the system. The baseline hemodynamic conditions were as follows: ∼125/80 mmHg aortic pressure, heart rate of 70 beats/min, cardiac output (CO) of ∼5.0 l/min, diastolic duration of ∼67% of the cardiac cycle, peak aortic flow rate of ∼28 l/min, and peak mitral E- and A-wave flow rates of ∼18 and ∼16 l/min (E/A ratio of ∼1.13), respectively. Within each of the LV models, the work input to the simulator via the linear piston pump was held constant. This was done by monitoring the pressure in the outer LV chamber and keeping the linear pistons' displacement constant across each LV model condition.

RESULTS

Changes in LV hemodynamics.

For all LV models tested, AR severity increased with increasing ROA (ranging from none to moderate AR; Table 1), leading to a decrease in the effective CO. Figure 4 shows that the peak aortic flow rate for all ROAs studied remained relatively constant at ∼28 l/min. The LV filling rate via the mitral valve (MV) was found to decrease during the E and A waves as ROA increased. This decrease in LV filling rate was more pronounced as LVDS increased. Figure 5 shows a difference in the temporal integration of LV filling curves via the MV of each experimental condition from its respective control case (no AR). It indicates the changes that occur in LV filling volume as LVDS increased. As the LVDS increased by a factor of ∼2, the corresponding reductions in LV filling volumes as AR was induced increased. This increase in the reduction of LV filling was more apparent for lower AR fractions.

Fig. 4.

Representative mitral and aortic flow curves for all AR levels: low left ventricular diastolic stiffness (LVDS) model.

Fig. 5.

Percent reduction in transmitral filling as a function of AR severity and LVDS. The legend represents LVDS.

Due to an increase in the back pressure from the aorta, the DP within the LV increased as ROA increased. The LV diastolic pressure (LVDP) was also observed to increase as LVDS was increased as depicted in Fig. 6. This rate of increase in LVDP as ROA increased was found to be similar across LV models.

Fig. 6.

Increase in LV diastolic pressure (LVDP) as a functions of AR severity and LVDS. The legend represents the severity of LV stiffness.

AR index.

The AR index was calculated using Eq. 1 below (28),

$$AR\hspace{0.17em}index=100\times \frac{DBP-LVEDP}{SBP}$$ (1)

where DBP is diastolic blood pressure in the aorta, LVEDP is the left ventricular end diastolic pressure, and SBP is the systolic blood pressure in the aorta. Figure 7 shows the change in the AR index as a function of AR severity and LVDS. The values of AR index measured in the study ranged from ∼42 to 16. As ROA increased, the AR index decreased. In the same fashion, it was observed that as LVDS increased, the AR index decreased. The reduction in AR index as a function of ROA was quantitatively similar when comparing all LV models studied.

Fig. 7.

AR index as a function of AR severity and LVDS. The horizontal line represents the cutoff value for survival at 1 yr as determined by Sinning et al. (28). The legend represents the severity of LVDS. The numbers above each individual bar are the aortic systolic/diastolic pressures in mmHg.

Additional SW requirement.

To determine how much more work the LV would have to generate to make up for the lost volume during AR, for each ROA, the SV was increased such that the effective CO was increased back to the original 5.0 l/min. For each of those cases, the SW was calculated using Eq. 2 below:

$$SW={\displaystyle {\oint }_c\hspace{0.17em}PdV}$$ (2)

where P and V are the pressure and volume in the LV model, respectively, and C indicates that the integration was performed over the closed pressure-volume loop.

Figure 8 illustrates the percent additional SW required by the LV to maintain a CO of 5.0 l/min as a function of ROA for each of the LV models. Each individual plot is the percentage difference from its respective baseline (with no ROA). Hence, an additional SW requirement of 10% indicates a 110% energy requirement was necessary, for that specific LV model, to maintain an effective CO of 5.0 l/min. A linear regression analysis was performed with a goodness of fit, R². The slope of the regression describes the fold change in work needed per centimeters squared of ROA. It was found that as LVDS increased, the slope of the linear regression increased. The slopes from Fig. 8 revealed that for the low, medium, and high diastolic stiffness models, the LV would need to work 5.1, 5.5, and 6.6 times harder, respectively, per square centimeter of ROA.

Fig. 8.

Percent additional stroke work required to increase the cardiac output back to 5.0 l/min as a function of AR severity.

DISCUSSION

In this study, we quantitatively evaluated the effects of LVDS on LV performance in the presence of AR. For each of the different levels of LVDS, AR was parametrically varied from none to moderate AR. The AR index was used as a metric to estimate AR tolerance by the LV as it has previously been shown to correlate strongly to high mortality (28).

Before the TAVR procedure is carried out, improved preprocedural data have the potential to determine a patient's ability to tolerate different levels of AR. This becomes increasingly important as TAVR indications expand towards lower risk patients (22). From previous studies, we understand that for patients to tolerate AR, the LV must be able to compensate for the extra volume returning from the aorta (4, 5, 8, 14, 17). In the case of patients with preexisting chronic AS, this compensatory mechanism is diminished (17). It is, therefore, important to estimate the LVDS presurgery to gain an indication of how much AR the LV may tolerate. Furthermore, there is a need to understand the physiology of the LV performance as a function of AR severity and LVDS.

In this study, we observed a decrease in transmitral flow as AR increased and this effect was exacerbated as the LVDS increased. Furthermore, the rate of filling of the LV also decreased with increasing LVDS. This observed phenomenon was due to the rate of equalization of pressures between the left atrium and LV as a function of cardiac cycle diastolic time (1, 20). The increase in the rate of pressure equalization is affected by the LV myocardium's ability to stretch during both passive and active filling (20, 25). As the regurgitant flow from the aorta is forced into the LV, the lack of myocardial diastolic compliance prevents the LV from stretching to accommodate the extra volume, hence raising the LVDP (Figs. 5 and 6) (2). This increase in LVDP is an increase in the preload which the atrial pressure must then overcome in order for adequate LV filling via the mitral valve to occur.

The values of AR index measured in the study ranged from ∼42 to 16. Figure 6 shows an increase in the LVDP as LVDS was increased. In Fig. 7, we clearly observe that higher LVDS decreased the AR index. Sinning et al. (28) correlated this index to survival of TAVR patients with AR and showed statistical evidence that a cutoff value of 25 was optimal for survival one year post-surgery. Based on this value, Fig. 7 shows that for the highest LVDS, even trace AR was sufficient to push the AR index close to the threshold. Every other AR severity case for this LV model fell below this cutoff value. For the low LVDS case, the AR index was above the cutoff until the AR severity was at a moderate level, where it fell just below the threshold. These results emphasize the effectiveness of LVDS as a metric to evaluate the AR tolerance of patients before aortic valve replacement. They indicate that patients with low LVDS could have a higher tolerance to post-intervention AR.

From the definition of the AR index (Eq. 1), either a large numerator (DBP − LVEDP) or a small denominator (SBP) is preferred: both would lead to higher AR index values (28). The reduction of the difference between DBP in the aorta and the LVEDP results from AR, as the pressures in the aorta decreases due to blood leaking into the LV (17). Simultaneously, the LVDP increases as the LV chamber is no longer sealed off from the much greater systemic pressures. This increase in LVDP is further intensified as LVDS increases (19). On the other hand, a lower SBP can be obtained by reducing the systemic resistance (or pressure). However, unlike acute AR, during chronic AR, the LV attempts to remodel such that it is able to push out the lost volume resulting from the backflow of blood. This inherently raises the SBP and, hence, lowers the AR index. It is, therefore, important to note that the AR index is only a complementary parameter to be used to assess AR severity.

Finally, to gain an idea of the extra workload imposed on the LV as a result of AR severity, Fig. 8 depicts the percent additional SW required as a function of ROA for each of the LV models studied. In basic terms, the LV SW is a multiplication of pressure and volume. Again, as the change in volumes across all the LV models was similar, the only parameter which affected the calculated SW was the difference in pressures. The results show that the LV model with the highest LVDS has to work ∼30% harder than the lowest stiffness model in the presence of AR. We speculate that this difference is too great for the LV to compensate for, which could be the reason why mild/moderate AR post-TAVR results in high mortality (11, 13, 21, 24, 29).

Limitations.

While the results of this study have important clinical implications, a few limitations need to be noted. Firstly, the LV models used in this work were isotropic in nature, meaning they were idealized and uniform in terms of LVDS and geometry. Secondly, this study was conducted using an idealized in vitro simulator, which does not contain the multiple disease confounders that exist in patients undergoing the TAVR procedure. However, an in vitro model such as this allows for fully controlled parametric studies to be performed, ensuring that the differences being observed are only a result of the altered parameter. Thirdly, central AR (not paravalvular) was modeled in this study; however, the metrics analyzed resulted from bulk hemodynamic parameters. Therefore, the results observed in this study should hold true for paravalvular AR as well. Finally, although this work focuses on the effect of LVDS and AR on LV performance, it should be noted that other factors, such as residual prosthesis stenosis, CO, LV size, etc., influence how well the patient might tolerate AR post-AS.

Conclusions.

This work has provided, in an in vitro model, evidence that LVDS affects the hemodynamic severity of AR. It was shown that increased LVDS substantially increases the SW requirement to maintain a forward SV of 70 ml/beat in the presence of AR. The results presented suggest that a TAVR patient's LVDS significantly impacts their ability to compensate for post-procedural AR and has the potential to predict TAVR patient outcomes. Within the limitations of this study, these findings suggest that preoperative measurement of LVDS could potentially be used to augment patient-specific assessment of TAVR risk. An investigation of this work using animal models would lend more insight to its generalization towards humans.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

I.O., G.K., and A.P.Y. conception and design of research; I.O. performed experiments; I.O. and V.R. analyzed data; I.O., V.R., P.M., G.K., and A.P.Y. interpreted results of experiments; I.O. and P.M. prepared figures; I.O. drafted manuscript; I.O., V.R., P.M., G.K., and A.P.Y. edited and revised manuscript; I.O., V.R., P.M., G.K., and A.P.Y. approved final version of manuscript.