Association of pre-Fontan hemodynamics with long-term outcomes after Fontan palliation: A study from the Pediatric Cardiac Care Consortium

Abstract

Background:

Long-term outcomes after Fontan vary widely. While pre-Fontan hemodynamics predict early failure, their association with long-term outcomes remains unclear. We hypothesized that pre-Fontan hemodynamics predict long-term risk of death or transplant.

Methods:

We analyzed data from the Pediatric Cardiac Care Consortium, a US-based multicenter registry, including patients undergoing first-time Fontan with pre-Fontan catheterization and long-term follow-up. Patients undergoing a Fontan procedure before 18 years of age at anytime between 1982 and 2011 were included in the study. Outcomes of interest were in-hospital Fontan failure (death or takedown) and post-discharge death or transplant, identified through matching with the National Death Index and the Organ Procurement and Transplantation Network through 2022. Associations between pre-Fontan hemodynamics and long-term risk for death or transplant were assessed using Kaplan-Meier survival curves and extended Cox regression.

Results:

Among 1,175 patients [736 (62.6%) male, 626 (53.3%) with systemic left ventricle (LV)], 1,111 were discharged with Fontan physiology. Over a median post-discharge follow-up of 20.6 years (IQR: 18.2–24.4), 85 deaths and 49 transplants occurred. Pre-Fontan mean pulmonary arterial pressure (mPAP) was the strongest hemodynamic predictor of post-discharge outcomes, with a continuous association and no clear inflection point; 25-year transplant-free survival declined from 83.7% (95%CI 77.6–88.3%) in the low mPAP tertile to 73.7% (95%CI 65.5–80.3%) in the highest (log-rank P=0.02). Each 1SD increase in mPAP was associated with 1.33-fold higher odds of in-hospital failure (aOR 1.33; 95%CI: 1.00–1.77, P=0.05) and a 2.2-fold higher hazard of death or transplant (aHR 2.20; 95%CI: 1.62–3.00, P<0.01) estimated at discharge. This hazard declined 3% per year post-discharge (aHR/year 0.97; 95%CI: 0.95–0.99, P<0.01) and resolved by 17 years in patients with systemic RV and by 23 years in those with LV. Additional independent risk factors included systemic RV vs LV (aHR 2.39; 95%CI: 1.65–3.46, P<0.01) and delayed Fontan completion (>4 vs. 2–4 years of age) (aHR 1.80; 95%CI: 1.25–2.60, P=0.02).

Conclusions:

Elevated pre-Fontan mPAP is a strong predictor of in-hospital and long-term post-Fontan risk of death or transplant. Systemic RV and delayed Fontan (>4 years of age) further increased risk. These findings support early Fontan consideration and ongoing hemodynamic surveillance to optimize long-term outcomes.

Introduction:

The Fontan procedure, first introduced in 1971, revolutionized the management of patients with single ventricle (SV) physiology by separating the pulmonary and systemic circulations, providing a palliative pathway for this group of complex congenital heart diseases.¹ Over the past five decades, modifications in surgical techniques and advancements in postoperative care have significantly improved survival.² Large-scale cohort studies from tertiary care centers,^3–7 US multicenter⁸ and international⁹ registries including meta-analysis¹⁰ have demonstrated variable long-term survival rates following Fontan palliation, influenced by surgical era, techniques, systemic ventricular morphology and inconsistent inclusion of transplants or Fontan takedown cases. The 2019 scientific statement from the American Heart Association summarized the global experience with the Fontan procedure, estimating a 30-year survival of approximately 85%.¹¹

Despite increasing use of non-invasive methods in pre-Fontan evaluation, cardiac catheterization remains the standard of care for assessing surgical candidacy, based on parameters such as mean pulmonary artery pressure (mPAP), pulmonary vascular resistance (PVR) and systemic ventricular end-diastolic pressure (SVEDP).^11,12 While prior studies have explored the relationship between pre-Fontan hemodynamics and early post-operative outcomes, their predictive value for long-term outcomes remains uncertain.¹³

As an increasing number of Fontan patients transition into adulthood, it is important to identify predictors of premature failure or death. A data-driven approach to risk stratification and long-term management could help optimize outcomes. Our study aims to evaluate the association between pre-Fontan hemodynamics and long-term outcomes (i.e., death and transplantation), providing insight into surveillance strategies and therapeutic interventions to improve post-Fontan outcomes. We hypothesized that elevated pre-Fontan mPAP independently predicts long-term risk of death or transplant, and that this prognostic effect differs by systemic ventricular morphology.

Methods:

Deidentified data and methodological details, including SAS code, are available from the corresponding author to qualified researchers upon reasonable request; however, National Death Index (NDI) and Organ Procurement Transplant Network (OPTN) matched data are subject to federal disclosure restrictions and cannot be shared.

This is a retrospective cohort study of patients with functionally single ventricle, who were enrolled in the Pediatric Cardiac Care Consortium (PCCC), a US-based registry of congenital heart surgeries, between 1982 and April 15, 2003 and underwent a Fontan procedure before 18 years of age at anytime between 1982 and 2011 (end of new patient enrollment in the registry). Prospective follow-up was conducted through matching with the NDI and the OPTN to capture death and heart transplant events through December 31, 2022.¹⁴ The study was approved by the Emory University Institutional Review Board with waiving the need for consent for patients enrolled in PCCC up to April 15, 2003 (date of stricter HIPAA rule implementation). This study used data from OPTN. The OPTN data system includes data on all donors, wait-listed candidates, and transplant recipients in the US, submitted by the members of the OPTN. The Health Resources and Services Administration, U.S. Department of Health and Human Services provides oversight to the activities of the OPTN contractor. This study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Study population and data collection:

The study included all SV patients with Glenn or Kawashima palliation, who were enrolled in the PCCC, after cardiac catheterization in preparation for the Fontan procedure between 1 and 21 years of age. Details on the rationale, design and baseline characteristics of this Fontan cohort, have been reported before.^14,15 Only patients undergoing lateral tunnel (LT) and extracardiac conduit (EC) were included, as atrio-pulmonary Fontan procedures are no longer relevant to contemporary outcomes. Additional inclusion criteria were availability of direct identifiers and enrollment before April 15, 2003. Clinical, surgical, and hemodynamic data were abstracted from the registry using standardized definitions.

Outcomes:

The primary outcome was all-cause mortality or transplant after hospital discharge following Fontan completion. Secondary outcomes included in-hospital Fontan failure, defined as in-hospital death, heart transplant or Fontan takedown (i.e. surgical deconstruction of the Fontan connections to an intermediate palliative circulation due to Fontan non-tolerance). Time to event was defined as the time after the day of discharge until transplant or death, whichever occurred first, and censored on December 31, 2022 (latest available data). Death and transplant events were ascertained through the PCCC or by matching with the NDI and OPTN records. Causes of death were obtained from the NDI-Plus and classified using International Classification of Diseases (ICD-9 and ICD-10) codes into three categories: congenital heart disease (CHD)-related, cardiovascular disease (CVD)-related or non-CHD/non-CVD-related as previously described.¹⁶

Statistical Analysis:

Pre-Fontan hemodynamic variables were derived from cardiac catheterization performed within 12 months before Fontan completion. Pulmonary parameters included mPAP, PVR and transpulmonary gradient (TPG). Systemic parameters included mixed venous oxygen saturation (SVO₂), cardiac index (CI) and systemic ventricle end-diastolic pressure (SVEDP). Variables were analyzed both categorically (by tertiles), and continuously [per 1 standard deviation (SD) increase].

Continuous variables were summarized as means ± SD or medians with interquartile ranges (IQR), as appropriate and categorical variables as counts and percentages. Comparisons were performed using chi-square for categorical variables (including p-trend tests for variables with a natural ordering, assessed by Mantel-Haenszel chi-square), and the t-test or Wilcoxon rank sum tests for continuous variables.

Univariable models were first evaluated for each covariate’s association with early (in-hospital) and long-term (post-discharge) outcomes. Variable selection for multivariable analysis was guided by directed acyclic graph (DAG) analysis, clinical relevance, and collinearity assessment as described before.¹⁴ All multivariable models were adjusted for age at Fontan (<2, 2–4, and >4 years), sex, surgical era (1982–1998, 1999–2002, 2003–2011), systemic ventricular morphology [right (RV), left (LV)], systemic oxygen saturation (SaO₂) and serum hemoglobin level (Hgb) (both by tertiles). Patients with non-classifiable ventricular type were excluded due to small sample size. Variables representing the same hemodynamic domain, pulmonary (mPAP, PVR, TPG) or systemic (SVEDP, CI, SvO2), or those with strong correlation (r>0.5) between them (such as mPAP and SVEDP) were not included in the same model.¹⁴ Because Fontan type correlated strongly with era,¹⁴ it was excluded from the primary model but was evaluated separately in sensitivity analyses. Interaction terms were included in the models to explore whether the associations of mPAP with outcomes differed between patients with systemic RV or LV. Missing values in this dataset ranged from 12% for SaO2 to 26% for PVR and were amounting to 19.6% for the mPAP and 16.5% for the SVEDP.¹⁴ To address missing data, we applied a combination of complete-case analysis, multiple imputation by chained equations (MICE), and inverse probability weighting methods (IPW) to ensure robustness of results. In-hospital outcomes were modeled as a binary outcome (death or Fontan take down) using generalized linear mixed models (mixed-effects logistic regression) with treatment center included as a random effect. Survival analysis was conducted using Kaplan-Meier curves and extended Cox proportional hazards model to evaluate risk factors for adverse outcomes. The log-rank P-values were adjusted for multiple comparisons using the Šidák correction. To account for variability across treatment centers, Cox models incorporating a shared frailty term were used, treating the center as a random effect. Proportional hazards assumptions were assessed using Schoenfeld residuals, and extended Cox models with time-varying effects were applied by including an interaction term between the variable and time when necessary.

Parsimonious models assessed the impact of heterotaxy and pacemaker presence on post-discharge outcomes using backward stepwise elimination. Variables were initially included in the model if they met a significance level of P< 0.20 in the univariable analysis and sequentially eliminated until all variables having P<0.05. Three different models were created: (1) base multivariable model including mPAP, systemic ventricular type, and Fontan age, (2) base model plus presence of heterotaxy, (3) base model plus presence of pacemaker.

Fine-Gray competing risk analyses were used to estimate the subdistribution hazard of CHD/CVD-related death or transplant, treating deaths from other causes as competing risks. In a complementary analysis, all-cause mortality was evaluated with transplant considered a competing event.

Mediation analysis was performed to examine whether elevated mPAP mediated the relationship between elevated SVEDP and long-term outcomes (death or transplant) after Fontan completion. Direct effect and indirect effects of SVEDP were estimated using the PROCESS macro for SAS under the Preacher and Hayes causal mediation framework¹⁷ adjusting for age at Fontan, sex, surgical era, systemic ventricular morphology, SaO₂, and Hgb concentration. Mediation was inferred when the 95%CI of the indirect effect excluded zero and no significant exposure–mediator interaction was present.

Statistical analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, North Carolina). A two-sided P-value <0.05 was considered statistically significant.

RESULTS

Patient Characteristics:

A total of 1461 patients met inclusion criteria. Among pulmonary hemodynamics, mPAP was selected as the primary predictor based on its strongest univariable and multivariable associations with outcomes. Based on mPAP availability, 1175 patients were included in the final analysis with 816 of them missing no hemodynamic values. Patient characteristics by mPAP tertiles (Low 3–9 mmHg, Middle 10–12 mmHg, High 13–24 mmHg) are presented in Table 1. In our cohort, 736 (62.6%) of patients were males, with a median age of 3.1 years (IQR 2.4 – 4.3 years) at Fontan surgery. Systemic LV morphology comprised 626 (53.3%) of the overall cohort. There is no trend association observed between surgical era and mPAP tertiles. Compared with systemic LV, patients with a systemic right ventricle (RV) were more often males [(349 (66.5%) vs. 372 (59.4%), P=0.01), operated in the most recent era [219 (41.7%) vs. 175 (27.9%), P<0.01), and more likely to have heterotaxy [71 (13.5%) vs. 33 (5.3%), P<0.01) or associated anomalies (asplenia/polysplenia, total anomalous pulmonary venous return and dextrocardia) (Table S1A).

Table 1.

Patient Characteristics by Mean Pulmonary Arterial Pressure (mPAP) Tertile in the Study Cohort

Patient characteristics	Total (n = 1175)	Low Tertile (3–9 mmHg) (n = 363)	Middle Tertile (10–12 mmHg) (n = 486)	High Tertile (13–24 mmHg) (n = 326)	P-value
Sex					0.28
Male	736 (62.6)	216 (59.5)	315 (64.8)	205 (62.9)
Female	439 (37.4)	147 (40.5)	171 (35.2)	121 (37.1)
Surgical Era (tertiles)					<.01*
Early (1987–1998)	385 (32.8)	145 (39.9)	142 (29.2)	98 (30.1)
Middle (1999–2002)	387 (32.9)	113 (31.1)	174 (35.8)	100 (30.7)
Late (2003–2011)	403 (34.3)	105 (28.9)	170 (35.0)	128 (39.3)
Age at Fontan in Years (IQR)	3.1 (2.4 – 4.2)	3.1 (2.4, 4.1)	3.0 (2.4, 4.2)	3.2 (2.4, 4.6)	0.48
Type of Systemic Ventricle					0.51
Systemic LV	626 (53.3)	203 (55.9)	262 (53.9)	161 (49.4)
Systemic RV	525 (44.7)	152 (41.9)	215 (44.2)	158 (48.5)
Non-classifiable/Unknown	24 (2.0)	8 (2.2)	9 (1.9)	7 (2.1)
Specific Diagnosis
Hypoplastic Right Heart	366 (31.1)	129 (35.5)	146 (30.0)	91 (27.9)	0.77
HLHS	255 (21.7)	71 (19.6)	109 (22.4)	75 (23.0)	0.48
Single Ventricle (other)	215 (18.3)	67 (18.5)	92 (18.9)	56 (17.2)	0.81
DORV	105 (8.9)	31 (8.5)	46 (9.5)	28 (8.6)	0.87
UCAVC	149 (12.7)	40 (11.0)	54 (11.1)	55 (16.9)	0.03
Complex TGA	85 (7.2)	25 (6.9)	39 (8.0)	21 (6.4)	0.66
Heterotaxy
Asplenia	75 (6.4)	16 (4.4)	34 (7.0)	25 (7.7)	0.15
Polysplenia	41 (3.5)	9 (2.5)	17 (3.5)	15 (4.6)	0.30
TAPVR	43 (3.7)	7 (1.9)	17 (3.5)	19 (5.8)	0.02
Dextrocardia	68 (5.8)	15 (4.1)	29 (6.0)	24 (7.4)	0.19
Pre-Fontan Procedures
A. Stage I
Norwood	309 (26.3)	90 (24.8)	133 (27.4)	86 (26.4)	0.70
DKS	118 (10.0)	45 (12.4)	44 (9.1)	29 (8.9)	0.20
AP Shunt Only	595 (50.6)	179 (49.3)	246 (50.6)	170 (52.2)	0.76
Pulmonary Arterial Band	196 (16.7)	57 (15.7)	80 (16.5)	59 (18.2)	0.68
B. Stage II
Glenn or Hemi-Fontan	1128 (96.0)	354 (97.5)	468 (96.3)	306 (93.9)	<.05
Kawashima	47 (4.0)	9 (2.5)	18 (3.7)	20 (6.1)	<.05
C. Pacemaker	39 (3.3)	6 (1.7)	21 (4.3)	12 (3.7)	0.09
Type of Fontan					0.02
Lateral Tunnel	419 (35.7)	149 (41.1)	168 (34.6)	102 (31.3)
Extracardiac Conduit	756 (64.3)	214 (58.9)	318 (65.4)	224 (68.7)
Fenestration	702 (59.7)	226 (62.3)	281 (57.8)	195 (60.0)	0.42
Post-Fontan Median Length of Stay in Days (IQR)	11 (8 – 19)	10 (8 – 16)	10 (8 – 17)	12 (9 – 22)	<.01
In-hospital Takedown	16 (1.4)	6 (1.7)	3 (0.6)	7 (2.2)	0.14
In-hospital Failure (death/takedown)	64 (5.4)	14 (3.9)	23 (4.7)	27 (8.3)	0.03†

LV: left ventricle, RV: right ventricle, HLHS: hypoplastic left heart syndrome, DORV: double outlet right ventricle, UCAVC: unbalanced common atrioventricular canal, TGA: transposition of great arteries, TAPVR: total anomalous pulmonary venous return, DKS: Damus-Kaye-Stansel, AP: aortopulmonary

^*P-trend=0.85;

^†P-trend=0.05

Across age groups at Fontan completion, mPAP, TPG, PVRi and CI were similar (Table S2). SVEDP, SaO₂, SvO₂ and Hgb increased modestly with age at Fontan completion (P<0.01 for all).

Pre-Fontan Hemodynamics and In-Hospital Outcomes:

Among 1175 patients with available mPAP 64 patients (deaths: 54; Fontan takedown: 16 with 6 of them among the reported deaths) experienced in-hospital Fontan failure, while 1111 patients were discharged alive with Fontan physiology. Patients in the higher mPAP tertile had longer post-Fontan hospitalization [Median (IQR): 12 (9 – 22) vs 10 (8 – 16) days, P<0.01], and higher incidence in Fontan failure (8.3% vs 4.7% in the middle and 3.9% in the low tertile, P=0.03 and P-value trend test=0.05) (Table 1). Patients with systemic RV had marginally longer post-Fontan hospitalization (11 vs. 10 days, P=0.05) (Table S1A). Pre-Fontan hemodynamics were similar between systemic RV and LV groups with only slightly higher mPAP in the systemic RV group (11.1 ± 3.1 mmHg vs. 10.8 ± 2.9 mmHg in the systemic LV group; P=0.03) (Table S1B).

After adjustment, each 1SD increase in mPAP was associated with 1.33-fold higher odds (95%CI: 1.00–1.77, P=0.05) of in-hospital Fontan failure (Table S3). No other variables reached statistical significance.

Pre-Fontan Hemodynamics and Long-term Outcomes:

Over a median follow-up of 20.6 years (IQR 18.2 – 24.4), 85 deaths and 49 transplants occurred. The 25-year survival for the overall cohort was 80.3% (95% CI, 76.7–83.4%). Survival differed across mPAP tertiles, ranging from 83.7% (95%CI 77.6–88.3%) in the low mPAP tertile to 73.7% (95%CI 65.5–80.3%) in the high tertile, with the middle tertile being 81.4% (95%CI:75.6–85.9%) (log-rank P value <.01) (Figure 1).

Figure 1.

Long-term transplant-free survival estimates post Fontan discharge by tertiles of mean Pulmonary Arterial Pressure (mPAP).

P-values are adjusted for multiple comparisons by the Šidák method.

Survival patterns differed by systemic ventricular morphology: in systemic LV patients, survival in the middle tertile resembled that of the low tertile, whereas in systemic RV patients, it paralleled that of the high tertile (Figure 2A, B).

Figure 2.

Long-term transplant-free survival estimates post-Fontan discharge by tertiles of mean Pulmonary Arterial Pressure (mPAP) of patients with systemic A) left ventricle (LV) and B) right ventricle (RV).

P-values are adjusted for multiple comparisons by the Šidák method.

In univariable analysis, each 1SD increase in mPAP or SVEDP (≈3 mmHg for both variables), systemic RV vs LV and Fontan completion at >4 years of age vs. 2–4 years of age were associated with higher risk of death or transplant (Table S4).

After adjustment, each 1SD increase in mPAP was associated with a 2.2-fold higher hazard for death or transplant at discharge (95%CI 1.62–3.00) (Figure 3). The association between mPAP and hazard of death or transplant was continuous, with no clear inflection point, but exhibited a time-dependent decline of 3% per year post-Fontan discharge (aHR 0.97 per year; 95%CI 0.95–0.99), across both systemic LV and RV groups. Consequently, the aHR gradually decreased, reaching 1.0 at 17 years post-discharge for patients with systemic RV and 23 years for those with systemic LV (Figure 4). Systemic RV morphology was associated with a 2.39-fold increased hazard compared to systemic LV (95%CI 1.65–3.46, P <0.01) for the same mPAP z-score. The interaction between systemic RV and mPAP was not statistically significant.

Figure 3.

Forest plot with logarithmic scales of adjusted hazard ratios (aHR) from multivariable analysis of post-Fontan death or transplant after discharge, highlighting the association with mean Pulmonary Arterial Pressure (mPAP).

RV: systemic right ventricle, LV: systemic left ventricle, SD: standard deviation, SaO2: arterial oxygen saturation, Hgb: hemoglobin. mPAP(1SD)*RV represents the interaction between a 1SD increase in mPAP and systemic RV (vs. LV). mPAP(1SD)*time represents the interaction between a 1SD increase in mPAP and each additional year after Fontan discharge.

^§Indicates statistical significance at P<.05 (all other P-values>0.1).

Figure 4.

Change in adjusted Hazard Ratio (aHR) over time with logarithmic scales per one standard deviation (1SD) increase in the mean Pulmonary Arterial Pressure (mPAP) by type of systemic ventricle.

For patients with systemic LV, the lowest bound of the 95%CI crosses 1.0 at 14 years and reaches 1.0 at 23 years post-Fontan (blue *). For those with systemic RV the corresponding time points are 10 and 17 years post-Fontan respectively (red *).

Fontan completion at >4 years of age vs. at 2–4 years of age remained a significant predictor of elevated hazard of death or transplant after Fontan discharge (aHR = 1.80; 95%CI 1.25–2.60, P<0.01). There was a trend toward lower hazard with earlier Fontan completion (1-<2 years of age), but without reaching threshold for statistical significance (aHR 0.52; 95%CI: 0.23–1.14, P=0.10). Surgical era was not independently associated with outcomes (Figure 3).

In parsimonious models adding comorbidities, heterotaxy (aHR 1.61;95% CI 1.00–2.58, P=0.05) and pacemaker (aHR 2.96; 95%CI 1.55–5.69, P<0.01) were independently associated with the hazard of death or transplant, without altering the mPAP-outcome relationship (Table S5).

Competing risk analysis accounting for non-CHD/non-CVD-related events did not alter the association between mPAP increase and the hazard of CHD/CVD-related death or transplant (Table S6A). Similarly, treating transplant as a competing risk had no significant impact on the association between mPAP increase and all-cause mortality (Table S6B).

A parallel model for SVEDP demonstrated a comparable association for each 1SD increase (aHR 2.15; 95%CI 1.47–3.16) and with a 3% annual decline in hazard (aHR 0.97 per year; 95% CI 0.95–0.99) (Figure 5). Adjusted mediation analysis revealed that mPAP partially mediated the relationship between SVEDP and long-term outcomes (Figure S1), accounting for 62% of the total association.

Figure 5.

Forest plot with logarithmic scales of adjusted hazard ratios (aHR) from multivariable analysis of post-Fontan death or transplant after discharge, highlighting the association with Single Ventricle End-diastolic Pressure (SVEDP).

RV: systemic right ventricle, LV: systemic left ventricle, SD: standard deviation, Hgb: hemoglobin, SVEDP(1SD)*RV represents the interaction between a 1SD increase in SVEDP and systemic RV (vs. LV). SVEDP(1SD)*time represents the interaction between a 1SD increase in SVEDP and each additional year after Fontan discharge.

^§Indicates statistical significance at P<.05 (low vs middle tertile Hgb comparison reached a P-value of 0.08; all other P-values>0.1).

Replacing era with type of Fontan (EC vs. LT and with or without fenestration), yielded no significant differences in long-term outcomes between Fontan types (Table S7). Sensitivity analyses using MICE and IPW confirmed the robustness of the main findings (Tables S8A and S8B). The adverse association of Fontan completion after 4 years of age with late outcomes was reproduced only in the multiple imputation model. Conversely, both approaches consistently demonstrated a survival advantage for Fontan completion between 1 and <2 years of age compared with the 2–4 year reference group (aHR 0.43; 95%CI: 0.23–0.80 with MICE, and 0.23; 95%CI: 0.10–0.52 with IPW; P<0.01 for both). Additionally, the IPW analysis identified the earliest era (1982–1998) as having the highest hazard of death or transplant (aHR 3.04; 95%CI: 2.07–4.46).

DISCUSSION

In this large multicenter cohort with one of the longest reported follow-ups, we identified mPAP at the time of Fontan completion as a strong, independent predictor of both early and late transplant-free survival post-Fontan. The association between mPAP and outcomes was continuous across its range, with no clear inflection point, and remained significant after adjusting for ventricular morphology, age at Fontan, and comorbidities. The impact of mPAP diminished over time, becoming neutral by 17 years post-Fontan discharge in patients with systemic RV and by 23 years in those with systemic LV. Although the interaction between systemic RV and mPAP was not statistically significant, the earlier attenuation of mPAP’s prognostic impact in patients with systemic RV likely reflects competing risks from intrinsic vulnerability and distinct reconstructive strategies, rather than better adaptation to pulmonary pressure load.^18–20 Progressive ventricular dilation, atrioventricular valve insufficiency, and greater energy loss through the reconstructed neoaorta, each largely independent of pulmonary pressure load, may overshadow the effect of mPAP on long-term Fontan performance and explain why its prognostic impact became neutral six years earlier in the RV subgroup.

SVEDP demonstrated a similarly strong association with long-term outcomes; however, mediation analysis revealed that nearly two-thirds of its effect was transmitted through mPAP, reinforcing the interdependence of ventricular filling pressures and pulmonary vascular load. This suggests that elevated SVEDP contributes to long-term risk primarily through its impact on mPAP and may represent a modifiable target for intervention.

Previous studies have variably linked adverse pre-Fontan hemodynamics (elevated mPAP, SVEDP, PVR and TPG) with early Fontan outcomes, particularly in relation to surgical candidacy and perioperative risks.^6,21–25 Elevated postoperative pulmonary arterial pressures have also been associated with adverse outcomes, though such data primarily reflect the body’s adaptation to the Fontan physiology.^15,26,27 However, only few studies have systematically evaluated the long-term prognostic significance of pre-Fontan hemodynamics beyond the first decade post-surgery. Prior work, identified thresholds for poor outcomes (mPAP> 15–17 mmHg; SVEDP>12 mmHg), but did not characterize how these risks evolve over time or differ by ventricular morphology.^6,21,28

In this study, we evaluated the association of the hemodynamics with the outcomes separately for the in-hospital and post discharge survival, recognizing that prognostic significance may change as the circulation adapts to the Fontan physiology.²¹ Our study extends prior observations on that subject by defining the time-dependent nature of pre-Fontan mPAP and SVEDP-related risk, showing that associations with mortality or transplant attenuate over time, earlier among patients with systemic RV than LV. Although unadjusted analyses suggested morphology-specific survival patterns at mid-range mPAP values, similar to the low tertile in systemic LV and the high tertile in systemic RV, we could not detect a statistically significant interaction between systemic ventricular type and mPAP with long-term outcomes. Nonetheless, these findings imply that systemic ventricular morphology and its clinical correlates may modulate tolerance to elevated mPAP. Independent of ventricular morphology, though, our findings demonstrate that even modest elevations in pre-Fontan mPAP, within ranges often deemed acceptable, are associated with increased long-term risk. Despite recent suggestions that pre-Fontan catheterization may be omitted without compromising outcomes,^12,22 our findings reaffirm the value of pre-Fontan hemodynamics in risk stratification and long-term surveillance planning.

The survival advantage observed in patients with a systemic LV reinforces prior findings from the PCCC and other registries.^{8,10,19,29,30} Although elevated mPAP conferred increased hazard in both morphologies, the earlier attenuation of its effect in patients with a systemic RV may reflect the influence of additional adverse factors in this high-risk group, supporting enhanced post-Fontan surveillance and a lower threshold for intervention.

The absence of a consistent era effect across models suggests that improvements in perioperative and ambulatory care may have not eliminated the intrinsic limitations of the Fontan circulation. Interestingly, outcomes appeared more favorable in the middle surgical era, but this trend did not reach statistical significance and was unrelated to the transition from LT to EC Fontan in the PCCC registry.¹⁴

We also found that earlier Fontan completion was associated with improved long-term survival. Patients undergoing Fontan between 2–4 years of age had better outcomes than those at older ages, while those between 1–<2 years demonstrated an additional survival advantage in sensitivity analyses. These findings support proactive surgical planning and suggest that the optimal window for Fontan completion may extend earlier than the commonly accepted 2–4 years. The optimal age for Fontan completion has long been debated, balancing the concerns of prolonged cyanosis against the presumed finite lifespan of the Fontan circulation (the “ticking clock theory”).³¹ Prior studies using administrative data have suggested that completion at 3–5 years may be ideal, though these analyses are often limited by confounding and selection bias.^32,33 Our study, using rigorously defined criteria and long-term follow-up, ⁷ confirms that delayed Fontan >4 years is associated with increased long-term mortality or transplant risk, consistent with a recent meta-analysis.¹⁶ Notably, our findings also suggest that Fontan completion before even 2 years of age may offer incremental benefit potentially by mitigating the detrimental effects of chronic hypoxia, pulmonary artery stenosis (often addressed at the time of Fontan), or aortopulmonary collateral formation.³⁴ A previous study observed a similar trend toward improved outcomes with younger Fontan age, although differences by unadjusted comparisons did not emerge until 15 years postoperatively.³⁵ With its exceptionally long follow-up, our study is uniquely positioned to quantify this delayed benefit, reinforcing the importance of early surgical timing in optimizing long-term survival.

These findings have several clinical implications. First, maintaining low mPAP - directly or by improving ventricular compliance - is critical for long-term survival. Second, the time-dependent decline in the prognostic value of pre-Fontan hemodynamics suggests the need for periodic hemodynamic reassessment starting around 10–15 years post-Fontan (earlier for patients with systemic RV) to inform management decisions such as use of pulmonary vasodilators or transplant listing. This reassessment should consider a broader evaluation of Fontan performance, within the context of the “Fontan paradox”, the coexistence of systemic venous hypertension and pulmonary arterial hypotension.³⁶ Over time, this paradox is exacerbated by rising SVEDP and mPAP. In that regard, lowering the mPAP in isolation may be insufficient without preserving adequate pressure gradients throughout the subsegments of the Fontan circulation. With several short and mid-term studies of pulmonary vasodilators having shown only modest benefits,^29,37–40 the long-term impact of lowering mPAP (or SVEDP) on preserving the Fontan circulation remains uncertain and warrants further investigation.⁴¹ Given the challenges of conducting such trials, longitudinal studies of the Fontan “natural history”, starting from pre-Fontan stage that is not affected by the post-Fontan adaptation may offer the most feasible path forward.

Limitations and Future Directions:

While this study offers important insights, supported by a large cohort and extended long follow-up, several limitations to causal inference merit consideration. The retrospective, observational design, is subject to residual confounding as analyses were restricted to data available within the registry. Approximately 20% of hemodynamic values were missing, and other clinically relevant variables such as atrioventricular valve regurgitation,¹⁹ ventricular function, burden of aortopulmonary collaterals and many co-morbidities were not captured. To mitigate these limitations, both IPW and MICE analyses were performed, and neither meaningfully altered the association between mPAP and outcomes, supporting the robustness of our findings.

Linkage with the NDI and OPTN minimized loss to follow-up, although some misclassification remains possible. Prior validation studies demonstrated sensitivity of 84.0% and 89.7% for the NDI and OPTN,⁴² respectively and a sensitivity of 94.7% for death ascertainment specifically in the Fontan cohort as previously reported.¹⁴ Any residual misclassification is likely nondifferential across hemodynamic strata and would bias results toward the null.

To strengthen causal interpretability, we used both design-based and analytical safeguards. (DAGs guided covariate selection based on known causal relationships and clinical knowledge.¹⁴ Multivariable adjustment was applied to further reduce confounding. In addition, we conducted a mediation analysis leveraging semi-longitudinal data to evaluate potential causal pathways linking systemic and pulmonary hemodynamics with long-term outcomes.

Other limitations include incomplete capture of post-Fontan reinterventions (e.g., valve repair or Fontan revision), lack of outpatient longitudinal data (e.g., medication use or functional status), and evolving perioperative and medical management strategies that could influence long-term outcomes. As race and ethnicity were not systematically captured in the PCCC, this analysis is unable to examine potential disparities among long-term survivors of the Fontan procedure. While mediation analysis suggested that mPAP partially mediates the effect of SVEDP on outcomes, the semi-longitudinal design limits definitive causal inference due to lack of temporality.

Although, the mPAP tertile cutoffs were registry-based, they align closely with those reported in other registries.³³ While the absolute range of mPAP may vary across centers due to factors such as catheterization conditions (e.g. use of positive pressure ventilation or sedation) or center-specific practices (e.g. management of antegrade pulmonary flow, collaterals or residual stenoses) such variability is likely nondirectional and would tend to bias results toward the null rather than introduce systematic error. Moreover, the continuous regression modeling per 1SD increase in mPAP enhances the generalizability of our findings across the observed study range (3–24 mmHg).

Of note and particularly relevant to the interpretation of surgical age associations with outcomes, the study was not specifically designed to evaluate associations with combination of Fontan types (LT vs EC, with or without fenestration). Given the complex interplay between pre-Fontan hemodynamics, surgical era trends, and center-specific practices, such analyses were beyond the scope of this report. However, the present findings will inform the design of a separate study specifically focused on the association between Fontan type and long-term outcomes.

Future research should reexamine these findings in contemporary cohorts receiving newer medical therapies (such as pulmonary vasodilators)^29,37–40 or structured exercise programs, an emerging strategy in Fontan management,^43,44 for their potential to modify long-term risk by lowering both mPAP and SVEDP. Studies are also needed to explore the role of serial mPAP assessment, including the use of implanted hemodynamic monitoring,⁴⁵ in tracking risk evolution post-Fontan and to define actionable thresholds and the utility of preemptive interventions for patients with borderline hemodynamics.

Our study lays the foundation for prospective validation studies to assess whether pre-Fontan mPAP-targeted therapies improve long-term survival. Developing risk prediction models that incorporate mPAP, ventricular morphology, and Fontan age may enhance patient selection and surgical decision-making, ultimately improving Fontan outcomes. A similar approach was taken in a Japanese study²³ that evaluated hemodynamic factors one-year post-Fontan and developed a predictive scoring system. This model identified low ejection fraction (<30%), systemic oxygen saturation (<80%), central venous pressure (>16 mmHg), mean systemic arterial pressure (<60 mmHg), and Fontan age >15 years as the highest risk factors for mortality. Applying a similar preoperative risk stratification could help tailor interventions for high-risk patients before Fontan completion.

Additionally, further exploration of novel pharmacologic strategies aimed at reducing pre-Fontan pulmonary pressures may improve post-Fontan outcomes. The FUEL trial,²⁴ which demonstrated improved anaerobic ventilatory thresholds with the use of udenafil in Fontan patients, supports the potential role of targeted vasodilator therapy in this population. Future randomized controlled trials are needed to assess whether pulmonary vasodilator therapy could improve long-term Fontan survival and reduce transplant rates.

Conclusions

Our findings provide compelling evidence that pre-Fontan mPAP is a strong, independent predictor of early Fontan failure and long-term transplant-free survival. Even within ranges traditionally considered acceptable, elevated mPAP conferred incremental risk, with additional vulnerability conferred by systemic RV morphology and delayed Fontan completion (>4 years of age). Importantly, the prognostic influence of mPAP waned over time, with a faster decline observed in systemic RV patients compared to those with systemic LV. This time-dependent attenuation supports ventricular morphology-specific approaches to long-term surveillance, with earlier hemodynamic reassessment potentially offering greater utility in RV-dominant anatomies. These results support the incorporation of mPAP into risk stratification algorithms, optimization of pulmonary vascular loading conditions before Fontan, and tailoring of post-discharge surveillance strategies by ventricular morphology. Future studies should explore whether these risks can be modified through targeted therapies, or longitudinal monitoring, particularly for patients with borderline hemodynamics. Despite advances in surgical technique and perioperative care, the absence of a consistent era effect highlights the persistent long-term challenges inherent to the Fontan circulation.

Clinical Perspective.

What is New?

• Pre-Fontan mean pulmonary arterial pressure (mPAP) independently predicts both early Fontan failure and long-term risk of death or transplant, with a continuous association and no inflection point.

• The prognostic impact of elevated mPAP diminishes over time, resolving earlier in patients with systemic right ventricle (RV) than in those with systemic left ventricle (LV).

• Systemic RV and delayed Fontan completion >4 years of age are associated with increased long-term risk of death or transplant.

What are the Clinical Implications?

• Pre-Fontan mPAP is valuable for risk stratification at the time of Fontan and for long-term follow-up surveillance.

• The time-dependent attenuation of pre-Fontan mPAP’s prognostic value supports periodic hemodynamic reassessment beginning 10–15 years post-Fontan (earlier in patients with systemic RV) to inform individualized management strategies in Fontan patients.

• Fontan completion before 4 years of age (and possibly as early as 1–2 years) may confer improved long-term survival and warrants consideration in surgical planning.

Non-standard Abbreviations and Acronyms

CHD

congenital heart disease

CI

cardiac index

CVD

cardiovascular disease

DAG

directed acyclic graph

EC

extracardiac conduit

IPW

inverse probability weighting

LT

lateral tunnel

LV

left ventricle

MICE

multiple imputation by chained equations

mPAP

mean pulmonary arterial pressure

NDI

National Death Index

OPTN

Organ Procurement Transplant Network

PCCC

Pediatric Cardiac Care Consortium

PVR

pulmonary vascular resistance

RV

right ventricle

SV

single ventricle

SVEDP

systemic ventricular end-diastolic pressure

SVO₂

systemic mixed venous oxygen saturation

TPG

transpulmonary gradient