Introduction

Left ventricular assist device (LVAD) implantation is progressively becoming a viable solution to treat heart failure and promote myocardial remodeling [1]. According to the Society of Thoracic Surgeons (STS)-Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS), 78.1% of patients receive an LVAD as destination therapy, 15.2% as bridge-to-candidacy (BTC), and 6.6% as bridge-to-transplant (BTT). At long-term follow-up, 44.2% of patients are alive on device 5 years following first LVAD implantation [2]. The increasing durability of current generation LVADs, together with an overall higher patient risk profile, highlights the importance of managing specific complications related to these devices [3]. Most commonly reported major adverse events after LVAD implantation include bleeding requiring surgery, gastrointestinal bleeding (GIB), neurological events, pump thrombosis, blood trauma due to excessive forces generated by the mechanical pump, device failure, infections, and right ventricular failure [4, 5]. LVAD exchange (LVADE) represents one of the possible therapeutic strategies for some of these complications. Historical data regarding LVADE reported in 2004 from the REMATCH trial included 29 LVADE in 23 patients (accounting for 33.8% of total LVAD recipients). The study compared the outcomes of patients affected by advanced heart failure treated with optimal medical therapy alone or LVAD implantation with the HeartMate-VE (HM-VE). All LVADE were performed with another HM-VE. The 1-year freedom and 2-year freedom from device replacement were 87% and 37%, respectively [6, 7]. Subsequently, multiple studies have compared results of LVADE in specific settings and defined the impact of improving technologies. The MOMENTUM 3 trial demonstrated a significantly lower 2-year LVADE rate after implantation with the HeartMate 3 (HM3, Abbott, Abbott Park, IL) when compared to the HeartMate 2 (HM2) (2.3% vs. 11.3%, respectively [p < 0.001]) [8]. These data are comparable with other published works [9, 10]. In this review, we discuss the therapeutic strategies when dealing with the indications of contemporary LVADE, decisional and technical aspects, and outcomes.

Main indications for device exchange

The specific indication for LVADE strongly impacts outcomes (Table 1). Depending on which LVAD generation is considered, the main indication for pump exchange is generally device thrombosis, which represents about 2/3 of all procedures [6, 11, 12]. A recent systematic review reported an incidence of device thrombosis between 2 and 11% of all LVAD recipients (less than 0.04 events per patient-year) [13]. Surgical treatment of this complication may be required because of unsatisfactory outcomes and high mortality reported in patients managed conservatively [14]. This is particularly important as increases in prophylactic antithrombotic therapy to mitigate thrombosis have been associated with a higher incidence of hemorrhagic cerebrovascular events and death [15]. A recent systematic review and meta-analysis showed that surgical pump exchange is superior to medical therapy with a higher success rate of pump thrombosis resolution (81.3% vs. 45.4%; p < 0.001), lower mortality rate (16.7% vs. 34.5%; p = 0.013), and lower recurrence rate (11.8% vs. 38.3%; p < 0.001) [16].

Table 1 Summary of indications, definitions, and outcomes for left ventricular assist device exchange [17]
Full size table

Regarding infection with need for LVADE, these patients tend to have worse outcomes than those surgically treated for device malfunction or thrombus [19]. Device infection represents 12–18% of LVADE indications and includes driveline infection (5 to 44% of patients), pump pocket infection (0 to 22% of patients), and refractory sepsis (0 to 33% patients) [13, 20]. It is difficult to define the best treatment strategy in cases of device infection because of the heterogeneity of this scenario (i.e., specific causative microorganism, portion of device affected, clinical context). The specific indications for LVADE in this context have yet to be defined. Nevertheless, a recent systematic review and meta-analysis demonstrated LVADE does not appear to confer an advantage as compared to conservative strategies. There were no significant differences in the overall mortality (exchange 17.6% (4.3–50.6) vs. non-exchange 23.3% (15.8–32.9), p = 0.67) and infection recurrence rates (exchange 26.7% (8.7–58.0) vs. non-exchange 38.6% (15.4–68.5), p = 0.56) [21]. Regardless, it remains important to clarify the role of technical aspects that can influence surgical outcome in device infection. These include surgical technique, grade of debridement, necessity for omentum wrapping, use of antibiotic-impregnated cement and implants, antimicrobial washout solutions, and duration of antibiotic therapy [22]. It has been also demonstrated that patients who underwent exchange after more than 150 days of active infection had worse outcomes than those who underwent exchange earlier [19].

Another non-infrequent indication for exchange is failure of LVAD components other than the pump (e.g., pump controller, battery, and monitor). This complication has been dramatically reduced by technological advancements, with more reliable devices in third-generation LVADs. Only 13 to 15% of device malfunctions are due to pump failure. However, failure of the integrated driveline can necessitate external repair but also LVADE if the involved portion of the driveline is too close to the skin exit site or at its junction with the LVAD [23].

Timing of LVAD exchange

In the majority of cases, LVADE must be done urgently while managing the patient with medical therapy in the interim. For most cases of device thrombosis, the patient retains partial LVAD output until the device is substituted [19]. Comorbid conditions must be optimized before LVADE to mitigate complications frequently associated with LVADE. Preoperative considerations and studies before LVADE are fundamental to successful outcome that have been reported by Adamson and colleagues [24]. In general, these steps include accurate diagnosis of LVAD failure, assessment of infection (fluorodeoxyglucose (FDG)-positron emission tomography (PET)/computerized tomography (CT) scan [25]), determination of native cardiac function and presence of valvular disease (e.g., echocardiogram, pulmonary artery catheterization), assessment of chest anatomy and presence of existing adhesions (e.g., chest CT scan), choice of a replacement LVAD, and consideration of a new driveline exit site [26].

From a clinical point of view, the preoperative steps to be considered are medical stabilization of the patient (e.g., vasopressors or inotropes and antibiotics if necessary), need for anticoagulation therapy, discontinuation of long-acting anticoagulant and antiplatelet therapy, and use of continuous heparin infusion, when necessary. According to a recent expert review, preoperative medical management should be targeted to the most dangerous complications of LVADE: discontinuation of long-acting agents that may suppress the sympathetic nervous system against the risk of vasoplegia (e.g., beta-blockers, angiotensin receptor blockers, neprilysin inhibitor/angiotensin-converting enzyme inhibitor); optimization of right ventricular function with inotropes, diuretics, or temporary mechanical support in the event of right heart failure; and optimization of coagulative function to mitigate bleeding risk [27]. The importance of CT scan with three-dimensional (3D) reconstruction during preoperative planning has also been highlighted in cases of LVADE for determining the optimal route of surgical access [27].

Technical aspects of LVAD exchange

The surgical approach to device exchange may vary. Specific surgical strategies for LVADE have been previously described [3, 28]. Table 2 summarizes the main factors determining surgical approach for LVADE.

Table 2 Main factors to plan the surgical approach for left ventricular assist device exchange
Full size table

Surgical access can range from traditional full-sternotomy or redo-sternotomy to an alternative minimally invasive technique. Novel surgical approaches have been developed to minimize complications related to re-do surgery in cases of LVADE. These are helpful in reducing surgical trauma, risk of blood loss, arrhythmic complications, and in decreasing intensive care unit and overall in-hospital length of stay. Additionally, the risk of right-sided heart failure can be reduced with less invasive techniques as the right ventricle remains in its natural position [3]. Implantation with less invasive techniques is becoming the gold standard due to these potential benefits combined with the non-inferiority of their surgical outcomes.

It is important to define the extent of the dysfunction affecting the device as it may be necessary to change only a portion of its components. The traditional approach through redo-sternotomy (as sternotomy is currently the most common access used during the first implantation) implies greater complexity due to presence of adhesions and risk of major bleeding. However, this technique allows the best surgical exposure with complete access to the entire outflow graft, allowing for revision of the inflow cannula angle relative to the heart, if needed. The use of a sternal-sparing less invasive approach through a left lateral thoracotomy with partial rib resection may also be feasible [29]. However, this approach implies a smaller surgical field with more difficult access to anatomic structures. This is the purpose for the presence of a longer remnant of the HVAD (Medtronic, Framingham, MA) outflow graft after LVADE: graft-to-graft anastomosis is typically performed over the acute margin of the right ventricle just behind the sternum []. It is also possible to combine this surgical access with a right anterior thoracotomy at the third intercostal space for a direct anastomosis on the aorta. For specific devices, LVADE has also been performed with subcostal access, as is the case for the HM2, with a non-muscle dividing approach being associated with lower pain burden [30]. This approach is reasonable when there is no inflow and outflow involvement (such as obstruction) and there is no need for any concomitant cardiac procedure, since the pump is located in the abdomen. It is a less invasive procedure requiring shorter operative time, shorter cardiopulmonary bypass time, fewer blood transfusions, shorter intensive care unit stay, and less postoperative complications than re-sternotomy [31]. This surgical approach may increase the risk of postoperative device infections, although this has been mainly observed when an extended one-J-incision (incision extends from the xiphoid process to the left midclavicular line, with transection of the rectus muscles, fascia, and ribs) has been performed [32].

Regardless of the specific technique, a few key technical issues should be considered. First, it is of unique importance to verify the correct positioning and angle of the inflow cannula in order to adequately position the new sewing ring []. Second, obliteration of the dead space surrounding the new LVAD pump should be considered using soft tissue coverage of the pump pocket with a bulky vascularized pedicled flap of the greater omentum [20].

Regarding circulatory support during surgery, LVADE can be both performed on-pump and off-pump. When cardiopulmonary bypass is established, venous and arterial cannulas of the extracorporeal circulation are usually placed in the common femoral artery and vein. The main advantage of performing the procedure on-pump is allowing for careful inspection of the left ventricle chamber for thromboembolic material or remaining trabeculae tissue to mitigate recurrent thrombosis or stroke risk [33]. When the whole pump is exchanged, the new driveline should be tunneled to the opposite site of the former driveline exit site in an effort to reduce the risk of infective complications [33].

Postoperative care in patients undergoing LVADE exchange is similar to that of primary LVAD implantation. Special attention to the risk of bleeding due to adhesions or extensive surgery must be considered. This can be managed by strict monitoring and diligent management of anticoagulation, especially in patients who previously experienced device thrombosis [29].

A recent expert review summarizes the best available evidence to consider during LVADE from HVAD to HM3 []. It is important to underline specific technical issues that must be considered when performing LVADE with those devices. This is particularly important regarding the possibility to maintain the sewing ring and the outflow graft of the previous pump. Multiple solutions have been developed to assist in avoiding the traumatic complete removal of the device. The HVAD inflow cannula has a larger diameter (20.6 mm) when compared with the HM3 (20.5 mm). Currently, the best option is complete removal of the existing sewing ring. The use of a rubber seal to obtain hemostasis at the inflow connection has been described as an alternative, though the long-term consequences remain unknown. The possibility of sewing the HM3 apical connector over the existing HVAD sewing ring has also been described []. A drawback of this technique is that the tip of the inflow cannula will be less inside the LV reducing the LV unloading and consequently the LVAD flow.

Surgical adaptation of the outflow graft is possible and must be considered since the HVAD outflow prosthesis diameter is smaller than the outflow of the HM3 (10 mm vs. 14 mm, respectively). The anastomosis between the two outflow grafts must be done consequently. The optimal solution is still considered to be the exchange of the entire device, including the outflow graft, but in vivo and in vitro studies suggest that outcomes are not influenced by the slightly higher resistance caused by the lower diameter of HVAD outflow graft attached to the aorta [32, 34]. The safety of leaving portions of the infected LVAD in place has not been described. Ultimately, it is important to note the unknown clinical consequences of this procedure on hemocompatibility risks, battery runtime, and pump performance [].

Results

LVADE is associated with variable operative mortality, ranging from 7 to 10% [6, 10]. Survival rate following LVADE has been found to be non-inferior to conservative treatment group (93% vs. 76%, p = 0.15) [35]. In a recent study, postoperative mortality at 30 days was comparable for patients undergoing LVADE and primary implantation [19]. Causes of death are not particularly device-specific and, therefore, are usually related to patient medical history, management, and etiology of device dysfunction. While the rates of each cause of death are similar to those expected with primary implantation, patients requiring exchange may have increased risk of postoperative coagulopathy, with higher incidences of cerebro-vascular accident and pump thrombosis [19]. Among the various complications that can follow LVADE, the most prevalent appears to be right heart failure. In a recent study conducted by Austin et al., right heart failure occurred in 33% of the patients, with no difference in device technology [6]. A recent observational study showed that among candidates awaiting heart transplantation on a durable LVAD, undergoing pump exchange doubled the risk of 1-year mortality [36].

Table 3 summarizes large series published focusing on patients undergoing LVADE. From 2004 to 2021, a total of 19 manuscripts were published, ranging from first- to third-generation devices. A total of 935 patients were included. Main indications for exchange were thrombosis (56%), device malfunction (28.8%), device infection (10.9%), and outflow graft obstruction or inflow graft malposition (0.3%). Exchange using the same technology occurred in 441 patients (60.9%) while using a different technology occurred in 283 (39.1%). Surgical approach included redo-sternotomy (53.1%), subcostal incision (40.2%), thoracotomy (4.5%), left anterior thoracotomy with subcostal approach (0.9%), abdominal (0.4%), or first sternotomy (0.20%). Complications included acute kidney injury (10.3%), infection (9.6%), right heart failure requiring right ventricular assist device (RVAD) (7.2%), stroke (6.5%), bleeding (2.0%), refractory ventricular tachycardia (0.7%), GIB (0.6%), device failure (0.6%), and heart failure requiring postoperative extracorporeal membrane oxygenation (ECMO) (0.6%). Overall mortality was 8.8%.

Table 3 Studies reporting patients undergoing Left ventricular assist device exchange
Full size table

Table 4 summarizes reports of LVADE cases series (< 10 patients). From 2012 to 2019, 6 studies with a total of 37 patients were included. Main indications for device exchange were thrombosis (51.3%), device infection (24.3%), driveline-related device malfunction (18.9%), and non-driveline-related device malfunction (5.4%). Exchange using a different technology was chosen in 25 patients (67.6%), while using the same technology occurred in 12 patients (32.4%). Surgical approaches included redo-sternotomy (29.4%), subcostal incision (26.5%), and thoracotomy (44.1%). Overall mortality was 8.1% (3/37).

Table 4 Case series reporting patients undergoing left ventricular assist device exchange
Full size table

Discussion

LVADE can be performed safely and with low surgical mortality using the same generation pump or exchanging in favor of a more recent device [6, 10, 11]. Although there are currently no guidelines as to the best strategy for LVADE, the choice of new device should be tailored to the individual patient’s risk profile, considering the patient’s unique factors and comorbidities [6]. This is particularly prudent when it is not possible to define the specific etiology prompting LVADE. Patients that develop complications despite optimal anticoagulation, antiplatelet therapy, appropriate pump speed and flow, and no other signs of infection are prone to develop the same complication after LVADE with a pump of the same generation [47]. It is also important to consider the specific risk profile of every device. Data suggesting improved stroke rate outcomes with the HeartMate 3 help inform device exchange choice in cohorts of patients at an increased risk of cerebrovascular events. Due to the improved hemocompatibility profile, LVAD upgrade to HM3 is attractive for patients with recurrent device thromboses [33]. It should be noted that data from the MOMENTUM 3 trial suggests improved survival benefit with HM3 compared to HVAD for LVADE regardless of primary implantation or device exchange. There are also some complications that have similar incidences in different devices, including right heart failure. In this case, the most important consideration is optimizing perioperative patient management to reduce this risk [6]. This discussion is now strongly influenced by the recent recall of the HVAD from the Food and Drug Administration (FDA) on June 3, 2021, which made the HM3 the only commercially available device at the moment [37]. The reasons for this were both delay or failure to restart after elective or accidental discontinuation of pump operation and the higher reported risk of stroke and all-cause mortality in HVAD recipients [27]. In particular, LVADE to the HM3 compared with exchange to an HVAD demonstrated superior late survival with the former, but using this strategy in elective, uncomplicated cases is not currently supported by enough evidence. The risk of death due to LVADE likely exceeds the risk of death remaining on a normally functioning HVAD device [11].