Implications of Etiologies of Intracranial Hemorrhage During Left Ventricular Assist Device Support

Abstract

Background:

Intracranial hemorrhage (ICH) is a frequent complication in patients with an implanted left ventricular assist device (LVAD) for advanced heart failure. Bloodstream infection is known to be associated with ICH in patients with LVAD, but its effects on ICH-associated mortality are unknown. We compared characteristics and mortality of infection-associated, traumatic, and spontaneous hemorrhages.

Methods:

Patients in an LVAD registry at a tertiary care center were reviewed for this cohort study. ICH included intraparenchymal hemorrhage (IPH), subarachnoid hemorrhage (SAH), and subdural hemorrhage (SDH). Hemorrhages were categorized into infectious, traumatic, and spontaneous by the presence or absence of concurrent device-associated infection or antecedent trauma.

Results:

Of 683 patients with an LVAD, 73 suffered ICH (10.7%). IPH was the most prevalent (72%), followed by SAH (27%) and SDH (23%) with multiple concurrent hemorrhage subtypes in 16 patients (22%). Median time from implantation to ICH was shorter in spontaneous ICH than in infection-associated ICH (100 days vs. 252 days, p=0.048). The prevalence of the different subtypes of ICH were similar between spontaneous and infection-associated ICH, but IPH was less prevalent (20% vs 76%, p <0.001) and SAH more prevalent (60% vs 22%, p=0.019) in traumatic ICH than spontaneous ICH. No differences were seen in mortality between the different etiologies of ICH.

Conclusions:

While spontaneous ICH occurred earlier after LVAD implantation than infection-associated ICH, no difference in mortality was seen between spontaneous, infection-associated, or traumatic ICH.

INTRODUCTION

Left ventricular assist device (LVAD) implantation is considered in patients with advanced cardiac failure refractory to medical therapy and may be used as a bridge to transplant as well as destination therapy. LVADs have demonstrated mortality benefits in appropriately selected patients¹, but remain associated with frequent complications including acute cerebrovascular injury^2,3. Neurologic complications are frequently seen in LVAD patients and intracranial hemorrhage (ICH) is associated with an increased risk of morbidity, mortality, and ineligibility to undergo heart transplantation compared to patients without ICH^4,5. While prior research has identified outcomes and risk factors for the development of ICH in LVAD patients⁶, the implications of different etiologies of ICH in LVAD patients have not been well characterized.

The high prevalence of LVAD-associated ICH is likely related to multiple device and patient factors. Anticoagulation to prevent device thrombosis places the patient at increased risk for ICH related to coagulopathy as well as a higher risk of ICH after trauma. In addition, infection is among the most common complications encountered post-LVAD implantation.⁷ Our prior study has shown infection is a strong risk factor for the development of LVAD-associated ICH (in particular subarachnoid hemorrhage [SAH]) potentially due to septic embolism and rupture of mycotic aneurysms⁸. Although the relationship between infection and LVAD-associated strokes has been described⁹, the impact of its outcomes after ICH is not known. We aimed to compare risk factors and outcomes of patients with LVAD who developed ICH either in the presence or absence of device-associated infection.

METHODS:

Consecutive patients undergoing LVAD implantation between October 2004 and April 2019 at a single tertiary care center were collected in a local prospective database (Electronic Data Interface for Transplantation). Prospectively collected data included preimplant demographics, medical history, substance use history, device type, and clinical status. The records of patients with an LVAD device who developed ICH were then retrospectively reviewed from time of implantation until their death, heart transplantation, or LVAD explantation. Inclusion criteria included age ≥ 18 years and implantation of a durable LVAD for either bridge to transplant (BTT) or destination therapy (DT). Exclusion criteria were placement of a biventricular assist device, total artificial heart, or short term mechanical circulatory support device. This study was approved by the local institutional review board (number 17–363).

All patients diagnosed with ICH were evaluated at the time of diagnosis by a board-certified vascular neurologist and underwent computed tomography (CT) imaging. CT images were reviewed by a neuroradiologist at the time of the stroke and retrospectively by an independent single neurology investigator (A.S.). Magnetic resonance imaging (MRI) was not performed due to incompatibility with LVAD devices. Details of the patients’ neurologic status at the onset of the ICH (including National Institutes of Health Stroke Scale [NIHSS], Glasgow Coma Scale, ICH score, and modified Fisher scale) were retrospectively collected. ICH included IPH, SAH, and SDH, and patients with multiple subtypes of hemorrhage were coded in each category. Information about antithrombotic therapy (none, antiplatelet therapy, anticoagulation, or a combination of antiplatelet and anticoagulation) at the time of ICH was collected.

Infection was defined according to INTERMACS adverse event criteria¹⁰ as a clinical event accompanied by fever, pain, drainage, and/or leukocytosis treated with antimicrobial agents. A positive culture was required unless strong clinical evidence led to the diagnosis and treatment by an infectious disease specialist at the time despite blood negative cultures. Infection-associated hemorrhage was defined as an ICH identified while the patient had an active device-associated infection. Device-associated infection categories included blood stream infection, pump infection, driveline infection, and pump pocket infection. All infections were treated with antibiotics guided by an infectious disease specialist consultation. A traumatic ICH was defined as an ICH preceded by significant head trauma and where etiology was determined to be traumatic by the evaluating vascular neurologist. Spontaneous ICH was defined as hemorrhage in the absence of infection or antecedent trauma.

Statistical Analysis

Patient and hemorrhage characteristics were compared between spontaneous, traumatic, and infectious hemorrhages. Comparisons of demographic and clinical characteristics were analyzed using Fisher’s exact test, Chi-square, or Mann–Whitney U test as appropriate. Kaplan-Meier survival analyses were performed using Breslow-Wilcoxon method for comparison of survival probabilities. A p value less than 0.05 was considered significant. Analyses were conducted using SPSS version 27.0 (Chicago, IL).

RESULTS:

Between October 2004 and April 2019, 683 patients underwent LVAD implantation and met inclusion criteria. Seventy-three (10.7%) of these patients developed ICH during their course of LVAD support. Of these hemorrhages, 10 (13.7%) were determined to be associated with trauma, 26 (35.6%) were associated with infection, and 37 (50.7%) were spontaneous ICHs. Preimplant demographics, medical history, substance use history, type of device, pre-implant clinical status, and the purpose for implantation (Table 1) did not differ between infectious and spontaneous hemorrhages except for history of hypertension, which was more common in patients with spontaneous ICH (91.9% vs 69%, p=0.038). Patients with traumatic ICH were older than those with spontaneous ICH (67 vs 60 years old, p = 0.012) and all patients with traumatic ICH were male (Supplemental Table 1). No difference was seen between different antithrombotic management strategies at the time of hemorrhage between different etiologies of ICH (Supplemental Table 2, p=0.981).

Table 1:

Characteristics of Patients who developed Intracranial Hemorrhage after LVAD Implantation

	All (n=73)	Trauma (n=10)	Infection (n=26)	Spontaneous (n=37)	P value Infection vs Spontaneous
Age at ICH, years, median (Interquartile range [IQR])	60 (50–68)	67 (63–70)	56 (46–66)	60 (50–67)	0.357
Male	59 (80.8%)	10 (100%)	22 (84.6%)	27 (72.9%)	0.362
Race: White	54 (73.9%)	7 (70%)	22 (84.6%)	25 (67.6%)	0.152
Race: African American	17 (23.3%)	3 (30%)	4 (15.4%)	10 (27.0%)	0.362
Indication for LVAD					0.797
Bridge to Transplant	30 (40.5%)	4 (40%)	10 (38.4%)	16 (43.2%)
Destination Therapy	43 (59.5%)	6 (60%)	16 (61.6%)	21 (56.8%)
Patient Medical History
Hypertension	59 (80.8%)	8 (80%)	18 (69%)	34 (91.9%)	0.038
Diabetes Mellitus	37 (50.7%)	7 (70%)	10 (38.4%)	20 (54.1%)	0.306
Cardiac Arrythmia	56 (76.7%)	8 (80%)	22 (84.6%)	26 (70.3%)	0.238
Chronic obstructive pulmonary disease	13 (17.8%)	0 (0%)	3 (11.5%)	10 (27%)	0.207
Peripheral vascular disease	3 (4.1%)	0 (0%)	2 (7.7%)	1 (2.7%)	0.564
Myocardial infarction	33 (45.2%)	4 (40%)	11 (42.3%)	18 (48.6%)	0.798
Stroke	7 (9.6%)	1 (10%)	4 (15.4%)	2 (5.4%)	0.220
Malignancy	13 (17.8%)	0 (0%)	6 (23.1%)	7 (18.9%)	0.757
Tobacco abuse	48 (65.8%)	5 (50%)	16 (61.5%)	27 (72.9%)	0.414
Alcohol abuse	11 (15.1%)	2 (20%)	5 (19.2%)	4 (10.8%)	0.469
Drug abuse	4 (5.5%)	0 (0%)	1 (3.8%)	3 (8.1%)	0.637
Coronary artery bypass	20 (27.4%)	2 (20%)	7 (26.9%)	11 (29.7%)	1.00
Any valve surgery	14 (19.2%)	0 (0%)	6 (23.1%)	8 (21.6%)	1.00
Percutaneous coronary intervention	25 (34.2%)	3 (30%)	7 (26.9%)	15 (40.5%)	0.296
Defibrillator	64 (87.7%)	10 (100%)	24 (92.3%)	30 (81.1%)	0.286
INTERMACS Profile					0.388
1	13 (17.8%)	2 (20%)	7 (26.9%)	4 (10.8%)
2	26 (35.6%)	2 (20%)	10 (38.5%)	14 (37.8%)
3	17 (23.3%)	2 (20%)	5 (19.2%)	10 (27.2%)
4	12 (16.5%)	3 (30%)	2 (7.7%)	7 (19.4%)
5	5 (6.8%)	1 (10%)	2 (7.7%)	2 (5.4%)
Device					0.569
HeartMate 2	40 (54.8%)	9 (90%)	11 (42.3%)	20 (54.1%)
HeartMate 3	10 (13.7%)	0 (0%)	4 (15.4%)	6 (16.2%)
HeartWare	23 (27.4%)	1 (10%)	11 (42.3%)	11 (29.7%)

The timing of ICH after LVAD implantation and mortality after ICH were compared between the different etiologies. (Table 2). ICH occurred earlier after LVAD implantation in spontaneous ICH than infection-associated ICH (median 100 days vs. 252 days, p=0.048). Spontaneous ICH was more likely to occur within the first 30 days after LVAD implantation than infection-associated ICH (27.0% vs 3.8%, p=0.017). There was no significant difference between the median time from ICH to death of spontaneous and infection-associated ICHs (2 days vs 11 days, p=0.638). Kaplan Meier survival curves were similar between spontaneous and infection-associated ICHs (Figure) at both 30 days (43.2% vs 53.8%, p=0.451) and one year (35.1% vs 26.9%, p=0.587). Subsequent conditional mortality analyses were performed to evaluate long term survival among patients who survived after their initial ICH. One year mortality among patients surviving beyond 30 days after ICH was 33.3%. No significant difference was seen in the conditional mortality of the ICH subtypes including IPH (36.4%), SAH (50.0%), and SDH (32.9%). When comparing patients who survived greater than 30 days after initial ICH, a trend towards increased mortality was seen in infection-associated vs spontaneous ICH (50% vs 18.8%, p=0.07) and IPH (54.5% vs 20%, p=0.10).

Table 2:

Timing and Mortality for All Intracranial Hemorrhages

	All (n=73)	Trauma (n=10)	Infection (n=26)	Spontaneous (n=37)	P value Infection vs Spontaneous
Median Time from Implantation to ICH, day (IQR)	204 (54–449)	120 (13–430)	252 (118–512)	118 (31–447)	0.048
ICH within 30 days of implantation	31 (42.5%)	3 (30%)	1 (3.8%)	10 (27.0%)	0.017
ICH, Glasgow Coma Scale, median (IQR)	10 (3–15)	14 (6–15)	11 (3–15)	9 (3–15)	0.406
30-day mortality after ICH	37 (50.6%)	4 (40%)	12 (46.2%)	21 (56.8%)	0.606
1-year mortality after ICH	49 (67.1%)	6 (60%)	19 (73.1%)	24 (64.9%)	0.587
Median Time from ICH to death, days (IQR)	6 (1–125)	39 (5–340)	11 (1–99)	2 (1–8)	0.638
ICH Conditional 1-year mortality	12/36 (33.3%)	2/6 (33.3%)	7/14 (50%)	3/16 (18.8%)	0.070

Figure:

Kaplan Meier survival curve at one year after intracranial hemorrhage comparing spontaneous and infectious hemorrhages. Traumatic hemorrhage not included due to low sample size. ICH: Intracranial Hemorrhage.

Among LVAD-associated ICH, IPH was the most frequent subtype (52, 71.2%) followed by SAH (23, 31.5%) and SDH (17, 23.2%). Sixteen patients (21.9%) had more than one concurrent subtype of ICH (for example, an IPH and SAH at the same time). When comparing spontaneous and infection-associated ICH, the prevalence of IPH (75.7% vs 84.6%, p=0.53), SAH (21.6% vs 34.6%, p=0.26), and SDH (27.0% vs 7.7%, p=0.19) were similar (Supplemental Table 2). Neurologic impairment by NIHSS score and Glasgow Coma Scale, IPH prognostic scale by ICH score¹¹, IPH volume (by ABC/2¹²), and SAH radiographic severity by modified Fisher Scale score¹³ were similar by etiologic subtype of ICH.

DISCUSSION:

The primary aim for the study was to report different timing, risk factors and outcomes between infection-associated ICH and spontaneous ICH. We demonstrated that spontaneous ICH occurred earlier after LVAD implantation compared with infection-associated ICH. However, the prevalence and mortality of various ICH subtypes did not significantly differ.

Our conditional mortality analyses suggest that some patients surviving the initial period after their ICH may be able to survive to go on to undergo transplantation. Infection is among the most common causes of morbidity and mortality in LVAD patients¹⁴ and there was a trend towards higher conditional mortality after infection-associated ICH than spontaneous ICH although this did not reach significance. Acute infection has been previous identified as a contributing factor for ICH in patients with an LVAD^15,16. Potential mechanisms of infection-associated ICH include infective endocarditis and infectious aneurysm. Previous research has demonstrated a bimodal distribution of the incidence of ICH after LVAD implantation with an initial peak in the perioperative period followed by another rise at approximately one year⁶. Part of this previously observed delayed occurrence of ICH after LVAD implantation may be related to development of device-associated infection. In our cohort, only one infection-associated ICH (4%) occurred within 30 days of LVAD implantation and median time from LVAD implantation to ICH was longer in infection-associated ICH (252 days) compared with spontaneous (100 days). Infection is frequently a delayed complication of LVAD implantation¹⁷ and longer duration of LVAD support significantly increases the risk of associated infection¹⁸. Furthermore, patients with long-term LVAD support have a higher prevalence of cerebral microbleeds¹⁹ and greater numbers of microbleeds are associated with higher prevalence of ICH among LVAD patients²⁰. In this latter study, patients with bacteremia or pump-pocket infection had a greater number of cerebral microbleeds, which may contribute to the association between acute infection and ICH.

Additional factors previously associated with LVAD-ICH include female sex^3,21 and coagulopathy/anticoagulation¹⁵. While the characteristics of patients in our study (including sex, race, medical and surgical history, type of LVAD device, and INTERMACS status) did not vary between spontaneous and infectious ICH, history of hypertension was more prevalent in spontaneous ICH than infectious (92% vs 69%). Hypertension has been associated with ICH among LVAD patients²². History of hypertension prior to LVAD implantation may also place patients at risk of ischemic stroke,²³ which could lead to development of ICH via hemorrhagic transformation after initiation of antithrombotic therapy.

Traumatic ICH is not uncommon in patients after LVAD implantation and is likely related to the continuous need for anticoagulation. In addition to anticoagulation, non-pulsatile blood flow is associated with autonomic dysregulation^24,25 which can lead to orthostatic hypotension²⁶. This orthostatic intolerance as well as the common occurrence of cardiac arrhythmias in this population²⁷ may contribute to the elevated risk for syncope with subsequent trauma in LVAD patients. Additionally, patients suffering traumatic ICH were on average older at time of ICH than spontaneous or infectious ICH and this may be related to an older population’s predisposition to falls.

One limitation of our study was the retrospective review of prospectively collected data. Additionally, our study represents the experience of a single center and may not be reflective of practice at other institutions. The ability to compare between different etiologies of ICH as well as ICH subtypes was limited due to low sample size. While CT scans were assessed by a neuroradiologist and neurologist obtained history and a study investigator reviewed the records and images, ICH etiologies may have been misclassified. It is difficult to truly ascertain whether all the traumatic ICH were in fact caused by the head trauma and it is instead possible that some of these may have been spontaneous ICH that subsequently resulted in head trauma. However, these patients were evaluated at the time by a vascular neurologist who felt that trauma was the most likely etiology. While there are multiple mechanisms by which device-associated infection may cause intracranial hemorrhage (including septic embolism and mycotic aneurysm), it is not possible to definitively know whether infection-associated hemorrhages were caused by one of these mechanisms or were spontaneous hemorrhages in patients with concomitant device-associated infection. Not all patients had blood cultures at the time of ICH (usually in cases of nonsurvivable ICH leading to withdrawal of care less than a day from presentation) and therefore some infection-associated ICH may have been misclassified as spontaneous due to lack of data.

CONCLUSIONS:

Mortality is high in patients with LVAD-associated ICH regardless of the presence or absence of acute device-associated infection or antecedent trauma. Spontaneous ICH occurred earlier after LVAD implantation in comparison to infectious ICH, but mortality and other characteristics of ICH and its subtypes did not vary between ICH etiologies. Further prospective and registry-based research is necessary to better understand characteristics and outcomes after different etiologies and subtypes of LVAD-associated ICH in order to understand prognosis and optimal clinical management.