HVAD Usage and Outcomes in the Current Pediatric Ventricular Assist Device Field: An Advanced Cardiac Therapies Improving Outcomes Network (ACTION) Analysis

Abstract

Advanced Cardiac Therapies Improving Outcomes Network (ACTION) is the first pediatric ventricular assist device (VAD) quality improvement network (46 centers). We aimed to describe outcomes with the HeartWare HVAD from ACTION centers. Patients with an HVAD implant in the ACTION registry (April 2018–April 2020) were analyzed. Baseline characteristics, adverse events, and survival were described. There were 50 patients implanted with a HVAD during the study period [36 cardiomyopathy, 8 congenital heart disease (CHD), and 6 other] and 21 (42%) had a prior sternotomy. Median age (range) was 12.9 years (3.4–19.1), body surface area was 1.3 m² (0.56–2.62), and weight was 41.8 kg (12.8–135.3). Most were INTERMACS profile 2 (n = 26, 52%). Mechanical ventilation and ECMO were used pre-HVAD in 13 (26%) and 6 (12%), respectively. Median time on VAD was 71 (5–602) days. Survival was 96% at 1 year; 3 deaths were recorded, all of whom had CHD (p = 0.001). Neither ECMO nor mechanical ventilation were associated with death (p > 0.29). Most frequent AEs were bleeding (n = 7, 14%) and infection (n = 7, 14%). Stroke was rare (n = 2, 4%). ACTION Network HVAD outcomes were excellent, with 96% survival at 1 year and only 4% occurrence of stroke. Major bleeding and infection were the most common adverse events.

Pediatric Heart Failure and VAD

Despite the increasing burden of acute and chronic heart failure in children, there remain few options for those who progress to end-stage heart failure, often necessitating listing for heart transplantation.^1–3 Before the availability of ventricular assist device (VAD) support as a bridge to transplantation, the waitlist mortality was unacceptably high, especially for younger patients. Utilization of VADs in children resulted in significant improvements in outcomes compared with the pre-VAD era.^4–7 This improvement in waitlist outcomes resulted in increased utilization of VAD support from approximately 4% in the era between 1993 and 2003 to 33% in the current era.^8–10

While the availability of the pulsatile Berlin Heart EXCOR VAD improved waitlist and posttransplant outcomes compared with extracorporeal membrane oxygenation (ECMO), waitlist morbidity and mortality remained high.^11–13 The development of continuous flow devices showed superior stroke free survival compared with pulsatile devices. However, the large size of the early generation continuous flow devices allowed only adult-sized or larger pediatric patients to be implanted and benefit from improved outcomes. Newer generation devices are considerably smaller and allow for implant in much smaller children. The HeartWare HVAD System (Medtronic) is currently the smallest implantable commercially available continuous flow VAD, which makes it attractive to those caring for children with heart failure. The HVAD allows small children an opportunity to be discharged from the hospital while awaiting transplant. The first worldwide report on the use of the HVAD showed that survival in children with a body surface area (BSA) <1 m² were no different than in children with a BSA >1 m².¹⁴ This analysis from the ACTION registry extends the recent reporting on HVAD-specific outcomes with high-quality, adjudicated, real-world data from a large multicenter cohort.

Methods

Design of the Network

The Advanced Cardiac Therapies Improving Outcomes Network (ACTION) was designed as a learning network as defined by the Institute of Medicine in 2007.¹⁵ The ACTION multicenter learning network was developed in 2017 with a mission to improve clinical outcomes and the patient/family experience for children with heart failure, currently consists of 46 member institutions (as of July 2020).

Overall data collection by ACTION has been described previously.¹⁶ All patients undergo informed consent before enrollment in ACTION registry data collection. The ACTION registry collects data prospectively from all VAD implantations in participating centers. All adverse events (AE) were prospectively defined and users were trained in the definitions. Comprehensiveness of data entry is verified by comparison to center-provided volume data and by internal data auditing. Additionally, key outcomes including stroke, cause of mortality, and device malfunction are 100% centrally adjudicated, as well as a 15% sampling of other AE. Feedback is provided to sites when adjudication results in re-classification of events. Completion rates of key fields are provided in Table (Supplemental Digital Content 1, https://links.lww.com/ASAIO/A594).

HVAD-specific Data Extraction

Data on patients with a history of HVAD implant were extracted from the ACTION registry for this analysis. ACTION centers submitted data from HVAD implants between April 1, 2018, and April 13, 2020. Demographics, preoperative and perioperative characteristics, and postoperative outcomes were described.

Statistical Methodology

Descriptive statistical methods included median (minimum-maximum range), and N (%). Time to event analysis was performed using the competing outcomes methodology. After VAD implantation, potential outcomes include death on device, transplantation, myocardial recovery with explantation, or remaining alive on the device. Kaplan-Meier estimates for each individual outcome are misleading because they are competing. Competing outcomes analysis allows for the simultaneous time-related analysis of multiple discrete outcomes and uses parametric modeling of these multiple discrete outcomes and a robust time-related analysis of multivariable risk factors, allowing for patient-specific predictions of outcomes after VAD implantation.¹⁷ This method has been previously described by McGiffin et al.¹⁸ AE were described using events per 100 patient-months. Adverse event definitions are in Document (Supplemental Digital Content, https://links.lww.com/ASAIO/A593). Patient characteristics and death were compared using univariable chi-square analysis or Fisher Exact Test.

Results

Preimplant Patient Characteristics

During the study period, there were a total of 50 HVAD implants in 50 patients. The median age was 12.9 years (3.4–19.1 years), with only two patients >18 years of age. Sex was female in 48%. Median BSA was 1.33 m² (0.56–2.6 m²) and the BSA was <1 m² in 28% of this cohort. Within the cohort, 24% of patients were <25 kg. The median weight was 42 kg (12.8–135.3 kg). The majority of patients (72%) had a diagnosis of dilated cardiomyopathy (DCM) and 16% had a diagnosis of congenital heart disease (CHD). Detailed preimplant characteristics are shown in Table 1.

Table 1. - Preimplant Characteristics

Total Patients and HVAD Implants (N, %)	50
Age in yrs (median, range)	12.9 [3.38–19.1]
Age >18 yrs (N, %)	2 (4%)
Gender—female (N, %)	24 (48%)
BSA (median, range)	1.33 [0.56–2.62]
Weight in kg (median, range)	41.8 [12.8–135.3]
Diagnosis (N, %)
Dilated cardiomyopathy without NMD1	34/50 (68%)
Dilated cardiomyopathy with NMD	2/50 (4%)
CHD, biventricular circulation	3/50 (6%)
CHD, Fontan circulation	5/50 (10%)
Fontan with systemic RV	3/5
Device strategy (N, %)
Bridge to transplant	29 (58%)
Bridge to candidacy/decision	18 (36%)
Bridge to recovery	2 (4%)
Chronic therapy	1 (2%)
INTERMACS profile at implant (N, %)	49 Subjects
1	11 (22%)
2	26 (52%)
3	10 (20%)
4 or greater	2 (4%)
Days on ECMO before implant (N, %)	6/50 (12%)
3 d	2 (4%)
6 d	1 (2%)
8–15 d	3 (6%)
Estimated GFR implant (median, range)*	125 (45.5–855.25)
Dialysis dependent at implant	1 (2%)
Total bilirubin at implant (median, range)	0.9 [0–9]
Diagnosis of PLE	0
Mechanical Ventilation (in 7 d before implant)	13 (26%)
TPN full/partial (in 7 d before implant)	20 (40%)
Inotropes (in 7 d before implant) (N, %)
No Inotrope	2 (4%)
1 Inotrope	22 (44%)
2 Inotropes	20 (40%)
3 Inotropes	6 (12%)
Prior sternotomies (median, range)	3 [1–6] for 21 patients

*Estimated GFR calculated using the Schwartz equation and reported as ml/min/1.73m².

BSA, body surface area; CHD, congenital heart disease; ECMO, extracorporeal membrane oxygenation; GFR, glomerular filtration rate; NMD, neuro-muscular dystrophy; PLE, protein losing enteropathy; RV, right ventricle; TPN, total parenteral nutrition.

The device strategy was bridge to transplant or candidacy in 94%. Intermacs profile at implant was profile 1 in 22% (n = 11), profile 2 in 52% (n = 26), and profile 3 in 20% (10). Only 2 patients were profile 4 or greater. ECMO support was required in 6 patients before implant and 2 patients had a history of temporary VAD placement before HVAD (1 Impella and 1 Rotaflow). Mechanical ventilation was required in 13 patients (26%) in the 7 days before implant and 48 patients (96%) were on at least 1 inotrope.

Postimplant Characteristics, Hospital Course, and Outcomes

The median device days were 71 (5–602). The median initial postimplant intensive care unit length of stay and hospital length of stay (before discharge or transplant) were 18 (3–190) and 38 (10–190) days, respectively. Temporary biventricular VAD support was required in 2 patients at time of HVAD implant [1 Centrimag (explanted at 35 days) and 1 Rotaflow (explanted at 3 days)], while one patient required Centrimag RVAD for right heart failure (RHF) on day 13 of HVAD and was transplanted after 18 days of BiVAD support. Systemic anticoagulation and antiplatelet use are shown in Table 2. There were 22 patients (44%) discharged with the device and 6 patients (12%) were transplanted during the same hospitalization as HVAD implantation. There were a total of 9 readmissions in 6 patients, including 5 unplanned readmissions for warfarin dose adjustment, observation for possible pump thrombosis, bilevel positive airway pressure initiation, abdominal pain, and observation. Competing outcomes are shown in Figure 1. There was a 94% positive outcome at study completion with 34 patients (68%) having undergone transplant, 11 (22%) alive on device, 1 (2%) explanted for recovery, 3 (6%) died, and 1 (2%) alive when transferred to a non-ACTION center.

Table 2. - Anticoagulant and Antiplatelet Agents Used

	N (% of total)
Unfractionated heparin	30 (60%)
Low-molecular weight heparin	6 (12%)
Bivalirudin	21 (42%)
Warfarin	36 (72%)
Aspirin	46 (92%)
Clopidogrel	0
Dipyridamole	0

Figure 1.: A, Competing outcomes analysis for Alive on device, transplanted, death, and recovery are shown. B, Percentages of patients with the competing outcomes are shown.

The only characteristic associated with death was a preimplant diagnosis of CHD (3/8 with CHD died vs. 0/36 with DCM; p = 0.001) (Table 3). All 3 of the deaths occurred in patients with a functionally univentricular heart. The causes of death on device included device malfunction (n = 1) and hemorrhagic stroke (n = 2).

Table 3. - Associations Between Patient Characteristics and Survival

	Group 1 (Died) N = 3	Group 2 (All Others) N = 47	p
Age in yrs (median)	16.3	12.9	0.475
Gender—male (n)	3	23	0.086
Female (n)	0	24
BSA (median)	1.64	1.3	0.791
Diagnosis (n)
CHD	3	5	0.001
DCM/myocarditis	0	36
Other	0	5
RCMP	0	1
ECMO preimplant (n)	0	6	0.509
Device strategy preimplant (n)
BTC	0	18	0.510
BTR	0	2
BTT	3	26
Chronic therapy	0	1
Prior medical support
Dialysis/RRT	0	1	0.799
Mechanical ventilation	0	13	0.290
Comorbidities
History of CVA	0	3	0.652
Neurologic brain injury not CVA	0	1	0.799
Total device duration (median)	24	72	0.624
Post implant hospital stay (median)	24	38	0.193

BSA, body surface area (m²); BTC, bridge to candidacy; BTR, bridge to recovery; BTT, bridge to transplant; CHD, congenital heart disease; CVA, cerebrovascular accident.; DCM, dilated cardiomyopathy; ECMO, extra corporeal membrane oxygenation; RCMP, restrictive cardiomyopathy; RRT, renal replacement therapy.

Adverse Events

There were 41 AE that occurred in 24 patients over 4323 device days. Table 4 shows specific AE and their rates. Major bleeding and major infection were the most common AE, at a rate of 5.64 events per 100 patient-months each. Of major bleeding events, 6 of 7 events occurred in the first month following implantation. Of bleeding events, sources included intrathoracic (3), epistaxis (1), wound (1), and other (2). Surgical management of bleeding was required in two events, one of which was a gastrointestinal bleed. There was a single device-related infectious adverse event in the form of a driveline infection, while all others were localized, nondevice related infections or sepsis. Respiratory failure, RHF, and neurologic events were the next most common AE, with three of each event. Of the three patients with RHF, two were management medically and one required temporary RVAD support. Of the neurologic events, there were two strokes (1 ischemic and 1 hemorrhagic) and one extra-axial hemorrhage. There was a single pump thrombosis event, but no device exchange was required.

Table 4. - Adverse Events

Event Type, n (%)	Events (n = 41)	Per 100 Patient-Months
Bleeding	7 (17%)	4.94
Major infection	7 (17%)	4.94
Localized nondevice infection	5	3.53
Sepsis	1	0.71
Driveline infection	1	0.71
Right heart failure	3 (7%)	2.12
Right heart failure requiring RVAD	1	0.71
Respiratory failure	3 (7%)	2.12
Total neurologic events (including stroke)	3 (7%)	2.12
Neurologic dysfunction	3 (7%)	2.12
Stroke	2	1.41
Extra-axial hemorrhage	1	0.713
Device malfunction	2 (5%)	1.41
Pump thrombosis	1 (2%)	0.71
Arrhythmias	2 (5%)	1.41
Renal dysfunction	2 (5%)	1.41
Pleural effusion	2 (5%)	1.14
Hemolysis	1 (2%)	0.71
Thromboembolism—venous	1 (2%)	0.71
Thromboembolism—arterial non-CNS	1 (2%)	0.71
Pericardial effusion with tamponade	1(2%)	0.71
Hepatic dysfunction	0	0.00
Pump malfunction requiring exchange	0	0.00
Other	2 (5%)	1.41

Percentiles rounded to nearest whole number.

CNS, central nervous system; RVAD, right ventricular assist device.

Discussion

This initial report from the ACTION registry experience with the HVAD implantable continuous flow ventricular assist device shows that survival one year following implantation was 96% and was 94% for the overall study period. Most patients in this series had a preimplant diagnosis of dilated cardiomyopathy and on inotropic support at time of implant. Overall, 74% of the cohort were either Intermacs profile 1 (cardiogenic shock) or 2 (progressive decline on inotropes), consistent with a relatively clinically ill patient population at time of implant. While most patients had a BSA of >1 m², successful bridge to transplantation was possible in patients with a BSA <1 m² and there was no association between death and BSA. Temporary RVAD support was infrequent, which is in line with previous outcomes from the Pediatric Interagency Registry for Mechanical Circulatory Support (Pedimacs) registry,¹⁹ but the median initial intensive care unit stay following implantation was 18 days. RHF was only reported in 6% of the ACTION cohort, which is substantially less than the Pedimacs registry, which reported nearly half of durable continuous flow LVAD patients had moderate to severe RHF early after implant. Whether this difference in reported RHF represents a difference in patient clinical profiles or eras is unclear as the definition of RHF used was similar. Nearly half of patients (44%) were discharged from the hospital after implant within ACTION, similar to the Pedimacs registry experience from 2012 to 2017, which reported 47% of children were discharged after continuous flow VAD support.²⁰ Readmission was more frequent in the Pedimacs experience (67 vs. 27%), which may be reflective of increased institutional experience over time with patient preparation for discharge and outpatient VAD management. A diagnosis of CHD was associated with death after implant, while preimplant ECMO was not associated with death. Conversely, CHD was not a risk factor for mortality in a recent Pedimacs report, while ECMO before implant was associated with higher mortality.¹¹ The overall small size of the cohort and the number of events within this registry report limit the ability to draw broad conclusions about associated risks. As the ACTION cohort increases with time, more distinct patterns of patient risk profiles may be apparent to help guide clinical decision making.

The results of this analysis suggest that the outcomes of HVAD support continue to improve.^6,9,11,21,22 Mortality by competing outcomes compared favorably to prior reports for durable continuous flow devices in children.^11,14,23 Additionally, AE were lower than previous reports despite a dedicated adjudication process in the ACTION registry. The Pedimacs registry reported that between 2012 and 2016, 10–12% of children supported with an intracorporeal continuous flow device experience stroke as opposed to only 4% in this series.^20,23 Similarly, neurologic dysfunction event rates were 4.9 events per 100 patient-months in the Pedimacs registry versus 2.12 events per 100 patient-months in ACTION.²³ Pedimacs has reported that 21–27% of implantable continuous flow device patients had an infectious adverse event, while only 14% in our cohort had an infectious adverse event (12–17.4 vs. 4.94 events per 100 patient-months).^20,23,24 Similarly, Pedimacs reported that 23–28% of implantable continuous flow device patients had a major bleeding event, while only 16% in our cohort had a reported major bleed with an event rate of 14.9–16.4 versus 4.94 events per 100 patient-months, respectively.^20,23 These differences in outcomes likely reflect important trends in HVAD utilization over time, including patient selection and perioperative management in children over different eras. Improved sharing of institutional experience, targeted quality improvement initiatives, and harmonization of VAD management, such as through the ACTION learning network, may also have positively affected outcomes in more recent years.

The current study is limited due the retrospective analysis of registry data; so only associations can be determined. Likewise, the number of patients and events are small, so this report is only descriptive in nature and there were not enough events to power meaningful multivariable analysis. Therefore, the association between CHD and death may not be significant when controlling for other variables. Finally, despite a rigorous adjudication process, the ACTION registry is dependent on its participating centers to report events to the registry, so under-reporting of AE is a possibility and data from each contributing site may have not been complete. Despite these limitations, this ACTION registry experience gives important insight to the trends in patient outcomes on continuous flow VAD support.

Conclusion

This initial description of HVAD use in the ACTION Network showed excellent survival for pediatric patients despite a heterogeneous patient population with respect to size and preimplant diagnosis. Bleeding and infection were the most common AE and stroke was infrequent, with event rates improved compared with previous era pediatric VAD registry reports. Hospital discharge after implant is possible in children with HVAD support, although only reported in about half of patients, suggesting opportunity for improvement in this population.

Acknowledgments

ACTION Learning Network Investigators (Listed alphabetically): Mohammed Absi, MD: Le Bonheur Children’s Hospital, Memphis, TN; David W. Bearl, MD: Monroe Carell Jr. Children’s Hospital at Vanderbilt, Nashville, TN; Umar Boston, MD: Le Bonheur Children’s Hospital, Memphis, TN; Jennifer Conway, MD: Stollery Children’s Hospital, University of Alberta, Edmonton, AB, Canada; Nhue L. Do, MD: Monroe Carell Jr. Children’s Hospital at Vanderbilt, Nashville, TN; John C. Dykes, MD: Stanford Children’s Health and Lucile Packard Children’s Hospital, Palo Alto, CA; Joshua Friedland-Little, MD: Seattle Children’s Hospital, Seattle, WA; Beth Hawkins, NP: Boston Children’s Hospital, Boston, MA; Osami Honjo, MD, PhD: The Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Aamir Jeewa, MD: The Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; John S. Kim, MD: University of Colorado Denver, Anschutz Medical Campus, Children’s Hospital of Colorado, Aurora, CO; Steven Kindel, MD: Children’s Hospital of Wisconsin, Milwaukee, WI; David M. Kwiatkowski, MD, MS: Stanford Children’s Health and Lucile Packard Children’s Hospital, Palo Alto, CA; Jodie Lantz, PCNS: University of Texas Southwestern Medical Center, Children’s Health, Dallas, TX; Angela Lorts, MD: Cincinnati Children’s Hospital, University of Cincinnati, Cincinnati, OH; Desiree Machado, MD: Shands Children’s Hospital, University of Florida Health, Gainesville, FL; Chad Y. Mao, MD: Children’s Healthcare of Atlanta, Atlanta, GA; Christopher E. Mascio, MD: Children’s Hospital of Philadelphia, Philadelphia, PA; Lindsay J. May, MD: Primary Children’s Hospital, Salt Lake City, UT; Mary Mehegan, RN: St. Louis Children’s Hospital, St. Louis, MO; Max B. Mitchell, MD: Children’s Hospital of Colorado, University of Colorado Denver, Anschutz Medical Campus, Aurora, CO; David L.S. Morales, MD: Cincinnati Children’s Hospital, University of Cincinnati, Cincinnati, OH; Robert A. Niebler, MD: Children’s Hospital of Wisconsin, Milwaukee, WI; Matthew J. O’Connor, MD: Children’s Hospital of Philadelphia, Philadelphia, PA; David M. Peng, MD: C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor, MI; Michelle Ploutz, MD: Primary Children’s Hospital, Salt Lake City, UT; David N. Rosenthal, MD: Stanford Children’s Health and Lucile Packard Children’s Hospital, Palo Alto, CA; Ming-Sing Si, MD: C.S. Mott Children’s Hospital, University of Michigan, Ann Arbor, MI; Joshua Sparks, MD: Norton Children’s Hospital, Louisville, KY; David Sutcliffe, MD: University of Texas Southwestern Medical Center, Children’s Health, Dallas, TX; Christina VanderPluym, MD: Boston Children’s Hospital, Boston, MA; Philip Thrush, MD: Ann & Robert H. Lurie Children’s Hospital, Chicago, IL; Farhan Zafar, MD: Cincinnati Children’s Hospital, University of Cincinnati, Cincinnati, OH; Matthew D. Zinn, DO: Children’s Hospital of Pittsburgh, Pittsburgh, PA.