Progesterone for Neurodevelopment in Fetuses With Congenital Heart Defects

Key Points Question Does maternal progesterone therapy in fetuses with congenital heart defects (CHD) improve neurodevelopmental outcomes? Findings In this randomized clinical trial of 102 mother-fetal dyads receiving maternal progesterone therapy, the treatment benefit of progesterone in the overall population was small and not statistically different from 0. Subgroup analyses suggested heterogeneity of treatment effect in Motor Scores by fetal sex and among fetuses with different types of CHD. Meaning These results suggest that continued research into the possible benefits of progesterone for neurodevelopmental outcomes for a subset of patients with CHD appears warranted.


eMethods 1.Exclusion Criteria
The exclusion criteria were: 1) major genetic or extra-cardiac anomaly other than 22q11 deletion; 2) language other than English spoken at home; 3) known sensitivity or listed contraindication to progesterone (known allergy or hypersensitivity to progesterone, severe hepatic dysfunction, undiagnosed vaginal bleeding, mammary or genital tract carcinoma, thrombophlebitis, thromboembolic disorders, cerebral hemorrhage, porphyria); 4) prescription or ingestion of medications known to interact with progesterone (e.g.bromocriptine, rifamycin, ketoconazole or cyclosporin); 5) maternal use of progesterone within 30 days of enrollment; 6) history of preterm birth or short cervix (defined as cervical length ≤ 25 mm at 18-24 weeks GA necessitating progesterone therapy; 7) multiple gestation; 8) maternal contraindication for magnetic resonance imaging (MRI); and 9) participants with a known history of non-compliance with medical therapy.

Family Reimbursement
For three study-related visits, families received financial reimbursement for per diem expenses, travel expenses (airfare, hotel, mileage, parking, tolls, etc.), and childcare.The family was additionally given a gift card ($50) after each visit.

Enrollment, Randomization, Study Drug Preparation, Administration and Adherence
A study nurse met weekly with the Fetal Heart Program to identify potential subjects.
Potential subjects were screened for eligibility at 24-28 weeks GA by a study nurse using the protocol inclusion and exclusion criteria.Permission to review the research study in person was asked of eligible subjects at a prenatal visit.If subjects agreed to a research review, the informed consent process was initiated to assure subjects were able to comprehend the purpose of the study, the study procedures, and the risk-benefit profile.
Participants who agreed to participate were 1:1 block-randomized to progesterone or placebo by fetal CHD diagnosis (hypoplastic left heart syndrome (HLHS), transposition of the great arteries (TGA), and other CHD (OTHER).The block size was four.The randomization scheme was created by the study statistician (MP) and uploaded into a REDCap database.Independent of the study team, a research pharmacist used the randomization table to assign eligible participants to either progesterone (8% gel) or placebo, a vaginal moisturizer.The study was double-blinded throughout; the statisticians were blinded to the identity of the intervention arms until the interim analysis requested by the DSMB in January 2020.The mothers received care in the Richard D Wood Jr. Center for Fetal Diagnosis and Treatment at the Children's Hospital of Philadelphia with planned delivery in the Garbose Family Special Delivery Unit.

Study Drug Preparation, Administration and Adherence
The treatments were packaged in identical prefilled applicators (Progesterone, 8% gel equivalent to 90 mg /applicator).Participants received the initial dispense of study drug at their initial visit and then monthly.Drug administration was initiated prior to 28 weeks GA.For this study, up to 28 weeks 3 days was considered as 28 weeks GA.Participants were to administer the medication vaginally twice daily until 39 weeks gestational age (GA) or delivery of the baby, whichever happened first.Participants were asked to record the date and time of each dose along with any side effects in a study diary.The study team met with mothers during each prenatal visit to dispense study drug, collect unused applicators, and answer questions.For study-related visits, families received financial reimbursement.
Adherence to the study drug administration was based on the returned applicators and the diary.Patients were asked to return all unused medication applicators at each prenatal visit.The number of doses taken was determined from the difference between the number of doses dispensed and returned unused.If a subject did not return their unused applicators, the number of doses taken was determined from their diaries.If the medication was stopped before 39 weeks GA due to delivery or study termination, the calculation was based on the total number of doses available from study enrollment up to medication termination.For 92 participants (90%), adherence was calculated as the percentage of doses taken (the number of doses available up until medication termination dispensed minus the number of doses returned) divided by the number of possible doses.Occasionally (n=7, 7%), calculated adherence exceeded 100%, apparently because some unused applicators were not returned after withdrawal or delivery.In this case, the reported adherence was truncated to 100%.For one subject who did not return any applicators, adherence was estimated solely based on the diary.Two participants did not return applicators or a diary and adherence could not be calculated.Adverse events were monitored throughout the study by regular contact with the participants, patient diaries, and review of the medical record.Maternal adverse event terms and grades were monitored and collected using the Common Terminology Criteria for Adverse Events, version 4.0 (CTCAE v.4) developed by the National Institutes of Health.Child adverse events were collected using an adverse event reporting system developed by the Pediatric Heart Network (established by the National Heart, Lung, and Blood Institute) for reporting adverse events in clinical trials for complex congenital heart disease. 5

Delivery and Neonatal Hospitalization
After delivery, a cord blood sample was collected for genetic testing.

Genetic Evaluation and APOE Genotyping
Clinical genetic testing was performed as indicated.While major genetic anomalies were an exclusion criterion at enrollment, universal prenatal recognition of these anomalies is difficult.
Thus, some enrolled infants received the diagnosis of a genetic syndrome only after the 18month evaluation or by genetic testing.Whole exome sequencing and a microarray were performed on the participants, if not obtained clinically.Pathologic or likely pathologic genetic syndromes or chromosomal abnormalities were identified.Based on the clinical evaluation and genetic testing, patients were classified as normal (no pathogenic or likely pathogenic variants identified), as abnormal (pathogenic or likely pathogenic variants related to CHD or neurodevelopment identified), or as suspect (variants of unknown significance were identified or there were concerns on the clinical examination).Variant calls were queried for single nucleotide polymorphisms to determine apo-lipoprotein E (APOE) genotype.Variant call files were obtained from the clinical and research testing laboratories filtered using BCFtools v1.9 to contain only two variants, rs429358 and rs7412.APOE haplotypes were subsequently defined based on individual rs429358/rs7412 genotypes as follows: C/T, ε1; T/T, ε2; T/C, ε3; C/C, ε4.Patients were classified as ε2 (ε2ε2 and ε2ε3), ε3 (ε3ε3), or ε4 (ε3ε4 and ε4ε4).Patients with ε2ε4 were excluded.

Outcome Variables
Neurodevelopmental

Per Protocol Analyses
Results for participants who used at least 80% or 90% of their intended dose for a 39-week pregnancy appears in Supplemental Table S9.Per-protocol models were adjusted for diagnosis, sex, and presence of genetic anomalies.In the >90% group, the mean improvement for the Motor Score for progesterone versus placebo was 4.0 units (90% CI -1.0, 8.9).For Language, the mean improvement after adjustment was 1.76 (90% CI -5.2, 8.7) and for Cognition 0.37 units (90% CI -5.1, 5.9).Results were similar for > 80%.

eReferences
1. de Onis M, Garza C, Victoria C, Onyango A, Frongillo E, Martines J.The assignment to HLHS, TGA, or OTHER was based on fetal echocardiography prior to 28 weeks gestational age.Cardiac Class was assigned post-natally based on clinical findings and management.However, there may be changes over the third trimester of pregnancy and with better post-natal assessment.One subject assigned to HLHS underwent a 2-ventricular repair (Class II) and one had single ventricle physiology without arch obstruction (Class III).Also, two of the subjects assigned to TGA, had TGA/VSD with arch obstruction not appreciated at the time of randomization.Also, the majority of the OTHER patients had arch hypoplasia/obstruction with anatomy suitable for 2-ventricular repair (Class II).One was found to have single ventricle physiology with arch obstruction after birth and underwent a Norwood procedure (Class IV).
a One fetus in the HLHS category died prior to the surgery necessary to delegate which cardiac class it would have been assigned.
b Three TGA fetus' withdrew prior to the study initiation.b Three participants with TGA (2 on progesterone and 1 on placebo) and one subject with HLHS (placebo) were missing operative data.These participants were included in the denominator when calculating percentages.
c Unit of measurement N (%) unless otherwise indicated.Two participants returned for the 18-month follow-up visit but were missing all BSID-III Scores.An additional subject returned for follow-up but had no BSID-III Motor Score.Cumulative mortality among all participants born on-study (left) and among those with an HLHS diagnosis (right panel).Follow-up ends at 18 months.All but one death, (a TGA subject allocated to progesterone), occurred in those with an HLHS diagnosis.

Figure S2: Common Maternal Adverse Events
Most common adverse events among mothers on Progesterone (right) and mothers on placebo (left).Adverse events were restricted to at least 5% prevalence among the population and further restricted to exclude those that were not related to either Progesterone or the Placebo.

eFigure 3: Common Infant Adverse Events
Most common adverse events among infants stratified by Progesterone (right) and Placebo (left).Adverse events were restricted to at least 5% prevalence among the population and further restricted to exclude those that were not related to either Progesterone or the Placebo.
outcomes were assessed using The Bayley Scales of Infant and Toddler Development-III; which provide composite scores for Cognitive, Language and Motor Scales (µ = 100 with σ = 15).For the Motor and Language composites, there are component scores: Motor (Fine and Gross) and Language (Expressive and Receptive), (µ = 10 with σ = 3).Higher scores indicate better skills. 5Additional variables of interest pre-specified prior to unblinding and data analysis include: GA of neonate at time of birth (both continuous and whether born earlier than 37 weeks GA, delivery type (vaginal or cesarean-section, either elective or emergent), placental weight and placental weight <10th percentile, birthweight/placental weight ratio, presence of placental infarction, placental fetal or maternal vascular malperfusion and placental fetal vascular thrombi, cardiac operation during first admission, age at first operation, use of cardiopulmonary bypass (CPB), weight at first operation, Cardiac Class, deep hypothermia © 2024 Gaynor JW et al.JAMA Network Open.circulatory arrest (DHCA) time, total support time (TST), antegrade cerebral perfusion time, need for extracorporeal membrane oxygenation (ECMO), additional operations with CPB, additional TST, additional DHCA, mortality (in-hospital, on-study and post-study, length of stay and selected complications for first surgical admission.

a
Two participants had antegrade cerebral perfusion time for first operation.Values were 49 minutes (placebo) and 56 minutes (progesterone).

c
One individual assigned to placebo was missing birthweight data.d Two individuals, one assigned to placebo and one assigned to progesterone, are missing placental weight data.© 2024 Gaynor JW et al.JAMA Network Open.

Baseline Characteristics of Mothers and Fetuses Stratified by Cardiac Diagnosis HLHS Other TGA
. Virzi L, Pemberton V, Ohye RG, Tabbutt S, Lu M, Atz TC, Barnard T, Dunbar-Masterson C, Ghanayem NS, Jacobs JP, et al.Reporting adverse events in a surgical trial for complex congenital heart disease: the Pediatric Heart Network experience.J Thorac Cardiovasc Surg.

Baseline Characteristics of Mothers and Fetuses Stratified by Cardiac Diagnosis HLHS Other TGA
© 2024 Gaynor JW et al.JAMA Network Open.eTable 1:

Baseline Characteristics of Mothers and Fetuses Stratified by Cardiac Diagnosis
a N (%) represented unless otherwise indicated.bImpaired MFE indicates mother with one or more of gestational hypertension, gestational diabetes, tobacco use or hypothyroidism at baseline or pre-eclampsia during the pregnancy.c Thirty-three participants were Cardiac Class I; 13 participants were Class II; two s were Class III; 50 were Cardiac Class IV. (See Supplemental Table S2) d Genetic classification, 22q11.2deletion, and APOE genotype are often not known until after birth.Three participants withdrew before birth, contributing to the missing data.e ε2 indicates homozygous for ε2 or heterozygous ε2/ ε3.ε3 indicates homozygous for ε3.ε4 indicates homozygous for ε4 or heterozygous ε4/ε3.© 2024 Gaynor JW et al.JAMA Network Open.

eTable 3: Results of Genetic Evaluation and Testing
© 2024 Gaynor JW et al.JAMA Network Open.

eTable 4: Reasons for Drug Termination Before 39 Weeks GA
a Percentage of participants who terminated study drug ≤39 weeks GA b 90% CI for the percentage difference using a Chi-square distribution eTable 5:

Delivery and Neonate Characteristics Outcomes a
Treatment effect.P-value is for a univariate test or in the case of multiple categories, a global test.Specific model indicated for each feature.All models stratified by cardiac diagnosis.Two participants (4%) had missing data for delivery type, induction of labor, birthweight, placental weight, and BW:PW.Postmenstrual age missing in 1 subject.Percentages based on N=50.
© 2024 Gaynor JW et al.JAMA Network Open.b d e Multinomial or logistic regression model.Values are odds ratios for progesterone versus placebo for each category versus the reference group (REF).fOrdinallogistic regression model.The youngest two GA (<32 weeks and 32-37wks 6 d) were combined for the regression model.The estimate is the odds of falling into the lower age categories for progesterone versus placebo e.g., the odds of being in a GA category less than 39 weeks is 0.6-fold smaller for progesterone versus placebo.gLinear regression model.Values are mean differences for progesterone vs placebo.h Linear regression model using log-transformed outcome variable.Values are ratio of BW:PW for progesterone vs placebo.I Unit of measurement (N %) unless otherwise indicated.© 2024 Gaynor JW et al.JAMA Network Open.

Treatment Effect Estimates for BSID-III Scales Using Multiple Imputation to Address Incomplete Outcome Data a BSID-III Scale Pooled Across All Diagnoses b Stratified by Diagnosis c
b Data pooled across all diagnoses c Analyzed using a linear model stratified by diagnosis © 2024 Gaynor JW et al.JAMA Network Open. a Analyzed using a linear model stratified by diagnosis, with sex, MFE, and genetic profile as additional covariates.© 2024 Gaynor JW et al.JAMA Network Open. a Unit of measurement N (%) unless otherwise indicated.