Cardiac Performance in the First Year of Age Among Preterm Infants Fed Maternal Breast Milk

Key Points Question What are the associations of exposure to mother’s own milk on cardiac performance during the first year of age in preterm infants? Findings This cross-sectional study of 80 preterm infants found that those with higher exposure to mother’s own milk had enhanced cardiac function and morphology on echocardiographic examination at age 1 year, with values approaching those of full-term infants in the control group. Meaning These findings suggest that there is a favorable association between consumption of maternal breast milk and cardiac performance at age 1 year in preterm infants.


Defining High vs. Low Mothers Own Milk Exposure Groups
The PROP study design defined high breast milk exposure as greater than 28 days (4 weeks) of consumption of mother's own milk (MoM) feeds before 36 weeks PMA. 8 The 28-day cut-off mark was chosen to 1) avoid any concerns with missing data and 2) not bias against infants with feeding difficulties (eFigure 1). In post-hoc analysis, we determined that to detect right ventricular (RV) (which is >3 months for an extreme preterm infant), we also analyzed the cohort based on their discharge nutritional status, MoM vs. bovine formula. To note, as per standard center protocol, infants born less than 29 weeks gestation received bovine fortifiers when they reach 100 milliliter per kilogram per day of feeds and remained on fortification through discharge.

Echocardiographic Measures
Echocardiograms were acquired with commercially available ultrasound imaging systems (Vivid 7 and E9; General Electric Medical Systems, Milwaukee, Wisconsin). One designated experienced pediatric cardiac sonographer obtained all the echocardiographic images using a phased array transducer (5-12 megahertz). 2 Echocardiographic measurements for LV and RV functional, morphometric parameters, and estimates of pulmonary hemodynamics were obtained in both term and preterm infants with each echocardiographic study. All measures were acquired and analyzed using previously published image acquisition and data analysis protocols from our laboratories. 2,9 All measures were generated according to guidelines of the American Society of Echocardiography 10,11 and recent validated protocols in neonates. 12 Echocardiographic data have previously been published from the preterm cohorts, 1,2,4,13,14 but the measurements had not been previously assessed in relationship to daily milk consumption. We have also previously demonstrated excellent reproducibility in term and preterm infants with each of the measures described in this section. 2,4,[13][14][15]

LV Function
LV function was characterized by the quantification of two dimensional-speckle tracking derived deformation imaging.
A. Deformation: LV free wall longitudinal strain (%) and systolic early, and late diastolic strain rate (1/sec) were assessed with two-dimensional speckle tracking echocardiography using previously published image acquisition and data analysis protocols from our laboratories. 2,9 Longitudinal strain is a measure of the maximal shortening of myocardial longitudinal length during systole compared to the resting length in diastole. The speed at which the myocardial wall deforms and then returns to baseline is measured as the time derivate of strain in systole and diastole, referred to as systolic and early and late diastolic strain rate, respectively. A frame rate to heart rate ratio between 0.7 and 0.9 frames/sec per beats per minute was utilized to optimize myocardial speckle tracking and mechanical event timing. 9 Peak strain for each index was measured as end-systolic strain at the closure of the aortic valve. 16 Two observers, who were blinded to the maternal and infant clinical and cardio-respiratory conditions, analyzed deformation using vendor customized commercially available software (EchoPAC; General Electric Medical Systems, Waukesha, WI, USA, version 112).

LV Morphology
LV morphology was assessed LV mass index (LVMi) and relative wall thickness (RWT) and according to previously published guidelines. 11 The estimation of LVMi and RWT were derived from LV measurements obtained by two-dimensional guided M-mode echocardiography using the parasternal short-axis view at the level of the papillary muscles. End diastole was defined as the time of maximum LV dimension. We measured end diastole interventricular septal thickness (IVSd), left ventricular (LV) posterior wall thickness (LVPWd), and LV dimension at end diastole (LVEDD) and end systole (LVESD) over three consecutive cardiac cycles and averaged the values.
A. LVM was estimated by the Devereux equation, LVM (grams) = 0.8{1.04 [(LVEDD + LVPWd + IVSd) 3 -(LVEDD) 3 ]} + 0.6. 17 We indexed LVM to allow comparisons between term and preterm infants with different body sizes. In this study, LVM was normalized to height to the allometric power of 2.7 to obtain LVMi. 18 Although whether to use height, weight, or BSA as the indexing term is unclear in children, it has been previously shown that LV mass centile curves indexed to weight, length, and BSA are all practical method to assess LV morphology in preterm infants.
Accordingly, we included these analyses to assess LV morophology. 19 We used the Haycock formula for calculation of body surface area: weight 0.5378 × height 0.3964 × 0.024265. 20 Since altered LV geometry in preterm infants is also associated with steroid use, 15 we also accounted for its postnatal use in the model. B. Relative wall-thickness (RWT) was calculated using two formulas: (1) RWT equals twice the posterior wall thickness (LVPWd) over LVEDD; (2) RWT equals the ratio of the sum of LVPWd and IVSd over LVEDD. 21 In the result section, we present RWT determined with twice the LVPWd over LVEDD rather than RWT equals the ratio of the sum of LVPWd plus IVSd due to the potential interference by tricuspid valve tissue and right ventricular trabeculations with septal measurement. 15

RV Function
The major contribution to ejection fraction and stroke volume for the RV during systole is provided by the dominant deep longitudinal fiber layer, which makes up 80% of the RV free wall thickness. In this study, RV performance was assessed using measures that assess changes in RV longitudinal shortening from one of three distinct approaches in neonates: 12  C. The same deformation approach utilized from the LV was applied to the RV, but assessment of RV longitudinal strain and strain rate images were acquired from an RV focused apical four chamber view according to previous published protocols. 2,13

RV Morphology
RV morphology was assessed with RV areas 1,23 and RV linear dimensions.

RV Afterload/Pulmonary Hemodynamics
We measured several different markers of RV afterload to account for its two main components, pulmonary vascular resistance and compliance. 24 Specifically, we measured pulmonary artery pressure derived by both the tricuspid regurgitation jet velocity, when available, and RV systolic time intervals.
A. Pulmonary artery systolic pressure was estimated by the tricuspid regurgitation jet velocity, when available B. Pulmonary artery systolic pressure was estimated by RV systolic time intervals, also known as pulmonary artery acceleration time (PAAT) and its ratio to right ventricle ejection time (RVET) to PAAT (PAATi). 14 PAAT and PAATi have been validated against cardiac catheterization in children and neonates, 24 and maturational patterns have recently been established in healthy children 25 and preterm infants. 14 Specific methods for acquisition are described elsewhere. 14 © 2021 El-Khuffash A et al. JAMA Network Open.

RV Function to RV Afterload Coupling
The RV and pulmonary arterial (PA) circulation function as one unit, commonly referred to as the RV-PA axis. 26 Recent evidence suggests an index of ventriculo-arterial coupling can serve as a comprehensive measure of RV-PA axis with different neonatal outcomes. 27 Accordingly, we assessed the relationship of TAPSE (a measure of RV longitudinal shortening) to PAATi (a comprehensive measure of RV afterload) as a surrogate of RV-PA coupling. 28,29

Common Neonatal Morbidities
The following additional common neonatal outcomes were evaluated: intraventricular hemorrhage (classified according to Papile Classification), 30

Sample Size Justification: Power Analysis
To overcome the limitations of enrolling enough patients at specific gestational ages with certain weights to detect statistically significant differences, our initial calculations on the basis of preliminary analysis and literature searches determined that we needed 150 patients to detect a 20% difference in LV systolic and RV systolic strain to expect moderate associations between MoM and strain (r = 0.3). Our design maintains 90% power for this effect size with n = 150 and 95% power with n = 170. The larger sample size allows power to detect variability in rates of change in our trajectory models, as well as 90% power for group differences of d = 0.43 and r = 0.20. The sample size of 170 assumes a missing rate of 10% (5% because of late deaths and 5% because of loss to follow-up) that would allow for a power of >80% to detect a significant association at an odds ratio of ≥1.30 for each 1-SD increase in RV and LV strain associated with gestational age. This calculation assumed no measurement errors and no correlation between strain and other covariates and a conservative 50% rate of BPD and 20% PH. In addition to account for the potential limitation of enrolling enough patients at specific gestational ages with certain weights to detect statistically significant differences, we decided to also generate maturational patterns and compare them between preterm-born infants in the high MoM, preterm-born infants in the low MoM, and term-born control infants at one year of age.

Characteristics of Cohorts
One hundred and eighty infants (80 born preterm and 100 born at term) were included in this study was 39.0 weeks (IQR 38.0 -40.0) and the median birth weight was 3165 g (IQR 2900-3760). Of the 100 term infants, 54 were white and 46 were black.

Clinical Data Reporting
The clinical information that was not described in Table 1 is detailed below in this supplement explanation. In this study, none of the preterm infants received inotropes, inhaled nitric oxide or other pulmonary vasodilators at any time point in the first year of age. Additionally, none of the preterm born infants were on respiratory support at 1 year CA. In our neonatal intensive care unit, all preterm infants born less than 29 weeks gestation received caffeine from birth through 34 weeks PMA, and none received it beyond 34 weeks PMA. Twenty-six preterm infants (33%) received postnatal steroids at least one time during their hospital course, of which 23 (88%) received them to treat lung disease beyond the first month of age. There was no significant difference in postnatal steroid administration based on MoM exposure (p=.13). Of the 23 infants that received postnatal steroids to treat lung disease beyond the first month of age, all were diagnosed with BPD. Of this subset, there was also no significant difference in postnatal steroid administration base on MoM exposure( p=.45). The three (12%) infants who did not develop BPD received steroids in the first month of age to facilitate a trial of extubation. At least one dose of diuretics was administered to 15 (18%) of the preterm infants during their neonatal course. All 15 were on a diuretic at 32 weeks PMA, but off by 36 weeks PMA and remained off at 1 year CA. There was no statistical difference in any of the cardiac indices between those infants that did and did not receive diuretics (p=.43), or postnatal steroids (p=.49), although the study was not adequately powered to answer these questions.

Dose Response: High vs. Low MoM Cut-off
Of 80 preterm-born infants, 42 (53%) were exposed to > 28 days of MoM (high MoM) and 38 (47%) were exposed to < 28 days of MoM (low MoM). eFigure 4 and eFigure 5 dichotomize the group by high vs. low MoM exposure at 1 year CA and allow for comparison of the primary cardiopulmonary outcomes to term infants.

Maturational Patterns of LV performance
LV function, as characterized by the magnitude of LV longitudinal strain, increased from 32 weeks to 1 year of age in both the preterm and term infants. of LV function were similar (p>.05) and LV morphology different (p<.05) at 32 weeks PMA with both exposure groups (eTable 2).

Maturational Patterns of RV Performance
The magnitudes of RV longitudinal strain, FAC, and TAPSE increased from 32 weeks PMA to 1 year CA in both the preterm and term infants (eFigure 6). RV morphology measures of systolic and diastolic areas, basal, mid-cavity, and apex-to-base linear dimensions also increased in both the preterm and term infants from 32 weeks postmenstrual age to 1 year CA (p<.01 for all measures). Compared to term infants, all preterm infants at 32 weeks PMA had decreased magnitudes of RV function and morphometric measures (P<.05 for all measures) (eTable 2).

Maturational Patterns of RV Afterload and Coupling
PAATi decreased from 32 weeks PMA to 1 year of CA in both the preterm and term infants (P<.05)

Additional Limitations
We did not explore factors that may influence a mother's ability to provide breast milk to her preterm infant during the neonatal period or at discharge. 31 Although preterm infants who received MoM at discharge in this study had enhanced cardiopulmonary function compared to infants who received formula at discharge, future work is needed to explore factors affecting breastfeeding outcomes in preterm infants and why a preterm infant would have higher exposure to MoM, as these factors themselves may also be linked to outcome measures.