Fetal hemodynamic response to aortic valvuloplasty and postnatal outcome: a European multicenter study
ABSTRACT
Objective
Fetal aortic stenosis may progress to hypoplastic left heart syndrome. Fetal valvuloplasty (FV) has been proposed to improve left heart hemodynamics and maintain biventricular (BV) circulation. The aim of this study was to assess FV efficacy by comparing survival and postnatal circulation between fetuses that underwent FV and those that did not.
Methods
This was a retrospective multicenter study of fetuses with aortic stenosis that underwent FV between 2005 and 2012, compared with contemporaneously enrolled natural history (NH) cases sharing similar characteristics at presentation but not undergoing FV. Main outcome measures were overall survival, BV-circulation survival and survival after birth. Secondary outcomes were hemodynamic change and left heart growth. A propensity score model was created including 54/67 FV and 60/147 NH fetuses. Analyses were performed using logistic, Cox or linear regression models with inverse probability of treatment weighting (IPTW) restricted to fetuses with a propensity score of 0.14–0.9, to create a final cohort for analysis of 42 FV and 29 NH cases.
Results
FV was technically successful in 59/67 fetuses at a median age of 26 (21–34) weeks. There were 7/72 (10%) procedure-related losses, and 22/53 (42%) FV babies were delivered at < 37 weeks. IPTW demonstrated improved survival of liveborn infants following FV (hazard ratio, 0.38; 95% CI, 0.23–0.64; P = 0.0001), after adjusting for circulation and postnatal surgical center. Similar proportions had BV circulation (36% for the FV cohort and 38% for the NH cohort) and survival was similar between final circulations. Successful FV cases showed improved hemodynamic response and less deterioration of left heart growth compared with NH cases (P ≤ 0.01).
Conclusions
We report improvements in fetal hemodynamics and preservation of left heart growth following successful FV compared with NH. While the proportion of those achieving a BV circulation outcome was similar in both cohorts, FV survivors showed improved survival independent of final circulation to 10 years' follow-up. However, FV is associated with a 10% procedure-related loss and increased prematurity compared with the NH cohort, and therefore the risk-to-benefit ratio remains uncertain. We recommend a carefully designed trial incorporating appropriate and integrated fetal and postnatal management strategies to account for center-specific practices, so that the benefits achieved by fetal therapy vs surgical strategy can be demonstrated clearly. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.
INTRODUCTION
A proportion of fetuses with aortic valve stenosis (AoS) will develop hypoplastic left heart syndrome before birth, requiring postnatal univentricular (UV) palliation1-3. Fetal aortic valvuloplasty (FV) has been developed for the treatment of AoS during the past 15 years, with the intention of improving fetal left heart hemodynamics and promoting growth to achieve biventricular (BV) circulation.
Single-center studies have reported BV circulation outcome in one-third to two-thirds of fetuses with AoS undergoing FV4, 5, and an international anonymized registry reported 43% of such cases as having BV circulation compared with 20% of those untreated6. In our previous study on the natural history (NH) of 147 fetuses with AoS, 33% of those fulfilling hypothetical FV selection criteria had BV circulation2.
The Fetal Working Group of the Association for European Paediatric and Congenital Cardiology established a retrospective European study to assess the benefits of FV.
In the current study, we present survival and circulation following FV and, uniquely, compare these with contemporaneously enrolled NH cases sharing similar characteristics at presentation but not undergoing FV.
METHODS
Six centers performing FV in Europe submitted their outcome data on fetal AoS and on their NH cases collected contemporaneously from January 2005 to May 2012, with follow-up until April 2017. A further 17 fetal centers in 13 countries submitted NH data over the same time period and the live-born children were treated at one of 16 postnatal centers. Inclusion criteria were usual atrial arrangement, concordant atrioventricular and ventriculoarterial connections, and stenosed, but still patent, aortic valve. No fetuses with non-cardiac congenital malformations were enrolled. No maternal conditions or multiple pregnancies were excluded.
NH data reported here were published recently2 and pre-intervention echocardiograms from 109 neonates were reported in a blinded study of surgical decision making7.
The Ethics Committee at Imperial College London considered the study as audit of practice and no ethical approval was required.
Morphological and physiological data were entered into a standardized form by fetal cardiologists in participating centers, as described previously2. A.K. added missing measurements from available clips. Data included right- and left-sided valve and ventricular dimensions, and cardiac Doppler, including aortic and mitral valve pressure drop. Doppler waveforms of systemic and pulmonary veins, the ductus venosus, across the foramen ovale and the aortic and ductal arches were assessed, and fetuses with bidirectional or retrograde flow along most of the transverse aortic arch were classified as having retrograde flow. Demographic data, technical procedures and follow-up data were collated by A.K. and A.O., who calculated gestational age Z-scores for cardiac dimensions8.
All analyzed FV fetuses had AoS as the major lesion, defined as stenotic, but patent, aortic valves with qualitatively depressed left ventricular function, and all but one had retrograde arch flow. Primary outcome measures were survival and circulation. Secondary outcome measures were changes in fetal hemodynamics and left heart growth.
All centers performed FV percutaneously under ultrasound guidance using needles of 15–20 cm in length and 18–16 gauge, and coronary artery balloons of 2.0–4.0 mm in diameter with balloon to aortic valve ratio of 0.7–1.3. Technically successful FV (successful FV) was defined as balloon inflation resulting in increased flow when a balloon is placed across the aortic valve, with or without new regurgitation4, 5. Procedure-related events were defined as demise, or delivery resulting in death within 24 h of FV. Appendix S1 provides further procedural and technical details and Table S1 includes outcomes for FV cases in chronological order, according to FV center with outcomes reported up to April 2017.
Propensity score
Propensity score was used to assess the likelihood of a fetus with AoS receiving FV, enabling retrospective pseudorandomization of enrolled cases in a two-stage process. First, propensity score was derived from clinically important variables. Eligible FV cases included successful FV, unsuccessful FV and FV-related demise. All liveborn cases were required to have had postnatal intervention for AoS and known outcome with adequate data. We excluded spontaneous intrauterine fetal demise (sIUFD) and termination of pregnancy. Propensity score selected 54/67 FV (43 successful FVs, five unsuccessful FVs and six FV-related demises) and 60/80 NH cases. Second, propensity score cases were weighted and restricted to those with a propensity score within designated limits (0.14–0.9) to provide comparative cohorts9 using inverse probability of treatment weighting (IPTW) analyses. The final (IPTW) cohort for analysis was 42 FV and 29 NH9.
We tracked physiological changes and growth from first or immediate pre-FV echocardiograms to delivery. Hemodynamic changes of Doppler profiles through the foramen ovale, mitral valve and aortic arch were documented. Table S2 describes the relative weighting assigned to each Doppler flow based on clinical consensus of the authors. This enabled a comparison of hemodynamic changes during pregnancy; these, as well as changes in left heart Z-scores, were compared between three propensity score cohorts: successful FV, unsuccessful FV and NH.
The postnatal surgical pathway was considered UV if the first surgery was a Norwood or hybrid procedure and it was considered BV if the first surgery was aortic valvuloplasty (balloon or surgical) or a Ross/Ross–Konno procedure. BV–UV conversion was initial BV circulation intent followed by subsequent UV surgery (Norwood or hybrid), independent of its timing. There were no UV–BV circulation conversions. Survival was compared for final BV and UV pathways.
Data analysis
Frequencies and descriptive statistics were used to summarize baseline characteristics for each cohort. We developed a propensity score for IPTW analyses to compute the average treatment effect of FV (whether FV was successful or not), accounting for potential confounding by observed baseline characteristics. Logistic model predictors used to calculate the propensity score included: gestational age at first scan; restrictive foramen ovale; aortic arch and foramen ovale flow directions; aortic and mitral valve diameter Z-scores; mitral valve inflow Doppler pattern; left ventricular inlet length Z-score; left–right ventricular inlet-length ratio; hydrops; and large-center effect for fetal and postnatal treatment. A large center was defined as one contributing to the study data on 10 or more of both FV and postnatal procedures. The aortic valve pressure gradient at presentation was left out of the propensity score model since it did not improve the balance of baseline covariates.
Weights were calculated as the inverse of the propensity score. To obtain acceptable balance in baseline covariates, we restricted all IPTW analyses to observations with a propensity score of 0.14–0.909.
Overall survival and BV circulation survival (from fetal therapy to successful surgery) were compared between the FV and NH cohorts, using an IPTW logistic regression model with cohort as a covariate. Estimated odds ratios (ORs) and 95% CI are reported. Secondary analyses were performed, adjusted for a subset of six covariates: gestational age at first scan; mitral valve inflow Doppler; mitral and aortic valve Z-scores; left–right ventricular inlet-length ratio; and hydrops.
Liveborn survival was compared between the FV and NH cohorts using Kaplan–Meier survival curves and Cox regression with IPTW, adjusting for circulatory type and clustering of postnatal surgical center. Similarly, we compared survival in four groups for final circulation, including only successful FV: FV–BV, FV–UV, NH–BV and NH–UV.
Differences between the pre-FV and last fetal echocardiograms in left heart growth and hemodynamics were compared among successful FV, unsuccessful FV and NH cohorts using linear regression with IPTW. Statistical significance was defined as P < 0.05. All analyses were conducted in Stata 14.2 (Stata Corp, College Station, TX, USA).
RESULTS
Entire FV cohort
Sixty-seven fetuses undergoing FV were reported from six centers. Median gestational age at referral for FV was 25 (range, 15–33) weeks, and the procedure was performed at median gestational age of 26 (21–34) weeks. There were 72 procedures performed: three had repeat FV, one was unsuccessful on both occasions and one repeat FV had been thought successful initially, 1 month previously. Interatrial-septum ballooning/stenting was performed in two cases after FV (repeated in one). Figure 1 shows outcomes for the entire FV cohort and Table S1 shows outcome data reported up to April 2017 in chronological order of procedure.
FV-related death occurred in 7/72 (10%) procedures, including six considered successful FV. Rare adverse events included left ventricular thrombus formation and balloon rupture. One serious maternal complication (placental abruption) resulted in delivery at 25 weeks' gestation. Fifty-nine fetuses had successful FV and 19/43 (44%) treated neonates had BV circulation.
Eight fetuses had unsuccessful FV (1/8 developed intact atrial septum resulting in fetal demise), of which 4/5 survivors were UV, and 1/5 with retrograde arch flow and monophasic mitral valve inflow had BV circulation and was alive at age 5.7 years (at the time of writing) without pulmonary hypertension.
Hydrops was present in 24/59 successful FV cases and resolved in 9/24 affected fetuses. The course and outcomes are presented in Table 1. One additional case presented with hydrops that did not resolve after unsuccessful FV at 21 gestational weeks and resulted in sIUFD.
| GA at FV (weeks) | GA at birth (weeks) | Circulatory outcome | Follow-up age (years) |
|---|---|---|---|
| Presented at first echocardiogram, resolved (n = 9) | |||
| 23 | 41.4 | CC | |
| 21 | 36.4 | UV dead | 0.003 |
| 22 | 39.0 | UV alive | 3.7 |
| 26* | 40.0* | UV alive* | 6.5* |
| 25 | 34.7 | BV alive | 10.1 |
| 26 | 39.1 | BV alive | 5.7 |
| 27 | 32.9 | BV alive | 6.7 |
| 28 | 34.4 | BV alive | 11.5 |
| 33 | 36.9 | BV alive | 11.4 |
| Developed after first echocardiogram, not resolved (n = 7) | |||
| 21 | — | sIUFD | |
| 23 | — | sIUFD | |
| 25 | — | TOP | |
| 26, 30 | — | CC | |
| 30* | 40.0* | UV alive* | 6.7* |
| 24 | 40.0 | BV alive | 4.9 |
| 30* | 33.6* | BV alive* | 7.5* |
| Presented at first echocardiogram, not resolved (n = 8) | |||
| 28 | — | sIUFD | |
| 29 | — | sIUFD | |
| 25 | 25.0 | FV-NND | 0.003 |
| 27 | 39.0 | NND (presurgery) | 0.027 |
| 26, 27 | 36.6 | BV–UV dead | 0.014 |
| 24* | 37.1* | BV dead* | 0.125* |
| 31* | 39.0* | BV dead* | 0.25* |
| 30* | 36.0* | BV–UV alive* | 5.5* |
- * Case included in inverse probability of treatment weighting cohort. BV, biventricular circulation; BV–UV, biventricular to univentricular conversion; CC, comfort care; FV-NND, neonatal death related to FV; GA, gestational age; NND, neonatal death; sIUFD, spontaneous intrauterine death; TOP, termination of pregnancy; UV, univentricular circulation.
Sustained hemodynamic improvement was documented in 29/43 (67%) successful FV cases undergoing postnatal procedures, with temporary improvement in another five (12%). One fetus with BV outcome improved initially, but subsequently developed an intact atrial septum and hydrops. Four with UV outcome showed no hemodynamic improvement, or deteriorated following FV; one had inadequate follow-up data to evaluate change. Seven out of eleven (64%) fetuses with subsequent BV–UV conversions showed sustained hemodynamic improvement after FV and two showed temporary improvement. Following unsuccessful FV, the five liveborns demonstrated no fetal hemodynamic improvement and only one achieved BV circulation.
Median gestational age at delivery in the FV cohort was 38.0 (range, 25.0–41.4) weeks, but 22/53 (42%) were delivered before 37 + 0 weeks, compared with 22/85 (26%) of the NH cohort2. Outcomes following premature delivery were similar in both cohorts, with two-thirds surviving (70% of which had BV circulation). Birth weight was < 10th centile in 11 in each cohort, but all but one of these delivered at term. The children underwent a median of three (range, 1–12) procedures. Six neonates had persistent pulmonary hypertension (one died from multiorgan failure before the procedure and five had BV procedures with one surviving to 3 years). Three children had late-onset pulmonary hypertension, in one case after a Norwood procedure, delaying the Glenn procedure, but the Fontan procedure was completed and the child was alive at age 5 years (at the time of writing), treated with sildenafil. The two others were BV–UV conversions (one early and one aged 18 months); both died. Seven of these nine were included in the weighted analysis.
Propensity score modeling and IPTW analysis
Table 2 compares the baseline characteristics of the first scan used to derive the propensity score model and IPTW cohort used in the weighted analyses, resulting in between-group balance on baseline characteristics with standardized differences of 0.14 or less. The postnatal circulatory outcomes for the liveborn weighted cohorts were similar (36% and 38%) (Table 3).
| Characteristic | Whole cohort (with sufficient data) | Inverse probability of treatment weighting cohort | ||||
|---|---|---|---|---|---|---|
| FV (n = 55) | NH (n = 80) | Standardized difference | FV* (n = 42) | NH* (n = 29) | Standardized difference | |
| Gestational age at scan (weeks) | 25.7 ± 3.7 | 25.4 ± 4.7 | 0.08 | 25.7 ± 3.7 | 25.5 ± 4.8 | 0.06 |
| Aortic valve diameter Z-score | −1.3 ± 1.34 | −1.62 ± 2.15 | 0.17 | −1.64 ± 1.32 | −1.51 ± 2.07 | −0.08 |
| Mitral valve diameter Z-score | −0.97 ± 1.96 | −2.21 ± 2.43 | 0.54 | −1.85 ± 1.99 | −2.00 ± 1.93 | 0.08 |
| LV:RV length ratio | 1.03 ± 0.23 | 0.97 ± 0.22 | 0.26 | 0.99 ± 0.23 | 0.99 ± 0.20 | 0.01 |
| FO right-to-left flow | 1/54 (2) | 26/72 (36) | −0.97 | 1 (4) | 1 (4) | −0.01 |
| AoA retrograde flow | 54/55 (98) | 42/80 (53) | 1.25 | 34 (97) | 35 (98) | −0.08 |
| Mitral valve biphasic flow | 17/55 (31) | 39/68 (57) | −0.55 | 7 (20) | 7 (19) | 0.02 |
| Hydrops | 12/55 (22) | 3/80 (4) | 0.56 | 2 (7) | 1 (4) | 0.14 |
| LV inlet length Z-score | −0.47 ± 1.55 | −1.04 ± 2.03 | 0.30 | −0.80 ± 1.74 | −0.69 ± 1.88 | −0.06 |
| Large center† | 15 (44) | 14 (38) | 0.11 | |||
| AoVPG (mmHg)‡ | 16.2 ± 14.3 | 15.0 ± 11.8 | 0.09 | |||
- Data are presented as mean ± SD, n/N (%) or n (%).
- * Numbers weighted by inverse probability.
- † Number of cases presenting for FV and/or postnatal surgery at one or more large center.
- ‡ Not included in propensity score model.
- AoA, aortic arch; AoVPG, aortic valve pressure gradient; FO, foramen ovale; LV, left ventricle; RV, right ventricle.
| Final circulation | FV | NH | Total |
|---|---|---|---|
| BV | 13 (36.1) | 11 (37.9) | 24 |
| BV–UV conversion | 10 (27.8) | 4 (13.8) | 14 |
| UV | 13 (36.1) | 14 (48.3) | 27 |
| Total | 36 | 29 | 65 |
- Data are given as n (%) or n.
- BV, biventricular; UV, univentricular.
Survival and circulatory outcomes
Overall survival was similar in FV and NH cohorts (OR, 1.57; 95% CI, 0.72–3.41; P = 0.25), as was BV circulation survival (OR, 1.31; 95% CI, 0.23–7.48; P = 0.76). Secondary analyses adjusting for additional covariates gave similar results (not shown).
The six procedure-related fetal deaths were removed to create survival analyses of livebirths. The age at which the first postnatal procedure was performed was similar in both cohorts; median age was 6 (range, 1–56) days for FV and 4 (range, 1–74) days for NH. IPTW Cox-regression analysis, adjusting for clustering due to surgical center, showed that FV conferred postnatal survival advantage after adjusting for circulation (hazard ratio (HR), 0.38; 95% CI, 0.23–0.64; P = 0.0001; Figure 2). The final circulations were compared for each cohort (after removing unsuccessful FV cases) and survival over 10 years' follow-up was similar (HR, 0.54; 95% CI, 0.14–2.08; P = 0.37; Figure 3). Pairwise comparisons of the marginal linear predictions are included in Table S3.
) and those that did not (
). FV group includes both technically successful and unsuccessful FV cases on intention-to-treat basis. Six procedure-related fetal deaths were removed to create survival analyses of live births. Actual number of included individuals at each time period is documented below curves (although weighted values were used in analysis and curves). Time zero represents birth. IPTW Cox-regression analysis, adjusting for clustering due to surgical center, showed FV conferred survival advantage after birth (hazard ratio, 0.38; 95% CI, 0.23–0.64; P = 0.0001), even after adjusting for circulation and postnatal treatment center.
); FV-BV (
); NH-UV (
); NH-BV (
). FV group comprises only technically successful FV cases. Actual number of individuals included at each time period is documented below curves. Time zero represents birth. IPTW Cox-regression analysis, adjusting for clustering due to surgical center, showed no difference in survival between postnatal circulations in FV cohort (hazard ratio, 0.54; 95% CI, 0.14–2.08; P = 0.37), after adjusting for postnatal treatment center.Fetal hemodynamics and left heart growth
The IPTW analysis shows hemodynamic improvement was significantly better following successful FV than following a failed attempt, but did not differ significantly from the NH group. However, left heart growth was significantly worse in the NH than in the successful FV group (Table 4). The small number of technically unsuccessful cases appeared to show similar left ventricular and aortic valve growth to those of the successful FV cohort, but had a significantly reduced mitral valve size by delivery. The hemodynamic and left heart growth data used to create the propensity score are included in Table S4.
| Parameter | FV technically successful | FV technically unsuccessful | Natural history | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | First | Last | Δ | n | First | Last | Δ | P * | n | First | Last | Δ | P * | |
| Aortic valve diameter Z-score | 32 | −1.29 ± 1.17 | −1.5 ± 1.55 | −0.21 ± 1.53 | 6 | −1.78 ± 1.75 | −3.04 ± 1.02 | −1.26 ± 1.68 | 0.09 | 19 | −2.38 ± 2.22 | −3.12 ± 2.40 | −0.74 ± 3.06 | 0.01 |
| Mitral valve diameter Z-score | 32 | −1.25 ± 1.91 | −1.24 ± 1.55 | 0.01 ± 1.98 | 5 | −2.07 ± 1.16 | −3.42 ± 1.46 | −1.35 ± 1.18 | 0.04 | 17 | −2.48 ± 1.60 | −3.41 ± 2.37 | −0.93 ± 2.78 | 0.02 |
| LV inlet length Z-score | 31 | −0.46 ± 1.67 | −1.14 ± 1.97 | −0.69 ± 1.33 | 6 | −1.63 ± 1.35 | −2.3 ± 2.01 | −0.67 ± 1.05 | 0.83 | 20 | −1.21 ± 1.73 | −3.21 ± 2.63 | −2.01 ± 2.58 | 0.002 |
| LV-EDD Z-score | 31 | 0.96 ± 1.92 | 0.36 ± 2.23 | −0.61 ± 2.01 | 5 | −0.86 ± 1.23 | −0.34 ± 1.34 | 0.52 ± 1.22 | 0.11 | 19 | 0.34 ± 1.70 | −1.64 ± 2.74 | −1.99 ± 2.64 | 0.006 |
| Hemodynamics weighted score | 26 | 14.15 ± 2.92 | 11.15 ± 4.88 | 3.00 ± 4.72 | 6 | 16.67 ± 2.04 | 16.25 ± 1.37 | 0.42 ± 1.02 | 0.04 | 16 | 14.03 ± 3.44 | 12.72 ± 4.32 | 1.31 ± 2.73 | 0.09 |
| Composite hemodynamic score | 15 (12.5–15) | 10.25 (9.5–15) | 2.75 (0–5) | 16.25 (15–17.5) | 16.25 (15–17.5) | 0 (0–0) | 15 (14.5–15) | 14.5 (10–15) | 0 (0–3.5) | |||||
- Data are presented as mean ± SD or median (interquartile range).
- * Comparison of difference between scans (Δ) with that in FV technically successful cases, calculated from IPTW regression.
- EDD, end-diastolic diameter; LV, left ventricle.
DISCUSSION
During the study period, the selection of cases for FV was evolving worldwide. Initially, fetal cardiologists hoped that FV could achieve BV circulation in fetuses with short left ventricles and endocardial fibroelastosis, compared with NH. Subsequent experience has shown that only selected fetuses appear to benefit; however, selection criteria remain elusive.
As a prospective randomized controlled trial was not feasible, we used the propensity score to provide pseudorandomization of our retrospective data. We observed similar proportions with BV circulation outcome in our IPTW intention-to-treat cohorts (36% for the FV and 38% for the NH cohort). IPTW logistic analysis showed that FV did not confer survival or circulatory benefits but, when procedure-related deaths were removed and Cox regression was adjusted for circulation and surgical center, FV reduced the risk of early postnatal death by two-thirds. Survival to 10 years' follow-up in this cohort was similar for those with final BV circulation and those with final UV circulation, with no deaths after 2.3 years.
FV was introduced into clinical practice without a trial, and many centers performed procedures without reporting outcomes. An international, anonymized registry was established recently to collect multicenter data, but currently lacks independent audit, making data validation difficult6. Contemporaneously matched controls (rather than choosing those with unsuccessful FV) and treatment randomization are missing from FV publications1, 4-6, 10, 11, providing only Level-3 evidence of treatment efficacy7.
Our study contributes to the global experience of FV and is strengthened by contemporaneously collected NH controls. Several important observations can be drawn. Firstly, our 10% procedure-related loss was similar to that found in single-center studies4, 5, 9 and less than the 17% reported in the anonymized registry6, highlighting the importance of experienced teams mentoring new FV centers. Secondly, fetal Z-scores demonstrated favorable anatomy for FV, indicating good case selection. Thirdly, the operators' evaluation of FV success was accompanied by objective changes in fetal hemodynamics; hydrops resolved in over one-third of affected fetuses, and two-thirds of all successful FV showed sustained hemodynamic improvement, including reversal of previously retrograde arch and foramen flow and new biphasic ventricular filling. These individuals had BV circulation outcomes more commonly.
The prevalence of premature delivery (< 37 + 0 weeks) has not been reported following FV. Premature delivery occurred in 42% of FV cases compared with 26% (22/85) of NH cases. From the limited maternity data that were collected, growth restriction (estimated fetal weight < 10th centile) was found not to be responsible for early delivery as it occurred almost exclusively in those delivering > 37 weeks. Early delivery may represent institutional practice (unsubstantiated by evidence) to avoid worsening left heart function and aortic valve closure and, in this study, was more frequent following fetal intervention. Delivery was not centralized to the site performing the FV and lack of familiarity in disease assessment may be a contributing factor.
Recognition that left heart growth remains suboptimal following successful FV has resulted in modifications to selection criteria. Although important in long-term ventricular function, the diagnosis and grading of endocardial fibroelastosis by ultrasound remains elusive due to poor accuracy in comparison with histology12. Although it was originally included as a predictive variable13, it has been removed recently due to the qualitative nature of grading and only modest interobserver reliability12, 14. Newer selection criteria include left ventricular inlet length Z-score > 0 at presentation and pressure generation ≥ 20 mmHg4, 5. While one group applied the 2010 criteria hypothetically to a small series and described it as predictive of outcome3, our larger dataset suggests otherwise. Of the 40 NH fetuses satisfying criteria for emerging hypoplastic left heart syndrome, 13 (33%) had BV circulation despite eight falling below the threshold score to be theoretically offered FV2. Importantly, our hemodynamic and left heart growth data suggest that outcomes following unsuccessful FV are not similar to NH, making unsuccessful FV cases unsuitable as controls.
Prospective fetal therapy trials for open meningomyelocele surgery and laser therapy for twin-to-twin transfusion syndrome15, 16 demonstrate that successful fetal procedures rely on case selection, technical prowess and integrated postnatal management. Therefore, refinement of FV selection criteria, unsupported by a trial, may increase the chance that FV is offered to those who would achieve BV circulation without fetal therapy, with the associated risks of procedure-related mortality and fetal and/or maternal morbidity.
We have previously discussed postnatal selection bias and its effects, which touches upon the ethos and ability of the entire pre- and postnatal team in decision making7. We note that BV circulation survival is relatively low in our study, similar to survival to hospital discharge of 58% (irrespective of FV) reported in a recent multicenter registry report, and less than that reported in a single center report6, 11. Poor outcome may be associated with the complexity of congenital AoS resulting in multiple procedures in addition to premature delivery.
Even though we accounted for surgical center variability in our analyses, unrecognized center-specific differences in delivering affected babies preterm to initiate earlier treatment, decision-making regarding postnatal management, skill and practice may potentially confound our results. The postnatal treatment centers had different postnatal strategies, in part because the range of surgical options was not available in all cardiac surgical centers in this study17-21. An aggressive approach may preclude later conversion to UV and result in early mortality and the risk of later pulmonary hypertension11, 17-22.
Limitations of the current study include its retrospective multicenter design with a limited cohort size. Low rates of prenatal diagnosis of AoS23 make a prospective, randomized FV trial challenging; therefore, we used IPTW in our study to minimize the deficiencies in our dataset. Although the number of postnatal cardiac centers may have introduced unrecognized confounding and bias into the assessment of the efficacy of FV in this study, the statistical model we used adjusted for clustering due to center influence.
In conclusion, we report improvements in fetal hemodynamics and preservation of left heart growth following successful FV compared to NH. While the proportion of those achieving a BV circulation outcome was similar in both cohorts, FV survivors showed improved survival, independent of final circulation, to 10 years' follow-up. However, FV is associated with a 10% procedure-related loss and increased prematurity compared with NH, and therefore the risk–benefit ratio remains uncertain.
We recommend a carefully designed trial, incorporating appropriate and integrated fetal and postnatal management strategies to account for center-specific practices, so that the benefits achieved by fetal therapy vs surgical strategy can be demonstrated clearly.
Acknowledgments
This study was devised by the Fetal Working Group of the Association for European Paediatric Cardiology. Financial support for site visits by A.K. and some statistical assistance was provided by the Fetal Working Group of the AEPC. Additional funding for statistical analysis was provided by Children's Heart Unit Fund (CHUF), Royal Brompton Hospital Charities (Registered Charity No: 1053584); Oberösterreichische Spitals AG; donations to The Fetal Center at Children's Memorial Hermann Hospital, University of Texas, Houston; and from the coauthors' private funds. A.K. was supported by a grant from the Department of Paediatric Cardiology, Heinrich Heine University Duesseldorf, Germany. M.M. and A.Ö. were supported by the Swedish Heart and Lung Foundation.
Contributors to the study including members of the Fetal Working Group of the AEPC
Joaquim Bartrons, Department of Paediatric Cardiology, Hospital Clinic of Barcelona, Sant Joan de Déu, Barcelona, Spain;
Frances Bulock and Suhair Shebani, Congenital and Paediatric Cardiac Service, Glenfield Hospital, Leicester, UK;
Sally Ann Clur, Academic Medical Center, Amsterdam, The Netherlands;
Ingo Daehnert, Clinic for Pediatric Cardiology, Heart Centre, University of Leipzig, Germany;
Giovanni Di Salvo, Department of Pediatric Cardiology, Second University of Naples, Monaldi Hospital, Naples, Italy;
Ruth Heying and Marc Gewillig, Department of Congenital and Pediatric Cardiology, University Hospital Gasthuisberg, Leuven, Belgium;
Els Grijseels and Laurens Koopmann, Departments of Prenatal Medicine and Pediatric Cardiology, Erasmus MC/Sophia Children's Hospital Rotterdam, The Netherlands;
Kaarin Makikallio, Aydin Tekay and Markku Leskinen, Departments of Obstetrics and Gynaecology and of Pediatrics, Oulu University Hospital, Oulu, Finland;
Nicky Manning and Nick Archer, Paediatric and Fetal Cardiology Service, John Radcliffe Hospital, Oxford, UK;
Renate Oberhoffer, The German Heart Centre, Munich, Germany;
Cristina Romeo, Department of Cardiology, Azienda Ospedaliera Bolzano, Bolzano, Italy;
Keld Ejvind Sørensen, Skejby University Hospital, Aarhus, Denmark;
Trevor Richens, Department of Cardiology, Royal Hospital for Sick Children, Yorkhill Division, Glasgow, UK;
Klaus Schmidt, Department of Paediatric Cardiology, Heinrich Heine University Duesseldorf, Germany;
Anna Seale and Victoria Jowett, Queen Charlotte's Hospital and Royal Brompton Hospital, London, UK;
Cecile Tissot, Pediatric Cardiology Unit, The Children's University Hospital of Geneva, Geneva, Switzerland;
Viktor Tomek, University Motol Hospital, Prague, Czech Republic;
Frank Uhlemann, Department of Pediatric Cardiology and Pediatric Intensive Care, Olgahospital, Stuttgart, Germany;
Niels Vejlstrup, Department of Cardiology, Rigshospitalet, Copenhagen, Denmark;
Jochen Weil, Clinic for Paediatric Cardiology, University Heart Centre Hamburg, Germany;
Adam Koleśnik, The Children's Memorial Health Institute, Cardiology Department, Warsaw, Poland;
Marzena Dębska, The 2nd Department of Obstetrics and Gynecology, The Center of Medical Postgraduate Education, Warsaw, Poland;
Agata Włoch, Department of Obstetrics and Gynecology, Medical University of Silesia, Katowice, Poland;
Paweł Dryżek, Department of Cardiology, Polish Mother Memorial Hospital Research Institute, Lodz, Poland;
Maciej Chojnicki, Department of Pediatric Cardiac Surgery, Mikolaj Kopernik Hospital, Gdansk, Poland.
