Male sex is an independent risk factor for poor neurodevelopmental outcome at 20 months’ corrected age, in human milk-fed very preterm infants: a cohort study

ABSTRACT Objective: To determine associations between sex and neurodevelopmental outcomes in human milk-fed very preterm infants, adjusted to early measured nutrient intakes and other neonatal cofactors. Methods: Consecutive inborn human milk-fed infants, with gestational age <33 weeks, were eligible. In-hospital energy and protein intakes have relied on measured human milk composition. The Bayley Scales of Infant Development II mental and psychomotor developmental indexes were used to assess neurodevelopment at 20 months’ corrected age. After univariate analysis, some covariables were used for linear multiple regression. Results: Thirty-two infants were included, with a mean (standard deviation) gestational age of 29.8 (1.8) weeks, and a median birth weight of 1168g (interquartile range 990-1419g). Minimum recommended intakes were achieved in 63.6% and 15.2% of infants for protein and energy, respectively. The mental and psychomotor developmental indexes were within normal limits in 93.8% of infants. The mean mental and psychomotor developmental indexes were significantly lower in males. Only male sex negatively and significantly affected the mental and psychomotor developmental indexes (B=-9.44; 95%CI: -17.64- -1.23; adjusted r2=0.17; p=0.026), adjusted to gestational age and measured energy intake. Conclusion: In human milk-fed very preterm infants, males had a significantly lower mental and psychomotor developmental indexes score at 20 months’ corrected age, adjusted for gestational age and measured energy intake.


❚ INTRODUCTION
Male sex has been described as an independent risk factor for adverse neurodevelopmental outcomes in preterm infants. (1,2) Some reasons have been proposed to explain this disadvantage in male preterm infants, such as impaired adaptive response to prenatal stress, (3) a pro-oxidant state in the placentae, (4) and specific morphological characteristics of the brain. (5,6) The type of feeding is an important independent factor for neurocognitive development. In very preterm infants, early human milk (HM) intake appears to be independently associated with better neurodevelopment. (7,8) It is controversial whether nutritional requirements in the neonatal period differ between male and female preterm infants, and whether neurodevelopment is affected. A study involving preterm infants did not find an interaction between diet and sex on neurodevelopment at 9 months' corrected age. (9) In contrast, other authors have suggested that suboptimal in-hospital nutrition in preterm infants can alter brain organization and neurocognitive outcome, with particular sensitivity in males. (10,11) In addition, improved neurodevelopmental outcome in female preterm infants receiving higher intakes of protein (10) and docosahexaenoic acid (12) has been reported.
Published studies assessing the association between early diet and neurodevelopmental outcome adjusted to sex, in HM-fed preterm infants, have not measured the composition of HM. Instead, nutrient intakes provided by HM have relied on its assumed composition, or on growth as surrogate of nutritional support. (10,11,(13)(14)(15) In this study, we hypothesized a better outcome in females, adjusted for in-hospital measured energy, protein and protein-to-energy ratio intakes, and other relevant neonatal covariables.

❚ OBJECTIVE
To determine the association between sex and neurodevelopmental outcome in human-milk fed very preterm infants.

❚ METHODS Study design and participants
This is a secondary analysis based on a birth cohort study primarily designed to evaluate the associations of in-hospital nutrient intakes, with body composition and neurodevelopmental outcome. In the present study, the association between sex and neurodevelopmental outcome at 20 months' corrected age was investigated. The study was performed in the neonatal intensive care unit of Maternidade Dr. Alfredo da Costa, Centro Hospitalar de Lisboa Central, Lisbon, Portugal. The study was approved by the Hospital Ethics Committee and is registered at ISRCTN (ID: 27916681). Informed written consent was obtained from the parents or legal representative of each infant.
The study protocol has been described elsewhere. (16) Briefly, consecutive inborn neonates at <33 weeks' gestation and who were HM-fed at least 80mL/kg/day (convenience criterion for tolerance to enteral feeding) were eligible for the study. Infants with major congenital malformations and triplets or more were not included. Infants with diagnosed inborn errors of metabolism, and those who were subsequently formula-fed >12.5% of the enteral volume intake, transferred, deceased, or unavailable for neurodevelopment assessment were excluded. In our unit, enteral feedings are administered every 3 hours (eight times per day); as a convenience criterion, the infants were considered to be predominantly HM-fed if no more than one of eight meals (12.5%) was replaced with formula.
The demographic and clinical independent variables recorded were sex, single or twin pregnancy, gestational age (GA), birth weight, small-for-GA (birth weight <10 th percentile), (17) measured protein, energy, and proteinto-energy ratio daily intakes, Score for Neonatal Acute Physiology − Perinatal Extension II (SNAP-PE II) score, (18) use of prenatal and postnatal corticosteroids, diagnosis of late-onset sepsis, (19) severe necrotizing enterocolitis (grade ≥3B), (20) severe periventricular/ intraventricular hemorrhage (grade ≥3), (21) multicystic periventricular leukomalacia, (22) days on invasive ventilation and on supplemental oxygen, retinopathy of prematurity (stage 3 or plus disease), (23) and chronic lung disease. (24,25) Gestational age was determined by an early prenatal ultrasonography or by the first day of the last menstrual period and, in the case of assisted reproductive technology, by adding two weeks to the conception age. (26)

Nutrition protocol
Infants were managed according to the neonatal intensive care unit nutrition protocol, based on international (27)(28)(29) and national (30,31) recommendations for neonatal parenteral nutrition (PN) and enteral nutrition. Briefly, PN was initiated within the first 2 postnatal hours with 2.5g/kg/day of amino acids and was increased up to 3.8-4.0g/kg/day; parenteral lipids were initiated within the first 24 postnatal hours with 1g/kg/day and increased up to 3g/kg/day. Early enteral trophic feeding (10 to 20mL/kg/day) was initiated within the first 2 to 4 postnatal days using HM; subsequently, enteral nutrition was increased as the PN was proportionally reduced. Until 35 weeks' corrected age exclusive HM (own mother's milk -OMM, or donor HM − DHM) was used. If the OMM was not sufficient after 35 weeks' corrected age, preterm formula was used owing to limited DHM stock. Nutrition was prescribed by physicians in collaboration with a nutritionist.
Donor human milk and OMM were stored frozen in the maternity milk bank. (16) For each infant, a 3mL homogenized sample of daily pools of OMM was analyzed using a mid-infrared HM analyzer (Miris AB, Uppsala, Sweden). The composition of DHM was always analyzed. The physicians, nutritionist, and psychologist were blinded to HM composition during the entire study period. When breastfeeding predominated (unknown volume intake and composition), the OMM composition analysis was suspended. An HM fortifier (Aptamil FMS ® , Milupa/Danone GmbH, Friedrichsdorf, Germany) was used when the HM intake was at least 100mL/kg/day. The standard fortification method was supplemented with modular protein (32) (Aptamil Protein Supplement powder ® , Milupa/ Danone GmbH, Friedrichsdorf, Germany) and/or modular medium-chain triglycerides (MCT OIL, SHS Nutricia/Danone ® , GmbH, Friedrichsdorf, Germany), considering the lowest reported HM protein content. (33) The administered volumes of OMM and DHM were used to calculate the energy, protein, and per intakes. (33,34) The volumes and powder weights of PN solutions and commercial products were also accounted for in these calculations. Following the nutrition protocol used in our unit, the minimum targeted daily intakes according to body weight, once the daily fluid intake of 140 to 150mL/kg was reached, were as follows: energy 110kcal/kg; protein (g/kg) 4.0 if <1000g, 3.7 if <1200g, 3.6 if <1800g, and 3.4 if >1800g; and protein-to-energy ratio of 3.6 if <1000g, 3.2 if <1800g, and 2.6 if >1800g. (27,29) The minimum recommended daily nutrient intakes for body weight, achieved in at least 75% of the days, after reaching a steady fluid intake, was assessed in each infant up to 35 weeks' corrected age.

Neurodevelopmental assessment
The mental developmental indexes (MDI) and psychomotor developmental indexes (PDI) of the Bayley Scales of Infant Development II (BSID-II) (35) were used to assess functional development in the study sample. This is a validated tool for infants up to 42 months of age. (36,37) The MDI measures cognitive, language, and personal-social functioning, and the motor scale measures control of gross and fine motor functions. (35) Reliability of scores of both scales was demonstrated in the normative sample of children, with internal consistency coefficients ranging from 0.78 to 0.93 for MDI and from 0.75 to 0.91 for PDI. (35) Test-retest score reliability in the normative sample is moderate to high for the approximate age range examined in the present study (r=0.91, for MDI, and r=0.79, for PDI, at 24 months of age). (35) The mean MDI and PDI scores were classified as within normal limits (≥85), mildly delayed (70 to 84), or severely delayed (<70). (38) The neurodevelopmental assessment was initially scheduled at 18 months' corrected age. However, at this age other assessments have been scheduled, and it was decided to delay the assessment 2 months (20 months' corrected age) to assess the neurodevelopment in better conditions.

Statistical analysis
Statistical analysis was performed using R version 3.4.0 (R Foundation for Statistical Computing, Vienna, Austria) and Statistical Package for the Social Sciences (SPSS) version 13 (SPSS Inc., Chicago, IL, USA). To determine the association between in-hospital nutrient intakes and neurodevelopment at 20 months' corrected age, a sample of 75 infants was estimated to detect a difference ≥11 in MDI or PDI and a standard deviation (SD) ≥8 points (39) in a normally distributed variable, with a significance level of 0.05 and 80% power. The normality of continuous numerical variables was einstein (São Paulo). 2019;17(3):1-8 tested using the Shapiro-Wilk test, and the data are expressed according to adequate central and dispersion measures. Univariate analysis was performed using the Student's t test, the Mann-Whitney U test, Pearson's r, or Kendall's tau-b, as appropriate. Categorical variables were described by their absolute values, and relative frequencies compared using the χ 2 or Fisher's exact test. When OMM composition analysis was not possible, a post-hoc analysis for imputation of missing values was performed, as previously described. (16,22,(24)(25)(26) The effect of covariables was first explored using univariate analysis, selecting associations with p<0. 15 and fulfilled all the assumptions, for linear multiple regression. The backward method was then used, with MDI and PDI as the dependent variables and sex as independent variable.

❚ RESULTS
The study was interrupted before the calculated sample size was reached, owing to logistical constraints. The period of enrollment was from February 1 st , 2014, to February 28, 2015 (13 months), during which 156 eligible infants were identified (Figure 1).
Neurodevelopment was assessed in 32 infants, with 26 very preterm (≥28 weeks) and 6 extremely preterm (<28 weeks). Their characteristics and clinical outcomes, and comparison between sexes are summarized in table 1.
All infants received a complete cycle of antenatal betamethasone. No cases of small-for-GA, postnatal steroids, severe necrotizing enterocolitis, multicystic periventricular leukomalacia, retinopathy of prematurity, and transferred or deceased infants were recorded. No significant differences between sexes were found in relation to protein, energy and protein-to-energy ratio intakes (Table 2), twins, SNAP-PE II, late onset sepsis, days on supplemental oxygen, chronic lung disease, postnatal steroids, severe necrotizing enterocolitis, and severe intra-periventricular hemorrhage. Compared with the 56-excluded formula-fed infants, the femaleto-male ratio did not differ significantly (p=0.82); however, the 33 infants who were enrolled had a lower GA ( During the hospital stay, infants were exclusively or predominantly HM-fed or breastfed. Fortified HM was started on postnatal day 7 (without differences between sexes) and generalized to all infants by postnatal day 28. The minimum recommended daily nutrient intakes was attained in 63.6% of infants for protein, 15.2% for energy, and 93.9% for protein-to-energy ratio. No differences between sexes were found either in initiation of fortified HM or in minimum recommended daily nutrient intakes. The median daily protein, energy, and protein-to-energy ratio intakes ranged from 2.7 to 4.2g/kg, 53.7 to 109.2kcal/kg, and 3.4 to 5.6, respectively.
In the entire sample, the mean (SD) score for MDI was 100.2 (11.5) and for PDI was 97.4 (8.0). The mean MDI score was within normal limits in 30 (93.8%) of infants, and mildly and severely delayed in two males, respectively. The mean PDI score was within normal limits in 30 (93.8%) of infants, and mildly delayed in one female and one male, respectively. No significant differences in MDI and PDI scores were found between extremely preterm and very preterm infants.
In the univariate analysis, the MDI distribution in females was skewed toward the upper scores. Mental developmental index and PDI in males, and PDI in both sexes, remained in a normal distribution.
The mean MDI score was significantly lower in males, and no significant differences were found in the mean PDI score between sexes (Table 3).     Thus, only covariables for MDI were analyzed. The MDI score was weakly and negatively correlated with GA (r=-0.274; p=0.129), and weakly and positively correlated with daily energy intake (r=0.289; p=0.109).
Only sex, GA, and daily energy intake were selected for linear multiple regression, because other covariables did not meet criteria to enter in the multivariable models. Neither daily energy intake nor GA significantly affected the MDI score; only male sex negatively and significantly affected (B=-9.44; 95%CI: -17.64--1.23; adjusted r 2 =0.17; p=0.026) the MDI score, adjusted to GA (Table 4). Specifically, the MDI score was 9.44 points lower (95%CI: -17.64--1.23) in males than in females.

DISCUSSION
In this study, it was found that among infants born very prematurely, males had a significantly lower MDI score (-9.44) than females at 20 months' corrected age, adjusted for GA and daily energy intake. This model explains 17% of the variation in MDI. These results were found in a context of a suboptimal nutrition, with only 63.6% and 15.2% infants attaining the minimum recommended protein and energy intakes, respectively. This reflects poor effectiveness of the fortification method used. (16) Despite early suboptimal nutrition, MDI and PDI were within normal limits in 93.8% of infants at 20 months' corrected age.
Several studies have reported that male sex is independently associated with poor neurodevelopmental outcomes in preterm infants at a wide range of GA. In late preterm and very preterm infants, significantly lower MDI scores at 24 months' corrected age have been reported in males. (40) In a large sample of very preterm infants from the National Institute of Child Health and Human Development Neonatal Research Network, male sex was found to be an independent risk factor for MDI <70 at 18 to 22 months' corrected age. (41) In another study including 797 infants born at 23 to 28 weeks gestation, male sex was significantly associated with cognitive delay at 24 months' corrected age. (42) Male sex is also a predictor of poor neurological outcome at extremely low GA, as demonstrated at 30 months' corrected age in surviving males from the EPICure cohort born at ≤25 weeks' gestation. (2) Neurological outcome in male preterm infants appears to be more sensitive to suboptimal nutrition than in females. (10,11) In our sample, a weak-to-moderate correlation had been previously reported between suboptimal nutrient intakes and weight gain velocity. (16) A large cohort study of very preterm infants, a loss of weight z-score during neonatal hospitalization was associated with a poor neurological outcome at 24 months' corrected age in males. (10,11) In another cohort of very low birth weight infants, it was found that PDI assessed at 24 months' corrected age increased with greater protein and energy intake during the first postnatal week, especially in males. (10,11) In our study, the MDI score was weakly and positively correlated with in-hospital energy intake in male preterm infants, although this effect was lost after adjustment to GA.
The particular reason for male vulnerability is largely unclear, but may be related to impaired adaptive response to prenatal stress, with potential influence on early brain development. (3) A pro-oxidant state was observed in the placentae of preterm male infants born within 72 hours of antenatal betamethasone exposure, compared with females, conferring a physiological postnatal disadvantage for males. (4) In our sample, all infants received antenatal steroids. In three-dimensional magnetic resonance images, a smaller interface between the cortical gray and white matter was revealed in very preterm males compared with females. (5) In a study assessing sex differences in brain volumes at 8 years of age, only males born preterm exhibited significantly reduced white matter compared with term males, whereas white matter volumes were equivalent in preterm and term females. (6) The present study has strengths that should be acknowledged. A homogeneous cohort of exclusively or predominantly in-hospital HM-fed preterm infants was studied. This is important because the early HM feeding is reported to be independently associated with better neurodevelopment. (7,8) In this context, this is the first study evaluating the association between sex and neurodevelopmental outcome, adjusted to measured nutrient intakes provided by HM, instead of estimated intakes or relying on growth as a surrogate of nutritional support. (10,11) However, this study also had limitations. Owing to premature termination of the study, the sample became undersized and, in turn, probably underpowered to detect significant associations between nutrient intakes and neurodevelopmental outcomes. Nevertheless, in multivariable analysis the studied sample size was powered enough to find a significant association between sex and neurodevelopment, adjusted for nutritional intake. The unexplained variation in MDI might be due to underestimated variables owing to the small sample size, or by non-controlled confounders, including socioeconomic background, maternal education, and environmental stimuli. (11,43) In our cohort, sex was unevenly distributed; however, this factor was adjusted for in the multivariable models. Finally, a bias for withdrawal exists because enrolled infants completing the study were significantly more immature, were more frequently singletons, and stayed longer in the hospital than those who were excluded. With this bias, more vulnerable infants to neurodevelopment impairment could be included. Nevertheless, the main independent variable was not affected because the female-to-male ratio did not differ significantly between the included and excluded infants.

❚ CONCLUSION
Male sex was found to be an independent risk factor for lower mental developmental index in human milkfed very preterm infants, in a context of a suboptimal nutrition. This study reinforces that, in human milk fed preterm infants, an accurate evaluation of the effect of nutrition on neurodevelopmental outcome should consider sex and rely on measured human milk composition instead of estimated intakes or growth as a surrogate of nutritional support.