Study design

Recruitment of the 1500 women begins in September 2015 with the aim of completing all assessments by May 2018. The women will be randomized into one of the three treatment groups:

• Control group: standard maternal milk formulation (product A; table 1) which provides a minimum of 2 mg GA per serving and 4 mg GA per day.

• Intervention group: CML-enhanced maternal milk (product B; table 1) formulated to provide a minimum of 4 mg GA per serving and 8 mg GA per day. CML-enhanced maternal milk (Anmum maternal milk, Fonterra Co-operative Limited, New Zealand) contains Beta Serum (Nuelipid supplied by Fonterra Co-operative Limited), a byproduct from the manufacturing of anhydrous milk fat and a source of MFGM complex lipids).

• Reference group: no maternal milk with standard obstetric care including prenatal folic acid supplementations.

All mothers and infants will be followed up until 12 months postdelivery.

Study enrolment will take place at the First Affiliated Hospital of Chongqing Medical University (FCQMU) and Chongqing Health Centre for Women and Children (CHC). After delivery, infant follow-up will take place at the CHC. In total, there are five clinic visits (see tables 2 and 3)—three will take place during pregnancy (11–34 weeks, at both the FCQMU and CHC) and two will take place in the infant period (6 weeks and 12 months, exclusively at CHC).

Recruitment

Recruitment posters promoting the study will be placed in the maternity clinics of two centres (FCQMU and CHC). Online advertisements are also being is produced and uploaded onto various social media platforms in China (eg, Wechat, Weibo and QQ).

Women who are 11–14 weeks pregnant attending their first prenatal visit at the FCQMU and CHC will be approached directly about participating. A qualified research nurse will determine whether the woman is eligible, according to the inclusion and exclusion criteria. If eligible, the research nurse will provide a detailed overview of the study, emphasising that declining to participate will have no negative impact on subsequent prenatal care. All information will be communicated in Mandarin, as this is the main language spoken in this region of China.

If the woman agrees to participate, written informed consent will be obtained, permitting us to enrol both her and her offspring into the study. This consent will explain that samples will be collected primarily for use in the current study, but which may also be used in the future for commercial purposes or for research unrelated to this study. Women will be randomised into one of the groups with a unique trial identity code number and allocated to a specific research nurse for the duration of the study, in an attempt to minimise the loss to follow-up rate.

Inclusion and exclusion criteria

Women who are 11–14 weeks pregnant will be recruited from maternity clinics. Further inclusion criteria include: gravidas aged 20–40 years, singleton pregnancy, living in Chongqing municipality and able to provide written informed consent. Exclusion criteria are: previous pregnancy with complications which resulted in delivery before 32 weeks, allergy to milk and severe milk aversion or lactose intolerance.

Withdrawal

All participants are free to withdraw from the study at any time, for any reason, and without any impact on future medical care. If a participant is withdrawn before completing the study, the date of withdrawal and the reason will be entered into the case report form (CRF). All previous collected data (including measures and samples) will remain in storage for later analysis. If a participant does not return for a scheduled visit, every effort (eg, phone call, WeChat and QQ) will be made to contact them. If the participant refuses to attend, she will be considered as having dropped out. If the participant withdraws from the study and withdraws consent, no intervention will be performed and no additional data will be collected.

Randomisation and allocation concealment

At clinic visit 1, all eligible participants will be randomised via the MedSciNet AB (Stockholm, Sweden) Interactive Web Response System, to either the control, intervention or reference group. The system will assign each participant to a unique trial number, consistent with either one of the treatment groups or the reference group. Randomisation will ensure equal allocation of participants to each of the groups.

The randomisation sequence for the treatment groups will be concealed until interventions are all assigned and recruitment, data collection and data cleaning are complete. The central list that details the numbered cans of product A or product B will be produced by Fonterra Co-operative Group Ltd, New Zealand.

Intervention and comparison

The intervention period begins at the time of recruitment (11–14 weeks of gestation) and ends when the baby is born, allowing the impact of maternal and fetal nutrition during gestation, a sensitive window for brain development, to be investigated.

Pregnant women will be assigned to either a reference group, or to one of two intervention groups, milk powder of product A or product B. The only difference between products A and B, both reduced-fat formulations, is the amount of CML contained in each, with product B having a restored level of CML to that of full-fat milk (table 1). Both treatment groups are instructed to consume 37.5 g of the milk powder twice a day, with clear instructions written on the can of milk powder. These instructions are verbally communicated by the research nurse. The reference group (as well as both treatment groups) will receive standard pregnancy care (including regular monitoring and supplementation with 400 µg of folic acid, based on general dietary recommendation for pregnancy).32 To improve study adherence, participants in the intervention groups will be asked to return, at each clinic visit, the product cans along with the forms provided to document daily consumption of the product. Furthermore, we will utilise popular Chinese social media platforms to provide weekly pregnancy health and nutrition suggestions, specific to pregnancy, and also to facilitate one-to-one communication between nurses and participants. Finally, reminder messages will be sent 3 days prior to each visit.

Follow-up will continue until the child is 12 months of age. This age was chosen as it enables the effect of an improved maternal nutrition during pregnancy on infant cognitive development, to be assessed. Cognitive development will be assessed via the Bayley Scales of Infant Development I (BSID-I). A further consideration in selecting the follow-up period was that after this point in infancy, it is thought that other influences affecting infant development (eg, variability of their own diet and environment) may potentially exert large effects on the outcome.

Blinding

Participants will be allocated to either product A (control), product B (intervention) or to the reference group. If the participant is randomised to one of the treatment groups, they wiil be blined to which of the two groups they have been allocated to. Investigators and the specific research nurse allocated to each enrolled woman will be aware of which participants have been randomised to standard care (vs Products A and B). However, they will remain blinded to the product A/B allocation from the time of randomisation until either the subject is unblinded or the database is unlocked. As for any other assessments (excluding the food frequency questionnaire and 24-hour food recall), staff will be completely blinded.

Measures and samples

The primary maternal outcome is serum complex lipid levels (eg, GA determined by ultra performance liquid chromatography–tandem mass spectrometer) in maternal blood. The secondary maternal outcomes are maternal nutritional status (eg, blood routine, micronutrient status in blood), clinical indicators of maternal health (eg, blood pressure, glucose tolerance), gestational weight gain and outcomes of pregnancy, including complications (eg, gestational diabetes mellitus, gestational hypertension). The primary offspring outcomes are infant cognitive development (assessed by BSID-I) and general infant health (infant growth, eg, height, weight and skinfold thickness). Secondary offspring outcomes include fetal size, growth and outcomes at birth. Additional samples include those for subsequent epigenetic analysis (eg, buccal smears of both mother and infant is taken for bisulfite sequencing), will be taken (tables 2 and 3).

Quality control

Source documents shall be filed at the site and may include, but are not limited to consent forms, current medical records, laboratory results and pharmacy records. The CRF will serve as source for some of these data. Only trial staff listed on the Site Responsibility Delegation Log will have access to trial documentation other than the regulatory requirements listed below. The CRF and all source documents, including progress notes and copies of laboratory and medical test results, will be made available at any time for review by the sponsor’s designed and inspection by relevant regulatory authorities.

All paper forms shall be filled in using gel pen. Errors shall be crossed out but not obliterated and the correction inserted, initialled and dated. The recruiting clinician shall sign a declaration ensuring accuracy of data recorded in the CRF. CRFs will be held securely in a locked cabinet in a locked room. Real-time electronic data will be created as soon as local data is imported into the database. Blood and/or serum samples will be collected and prepared by employees of the institution and stored in a regulated −80°C freezer.

Data management

At the time of randomization, each participant will be assigned a unique trial identity code number. The number will be for use on the CRF, other trial documents and the password-protected electronic database (MedSciNet). The documents and database will also incorporate their initials and date of birth. The investigators will also keep a screening log with details of all patients screened for the trial (including those who were and were not subsequently randomised) with their name, date of birth, local hospital number and trial number.

The investigators and institutions will permit monitoring, audits, institutional ethics committee/institutional review board review, and regulatory inspections, providing direct access to source data/documents. CRFs will otherwise be restricted to those personnel approved by the Principal Investigator and recorded on the ‘Site Responsibility Delegation Log’.

Statistical methods

On the basis of previously reported serum GA measures,30 33 a sample size of 500 participants in each group is sufficient to detect a difference of 25% SDs, with a study power of 80%, and at a significance level of 5%. The loss to follow-up rate is expected to be 20%. This sample size enables adjustment for three pair-wise comparisons.

Treatment evaluation will be performed on the principle of intention to treat. Statistical analyses will be conducted using SAS V.9.3 software (SAS Institute, Cary, North Carolina, USA). All statistical tests will be two-sided, with significance at p<0.05. Baseline characteristics of all randomised participants will be first summarised for each group using descriptive statistics. Continuous variables will be reported as numbers of observed and missing values, means and SD or medians and IQR, depending on their distributions. Categorical variables will be described as frequencies and percentages.

Analysis of covariance regression models will be used to evaluate the main treatment effects on the primary outcome at each scheduled visit, adjusting for important baseline confounding variables (if any). Model-estimated means will be presented for each group, and their differences will be tested. A repeated measures analysis will be conducted as a secondary approach in order to investigate changes over time. This will utilise participant data collected at the multiple visits. Sensitivity analyses will be considered on the primary outcome with missing data, using multiple imputations and/or other prespecified imputation methods when deemed appropriate (ie, data missing at random). A similar approach will be applied to other continuous secondary outcomes. Generalised linear models will be used for categorical outcomes, using specific link functions.

Data monitoring

A trial data safety and monitoring committee has been convened for this study. All members are independent of the study team. This committee will oversee all ethical and safety issues in accordance with current regulations and the Terms of Reference for the group. The Sponsor reserves the right to discontinue this trial at any time for failure to meet expected enrolment goals, in the interests of safety, or any other administrative reasons. The Sponsor shall take advice from the data and safety monitoring committee as appropriate in making this decision.

Potential complications/side effects

CML as a source of GA and phospholipids are a natural component of milk and found in the MFGM surrounding milk fat globules. The milk powder and other ingredients used in the formulation of our study products have been studied previously in both animal models and human studies (e.g. in infants30 and in non-pregnant and pregnant women in New Zealand,34 Singapore35 and Indonesia36) and are now commercially available. There is no known risk to human participants from any of these ingredients.

For the purpose of this study, an ‘adverse event’ (AE) includes any condition detected or diagnosed after the intervention has been administered and has a possible, probable or definite causal relationship with the intervention. A ‘serious adverse event’ is an AE as defined above that occurs after having received the intervention, and which results in a life-threatening outcome, disability, incapacity or death. All AEs will be assessed for seriousness by the treating physician. Those events that are considered to be serious will also be assessed in order to identify causality. All serious AEs will be recorded and monitored until either resolution, stabilisation or until it has been shown that the study intervention is not the cause. They will also be recorded and reported to the relevant authority and research ethics committee as part of the annual reports. Serious adverse reactions will also be reported to the sponsor immediately.

Auditing

Fonterra Co-operative Group Ltd, New Zealand, or its delegate will carry out regular monitoring and auditing to ensure that all procedures are being adhered to, subject information and all data collection systems are robust, there is appropriate recording of and responses to adverse events and furthermore, that sample handling, processing and storage are correct according to the standard operating procedures.

Confidentiality

Individual participant medical information obtained as a result of this study is considered confidential and disclosure to third parties is prohibited. Participant confidentiality will be ensured by assigning each subject a unique study number throughout the study. All computer entry and networking programmes will be done with coded numbers only. Medical information may be given to the participant’s medical team and all appropriate medical personnel responsible for the participant’s welfare. Clinical information will not otherwise be released without written permission of the subject. Participants will not be identifiable by name or personal details on any report forms or published material.

Access to data

After data collection is complete, the database will be locked, after which time no amendments will be possible. Thereafter, data will only be available to the investigators for statistical analysis. Data will, however, be made available for inspection on request by the participating physicians, the research ethics committee and the regulatory authorities.

Protocol amendments

Any protocol amendments will be reported, reviewed and initially approved by the study investigators and sponsor. These amendments will then be submitted to the Ethics Committee of Chongqing Medical University for approval. Participants will be informed of any protocol amendments. If a participant becomes dissatisfied with any protocol amendment, they are free to withdraw from the study at any time and without any impact on future medical care.