More than 20 years’ experience of left ventricular assist device implantation at a non-transplant Centre

Abstract

Objectives. Over recent decades implantable left ventricular assist devices (LVAD) have increased the possibility of improved survival in patients with advanced heart failure who also benefit from a better quality of life. The aim of this retrospective survey was to review the clinical results of LVAD implantation at a low-volume non-transplant centre (Linköping, Sweden) between 1993 and 2016. Our aim was also to assess the mortality and morbidity rates associated with implantation of three LVAD versions at our centre, and to compare our results with those from transplant centres. Design. A retrospective cohort study was performed examining the medical records of patients who had a HeartMate^® (HMI, HMII, HMIII) LVAD implanted as a bridge to heart transplantation (BTT) or as destination therapy (DT) at the University Hospital, Linköping. Results. Our main finding was a survival to heart transplantation rate of 82% among our BTT LVAD patients. The most common adverse event among our patients was infection. A higher frequency of temporary dialysis was seen in the HMII group compared to the HMI group, and the frequency of right ventricular failure was higher in our HMII material. Conclusions. Our data suggests that patients requiring long-term LVAD support can safely have their device implanted and cared for at a non-transplant centre.

Keywords:

• LVAD
• long-term
• HeartMate
• low-volume
• non-transplant centre

Introduction

Over the recent decades, the implantable left ventricular assist device (LVAD) has increased the possibility of improved survival with increased quality-of-life in patients with advanced heart failure. The implantation and management of LVADs at non-transplant centres has been somewhat controversial due to concerns about adequate training, experience and resources [1]. With the development of continuous flow pumps, and after the commercial approval of the HeartMate^® II LVAD (Thoratec Inc, Pleasanton, CA) for bridge to transplantation and as destination therapy, the number of implantations at non-transplant centres has grown [2,3,4].

The University Hospital in Linköping is a teaching hospital where approximately 800 cardiac and 250 general thoracic surgical procedures are performed each year. Our referral centre for heart transplantation is the University Hospital in Lund. Lund and Linköping are the two surgical centres for patients with advanced heart failure in the south respectively south-east regions of Sweden. The total number of heart transplantations in Sweden has increased, and by 2016 there were 6.4 heart transplantations per million inhabitants.

The aim of this retrospective survey was to review the clinical results of LVAD implantation performed at our non-transplant, low-volume centre between 1993 and 2016, to assess the mortality and morbidity rates associated with the implantation of three LVAD versions, and to compare our results with those from contemporary studies.

Material and method

A retrospective cohort study was performed examining the medical records of patients who had a LVAD implanted as a bridge to heart transplantation (BTT) or as destination therapy (DT) at the University Hospital, Linköping, between February 1993 and November 2016. The HeartMate^® (version I, II and III) was the LVAD used during this time period. The HMI^® was implanted between February 1993 and May 2005, the HMII^® from June 2005 to February 2016 and thereafter the HMIII^® was used. Ethics approval was granted by the Regional Ethical Committee.

Study patients

The local database was retrospectively reviewed and all patients who required a LVAD during the study period were identified.

At the time of decision to implant a LVAD, the majority of patients had been accepted for heart transplantation (Tx), others after contact with the transplant centre in Lund.

Data collection

Baseline characteristics, preimplantation clinical values, and outcomes were obtained from the medical records.

The primary outcome was mortality. Secondary outcomes were morbidity, length of stay on the intensive care unit (ICU), and total length of stay in hospital.

Statistical analysis

For comparisons between the HMI and HMII groups, ordinal data were analysed using the non-parametric Mann-Whitney U-test. A p value <.05 was considered statistically significant. The HMIII patients are presented separately in the tables. No statistical analysis was carried out on this group since it was too small.

Results

Preoperative demographics and aetiology

Demographics and aetiology are presented in Table 1. A total of 49 patients had a LVAD implanted as a bridge to heart transplantation (n = 47) or destination therapy (n = 2). Two patients had a LVAD implanted for extended BTT. One extended BTT case was a cancer patient who, after 5 years of freedom from cancer, was accepted for heart transplantation. The other extended BTT case is still receiving pump support while on the transplantation waiting list. One of the BTT patients was re-evaluated and assigned destination therapy. Sixteen patients received a HMI^® and 30 patients a HMII^®. To date only three patients have had a HMIII^® implanted and are therefore briefly presented separately and in Tables 1 and 2. Between 1993 and November 2016, eighty-seven patients from the south-east region had received a heart transplant, of which thirty-five patients had a LVAD as a bridge to transplantation.

Table 1. Demographics and aetiology, n (%).

	HMI n = xx (%)	HMII n = xx (%)	p	HMIII n = xx
Number of patients	16	30		3
Male	12 (75)	23 (77)	.91	1
Age, median years [range]	49 [15–63]	54 [26–62]	.23	59 [51–65]
Non-ischaemic cardiomyopathy	12 (75)	21 (70)	.73	2
Ischaemic cardiomyopathy	4 (25)	8 (27)	.91	1
Myocarditis	0	1 (3)	.49	0
Left ventricular heart failure	14 (87)	24 (80)	.54	2
Biventricular heart failure	2 (13)	6 (20)	.54	1
Previous cardiac surgery	3 (19)	7 (23)	.73	0
Previous myocardial infarction	4 (25)	10 (33)	.57	1
NYHA, median [range]	4 [3–4]	4 [3–4]	.73	3
Duration of heart failure, median years [range]	1.6 [0.03–12]	4 [0.02–21]	.35	4 [0.03–21]
Previous CVI	1 (6)	2 (7)	.69	0
Preoperative temporary assist device	2 (12)	4 (13)	.95	1
Preoperative mechanical ventilation	3 (19)	4 (13)	.64	1
Euroscore II, median [range]		9.4% [2.4–44.4]		25.5 [12–26]
INTERMACS level (median)	3	4	.07	3
Level 1 Critical cardiogenic shock, n (%)	2 (12)	4 (13)		1
Level 2 Progressive decline, n (%)	3 (19)	3 (10)		0
Level 3 Stable but inotrope dependent, n (%)	8 (50)	6 (20)		2
Level 4 Non-Inotrope dependent, n (%)	3 (19)	17 (57)		0

NYHA: New York Heart Classification; CVI: cerebrovascular insult.

Table 2. Postoperative outcome, n (%) or median days [range].

	HMI n = 16	HMII n = 30	p	HMIII n = 3
Mortality before Tx	1 (6)	7 (25)	.13
ICU-stay	15 [6–79]	15 [4–43]**	.59	9 [7–19]
Length of stay hospital	89 [37–249]	39 [14–105]**	.00006*	30 [20–32]
Pump time to Tx	232 [55–873]	160 [65–2484]	.8
Ongoing BTT	0	1(3)		3
Ongoing time	0	1252		135 [34–233]
Destination therapy (DT)	0	2 (7)		0
Ongoing DT		0
Time on DT		638 [159–1117]

*Significant, p < .05.

**One patient excluded due to mortality in the OR (n = 29).

Tx: Heart transplant.

The dominant indication for implantation was non-ischaemic cardiomyopathy. Fourteen patients (87%) in the HMI group and 24 (80%) in the HMII group had isolated left ventricular failure, and the others biventricular failure. Preoperative short-term mechanical assist devices (IABP or Impella^®) were used in 2 (12%) HMI patients and 4 (13%) HMII patients, as bridge-to-bridge (BTB) support. No significant differences in demographics or comorbidity were seen between the two groups. The median INTERMACS score was 3 in the HMI group and 4 in the HMII group (p = .07).

Outcomes

Mortality

Outcomes on mortality and adverse events are presented in Tables 2 and 3. Eight (19%) BTT patients died before heart transplantation, 1 patient (6%) in the HMI group and 7 patients (25%) in the HMII group (p = .13). Apart from one, all deaths occurred within 90 days after LVAD implantation.

Table 3. Adverse event rates for HMI and HMII patients*.

	HMI 11 pt-yrs			HMII 26 pt-yrs
Adverse events	n (%)	Events, n	Events/pt-yrs	n (%)	Events, n	Events/pt-yrs	p
Right heart failure	4 (25)	4	0.36	12 (40)	12	0.47	.32
Reoperation bleeding	1 (6)	1	0.09	8 (27)	11	0.43	.10
Pump thrombosis	0	0	–	5 (17)	6	0.23	.09
Major CVI	2 (13)	3	0.27	5 (17)	5	0.19	.51
Thromboembolism	2 (13)	3	0.27	3 (10)	3	0.12	.81
Haemorrhage	0	0	–	2 (7)	2	0.07	.31
Gastrointestinal bleeding	0	0	–	1 (3)	1	0.04	.49
Septicaemia/Infection	9 (56)	9	0.81	12 (40)	14	0.54	.42
Device infection	4 (25)	5	0.45	8 (27)	13	0.50	.91
Temporary dialysis	0	0	–	6 (20)	6	0.23	.06

*HMIII is not listed here due to few events related to these patients.

pt-yrs: patient-years.

The two patients on DT died 159 and 1117 days after pump implantation respectively.

Support times

The time between LVAD implantation and transplantation was a median of 232 (55 to 873) days in the HMI group and 160 (65 to 2484) days in the HMII group (p = .8). One patient has been on pump support 1252 days and is still on the transplantation waiting list.

Nine patients with a HMI vented electrical system were discharged to out-patient care compared to all patients in the HMII group. Six patients with a pneumatic HMI device were hospitalized, due to device design, until Tx, for a median of 89 days (55 to 173).

Length of stay on the ICU was a median of 15 (6 to 79) days in the HMI group and a median of 15 (4 to 43) days in the HMII group (p = .59). Thirteen (43%) HMII patients with prolonged ICU stay suffered from respiratory failure requiring ventilator support >1 week or tracheostomy. Six patients (20%) in the HMII group had acute kidney injury (AKI) requiring temporary dialysis compared to none in HMI group (p = .06).

Total length of stay in hospital was a median of 89 (37 to 249) days in the HMI group and median of 39 (14 to 105) days in the HMII group (p < .001).

Haemorrhagic complications

Eight (27%) of the HMII patients were reoperated within the first postoperative week due to bleeding compared to 1 (6%) in the HMI group (p = .1). One HMII patient needed blood transfusion due to epistaxis three months after implant and a successful cholecystectomy was also performed on this patient. He had CYP2C9 polymorphism resulting in sensitivity to warfarin. Another HMII patient suffered from gastrointestinal bleeding four months after implantation.

All HMII patients received low-molecular heparin which was subsequently substituted by warfarin when appropriate. The frequency of warfarin treatment in the HMI group was 18% because of the device’s design that required less anticoagulation [3]. Antiplatelet therapy with acetylsalicylic acid was used (HMI 75%, HMII 100%), and 20% of the HMII patients also received clopidogrel.

Cerebrovascular insult

Two (13%) HMI patients had a major cerebrovascular insult compared to 5 (17%) HMII patients (p = .51). One HMI patient developed left-sided paresis and grand mal seizures three days after implant, and another had an occipital embolism three months after surgery that led to imminent intracranial herniation requiring neurosurgical intervention. One HMII patient died from a right-sided cerebral haemorrhage after 80 days on pump; probably due to anticoagulation treatment. Another patient suffered an occipital thrombosis and a third patient had infarction of the spleen and a TIA, both were urgently transplanted.

Death due to pump thrombosis-related CVI was seen in two HMII patients. Another patient successfully received a stent in an obstructed outflow graft day 815. However five days later he had an acute subdural bleed due to amyloid plaque and needed neurosurgery. He survived, recovered and is now on the transplantation waiting list again.

Pump thrombosis

Five (17%) HMII pump thromboses were confirmed by clinical presentation, laboratory tests indicating haemolysis, and pump speed change testing (i.e. a ‘‘ramp’’ study) [5]. One HMII patient died of cerebral infarction due to pump thrombosis embolism 274 days after implantation. Another had a previously unknown antiphospholipid syndrome and developed pump thrombosis causing a CVI day 52. Due to pump malfunction, this patient developed multiple organ failure, and died day 72.

Two patients with pump thrombosis without systemic embolisation received new HMII^® after 377 and 77 days respectively. One of them developed a new pump thrombosis and was urgently transplanted.

Infection

Culture-verified infection, or sepsis (defined as two of the following: positive blood culture; temperature >39 or <35 °C; or leukocytosis >12 or <3 × 10*9/L) during hospital stay, was the most common adverse event per patient-years among all LVAD patients. It was more frequently seen in the HMI group, 9 patients (56%), than in the HMII group, 12 patients (40%), p = .42. One HMII patient had several episodes of septicaemia caused by methicillin-resistant S. aureus, but successfully underwent Tx after 185 days.

In the HMI group, 3 (19%), had culture-verified driveline infection and one was reoperated due to infection in the pump pocket. Among the HMII patients, 8 (27%) developed a cultured-verified driveline infection (p = .42) that was treated with intravenous antibiotics. Early infection with coagulase-negative Staphylococci occurred 3-5 weeks postoperatively, and Staphylococcus aureus driveline infection occurred 3–9 months after surgery.

Pump endocarditis

HMI^® pump endocarditis involving the inflow biological valve conduit occurred in two patients (13%). The first patient died in the OR during pump replacement because of severe right ventricular failure (79 pump days). The other patient had the pump explanted after 232 days and a new HMI^® successfully implanted. The patient was transplanted after a further 641 days on the LVAD.

Right ventricular (RV) failure

Twelve (40%) HMII patients developed clinical signs of right ventricular (RV) failure that was confirmed by invasive and non-invasive measurements. They were treated with inodilators and inhaled pulmonary vasodilators more than 5 days postoperatively, or by a right ventricle (RV) assist device. In the HMI group, 4 (25%), developed RV failure (p = .32).

Four (33%) HMII patients with RV failure died before transplantation. Three of the deceased were in need of a RV assist device. The first HMII patient died of persistent vasoplegia in the OR after acute RV failure despite implantation of a RV assist device. The second had myocarditis and a transient episode of RV failure in the perioperative period. He received a RV assist device after 42 days but died during the procedure due to bleeding complications. The third had a preoperative Impella^® device improving his moderate preoperative RV failure, elevated PA-pressure and increased PVR. He received a RV assist device at the same time as the LVAD. After 16 days, weaning from the RV assist device was possible and necessary due to pump thrombosis. He was initially haemodynamically stable on low-dose epinephrine. However, five days later he suffered a circulatory collapse due to right ventricular failure and died day 25.

Prior to implantation, another HMII patient had an elevated PA pressure, increased PVR, and ascites. After surgery he received pharmacotherapy for RV failure, including inhaled pulmonary vasodilators, but he eventually died from multiorgan failure. One HMII DT patient had known RV failure prior to implantation, with liver dysfunction and ascites. He died 159 days after LVAD implantation.

Pump failure

In the HMI group, one pneumatic device had to be replaced by an electrical device after pump failure. One HMII DT patient died day 1117 after implantation due to cable rupture from excessive wear.

HMIII

To date, three patients have received a HMIII^® pump. Demographics are shown in Table 1 and outcome in Table 2. Patient no 2 had ischaemic-related cardiogenic chock and both Impella^® and ECMO as BTB. She was successfully bridged to LVAD support after 11 days.

At the time of writing, all three patients are on pump support after 34, 135 and 233 days’ support, and have been discharged from hospital without major adverse events. Patient no 1 had epistaxis during week 1 that required intervention. Patient no 3 needed a reoperation day 2 due to bleeding. No source was found, but tests showed minor coagulopathy.

Discussion

Our main finding was an overall survival to heart transplantation rate of 82%. One patient with extended BTT regime is still on pump support after 1252 days. Three HMIII^® pumps have so far been implanted and are still in use with no major adverse events.

Fifteen patients (94%) survived to transplantation in the HMI group. This figure compares well to the INTERMACS registered one-year survival or survival to transplant rate of 76% 2007–2008 for all patients in the register with a pulsatile LVAD as bridge to transplantation [6].

For patients in the HMII group, the survival to transplantation rate was 75% including one extended BTT patient still on the waiting list. This is lower than the INTERMACS register’s 80% 1-year survival with continuous flow LVADs support, between 2008 and 2014 [7]. This might be due to chance arising from our rather small material. A relatively high right ventricular failure rate was observed, which is a poor prognostic factor, and may contribute to our slightly lower survival rate.

The study includes seven patients (23%) with INTERMACS Levels 1–2 and six patients with Level 3. Boyle et al showed that patients with INTERMACS Levels 1 and 2–3 had survival rates of 51% and 69% respectively. This should be compared with a 95% survival rate for patients with Levels 4–7 [8]. In the Sixth INTERMACS Report, Kirklin et al. showed that patients with INTERMACS Levels 1 and 2 that receive a LVAD continue to have poor survival rates. Compared to the early days, the proportion of patients receiving a HM^® pump in a stable, inotrope-dependent state (Level 3) has increased [9], but our material is small and one death has a large effect on outcome.

Cause of death among our patients follows the same pattern as that in the INTERMACS register where cerebrovascular insult, multiple organ failure and right ventricular failure are the most common causes of death the first 3 months after implantation [7,9]. In our material there were two deaths due to cerebrovascular insult, both related to pump thrombosis; day 72 and day 274. One patient died of cerebral haemorrhage caused by anticoagulation day 80. After 24 months, infection and MOF increase as causes of death, whilst intracranial causes remain at the same level [7,9]. Over the years, high CVI and pump thrombosis rates have led to changes in anticoagulation strategies, and intracranial haemorrhage has become a more prominent cause of morbidity/mortality than infarct in patients with continuous flow devices [3]. The frequency of thrombocyte inhibitor therapy over time was the same in all three groups, anticoagulation with warfarin was introduced following HMII^® implantation, and in selected cases some other form of thrombocyte inhibitor was used.

We observed a 40% postoperative right ventricular failure rate among the HMII patients. Five of the deceased patients had right ventricular failure and died after a median of 36 (0 to 79) days. This is high compared to other studies where rates between 7% and 25% have been reported [4,6,10,11]. RV failure after LVAD is a poor prognostic factor and results in an up to 6-fold increase in mortality risk, and is one of the main reasons for prolonged hospitalisation [1]. Cowger et al validated the HeartMate II Risk Score and concluded that preoperative age, serum albumin, creatinine, INR and centre implantation experience were factors affecting survival [12]. Kirklin et al. also included Model of End-stage Liver Disease (MELD), including raised serum bilirubin, ascites and high body mass index as predictors of outcome [9,13,14]. It is important to select suitable patients. Three of our LVAD patients that died of RV failure had clinical signs of RV failure, such as liver dysfunction and ascites, prior to implantation. Several studies have shown that patients with biventricular failure have a worse outcome after LVAD implantation, with higher on-pump mortality, a poor survival to transplantation rates and lower survival after transplantation [14,15]. Among our data we can see some LVAD patients that would not have been candidates for LVAD treatment today due to right ventricular failure pre-implant. This probably contributes to the high event rate of RV failure and mortality rate observed. Some of these patients would have been BIVAD candidates but others not though they were not BTT candidates, only DT candidates. This experience has made the team more cautious about these patients and treatment with LVAD is not an alternative for RV failure patients. Today more frequent discussions with BIVAD centres are made concerning the above mentioned patient category. Transplant candidates, with pulmonary hypertension (PH) and an elevated PVR that is reversible in a maximal dilation test, have been shown by i.e. Liden et al. to reduce elevated PVR during LVAD treatment prior to Tx [16]. A full workup around the patient is crucial in order to decide on LVAD implantation and it should not be a short-term alternative assist device in patients in need of urgent Tx.

Regarding total length of hospital stay, device design and function differ between the HMI, HMII and HMIII pumps, which, together with changes in hospital routine, explains the longer admission times after HMI implantation [3]. It is more difficult to explain why length of stay on the ICU does not differ significantly between the HMI and HMII groups, though we speculate that the higher prevalence of haemorrhagic complications, temporary dialysis and right ventricular failure in the HMII group might have affected length of stay on the ICU.

Cowger et al, using the INTERMACS database, showed that centres with a 5-year implant volume less than 15 had higher 90-day mortality rates [12]. A retrospective analysis by Lietz et al. [17] on the impact of centre volume on the outcome of LVAD implantation as destination therapy concluded that 1-year survival seemed to improve as centres gained experience, i.e. >9 destination therapy patients. However, they were not able to conclude which aspects played the most critical role: patient selection, approach to surgical and postoperative care, or long-term medical management [17]. Both authors strongly emphasize the multidisciplinary approach when caring for these heart failure patients [12,17]. A review by Chowdhury et al. [18] came to the conclusion; that individual surgeon volume has a greater effect on outcome than centre volume. They also concluded that the contribution of each specialist member of a multidisciplinary team responsible for the patient, independently improved patient outcome [18]. In our department we have a multidisciplinary team comprising representatives from the Departments of Cardiology, Cardiac Surgery, Anaesthesia and Perfusion, all of whom participate in perioperative care. Weekly meetings are held, and our Cardiology department has all facilities for both pre- and post-transplant patient care. We consider the teamwork the most crucial factor for a LVAD programme to work at a non-transplant centre together with a close collaboration with a transplant centre. An increasing amount of LVAD candidates, not the least for DT, require additional LVAD centres. For the patients the closeness to a hospital with LVAD programme is a major asset regarding follow-ups and readmissions. Although drawbacks might be few patients making it harder to gain experience. The need for a dedicated team and not very seldom the involvement of other specialist such as neurologists, infection disease specialists etc. makes it harder for the often smaller non-transplant hospital to live up to the same excellent standard as the transplant centres. Also, as discussed in an article by Rossing et al. the readmissions are frequent and resource intensive, though rather short, with a length of stay of 5 days per admission [19].

Our outcomes are comparable with contemporary multicentre studies and outcome data in the INTERMACS register [4,9,10,20,21]. Complications and their prevalence are acceptable and similar to those in recent publications [4,9,19,22]. The overall event rate per patient-years was higher in the HMII group. Rossing et al. show a mean readmission rate of 2.9 per patient per year with infections as the most common cause [19]. This corresponds well to our material where infection and device infection were the most frequent events. Time to transplantation in our material was shorter than the average of around 200 days reported in the literature [23]. Again the small number of patients in this survey, and thereby the impact of a single divergent patient, may be a reason.

Our LVAD programme provides stable outpatient back-up for patients in the vicinity with ongoing LVAD support; probably increasing their quality of life.

Our data suggest that patients requiring long-term LVAD support can have their device safely implanted and cared for in a non-transplant centre.

Limitations

A limitation of this study is its retrospective design. The small sample size is also a major limitation. Others include a varying quality of the documentation and absence of laboratory data for some of the early HMI patients. The long inclusion time might lead to many biases and changes in treatment strategies influenced by different caretakers, technical advancement and gained experience.

Conclusion

Our main finding was a survival to heart transplantation rate of 82% among our BTT LVAD patients. We observed no significant difference in length of stay on the ICU between the HMI and HMII groups, though there was a significantly shorter total length of stay in hospital in the HMII group; probably due to changes in patient care over time. The most common adverse event among our patients was infection. A higher frequency of temporary dialysis was seen in the HMII group than in the HMI group, and the frequency of right ventricular failure was higher in our HMII material than those reported from contemporary studies. No significant difference concerning mortality or adverse events was observed between the groups. Our data suggest that patients requiring long-term LVAD support can have their device safely implanted and cared for in a non-transplant hospital.

Disclosure statement

The authors report no conflicts of interest.