The Effects of an Infant Formula Enriched with Milk Fat Globule Membrane, Long-Chain Polyunsaturated Fatty Acids and Synbiotics on Child Behavior up to 2.5 Years Old: The COGNIS Study

nutrients Article The Effects of an Infant Formula Enriched with Milk Fat Globule Membrane, Long-Chain Polyunsaturated Fatty Acids and Synbiotics on Child Behavior up to 2.5 Years Old: The COGNIS Study Ana Nieto-Ruiz 1,2,3,4 , Estefanía Diéguez 1,2,3 , Natalia Sepúlveda-Valbuena 1,2,5 , Florian Herrmann 1,2, Tomás Cerdó1,2,3 , Francisca López-Torrecillas 4,6 , Roser De-Castellar 7, Jesús Jiménez 7, Miguel Pérez-García4, María T. Miranda 8, Andrés Catena 4, JoséA. García-Santos 1,2,3, Mercedes G. Bermúdez 1,2,3 and Cristina Campoy 1,2,3,9,* 1Department of Paediatrics, School of Medicine, University of Granada, Avda. Investigación 11, 18016 Granada, Spain; [email protected] (A.N.-R.); [email protected] (E.D.); [email protected] (N.S.-V.); dr[email protected] (F.H.); [email protected] (T.C.); [email protected] (J.A.G.-S.); [email protected] (M.G.B.) 2 EURISTIKOS Excellence Centre for Paediatric Research, Biomedical Research Centre, University of Granada, 18016 Granada, Spain 3Instituto de Investigación Biosanitaria ibs. GRANADA, Health Sciences Technological Park, 18012 Granada, Spain 4Mind, Brain and Behaviour Research Centre-CIMCYC, University of Granada, 18011 Granada, Spain; [email protected] (F.L.-T.); [email protected] (M.P.-G.); [email protected] (A.C.) 5Nutrition and Biochemistry Department, Faculty of Sciences, Pontificia Universidad Javeriana, 110231 Bogotá, Colombia 6Department of Personality, Evaluation and Psychological Treatment, School of Psychology, University of Granada, 18011 Granada, Spain 7Ordesa Laboratories, S.L., 08820 Barcelona, Spain; Roser[email protected] (R.D.-C.); [email protected] (J.J.) 8Department of Biostatistics, School of Medicine, University of Granada, 18016 Granada, Spain; [email protected] 9 Spanish Network of Biomedical Research in Epidemiology and Public Health (CIBERESP), Granada’s Node, Institute of Health Carlos III, 28029 Madrid, Spain *Correspondence: [email protected]; Tel.: +34-629-308-695 Received: 29 October 2020; Accepted: 14 December 2020; Published: 15 December 2020   Abstract: Although early life nutrition influences brain development and mental health, the long-term effects of supplemented infant formula on children ´ s behavior remain unclear. We analyzed the effects of a bioactive nutrients-enriched-infant formula on children’s behavior up to 2.5 years, compared to a standard infant formula or breastfeeding. Current analysis involved 70 children who were fed a standard infant formula (SF, n=29) or a bioactive compounds enriched-infant formula (EF, n=41), during their first 18 months of life, and 33 breastfed (BF) children (reference group) participating in the COGNIS study. Behavioral problems were evaluated using the Child Behavior Checklist at 18 months and 2.5 years. Different statistical analyses were performed using SPSS. EF children aged 2.5 years presented fewer pathological affective problems than SF children. Besides, SF children were classified more frequently as bordering on internalizing problems than BF children. Rates of externalizing problems were increased in SF infants compared to EF and BF infants. Higher maternal IQ was found to have beneficial effects on internalizing and total problem rate in their offspring at 18 months of life; finally, higher maternal educational level was related with fewer ADHD problems in children at 18 months, as well as internalizing, externalizing, total and anxiety problems in children aged 2.5 years. Our analysis suggests that enriched infant formula fed infants seem to show fewer behavioral problems up to 2.5 years compared to a standard infant formula-fed infants. In addition Nutrients 2020,12, 3825; doi:10.3390/nu12123825 www.mdpi.com/journal/nutrients Nutrients 2020,12, 3825 2 of 16 to type of early feeding, maternal IQ and educational level seem to play a key role on children behavioral development. Keywords: infant formula; milk fat globule membrane; long-chain polyunsaturated fatty acids; bioactive compounds; children behavior; breastfeeding 1. Introduction Brain development is particularly vulnerable to adverse prenatal and postnatal early-life events [ 1 ]. Among them, inadequate nutrition during the first 1000 days of life may have a negative influence on brain growth and functional development, including cell proliferation, synaptogenesis and myelination [ 2 ], thereby affecting cognitive development and behavioral performance later in life [ 3 , 4 ]. Breastfeeding is the gold standard for infant feeding due to its multiple shortand long-term health benefits for both child and mother [ 5 ]. In fact, breastfeeding confers protection against breast and ovarian cancer and contributes to birth spacing; it might also protect against type 2 diabetes in lactating mothers. Moreover, human milk provides protection against child infections and probably reduces the risk of developing overweight and diabetes along life [ 6 ]. There is also scientific evidence for the positive effects of breastfeeding on cognitive function [ 7 ], although its beneficial effects on emotional regulation and behavior development still remain unclear [ 8 – 11 ]. Conversely, the use of infant formula for non-breastfed or partially breastfed infants is widely extended in high income countries, probably related to socioeconomic circumstances, such as educational level or employment status [ 6 ]. In light of current advances in food technologies, infant formulas are being continuously improved trying to resemble human milk in terms of optimal children’s growth and neurodevelopment [ 12 ]. Long-chain polyunsaturated fatty acid (LC-PUFA) enriched formulas have been approved and recommended by international authorities [ 13 ] because of their association with better visual acuity and cognitive development later in life [ 14 – 16 ]. However, potential long-term effects of these type of infant formulas on neurodevelopment and behavior are a current matter of discussion [17]. There is also an increasing scientific interest in infant formula enriched with milk fat globule membrane (MFGM) [ 18 – 20 ], a complex membrane structure composed of proteins (1–4% of the total content of milk protein), enzymes and lipids [ 21 ]. Timby et al. found that infants fed with an MFGM-supplemented infant formula showed higher scores in cognitive domains than those fed with a standard formula at 12 months of age [ 22 ]. Furthermore, a study performed in preschool children aged 2.5–6 years, who daily received phospholipid-rich MFGM formula over four months, showed reduced scores in internal, external and total behavioral problems [23]. Other important bioactive components of human milk are preand probiotics, which facilitate a healthy structural and functional development of gut microbiota in the offspring [ 24 , 25 ]. It is known that intestinal microbiota might contribute to neural network shaping and response to neurotransmitters, probably modulating brain development and neurocognitive function [26,27]. Until now, most studies have demonstrated the beneficial effects of one single bioactive compound supplementation in infant formulas [ 22 ]. However, new infant formulas are currently developed based on the addition of different human milk-like bioactive compounds, including LC-PUFAs, MFGM, synbiotics (preand probiotics), as well as human milk oligosaccharides, sialic acid, nucleotides, or gangliosides, among others. It is complicated to evaluate the effect of a single nutrient, so possible beneficial effects might be due to the synergistic effect of all formula components. Bearing in mind these considerations, the current study analyzed the effects of an infant formula supplemented with specific functional nutrients on behavioral development in healthy children up to 2.5 years old, compared to those fed with a standard formula or human milk. Nutrients 2020,12, 3825 3 of 16 2. Materials and Methods 2.1. Study Design and Subjects The COGNIS study (A Neurocognitive and Immunological Study of a New Formula for Healthy Infants) is a prospective, randomized double-blind, nutritional intervention study, registered at www.ClinicalTrials.gov, Identifier NCT02094547. The primary outcome of the COGNIS study was the neurocognitive development of children up to 6 years old, while secondary outcomes included behavior and temperament, infant growth, infectious and allergic events, among others. Detailed study design has been published elsewhere [ 28 , 29 ]. Briefly, 220 healthy full term babies were involved in the study; from these, 170 infants aged between 0–2 months old were randomized using a mathematical statistical method (ratio 1:1) to receive, during their first 18 months of life, either a standard infant formula (SF n=85), or an experimental infant formula (EF n=85) containing MFGM components [10% of total protein content (wt:wt)], synbiotics [Fructooligosaccharides: Inulin, proportion 1:1; Bifidobacterium longum subsp. infantis CECT7210 (Bifidobacterium infantis IM1) and Lactobacillus rhamnosus LCS-742], LC-PUFAs, gangliosides, sialic acid and nucleotides. Both infant formulas were provided by Laboratorios Ordesa, S.L. (Barcelona, Spain), and were developed according to the guidelines of the Committee on Nutrition of the European Society for Paediatric Gastroenterology, Hepatology and Nutrition (ESPGHAN) [ 30 ], as well as international and national recommendations for the composition of infant formulas (Table S1). Infants received initiation formula up to 6 months of age; follow-on formula was given between 6–18 months of age. Infants who were included in both formula groups received 30 days maximum of exclusive breastfeeding, followed by exclusive or >70% daily formula intake. As a reference group, 50 exclusively breastfed (BF) infants for at least 2 months, were included between 0–6 months of age. Inclusion criteria : eligible infants were healthy term infants [37–41 weeks gestational age (GA)], with adequate birth weight for GA (between 3–97 percentile), normal APGAR score (7–10) and umbilical pH ≥ 7.10. They must have had the availability to continue throughout the study period, and parents or legal guardians had to have signed the informed consent. The exclusion criteria were defined as follows: infants participating in another study; infants who suffered neurological disorders (hydrocephalus, perinatal hypoxia, intraventricular hemorrhage, neonatal meningitis, septic shock, West Syndrom, . . . ) or gastrointestinal disturbances (mainly cow’s milk protein allergy/intolerance or lactose intolerance); maternal pathological background (neurological diseases, mental illness, metabolopathies (type 1 diabetes mellitus), chronic diseases (hypothyroidism), maternal malnutrition or prenatal infections (TORCH complex,..)), mothers who received drug treatments during pregnancy or lactation which are potentially harmful for neurodevelopment (anxiolytics, antidepressants, . . . ); or parents’ impossibility to continue through the study. Regarding withdrawal criteria , those infants who met the following criteria after randomization were excluded from the study: infants fed with another infant formula (different from SF or EF) for a week or more; breastfed infants with formula intake >25% before 6 months; formula fed infants with human milk intakes higher than 25% beyond the third month of life; any adverse event that could interfere with study follow-up, cow’s milk protein allergy/intolerance or lactose intolerance, infant formula intake rejection or neurological disorder. The current study involved 132 infants at 18 months of life (SF =47; EF: 48; BF =37) and 103 children at 2.5 years old (SF =29; EF: 41; BF =33). A detailed flow chart of the recruited participants up to 2.5 years old is shown in Figure 1. Nutrients 2020,12, 3825 4 of 16 Nutrients 2020, 12, x FOR PEER REVIEW 4 of 17 Figure 1. Participant flowchart from recruitment to 2.5 years old. SF, standard infant formula; EF, experimental infant formula; BF, breastfed infants; D, drop-outs; E, exclusions, N, sample size. Up to 18 months of life, a total of 40 infants were excluded in the SF and EF groups as follows: 24 were excluded in the SF group (1 infant due to perinatal hypoxia, 1 infant had growth deficiency, 15 infants met the withdrawal criteria referring to formula intake, 2 infants were colic, 3 were excluded due to lactose intolerance, 1 infant was due to digestive surgical intervention, and 1 infant suffered hydrocephalus); 16 infants were excluded in the EF group (2 infants presented growth deficiency, 2 infants lactose intolerance, 11 infants met the withdrawal criteria referring to formula intake, and 1 was excluded due to epileptic seizure). Furthermore, one infant of the BF group was excluded because they were not breastfed. * BF infants were randomized between 0–6 months of age. § 1 mother did not fill the child behavior checklist (CBCL) test at 18 months old; ¥ 8 mothers did not complete the CBCL test at 18 months old; † 5 mothers did not fill the CBCL test at 2.5 years old; ‡ 2 mothers did not fill the CBCL test at 2.5 years old; # 2 mothers did not complete the CBCL test at 2.5 years old. Sixteen children at 2.5 years old did not show up at the evaluation (described as “did not attend”). During the follow-up visit at 18 months, dropouts were considered those who did not continue to participate in the study: mainly due to not wanting to continue in the study or not being able to come to the follow up visits, usually for a change in place of residence or workplace conditions of the parents. 2.2. Ethics, Consent, and Permissions This study was performed according to the updated Declaration of Helsinki Principles [31], and all procedures were approved by the Research Ethical Committee of the University of Granada and the Bioethical Committees for Clinical Research of the University San Cecilio and Mother–Infant Hospitals in Granada (Spain). All families were informed about all procedures during the follow-up and a signed informed consent was obtained from each parent or legal guardian before involving each child in the study. 2.3. CBCL Test (Child Behavior Checklist) To evaluate child behavior and psycho-emotional disorders at 18 months and 2.5 years old, the Child Behavior Checklist (CBCL) (validated Spanish version for 1.5–5 years), was filled out by parents or caregivers [32]. This questionnaire is composed of 101 items divided into two scales (internalizing and externalizing problems) and a total problems scale. The internalizing problems scale contains four syndromic subscales: emotionally reactive, anxious/depressed, somatic Figure 1. Participant flowchart from recruitment to 2.5 years old. SF, standard infant formula; EF, experimental infant formula; BF, breastfed infants; D, drop-outs; E, exclusions, N, sample size. Up to 18 months of life, a total of 40 infants were excluded in the SF and EF groups as follows: 24 were excluded in the SF group (1 infant due to perinatal hypoxia, 1 infant had growth deficiency, 15 infants met the withdrawal criteria referring to formula intake, 2 infants were colic, 3 were excluded due to lactose intolerance, 1 infant was due to digestive surgical intervention, and 1 infant suffered hydrocephalus); 16 infants were excluded in the EF group (2 infants presented growth deficiency, 2 infants lactose intolerance, 11 infants met the withdrawal criteria referring to formula intake, and 1 was excluded due to epileptic seizure). Furthermore, one infant of the BF group was excluded because they were not breastfed. * BF infants were randomized between 0–6 months of age. § 1 mother did not fill the child behavior checklist (CBCL) test at 18 months old; ¥ 8 mothers did not complete the CBCL test at 18 months old; † 5 mothers did not fill the CBCL test at 2.5 years old; ‡ 2 mothers did not fill the CBCL test at 2.5 years old; # 2 mothers did not complete the CBCL test at 2.5 years old. Sixteen children at 2.5 years old did not show up at the evaluation (described as “did not attend”). During the follow-up visit at 18 months, dropouts were considered those who did not continue to participate in the study: mainly due to not wanting to continue in the study or not being able to come to the follow up visits, usually for a change in place of residence or workplace conditions of the parents. 2.2. Ethics, Consent, and Permissions This study was performed according to the updated Declaration of Helsinki Principles [ 31 ], and all procedures were approved by the Research Ethical Committee of the University of Granada and the Bioethical Committees for Clinical Research of the University San Cecilio and Mother–Infant Hospitals in Granada (Spain). All families were informed about all procedures during the follow-up and a signed informed consent was obtained from each parent or legal guardian before involving each child in the study. 2.3. CBCL Test (Child Behavior Checklist) To evaluate child behavior and psycho-emotional disorders at 18 months and 2.5 years old, the Child Behavior Checklist (CBCL) (validated Spanish version for 1.5–5 years), was filled out by parents or caregivers [ 32 ]. This questionnaire is composed of 101 items divided into two scales (internalizing and externalizing problems) and a total problems scale. The internalizing problems scale contains four Nutrients 2020,12, 3825 5 of 16 syndromic subscales: emotionally reactive, anxious/depressed, somatic complaints and withdrawn. The externalizing problems scale is composed of two syndromic subscales: attention problems and aggressive behavior. In this study, externalizing and internalizing scales were used as substitutes for syndromic scales. Furthermore, the CBCL test also measures the following diagnostic and statistical manual of mental disorders (DSM)-oriented scales: affective problems, anxiety problems, pervasive developmental problems, attention deficit hyperactivity disorders (ADHD) and oppositional defiant problems. A Likert-type scale is used, with responses 0 (“not true”), 1 (“somewhat or sometimes true”) or 2 (“very true or often true”), taking into account the two months previous to the assessment. The total direct score on each scale was obtained automatically with Achenbach system of empirically based assessment (ASEBA) software. In order to perform a longitudinal study, direct scores on the total, externalizing, internalizing, and DSM-oriented scales were used (mean normative values: total problems =33.5; externalizing problems =14; internalizing problems =9; affective problems =2.2; anxiety problems =3.5; pervasive developmental problems =3; ADHD =5.5, and oppositional defiant problems =4 [ 33 ]. Additionally, typical scores (T) in each scale were divided, according to CBCL test standards, into three categories: normal (no behavioral problems, below the 93rd percentile: T<60), borderline to clinical range (behavioral problems, 93rd to 97th percentiles: 60 ≤ T ≤ 63) and clinical/pathological (behavioral problems, above the 97th percentile: T ≥64) [33,34]. 2.4. Statistical Analysis All statistical analyses were performed using IBM ® SPSS Statistics ® program, version 22.0 (SPSS Inc., Chicago, IL, USA). Normally distributed variables were expressed as the mean ± standard deviation (SD), and non-normal variables as the median and interquartile range (IQR). Categorical variables were presented as frequencies and percentages. Differences in CBCL scores among SF, EF and BF groups were contrasted using Chi-squared or Fisher’s exact tests for categorical variables, depending on the number of subjects in the analysis. In addition, analyses of group comparisons using a one-way analysis of covariance (ANCOVA) including maternal educational level and IQ, smoking during pregnancy, paternal educational level, place of residence and socioeconomic status as confounders, were performed. Furthermore, a generalized linear mixed model (GLMM) for repeated measures was developed to identify longitudinal differences between study groups. Bonferroni-corrected post-hoc comparisons were used to identify significant pair-wise group differences (corrected pvalue <0.05). Finally, a Wald test for logistic regression was performed to evaluate the effects of potential confounders on the likelihood (odds ratios (ORs) and 95% confidence interval (CI)) of having Borderline/Clinical values in the CBCL test. At 18 months, logistic regression models were performed, including one CBCL outcome as a dependent variable (internalizing, externalizing and total problems; affective, anxiety, pervasive developmental, ADHD and oppositional defiant problems) and the following confounder variable list as independent variables: maternal educational level and IQ, smoking during pregnancy, paternal educational level, place of residence, and the COGNIS groups. At 2.5 years old, logistic regression models were accomplished including the same CBCL outcomes mentioned above as dependent variables, and maternal educational level, socioeconomic status, place of residence and the COGNIS groups as independent variables. pvalues <0.05 were considered statistically significant. Statistical power was calculated for the current study applying the following equation [35]: n=2 Zα+Z2β δ0!2 . (1) Nutrients 2020,12, 3825 6 of 16 3. Results 3.1. Characteristics of the COGNIS Study Participants up to 2.5 Years Old Background and baseline characteristics of parents and children are shown in Table 1. At 18 months of life, BF infants’ mothers presented a higher educational level (p<0.001) and IQ (p=0.030) compared to mothers of both formula groups, and EF infants, respectively. Moreover, a higher percentage of the SF group mothers smoked during pregnancy with respect to mothers of the BF group (p=0.049). Paternal educational level was also higher in the BF group compared to the EF group (p=0.013); BF participants more frequently resided in rural areas (p=0.037). At 2.5 years, significant differences in maternal educational level, place of residence and socioeconomic status were found between study groups. In fact, mothers of BF infants showed a higher educational level (p=0.001) with respect to mothers of both infant formula groups. SF participants more frequently resided in urban areas than BF infants (p=0.012). Concerning socioeconomic status, those parents whose children were breastfed had higher status compared to EF-fed infants ´ parents (p<0.001). Due to the COGNIS study design, days of breastfeeding significantly differed between BF and formula study groups (p<0.001). No other significant differences in baseline characteristics were found in parent–child pairs participating in the COGNIS study. In order to reduce the possibility that selection bias could affect study conclusions, we analyzed background variables in the missing sample (i.e., those children who were excluded or withdrew from the study during the follow-up time). It is noteworthy that at the beginning of the study there were no differences between infant formula groups regarding baseline characteristics, as previously reported [ 25 ]. Considering drop-out subjects, differences among COGNIS groups were only found in maternal educational level: mothers of BF infants presented higher educational level compared with SF and EF infant’s mothers. Moreover, the exclusion rate was similar in both infant formula groups, being statistically different compared to the BF group (data not shown). 3.2. Effects of Type of Early Nutrition on Behavior Development in COGNIS Infants at 18 Months and 2.5 Years Old Table 2shows the association between the type of feeding and CBCL scores, categorized in each scale into normal, borderline, and pathological outcomes according to the standard test. No differences were found between COGNIS groups at 18 months of life. However, at 2.5 years old, SF-fed children were classified more frequently as borderline on internalizing problems than BF children (SF: 24.1%; BF: 3.0%; p=0.042). Moreover, EF children less frequently presented clinical pathological affective problems compared to SF-fed children at 2.5 years old (EF: 0.0%; SF: 13.8%; p=0.026). Overall, the percentage of EF children who were classified as normal behavior was similar to that of the BF children. We next evaluated the effects of type of feeding on CBCL scores in infants at 18 months and 2.5 years old (Table S2). At 18 months of life, no differences were found between SF, EF, and BF infants. Nevertheless, at 2.5 years old, results showed that children who received SF presented higher scores in internalizing (p=0.035) and total problems (p=0.017), as well as ADHD (p=0.039), compared to those who were breastfed. Interestingly, children who received EF or BF showed lower scores in externalizing problems (p=0.005) than SF-fed children. Overall, CBCL scores did not differ between children who received EF and breastfed infants. However, those significant differences did not remain after adjustment for maternal educational level, socioeconomic status and place of residence. Nutrients 2020,12, 3825 7 of 16 Table 1. Parents and children baseline characteristics depending on their type of feeding. Parents Characteristics 18 Months 2.5 Years Old SF (n=47) EF (n=48) BF (n=37) p1SF (n=29) EF (n=41) BF (n=33) p1 Maternal age (years) 32 (27–35) 32 (27.5–36) 34 (30–38) 0.12 33(29.5–3) 32 (29–36) 34(31–38.5) 0.27 Maternal pBMI, (kg/m2)24.3 ±3.9 24.8 ±4.4 24.5 ±3.2 0.86 24.2 ±3.5 24.3 ±4.3 24.9 ±3.2 0.71 Maternal educational level Primary 9(19.1%) a,b 13 (27.1%) b1 (2.7%) a <0.001 4 (13.8%) 6 (14.6%) 1 (3.0%) 0.001 Secondary 14 (29.9%) a8 (16.7%) a,b 2 (5.4%) b11 (37.9%) a10 (24.4%) a,b 2 (6.1%) b VT 9 (19.1%) 15 (31.2%) 9 (24.3%) 4 (13.8%) 15 (36.6%) 8 (24.2%) University 15 (31.9%) a12 (25.0%) a25 (67.6%) b10 (34.5%) a10 (24.4%) a22 (66.7%) b Maternal IQ (points) 104.7 ± 14.4 a,b 101.1 ±14.2a109.4 ±13.5 b0.030 105.4 ±15.5 102.4 ±12.9 109.5 ±14.1 0.10 Smoking during pregnancy No 31 (73.8%) a37 (84.1%) a,b 34 (94.4%) b 0.049 21 (77.8%) 33 (82.5%) 31 (93.9%) 0.16 Yes 11 (26.2%) a7 (15.9%) a,b 2 (5.6%) b6 (22.2%) 7 (17.5%) 2 (6.1%) GWG (kg) 6.5 (4–10) 5 (3–9) 6.3 (4.5–9) 0.80 5 (3–8) 5 (3–8.7) 6 (4.2–8.7) 0.42 Type of delivery Vaginal 34 (72.3%) 33 (68.8%) 27 (73.0%) 0.89 23 (79.3%) 27 (65.9%) 25 (75.8%) 0.46 Caesarean 13 (27.7%) 15 (31.2%) 10 (27.0%) 6 (20.7%) 14 (34.1%) 8 (24.2%) Postpartum Depression No 38 (80.9%) 39 (81.2%) 31 (83.8%) 0.93 22 (75.9%) 34 (82.9%) 28 (84.8%) 0.63 Yes 9 (19.1%) 9 (18.8%) 6 (16.2%) 7 (24.1%) 7 (17.1%) 5 (15.2%) Paternal educational level Primary 16 (34.8%) a,b 23 (48.9%) b6 (16.2%) a 0.013 10 (34.5%) 16 (39.0%) 6 (18.2%) 0.45 Secondary 13 (28.3%) 6 (12.8%) 5 (13.5%) 6 (20.7%) 8 (19.5%) 5 (15.2%) VT 7 (15.2%) 8 (17.0%) 11 (29.8%) 6 (20.7%) 9 (22.0%) 10 (30.3%) University 10 (21.7%) 10 (21.3%) 15 (40.5%) 7 (24.1%) 8 (19.5%) 12 (36.4%) Paternal IQ (points) 105.9 ±13.8 106.2 ±15.2 107.2 ±13.3 0.93 107.8 ±14.1 104.8 ±15.1 106.9 ±12.9 0.70 Socioeconomic status Low N/A N/A N/A6 (21.4%) a,b 13 (31.7%) b1 (3.0%) a <0.001 Mid–Low N/A N/A N/A 14 (50.0%) 21 (51.2%) 11 (33.3%) Mid–High N/A N/A N/A7(25.0%) a,b 7 (17.1%) b14 (42.4%) a High N/A N/A N/A1 (3.6%) a,b 0 (0.0%) b7 (21.3%) a Place of residence Urban 23 (48.9%) a14 (29.2%) a,b 9 (24.3%) b 0.037 18 (62.1%) a13 (33.3%) a,b 9 (27.3%) b 0.012 Rural 24 (51.1%) a34 (70.8%) a,b 28 (75.7%) b11 (37.9%) a26 (66.7%) a,b 24 (72.7%) b Siblings 0 21 (44.7%) 25 (52.1%) 20 (54.1%) 0.65 9 (31.0%) 16 (39.0%) 12 (36.4%) 0.79 ≥1 26 (55.3%) 23 (47.9%) 17 (45.9%) 20 (69.0%) 25 (61.0%) 21 (63.6%) Gestational age (weeks) 40 (39–41) 40 (39–40) 40 (39–41) 0.50 40 (39–41) 40 (37.5–40.5) 40 (38.5–41) 0.67 Newborn characteristics Newborn weight (g) 3350.1 ±413.7 3241.5 ±434.4 3395.7 ±381.1 0.20 3344.8 ±462.6 3297.3 ±569.0 3374.2 ±392.6 0.74 Newborn length (cm) 51 (49.5–52) 50.5 (49–52) 51 (50–52) 0.52 51 (49.2–52.9) 51 (49–52) 51 (50–52) 0.70 Newborn HC (cm) 34.5 (34–36) 34 (34–35) 35 (33.8–35) 0.15 35 (34–35.7) 34 (33.5–35) 35 (33.6–35.7) 0.25 Newborn Gender Boy 27 (57.4%) 29 (60.4%) 15 (40.5%) 0.16 18 (62.1%) 25 (61.0%) 13 (39.4%) 0.06 Girl 20 (42.6%) 19 (39.6%) 22 (59.5%) 11 (37.9%) 16 (39.0%) 20 (60.6%) Breastfeeding lactation (days) 10 (0–21) a11.5 (0.5–25) a420 (270–540) b<0.001 15 (5–43.5) a14 (2.5–29.5) a390 (270–765) b<0.001 Data are presented as n(%) for categorical data, mean ± SD for parametrically distributed data, and median (IQR) for non-parametrically distributed data. 1 p-values for overall differences between COGNIS-groups. ANOVA for normally distributed variables, Kruskal–Wallis rank-sum test for non-normal continuous variables and Chi-squared or Fisher’s exact test for categorical variables. Values which do not share the same suffix (ab) are significantly different in a Bonferroni post-hoc test. Bold: p-value <0.05. BF, breastfed infants; EF, experimental infant formula; GWG, gestational weight gain; HC, head circumference; IQ, intelligence quotient; IQR, interquartile range; Mid, middle; N/A, not available (data not recorded); pBMI, pre-conceptional body mass index; SD, standard deviation; SF, standard infant formula; VT, vocational training. Nutrients 2020,12, 3825 8 of 16 Table 2. Association between early nutrition and CBCL clinical clusters problems in children at 18 months and 2.5 years old. CBCL Scales 18 Months 2.5 Years Old SF (n=47) EF (n=48) BF (n=37) Fisher’s Exact Test p1SF (n=29) EF (n=41) BF (n=33) Fisher’s Exact Test p1 Internalizing Problems Normal 41 (87.2%) 41 (85.4%) 33 (89.2%) 2.606 0.65 17 (58.7%) 33 (80.5%) 28 (84.9%) 9.516 0.042 Borderline 5 (10.6%) 3 (6.3%) 3 (8.1%) 7 (24.1%) a6 (14.6%) a,b 1 (3.0%) b Pathology 1 (2.1%) 4 (8.3%) 1 (2.7%) 5 (17.2%) 2 (4.9%) 4 (12.1%) Externalizing Problems Normal 39 (83.0%) 41 (85.4%) 32 (86.5%) 4.220 0.37 22 (75.9%) 36 (87.8%) 29 (87.9%) 6.470 0.15 Borderline 5 (10.6%) 3 (6.3%) 5 (13.5%) 1 (3.4%) 3 (7.3%) 3 (9.1%) Pathology 3 (6.4%) 4 (8.3%) 0 (0.0%) 6 (20.7%) 2 (4.9%) 1 (3.0%) Total Problems Normal 40 (85.1%) 39 (81.3%) 31 (83.8%) 1.659 0.83 19 (65.5%) 33 (80.5%) 28 (84.8%) 6.970 0.13 Borderline 5 (10.6%) 4 (8.3%) 4 (10.8%) 3 (10.3%) 6 (14.6%) 2 (6.1%) Pathology 2 (4.3%) 5 (10.4%) 2 (5.4%) 7 (24.2%) 2 (4.9%) 3 (9.1%) Affective Problems Normal 45 (95.7%) 42 (87.5%) 35 (94.6%) 4.386 0.35 25 (86.2%) 40 (97.6%) 28 (84.8%) 8.491 0.026 Borderline 2 (4.3%) 2 (4.2%) 1 (2.7%) 0 (0.0%) 1 (2.4%) 3 (9.1%) Pathology 0 (0.0%) 4 (8.3%) 1 (2.7%) 4 (13.8%) a0 (0.0%) b2 (6.1%) a,b Anxiety Problems Normal 46 (97.9%) 46 (95.8%) 35 (94.6%) 1.948 0.94 24 (82.8%) 36 (87.8%) 29 (87.9%) 1.700 0.92 Borderline 1 (2.1%) 1 (2.1%) 1 (2.7%) 1 (3.4%) 1 (2.4%) 0 (0.0%) Pathology 0 (0.0%) 1 (2.1%) 1 (2.7%) 4 (13.8%) 4 (9.8%) 4 (12.1%) Perv. Develop. Problems Normal 44 (93.6%) 43 (89.6%) 33 (89.2%) 2.027 0.78 23 (79.3%) 34 (82.9%) 32 (97.0%) 5.207 0.25 Borderline 1 (2.1%) 3 (6.3%) 1 (2.7%) 4 (13.7%) 4 (9.8%) 1 (3.0%) Pathology 2 (4.3%) 2 (4.2%) 3 (8.1%) 2 (6.9%) 3 (7.3%) 0 (0.0%) ADHD Normal 44 (93.6%) 46 (95.8%) 35 (94.6%) 1.100 0.97 26 (89.7%) 39 (95.2%) 33 (100%) 3.589 0.46 Borderline 2 (4.3%) 1 (2.1%) 1 (2.7%) 1 (3.4%) 1 (2.4%) 0 (0.0%) Pathology 1 (2.1%) 1 (2.1%) 1 (2.7%) 2 (6.9%) 1 (2.4%) 0 (0.0%) Op. Def. Problems Normal 45 (95.7%) 46 (95.8%) 37 (100%) 3.118 0.78 28 (96.5%) 37 (90.3%) 32 (97.0%) 2.397 0.85 Borderline 2 (4.3%) 1 (2.1%) 0 (0.0%) 0 (0.0%) 1 (2.4%) 0 (0.0%) Pathology 0 (0.0%) 1 (2.1%) 0 (0.0%) 1 (3.5%) 3 (7.3%) 1 (3.0%) Data are presented as n(%) for categorical data. Fisher’s exact test for categorical variables. 1 p-values are comparisons between COGNIS-groups. Values which do not share the same suffix (ab) are significantly different in a Bonferroni post-hoc test. Bold: p-value <0.05. ADHD, attention deficit/hyperactivity disorders; BF, breastfed infants; EF, experimental infant formula; Op. Def., oppositional defiant; Perv. Develop., pervasive developmental; SF, standard infant formula. Nutrients 2020,12, 3825 9 of 16 Afterwards, a longitudinal study of behavior development up to 2.5 years of age was conducted using a GLMM of repeated measures (Table 3and Figure 2). Significant differences were found between study groups in internalizing, externalizing, total, ADHD, and oppositional defiant problems (Table 3). As shown in Figure 2, SF children presented increased rates in internalizing (p=0.047) (Figure 2A), total (p=0.044) (Figure 2C), ADHD (p=0.036) (Figure 2D), and oppositional defiant problems (p=0.030) (Figure 2E) compared to BF children during their first 2.5 years of age. Interestingly, our results seemed to suggest a similar score increase between EF children and BF children. In addition, children fed with SF showed a significantly higher score increase in externalizing problems than those fed with EF or BF (p=0.024) (Figure 2B). It is important to note that lower CBCL scores are associated with better behavior development, i.e., fewer behavioral problems. Table 3. Longitudinal study of CBCL scores up to 2.5 years of age in COGNIS participants. CBCL Scales 18 Months 2.5 Years F(df) p1p2p3ηp2 Internalizing Problems SF 8.7 ±4.6 12.6 ±9.9 a 3.041(2,98) 0.011 0.16 0.047 0.058 EF 8.0 ±6.7 9.0 ±6.0 a,b BF 6.8 ±5.5 7.5 ±6.0 b Externalizing Problems SF 14.6 ±6.2 16.9 ±7.9 a 3.885(2,98) 0.77 0.13 0.024 0.073 EF 12.5 ±6.9 12.2 ±6.8 b BF 12.4 ±6.6 11.1 ±7.7 b Total Problems SF 37.9 ±13.9 46.1 ±26.3 a 3.222(2,98) 0.15 0.16 0.044 0.062 EF 33.0 ±21.3 34.4 ±17.5 a,b BF 32.2 ±16.9 31.1 ±19.6 b Affective Problems SF 2.6 ±1.5 3.4 ±2.6 0.403(2,98) 0.14 0.07 0.67 0.008 EF 2.8 ±2.9 2.4 ±1.6 BF 2.2 ±1.8 2.9 ±2.2 Anxiety Problems SF 3.6 ±1.7 4.7 ±3.0 0.500(2,98) 0.005 0.34 0.61 0.010 EF 3.3 ±2.2 4.2 ±3.0 BF 3.5 ±2.2 3.7 ±3.0 Perv. Develop. Problems SF 3.5 ±2.5 4.4 ±3.7 1.710(2,98) 0.25 0.42 0.19 0.034 EF 3.3 ±3.5 3.8 ±3.0 BF 3.0 ±2.6 2.8 ±2.1 ADHD SF 5.8 ±2.5 6.1 ±2.6 a 3.442(2,98) 0.11 0.06 0.036 0.066 EF 4.9 ±2.3 4.6 ±2.7 a,b BF 5.2 ±2.6 3.9 ±2.7 b Op. Def. Problems SF 3.6 ±2.1 4.5 ±2.0 a 3.630(2,98) 0.034 0.39 0.030 0.069 EF 2.9 ±1.8 3.5 ±2.5 a,b BF 2.9 ±1.9 2.9 ±2.3 b Data are presented as mean ± SD. SF (n=29); EF (n=39); BF (n=33). p-values, F-values (F) and effect sizes ( ηp2 ) were obtained from a generalized linear mixed model of repeated measures: 1 shows differences between time points; 2differences between time points according to the COGNIS groups; 3longitudinal differences between the COGNIS groups. Values which do not share the same suffix (ab) are significantly different in a Bonferroni post-hoc test. Bold: p-value <0.05. ADHD, attention deficit/hyperactivity disorders; BF, breastfed infants; df, degrees of freedom; EF, experimental infant formula; Op. Def., oppositional defiant; Perv. Develop, pervasive developmental; SF, standard infant formula. Mean normative values: total problems =33.5; externalizing problems =14; internalizing problems =9; affective problems =2.2; anxiety problems =3.5; pervasive developmental problems =3; ADHD =5.5 and oppositional defiant problems =4 [33]. Nutrients 2020,12, 3825 16 of 16 42. Lönnerdal, B. Infant formula and infant nutrition: Bioactive proteins of human milk and implications for composition of infant formulas. Am. J. Clin. Nutr. 2014,99, 712S–717S. [CrossRef] 43. Sauer, P.J.J. Can we define an infant’s need from the composition of human milk? Am. J. Clin. Nutr. 2013 ,98, 521S–528S. 44. Qawasmi, A.; Landeros-Weisenberger, A.; Leckman, J.F.; Bloch, M.H. Meta-analysis of long-chain polyunsaturated fatty acid supplementation of formula and infant cognition. Pediatrics 2012 ,129, 1141–1149. [CrossRef] [PubMed] 45. Pecoraro, L.; Agostoni, C.; Pepaj, O.; Pietrobelli, A. Behind human milk and breastfeeding: Not only food. Int. J. Food Sci. Nutr. 2017,69, 641–646. [CrossRef] [PubMed] 46. Liu, J.J.; Green, P.; John Mann, J.; Rapoport, S.I.; Sublette, M.E. Pathways of polyunsaturated fatty acid utilization: Implications for brain function in neuropsychiatric health and disease. Brain Res. 2015 ,1597, 220–246. [CrossRef] [PubMed] 47. EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA). Scientific Opinion on the essential composition of infant and follow-on formulae. EFSA J. 2014,12, 3760. [CrossRef] 48. Pärtty, A.; Kalliomäki, M.; Wacklin, P.; Salminen, S.; Isolauri, E. A possible link between early probiotic intervention and the risk of neuropsychiatric disorders later in childhood: A randomized trial. Pediatr. Res. 2015,77, 823–828. [CrossRef] 49. Shaaban, S.Y.; El Gendy, Y.G.; Mehanna, N.S.; El-Senousy, W.M.; El-Feki, H.S.A.; Saad, K.; El-Asheer, O.M. The role of probiotics in children with autism spectrum disorder: A prospective, open-label study. Nutr. Neurosci. 2018,21, 676–681. [CrossRef] 50. Sonego, M.; Ll á cer, A.; Gal á n, I.; Sim ó n, F. The influence of parental education on child mental health in Spain. Qual. Life Res. 2013,22, 203–211. [CrossRef] 51. Kwok, M.K.; Leung, G.M.; Schooling, C.M. Breast feeding and early adolescent behaviour, self-esteem and depression: Hong Kong’s “Children of 1997” birth cohort. Arch. Dis. Child. 2013,98, 887–894. [CrossRef] 52. Liu, J.; Chen, X.; Lewis, G. Childhood internalizing behaviour: Analysis and implications. J. Psychiatr. Ment. Health Nurs. 2011,18, 884–894. [CrossRef] 53. Eisenberg, N.; Cumberland, A.; Spinrad, T.L.; Fabes, R.A.; Shepard, S.A.; Reiser, M.; Murphy, B.C.; Losoya, S.H.; Guthrie, I.K. The relations of regulation and emotionality to children’s externalizing and internalizing problem behavior. Child Dev. 2001,72, 1112–1134. [CrossRef] 54. Gustavson, K.; Ystrom, E.; Stoltenberg, C.; Susser, E.; Sur é n, P.; Magnus, P.; Knudsen, G.P.; Smith, G.D.; Langley, K.; Rutter, M.; et al. Smoking in Pregnancy and Child ADHD. Pediatrics 2017 ,139, e20162509. [CrossRef] [PubMed] 55. Malanchini, M.; Smith-Woolley, E.; Ayorech, Z.; Rimfeld, K.; Krapohl, E.; Vuoksimaa, E.; Korhonen, T.; Bartels, M.; van Beijsterveldt, T.C.E.M.; Rose, R.J.; et al. Aggressive behaviour in childhood and adolescence: The role of smoking during pregnancy, evidence from four twin cohorts in the EU-ACTION consortium. Psychol. Med. 2019,49, 646–654. [CrossRef] [PubMed] 56. Grossi, E.; Migliore, L.; Muratori, F. Pregnancy risk factors related to autism: An Italian case-control study in mothers of children with autism spectrum disorders (ASD), their siblings and of typically developing children. J. Dev. Orig. Health Dis. 2018,9, 442–449. [CrossRef] [PubMed] 57. Van den Bergh, B.R.H.; van den Heuvel, M.I.; Lahti, M.; Braeken, M.; de Rooij, S.R.; Entringer, S.; Hoyer, D.; Roseboom, T.; Räikkönen, K.; King, S.; et al. Prenatal developmental origins of behavior and mental health: The influence of maternal stress in pregnancy. Neurosci. Biobehav. Rev. 2017,117, 26–64. [CrossRef] Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).