Heart Failure Association of the European Society of Cardiology position paper on the management of left ventricular assist device-supported patients for the non-left ventricular assist device specialist healthcare provider: Part 2: at the emergency department

Incidence

Significant bleeding complications are quite common in LVAD carriers, incidence varying depending on the source, but ranging between 22% and 32% during LVAD support or 0.35–0.65 events per patient-year.^{1, 2} These bleeding episodes were rarely fatal: major bleeding was the cause of death in 2% of the total number of deaths in LVAD patients.³ Recently, a special focus has been placed on the interaction of blood components and the device, that is, haemocompatibility, with increasing awareness of the importance of these interactions on adverse events.⁴

Comparisons of the haemocompatibility parameters of different LVADs have been examined, and while pump thrombosis has been present less often in the newest centrifugal-flow device, the incidence of bleeding was quite similar between the investigated devices (HMII™, HM3™, and HW™).^{5, 6} On the other hand, the recent analysis by Kormos et al. noted a significantly lesser incidence of gastrointestinal bleeding (GIB) in patients implanted with a centrifugal-flow pump compared with those who received an axial flow pump.^{3, 4} Similar findings were observed in the MOMENTUM trial, where a significantly lower incidence of any bleeding, and GIB in particular, was seen with the centrifugal pump (HM3 vs. HMII).⁴

Significance and long-term repercussions of bleeding

Although bleeding complications are usually not fatal, they are yet clinically significant. Aside from the obvious impairment of quality of life, bleeding usually leads to adjustments in anticoagulant and anti-platelet therapy, which could potentially increase the probability of pump thrombosis.

Additionally, some of the bleeding episodes will be significant enough to require blood transfusions at an average of 2–4 units of packed red blood cells per admission.⁷ This is particularly relevant for the bridge-to-transplant candidates, in whom leuco-irradiated products should be used to reduce the risk of allo-sensitization caused by the excessive exposure to blood products that could result in low availability of appropriate heart donors or higher risk for humoral rejection after heart transplantation (HTx).^{8, 9}

Pathophysiology of bleeding

An obvious contributing factor to the bleeding diathesis is the mandatory anticoagulant and anti-platelet therapy, required for the normal function of the devices.

Other possible contributing factors have been investigated. A lot of attention has been devoted to the interplay of the pump with the blood, unified under the name haemocompatibility.⁶

One of the examined pathohistological mechanisms for the induction of bleeding in ventricular assist device-supported patients is the acquired von Willebrand syndrome (AvWS). von Willebrand factor (vWF) is large multimeric glycoprotein, crucial for the process of haemostasis at the site of vascular injury, where it promotes platelet aggregation and the formation of platelet plug, also serving as a carrier protein for FVIII and protecting it from proteolysis.^{10, 11} Patients with impaired function of vWF, that is, patients with von Willebrand disease, are prone to bleeding complications. This has previously been described also in patients with aortic stenosis (Heyde's syndrome) and severe heart failure (HF) patients: vWF may already be degraded in some patients with severe HF at the time of LVAD implant, and these patients tend to bleed more.¹² A similar pathological process is seen in LVAD carriers.¹³

According to the current literature, most all of the LVAD patients have loss of vWF activity,¹⁴ which seems to resolve after LVAD explant or HTx.¹⁵ Different mechanisms have been hypothesized to contribute to the AvWS in LVAD patients. It is thought that increased shear stress in the LVAD circulation may cause the vWF to disentangle, thus exposing domains promoting proteolysis and degradation of vWF,¹¹ which then leads to reduced platelet aggregation. But despite some extent of AvWS being present in most LVAD patients,¹³ not all experience bleeding events.

von Willebrand factor is also considered an inhibitor of angiogenesis,¹⁶ meaning loss of function of vWF could be implicated in promotion of angiodysplasia and arteriovenous malformations.¹⁷ Bansal et al. conducted a prospective multicentre clinical trial, which demonstrated lower amount of vWF monomer degradation in the HM3 group.¹² Despite all that, it is yet unclear if assessment for inherited or acquired coagulation disorders is warranted.

Platelet damage and dysfunction observed in LVAD patients,^18-20 thought to be related to shear forces in the LVAD, could potentially add to the increased risk of bleeding in this population.

Arteriovenous malformation is anatomically the most frequently reported cause of GIB. Right HF is another potential risk factor for GIB through the enhanced arteriovenous malformation formation due to elevated portal vein pressure and coagulopathy associated with congestive hepatic failure. Non-pulsatile flow was considered a potential predisposing factor for the development of GIB through the formation of small intestinal angiodysplasia²¹; however, studies with partial ventricular support with fully preserved arterial pulse still reported a high incidence of GIB.²²

Improved haemodynamics might have a favourable effect on suppression of GIB, achieved through the use of optimal medical therapy or optimizing LVAD speed through ramp testing²³ and medications to control the blood pressure such us angiotensin-converting enzyme inhibitor and angiotensin receptor blocker²⁴ and digoxin that was found to be associated with a decreased incidence of angiodysplasia-related GIB in patients with continuous-flow (CF)-LVADs.²⁵

The favourable effects of omega-3 unsaturated fatty acids on angiogenesis and inflammation have been previously described,²⁶ and in a small study, authors demonstrated that LVAD patients regularly receiving omega-3 acids had a greater freedom from GIB compared with controls.²⁷

Risk factors for bleeding in the left ventricular assist device patient

Some patient populations are at increased risk of bleeding. In the case of GIB, risk factors include older age,²⁸ lower body mass index, history of GIB prior to LVAD implant, smoking, elevated international normalized ratio (INR), and a low platelet count.²⁹ Patients with pre-implant right ventricular failure have been prone to development of bleeding complications, possibly due to hepatic congestion and subsequent coagulopathy. Female sex has also been associated with an elevated incidence of all bleeding complications after LVAD implant.²⁸

Diagnosis and treatment options

The extent and the selection of diagnostic and therapeutic modalities will vary according to the bleeding location, as well as its severity and the clinical presentation of the patient.

Gastrointestinal bleeding

Gastrointestinal bleeding is most frequently seen in LVAD patients,¹⁰ with the incidence ranging between 25% and 35% of patients with events on LVAD support or 0.31–0.56 events per patient-year,^{4, 5, 32} and is one of the major causes of readmission of LVAD patients.¹⁰ GIB can occur in any part of the gastrointestinal (GI) tract, but most often in the upper portion, proximal to the Treitz ligament.³⁰

Non-gastrointestinal bleeding

Gynaecological bleeding

Given the relatively small proportion of female LVAD patients: around 20%,^{3, 43} the data on management of gynaecological bleeding are understandably scarce.

Women seem to have a greater incidence of non-gynaecological types of bleeding complications (including gastrointestinal, naso-oropharyngeal, and ICH) compared with men, but this did not translate into worse survival.^{2, 28, 44} In an analysis by Yavar et al., 10% of patients experienced gynaecological bleeding requiring transfusions or surgical interventions.²⁸ Depending on the severity, such episodes might require interruption of anticoagulant and anti-platelet treatment, hormonal therapy, or, in extreme cases, surgical treatment.

The aetiology of increased bleeding diathesis in women is not yet elucidated.

Other sources of bleeding

Other forms of bleeding that might be encountered in LVAD carriers include mostly bleeding from the nasal or oral mucosa, seen usually as epistaxis, after teeth extraction, skin bruises, or haematuria.²⁸ Although these bleeding episodes are usually treated by other specialists (otolaryngology, oral surgeons, urologists, etc.), it is advised to include a cardiologist specialized in managing LVAD patients in the team responsible for deciding on interruption of anticoagulant or anti-platelet therapy.

Prevention

Several preventive measures can be undertaken in order to reduce the number and intensity of bleeding episodes.

Optimal management of anticoagulation is of utmost importance: keeping the INR values within the therapeutic range could help decrease the risk of bleeding complications. For this end, patient education and nutritional counselling are very helpful if the oscillations of INR levels are suspected to be associated with food interactions (e.g. green and leafy vegetables). Detailed guidelines on the management of oral anticoagulation in this population are available.^{45, 46}

Coumarin oral anticoagulants such as warfarin are metabolized via the CYP2C9,⁴⁷ which is also a common pathway for several other drugs commonly used in this patient population (antidiabetic medication, angiotensin receptor blockers, anticonvulsive medication, etc.).

Given the common occurrence of polypharmacy in these patients, caution must be exercised regarding the drug–drug interactions. In patients with special difficulties maintaining the INR range, genetic counselling with the identification of the patient's metabolic profile is possible.⁴⁷ Devices for home monitoring of INR are available and could potentially be beneficial in achieving optimal INR control.

The excellent results of the MOMENTUM trials,⁴ demonstrating significantly reduced burden of device thrombosis in HM3 carriers, raised the potential need for decreased level of anticoagulation. This was examined in the MAGENTUM 1 trial, which evaluated lower target INR levels (target INR 1.5–1.9) in 15 HM3 carriers after the sixth post-operative week. Primary endpoint (event-free survival) was achieved in 93 ± 6%, but due to the small sample size. This study needs further validation.⁴⁸

Ischaemic stroke

Ischaemic strokes can occur because of thrombi that pass from the heart through the device and into the brain or could develop in unique locations, such as the aortic surface of the valves within the sinus of Valsalva, the carotid bulb, and in some cases septic emboli as a result of infections.^{58, 59}

The most consistent risk factor for ischaemic stroke in CF-LVADs is systemic infection, which has increased the risk of stroke by nearly two-fold (39.5% vs. 19.3%, P = 0.003).^60-62 In the INTERMACS registry, independent preoperative predictors of ischaemic stroke were female sex and previous cardiac operation.⁵² However, when considering post-operative risk factors, infection and GIB significantly increased the risk of stroke.⁵²

One of the most feared complications of any ischaemic stroke is bleeding into the area of the infarcted brain (haemorrhagic conversion/transformation). Consequently, the risk assessment for resuming anti-thrombotic agents is often guided by known risk factors in the undifferentiated ischaemic stroke patient, including infarct size, cardioembolic aetiology, vessel recanalization, and heparin treatment.^63-65

Haemorrhagic stroke

Intracranial haemorrhage is a complication, which usually results in significant long-term consequences for the patient. An INTERMACS analysis looking into the incidence of stroke in LVAD carriers showed that 10% had at least one stroke, of which nearly 50% were haemorrhagic, leading to a significantly worse survival compared with those who suffered an ischaemic stroke.⁵² In this analysis, 45% of patients survived 1 month after the ICH, while only 30% survived 1 year. In a multivariable analysis, heparin-induced thrombocytopenia, use of intra-aortic balloon pump, female sex, and primary cardiac diagnosis were significant predictors of ICH. Female sex has been a predisposing factor in other analysis.^{8, 28, 66} Whereas GIB might facilitate a sooner transplantation in LVAD carriers, ICH more often renders the patient ineligible for transplant, depending on the residual morbidity.⁵²

There are several possible causes of haemorrhagic stroke in patients with LVAD, such as anticoagulation, haemorrhagic conversion after infarction, the acquired von Willebrand syndrome as a result of mechanical destruction, and proteolysis of high-molecular-weight multimers of von Willebrand factor 8 and rupture of histologically fragile vessels caused by the non-physiological continuous pulseless flow.^{8, 67} It is often unclear why a specific CF-LVAD patient develops a specific ICH, as coagulopathy alone appears neither necessary nor specific. Intra-cerebral haemorrhages occur in the absence of endocarditis with subtherapeutic INRs, whereas supratherapeutic INR levels do not necessarily lead to haemorrhagic stroke.⁶⁷

The aetiology of ICH is not completely elucidated. The presented rates of haemorrhagic strokes are significantly higher compared with other populations treated with warfarin, and other studies demonstrated that spontaneous ICH has been noted in all INR ranges.^68-70 Interestingly, all the patients who developed ICH had concomitant AvWS.⁷¹

The incidence of haemorrhagic stroke has also been shown to reduce with the newest generation of centrifugal-flow devices.⁴ This leads to the conclusion that there could be several predisposing factors for ICH.

While it is long known from normal pulsatile physiology, elevated BP during LVAD support has recently been associated with a higher incidence of intracranial bleeding, and subsequent mortality, which could indicate BP control as a beneficial measure in preventing ICH.^{72, 73} Given the established difficulties in obtaining BP measurements in LVAD carriers, an alternative to the traditional electronic BP measuring devices has been investigated, and Doppler-derived BP measurements have shown to correlate well to invasive systolic BP measuring.⁷⁴ According to the International Society for Heart and Lung Transplantation guidelines for mechanical circulatory support, level of evidence C (expert opinion), the Doppler-derived BP is to be maintained at ≤80 mmHg.⁷⁵

The correlation between hypertension and neurological events was not so obvious in the INTERMACS analysis. Such a correlation might be device specific as presented in the ENDURANCE supplemental study showing that BP management is associated with reduced stroke rates in HW subjects.⁶¹ Maintaining BP in the reference range was associated with better overall outcomes.⁵¹ The authors stress the importance of avoiding too aggressive BP management with hypotension as well as hypertension.

Strict management of BP maintaining MAP at 70–90 mmHg⁴⁹ and anticoagulation targeting INR levels of 2–2.5⁴⁹ are expected to be beneficial preventive strategies for several complications of LVAD treatment, but this does not necessarily guarantee absence of bleeding complications⁴⁹ (see Part 3 for more on anticoagulation in LVAD-supported patients).

A study by Trachtenberg et al. indicated blood stream infections as a predisposing factor for neurological events, both ischaemic and haemorrhagic.⁷⁶

Survival after the haemorrhagic stroke is significantly reduced and was 45.3%, 34.8%, and 30.3% at 1 month, 6 months, and 1 year, respectively.⁵²

Cerebral microbleeds shown by brain magnetic resonance imaging are predictive of symptomatic cerebral haemorrhage in patients with various causes.⁷⁷

In general population studies, it has been reported that several factors (e.g. male sex, older age, and elevated BP) show positive correlations with the number of cerebral microbleeds,^{67, 78} while correlations with aspirin and warfarin administration, which are mandatory for patients with LVAD, remain controversial.⁷⁹

One small-scale study assessed a treatment protocol for ICH in LVAD patients, which included observation alone, observation with suspension of anticoagulation, observation with reversal of anticoagulation, or reversal of anticoagulation plus surgical intervention.

Reversal of warfarin was performed with fresh frozen plasma (FFP), prothrombin complex concentrate, vitamin K, or a combination. Prothrombin complex concentrate is preferred in neurosurgical emergencies, given its faster effect.⁷¹

Depending on the severity of presentation, some patients will require surgical treatment, such as a craniotomy with drainage.⁵³ Ramey et al. suggest terminating anti-platelet treatment in patients with established AvWS, but this is not supported by data.⁷¹

Routine administration of platelets is debatable in ICH, because adverse outcomes have been noted, but it might be requested in patients undergoing neurosurgical intervention.⁷¹

The significance of radiological screening of potential risk factors for ICH is not established. A multidisciplinary approach to treatment and decision-making might be optimal in this setting.

Treatment

The assessment of all stroke patients is performed using standardized examination tools, such as the National Institutes of Health Stroke Scale, and a measure of consciousness, such as the Glasgow Coma Scale, for the sake of monitoring the patient with reproducible, objective measurement tools. Intravenous recombinant tissue plasminogen activator is highly effective for improving functional outcomes in general acute ischaemic stroke patients, but there are no trials in LVAD-related stroke patients. Thrombolysis may be ineffective because thrombi from within the CF-LVADs may have a composition (fibrin and denatured protein) that is not been amenable to thrombolysis.

Patients with CF-LVADs are likely to be at significantly higher risk of symptomatic haemorrhagic transformation after systemic thrombolysis. Firstly, patients with CF-LVADs are often taking warfarin and typically at least one anti-platelet agent, and tend to have larger territorial infarcts, which are all associated with a higher risk of haemorrhagic conversion in several prediction scores. Secondly, the association of systemic infection with ischaemic stroke increases the likelihood a septic embolism, and thrombolysis in patients with bacterial endocarditis leads to high rate of haemorrhagic complications (close to 20%).⁸⁰ Thirdly, patients with CF-LVADs have an acquired vWF deficiency that would increase the risk of haemorrhagic transformation.⁸

Because of the high risk for haemorrhagic transformation, endovascular stroke therapy may provide specific clinical and diagnostic benefits in patients with a CF-LVAD, avoiding systemic fibrinolytic therapy in patients treated with anticoagulation and anti-platelets. However, the risk of haemorrhagic transformation with endovascular stroke therapy appears like that of systemic thrombolysis.⁸¹

Treatment decisions for acute haemorrhagic stroke patients are based on whether the patient has had a primary haemorrhagic stroke (intra-cerebral haemorrhage and subarachnoid haemorrhage) or haemorrhagic transformation. Other factors include the anti-thrombotic agents used, coagulation profile, and platelet count. Regardless of aetiology, the first focus of treatment is to establish the need for mechanical ventilation due to impaired alertness, in parallel with preventing expansion of the haematoma. ICH is associated with a very high in-hospital mortality of about 20%.⁶⁷

Currently, the mainstay of therapy in haemorrhagic stroke is reduction of BP and reversing coagulopathy, in line with safety concerns regarding BP targets with specific devices. The acute treatment of BP in the CF-LVAD patient with an ICH is not well established in the guidelines for stroke, but it would be reasonable to lower mean arterial pressure to <90 mmHg, as suggested for primary stroke prevention in CF-LVAD.⁸²

The decision to reverse a coagulopathy is more complex. The most recent ICH guidelines from the American Stroke Association suggest reversing warfarin with prothrombin complex concentrates or FFP without a specific INR target.⁸³

The decision about when to resume anticoagulation in the ischaemic stroke patient without endocarditis is driven by the presence, and likelihood, of developing haemorrhagic conversion in the general ischaemic stroke patient. Generally, resume warfarin at the same time as aspirin (without a heparin bridge) within 24 h if haemorrhagic conversion is absent and the total infarct size is <50% of the middle cerebral artery territory (given the association of large infarct size with haemorrhagic conversion).⁶⁷

If the patient is at high risk for haemorrhagic conversion, such as with large infarcts with petechial bleeding, a head computed tomography (CT) is repeated at 5 days and warfarin is resumed at that point if there is no evidence of haemorrhagic conversion.⁶⁷

After a primary ICH in the general stroke patient, long-term anticoagulation can be resumed for clear indications if there is no lobar ICH due to cerebral amyloid angiopathy.

Early decompressive haemicraniectomy for large hemispheric ischaemic strokes with somnolence and intact brainstem reflexes (except loss of one pupillary response) provides substantial mortality benefit (absolute risk reduction 78–29%) in patients <60 years of age.⁸⁴ However, confounding the care of patients with CF-LVADs, haemicraniectomy requires withholding anticoagulation to prevent surgical site bleeding until 4–8 weeks afterward when the cranioplasty is placed. Post-operative management with anti-thrombotic agents and long-term functional prognosis in large strokes increases the complexity of the decision on whether to perform haemicraniectomy.

There is a strong need to continue research into the risk factors, natural history, complication rates, and acute stroke treatment for patients with CF-LVADs.

The efficacy of treatment after stroke is also not well known, with small series documenting results after reversing anti coagulation in ICH, but not outcomes beyond death such as functional statistics.⁸⁵ This is notable because stroke can directly lead to death but more frequently leads to significant disability. The latter can be particularly important to patients with LVAD in whom complications from limited mobility can be substantial, including loss of eligibility for transplantation and infection.

Diagnosis

Inflow cannula and intra-device thrombus

Left ventricular assist device thrombus that is located within the pump has a very distinct appearance (Figure 1); once seen, one can better understand the catastrophic consequences of this event.⁸⁸

Inflow cannula and intra-pump thrombus are suspected in the setting of any of the following: signs of worsening HF (progressive dyspnoea on exertion), acute changes in the LVAD parameters (‘low-flow’ alarms and LVAD power spikes), haemolysis (lactic dehydrogenase increase, haemoglobin drop plasma-free haemoglobin rise, and darker urine), and signs of distal embolization.

Once an LVAD thrombus is suspected, one should explore the LVAD alarms and power spikes: if the decrease of power and flow happens in a short period of time, an inflow pump thrombosis is the more reasonable diagnosis. This suspicion of intra-device or inflow cannula thrombus can be confirmed using the following tools:

• Transthoracic echocardiogram—ramp studies:

  • Left ventricle enlargement
  • Opening of the aortic valve in each cycle
  • Worsening mitral regurgitation
  • Inflow cannula turbulence
  • Concern for a possible seen thrombus

• CT with angiography that will demonstrate no outflow graft obstruction

Treatment

If an outflow graft obstruction was diagnosed and a percutaneous approach is applied, when balloon inflating does not suffice, the use of Wallstent is preferred. The physiological effect of reliving the outflow graft occlusion can be seen immediately after deploying the stent by a reduction of the power consumption and an increase in the pump flow.

A word of caution regarding the use of thrombolysis must be cleared. Its efficacy in treating LVAD thrombosis has been questioned along the years and was debated because of increased risk of bleeding; however, data were scarce. Lately, Seese et al.⁸⁴ have demonstrated that out of 26 patients that received initial thrombolytic therapy, in only three patients a successful treatment was achieved. Conversely, three patients suffered from major bleeding and two patients had a new stroke. Furthermore, most patients undergoing thrombolytic therapy underwent subsequent device exchange.⁸⁴ Hence, thrombolysis for an HMII pump thrombosis should only be applied as a last resort due to the limited efficacy of thrombolysis and the related increased risk of major bleeding or stroke and surgical pump exchange is preferred. Regarding the management of an HW pump thrombosis, as of 3 June 2021, the distribution of the HW has been discontinued; therefore, in case of the need for pump exchange, the HW would have to be replaced with the HM3. A surgical procedure for such an LVAD exchange is not established yet, and most centres do not have the needed experience in performing such an exchange. Thus, the non-surgical approach to HW pump thrombosis must be exploited as far as possible.