Sex Differences in the Association of Pre-Transfusion Hemoglobin Levels with Brain Structure and Function in the Preterm Infant

Abstract

OBJECTIVE

To assess sex-specific differences in early brain structure and function of preterm infants after red blood cell (RBC) transfusions.

STUDY design

A single-center subset of infants with a birth weight <1000 g and gestational age 22-29 weeks were enrolled from the NICHD Neonatal Research Network Transfusion of Prematures (TOP) Trial. Hemoglobin concentration obtained directly prior to each transfusion (ptHb) was obtained longitudinally throughout each infant’s neonatal intensive care unit stay and used as a marker of degree of anemia (n=97). Measures of regional brain volumes using magnetic resonance imaging (MRI) were obtained at ~40 weeks postmenstrual age or at hospital discharge, if earlier (n=29). Measures of brain function were obtained at 12 months’ corrected age using the Bayley Scales of Infant & Toddler Development, 3rd Edition (Bayley-III) (n=34).

RESULTS

PtHb positively correlated with neonatal cerebral white matter volume in males (B=+0.283, p=0.006), but not females (B=−0.099, p=0.713), resulting in a significant sex interaction (p=0.010). Bayley-III gross motor scores and a pooled mean score were significantly lower in association with higher ptHb in females (gross motor score: B=−3.758, p=0. 013; pooled mean score: B=−1.225, p=0.030), but not males (gross motor score: B=+1.758, p=0.167; pooled mean score: B=+0.621, p = 0.359). Higher ptHb was associated with descriptively lower performance on multiple Bayley-III subscales in females, but not males.

CONCLUSION

This study demonstrates sex-specific associations between an early marker of anemia and RBC transfusion status (i.e. ptHb) with both neonatal white matter volumes, and early cognitive function at 12-months-old, in preterm infants.

Despite the increase in survival of preterm infants over the last several decades, preterm infants remain at risk for several short-term and long-term morbidities, including deficits in motor, sensory, cognitive, and behavioral domains ^1–7. Therefore, understanding early brain developmental outcomes may be critically important in determining early interventions for these infants. One possible etiology of altered neurodevelopment in the first weeks of life, as well as an early marker of risk that can be targeted for intervention, is the incidence of neonatal anemia, and its treatment: red blood cell (RBC) transfusion. Until recently⁸, it was not clear whether minimizing anemia and maintaining a higher, but normal, hemoglobin (Hb) level through transfusions, or tolerating anemia and minimizing transfusions, was better for the developing brain, as several studies have had variable results ^9–12.

the National Institute of Child Health and Human Development (NICHD) Neonatal Research Network Transfusion of Prematures (TOP) trial published the findings that the primary outcome of death or neurodevelopmental impairment (NDI) was not different in extremely low birth weight (ELBW) infants randomized to either a low- or high- hemoglobin threshold for RBC transfusions¹³. However, most neurodevelopmental outcomes for premature birth are far less severe than NDI, such as deficits in learning, attention, and behavioral issues. Moreover, these deficits are common, occurring in 50-70% of children born with very low birth weight ^{14, 15}, and can have lifelong, significant impacts on functioning. Another important consideration in evaluations of transfusion status on neurodevelopmental outcomes is exploration of sex-specific outcomes. Sex differences are critical biological variables in investigations of brain structure and function ^16–18. we have reported on sex-specific findings of prematurity ¹⁹ as well as in the context of the impact of transfusion status on brain structure and function outcomes ^20–23.

The current study evaluates the impact of transfusion status on optimal brain structure and function in a subset of neonates from the TOP trial who participated at the University of Iowa. Neonatal pre-transfusion hemoglobin (ptHb) levels were used to predict brain structure from term-equivalent MRI scans (~ 40 weeks’ gestational age) and functional outcomes at 12 months utilizing the Bayley Scales of Infant & Toddler Development, 3^rd Edition (Bayley-III). All analyses evaluated sex-specific differences.

METHODS

The current study was carried out at the University of Iowa Stead Family Children’s Hospital. Infants were enrolled in the multicenter NICHD Neonatal Research Network Transfusion of Prematures (TOP) Trial, a large two-armed randomized clinical trial comparing liberal and restrictive transfusion guidelines based on hemoglobin threshold in extremely low birth weight infants²⁴. Infants fulfilled the following inclusion criteria: birth weight ≤1000 g and gestational age ≥22 weeks but <29 completed weeks, and <48 hours of age. Exclusion criteria included: considered nonviable by attending neonatologist; cyanotic congenital heart disease; parental opposition to transfusion of blood; parents with hemoglobinopathy or congenital anemia; fetal transfusion; twin-to-twin transfusion syndrome; isoimmune hemolytic disease; severe acute hemorrhage, acute shock, sepsis with coagulopathy, or need for perioperative transfusion; prior blood transfusion on clinical grounds beyond the first 6 hours of life. From the 124 eligible infants, two infants received no RBC transfusions, one infant passed away prior to the end of the study, and 24 infants had no longitudinal hemoglobin measurements (see [Figure 1; available at www.jpeds.com] for flow diagram of participants). There were a total of 29 infants that were included in the included in final neonatal brain structure analyses and 34 infants that were included in final 12-month-old cognitive function analyses. 9 infants were included in both brain structure & cognitive function analyses.

Figure 1. CONSORT Flow Diagram of Study Participants.

29 infants were included in final neonatal brain structure analyses;

34 infants were included in final 12-month-old cognitive function analyses;

9 infants were included in both brain structure & cognitive function analyses.

A standard transfusion volume of 15 ml/kg was given according to criteria determined by the TOP trial protocol ²⁵. All RBCs transfused were tested and screened according to the policies of the hospital’s blood bank. Blood samples used in this study were left-over from clinically-indicated blood draws obtained by venous, arterial, or prewarmed heel capillary sampling, or from indwelling umbilical catheters, if present. Adequate clinically-indicated blood samples for all infants were obtained longitudinally throughout the hospital stay according to unit policy, at least weekly for the first month, and were not timed in any prescribed way relative to the blood transfusions. Clinical blood samples were stored at 4 degrees Celsius and processed (plasma was separated) within 4 hours of sample collection time. Plasma samples were subsequently stored at −80 degrees Celsius until analysis. All of the parents in this study agreed to have their infants’ scavenged blood assessed for cytokines and biomarkers. Hemoglobin obtained directly prior to each transfusion – termed the pre-transfusion Hb (ptHb) – was used as a marker of the degree of anemia prior to transfusion in each infant. Lower ptHb is therefore associated with lower transfusion threshold strategies (e.g. waiting for an infant’s hemoglobin to drop to the lower limit of normal range before transfusing the infant), whereas higher ptHb is generally associated with higher transfusion thresholds (e.g. transfusing infants at the upper limit of normal hemoglobin range).

Brain Structure

All infants underwent an unsedated MRI scan of their brain at approximately term equivalence (~40 weeks’ postmenstrual age) or shortly before discharge from the hospital, if earlier. An established protocol was used, in which infants were fed, then rocked to sleep by their mothers. Once asleep, they were carefully transferred to the MRI scanner. All procedures were approved by the local institutional review board, and parents signed a consent form prior to participation. Parents were compensated monetarily for the participation of their infant. Two 3T Siemens MRI scanners (TIM Trio and Skyra) were used in this study.

Structural Measures of Interest

The following regional volumes were extracted from the label maps for each neonate: white matter (WM), cerebral gray matter (GM), unmyelinated cerebral white matter (UWM), venous blood, cerebellar GM, cerebellar WM, basal ganglia, thalamus, hippocampus, and cerebrospinal fluid (CSF). Additional composite brain variables were created manually, including total intracranial volume (ICV), a sum of all regional volumes; and cerebellar tissue, a composite of cerebellar GM and WM.

Cognitive Function

The Bayley Scales of Infant Development, 3rd Edition (Bayley-III), were used to evaluate early cognitive function in these infants. Infants in the current study returned for a follow-up visit at approximately 12 months’ corrected age to undergo Bayley-III, including the sub-scale assessments of cognition, fine motor, gross motor, receptive language, and expressive language function.

Statistical Analyses

Mean ptHb concentrations, averaged throughout the hospital stay, were used to evaluate the relationship of transfusion status with measures of brain structure from MRI scans at term equivalent, and function at 12 months-corrected age.

SAS software (version 9.4, Cary, NC) was used to evaluate the association of mean ptHb with neonatal brain structural measures by robust linear regressions. All tissue-specific brain areas (WM, GM, and UWM) were reported as ratios to total ICV, as this accounts for the global decrement in volume by performing comparisons on the proportional measure. In order to reduce the influence of outliers in this small sample as well as account for wide leverage points in the data, robust linear regression models with MM-estimations were used in these analyses. All analyses controlled for birth weight. Each infant’s Bayley-III scale scores were assembled into a 5-entry vector for use as the dependent variable in repeated-measures analysis of variance (ANOVA), accounting for correlation among the scales. The effect of the primary independent variable, ptHb, was allowed to vary by sex and among scales. The ANOVA model also included covariate effects of gestational age, birth weight, and chronological age. Sex- and scale-specific estimates of the linear trend vs. ptHb, with 95% confidence limits, were calculated from the fitted model, adjusting to mean covariate values. Pooled sex-specific trends were obtained by averaging over the five scales and adjusting to a common scale mean. Robust regression identified and down-weighted 15 extreme datapoints using the Talworth criterion with 90% efficiency. P<0.05 was taken as the threshold for statistical significance.

RESULTS

Infants in this study were enrolled as part of the multicenter NICHD Neonatal Research TOP Trial (NCT01702805)²⁴. Enrollment dates for the entire multi-site clinical trial were between October 8, 2012 and January 2020. The infants included in this study were enrolled from a single-center site between February 6, 2013 and February 18, 2017. (Table i; available at www.jpeds.com) shows the demographics of all infants with hemoglobin measurements during the hospitalization (97 infants), infants with Bayley-III cognitive assessments at 12 months (34 infants), and infants with neonatal MRI brain imaging scans (29 infants). There were no significant differences between males and females in birthweight (BW), gestational age (GA), chronological or corrected age at follow-up cognitive assessment, or incidence of intraventricular hemorrhage (IVH) or periventricular leukomalacia (PVL). (Table II; available at www.jpeds.com) shows all measures of hemoglobin obtained longitudinally throughout the hospital stay. There were no significant differences in any hemoglobin measures between sexes.

Table 1.

Demographics of Infant Participants

	All Infant Participants With Hemoglobin Measurements during NICU Hospitalization (N=97)				Infant Participants with Neonatal MRI Brain Imaging [at ~Term-Equivalence] (N=29)				Infant Participants with Bayley-III Cognitive Assessment at 12-months-old (N=34)
	Male (N=43)	Female (N=54)	All Infants (N=97)	p / M=F *	Male (N=19)	Female (N=10)	All Infants (N=29)	p / M=F *	Male (N=18)	Female (N=16)	All Infants (N=34)	p / M=F *
Birth Weight (g)
Mean (SD)	725 (163)	696 (174)	709 (169)	p = 0.400	765 (153)	798 (218)	776 (175)	p = 0.673	734 (176)	658 (192)	698 (185)	p = 0.237
Median [Min, Max]	747 [279, 1010]	687 [379, 997]	712 [279, 1010]		785 [394, 1010]	896 [412, 988]	801 [394, 1010]		765 [394, 1010]	633 [395, 984]	694 [394, 1010]
Gestational Age (wks)
Mean (SD)	25.4 (1.61)	25.3 (1.69)	25.4 (1.65)	p = 0.629	25.5 (1.25)	26.2 (1.06)	25.8 (1.21)	p = 0.137	25.3 (1.54)	24.8 (1.42)	25.1 (1.49)	p = 0.296
Median [Min, Max]	25.7 [22.4, 28.4]	25.5 [22.3, 28.4]	25.7 [22.3, 28.4]		25.9 [23.1, 28.6]	26.1 [24.0, 28.1]	26.0 [23.1, 28.6]		25.8 [22.4, 27.4]	25.0 [22.4, 26.9]	25.4 [22.4, 27.4]
12-Month Chronological Age (months)
Mean (SD)	16.3 (1.49)	16.0 (0.758)	16.1 (1.15)	p = 0.447	16.7 (2.03)	15.9 (1.43)	16.4 (1.83)	p = 0.377	15.8 (0.825)	15.8 (0.476)	15.8 (0.669)	p = 0.784
Median [Min, Max]	15.8 [14.4, 20.3]	15.9 [14.9, 18.4]	15.9 [14.4, 20.3]		15.7 [14.4, 20.3]	15.3 [14.9, 18.4]	15.6 [14.4, 20.3]		15.6 [14.4, 17.8]	15.8 [15.0, 16.6]	15.7 [14.4, 17.8]
12-Month Corrected Age (months)
Mean (SD)	12.5 (1.36)	12.3 (0.826)	12.4 (1.10)	p = 0.508	13.0 (1.75)	12.3 (1.48)	12.8 (1.64)	p = 0.449	12.1 (0.877)	12.0 (0.464)	12.1 (0.697)	p = 0.864
Median [Min, Max]	12.0 [10.6, 15.8]	12.1 [11.3, 14.9]	12.1 [10.6, 15.8]		12.2 [10.6, 15.8]	11.7 [11.3, 14.9]	12.1 [10.6, 15.8]		11.9 [10.6, 14.3]	12.1 [11.3, 13.0]	11.9 [10.6, 14.3]
Intraventricular Hemorrhage (IVH), Germinal Matrix Hemorrhage (GMH), and Periventricular Leukomalacia (PVL) Grading				p = 0.890				p = 0.161				p = 0.492
Normal (No IVH/PVL)	31 (72.1%)	43 (79.6%)	74 (76.3%)		13 (68.4%)	8 (80.0%)	21 (72.4%)		13 (72.2%)	14 (87.5%)	27 (79.4%)
Grade I IVH or GMH	6 (14.0%)	4 (7.4%)	10 (10.3%)		3 (15.8%)	2 (20.0%)	5 (17.2%)		3 (16.7%)	1 (6.2%)	4 (11.8%)
Grade II IVH or GMH	3 (7.0%)	2 (3.7%)	5 (5.2%)		1 (5.3%)	0 (0%)	1 (3.4%)		1 (5.6%)	0 (0%)	1 (2.9%)
Grade III IVH or GMH	3 (7.0%)	1 (1.9%)	4 (4.1%)		1 (5.3%)	0 (0%)	1 (3.4%)		1 (5.6%)	1 (6.2%)	2 (5.9%)
Grade IV IVH or GMH -- periventricular hemorrhagic infarction	0 (0%)	3 (5.6%)	3 (3.1%)		0 (0%)	0 (0%)	0 (0%)		0 (0%)	0 (0%)	0 (0%)
PVL Present	0 (0%)	1 (1.9%)	1 (1.0%)		1 (5.3%)	0 (0%)	1 (3.4%)		0 (0%)	0 (0%)	0 (0%)

^*Independent Samples Student T-test for equal mean between males & females.

Table 2.

Measures of Hemoglobin in the Neonatal Intensive Care Unit (Bayley-III Participants Only)^*

	Males (N=17)	Females (N=13)	Total (N=30)	p / M=F *
Pre-transfusion Hemoglobin(ptHgb)				0.764
Mean (SD)	10.4 (0.907)	10.3 (1.40)	10.4 (1.13)
Median [Min, Max]	10.7 [9.12, 12.1]	10.1 [7.70, 12.3]	10.4 [7.70, 12.3]
Lowest Hemoglobin
Mean (SD)	8.69 (0.978)	8.83 (1.23)	8.75 (1.09)	0.730
Median [Min, Max]	8.60 [6.80, 10.3]	9.30 [6.60, 10.7]	8.70 [6.60, 10.7]
Highest Hemoglobin				0.770
Mean (SD)	16.9 (1.70)	16.8 (1.44)	16.8 (1.56)
Median [Min, Max]	16.7 [14.5, 20.3]	16.3 [15.2, 20.1]	16.5 [14.5, 20.3]
Average Hemoglobin (throughout hospital stay)
Mean (SD)	12.4 (0.944)	12.4 (0.755)	12.4 (0.847)	0.974
Median [Min, Max]	12.8 [10.6, 13.8]	12.7 [11.4, 13.4]	12.7 [10.6, 13.8]
Number of Transfusions
Mean (SD)	7.00 (4.03)	6.69 (3.75)	6.87 (3.85)	0.831
Median [Min, Max]	6.00 [2.00, 17.0]	6.00 [1.00, 15.0]	6.00 [1.00, 17.0]

^*Independent Samples Student T-test for equal mean between males & females; Four Bayley-III participants had missing hemoglobin values and are excluded from this analysis

Neonatal Brain Imaging & Baseline Cognitive Performance

(Table III; available at www.jpeds.com) and (Table IV; available at www.jpeds.com) demonstrate the average neonatal brain imaging variables and Bayley-III scores on all five subscales, respectively, by sex. There were no significant differences in neonatal brain structure or baseline Bayley-III cognitive performance between males and females in this small sample.

Table 3.

Neonatal MRI Brain Variables

	Male (N=19)	Female (N=10)	All Infants (N=29)	p / M=F *
Volume (in cubic centimeters, cc, x 103) a
Total Intracranial Volume (ICV)
Mean (SD)	466(66.3)	465 (112)	466 (82.9)	p = 0.994
Median [Min, Max]	452 [363, 586]	439 [324, 650]	450 [324, 650]
[Brain Tissue Type] Volume % Total ICV b
White Matter
Mean (SD)	3.18 (0.324)	3.12 (0.325)	3.16 (0.320)	p = 0.653
Median [Min, Max]	3.13 [2.61, 3.97]	3.20 [2.52, 3.55]	3.19 [2.52, 3.97]
Gray Matter
Mean(SD)	33.09 (5.61)	32.05 (5.42)	32.73 (5.47)	p = 0.636
Median [Min, Max]	34.28 [21.74, 42.10]	29.63 [26.05, 42.79]	32.73 [21.74, 42.79]
Unmyelinated White Matter
Mean (SD)	35.36 (5.70)	35.21 (6.54)	35.31 (5.89)	p = 0.949
Median [Min, Max]	34.56 [23.99, 45.92]	35.44 [26.92, 45.39]	35.20 [23.99, 45.92]
Cerebellar Tissue
Mean (SD)	4.41 (0.602)	4.41 (0.700)	4.41 (0.625)	p = 0.984
Median [Min, Max]	4.44 [3.35, 5.52]	4.27 [3.61, 5.64]	4.39 [3.35, 5.64]

^*Independent Samples Student T-test for equal mean between males & females.

^aVolumes expressed in cubic centimeters (cc’s) multiplied by 10³

^bVolume (in cc’s) of specific brain tissue type to ICV Ratios, expressed as % of total ICV; this accounts for the global decrement in volume, by performing comparisons on the proportional measure

Table 4.

Bayley-III Cognitive Assessment Subscale Scores

	Male (N=18)	Female (N=16)	All Infants (N=34)	p / M=F *
Cognitive Raw Score (Corrected Age)
Mean (SD)	39.8 (10.3)	38.4 (5.01)	39.2 (8.18)	p = 0.614
Median [Min, Max]	41.0 [4.00, 53.0]	37.5 [31.0, 48.0]	40.0 [4.00, 53.0]
Receptive Communication Raw Score (Corrected Age)
Mean (SD)	13.7 (2.78)	14.0 (1.79)	13.8 (2.32)	p = 0.719
Median [Min, Max]	14.0 [8.00, 19.0]	14.0 [10.0, 17.0]	14.0 [8.00, 19.0]
Expressive Communication Raw Score (Corrected Age)
Mean (SD)	14.7 (4.06)	13.7 (4.19)	14.2 (4.09)	p = 0.495
Median [Min, Max]	15.0 [4.00, 22.0]	12.5 [7.00, 22.0]	15.0 [4.00, 22.0]
Fine Motor Raw Score (Corrected Age)
Mean (SD)	26.2 (6.54)	26.4 (1.90)	26.3 (4.87)	p = 0.895
Median [Min, Max]	27.0 [2.00, 32.0]	27.0 [21.0, 30.0]	27.0 [2.00, 32.0]
Gross Motor Raw Score (Corrected Age)
Mean (SD)	38.9 (7.24)	31.6 (13.3)	35.6 (10.9)	p = 0.078
Median [Min, Max]	40.0 [17.0, 48.0]	36.0 [5.00, 45.0]	39.0 [5.00, 48.0]

^*Independent Samples Student T-test for equal mean between males & females.

Pre-transfusion Hb Prediction of Brain Structure at Term Equivalent

Table V shows the results of the regression analyses of ptHb with neonatal MRI brain variables. There were no significant associations between ptHb and total intracranial volume, cerebellar tissue, GM, or UWM in males or females. PtHb was significantly positively associated with WM in all infant participants (B = +0.268, F = 19.00, p < 0.001); that is, the higher ptHb, the greater the white matter volume. However, this was exclusively driven by the males, as ptHb was positively correlated with white matter volume in males (B = +0.283, F = 10.90, p = 0.006). For females, there was no relationship of ptHb with white matter volume (B = −0.099, F = 0.14, p = 0.713), resulting in a significant sex interaction (F = 8.26, p=0.010). Figure 2 shows this regression model visually, demonstrating that there is a significant difference in white matter volumes in association with ptHb between males and females. A post-hoc sensitivity analysis, in which the same regression analyses and sex interactions were performed after excluding infants diagnosed with any grade of IVH or PVL, was unable to be performed due to small sample numbers.

Table 5.

Robust Regression Analysis of Covariance of Pre-Transfusion Hgb (PtHb) with Neonatal MRI Brain Variables^*

	Overall				Male (N=16)				Female (N=7)				Sex Interaction
	B†	SE	F	p	B†	SE	F	p	B†	SE	F	p	(p / M=F)
Volume (cc’s)
Total Intracranial Volume (ICV)	−32489.61	22216.83	2.14	0.159	−37182.84	23263.43	2.55	0.134	5829.96	88614.84	0.00	0.951	0.764
[Brain Tissue Type] Volume (expressed as % of ICV)
White Matter (WM)	0.268	0.062	19.00	<0.001	0.283	0.086	10.90	0.006	−0.099	0.267	0.14	0.713	0.010
Gray Matter (GM)	−0.009	1.382	0.00	0.995	−0.433	1.503	0.08	0.778	1.204	3.426	0.12	0.743	0.497
Unmyelinated White Matter (UWM)	−0.745	1.612	0.21	0.649	0.399	1.816	0.05	0.829	−3.851	5.093	0.57	0.492	0.239
Cerebellum	0.066	0.158	0.18	0.680	0.114	0.191	0.36	0.560	−0.076	0.512	0.02	0.890	0.632

^*Controlling for birth weight; Six participants had missing hemoglobin values and are excluded from this analysis

^†B, Beta coefficient estimate, slope of relationship between predictor (pre-transfusion hemoglobin, PtHb) and outcome variable (e.g. brain area volume); SE, Standard Error of Beta Coefficient

Figure 2. Regression Model of White Matter Volume (expressed as ratio to total intracranial volume [ICV]) with Mean Pre-Transfusion Hemoglobin Level (g/dL).

Controlling for birth weight

• Sex Interaction:	p = 0.010*
• Females Only:	p = 0.713
• Males Only:	p = 0.006*

^*

p<0.05

Pre-transfusion Hb Prediction of Cognitive Function at 12 months Corrected Age

Table VI shows the results of the multivariate analyses of ptHb with all five Bayley-III subscales, as well as a pooled trend estimate. In every case the estimated trends for male and females went in opposite directions. Figure 3 displays this data visually by sex. Elevations of ptHb were associated with descriptively lower performance on four out of the five Bayley-III subscales in females, including cognitive, expressive language, and fine and gross motor function. Of the five individual subscales, only gross motor function showed a significant sex interaction with ptHb (B = +5.516, SE = 1.864, p = 0.007): that is, an increase in ptHb was associated with lower scores on Bayley-III gross motor function in females only (−3.758 Bayley-III score for every 1 g/dL increase in ptHb; SE = 1.395, p = 0.013), but not males (B = +1.758, SE = 1.232, p = 0.167). The overall mean Bayley score also showed significant sex-specific associations with ptHb levels (B = +1.846, SE = 0.851, p = 0.041): females had lower mean Bayley scores with higher ptHb (B = −1.225, SE = 0.528, p = 0.030), but males’ mean Bayley scores showed no association with ptHb levels (B = +0.621, SE = 0.664, p = 0.359). A post-hoc sensitivity analysis was subsequently conducted, in which the same regression analyses and sex interactions were performed after excluding infants diagnosed with any grade of IVH or PVL. This analysis did not change the results of any cognitive sub-scale or composite scores regression or sex interaction significance.

Table 6.

Pooled Regression Analysis of Covariance of Pre-Transfusion Hgb with Five Bayley Subscales^*

	Male (N=17)				Female (N=13)				Sex Interaction
	B†	SE	t	p	B†	SE	t	p	Difference**	SE	t	(p / M=F)
Cognitive	0.959	1.167	0.82	0.420	−1.264	0.837	−1.51	0.145	2.223	1.436	1.55	0.135
Expressive Language	0.360	0.972	0.37	0.714	−1.252	0.725	−1.73	0.098	1.613	1.212	1.33	0.196
Receptive Language	−0.428	0.759	−0.56	0.578	0.264	0.537	0.49	0.628	−0.692	0.938	−0.74	0.468
Fine Motor	0.455	0.683	0.67	0.512	−0.115	0.502	−0.23	0.822	0.569	0.846	0.67	0.508
Gross Motor	1.758	1.232	1.43	0.167	−3.758	1.395	−2.69	0.013	5.516	1.864	2.96	0.007
Mean	0.621	0.664	0.94	0.359	−1.225	0.528	−2.32	0.030	1.846	0.851	2.17	0.041

^*Controlling for birth weight, GA, and chronological age at cognitive assessment; Four participants had missing hemoglobin values and are excluded from this analysis

^†B, Beta coefficient estimate, fitted slope of relationship between predictor (ptHb) and outcome variable (Bayley subscale raw score or mean)

SE, Standard Error of Beta Coefficient

^**Difference in Beta coefficients between males and females

Figure 3. Pooled Regression Models of Cognitive Measures at 12 months old with Mean Pre-Transfusion Hemoglobin Level (g/dL).

All data points and trend lines are adjusted to mean gestational age, birth weight, and chronological age

1. Regression Models of Five Individual Subscales (cognitive, expressive language, receptive language, fine motor, gross motor) of Bayley Scales of Infant & Toddler Development, 3^rd edition (Bayley-III) with Mean Pre-transfusion Hemoglobin Level
2. Pooled Regression Model of Bayley Scores with Mean Pre-transfusion Hemoglobin Level; Adjusted to common scale mean

DISCUSSION

The current study demonstrates the sex-specific differences in both brain structure at term-equivalent age and brain function at 12 months of age in response to hemoglobin level in the preterm infant. Neonatal cerebral white matter volume was directly related to hemoglobin level in males only, in whom higher ptHb was associated with higher white matter volumes, while in females there was no such association. A converse pattern was found in the 12-month developmental scores, in which higher ptHb predicted lower (worse) Bayley-III gross motor and overall mean scores in females only, while in males there was no such association. Taken together, these results suggest that males with the lowest hemoglobin levels would have the worst outcomes, while females with the highest hemoglobin levels may have the worst outcomes. This pattern of sex-specific findings as a function of pre-transfusion hemoglobin (ptHb) level is shown in (Figure 4; available at www.jpeds.com).

Figure 4. Sex-Specific Outcomes Based on Hemoglobin Levels.

Hypothesis of Sex-Specific Effect of Hemoglobin Level on Brain Development

The current findings from short term outcomes in infancy are supportive of our findings evaluating long-term outcomes of infants assigned to higher or lower hemoglobin thresholds for transfusion, where females in the higher hemoglobin threshold group had the worst outcomes in regard to both brain structure (volume of white matter) and brain function (cognitive skills) ^20–22. Also, in support of this sex-specific hypothesis is a study in which cognitive outcomes were studied in premature infants who were randomized to delayed cord clamping (DCC) or early cord clamping (ECC). These two groups can be considered parallel to our groups in which DCC would be similar to a higher threshold group and ECC would be analogous to a lower threshold group. At 12-month follow-up, scores on the Ages and Stages Questionnaire were 12 points lower in the females in the DCC compared with females in the ECC group while males, in the ECC group had scores that were 5 points lower than the boys in the DCC group, resulting in a significant sex by group interaction ²⁶. This is the same pattern of poorest outcome in higher threshold females and lower threshold males found in both our short term outcomes (current study) and long-term outcomes studies of infants randomized to higher or lower hemoglobin thresholds for transfusion.

Among the important factors to consider in evaluations of transfusion status on neurodevelopmental outcomes are methodology of analysis and exploration of sex-specific outcomes. In regard to methodology, instead of defining binary categories of liberal versus restrictive, and death/NDI versus no death/NDI, a more powerful approach is identification of continuous measures of transfusion status and neurodevelopmental outcomes rather than categorical variables. That is, the continuous measures used in this study provided a more granular approach to outcomes by identifying measures that relate to the variability in transfusion status and predict the spectrum of functional outcomes. Secondly, sex differences are critical biological variables in investigations of brain structure and function ^16–18. As such, we have reported on sex-specific findings of prematurity ¹⁹ as well as in the context of the impact of transfusion status on brain structure and function outcomes ^20–23. We found that females have worse brain structural and functional outcomes with liberal transfusion strategies OBJ. Given the pattern of sex-specific outcomes, it is likely that there are sex-specific mechanisms that underlie the different patterns of response to RBC transfusion. (Figure 5; available at www.jpeds.com) features a hypothesis in which higher transfusion thresholds in females may lead to an inflammatory process affecting brain development, whereas in males with lower transfusion thresholds, anemia is the driver for impaired brain development.

Figure 5. Potential Sex-Specific Mechanisms Underlying Neurodevelopmental Outcomes of Transfusion.

Schematic representation of potential mechanisms of brain injury, in which females with liberal transfusions leads to an inflammatory process affecting brain development, whereas males with restricted transfusions points to anemia as the driver for impaired brain development.

In regard to females and cerebral white matter maldevelopment, two major pathologic hallmarks of injury are activation of microglia and depletion of “premyelinating” oligodendrocytes ²⁷. Premyelinating oligodendrocytes are highly vulnerable to injury by excitotoxicity, free radicals, and proinflammatory cytokines. Perinatal inflammation has been proposed as a risk to white matter in particular and a potential precursor to later neurodevelopmental deficits. , pro-inflammatory cytokine production and endothelial activation after RBC transfusion, often referred to as transfusion-related immunomodulation, has been proposed to underlie several neonatal morbidities ^{28, 29}. Two potential factors are that white matter in females is more vulnerable to inflammation, or that inflammatory processes are more robust in females. Our study evaluating the same cohort of TOP infants as described here found a sex-specific difference in the neonatal inflammatory response to RBC transfusion, showing a rise in monocyte chemoattractant protein-1 (MCP-1), interleukin-6 (IL-6), and interferon gamma-induced protein 10 (IP-10), all of which were limited to females ³⁰. Further, elevations of MCP-1, which has been shown to be associated with brain injury in premature infants ³¹, were associated with lower Bayley-III cognitive, language, and motor scores at 12 months in these preterm infants. These findings, in conjunction with the current findings, lend support to the notion that the mechanism by which female premature neonates are more vulnerable to abnormal brain development is via inflammation.

A potential distinctive mechanism for abnormal brain development after transfusion in males is anemia, which results in an overall lack of vital substrates such as oxygen and iron. Iron deficiency has been shown to be vital for brain development, and even initial iron deficiency that is later corrected can have long-term impacts ³². Moreover, studies have shown that male infants have lower iron stores than females ^{33, 34}. In addition, an animal model of chronic hypoxia showed sex-specific deficits in which male brain development was highly vulnerable whereas females were mostly unaffected ³⁵. These findings support the theory that for males, factors driving poor outcomes in those with restricted transfusions may be due to anemia. A limitation of the current study is that we are unable to answer the question of whether or not anemia itself is associated with the observed deficits in brain structure and function, as most extremely low birth weight infants will receive at least one RBC transfusion during their neonatal intensive care hospital stay.

One significant limitation of the current neurocognitive and brain structural associations with ptHb in this study were the small sample sizes in regression analyses for a number of reasons: marked loss to follow-up at 12 months for Bayley assessment, inherent difficulty in imaging neonates without sedation, and the exclusionary criterion for participation in brain imaging in this study at an institution with surgical clips used for patent ductus arteriosus (PDA) closures. Therefore, there is the potential for selection bias introduced by this low follow-up rate. In order to reduce the influence of outliers in this small sample, as well as account for wide leverage points in the data, robust linear regression models with MM-estimations were used in these analyses to account for the small sample sizes. It is important to note that the significant interactions demonstrating structural and functional outcomes (i.e. white matter volume or Bayley-III score) by sex, especially in the small sample size of this study, add weight to the hypothesis that the associations observed are truly sex-specific. Larger sample sizes may provide further evidence of these associations by allowing for identification of sex-specific differences in brain structural drivers of cognitive outcomes, or structure-function correlations. However, it should also be highlighted that there were no significant differences between participants and non-participants (for both the brain structural and cognitive outcome arms of the study) in all demographic characteristics, including gestational age, birth weight, age at testing, presence of IVH/PVL, or measures of hemoglobin in this sample. These are all potential confounders in brain outcome studies, but with no differences demonstrated between infants that were included in the follow-up studies and those that were not, there may be increased confidence that the results of this study are generalizable to other preterm infants. Another limitation in this study was that there were more males (N=19) than females (N=10) with high-quality MRI scans. Although females had descriptively lower WM volumes in association with increased ptHb in this study, there remains the possibility that the increased structural WM volume observed in the males is the norm for all premature infants, rather than a specific neurodevelopmental sequela for the preterm males in the higher threshold group in this study. A third limitation to the current study includes the fact that although the pre-transfusion hemoglobin is a readily available and easily measured serum marker in the clinical setting, the actual effects of hemoglobin on the brain are likely multifactorial. However, an averaged value of ptHb throughout the hospitalization may be a better continuous marker of the overall oxygen-deficient state, or lack thereof, of the infant over several weeks to months during a critical time in development it may also represent the sustained result of increased transfusions, e.g. a mounted inflammatory response, rather than a transient response to another biologic process in the infant.

This study investigated the potential association of average ptHb, which was obtained as a longitudinal measurement throughout the hospitalization, with structural brain volumes in the neonatal period in preterm infants, and cognitive functioning at 12 months old. Our results show sex-specific findings of the poorest outcomes in males managed with lower hemoglobin thresholds for transfusion and females with higher transfusion thresholds. Possible mechanisms for these findings were explored. Future investigations will be necessary to validate the sex-specific differences we observed in this subset by studying larger numbers of subjects, as well as to elucidate the mechanisms of adverse brain outcomes, both in the neonatal period and later in development.