Development of Aortic Insufficiency in Patients Supported With Continuous Flow Left Ventricular Assist Devices

Abstract

Development of aortic insufficiency (AI) in patients supported with continuous flow left ventricular assist devices (LVAD) can adversely affect pump performance. In this study, we examined the incidence of new AI after LVAD implant at our institution. Pre- and postoperative echocardiograms of 66 patients who received HeartMate II or Heartware LVAD at our institution since June 2008 were reviewed for presence of new AI. Median LVAD support duration was 221 days. New AI developed in 6 patients (9.5%) after a median time of 374.5 days of support. There were no cases of severe or symptomatic AI. There was no significant difference between the AI incidence between HeartMate II and Heartware recipients. For patients who remained on LVAD support at 6 and 12 months, freedom from AI was 100% and 68.4%, respectively. Age, destination therapy status, and duration of support were predictors of new AI after LVAD implant. In conclusion, AI develops frequently during long-term support with continuous flow LVADs, particularly in those supported for longer than 6 months. As we move to the era of long-term LVAD support and destination therapy, further studies with longer follow-ups are required to determine the progression and clinical significance of AI in these patients.

Continuous flow left ventricular assist devices (LVAD) are used to treat patients with end-stage left ventricular failure as bridge to transplantation and increasingly as destination therapy.^1,2 These new generation devices have the advantage of smaller size, increased device durability, and decreased incidence of adverse events.³ However, the physiologic impact of long-term continuous, nonpulsatile flow is not fully understood. Specifically, the behavior of the native aortic valve in this setting is of particular interest. It is obvious that aortic insufficiency (AI) in the context of a continuous flow LVAD results in a circulatory loop where a portion of pump output returns through the incompetent aortic valve back into the pump, thus reducing effective forward flow.⁴ Presence of significant AI at the time of LVAD implant is regarded as an indication for a concomitant procedure on the aortic valve to render it competent.⁵ Similarly, development of significant de novo AI in LVAD supported patients may be an indication for subsequent intervention to improve pump efficiency. In this study, we sought to determine the prevalence and progression of new AI in patients implanted with continuous flow LVADs at our institution. We also sought to determine the risk factors for developing AI continuous-flow LVAD supported patients.

Methods

Echocardiograms (n = 195) of 66 consecutive patients who received HeartMate II (Thoratec, Pleasanton, CA) (HMII, n = 58) or Heartware (HeartWare Inc, Framingham, MA) (HW, n = 8) LVAD from June 2008 to October 2010 at our institution were retrospectively reviewed. Studies obtained within 1 month preoperatively were considered as baseline. Postoperative echocardiograms were reviewed at 3 monthly intervals. The degree of AI was graded as 0 = none, 1+ = mild, 2+ = mild-to-moderate, 3+ = moderate and 4+ = severe. AI was regarded significant if mild-to-moderate (2+) or greater. Patients with concomitant or previous aortic valve replacement (n = 2) or significant baseline AI (n = 1) were excluded from analysis. In addition to AI severity, aortic root diameter at baseline and opening of the aortic valve after implant were recorded. Baseline characteristics of the patients included in the study were obtained by systematic chart review.

Statistical Analysis

Categorical variables are presented as frequencies and percentages and were analyzed using χ2 test. Continuous variables are presented as means with standard deviations and were analyzed using Student’s I t-tests. Kaplan-Meier analysis was performed to demonstrate freedom from AI. In this analysis, censoring events were transplantation and death.

Results

Demographics

Baseline characteristics of the cohort are depicted in Table 1. There were 55 patients with the HMII and eight patients with the HW device. There were 36 bridge-to-transplant and 27 destination therapy patients. Forty-seven (74.6%) of the patients were in Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) category 1 and 2 at the time of implant.

Table 1: Baseline Characteristics

Development and Progression of AI

Mean duration of LVAD support was 314.5 ± 288.7 days (median 221.5 days, range 1–1142 days). For HMII patients, mean duration of support was 336.5 ± 299.5 days (median 265.5 days, range 1–1142 days). For HW patients, mean duration of support was 158.25 ± 107.1 days (median 134.5 days, range 44–390 days). New AI developed in 6 of 63 LVAD supported patients (9.5%). The median time to AI development was 374 days (range 152–582 days). Kaplan-Meier analysis (Figure 1) demonstrated that for patients who remained on LVAD support at 6 and 12 months, freedom from AI was 100% and 68.4%, respectively (numbers at risk of 38 and 19, respectively). The AI grade increased progressively with time, although there were no cases of severe (>3+) AI (Figure 2A). In those with new AI, left ventricular systolic and diastolic dimensions remained stable throughout the observation period (Figure 2, 2C and 2D) and this was associated with absence of symptoms attributable to AI.

Figure 1: Kaplan–Meier analysis of freedom from aortic insufficiency (AI) after left ventricular assist device implant.

Figure 2: A: Progression of aortic insufficiency (AI) grade in patients with new AI after continuous flow left ventricular assist device (LVAD) implant. B: Changes in aortic sinus diameter in patients with new AI after continuous flow LVAD implant. C: Changes in left ventricular (LV) systolic diameter in patients with new AI after continuous flow LVAD implant. D: Changes in LV diastolic diameter in patients with new AI after continuous flow LVAD implant. Numbers 1 to 6 refer to the patients with new AI.

Risk Factors for Development of AI

The patients who developed AI were on average a decade older than those who did not (Table 2). All cases of AI were in those whose indication for LVAD implant was destination therapy. The duration of support in those who developed AI was also significantly longer (770 ± 262 vs. 293 ± 271 days, p = 0.03). None of the patients who developed AI had a preimplant history of hypertension, whereas 52.6% of patients who did not develop AI had history of hypertension (p = 0.045). Even though all cases of new AI occurred in those patients receiving a HMII device, there was no statistically significant difference between the AI prevalence between HMII and HW recipients (6/55 (10.9%) vs. 0/8 (0%), p = 0.32). AI occurred with similar frequency in those LVAD supported patients in whom the aortic valve opened versus those that did not (2/36 vs. 4/30, p = 0.27). There was no difference between aortic root diameters at baseline between those who developed AI and those who did not (3.5 ± 0.41 vs. 3.38 ± 0.41, p = 0.53) and aortic root diameter remained stable with progression of AI (Figure 2B).

Table 2: Risk Factors for Development of AI

Disposition

One of the LVAD-supported patients with new AI died of unrelated cause after 805 days of support and 224 days after discovery of AI. The remaining five LVAD supported patients with new AI remain alive and asymptomatic after mean of 331.6 ± 224.9 days since discovery of AI (median 262 days, range 85–652 days).

Discussion

In this study, we investigated the prevalence of de novo AI in patients treated with continuous flow LVADs. We found that AI develops frequently during long-term support with these devices, particularly in those supported for longer than 6 months. In fact, the prevalence of new AI in patients who remain on LVAD support for more than 1 year was 31.6%, a finding which coincides with other published series. Pak et al.6 reported that AI developed in a quarter of patients supported by HMII for more than 1 year. Similarly, Cowger et al. reported that, in their series, freedom from moderate or worse AI at 12 months after LVAD implant was 74%.⁷

The high incidence of new AI in LVAD supported patients is significant because these devices are increasingly used for destination therapy with anticipated support duration of several years.² The Randomized Evaluation of Mechanical Assistance in Treatment of Chronic Heart Failure (REMATCH) demonstrated survival advantage of LVAD support over maximal medical therapy in patients with end-stage heart failure.⁸ However, early device malfunction and failure limited the application of LVADs for long-term support and destination therapy. The advent of new generation pumps has resolved many of the issues resulting in early device failure and has made destination therapy a realistic option for patients with end-stage heart failure.⁹

The new generation rotary pumps produce a continuous nonpulsatile flow. The precise physiologic response to a nonpulsatile circulation is not fully understood but this phenomenon may be relevant in pathogenesis of aortic valve incompetence. It is conceivable that loss of pulsatility and presence of sustained high pressure in the aortic root throughout the cardiac cycle may result in commissural fusion and aortic root dilatation. There is evidence from previous studies that commissural fusion of the native aortic valve leaflets^10,11 and aortic root dilatation^6,7 are associated with development of AI during continuous flow LVAD support. It is of note that in our study, we did not demonstrate an association between aortic root diameter at baseline and follow up with development of AI. Also and in contrast to other reported series,^6,7 we did not find increased new AI in those patients whose aortic valve did not open during LVAD support.

It is obvious that significant AI in the context these pumps can adversely affect pump performance and if progressive can ultimately result in recurrent heart failure symptoms. It is of note that all patients in our cohort who developed AI were in the destination therapy group with significantly longer support duration and were accordingly older than AI-free patients. The other preoperative variable that appeared to predict postimplant AI was a negative history for treated hypertension, although there was no significant difference between immediate preoperative blood pressures in the two groups.

All cases of new AI occurred in patients receiving the HMII device as opposed to the HW device. However, this difference was not statistically significant. Moreover, given that most new AI develops after the first year of support, we would not anticipate the same frequency of AI in the HW group considering the shorter duration of support on this device in our cohort.

A potential limitation of this study is the difficulty of quantifying AI severity in the context of a continuous flow pump. Traditional means of assessing AI severity such as pressure half-time measurement may not be applicable given the continuous flow of regurgitation.

In conclusion, our data demonstrates that new AI develops frequently during long-term support with continuous flow LVADs, particularly in those supported for longer than 12 months. As we move to the era of long-term LVAD support and destination therapy, further studies with longer follow-ups are required to determine the progression and clinical significance of AI in these patients. Nevertheless, with increasing application of continuous flow pumps for destination therapy, a heightened degree of vigilance to detect and address AI at the time of device implant is advisable.