The Majority of Pediatric Fontan Patients Have Excellent Post-Transplant Survival

Structured Abstract:

Objective:

Many pediatric Fontan patients require heart transplant, but this cohort is understudied given difficulty identifying these patients in national registries. We sought to characterize survival post-transplant in a large cohort of pediatric Fontan patients.

Methods:

UNOS & PHIS were used to identify Fontan heart transplant recipients age<18 years (n=241) between 2005-2022. Decompensation was defined as the presence of ECMO, ventilation, hepatic/renal dysfunction, paralytics, or TPN at transplant.

Results:

Median age at transplant was 9[IQR=5-12] years. Median waitlist time was 107[37-229] days. Median volume across 32 center was 8[3-11] cases. Nearly half (n=107, 45%) of recipients had 1A/1 initial listing status. Sixty-four (28%) were functionally impaired at transplant, 10 (4%) were ventilated, and 18 (8%) had VAD support. Fifty-nine patients (25%) had hepatic dysfunction & 15 (6%) had renal dysfunction. Twenty-one (9%) were TPN-dependent. Median postop stay was 24[14-46] days and in-hospital mortality was 7%. Kaplan-Meier analysis showed 1- & 5-year survival of 89%[95%CI=85-94%] & 74%[81-86%], respectively. Kaplan-Meier of Fontan patients without decompensation (n= 154) at transplant demonstrated 1- & 5-year survival of 93%[88-97%] and 88[82-94%]. In-hospital mortality was higher in decompensated patients (11% vs. 4%, p=0.023). Multivariable analysis showed that decompensation predicted worse PTS (HR=2.47 [1.16 – 5.22], p=0.018), whereas older age at transplant predicted superior PTS (HR=0.89/year [0.80 – 0.98], p=0.019).

Conclusions:

Pediatric Fontan post-transplant outcomes are promising, although early mortality remains high. For non-decompensated pediatric patients at transplant without end-organ disease (>63% of cohort), early mortality is circumvented and post-transplant survival is excellent and similar to all pediatric transplantation.

Introduction

Over the past five decades, Fontan palliation has emerged as mainstay of surgical treatment for children born with single ventricle physiology.¹ The Fontan procedure has afforded decades of life to patients with disease processes that would otherwise be universally fatal in early childhood, with over 80% of Fontan patients surviving >20 years after surgery.² However, the altered physiology of total cavopulmonary connection leads to several downstream sequelae that result in “Fontan failure.”³ Up to 10% of Fontan patients will die or require transplant in childhood, and over half will progress to that stage by the age of 40.³ Many Fontan patients have become critically ill by the time they undergo transplant—including development of hepatic or renal dysfunction, inotropic or ventilatory requirements, or failure to thrive—despite convincing evidence that these modifiable risk factors predict worse post-transplant survival.^{4, 5} Although transplant is frequently required in this population, the majority of data available on long-term outcomes is limited to single-institution case series because of difficulty identifying these patients in national registries.^6-23 The United Network for Organ Sharing (UNOS) database, for example, which collates donor and recipient data on all transplants in the United States, lacks granular data on patient diagnosis beyond a binary variable indicating “congenital heart disease.” Other, larger or multi-institutional studies have lacked long-term survival data or combine pediatric and adult Fontan patients, despite potentially differing mechanisms of failure between these two cohorts.²⁴ As such, the authors sought to identify a large, multi-institutional cohort of pediatric Fontan patients who underwent heart transplant in order to better elucidate their long-term outcomes.

Methods

The aim of this retrospective, multi-institutional study was to describe post-transplant survival in pediatric Fontan patients. The primary outcome of interest was post-transplant survival. Secondary outcomes evaluated included intensive care unit (ICU) length of stay, hospital length of stay, postoperative dialysis requirement, rejection during index hospitalization, and in-hospital mortality. Additional analyses evaluated the impact of multiple modifiable risk factors at time of transplant on short- and long-term survival. The study received institutional review board approval (CCHMC IRB# 2018-6837; date of approval: 10/29/2018) and was granted a waiver of informed consent as all data included in both databases utilized are publicly available and de-identified.

Given the limitations of the UNOS database with regard to preoperative diagnosis discussed above, the authors used methodology previously described and validated to merge the UNOS database with the Pediatric Health Information System (PHIS) database, an administrative database of 49 major pediatric hospitals in the United States with billing & coding information on all admissions to those hospitals.²⁵ The granular detail available in the PHIS database allows it to better identify patient diagnoses, and merging the dataset with the UNOS database affords the ability to harness the extensive survival data available therein. In previously performed studies validating this method, there were no clinical or demographic differences noted between the UNOS/PHIS merged cohort and the overall UNOS cohort.²⁶

All patients age <18 years with an inpatient admission in the PHIS database who had an International Statistical Classification of Diseases and Related Health Problems (ICD) 9 procedure code (35.94) or ICD-10 procedure code (021608Q; 021609Q; 02160AQ; 02160JQ; 02160KQ; 02160ZQ; 02160ZR) indicating performance of a Fontan procedure between 2005-2022 were first identified. Subsequently, all patients age <18 years in the UNOS database who underwent heart transplant during the same time were identified. The two databases were merged; only patients identified in both datasets were included in the final analysis (n=241).

A number of clinical and demographic factors were described at the time of transplant. The transplant experience was divided evenly across the period of study into three eras (Era 1 = 2005-2010, Era 2 = 2011-2016, Era 3 = 2017-2022). Race & ethnicity groups were determined from variables available in UNOS. Renal dysfunction was determined by estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m², as calculated using the bedside Schwartz equation from creatinine level. Hepatic dysfunction was defined as total serum bilirubin >1.2 mg/dL at transplant. Patients were determined to be dependent on total parenteral nutrition (TPN) or requiring paralytics by clinical transaction classification (CTC) codes included in PHIS—patients with charges for paralytic agents ≥3/5 pre-transplant days were considered to require paralytics and patients with charges for “fat emulsions” or “hyperalimentation solutions” without any meals or enteral nutrition recorded for 5/5 pre-transplant days were considered TPN dependent. Functional status was defined by the patient’s assigned Karnofsky Performance Scale (KPS) rating given in UNOS and divided into groups based on functional ability. Inotropic support was provided as a binary (Y/N) variable at UNOS indicating the use of inotropes (dobutamine, dopamine, epinephrine, or norepinephrine) at time of transplant. Ventricular assist devices (VAD) included left, right, and biventricular support.

Statistical analyses were performed in IBM SPSS Statistics for Windows, version 27 (IBM Corp., Armonk, NY, USA) and R (R Foundation for Statistical Computing, Vienna, Austria). Patient characteristics were first described for the overall cohort. The cohort was then divided into two groups; “decompensated” and “non-decompensated.” Patients were considered clinically decompensated if they presented with any one of the following at the time of transplant: extracorporeal membrane oxygenation (ECMO) requirement, ventilator support, hepatic dysfunction, renal dysfunction, induced paralysis, or TPN dependence. They were considered non-decompensated if they presented with none of the above factors. Given the high rate of inotropic support in the overall cohort, this was not used as dividing variable. Comparative analyses between the two cohorts were coducted using Mood’s median test for continuous variables and Pearson’s chi-square test with Yates’ continuity correction or Fisher’s exact test (if cross-tabulation cell counts were less than 5) for discrete variables. Post-transplant survival was then modeled using Kaplan-Meier (KM) analysis for the overall cohort as well as the sub-cohorts. Survival was compared between cohorts using the Wilcoxon test. Conditional 6-month and 1-year survival were modeled as well for all groups. Subsequently, univariable & multivariable Cox proportional hazards regressions were used to determine significant predictors of post-transplant survival. Backward stepwise elimination using a p-value of 0.10 as a cutoff for removal from the model was used for variable selection when building multivariable regressions. Two multivariable Cox regression models were performed to determine predictors of post-transplant survival. Both models included age, male sex, Caucasian race, operative era, ischemic time, 1A/1 listing status, inotropic support, blunt trauma donor mechanism of death, waitlist time, and VAD at time of transplant as initial variables. Model A included clinical decompensations separately (TPN, renal dysfunction, hepatic dysfunction, ventilator support, and ECMO support), whereas model B included a binary variable of decompensation (Y/N).

Results

Of the 5,554 heart transplants performed in PHIS-reporting hospitals during the study period, 241 were identified as having undergone prior Fontan palliation. All 241 Fontan patients were subsequently identified in the UNOS database. The median age at time of transplant was 9 [IQR 5 – 12] years and the median time from the final stage palliation was 5.2 [2.0 – 9.3] years (Table 1). Twenty-two patients (9%) had undergone Fontan revision at some point prior to transplant. The most common fundamental diagnosis was hypoplastic left heart syndrome (HLHS; n=144, 60%) followed by double outlet right ventricle (DORV; n=25, 10%) and atrioventricular canal defects (n=15, 6%). Other indications represented in the cohort were tricuspid atresia, pulmonary atresia without ventricular septal defect (VSD), common ventricle, and congenitally corrected transposition of the great arteries (TGA). The number of Fontan transplants increased by era, with 29 performed in Era 1 (12%), 94 in Era 2 (39%), and 118 in Era 3 (49%).

Table 1.

Demographic & clinical characteristics of Fontan HTx patients as well as postoperative outcomes. kg, kilogram; ECMO, extracorporeal membrane oxygenation; KPS, Karnofsky Performance Scale; TPN, total parenteral nutrition; VAD, ventricular assist device; PRA, panel-reactive antibody; VSD, ventricular septal defect; TGA, transposition of great arteries; ICU, intensive care unit.

	Fontan (n = 241)
Age at transplant (years)	9 [5 – 12]
Gender (Male)	147 (61%)
Race
Caucasian	181 (75%)
African American	31 (13%)
Asian/Pacific Islander	1 (0%)
American Indian	0 (0%)
Other	28 (12%)
Ethnicity (Hispanic)	38 (16%)
Weight (kg)	23.9 [17.0 – 38.4]
Transplant year	2016 [2014 – 2019]
Era
2005 – 2010	29 (12%)
2011 – 2016	94 (39%)
2017 – 2022	118 (49%)
Center Experience (cases)	8 [3 – 11]
Days on waitlist	107 [35 – 229]
ECMO at transplant	2 (1%)
Functional Status
Performs normal activity (KPS=80-100%)	70 (29%)
Capable of self-care (KPS=60-70%)	98 (41%)
Requires care/assistance (KPS=40-50%)	43 (18%)
Incapable of self-care (KPS= 10-30%)	21 (9%)
Age <1 year/unknown	9 (4%)
Ventilation at transplant	10 (4%)
Inotropes at transplant	117 (49%)
Renal dysfunction at transplant	15 (6%)
Pre-transplant dialysis	5 (2%)
Liver dysfunction at transplant	59 (25%)
Concomitant liver transplant	5 (2%)
Paralytics	2 (1%)
TPN-Dependence	21 (9%)
VAD at transplant	18 (8%)
Peak calculated PRA	1 [0 – 57]
Ischemic time (hours)	4.1 [3.5 – 4.7]
UNOS Status 1A/1 at listing	107 (45%)
Time from Fontan (days)	1,905 [728 – 3,412]
Prior Fontan revision	22 (9%)
Indication for Fontan
Hypoplastic left heart syndrome	144 (60%)
Tricuspid atresia	10 (4%)
Pulmonary atresia w/o VSD	9 (4%)
Common ventricle	10 (4%)
Double outlet right ventricle	25 (10%)
Atrioventricular canal defects	15 (6%)
Corrected TGA	4 (2%)
Combined defect	4 (2%)
Unknown defect	7 (3%)
Postoperative length of stay (days)	24 [14 – 461
ICU length of stay (days)	14 [6 – 361
Postoperative dialysis requirement	27 (11%)
Rejection during index hospitalization	47 (20%)
In-hospital mortality	16 (7%)

The majority of the cohort was male (n=147, 61%) and white (n=181, 75%). Median number of cases performed at a single center over the study period was 8 [3 – 11]. Patients spent a median of 107 [35 – 229] days on the waitlist and 107 (45%) were UNOS status 1A/1 at the time of transplant. The majority of patients had a KPS of 60% or greater (n = 168, 70%). Ten patients (4%) were on a ventilator at transplant, 2 (1%) were paralyzed, 2 (1%) were on ECMO, 15 (6%) had renal dysfunction, and 5 (2%) were on dialysis. Fifty-nine (25%) patients had liver dysfunction at transplant, 117 (49%) required inotropic support at time of transplant, 21 (9%) were on TPN, and 18 (8%) had a VAD.

For the overall Fontan cohort, postoperative length of stay was 24 [14 – 46] days and ICU length of stay was 14 [6 – 36] days. Twenty-seven (11%) patients required dialysis postoperatively after their transplant and 47 (20%) had an episode of rejection during their index hospitalization. In-hospital mortality during the index transplant admission was 7% (n=16).

When divided into sub-groups, 87 patients (36%) met criteria to be considered clinically decompensated as described above (Table 2). Notably, in comparing the two groups (decompensated vs. non-decompensated), the patients were demographically similar with similar age at transplant (8 [4 – 14] years vs. 9 [5 – 12] years, p=0.938) and similar gender (n=56, 64% vs. n=91, 59%; p=0.503) & racial (56 Caucasian patients, 64% vs. 101 Caucasian patients, 66%; p=0.960). Transplant surgeries were similarly divided between eras as well for both groups (p=0.345). Although there was a trend toward shorter waitlist time for decompensated (75 [18 – 151] days) vs. non-decompensated (117 [49 – 251] days), this was not a statistically significant finding (p=0.134). The patients in the decompensated group, as expected, had worse functional status at the time of transplant, with only 56% (n=49) having a KPS of 60% or greater, relative to 77% (n=119) in the non-decompensated group (p=0.002). UNOS listing status differed between groups as well, with decompensated patients more likely to be listed as status 1A/1 than non-decompensated patients (n=50, 60% vs. n=57, 37%; p<0.001). In-hospital mortality was greater in the decompensated group vs. non decompensated group (n=10, 11% vs. n=6, 4%; p=0.023), post-operative length of stay was longer (28 [20 – 56] days vs. 21 [13 – 41] days, p=0.039), and there was a trend toward longer ICU stay as well (18 [9 – 54] days vs. 13 [5 – 33] days; p=0.053), although the latter finding lacked statistical significance. Although dialysis requirements (n=13, 15% vs. n=14, 9%; p=0.229) were higher in the decompensated group, this finding was not statistically significant.

Table 2.

Demographic & clinical characteristics of critically ill vs. non-critically ill Fontan HTx patients as well as postoperative outcomes. KPS, Karnofsky Performance Scale; VAD, ventricular assist device, PRA, panel reactive antibody; ICU, intensive care unit.

	Non-Decompensated (n=154, 64%)	Decompensated (n=87, 36%)	p-value
Age at transplant	9 [5 – 12]	8 [4 – 14]	0.938
Gender (Male)	91 (59%)	56 (64%)	0.503
Race
Caucasian	101 (66%)	56 (64%)	0.960
African American	23 (15%)	8 (9%)	0.281
Ethnicity (Hispanic)	26 (17%)	22 (25%)	0.161
Weight (kg)	24 [18 – 38]	24 [16 – 40]	0.798
Transplant year	2017 [2014 – 2019]	2016 [2013 – 2019]	0.669
Era			0.345
2005 – 2010	15 (10%)	14 (16%)
2011 – 2016	62 (40%)	32 (37%)
2017 – 2022	77 (50%)	41 (47%)
Days on waitlist	117 [49 – 251]	75 [18 – 151]	0.134
Functional Status			0.002
Performs normal activity (KPS=80-100%)	48 (31%)	22 (25%)
Capable of self-care (KPS=60-70%)	71 (46%)	27 (31%)
Requires care/assistance (KPS=40-50%)	25 (16%)	18 (21%)
Incapable of self-care (KPS= 10-30%)	8 (5%)	13 (15%)
Age <1 year/unknown	2 (1%)	7 (8%)
Inotropes at transplant	73 (47%)	44 (51%)	0.735
VAD at transplant	7 (5%)	6 (8%)	0.610
Peak calculated PRA	0% [0% – 50%]	21% [0% – 64%]	0.262
Ischemic time (hours)	4.1 [3.4 – 4.5]	4.2 [3.5 – 4.9]	0.558
UNOS Status 1A/1 at listing	57 (37%)	50 (60%)	<0.001
Time from Fontan (days)	2095 [920 – 3419]	1658 [620 – 3365]	0.248
Prior Fontan revision	15 (10%)	7 (8%)	0.837
Postoperative length of stay (days)	21 [13 – 41]	28 [20 – 56]	0.039
ICU length of stay (days)	13 [5 – 33]	18 [9 – 54]	0.053
Postoperative dialysis requirement	14 (9%)	13 (15%)	0.229
Rejection during index hospitalization	30 (19%)	17 (20%)	0.999
In-hospital mortality	6 (4%)	10 (11%)	0.023

Upon survival analysis (Figure 1), a KM curve of post-transplant survival for the overall cohort demonstrated mean survival of 11.3 [95% CI = 10.5 – 12.2] years, with 1-year and 5-year survival of 89% [85 – 93%] and 81% [75 – 88%], respectively (Figure 2). The mean six-month conditional survival was 12.6 [11.8 – 13.4] years and the mean one-year conditional survival was 12.7 [11.9 – 13.5] years. Six-month conditional survival at 5 years was 90% [85 – 97%] and one-year conditional survival at 5 years was 91% [86 – 97%]. When comparing the sub-cohorts, decompensated patients had a significantly worse mean survival relative to non-decompensated patients (9.8 [8.3 – 11.4] years vs. 12.4 [11.6 – 13.2] years, p=0.004) with one-year survival of 83% vs. 92% and five-year survival of 74% vs. 88% (Figure 3). However, when looking at conditional survival in the sub-cohorts, differences in mean survival significantly improved with six-month conditional survival (11.9 [10.4 – 13.4] years vs. 13.2 [12.6 – 13.8] years; p=0.301) and one-year conditional survival (11.9 [10.4 – 13.4] years vs. 13.4 [12.8 – 13.9] years; p=0.118) and were no longer statistically significant.

Figure 1.

Graphical abstract depicting study design and key outcomes. (CI 95%)

Figure 2.

Kaplan-Meier curves of post-transplant survival in years for the overall Fontan cohort. (A) Overall post-transplant survival. (B) Post-transplant survival conditional on survival at six months. (C) Post-transplant survival conditional on survival at one year. %, percent; #, number. (CI 95%)

Figure 3.

Kaplan-Meier curves of post-transplant survival in years comparing the decompensated and non-decompensated Fontan cohorts. (A) Overall post-transplant survival. (B) Post-transplant survival conditional on survival at six months. (C) Post-transplant survival conditional on survival at one year. %, percent; #, number. Non-Decomp., non-decompensated; Decomp., decompensated. (CI 95%)

In model A, which included each clinical decompensation as a separate predictor variable, TPN (HR=9.75, [95% CI=2.39 – 39.82], p=0.002), renal dysfunction (HR=6.14 [1.85-20.36], p=0.003), and listing status 1A/1 (HR=3.28[1.17-9.19], p=0.024) were significant predictors of post-transplant mortality whereas inotropic support (HR=0.22 [0.06 – 0.089], p=0.033) and older age (HR=0.88/year [0.80-0.97], p=0.011) was associated with decreased post-transplant mortality (Table 3). In model B, in which clinical decompensation was treated as a composite binary variable, the decompensation variable was predictive of post-transplant mortality (HR=2.47 [1.16 – 5.22], p=0.018) and older age (HR=0.89/year [0.80 – 0.98], p=0.019) was associated with decreased post-transplant mortality.

Table 3.

Univariable and multivariable Cox proportional hazards regression of post-transplant survival. Variables were included both multivariable models except where noted. ^aIncluded in multivariable model A only. ^bIncluded in multivariable model B only. TPN, total parenteral nutrition; VAD, ventricular assist device; ECMO, extracorporeal membrane oxygenation; HR, hazard ratio; CI, confidence interval.

Variable	Univariable Model		Multivariable Model A		Multivariable Model B
	HR[95% CI]	p-value	HR[95% CI]	p-value	HR[95% CI]	p-value
Age (years)	0.92 [0.85 – 1.00]	0.058	0.88 (0.80 – 0.97)	0.011	0.89 [0.80 – 0.98]	0.019
Sex (male)	0.96 [0.50 – 1.87]	0.913
Race (Caucasian)	1.07 [0.53 – 2.13]	0.858
Operative Era; 2005-2010 vs…
2011-2016	0.50 [0.20 – 1.22]	0.128
2017-2022	0.67 [0.27 – 1.67]	0.393
Ischemic time (hours)	1.09 [0.80 – 1.49]	0.578
Status 1A/1	1.55 [0.79 – 3.02]	0.200	3.28 [1.17 – 9.19]	0.024
Inotropes	0.61 [0.31 – 1.19]	0.147	0.44 [0.21 – 0.94]	0.033
Hepatic dysfunctiona	1.45 [0.71 – 2.95]	0.305
Renal dysfunctiona	2.48 [0.96 – 6.42]	0.061	6.14 [1.85 – 20.36]	0.003
TPN requirementa	2.19 [0.89 – 5.41]	0.088	9.75 [2.39 – 39.82]	0.002
Ventilatora	1.11 [0.26 – 4.64]	0.889
VAD	1.13 [0.27 – 4.73]	0.872
ECMOa	0.05 [0.01 – 260]	0.652
Waitlist time (days)	1.00 [1.00 – 1.01]	0.089
Donor cause of death (blunt trauma)	1.38 [0.71 – 2.68]	0.343
Decompensationb	2.56 [1.32 – 4.98]	0.005			2.47 [1.16 – 5.22]	0.018

Discussion

The present study demonstrates a large, multi-institutional cohort of Fontan patients with long-term post-transplant survival data. Although prior studies have evaluated the outcomes in this cohort, multi-institution data linked with long-term survival has been lacking. The post-transplant survival in the overall cohort was promising, with one- and five-year survival above 80%. There has been a historical reticence to refer Fontan patients for heart transplant due to a perceived higher risk of poor outcomes in this population, an assumption based on the increased surgical complexity of heart transplant in a patient with a single ventricle.²⁷ However, the findings in the above analysis—with survival rates similar to if not better than the one and five year survival rates for all pediatric heart transplants in the Registry of the International Society for Heart and Lung Transplantation (ISHLT)—lend credence to the belief that optimal outcomes can be achieved in specialized centers and to the shift over the past twenty years toward increasing referral and transplantation.²⁸

There was a trend in our study toward significantly more cases performed by era, which was to be expected given increasing specialized center comfort with transplant in these patients. Although functional status at the time of transplant was fairly good (>70% with a KPS score ≥60%), there were several patients with one or more clinical decompensations by the time of transplant. The most common of these, as expected in a Fontan cohort, was liver dysfunction, but a number of patients had developed other known modifiable risk factors for poor post-transplant survival by the time of transplant as well, including ventilator support, renal dysfunction, and parenteral nutrition. When the cohort was divided between patients with and without these modifiable risk factors, it became evident that although the two cohorts were demographically similar by age, race/ethnicity, patient weight, and era of transplant, they had significantly worse functional status at time of transplant. They also had higher listing status at the time of transplant. Patients in the decompensated cohort had significantly longer postoperative length of stay and were less likely to survive their initial hospital visit.

As discussed above, the overall cohort demonstrated survival at one and five years that was on par with estimates for all pediatric heart transplant recipients. Perhaps more remarkable, the five-year survival rates conditional on six-month and one-year survival were both greater than 90%, suggesting a large percentage of the mortality in this cohort arises from relatively early postoperative complications. To improve overall post-transplant Fontan survival, then, it is necessary to target the root causes of these early adverse outcomes, which can be achieved by optimizing patients, transplant timing, operative techniques, and early postoperative management. At the authors’ institution, we advocate early referral to heart failure specialists for patients with congenital heart disease to discuss advanced heart failure therapies. This includes patients with progressive NYHA Class III-IV symptoms, exercise intolerance, frequent hospitalizations, evidence of end organ dysfunction, need for inotropic support, worsening frailty, or ventricular tachyarrhythmias. Because of the often insidious progression of heart failure in this patient cohort, early referral allows for ideal partnering between the primary cardiologist, heart failure specialist, and patient for consistent testing and discussion regarding therapeutic options. The Advanced Cardiac Therapies Improving Outcomes Network (ACTION) has released a protocol with guidance for when to refer to a heart failure specialist.²⁹ For critically ill patients that have already progressed to Intermacs profile 1 or 2, the authors avoid the phenomenon of “limping to transplant” by deferring listing when possible until patients can be appropriately rehabilitated—off the ventilator, taking enteral nutrition, and discharged home if possible. Multiple previous studies have shown that VAD placement in patients with congenital heart disease and Fontans in particular can ameliorate these modifiable risk factors including improving functional status and decreasing rates of mechanical ventilation, inotropic support, renal dysfunction, and hepatic dysfunction.^30,31 On the other hand, a recent analysis of the authors’ institutional data found that patients on VAD support with three or more persistent modifiable risk factors at time of transplant had an in-hospital mortality of 50%, a sobering reminder that it may not be prudent to offer transplant to patients in this setting.³²

In the comparative Kaplan-Meier analysis of the decompensated and non-decompensated cohorts, there was significantly worse five-year and overall post-transplant survival in the decompensated cohort than the non-decompensated cohort. These findings demonstrate that pediatric Fontan patients who are medically optimized at the time of transplant and have not clinically decompensated at the time of transplant fare better post-transplant than those who have developed modifiable risk factors for poor outcomes. These patients in particular are vulnerable to early postoperative morbidity and mortality. When looking at the five-year and overall survival rates conditional on six-month survival, there was no longer a statistically significant difference in post-transplant survival relative to the non-decompensated patients, suggesting that these patients fare worse in the early postoperative period but after the early time point their survival curve generally matches those of other Fontan patients.

In the multivariable Cox regression models, the first model demonstrated pre-transplant TPN dependence, renal dysfunction, and higher listing status to be predictive of increased post-transplant mortality, whereas pre-transplant inotropic support & older age were found to predict decreased post-transplant mortality. Although the former findings were expected and likely reflect deconditioning of patients who require parenteral nutrition, having end-organ dysfunction, and who are more acutely ill having poorer postoperative outcomes, the latter finding was surprising. Inotropic support was common in the overall cohort and is frequently used in failing Fontan patients. Although difficult to demonstrate definitively, it is likely that inotropic support in this analysis served as a marker for the mechanism of Fontan failure; that is, patients who were prescribed inotropic medications likely had ventricular dysfunction which would readily respond to heart transplant whereas patients with right-sided Fontan failure and preserved ventricular function may not have indication for inotropic medication and may be more likely to have persistent complications after heart transplant as well. In the sensitivity analysis in Model B using clinical decompensation as a binary variable, decompensation was predictive of worse post-transplant survival even when controlling for a number of other clinical and demographic variables. In both models, older age at transplant predicted superior post-transplant outcomes, which may reflect the fact that Fontan patients who require transplant at younger ages may have an accelerated pathologic response to the altered physiology of a Fontan circulation (e.g., challenging pulmonary vasculature), which may not respond to transplant alone.

Limitations

The present study has several important limitations that should be considered when evaluating the findings. First, using retrospective data such as the data included in UNOS and PHIS is inherently subject to reporting, selection, and confounding biases. Although UNOS receives mandatory reporting on all transplants performed in the United States and is less liable to selection bias, the PHIS network is limited to only the 49 reporting hospitals included. Notably, a clear selection is that all the selected hospitals are pediatric centers, so the findings on survival can only be extrapolated to similar centers that may be more specialized and familiar with care of complex Fontan patients than other hospitals. However, the PHIS data was indispensable to being able to identify the patients in the UNOS database, and, as mentioned above, the linkage analysis between UNOS and PHIS has previously been validated with no significant differences between the linked cohort and the overall UNOS cohort. Another limitation to the study is that the sample size limited us from making more in-depth comparisons and characterizations of particular clinical decompensations. Furthermore, given the inability to conduct chart review on the participants in the study, more granular details of prior medical, catheter-based, and surgical interventions, exact dosing of medications such as inotropes at the time of transplant, echocardiographic data, type of Fontan circulation failure, and the like are unfortunately lacking from the analysis. Our study treated clinical decompensation as a binary variable as it was underpowered to evaluate any “dose-response” of having multiple decompensations; further research should investigate the impact of having multiple coexisting risk factors in the Fontan cohort. As experience with transplant in this population grows, additional research should focus on exploring these details further. Finally, given that the cohort in this analysis was limited to pediatric Fontan patients only, the conclusions drawn here should not necessarily be generalized to adult Fontan patients.

Conclusions

The findings of the above analysis demonstrate that excellent post-transplant survival is achievable in pediatric patients with Fontan failure, particularly when timing of transplant is optimized. These findings should serve to challenge the unfortunate dogma that listing for transplant in these patients represents a failure of palliation. If pediatric patients with failing Fontan circulations are referred early for transplant and medically optimized before they develop modifiable risk factors such as ECMO support, ventilator support, TPN dependence, or end organ dysfunction, they will have superior post-transplant outcomes. Conversely, for pediatric Fontan patients who have already progressed to developing modifiable risk factors whose physiology & anatomy would be appropriate for VAD implantation, mechanical circulatory support should be considered to allow for potential rehabilitation of these risk factors prior to transplant. Early referral for transplantation for pediatric Fontan patients and transplantation before the onset of clinical decompensation has the potential to achieve one- and five-year survival rates of greater than 90%.

Central Picture Legend:

Post-transplant survival of pediatric Fontan patients by clinical decompensation. (CI 95%)

Central Message:

Pediatric Fontan post-transplant outcomes are promising, although early mortality is high. For non-decompensated patients at transplant without end-organ disease, early mortality is circumvented.

Perspective Statement:

Heart transplant is increasingly being recognized as a feasible and effective therapy in pediatric failing Fontan patients. Survival data is limited in this cohort. The authors present a multi-institutional study of post-transplant survival after Fontan that finds high quality post-transplant outcomes are achievable, particularly when patients are transplanted before onset of clinical decompensation.

Glossary of Abbreviations

ACTION

Advanced Cardiac Therapies Improving Outcomes Network

CTC

clinical transaction classification

DORV

double outlet right ventricle

ECMO

extracorporeal membrane oxygenation

eGFR

estimated glomerular filtration rate

HLHS

hypoplastic left heart syndrome

ICD

International Statistical Classification of Diseases and Related Health Problems

ICU

intensive care unit

ISHLT

International Society for Heart and Lung Transplantation

KPS

Karnofsky Performance Scale

PHIS

Pediatric Health Information System

TPN

total parenteral nutrition

UNOS

United Network for Organ Sharing

VAD

ventricular assist device

VSD

ventricular septal defect