Patent foramen ovale can be difficult to diagnose with intraoperative transesophageal echo because left atrial pressures are usually greater than right atrial pressures. Even a bubble Doppler will fail to elucidate the potentially problematic patent foramen ovale. In the post-LVAD insertion patient, the left atrial pressures are lower than the right atrial pressures, which fosters a right-to-left shunt and produces a diagnostic triad of the missed patent foramen ovale: hypoxia, which is unresponsive to increased oxygen concentrations, a clear chest X-ray, and a bubble Doppler that shows a right-to-left shunt. Patent foramen ovale has certainly been closed during the postoperative period using both Amplatzer and BLANK devices, but is certainly much more complicated and potentially more of a thrombotic liability than simply opening the atrial septum to close the potentially problematic foramen.

We use a simple diagnostic test in the operating room prior to the insertion of LVAD. By compressing the pulmonary artery, the right atrial pressure increases and the left atrial pressure decreases and the atrial septum shifts to the left. Agitated saline, bubble Doppler, injected into the right atrium at that time will clearly demonstrate the presence of a potentially dangerous patent foramen. The incidence of patent foramen ovale in the general population is approximately 28%. If the atrium is inspected in every case, that is the approximate percentage that will be closed. Interestingly, in a reported analysis of multi-institutional results, only a 6% incidence of PFO closure is reported [5]. In our experience, using the pulmonary artery compression technique, we were able to reduce our patent foramen ovale closure significantly – down to approximately 20%, eliminating unnecessary pump time and incisions [4].

Mitral valve regurgitation is common in heart failure patients, but the need to correct mitral regurgitation in patients receiving LVADs remains controversial [6]. Insight comes from small, often single-centered, studies. Results of ongoing controlled studies are not yet available.

Most MR associated with cardiomyopathy is a consequence of LV dilation with resultant chordae tethering and mitral annular dilation. With LVAD decompression, LV dimensions decrease and may allow mitral leaflet coaptation and make MR insignificant during device support. Morgan [7] found that continuous-flow LVAD implantation significantly decreased the severity of MR (moderate-severe) from 76% preoperatively to 8% at 1 and 6 months postoperatively. However, there is still potential risk for residual MR when device support cannot be optimized to provide low, left-sided filling pressures, such as when there is suction of the interventricular septum over the inflow cannula at higher speeds or the need for intermittent AV opening [8]. Right ventricular failure is an important cause of early and late mortality and morbidity after LVAD implantation. Mitral regurgitation negatively affects RV function, LVAD filling, and possibly survival [9].

There are many proposed theoretical advantages of having a competent mitral valve in LVAD patients. In the early postoperative period when fluid balance is dynamic and often coupled with elevated pulmonary resistance, any contribution to the right ventricular afterload can result in increasing right heart failure. Early persistent mitral valve insufficiency is more likely to cause increases in afterload if the continuous-flow pumps are set to allow the ventricle to produce pulsatile pressure and guarantee adequate ventricular volumes for the LVAD to aspirate. The pressure can in turn be transmitted to the left atrium and reflected as an increase in the total pulmonary vascular resistance [10].

Experimentally, a competent valve can provide an increased pulsatility index [11]. LVAD filling and LVAD flow may affect pump thrombosis rates. In the presence of aortic insufficiency, a competent mitral valve will protect the pulmonary bed. If LVAD flows are kept low to permit aortic valve opening (in hopes of maintaining aortic valve integrity), pulsatility can again be transmitted retrograde to create a substrate for right ventricular failure. Finally, a competent mitral valve will facilitate weaning a patient with ventricular recovery.

Our bias has been to repair mitral valve regurgitation in any patient with greater than 2+ mitral regurgitation or in any patient with structural valvular problems (including ruptured chordae) or in functional regurgitation due to tethering of the leaflets and/or annular dilatation. Our technique for repairing the mitral valve is to perform the valve repair, almost always only requiring a posterior flexible annuloplasty in the beating heart, following the creation of an apical core for the LVAD. A vent placed through the apex and retractors on the annulus allows the heart to beat with no chance of embolization, excellent visibility, and only one case where aortic insufficiency necessitated the use of cardioplegia. A limited experience with transapical repair has been reported [12], and pre-existing mitral prostheses have been successfully left in place [13].

In our early experience from 1991 to 2008, we repaired approximately 17% of our mitral valves, and today our repair rate has risen to approximately 55%. As seen in the graph below (◘ Fig. 33.4), a retrospective analysis of 221 HeartMate II patients, 41% of whom received mitral valve repair or replacement, the early survival was unchanged, and there is a tendency for late survival to separate with favoring the mitral valve group [14].

Some patients supported by LVADs will continue to have or develop mitral regurgitation after LVAD [15]. For us and others, mitral valve annuloplasty associated with an LVAD does not increase the operative mortality, reduces long-term mitral regurgitation, and may decrease immediate and long-term RV failure rates, possibly providing a long-term advantage [16].

Our first experience with aortic insufficiency at the time of LVAD insertion was in 1993 [17]. This was unexpected pre-op and was treated with a bioprosthesis replacement of the aortic valve. At the time of transplantation, the prosthesis was occluded, and there was red thrombus below the valve, demonstrating the risk of embolism. The 2013 ISHLT guidelines for mechanical support state that more than mild aortic insufficiency should prompt consideration for surgical intervention during device implantation. In the operating room, we assess the degree of aortic insufficiency preoperatively, then at intervals following institution of cardiopulmonary bypass, and finally during LVAD support. During each interval, aortic insufficiency can become increasingly more prominent. Initial echoes examining patients for aortic insufficiency may fail to demonstrate a significant regurgitant leak because of high end-diastolic pressures and low aortic diastolic pressures and the consequent low gradient jet. Cardiopulmonary bypass will reduce the left ventricular end-diastolic pressure significantly and increase the jet. With a functioning left ventricular assist device in place, aortic flows generally increase, and end-diastolic pressures decrease, making the gradient even more visible on an echo.

Any leak more than trace for us with the current technology prompts us to repair the valve. As seen in the table above, the incidence of aortic insufficiency at 12 months in LVAD patients ranges between 25 and 70% (◘ Fig. 33.5). Although this is often not clinically significant, especially in the bridge to transplant population, these estimates are over a relatively short period of time. Aortic insufficiency tends to be progressive and is usually associated with older patients who have a natural tendency to develop valvular insufficiency over time.

Other risk factors are LVAD support for over 1 year, less aortic valve opening, preoperative aortic root dilatation, increased aortic diameters, and DT status [18]. Some have shown a potential advantage of pulsatility which occurs with left ventricular reverse remodeling which permits intermittent opening of the aortic valve [19]. Despite the fact that periodic aortic valve opening appears to reduce the incidence of insufficiency in LVAD patients, late de novo aortic insufficiency is known to have occurred (unsure incidence) in patients supported by the Jarvik 2000 FlowMaker device which allows opening of the native valve for 8 s every 64 s. The decision to repair or close an aortic valve should be made at the time of LVAD implantation, as severe AI may lead to heart failure and is not likely to be corrected without repeat surgery. Catheter interventions to correct late acquired aortic insufficiency have been used successfully, but not in a predictable way [20]. Amplatz catheters have embolized, as have Sapien valves [21] and Core valves. Although aortic insufficiency might be eliminated, it is unknown whether the architecture on the aortic side of the device will create the long-term liability of a thrombogenic profile and their durability is unknown in this situation. In our own examination of 221 patients with 17% aortic closure rate, there is no difference in early mortality between the groups and the survival of the two (closure and non-closure groups) (◘ Figs. 33.6 and 33.7).

This is in contrast to the papers by John Robinson who used the INTERMACS database to illustrate a higher mortality in the AV closure group compared to repair and replacement. It is noteworthy that the observed mortality is in the early perioperative period, which brings management styles into question. Our own survival curve is superimposed over the no AV procedure with no early liability (◘ Fig. 33.8).

A variety of aortic valve closure techniques have been used over time. Initial closure of the aortic valve leaflets at the nodule of Aranzio was used occasionally with some success, but was abandoned in the early days because of the periodic ejection of the heart, which would cause the valve to tear. The technique was modified by Soon Park who has had success using central closure with felt stitches with the intent of maintaining an outflow option for the ventricle in cases of LVAD failure. The procedure is especially recommended for patients expected to be supported for long periods of time, and based on Fukahara’s [22] findings, consideration of CAVC as a prophylactic repair in DT, elderly, and female patients may be appropriate, even if they have mild-degree native AI with no increase in early mortality but with a possible late failure rate of 8% in a small series. Central stitches with felt can create a fibrous reaction which ultimately may seal the valve aperture, thereby eliminating the ventricular escape aperture.

This was further modified by Morgan [7] using multiple sutures. We have personally observed recurrent aortic insufficiency peripheral to the central closure following a “Park” stitch placed in a patient with a HeartMate I device. We subsequently successfully closed the valve surgically using our own evolved technique, described by Adamson [23], which incorporates three felt strips rather than individual pledges to expedite closure of the valve, as seen in the figure below (◘ Fig. 33.9).