The NUTRISHIELD study is a parallel group, non-randomized, observational study performed at the Division of Neonatology of the University & Polytechnic Hospital La Fe (HULAFE), including TI and PI and their mothers, covering the period from birth to 24 months of infant's age. In addition, mothers providing DHM to study participants are also included.

The study protocol has been approved by the Scientific and Ethics Committee for Biomedical Research (CEIm) of the HULAFE (#2019-289-1). All methods have been performed in accordance with the relevant guidelines and regulations and written permission has been obtained from mothers or legal representatives by signing an informed consent form.

Confidentiality of subjects is maintained during the study. All participants have been assigned a code and data allowing personal identification will not be shared at any time. Participants may withdraw consent for participating in the study at any time.

Participants were recruited at the end of pregnancy (when admitted to the obstetric ward) or within one week after birth in case of hospitalized infants at HULAFE. Recruitment started in October 2020 and was completed in July 2022. Three mother-infant groups were enrolled, i.e., PI fed with OMM (PI-OMM), PI fed with pasteurized DHM (PI-DHM), and TI receiving OMM (TI-OMM), and their mothers.

The inclusion criteria were (i) acceptance of the mother to participate and sign an informed consent form, (ii) a gestational age <32 weeks for the group of PI and >37 weeks for the group of TI, and (iii) exclusive consumption (i.e., >80% of total intake) of either OMM or DHM at time point CEN (complete enteral nutrition, 150 mL/Kg/day) for PI or RBW (recovery of birth weight) for TI. Although the type of feeding of PIs included in either group (PI-OMM or PI-DHM) may have changed at later time points, they remained in the same study group, so the effect of feeding of each type of milk at CEN and its long-term effects can be studied. Likewise, the use of infant formula and the initiation of weaning at later time points was recorded without affecting follow-up assessments. The exclusion criteria were (i) non-compliance with any of the inclusion criteria, (ii) the requirement of a special diet for the mother (e.g., celiac disease, diabetes) or maternal consumption of probiotics, (iii) the need of intestinal surgery, severe congenital malformations, or chromosomopathies of the infant, (iv) mother's residence outside the Valencian Community and (v) severe language barriers hampering the collection of necessary data from mothers not speaking Spanish and/or English.

Sample size estimation was performed for the primary outcome (i.e., % weight gain/month from birth until hospital discharge of PI fed with DHM and OMM) and was based on result of a previous study (42). Considering the median % weight gain/month (interquartile range, IQR) of 52% (IQR: 30) in the PI-DHM group and 63% (IQR: 21) in the PI-OMM, 18 infants per group were needed to achieve a power of 90% with an alpha error of 5% between the % weight gain/month of PI exclusively receiving either OMM or DHM at CEN.

On the other hand, HM donors were recruited during their regular visits at the hospital's HM Bank of the Valencian Community. All participants met ordinary criteria for HM donation (e.g., negative screening results of a series of transmissible diseases, toxic habits, or some chronic medication) and accepted to sign an informed consent form and participate in the study.

Samples from lactating mothers and their infants are collected at different timepoints, as shown in Figure 1. At birth, cord blood, and urine, and faeces from both infants and mothers are collected when possible. When PIs achieve CEN and TIs RBW, as well as one, two, three, and six months after delivery, urine and faeces from mothers and infants, as well as, HM from lactating mothers, are collected. Infants' and mothers' buccal swabs are collected at month six. During hospitalization of participants, samples are collected at the hospital by healthcare providers. After discharge, samples are collected at home and either transported to the hospital by study participants or staff of the NUTRISHIELD study. Mothers receive Standard Operation Procedures (SOPs) with detailed instructions and all the required material for sample collection at home. Additionally, hospital's staff instructs mothers on how to collect and store samples correctly.

Samples collected at home are stored at 4°C during a maximum of 24 h before their transport to the hospital on ice, where they are registered, aliquoted, labelled, and stored at −80°C until analysis. A HM and faeces aliquot is directly stored in a tube containing a microbiome preservative solution. For DHM, collection times respective to delivery are heterogeneous, since from each donor several aliquots are collected over time, pooled, and then subjected to Holder pasteurization in batches (i.e., 62.5°C during 30 min followed by rapid cooling to 4°C). The lactation time was estimated as the mean (standard deviation, SD) of time elapsed during the collection of different aliquots from one DHM batch, being 16 (SD 12) days.

Mothers' and infants' buccal swab samples are collected by thoroughly rubbing a flocked sterile swab up and down the inner side of both cheeks for 30 s each. Swabs are cut and stored in a sterile tube at −80°C until further analysis. Discarded arterial and venous blood from the umbilical cord is collected in EDTA blood sample tubes after the placenta is delivered and separated from the baby. Blood is centrifuged (1,300 × g for 10 min at 20°C) and the upper plasma layer is stored in opaque vials at −80°C until analysis. Mother's first morning urine is collected in a polypropylene container, and infant's urine is collected by placing sterile cotton pads in the diaper and, after urinating, squeezing them with a sterile plastic syringe into sterile tubes (see Figure 2). Mother's faeces are collected in a polypropylene container, and infant's faeces are collected directly from the diaper into sterile tubes using sterile tweezers.

Mothers express milk using breast milk pumps following an SOP employed routinely in the hospital and the HM bank. The removable parts of the breast milk pump as well as the collection bottles are sterilized before their use. HM is collected at least three hours after breastfeeding. Mothers are required to wash their hands with soap and water and clean nipples with water. HM should be preferably expressed between 7 and 10 a.m., full expression of one breast is required (a minimum volume of 50 ml is recommended), and details regarding the extraction method (e.g., breast pump brand), date, and extracted volume are registered. For DHM samples, aliquots of each pooled DHM sample are collected before and after Holder pasteurization.

For all three study groups, information obtained from clinical records is collected. Anthropometric data (e.g., weight, height, and head circumference), sociodemographic, dietary, other relevant information related to lifestyle (e.g., smoking habits, sleep hours, and physical activity), and psychosocial status are provided through questionnaires by participating mothers as summarized in Table 1. As shown in Figure 1, 24 h dietary recalls are recorded at all timepoints except at birth, food frequency questionnaire (FFQ) at month 1, maternal psychosocial status at CEN/RBW and month six (corrected age for PI groups), and infant neurodevelopment at months six and twenty-four. The questionnaires are administered online, on paper, or through direct interviews with the mothers.

Regarding the 24 h recall, trained researchers ask for all foods and beverages participants consumed during the previous day, using the multiple-pass method (43). Recall data are analyzed in terms of nutrients using the dietary analysis software Nutritionist Pro™ (2007, Axxya Systems, Texas, USA). Additionally, dietary intake is grouped into food groups, (i.e., fruits, vegetables, bread/starch, meat/high fat, meat/medium fat, meat/low fat, meat/very low fat, milk/non-fat, milk/low fat, milk/full fat, eggs, fish, soy, dairy products, soft drinks, and other carbohydrate-rich foods).

The FFQ is administered by trained personnel, and it comprises 142 questions on the consumption of foods that are commonly eaten by the Spanish population throughout a year, including dairy products, cereals, fruits, vegetables, meat, fish, legumes, added fats, alcoholic beverages, stimulants, and sweets. Using a 9-grade scale (“never or less than 1 time/month”, “1–3 times/month”, “1 time/week”, “3–4 times/week”, “5–6 times/week”, “1 time/day”, “2–3 times/day”, “4–5 times/day”, “≥6 times/day”) participants are required to indicate the absolute frequency of consuming a certain amount of food, expressed in grams, milliliters or in other common measures, such as slice, tablespoon, or cup, depending on the food. The previous month is set as the timeframe. The FFQ is an easy-to-use questionnaire and is not expected to increase the burden of lactating mothers.

Based on the FFQ-responses, adherence to the Mediterranean diet is evaluated by using the MedDietScore, a composite score calculated for each participant (44). For food groups presumed to be part of the Mediterranean pattern (i.e., those with a recommended intake of 4 servings per week or more, such as non-refined cereals, fruits, vegetables, legumes, olive oil, fish, and potatoes) higher scores are assigned when the consumption is according to the rationale of the Mediterranean pattern, while lower scores are assigned when participants report no, rare, or moderate consumption. For the consumption of foods presumed to be eaten less frequently within the Mediterranean diet (i.e., consumption of meat and meat products, poultry, and full fat dairy products), scores are assigned on a reverse scale. As this study is focused on lactating mothers, the original score was modified by removing the alcohol consumption component. Thus, the range of this modified MedDietScore is between 0 and 50, with higher values of the score indicating greater adherence to the Mediterranean diet.

Maternal psychosocial status is measured at recruitment and at six months, including social support [Multidimensional Scale of Perceived Social Support, MSPSS (45)], family functioning [Family Adaptability and Cohesion Evaluation Scale, FACES (46)], perceived stress [Perceived Stress Scale -10, PSS-10 (47)], anxiety [State-Trait Anxiety Inventory, STAI (48)], depression [Edinburgh Postnatal Depression Scale, EPDS (49), Beck Depression Inventory (BDI) (50)] symptoms, and traumatic life events [Traumatic Experience Questionnaire, TEQ (51, 52)]. Mother-infant postpartum bonding [Postpartum Bonding Questionnaire, PBQ (53)] and parental stress [Parenting Stress Index, PSI (54)] are also examined.

As for the infants, at six and twenty-four months (corrected age for PIs) assessment of neurodevelopment is conducted during a hospital visit by a trained psychologist and psychiatrist from the team. Parents complete the Ages & Stages Questionnaires, Third Edition (ASQ-3) (55) for psychomotor development evaluation and the Infant Rothbart's Temperament Questionnaire (IBQ) (56) for infant temperament. The ASQ is a well standardized instrument used in clinical and research practice to examine children's psychomotor development gathering items in five domains: communication, gross motor, fine motor, problem solving, and personal-social. Clinical observation and Merrill-Palmer revised scales (57) for development are applied during the session.

Mother's concerns and attitudes to breastfeeding are also registered through the Iowa Infant Feeding Attitude Scale (IIFAS) (58) at CEN/RBW and six months.

Milk donors are asked to provide sociodemographic, medical, dietary, and other lifestyle data. PSS-10 is evaluated when entering the study in order to correlate with hormones and other compounds present in the milk. In particular, each time a batch of milk is prepared, a FFQ is administered (representing the period of time that the batch of milk covered) and the donors are asked to provide medical, anthropometric, and lifestyle data.

A range of state-of-the-art as well as novel laboratory methods are employed for the analysis of the collected biological samples, as summarized in Table 2.

Due to the special requirements of the study, three sensor prototypes enabling the determination of complementary parameters in HM and urine samples are being developed and benchmarked within the NUTRISHIELD study.

The Student's t-test, Wilcoxon ranksum test or Χ² test with α = 0.05 will be used for between-group comparisons and fold changes (FC) will be calculated as the ratio of means or medians between groups in accordance to the underlying distribution of the data. Pearson's linear or Spearman rank correlation coefficients will be determined between continuous variables. If necessary, partial correlation coefficients adjusting for known confounding factors will be computed. The false discovery rate (FDR) from the p-values of multiple-hypothesis testing will be estimated using the Benjamini and Hochberg procedure (86) and adjusted p-values <0.05 will be considered statistically significant.

One of NUTRISHIELD secondary objectives is the development of a personalized nutrition system based on measured biomarkers. The system is to be made available to medical personnel, as well as mothers, to improve infant-mother dyads' health and wellbeing. The results and the associations revealed from the study will be used to train a personalized nutritional algorithm with data obtained from clinical settings. This will be used to build and validate the NUTRISHIELD platform, describing how a theoretical framework designed and fed by the patients' data translates to clinical practice. Toward that goal, a comprehensive data integration system has been developed as the Clinical Trial App (CTA). The CTA is a system intended for personalized medicine/nutrition, implementing the following main features: (i) data acquisition from questionnaires, (ii) data acquisition from laboratory biomarker analysis, (iii) processing data to train machine learning algorithms, and (iv) deploying trained algorithms to produce reports for final users.

Figure 3 shows the data flow and functionalities of the CTA. Questionnaires data are collected by medical staff directly using the phone app. Alternatively, questionnaires that have been collected on paper or on spreadsheets, can be uploaded as well. All samples can be identified with a unique QR code and, at the moment of collection, the phone app can be used to scan the QR code and input sample data. This way the samples can travel between labs with the QR code as only identification. The QR code can be scanned again to associate the results with the original sample when uploading it through the web portal. The collected data is stored on a database and used to feed the personalized recommendation algorithm. This algorithm produces a report and feeds it back to the phone app.

Using the CTA, all study data is consolidated in a structured database, of which an anonymized version can be later stored on public repositories for re-use in further research.