Medical complications in patients with LVAD devices

Abbreviations

AI:  aortic insufficiency

AUS: anaemia of undetermined source of bleeding

AvWS: acquired von Willebrand syndrome

BTT: bridge to heart transplant

CF-LVAD: continuous flow left ventricular assist device

CTA: computed tomography angiography

DLI: driveline infection

DT: destination therapy

EPPY: event per patient year

GIB: gastrointestinal bleeding

INR: international normalised ratio

INTERMACS: Interagency Registry for Mechanically Assisted Circulatory Support

LDH: lactate dehydrogenase

LVAD: left ventricular assist device

PT: pump thrombosis

RVF: right ventricular failure

VCE: video capsule endoscopy

vWF: von Willebrand factor

Introduction

The use of a left ventricular assist device (LVAD), as a bridge to heart transplant (BTT), as a destination therapy (DT), or as a “bridge to recovery”, has become an important option for the treatment of patients with advanced heart failure refractory to medical therapy. The two leading LVAD devices currently on the market are the axial continuous-flow HeartMate II (Thoratec Corp., Pleasanton, CA, USA) and the centrifugal continuous-flow HeartWare (HeartWare International, Inc., Framingham, MA, USA). In Europe, the HeartMate II was authorised for BTT and for DT in November 2005, while the HeartWare received approval for use in patients as a BTT in 2009 and as a DT in 2012. In the USA, the HeartMate II has been approved for BTT since 2008 and as a DT since 2010, while the HeartWare has been approved as a BTT since 2010 and is still awaiting approval for DT [1].

The use of LVADs as a DT has increased in past years and currently represents approximately 46% of all LVAD placements; so people are on LVAD support for longer durations. The median time of support in patients awaiting heart transplantation is around 300 days (IQR, 147-537 days). The number of implanted LVADs is now approaching the number of heart transplantations performed in the USA (1]. Continuous-flow LVADs (CF-LVAD) have accounted for virtually 100% of devices in patients receiving DT since 2010 and actuarial survival is currently 81% at one year, 70% at two years, 60% at three years, and 48% at four years [1]. The transition from pulsatile to CF-LVADs has been associated with a significant decline in overall rates of adverse events, improved durability and much better long-term survival for both the BTT and DT indications. However, with the increasing usage of LVADs and longer survival, a substantial number of patients experience adverse events and complications. LVAD-related complications can occur in up to 60% of patients by six months post-implantation, and, by two years, 80% of patients experience at least one adverse event [2]. Unplanned hospital readmissions are common. It has been shown that patients were readmitted, on average, 2.2 times during their 11-month median follow-up time. The median time to readmission was 35 days after an initial discharge [3].

Major, non-surgical, adverse events and complications with the LVAD include bleeding complications, device thrombosis, ischaemic and haemorrhagic strokes, renal impairment, multi-organ failure and infections, which have been the primary causes of death in some series. With regard to timing, complications after LVAD placement can be early (less than 30 days after placement], or late (occurring more than 30 days after placement).

Bleeding complications

Bleeding is the most common adverse event after LVAD implantation. Patients with LVADs require antiplatelet and anticoagulant therapy, which predisposes them to bleeding complications. Bleeding that occurs in the first 14 days after the implantation is mostly related to surgery. Causes of later bleeding include the development of arteriovenous malformations, hepatic dysfunction from post-implant right ventricular failure and acquired von Willebrand syndrome. Non-surgical, early bleeding (within 30 days after implant), occurs in anywhere from 20 to 40% of the patients. The incidence of haemorrhagic events within six months of discharge is 13% [4]. Determining the potential causes and risk factors for bleeding is essential to improve overall outcomes as well as the quality of life of LVAD recipients.

Gastrointestinal (GI) bleeding occurs at a median of 33 days from surgery (range: 1 to 530 days), with the greatest risk within the first post-operative month. It is the most common cause of 30-day readmission [3]. The cumulative risk of GI bleeding for patients receiving the HeartMate II and the HeartWare is 21%, 27% and 31%, at one, three, and five years, respectively. Although previous reports found that the upper GI tract is the most common site of GI bleeding in LVAD recipients [5], obscure GI bleeding remains a frustrating condition and is one that is frequently encountered in the LVAD population. A recent, small, retrospective study proved video capsule endoscopy (VCE) to be a safe and feasible option in clarifying an obscure GI bleeding. The small intestine was the most common site of positive VCE findings (75%). The predominant positive VCE findings were small intestinal bleeding with no source or lesion identified (50%) and small intestinal angiodysplasias (33.3%). There was no electromagnetic interference of either VCE or LVAD and no capsule retention identified in any patients. The diagnostic yield of the study was 40%. However, it was performed, on average, 6.2 days after admission - after coagulopathy was corrected and after other endoscopic procedures failed to identify the cause of obscure GI bleeding [6].

Factors contributing to GI bleeding may be related to the increased shear stress and increased intraluminal pressure leading to the development of angiodysplasia. Another possible explanation for the high GI tract bleeding rate among CF-LVAD recipients is an acquired von Willebrand syndrome (AvWS) secondary to mechanical damage on red blood cells within the pump. It is believed that the rotating components of CF-LVADs may cause von Willebrand factor (vWF) deformity and subsequent cleavage of the high-molecular-weight multimers into smaller ones that are cleared from the bloodstream, resulting in the loss or reduction of the large vWF multimers that are essential for promoting platelets [7]. Recent studies found that all patients developed the typical laboratory findings of an AvWS after LVAD implantation, but not all of them had bleeding complications [8, 9]. These findings suggest that AvWS alone is not sufficient for the development of bleeding complications after LVAD implantation.

Another important site of bleeding is bleeding into the central nervous system. Central nervous system bleeds occur relatively late. In the HeartMate II, DT trial, there was an 11% risk of haemorrhagic stroke in the first two years after LVAD placement, with haemorrhagic stroke being the leading cause of death among patients with a CF-LVAD [10]. In a pooled analysis of 734 patients, a significantly higher incidence of haemorrhagic stroke was found in patients treated with the HeartWare in comparison to the HeartMate II [5]. In a recent single-centre retrospective review of 114 patients with the HeartMate II, 5% of patients experienced an intracranial haemorrhage [11]. Proportionally, more patients on aspirin 325 mg had a haemorrhagic intracranial event in comparison with those on aspirin 81 mg plus dipyridamole or aspirin 81 mg. High-dose aspirin in HeartMate II patients treated concomitantly with warfarin was associated with an increased hazard of bleeding but did not reduce thrombotic events [11].

An important reason for early hospital readmission, post-implant discharge, is anaemia of an undetermined source of bleeding (AUS) that results in any red blood cell transfusion. AUS importantly accounts for 20% of all bleeding events and for a significant proportion of transfusions when a definitive source of blood loss is not apparent. AUS may represent thoracic bleeding not requiring reoperation or occult GI bleeding with no endoscopic source, which highlights the challenges of evaluating GI bleeding in LVAD patients [12]. Capturing these data is extremely important, given the risks of red cell transfusion. The data show that using current definitions of the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) alone underestimates the bleeding risk in LVAD patients.

Efforts to reduce the bleeding rate are imperative and urgently needed. Strategies for lowering the incidence and severity of bleeding complications include: lowering the international normalised ratio (INR) goals, reducing the use of antiplatelet agents, and altering pump speed to allow pulsatile flow. A recent study showed that reduced pulsatility in CF-LVADs led to an increase in bleeding complications. Those patients in the low pulsatility index group had a hazard ratio of 4.06 (p=0.04) when compared to the high pulsatility group [13].

The ideal approach in a patient at an increased bleeding risk remains elusive and will always depend heavily on a combination of patient-related and device-related factors and the weighing of risks of bleeding risk against thrombotic risks. A patient’s clinical status will often necessitate a temporary change in INR target, often by decreasing the target INR or temporarily holding anticoagulation to stop significant or even life-threatening bleeding. Whereas the trial by the HeartMate II investigators specified an INR goal of 2.0 to 3.0, the relatively greater burden of haemorrhagic events in this trial (and the HeartMate II BTT trial) fostered the conclusion that a target INR of 1.5 to 2.5 (in addition to aspirin therapy) might be safer in patients at increased bleeding risk [14]. Nonetheless, this advantage comes at the expense of a significantly higher risk for thrombotic events [15].

Thromboembolic events

Despite antithrombotic treatment, thromboembolic events are common. They include: cerebrovascular accident, transient ischaemic attack, arterial non-central nervous system embolism, or pump thrombosis (PT).

Neurologic events remain one of the most dreaded complications of LVAD support and are most often the primary cause of death [14]. The reported incidence of ischaemic stroke during support with the HeartMate II, as either BTT or DT, is 0.064 – 0.082 events per patient per year (EPPY) [16], compared to a stroke rate of 0.013 to 0.035 EPPY reported for patients with advanced heart failure and no LVAD support [10]. The ischaemic stroke rates reported for the HeartWare were 0.11 EPPY [4]. LVAD-supported patients who experience a stroke have twice the risk of death as compared with stroke-free patients [17]. Patients received combined antithrombotic therapy with warfarin (target INR 1.5-2.0 IU) and 81 mg of aspirin. Multivariate analysis showed that diabetes, aortic clamping, duration of support, and INR values were independent predictors of stroke. Mean INR at the time of stroke was subtherapeutic in all patients with embolic strokes. Patients with diabetes were 6.36 times more likely to have a stroke than those without diabetes. Complete aortic clamping with cardioplegic arrest compared with partial aortic clamping with a side-biting clamp was associated with a significantly higher incidence of stroke and was an independent predictor of stroke [16].

Atrial fibrillation (AF) is a well-established risk factor for thromboembolic complications and is frequently found in patients with advanced heart failure, including patients undergoing LVAD implantation. There are scant data regarding the impact of preoperative AF on outcomes after LVAD implantation. However, a recent retrospective analysis of the INTERMACS data of primary CF-LVADs suggests that preoperative AF may not increase the risk of post-operative thromboembolic complications or patient mortality at midterm follow-up [18]. This indicates that the usual postoperative antithrombotic strategy is most likely appropriate also in patients with AF undergoing LVAD implantation.

Both HeartWare and HeartMate II patients are susceptible to PT. PT can cause life-threatening device malfunction and embolic strokes. The rate of PT at three months post-LVAD implantation has unexpectedly increased from 2.2% before March 2011 (a period from 2004 – 2013) to 8.4% after March 2011 (a period from 2011 – 2013) in patients receiving the HeartMate II [16], whereas the incidence of HeartWare thrombosis remains stable. Pump thrombus requiring exchange occurred in 4% of patients with HeartWare at a rate of 0.04 EPPY, and the overall incidence and prevalence of suspected pump thrombus were 0.08 EPPY and 8.1%, respectively [19].

Laboratory tests have a crucial role in the screening and diagnosis of PT. Even early stages of LVAD thrombosis cause haemolysis that can be identified with elevated lactate dehydrogenase levels (LDH), indirect bilirubin, and plasma-free haemoglobin levels. However, these levels can be influenced by many factors and, by using the currently accepted cut-off values, many patients are exposed to potentially unnecessary imaging studies, along with their adverse effects. Non-invasive methods have been shown to help detect PT but with limited accuracy. Echocardiography can offer indirect evidence under some circumstances, including imaging left ventricular thrombus resulting in inflow cannula occlusion, or measuring reduction in diastolic flow velocity across the cannula and increased systolic to diastolic flow velocity ratio. An echocardiographic “ramp study” has been proven to be highly sensitive and specific in the detection of LVAD thrombosis when used in conjunction with LDH levels in axial pumps, but not in centrifugal pumps [20]. The exact role of routine computed tomography angiography (CTA) is unclear. CTA is helpful in patients with an unexplained elevated LDH, in order to rule out other causes that may increase this marker and in order to assess patients with a suspected malposition of one of the pump parts. Further, routine use of CTA is limited due to the risk of contrast nephropathy. Currently, the ultimate diagnosis of PT is a pathological report following LVAD exchange, explant, heart transplantation, or death.

Many antithrombotic regimens have been proposed to prevent PT. However, the data to support any therapy are extremely limited. A recent study aimed to assess outpatient warfarin management: analysing 11,000 outpatient INRs among 249 outpatients demonstrated that thrombotic outcomes (suspected PT and ischaemic stroke) were highest among the lowest INRs (<1.5 [0.40 thrombotic EPPY]), but INR values of 1.5 to 1.99 also had high rates (0.16 thrombotic EPPY). There was a lack of a statistically significant association between INR and thrombotic events in the time span from LVAD implantation to three months post-LVAD implantation, suggesting that early PT may be affected by INR-independent events during the index hospitalisation, such as operative or device characteristics and intensity of postoperative bridging anticoagulation. Conversely, after the early post-surgical period, anticoagulation intensity predicted PT. The results of their weighted analysis indicate that a target INR of 2.6 is optimal for avoiding both bleeding and thrombotic complications and therefore minimising mortality. Their findings support current practice and provide reassurance that the target INR range of 2.0 to 3.0, to minimise all significant adverse events, is supported by data [15].

In answer to past research showing early PT rates of up to 8.4% in patients receiving the HeartMate II, the non-randomised, prospective, multicentre, single-arm Prevention of HeartMate II Pump Thrombosis Through Clinical Management (PREVENT) trial recently examined 300 patients who received the device at 24 centres across the USA and who agreed to follow simple clinical practice management recommendations [21]. It showed a 2.9% rate of confirmed pump thrombosis at three months post-implant and 4.8% at 6 months. The recommended practices focused on implantation technique (maximising flow through the LVAD), anticoagulation regimen (post-operative heparin bridging, a goal INR of 2.0-2.5, early initiation of warfarin and aspirin therapy), early optimal speed management (>9,000 rpm), and blood pressure management (a mean arterial pressure <90 mmHg).

At present, the ideal strategy for treating thrombosis in contemporary devices has yet to be defined. Medical therapy treatment for PT often consists of unfractionated heparin, direct thrombin inhibitors, thrombolysis (local or systemic), and glycoprotein IIb/IIIa inhibitors such as eptifibatide used individually or in combination. Intravenous thrombolysis with alteplase may be used for patients with PT who are poor candidates for re-do procedures. However, these therapies are not benign and are associated with severe side effects, mostly bleeding. Moreover, it is unclear which patients will respond to medical therapy, and the common practice in most centres weighs the risks associated with pump exchange versus thrombolytic therapy in an individualised manner. In case of a clinical suspicion of PT when a patient is haemodynamically unstable, the recommended treatment of PT is LVAD exchange or urgent heart transplant after the patient is hemodynamically stabilized [22]. A recent study showed that, when PT was managed by heart transplantation or pump replacement, patients had a mortality rate similar to individuals without PT. In patients who had suffered from PT and were treated medically (did not undergo a heart transplantation, or pump replacement), mortality was twice as high within six months after diagnosis of PT (16.8% [95% CI, 14.3 to 19.6] versus 35.6% [95% CI, 22.9 to 50.7] at 180 days) [23].

Infections

Infections are a common cause of morbidity and are the second most common cause of death in patients who survive the initial six months on CF-LVAD support. Infections are also one of the leading causes of readmission in these patients. The rates of LVAD-related infections are high, ranging from 30 to 50%. Recent data from the INTERMACS registry indicated that pneumonia and sepsis are the most common infectious complications in patients supported with LVADs (23% and 20%, respectively), followed by driveline site infections (DLIs), which occur in approximately 19% of LVAD recipients within one year after implant [1]. Pump interior infections and pocket infections are uncommon events.

Infections in HeartWare patients are lower in comparison to HeartMate II patients and occur in about 17% of patients [24]. The reason for a lower incidence of infections in HeartWare patients is most probably that HeartWare is implanted within the pericardial space without requiring a pump pocket.

The risk of developing bloodstream infection and subsequent sepsis is highest in the perioperative period. Patients usually present with fever, leucocytosis and, in more severe cases, septic shock. Occasionally, patients may present with septic embolization to distant sites or new incompetence of pump inflow or outflow valves.

Percutaneous DLIs is a late-onset infection, causing the majority of bloodstream infections in LVAD patients, with 85% of these infections reported to occur at more than 30 days after device implantation. An average time-to-occurrence of DLIs is approximately six months after LVAD implantation. After a DLI, mortality has been shown to be as high as 9.8% at six months and 31% at 12 months [21]. Risk factors for DLI development are multifactorial and involve patient factors, such as obesity, nutritional status, age, comorbidities [diabetes mellitus, chronic kidney disease, and depression); surgical technique, such as longer interfacial tunnelling of the driveline and immobilisation of the driveline with an anchoring procedure; and wound care, including patient education and the need to avoid trauma. DLIs may remain superficial, spread deeper along the driveline path and into the pocket or pump itself, or deepen within the abdominal wall to form an abscess.

The most common causative organisms causing infections related to LVADs are predominantly skin organisms, including staphylococcus species (staphylococcus epidermidis and staphylococcus aureus), corynebacterium species, followed by pseudomonas species and enterobacteriaceae, which become even more prominent with longer time on LVAD support. They are very difficult to eradicate [25].

In patients with suspected infections, diagnostic investigations should include prompt culture of drainage from the driveline exit site, three sets of blood cultures, chest radiography, and echocardiography. Ultrasound or computed tomography are frequently used to diagnose collections of fluid around the driveline, pump, or pump pocket, and may also be used to guide aspiration or debridement. Currently, no formal consensus exists on the treatment of DLIs. Neither the initial antibiotic therapy, nor the length of antibiotic therapy is standardised. However, most centres utilise two to four weeks of antimicrobial therapy with or without surgical debridement for DLI. The current recommended empiric antibiotic choice for early localised DLIs is doxycycline, 100 mg twice a day, for 14 days with antibiotic choice adjusted as needed based on culture results [26]. Another regimen includes ciprofloxacin, 500 mg twice a day, and doxycycline, 100 mg PO twice a day, for 10 days [24]. Although two weeks of antimicrobial therapy may be sufficient for a superficial DLI, deeper or more extensive DLI, especially those associated with bacteraemia, may require more prolonged intravenous antimicrobial therapy and surgical debridement. Deterioration on broad spectrum antibiotics should raise the clinical suspicion of fungal infection, which is rare but carries a very high mortality; as such, antifungal therapy should be considered. In patients with recurrent LVAD-related infections, chronic suppressive antibiotic therapy is used; however, studies indicate that approximately one third of patients have recurrence despite antibiotic use. In addition, recurrences are common with device exchange [27]. In a BTT patient with a DLI, an expedited heart transplant listing may be a reasonable option, since studies have shown that patients have no increase in mortality post-transplant, in spite of immunosuppressive therapy. However, patients with sepsis due to DLIs are less likely to be bridged to transplant [28]. More novel therapies focus on salvage of the device by way of serial washouts and instilling drug-eluting antibiotic impregnated beads into the wound. The wound is then serially debrided until clean and closed. This technique is better suited to patients whose device cannot be removed, patients who are poor candidates for cardiac surgery, or patients who have failed conventional prior treatments [29].

Mild and moderate infections can be closely monitored at weekly clinic visits. If the patient has signs of systemic infection, in-patient treatment should be considered.

Aortic insufficiency

De novo aortic insufficiency (AI) is a frequent occurrence in patients supported with LVAD, ranging between 11% and 42% [30]. The mechanism involved in LVAD-induced AI is multifactorial and related to variations in blood flow and pressure in the aortic root. Incorrect angle between the LVAD outflow graft and ascending aorta, may potentially cause aortic root weakening, dilation, and aortic valve cusp malcoaptation. Furthermore, high-pressure and velocity jets of regurgitated blood volumes contacting the root side of a closed aortic valve may result in valvular damage and degeneration. Another contributing mechanism in the development of AI in LVAD patients is change to the aortic wall because of shear stress and high diastolic luminal pressures. Once AI is developed in a patient with LVAD, it is clear that valvular incompetence decreases pump efficiency and can lead to worsening heart failure.

Almost all of the published studies cover the development of AI in patients supported with the HeartMate II device. In contrast, a recent study from Germany showed that AI appears to be very rare in patients supported with the HeartWare. In this study, after a median LVAD support duration of 408 days (77-1,250 days), 24% patients had trace or mild AI, and one patient developed moderate AI (3%). None of the patients developed severe AI and/or required surgical or other procedural intervention of AI [31]. The European version of the HeartWare offers intermittent low-speed software, or a Lavare cycle that may facilitate intermittent opening of the AV; this software is not approved in the USA.

Right ventricular failure

Right ventricular failure (RVF) is a serious complication, as mechanical support is provided to the left ventricle only. RVF occurs in approximately 11% of patients after the insertion of the LVAD [1]. RVF can result from mechanical aetiologies (septal shift to the left from high pump speed or a rightward shift due to low pump speed), volume overload, afterload (pulmonary hypertension), or intrinsic ventricular failure. Some patients develop RVF suddenly after the LVAD insertion. Although some of them may have had some degree of right ventricular dysfunction before the surgery, the RVF does not become apparent until after surgery when there is an obvious imbalance between the newly supported left ventricle and the failing right ventricle. Elevated central venous pressure, elevated pulmonary vascular resistance, and pulmonary artery pressures with reduced LVAD flows and RV cardiac output indicate acute RVF. Affected patients require intensive medical therapy directed at preventing hypoxia aimed at preventing pulmonary vasoconstriction and optimising oxygen delivery to the heart and focused on the causative factors. Intrinsic RVF can be managed with a variety of inotropic agents, including dobutamine, epinephrine, and milrinone. Pulmonary vasodilators, such as inhaled nitric oxide and inhaled prostacyclin, should be considered in settings of elevated pulmonary pressures. When medical therapy is ineffective, patients require insertion of a right VAD. The temporary RV support is maintained until the recovery of the right ventricle; otherwise the patient requires the long term RVAD or total artificial heart. Prior studies have shown that early initiation of biventricular support may be associated with improved survival [32].

Limitations

There are several limitations when reporting medical complications in LVAD patients. Although we have some data from randomised, controlled trials, studies cannot be unblinded. In addition, studies include relatively small samples of highly selected patients treated at a limited number of participating centres under well-controlled trial settings. Further, the reporting of event rates is not standardised, so data are difficult to compare across different studies. All these reasons make estimates of rare events difficult and not easily generalizable to the usual clinical setting.

Conclusions

For populations of carefully selected patients, LVADs improve survival, health-related quality of life and functional status. However, LVAD therapy remains associated with major adverse events in a significant percentage of patients. The incidence of adverse events varies between devices (HeartMate II and HeartWare), as well as between indications for LVAD implantation (DT or BTT).

Haemorrhagic and ischaemic stroke are among the most dreaded complications. Stricter control of post-operative anticoagulation may impact significantly on the incidence of this devastating complication. Studies show a decline in bleeding event rates over time; however, bleeding and anaemia requiring transfusion are common and associated with an increased risk of mortality. Infection remains one of the major contributors to morbidity. Strategies to prevent infection are lacking and are not very successful. Standardised antibiotic therapy is not well established, and infection can be difficult to eradicate.

Control or elimination of defined risk factors, improved and novel surgical implantation techniques, and advances in LVAD technology are likely to reduce device-related complications in the future.