Impact of breast milk composition on nutritional status of term neonates with congenital heart defects: A prospective study

Abstract

BACKGROUND

Congenital heart defects (CHDs) are the most common congenital malformations, often requiring surgical intervention during the early neonatal period. Although most children with CHDs are born at full term with appropriate birth weights, many of them develop malnutrition during the first weeks of life, resulting in complications after cardiac surgery. This study aimed to provide new insights into personalized nutritional interventions based on variations in breast milk (BM) composition. We hypothesized that the nutritional and energy composition of BM is associated with the growth trajectory during the late neonatal period in neonates with CHDs, and it may guide individualized nutrition support.

AIM

To assess the BM composition in mothers of neonates with CHD and evaluate its role in the nutritional status of these neonates.

METHODS

In this single-center prospective non-interventional study, we analyzed the BM composition in 35 mothers of neonates with CHDs (25 operated, 10 non-operated) at V1 (2-5 days), V2 (7-12), and V3 (14-40), and also in 21 mothers of healthy infants at V1 (n = 21) and V2 (n = 12), using the Miris Human Milk Analyzer. The weight-for-age z score (WAZ) at discharge was assessed in infants with CHD. Associations between the BM composition, nutrient intake, postoperative course, and WAZ change were evaluated.

RESULTS

Only the colostrum fat concentration was significantly lower in mothers of operated neonates [2.4 (2.0-2.6) g/100 mL] compared with mothers of non-operated neonates [2.9 (2.2-3.1) g/100 mL] and controls [2.7 (2.3-3.3) g/100 mL] (P = 0.045). Operated neonates exhibited a greater decrease in WAZ from birth to discharge than non-operated neonates (P = 0.008). Postoperative WAZ decline was associated with the postoperative period severity and delayed reintroduction of BM feeding. Higher calorie and fat intake on postoperative day 14 was significantly associated with a smaller WAZ decrease (P < 0.05). Later postoperative achievement of full enteral feeding independently predicted WAZ decline. In the operated group the colostrum protein content positively correlated with weight gain during the early neonatal period.

CONCLUSION

Colostrum from mothers of operated neonates with CHD had reduced fat content. BM composition did not affect WAZ trajectories. However, delayed achievement of full enteral feeding independently predicted WAZ decline.

Key Words: Breast milk composition; Congenital heart defects; Term neonates; Nutrition; Weight-for-age z score trajectories; Enteral feeding

Core Tip: This single-center prospective non-interventional study evaluated breast milk composition in mothers of neonates with congenital heart defects and its relationship with neonatal nutritional status. The colostrum fat concentration was significantly lower in mothers of operated neonates than in mothers of non-operated neonates. Operated neonates showed a decline in weight-for-age z score (WAZ) from preoperative values to discharge. The delayed achievement of full enteral feeding postoperatively independently predicted WAZ decline while higher calorie and fat intake on postoperative day 14 was associated with less WAZ reduction. In operated neonates higher colostrum protein content correlated with early weight gain.

INTRODUCTION

Congenital heart defects (CHDs) are the most common congenital developmental anomalies with the majority of cases requiring surgery during the early neonatal period[1]. Physical development during infancy is essential with its disturbance affecting other aspects of child development, including their cognition and social behavior later in life. Although most children with CHDs are born with a weight appropriate for gestational age[2], CHDs may affect fetal growth[3,4]. Many infants suffering from CHDs develop malnutrition during the first months of their lives[5,6]. Malnutrition is more frequent in patients with cyanotic congenital heart disease, especially in those with pulmonary hypertension[5,6]. Among the causes of postnatal malnutrition are increased metabolic demand[7], critical condition, hemodynamic features[8], gastrointestinal complications[9], as well as feeding practice variations among different centers. Conversely, malnutrition and growth deficiencies lead to complicated cardiac surgery and postoperative complications[10] in addition to having lifelong consequences. In infants with single-ventricle physiology, a low height trajectory was associated with worse neurodevelopmental outcomes[11]. To achieve appropriate nutritional results, both preoperative and postoperative nutrition protocols have been developed by cardiac surgical centers[12-16].

Human breast milk (BM) is the preferred source of nourishment for infants. In a cohort of infants with CHDs, an unfortified human milk diet showed a protective effect against gastrointestinal complications and necrotizing enterocolitis (NEC)[17,18]. There is also data on the benefits of human milk on neurocognition in neonates with CHDs[18].

Due to high energy expenditure and the need for fluid intake restriction in these infants, a few studies on high-calorie formulas and fortified BM were undertaken with discordant results[19-23]. In the study by Lin et al[23], the weight, head circumference, height, blood albumin level, and prealbumin level of the preoperative human milk fortifier (HMF) group were significantly higher after the first month post-intervention compared with the exclusive breastfeeding group. Sahu et al[22] demonstrated trends of reduced length of mechanical ventilation, duration of stay in the intensive care unit, and hospital infection rate and mortality rate in the fortified BM group postoperatively from day 1 to day 10. Nutrition fortification may have long-term effects on a child’s growth and nutritional status. In the study by Murray et al[24], preoperative nutrition fortification among infants with CHDs was associated with more rapid growth in the first 30 postoperative days and lower body mass index percentiles for age at 10 years with a trend toward a higher prevalence of overweight or obesity in the unfortified group[24].

We did not find any publications providing data on BM composition in mothers of infants with CHD although the issue has been widely discussed in the population of preterm neonates[25,26]. Such data can help characterize factors influencing BM composition and provide a rationale for personalized nutritional interventions in this category. Thus, it is necessary to search for possible intervention targets capable of improving nutritional status in CHD infants while assessing the risks and benefits of fortified feeds at different levels[27]. This study aimed to assess BM composition in mothers of neonates with CHDs and evaluate the role of BM in the nutritional status of these neonates.

MATERIALS AND METHODS

Study design, participants, and timepoints

From May 2024 to March 2025, a single-center prospective cohort non-interventional study was conducted at the Perinatal Center of the Almazov National Medical Research Center. This study was approved by the ethics committee of Almazov National Research Center (ID: 1505-25) and conducted in accordance with the principles outlined in the Declaration of Helsinki. Written informed consent was obtained from all mothers who participated in the study.

Eighty-four neonates with CHDs received treatment at the Almazov National Medical Research Center during the study period (Figure 1A). In this study we included full-term neonates with CHDs who received partial or exclusive BM feeding during the early neonatal period. Our non-inclusion criteria were maternal decline to participate, concurrent congenital anomalies of the gastrointestinal tract or central nervous system, and severe intranatal asphyxia. Our exclusion criteria for all neonates were hospital length of stay less than 10 days, chromosomal abnormalities, and the mother’s refusal to continue participating in the study. Our exclusion criteria for neonates who required cardiac surgery were cardiac surgery after 15 days of life, early postoperative mortality following surgical correction of CHDs, prolonged stay in the neonatal intensive care unit (NICU). Following the application of all eligibility criteria, a total of 35 neonates with CHDs were enrolled (Figure 1A). Twenty-five of them required cardiac surgery during the neonatal period (hereinafter referred to as the “operated group”). Ten neonates with CHDs did not require surgery but underwent prolonged hospitalization (hereinafter referred to as the “non-operated group”).

Figure 1 Study design. A: Group formation flow chart; B: Timepoints of observation and sample collection. Operation day of life is presented as median (interquartile range). BM: Breast milk; NICU: Neonatal intensive care unit; DOL: Day of life; POD: Postoperative day; V1: Colostrum; V2: Transitional milk; V3: Mature milk.

Gestational age, maternal age, and pregnancy-related complications were recorded for all study participants. Neonatal data analyzed included the timing of BM feeding initiation, dietary composition, enteral feeding volume, BM feeding volume, enteral and total nutrient intake, body weight, and the weight-for-age z score (WAZ) at defined timepoints (Figure 1B). Associations between clinical variables, the nutrient composition of maternal BM, and the trajectory of WAZ until discharge were analyzed.

In the operated group associations were also evaluated between intraoperative parameters (procedure duration and the cardiopulmonary bypass interval), the inotropic support interval (days), the postoperative assisted ventilation interval (days), postoperative comorbidities, the time of BM feeding reintroduction after surgery, enteral feeding volume, BM feeding volume, enteral and total nutrient intake, the nutrient composition of maternal BM, and the postoperative WAZ trajectory.

Due to the different courses of the late prenatal period and the different durations of hospitalization, correlations between the BM nutrient content and weight dynamics were assessed within the groups.

BM samples were collected from mothers at three physiologically relevant lactation stages: V1 (colostrum), V2 (transitional milk), and V3 (mature milk) (Figure 1B). The timing of neonatal clinical assessments broadly corresponded to these lactation stages, enabling the evaluation of associations between BM composition and early postnatal growth trajectories. For BM composition analyses, BM samples from mothers of healthy neonates (control group) were collected at two timepoints: V1 (n = 21) and V2 (n = 12).

Clinical variables

All neonates underwent routine postnatal clinical assessment and echocardiographic examination after birth performed by a pediatric cardiologist to confirm the presence of CHDs. Critical CHDs were defined as structural heart malformations requiring surgical or catheter-based intervention within the first month of life or those constituting ductus arteriosus-dependent pulmonary or systemic circulation.

The duration of antibiotic therapy (days) was recorded for the preoperative and postoperative periods. The stage of NEC was defined by the Bell staging criteria modified by Walsh and Kliegman[28]. NEC was managed per the clinical guidelines approved by the Ministry of Health of the Russian Federation[29]. Pneumonia in neonates was diagnosed based on a combination of clinical, laboratory, and radiological criteria in accordance with the clinical guidelines approved by the Russian Society of Neonatologists and the Russian Association of Perinatal Medicine Specialists[30].

Outcomes

The primary outcome of the study was the change in WAZ from birth to discharge in all neonates with CHDs and from the preoperative period to discharge in operated neonates with CHDs along with the analysis of associations between BM macronutrient composition and neonatal ΔWAZ. Secondary outcomes included the incidence of hypotrophy at discharge (defined as WAZ < -2). Additional outcomes included associations between postoperative nutrient intake, postoperative period parameters, and ΔWAZ in operated neonates with CHDs during the postoperative period.

BM composition analysis

The macronutrient composition and energy content of BM were analyzed by mid-infrared spectroscopy using the Miris Human Milk Analyzer. Before these analyses samples were warmed at 40 °C in a Miris Heater and homogenized with a Miris Ultrasonic Processor. The contents of protein (g/100 mL), fat (g/100 mL), carbohydrates (g/100 mL), energy (kcal/100 mL), and solids (g/100 mL) were measured. Calibration and quality control were performed as per the manufacturer’s guidelines.

WAZ calculation

WAZ were calculated using the World Health Organization Child Growth Standards in neonates with CHDs longitudinally at different timepoints. WAZ was calculated using the LMS method according to the formula: Z = [((x/M)^L) – 1]/(S × L), where x is the infant’s weight in kilograms, M is the median weight for the infant’s age and sex, L is the Box-Cox transformation power (to account for skewness), and S is the generalized coefficient of variation. The L, M, and S parameters were obtained from the World Health Organization Child Growth Standards[31]. Postnatal hypotrophy at discharge was defined as WAZ < –2.

Evaluation of enteral feeding volume and energy and nutrient intake in neonates

Each neonate’s intake of protein (g/kg/day), fat (g/kg/day), carbohydrates (g/kg/day), energy (kcal/kg/day), and enteral (total and BM) feeding volume (mL/kg/day) was calculated. Energy consumption was calculated as fat (g/kg/day) × 10 + protein (g/kg/day) × 4 + carbohydrates(g/kg/day) × 4.

Enteral nutrient intake was calculated based on the measured composition of the BM, analyzed using the Miris Human Milk Analyzer, and the daily volume of BM received (mL/kg/day). In the case of direct breastfeeding, the volume of BM intake was estimated by weighing the neonate immediately before and after feeding with the difference in weight (grams) assumed to correspond to the volume of milk consumed (mL), based on the assumption that 1 g ≈ 1 mL of milk.

In neonates receiving mixed feeding (BM and formula), nutrient intake from each source was calculated separately and then summed to determine total enteral intake. Formula-based nutrient intake was calculated using manufacturer-provided nutritional composition data and daily formula intake volumes.

Target nutrient and energy requirements were guided by Russian clinical guidelines for neonatal parenteral nutrition[32]. These guidelines recommend daily intakes of amino acids (AA) (2.0-2.5 g/kg), lipids (3 g/kg), and glucose (up to 12 g/kg) for neonates with birth weight > 2000 g, aiming to achieve a total energy intake of approximately 105–120 kcal/kg/day.

Statistical analysis

Statistical analyses were performed using Python 3.12.5. Before continuous variables were compared between two groups, the normality of their data distribution within each group was assessed using the Shapiro-Wilk test. If both groups demonstrated a normal data distribution (P > 0.05), comparisons were made using the independent samples t test. If the data distribution was not normal (P ≤ 0.05), the non-parametric Mann-Whitney U test was used. Continuous variables were presented using medians with interquartile ranges (Q1–Q3). Categorical variables were presented using frequencies with percentages and compared using Fisher’s exact or the χ² test according to their sample sizes.

Longitudinal analyses of the total nutrient, energy intake and enteral feeding volume was performed using a linear mixed-effects model. The model was fitted using the “statsmodels” Python package.

To identify factors associated with the change in WAZ from the preoperative period to discharge, a multiple linear regression analysis was performed in neonates of the operated group. The dependent variable was the change in WAZ from the preoperative period to discharge (was calculated according to the formula: WAZ change = WAZ at discharge – WAZ before surgery). Independent variables included continuous nutritional and clinical predictors as well as categorical variables. Categorical variables were one-hot encoded, and continuous variables were normalized using min-max scaling to standardize their range. At first univariable models were fitted. Variables with statistically significant associations in the univariable analysis were included in a multivariable model. An ordinary least squares regression model was fitted using the “statsmodels” package in Python.

Spearman’s correlation coefficients were used to analyze the correlations between BM composition, duration of BM feeding, and growth characteristics in neonates with CHDs.

To explore the patterns in mature BM composition in relation to the presence of postnatal hypotrophy at discharge, the principal component analysis was performed. This analysis included the following nutritional components: Fat; total protein; and energy (kcal). All variables were standardized using z score normalization. The first two principal components were visualized in a scatterplot.

RESULTS

Clinical characteristics of groups

There were no statistically significant differences between the operated and non-operated groups in key perinatal characteristics, including birth weight, WAZ at birth, gestational age, intrauterine growth restriction, maternal age, mode of delivery, and perinatal complications such as gestational diabetes and preeclampsia (Table 1).

Table 1 Clinical characteristics of term neonates with congenital heart defects.

	Operated (n = 25)	Non-operated (n = 10)	P value
Birth weight (g)	3340.0 (3130.0-3620.0)	3640.0 (3227.5-3895.0)	0.201
WAZ at birth	0.07 (-0.28-0.55)	0.72 (-0.25-1.36)	0.228
IUGR	1 (4.0)	1 (10.0)	0.524
Gestational age, weeks	39.6 (39.0-40.0)	40.00 (39.1-40.4)	0.687
Male neonates, female neonates	13, 12	3, 7	0.285
Age of mother (years)	29.0 (25.0-33.0)	31.00 (23.0-36.5)	0.609
Mothers from North Caucasian ethnic groups	14 (56.0)	4 (40.0)	0.471
Cesarean section	6 (24.0)	1 (10.0)	0.644
Gestational diabetes	1 (4.0)	3 (30.0)	0.061
Preeclampsia	2 (8.0)	2 (20.0)	0.561
Age at surgery, DOL	6.0 (5.0-8.5)
Aristotle score	7.0 (6.0-10.0)
Palliative surgery	7 (28.0)
Operation duration (minute)	240.0 (185.0-305)
Cardiopulmonary bypass	18 (72.0)
Cardiopulmonary bypass duration, min	115.0 (0.0-180.5)
Max. VIS 0-24 h after surgery	15.0 (5.0-21.0)
Max. VIS 24-48 h after surgery	8.0 (5.0-16.5)
Length of inotropic therapy, days	2.0 (1.0-4.5)
Length of assisted ventilation, days	2.0 (1.0-4.0)
Extracorporeal membrane oxygenation
Days in NICU after surgery	5.0 (4.0-9.0)
Discharge from NICU, DOL	13.0 (10.0-15.0)
Start of enteral feeding after surgery (POD)	1.0 (1.0-1.0)
Pneumonia (during hospital stay), N	10 (40.0)	3 (30.0)	0.709
NEC (during hospital stay)	6 (24.0)	1 (10.0)	0.644
Full enteral feeding		9.00 (3.2-14.8)	0.002
DOL	22.0 (16.0-32.0)
POD	8.5 (15.0-26.0)
RBC transfusions, from birth to discharge (except intraoperative)	10 (40.0)	0 (0.0)	0.034
Transfer to mother’s ward, DOL	28.0 (20.0-34.0)	10.0 (5.3-15.7)	0.003
Start of BM feeding postoperatively
DOL	16.0 (13.0-28.0)
POD	11.00 (5.0-19.0)
Discharge, DOL	33.5 (27.0-37.5)	20.50 (17.2-27.0)	0.015
Discharge, POD	26.00 (16.0-31.0)

Data are presented as n (%)/median (25^th-75^th percentiles). WAZ: Weight-for-age z score; IUGR: Intrauterine growth retardation; VIS: Vasoactive inotropic score; NICU: Neonatal intensive care unit; NEC: Necrotizing enterocolitis; RBC: Red blood cells; DOL: Day of life; POD: Postoperative day.

Operated neonates underwent cardiac surgery at a median age of 6.0 days (interquartile range: 5.0-8.5 days). Operated neonates had significantly higher frequency of red blood cell transfusions [10 (40.0%) vs 0, P = 0.034], longer hospital stay [33.5 (27.0-37.5) vs 20.50 (17.2-27.0), P = 0.015]. The prevalence of comorbidities such as pneumonia (40.0% vs 30.0%; P = 0.709) and NEC (24.0% vs 10.0%; P = 0.644) did not differ significantly between the two groups.

BM feeding initiation and types of diet

BM feeding was initiated later in the operated group than in the non-operated group: On day of life (DOL) 1.0 (1.0-2.0) vs DOL 1.0 (1.0-1.0), respectively (P = 0.066). In the early neonatal period (≤ DOL 7) during the initiation and consolidation of lactation in mothers, 15 (60.0%) neonates of the operated group and 7 (70.0%) neonates of the non-operated group received supplemental formula (P = 0.709) (Figure 2). In cases of mixed feeding (BM combined with formula), the choice of formula (preterm, term, or hydrolyzed) was based on the clinical scenario. Based on the χ² test, the distribution of diet was statistically different in non-operated group on DOL 3 (P = 0.010) and DOL 7 (P = 0.021) compared with the operated group.

Figure 2 Types of enteral feeding substrates in operated and non-operated term neonates with congenital heart defects during hospitalization. Mixed feeding combines breast milk and formula. DOL: Day of life; POD: Postoperative day.

In the early postoperative phase, operated neonates were initiated on hydrolyzed formula as per the institutional protocol. Feeding with BM was reintroduced on postoperative day (POD) 11.0 (5.0-19.0). The reasons for the later postoperative initiation of BM feeding were NEC during the early postoperative period (n = 4), severe hemodynamic instability with the need for prolonged NICU stay (n = 2), and chylothorax (n = 1). NEC stage IIA developed in 4 (16.0%) neonates in the early postoperative period (POD 1-5). Before NEC onset 3 patients had not been fed, and 1 patient had been fed with hydrolyzed formula.

By POD 14, 7 (28.0%) operated neonates had transitional symptoms of feeding intolerance: Regurgitation (n = 3); watery stools and bloating (n = 3); and hematochezia (n = 1). Therefore, they needed various approaches for enteral feeding. Untargeted BM fortification with a multinutrient HMF was performed in eight neonates (six and two neonates from the operated and non-operated groups, respectively) in a standard dosage as part of routine clinical care at the Almazov National Medical Research Center and was not influenced by the research objectives. One operated neonate demonstrated feeding intolerance to fortified BM (flatulence, diarrhea), which led to the termination of fortification, culminating in symptom reduction. Two (8.0%) operated neonates were exclusively formula-fed at discharge due to lactation failure (were fed with term formula).

BM composition

The colostrum fat content differed significantly among mothers of the operated neonates, the non-operated neonates, and healthy neonates (P = 0.045) (Figure 3A, Supplementary Table 1). The highest median fat concentration was observed in colostrum from mothers of non-operated neonates [2.9 (2.2-3.1) g/100 mL], followed by colostrum in mothers of the control group [2.7 (2.3-3.3) g/100 mL], and then colostrum in mothers of the operated group [2.4 (2.0-2.6) g/100 mL]. No significant differences in total protein, carbohydrates, energy (kcal), or true protein contents were found among the groups (Figure 3B-E).

Figure 3 Colostrum, transitional milk, and mature milk composition in mothers of operated and non-operated term neonates with congenital heart defects and mothers of healthy neonates. A: Fat concentrations; B: Total protein concentrations; C: True protein concentrations; D: Carbohydrate concentrations; E: Total kilocalories (kcal). P values are presented from Kruskal-Wallis tests. V1: Colostrum; V2: Transitional milk; V3: Mature milk.

In transitional milk values of fat, carbohydrates, and kilocalories were generally higher compared with those of colostrum (Figure 3). The caloric content of transitional milk was lowest in the operated group, followed by the control group, and then the non-operated group; however, the difference among the groups was not statistically significant (P = 0.172).

Regarding mature milk, only the operated and non-operated groups were compared. No statistically significant differences were observed for any nutrient between these two groups. The median nutrient levels remained relatively stable compared with transitional milk (Figure 3). Calorie content below 67 kcal/100 mL in mature milk was observed in 7 (20.0%) mothers of neonates with CHD: 5 mothers of neonates in the operated group and 2 mothers of neonates in the non-operated group (P = 1.0). These low-calorie BM samples (n = 7) had significantly lower fat content (2.6 ± 0.6 g/100 mL vs 4.4 ± 1.3 g/100 mL, P < 0.001) but similar protein content (1.2 ± 0.2 g/100 mL vs 1.5 ± 0.4 g/100 mL, P = 0.084) compared with higher calorie BM samples (n = 28). Total protein lower than 1.2 g/100 mL in mature milk was observed in 7 (20.0%) mothers of neonates with CHD: 6 mothers of neonates in the operated group and 1 mother of a neonate in the non-operated group (P = 0.644).

Assessment of nutrient intake in neonates

Operated neonates had overall lower enteral feeding volume than non-operated neonates (Figure 4A, Supplementary Figure 1A). BM feeding volumes increased over the first week of life in both operated and non-operated infants; however, non-operated neonates achieved significantly higher volumes at DOL 3 and DOL 5 (P = 0.036 and P = 0.020, respectively) (Figure 4B, Supplementary Figure 1B).

Figure 4 Enteral feeding volume and total (enteral and parenteral) nutrition trajectories in term neonates with congenital heart defects, comparing operated and non-operated groups during hospitalization. A: Enteral feeding volumes; B: Breast milk feeding volumes; C: Total fat intake; D: Total protein intake; E: Total carbohydrate intake; F: Total kilocalorie intake. DOL: Day of life.

Operated neonates did not show a significant difference in total fat, total protein, total carbohydrates, and total energy (kcal) intake before surgery [operation DOL 6 (interquartile range 5-8)] compared with the non-operated group (Figure 4C-F, Supplementary Figure 1C-F). In contrast, operated neonates had lower fat, protein, and energy intake in the early postoperative period while overall carbohydrate intake was similar to the non-operated group (Figure 4C-F).

Additionally, we evaluated the impact of enteral feeding volume (mL/kg/day), breastfeeding volume, and perioperative changes in nutrient and caloric intake on hypotrophy at discharge in operated neonates. Both subgroups (with and without hypotrophy at discharge) showed gradual increases in enteral intake over time. No statistically significant differences in enteral feeding volume, BM intake, or macronutrient and energy intake were observed between the neonates with and without hypotrophy at discharge on DOL 1 and DOL 3 (Figure 5). However, by DOL 5 neonates who did not develop hypotrophy had significantly higher protein and kilocalorie intakes.

Figure 5 Enteral feeding volume and total (enteral and parenteral) nutrition profiles in operated neonates with congenital heart defect, compared by hypotrophy status at discharge. A: Enteral feeding volumes; B: Breast milk feeding volumes; C: Total fat intake; D: Total protein intake; E: Total carbohydrate intake; F: Total kilocalorie intake. DOL: Day of life; POD: Postoperative day.

In the early postoperative period, neonates with hypotrophy had significantly lower enteral feeding volumes on POD 5 (P = 0.004), on POD 10 (P = 0.009), and on POD 14 (P = 0.018) and lower protein intake on POD 5 (P = 0.031) (Figure 5A and D). Moreover, neonates with hypotrophy had a higher incidence of enteral feeding intolerance in the early postoperative period (4/8 (50.0%) vs 1/17 (5.9%), P = 0.024). In contrast, neonates without hypotrophy had significantly higher fat and energy intake on POD 5 (P < 0.001 and P = 0.004, respectively) and on POD 10 (P = 0.011 and P = 0.037, respectively) (Figure 5C and F). These neonates also received significantly more BM on POD 10 (P = 0.005) and POD 14 (P = 0.024) (Figure 5B). No significant difference in carbohydrate intake was observed (Figure 5E).

Assessment of neonatal WAZ and their associations with BM composition

At birth there were no significant differences in WAZ between operated and non-operated neonates with CHD (P = 0.228) (Figure 6A). At DOL 14 neonates of the operated group had lower WAZ compared with the non-operated group (P = 0.036). The mean weight gains during hospitalization (from birth to discharge) did not differ significantly between the operated [10.3 (5.8-13.6) g/day] and non-operated [11.2 (8.0-13.7) g/day] groups (P = 0.675). However, the decrease in WAZ was greater in operated neonates [-1.35 (-1.58 to -0.98) vs -0.98 (-1.11 to -0.85), P = 0.039].

Figure 6 Dynamics of the weight-for-age z score in term neonates with congenital heart defects during hospitalization and at discharge. A: Hospitalization; B: Discharge. DOL: Day of life; POD: Postoperative day; OP: Operated group; NOP: Non-operated group; WAZ: Weight-for-age z score.

At discharge operated neonates had a significantly lower WAZ than non-operated neonates [-1.22 (-2.06 to -0.81) vs -0.13 (-0.59 to 0.18), P = 0.004] (Figure 6B). At discharge the incidence of hypotrophy was 8 (32.0%) patients in the operated group and 1 (10.0%) patient in the non-operated group (P = 0.235).

Mothers of neonates with hypotrophy at discharge had no statistically significant differences in mature milk composition (Supplementary Figure 2). Operated neonates with CHD who developed hypotrophy by discharge had been characterized by significantly lower birth weight [3110.0 (2900.0-3290.0) g vs 3435.0 (3237.5-3695.0) g, P = 0.017] and birth WAZ [-0.47 (-1.0 to -0.1) vs 0.42 (-0.02 to 0.89), P = 0.009] compared with those without hypotrophy. No significant differences in gestational age were observed. Neonates with hypotrophy received significantly longer courses of postoperative antibiotic therapy [17.0 (15.0-28.0) days in total until discharge vs 6.0 (3.0-10.0) days, P < 0.001] and required a longer NICU stay [9.0 (6.0-17.0) days vs 4.5 (3.0-5.2) days, P = 0.010]. Neonates with hypotrophy underwent shunt placement more often (44.4% vs 0%, P = 0.010) and showed a higher incidence of postoperative pneumonia (66.7% vs 21.1%, P = 0.035). There were no differences in the duration of BM feeding during the preoperative [4.0 (4.0-7.0) days vs 5.0 (2.8-7.2) days, P = 0.909] and postoperative periods [12.0 (10.0-23.0) days vs 10.0 (8.8-18.5) days, P = 0.334] in neonates with hypotrophy compared to neonates without hypotrophy. BM fortification was employed in neonates with hypotrophy more often (44.4% vs 6.2%, P = 0.040).

During the early neonatal period, there were moderate positive correlations between weight gain (from birth to preoperative day) and the percentage of BM feeding days in the preoperative period (r = 0.76) among neonates of the operated group (Figure 7A). There were strong positive correlations between the colostrum protein content and early weight gain (from DOL 3 to preoperative day) in the operated group (r = 0.82). However, neonates of the non-operated group had moderate positive correlations between colostrum fat and energy content and daily weight gain from birth to DOL 7, DOL 3 to DOL 7, and ΔWAZ from birth to DOL 7 (Figure 7B).

Figure 7 Spearman correlation analysis of relationships between maternal breast milk macronutrient content and growth parameters in neonates with congenital heart defects. A and B: Correlations for colostrum are presented separately for operated and non-operated neonates; C and D: Mature breast milk are presented separately for operated and non-operated neonates. Color scale indicates Spearman's r; ^aP < 0.05, ^bP < 0.01.

During the late neonatal period, there were no statistically significant associations among the mature BM nutrient content, energy composition, and postoperative WAZ dynamics and the mean weight gain in the operated group (Figure 7C). However, there was a moderate negative correlation between the time of the postoperative reintroduction of BM feeding (r = −0.45) and ΔWAZ from the preoperative value to discharge (therefore, later BM reintroduction was associated with greater WAZ decline; Figure 7C). Non-operated neonates had moderate negative correlations between the total (r = -0.46) and true protein content (r = -0.42) and mean weight gain in BM feeding, but they were not significant (Figure 7D).

In the operated group associations between the BM nutrient content of mature milk and the change in WAZ from the preoperative period to discharge (ΔWAZ) were also analyzed using univariable linear regression analyses (Table 2). No significant associations were identified between the content of fat, total protein, true protein, carbohydrates, and kilocalories in mature BM and ΔWAZ. However, the later subsequent postoperative reintroduction of BM feeding (P = 0.001) and later age at exclusive BM feeding (P = 0.001) were significantly associated with greater WAZ decline. Higher total (enteral and parenteral) calorie and fat intake on POD 14, greater enteral feeding volume on POD 14 contributed to less decrease in WAZ (P < 0.05).

Table 2 Associations of variables with delta of weight-for-age z score from the preoperative period to discharge in operated neonates with congenital heart defects.

	Univariable regression
	Coefficient	Standard error	T	P >|t|
General characteristics and postoperative morbidities
WAZ at birth	-0.36	0.68	-0.53	0.599
Gestational age, (weeks)	-0.69	0.65	-1.06	0.299
Length ABT preoperatively (days)	-0.72	0.44	-1.65	0.113
Length ABT postoperatively (days)	-2.26	0.46	-4.94	0.001
Full EF, POD	-2.32	0.41	-5.60	0.001
Pneumonia postoperatively1	-0.67	0.31	-2.17	0.041
NEC postoperatively1	-0.15	0.39	-0.39	0.699
Regurgitation postoperatively1	-0.81	0.48	-1.68	0.106
Day of surgery	-0.41	0.64	-0.64	0.528
Days in NICU	-1.98	0.62	-3.20	0.004
Transfer to mother’s ward, POD	-1.97	0.52	-3.82	0.001
Days of hospitalization postoperatively	-2.26	0.44	-5.10	0.001
Total intake on POD 14
Fat (g/kg/day)	1.58	0.71	2.23	0.036
Protein (g/kg/day)	-0.77	0.60	-1.28	0.214
Carbohydrates (g/kg/day)	-0.79	0.47	-1.69	0.106
Energy (kcal/kg/day)	1.61	0.78	2.08	0.049
EF volume (mL/kg/day)	1.05	0.46	2.28	0.032
Total intake before surgery
Fat (g/kg/day)	0.31	0.75	0.42	0.679
Protein (g/kg/day)	0.05	0.66	0.07	0.945
Carbohydrates (g/kg/day)	0.08	0.88	0.09	0.927
Energy (kcal/kg/day)	0.17	0.65	0.26	0.797
EF volume DOL 7 (mL/kg/day)	1.03	0.53	1.93	0.066
Perioperative characteristics
Aristotle score	-0.69	0.55	-1.27	0.217
Bypass length (minutes)	-0.65	0.48	-1.35	0.190
Operation length (minutes)	-0.47	0.61	-0.78	0.442
VIS 0-24 h postoperatively, max	-0.74	0.57	-1.31	0.203
VIS 24-48 h postoperatively, max	-1.72	0.72	-2.40	0.025
Inotropes length (days)	-1.76	0.68	-2.57	0.017
Ventilation length (days)	-2.05	0.62	-3.32	0.003
Shunt (BTS/Norwood/DKS)1	-1.33	0.36	-3.74	0.001
Bypass1	-0.73	0.36	-2.05	0.051
Delayed sternal closure1	-0.61	0.35	-1.73	0.097
Feeding intolerance, early postoperatively	-0.87	0.34	-2.52	0.019
RBC transfusions1	-0.57	0.32	-1.82	0.082
Mature mother’s milk composition
Fat (g/100 mL)	-0.69	0.78	-0.90	0.378
Total protein (g/100 mL)	0.80	0.78	1.02	0.318
True protein (g/100 mL)	0.93	0.79	1.17	0.254
Carbohydrates (g/100 mL)	-0.17	0.78	-0.23	0.824
Energy (kcal/100 mL)	-0.60	0.76	-0.79	0.436
Characteristics of feeding with mother’s milk
HMF/preterm formula postoperatively1	-0.82	0.33	-2.51	0.020
Mean weight gain in BM feeding postoperatively (g/day)	1.25	0.73	1.71	0.101
Start of BM feeding, POD	-1.97	0.50	-3.98	0.001
BM feeding only, POD	-2.21	0.45	-4.91	0.001
Days of BM preoperatively	-0.82	0.68	-1.22	0.236
Days BM postoperatively	-1.26	0.57	-2.21	0.037
BM feeding, % of POD	0.41	0.58	0.71	0.485
BM feeding, % of hospitalization	0.23	0.65	0.34	0.734
	Multivariable regression
Length ABT postoperatively (days)	-1.18	0.54	-2.17	0.041
Full EF, POD	-1.54	0.53	-2.94	0.008

¹Indicates the presence of the condition, coded as a binary variable. WAZ: Weight-for-age z score; Length ABT: Duration of antibacterial treatment; EF: Enteral feeding; DOL: Day of life; POD: Postoperative day; NEC: Necrotizing enterocolitis; NICU: Neonatal intensive care unit; VIS: Vasoactive inotropic score; BTS: Modified Blalock-Taussig shunt; DKS: Damus–Kaye–Stansel procedure; RBC: Red blood cells; HMF: Human milk fortification; BM: Breast milk.

In the univariable analysis several postoperative course features showed significant associations with greater WAZ decline from the preoperative period to discharge: Longer postoperative antibiotic therapy (P = 0.001); operation with shunt formation (modified Blalock-Taussig shunt or the Norwood procedure or the Damus–Kaye–Stansel procedure) (P = 0.001); pneumonia during the postoperative period (P = 0.041); longer NICU stay (P = 0.004); later achievement of full enteral feeding postoperatively (P = 0.001); later transfer to the mother’s ward (P = 0.001); and longer postoperative hospitalization (P = 0.001) (Table 2).

Factors with statistically significant associations in the univariable analysis were included in a multivariable model (Table 2) in which only longer postoperative antibiotic therapy (β = −1.18, P = 0.041) and later achievement of full enteral feeding after surgery remained significantly associated with greater WAZ decline (β = −1.54, P = 0.008), identifying them as a potential independent risk factors. Other variables lost their statistical significance after adjustment, suggesting the effects of confounders.

DISCUSSION

Infants with critical CHDs often suffer from malnutrition and growth retardation[5,6]. The neonatal period is critical for organ growth and development. Therefore, CHD-related malnutrition during this period may have long-term consequences. Palleri et al[33] demonstrated that moderate and severe CHDs negatively affected the growth of children even in a tertiary center in a high-income country where surgery is performed early in life and proper nutritional support is guaranteed[33]. Even basal metabolic requirements and target caloric intakes can hardly be met throughout the perioperative hospital stay[34]. Malnutrition is associated with adverse clinical outcomes in children and adults alike: Prolonged postoperative wound healing; myocardial dysfunction; vascular endothelial damage; decreased muscle function; and an increased risk of postoperative infections[6,7,10,35].

Human BM is the preferred substrate for enteral feeding in infants, and it has demonstrated a protective effect against gastrointestinal complications and NEC in infants with CHD[17,18]. In the study by Davis et al[36], human milk feeding of infants with CHD was not associated with a reduced risk of gastrointestinal symptoms, including NEC. A BM diet was associated with improved growth rates over time. Moreover, direct breastfeeding was associated with a reduced incidence of gastrointestinal distress and bloody stools without any subsequent risk of malnutrition[36].

We focused on the search for the rationale for possible nutritional interventions within the CHD cohort under routine care in a single-center setting in the context of BM feeding. The diversity of enteral feeding diets reflected the clinical complexity of nutritional management in neonates with CHD, particularly during the postoperative period. In most of the neonates in our cohort, BM feeding was maintained by discharge, either exclusively or in combination with formula. We did not access issues of lactation support specifically; however, we believe these results highlight the success of efforts made by both mothers and health workers to support breastfeeding under stressful conditions. This reflects the concern of implementing measures of maternal support, prenatal breastfeeding education, mothers’ access to a hospital-grade breast pump and facilities, and medical staff education in cardiac surgical units[37].

To optimize energy provision some feeding protocols propose an increase in the energy density of formulas in the postoperative period[27,38] and a few BM fortification[22]. In our cohort in the cases of insufficient BM, preterm formula was administered mostly in the operated group. BM fortification with multinutrient HMF was used in term infants with CHD on an individual basis. Although one neonate exhibited intolerance to fortified BM, overall clinical tolerance was favorable. The available data on BM fortification in preterm infants is comprehensive. A previous meta-analysis showed that individualized (either targeted or adjustable) fortification of enteral feeds in very low birth weight infants accelerates growth in terms of weight, length, and head circumference during the intervention compared with standard non-individualized fortification[39]. To the best of our knowledge, a targeted approach to BM fortification in term infants with CHD has not been studied yet.

The present study aimed to investigate the BM composition in mothers of neonates with CHD and to explore its potential influence on nutritional status at discharge. To our knowledge this is the first investigation to evaluate the macronutrient profile of BM in this population and its association with neonatal growth outcomes.

We found inter-individual variability in BM composition. One of the remarkable findings of our study was the lower fat content observed in the colostrum of mothers of neonates requiring surgery. First of all, the fat component of BM shows the highest inter-individual variation among the three major macronutrients (fats, proteins, and carbohydrates) with a coefficient of variation of 37.3%[40,41]. The fat content also varies significantly among BM samples produced by a particular female[42]. These facts may affect our data on the fat content of colostrum. Second, among maternal factors that influence the composition of BM, maternal stress caused by the infant’s critical condition should be considered. The effect of psychosocial stress on lactation is not adequately described. Stress negatively affects breastfeeding initiation, duration, and exclusivity; however, to the best of our knowledge, only a few studies have examined the impact of maternal stress on BM composition[43]. Levels of secretory immunoglobulin A, beta-endorphin, and cortisol in BM have been examined with controversial results[44-47]. Several studies have found no changes in the macronutrient composition of BM in association with maternal stress in the peripartum period[48,49]. Depressive symptoms early in pregnancy (below 20 weeks of gestation) were associated with lower postpartum BM docosahexaenoic acid concentrations[50]. No statistically significant associations were found between n-3/n-6 fatty acids in BM and antenatal anxiety and/or depression as well as postpartum depression[51].

As a hormone that regulates lipid metabolism, cortisol potentially affects the lipid content and general composition of human milk. Acutely, cortisol may contribute to the utilization of different lipid sources for the production of human milk lipids[52,53]. The use of antenatal corticosteroids may increase the total lipid content of human milk following premature birth[54]. Milk cortisol has been shown to significantly correlate with the mother’s plasma cortisol levels[55]. No correlation was found between milk glucocorticoids and the total fat content of milk[56,57]. However, the influence of cortisol could be observed when the composition of individual fatty acids in milk was investigated[57]. Conversely, our data demonstrated lower fat content in the colostrum of mothers of operated infants, highlighting the need for further investigation.

In both human and animal plasma, AA levels decrease due to stress[58,59]. Maternal plasma AA levels may likely influence BM AA levels[60]. A study conducted in mice revealed that maternal stress resulted in lower AA concentrations in maternal plasma and reduced growth of the offspring although the concentrations of the AAs asparagine and alanine were increased[61]. In a human study maternal stress during the postpartum period was associated with BM AA composition changes. The concentrations of protein-bound AAs (but not free AAs) in BM were higher in the high-stress group compared with the control group and were positively associated with BM cortisol concentrations[62].

The point of interest is the role of genetics in the mother and her BM and CHD infant dyad. CHD is mediated by complex genetics[63]. Hundreds of genes are implicated, including those coding for cardiac transcription factors, signaling pathways, and structural proteins. Both chromosomal aneuploidies (such as trisomy 21 and 22q11 deletion) and rare copy number variants contribute to syndromic and non-syndromic forms of CHD[63–65]. The interplay between growth restriction and CHD is still under investigation. Certain genetic syndromes that cause CHD also commonly feature short stature[64,65]. Hypotrophy is typically not the direct result of the same gene but rather of the overall impact of the syndrome on fetal growth and development.

In our center short-term routine genetic evaluation options are restricted to karyotyping for all CHD cases, and more specific genetic tests are performed for patients with multisystem congenital defects. Therefore, in our study we excluded infants with chromosomal abnormalities. We also did not include those with concurrent congenital anomalies of the gastrointestinal tract or central nervous system from a specific feeding issue point of view. Multiple studies confirm that genetic differences in mothers (such as variants in FADS1/FADS2, MTHFR, SLC30A2/ZnT2, FUT2, and others) significantly impact the concentration of nutrients, fatty acids, zinc, oligosaccharides, and other components in BM[66–69].

Although both maternal and CHD genetics are well-studied independently, there is no published evidence of a direct genetic correlation between the causes of CHD in the infant and the specific makeup of maternal BM or whether mothers of infants with genetic CHD have different BM compositions due to shared genetics. Johnson et al[70] evaluated genetic influences on gene regulation in milk and identified pathways linking milk gene expression with milk composition and infant gut health. In their study higher PER2 gene (the core circadian clock gene) expression correlated with lower milk volume with a higher percentage of milk fat[70]. Studying interactions between maternal genetics and breastfeeding in special infants’ conditions, i.e. CHD, is a future research orientation.

Operated neonates in our study showed a significantly greater decrease in WAZ from birth to discharge compared to non-operated neonates with CHDs, a finding that was expected. The incidence of postnatal hypotrophy (WAZ < −2) at discharge did not differ significantly between the operated (8/25) and non-operated (1/10) groups. The principal component analysis of whole BM composition (protein, carbohydrates, fat, and kcal) demonstrated no differences in mature milk composition in mothers of neonates with a hypotrophy at discharge. Operated neonates with CHD who developed hypotrophy by discharge had been characterized by significantly lower birth weight and birth WAZ. They also had more complex postoperative courses requiring enhanced nutritional and clinical management.

Neonates in the operated group showed strong positive correlations between colostrum total and true protein content and WAZ. The growth trajectory of healthy breastfed infants is known to differ from that of babies fed by formula and may be influenced by the human BM composition[71,72]. Although in studies on healthy infants regarding the effects of BM composition on growth are focused on avoiding excessive weight gain and subsequent obesity, data from these studies could be harnessed in elucidating regularities for those predisposed to malnutrition. In studies of “healthy” mother-infant dyads, BM carbohydrate and protein intakes were positively correlated with earlier infant growth gains [72-74].

In neonates of the operated group, the nutrient and energy composition of mature BM was not significantly associated with postoperative WAZ dynamics and the mean weight gain in either Spearman’s correlation analysis or the univariable linear regression analysis.

Among nutritional variables higher total calorie, fat intake, and enteral feeding volume on postoperative day 14 correlated with smaller WAZ decline, whereas delayed reintroduction of BM feeding and delayed exclusive BM feeding were associated with greater WAZ losses. According to the institutional protocol, the time required to switch from hydrolyzed formula to BM reflects the severity of the postoperative course. Conversely, this underscores the importance of early nutritional restoration in neonates with CHD, supporting the clinical approach that advocates the initiation of early enteral feeding postoperatively[15,22,34,75,76].

Univariable predictors of greater decline of WAZ from the preoperative period to discharge were clinical features indicating the severity of the condition in the postoperative period, such as longer postoperative antibiotic therapy, surgery with shunt formation, pneumonia in the postoperative period, longer NICU stay, later achievement of full enteral feeding postoperatively, later transfer to the mother’s ward, and longer postoperative hospitalization. In the multivariable analysis, only length of postoperative antibacterial treatment and later achievement of full enteral feeding postoperatively remained an independent predictor of growth decline, suggesting that these may mediate the impact of other postoperative complications. Therefore, achieving full enteral feeding early (ideally within 2 weeks postoperatively) appears crucial to minimizing postoperative growth failure.

However, our study covered a short observation period from birth to discharge from the hospital after surgery. A number of other studies analyzed mixed cohorts of children of different ages with a longer observation period post-discharge. A decrease in WAZ was revealed at discharge; however, operated children had successive increments in WAZ post-discharge compared with non-operated children[77]. In addition, an increase in WAZ was noted in children with simple CHDs (such as ventricular septal defect, atrial septal defect, patent ductus arteriosus, and pulmonary stenosis) compared with those with complex CHDs[78]. The data obtained indirectly indicate the influence of hemodynamic disorders on the growth trajectories of children with CHD.

In some studies, the features of the postoperative period were predictors of the WAZ trajectory. A longer hospital stay, the need for nutrition/supplementary feeding through a gastric tube at the time of discharge, and decreased appetite were associated with the absence of WAZ gain. On the contrary, a biventricular physiology of CHD was associated with persistent WAZ above zero[79]. According to analyses of a mixed pediatric population, the predictors of malnutrition were anemia[80,81], lower fat intake[80], and characteristics indicating the severity of hemodynamic disorders (congestive heart failure[80–82], low SaO₂, prolonged CHD symptoms, CHD type, and modified Ross scale scores of ≥ 7)[80]. Mignot et al[82] noted a high incidence of malnutrition in infants with CHD (43%). The predictors were low birth weight, CHD with increased pulmonary blood flow, heart failure, and the number of hospitalizations during the first year of life.

WAZ and its postoperative trajectory have long-term consequences for the growth of children with CHD. The predictors of underweight at 1 year after surgical treatment of CHD were the WAZ before surgery, at 1 month after surgical treatment of CHD, and at discharge. Severe malnutrition (wasting status) was associated with the length of hospital stay, formula feeding, and the weight-for-height z score at discharge[83]. Malnutrition during the preoperative period was associated with a lower WAZ during the postoperative period[78].

Several studies have analyzed the impact of WAZ on outcomes in children with CHD. An analysis of a mixed cohort of children aged 1 month to 10 years demonstrated that each one-unit decrease in WAZ was associated with an increase in the mortality odds ratio to 1.33 (1.25–1.41)[10]. In children who underwent Glenn surgery, a lower WAZ was a predictor of a longer hospital stay[84].

Limitations of the study

This study had several limitations. First, the relatively small sample size (n = 35) limited the statistical power and generalizability of the findings. The training/test data split was not performed because of sample size. Another possible limitation of the study was that enteral feeding strategies were individualized for each neonate. It introduced variability that may affect the interpretation of results. Variation in timing and frequency of milk sampling may also influence observed associations.

CONCLUSION

Only colostrum from mothers of operated neonates had a reduced fat concentration. In contrast, no differences were found in the transitional and mature BM composition among samples from mothers of operated neonates with CHD, non-operated neonates with CHD, and controls. No associations were revealed between the mature BM composition, the initiation of BM feeding after surgery, and the decrease in WAZ from the preoperative period to discharge in the operated neonates. In operated neonates the colostrum protein content was associated with weight gain in the early neonatal period. Our univariable linear regression analysis identified several significant predictors of postoperative WAZ decline, reflecting the severity of the postoperative condition. These included prolonged postoperative antibiotic therapy, surgery involving shunt placement, postoperative pneumonia, prolonged NICU stay, the delayed achievement of full enteral feeding, delayed transfer to the mother’s ward, and prolonged postoperative hospitalization. However, only the postoperative antibiotic therapy and delayed achievement of full enteral feeding after surgery remained significantly associated with a greater decline in WAZ in the multivariate regression analysis. Our findings indicated that the BM composition alone does not fully account for postoperative growth faltering in neonates with CHD, underscoring the importance of investigating metabolic dynamics during surgical recovery to develop optimized, physiology-based nutritional strategies for growth outcome enhancement.