Variable
Cohort 1 (BSA < 0.7 m2)
Cohort 2 (BSA 0.7–1.5 m2)
Age in months – median (range)
11.7 (2.6–45.6)
111.2 (50.8–191.8)
Weight in kilograms – median (range)
9.2 (3.6–13.6)
30.7 (16–58.1)
Congenital heart disease – number (%)
3 (12)
6 (25)
Dilated cardiomyopathy/myocarditis – number (%)
19 (79)
17 (71)
INTERMACS profile – number (%)
1
11 (46)
13 (54)
2
13 (54)
11 (46)
Preoperative ECMO – number (%)
6 (25)
8 (33)
VAD configuration – number (%)
LVAD
17 (71)
14 (58)
Biventricular VAD
7 (29)
10 (42)
Survival was significantly higher for each of the EXCOR cohorts than for the historical ECMO control groups (log rank p < 0.001 for cohort 1 and 2 compared to ECMO) (◘ Fig. 37.1). Furthermore, 90% of patients supported with the EXCOR survived to transplant or were weaned from the device compared to 67–75% of ECMO matched controls. The most common serious adverse events were major bleeding (46%), infection (56%), and stroke (29%). Adverse events rates were 0.07 events/patient day for cohort 1 and 0.08 events/patient day for cohort 2. The cumulative survival benefit and adverse event profile of the EXCOR met the burden of proof for HDE approval by the FDA of probable benefit and safety, and thus the EXCOR was given HDE approval in December 2011. HDE approval requires less than 4000 implantations a year and that the implanting institutions keep an active IRB protocol. The official approval stated, “EXCOR is intended to provide mechanical circulatory support as a bridge to cardiac transplantation for pediatric patients…..who are candidates for cardiac transplant….” One should note that the FDA chose not to limit the use of the EXCOR to the parameters of the study cohort and allowed wide application of the EXCOR (i.e., patients with complex congenital heart disease). It also did limit the use of the device to transplantation centers.
Fig. 37.1
Kaplan-Meier estimates of freedom from primary end point (death for ECMO cohort, death or weaning from device with unacceptable neurologic outcome for EXCOR cohorts). Cohort 1 includes patients with a body surface area < 0.7 m2, and cohort 2 includes patients with a body surface area 0.7–1.5 m2 (From Fraser et al. [9])
Subsequent analyses of the IDE cohort and patients who did not meet the study inclusion criteria and who received the device under compassionate use protocols further explored risk factors for poor outcome including the complex congenital heart disease and the impact of device use on post-transplant survival.
37.3 Combined IDE and Compassionate Use Data
The data from the IDE study was quite definitive to demonstrate the superiority of the EXCOR over ECMO in bridging patients to transplantation. However, it did not provide a complete picture of the US EXCOR experience, “the real-world experience,” as it only reported on a third of all the patients implanted with the EXCOR during the study period. Also, it did not include important patient cohorts such as those with complex congenital heart disease or with significant end-organ dysfunction. Nor did it represent a wide sample of all institutions placing the device. The broader perspective was provided by a comprehensive analysis of the combined IDE and compassionate use data [6]. This study included all children who received an EXCOR in the USA or Canada between May 2007 and Dec 2010. The pump sizes available and adverse event definitions were the same as for the IDE study.
Two hundred and four children were enrolled (only one in Canada) during the study period, of which 95 (47%) patients received the EXCOR at a non-IDE center. Compassionate-use patients comprised 67% of the entire cohort and were younger, smaller, more likely to have congenital heart disease, more likely to be supported on ECMO and had worse end-organ function (◘ Table 37.2).
Table 37.2
Patient characteristics for IDE criteria and compassionate use EXCOR cohorts
Variable | IDE criteria cohort (n = 68) | Compassionate use cohort (n = 136) |
|---|---|---|
Age in years – median (IQR) | 2.1 (0.6–7.1) | 1.4 (0.4–4.4) |
Weight in kilograms – median (IQR) | 10 (6.5–16.6) | 9.4 (5.7–14.8) |
Congenital heart disease – number (%) | 10 (15) | 49 (36) |
Single Ventricle – number (%) | 0 (0) | 19 (14) |
Preoperative ECMO – number (%) | 18 (27) | 65 (48) |
GFR categories – number (%) | ||
<30% predicted | 0 (0) | 11 (8) |
30–99% predicted | 11 (16) | 32 (24) |
>99% predicted | 57 (84) | 93 (68) |
Of the 204 children supported with the EXCOR during the study, 75% survived to transplant or recovery. Children in the compassionate use cohort were less likely to reach transplantation (53% versus 85%) and were more likely to die (34% versus 7%; p < 0.01). Predictors of early mortality while supported on the EXCOR included lower weight, higher bilirubin, and BiVAD support, while bilirubin extremes and decreased glomerular filtration rate (GFR) predicted late mortality. Decreased GFR was the strongest overall predictor of death. Of note, death also occurred more commonly in children with congenital heart disease, children on ECMO at the time of implantation, at non-IDE centers, and at centers with lower volume; however, these differences were not statistically significant after adjusting for the other risk factors for mortality.
The most common cause of death (n = 51) in the study was neurologic insult (n = 17, 33%), with thromboembolic strokes significantly outnumbering hemorrhagic stroke. Stroke severity data was only available for the IDE cohort; thus, further information regarding stroke severity could not be gleaned from this data. Neurologic insult was followed by respiratory failure (n = 6, 12%), bleeding (n = 3, 6%), multisystem organ failure (n = 3, 6%), right ventricular failure (n = 3, 6%), and other less common events. Of note, death due to device malfunction occurred in only one patient.
Most interesting is that even though mortality differed significantly between the IDE and non-IDE patients, the incidence of all morbidities was not different, in fact almost identical. Centers with increased experience (VAD implants >5 for an institution) was associated with significantly improved survival compared to institutions with less experience. Therefore, it was not the avoidance of complications but the recognition of complications and the ability to mitigate them in experienced centers that was associated with improved outcomes. This experience, we hypothesize, was a result of the formation of a multidiscipline team with protocol-driven care.
This study [6] was not only notable for providing the largest, most comprehensive analysis of EXCOR support to date, but it also identified the importance of neurologic insults on clinical outcomes in addition to identifying risk factors for mortality while supported on the EXCOR. Subsequent analyses would provide more in-depth analysis of neurologic events [10], patient size [11], and BiVAD use [12] on outcomes.
37.4 Neurologic Events
Neurologic insult was the most common cause of mortality among patients supported with an EXCOR in both the IDE study [9] and combined IDE/compassionate use study [6]. Neurologic dysfunction was also a frequent cause of morbidity (29% of patients suffered a neurologic insult) among patients who survived to transplant in these cohorts. Given the importance of this topic, Jordan et al. [10] specifically analyzed the stroke data available from the combined IDE and compassionate use cohorts.
Of the 204 children included in the study, 59 (29%) experienced at least one neurologic event (total of 73 events) for a total event rate of 0.51 events/100 patient days. The majority of events were ischemic strokes (n = 47, 89%) with few hemorrhagic strokes (n = 5, 9%). The risk for a neurologic event was not evenly dispersed as the majority of events (45 [62%]) occurred during the first month of support. Hazard function analysis showed the risk per patient day was highest during the perioperative period and slowly decreased to “baseline” by day 50 of support.
Assessing the severity of neurologic impairment following the events was limited to some degree by the lack of chronologic neurologic testing, only 25 of the patients who experienced a neurologic event had at least one Pediatric Stroke Outcome Measure (PSOM) completed, and only 15 had PSOM data following the event. That said, gross estimates of the severity could be gleaned from what data did exist. Of the patients with long-term PSOM data following the event, 11 of 22 (50%) had no or mild PSOM abnormalities. Furthermore, 56% of patients who experienced an event subsequently underwent heart transplantation.
Attempts to identify risk factors for stroke using pre-implant demographic and clinical variables yielded limited results. Female gender and a history of pump change due to thrombus were the sole risk factors for neurologic insult identified in the final multivariable model. The pump exchange data was further limited by the lack of data regarding the timing of pump change and the neurologic insult. Single centers have demonstrated improvement in stroke-related outcomes with increased institutional experience [13]. Centers have also explored alternate management strategies including anti-inflammatory therapy [14] and more frequent device pump exchanges [15] to try to decrease the risk of stroke; however, the success of these interventions have not yet been demonstrated. Finally, recent data from the Berlin database demonstrates there is considerable variation (tenfold) in the incidence of stroke [16] and the risk was not explained by center volume. This suggests it may be possible to improve outcomes through shared learning and establishing best practices from centers with the lowest risk of stroke.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
