Background: Children with acute malnutrition have immature gut microbial communities compared to age-matched children with healthy growth that can not be repaired by existing therapeutic foods (Subramanian et al., Nature. 2014). Hence, investigators' recent work in Bangladesh has focused on developing Microbiota-Directed Complementary Foods (MDCF) containing locally-available food ingredients, that repair the gut microbiota of children with acute malnutrition to a configuration that resembles that of healthy children living in the same urban community (Raman et al., Science 2019; Gherig et al., Science 2019).

The investigators recently completed a randomized, controlled proof-of-concept (POC) study of current lead microbiota-directed complementary food, MDCF-2 compared to a standard ready-to-use supplementary food (RUSF), in Bangladeshi children with moderate acute malnutrition (MAM)(Chen et al., N Engl J Med 2021). Children who received MDCF-2 for 3 months exhibited significantly greater repair of their gut microbial communities and faster rates of ponderal growth compared to those treated with RUSF (a formulation that was not designed based on knowledge of its effects on the gut microbiota). The superior effect of MDCF-2 on gut microbiota repair and weight gain was even more notable as the RUSF was significantly more energy dense than MDCF-2.

In the aforementioned POC study, MDCF-2 was prepared fresh daily in icddr,b field kitchens prior to distribution and supervised administration to study participants. The lack of a shelf-stable, bio-equivalent formulation of MDCF-2 limits the ability to perform larger studies in Bangladesh as well as in other geographic settings. This pre-POC study in Bangladeshi children with MAM will assess the bioequivalence of MDCF prototypes that the investigators have developed with the potential for improved storage stability compared to current MDCF-2, using the degree of microbiota repair after 4-weeks of treatment as the primary outcome.

Objective: To develop a scalable, shelf-stable formulation that is bioequivalent to MDCF-2 with respect to microbiota repair in 8-12 month-old Bangladeshi children with MAM after 4-weeks of treatment.

Methods: A 5-arm, randomized single-blind pre-POC study will be conducted in 8-12-month-old Bangladeshi children with MAM to compare the efficacy of alternative MDCF formulations in repairing their gut microbiomes compared to the repair produced by the current kitchen prepared MDCF-2 formulation.

Arm 1 - Reference control: kitchen-prepared MDCF-2

Arm 2 - Ready-to-use supplementary food

Arm 3 - Individually packaged, pre-measured sachets of MDCF-2 ingredients, combined and reconstituted in the home setting prior to consumption.

Arm 4 - MDCF-2 shelf-stable foil pouch formulation with green banana powder.

Arm 5 - MDCF shelf-stable foil pouch formulation with sweet potato instead of green banana.

Study site: The study will be conducted among residents of an under slum community in Baoniabad of Mirpur, one of the 21 administrative units of the nation's capital, Dhaka. Mirpur is selected as the study site because it is inhabited by poor and lower middle-class families, residential and sanitary conditions are typical of any congested urban settlement. Additionally, the investigators have established several field sites in the area and they have ongoing research activities in the community.

Enrollment: A trained Field Research Assistant (FRA) will explain the study in detail, answer any questions from the parent(s), and invite the parent(s) to enrol the child in the study. If the parent(s) are interested to volunteer in the study, the designated staff will proceed to screening and consenting. Screening will consist of a review of the inclusion and exclusion criteria listed above. If the subject is eligible to participate, the process will proceed to consenting consisting of a thorough review of the written consent form in a manner appropriate for the child's parents' literacy level. Prior to signing the consent form, the parents of the participants will have an opportunity to ask any questions about the study. If the FRA determines that participants have demonstrated adequate comprehension of the study, the consent form will be signed by the FRA and the child's parent(s). If the parent(s) are not sufficiently literate to read and/or sign the consent form, consenting and a thumbprint signature will be obtained in the presence of a witness who is not associated with the study. The child's parent(s) will be provided with a copy of the signed consent form.

Children will undergo thorough clinical assessment and laboratory tests (if needed) to exclude any organic diseases or any other causes of secondary malnutrition. Detailed inclusion/exclusion criteria are provided separately below.

At the beginning of the study, information will be collected on the demographic characteristics of the participants (family income, standard/type of housing, family structure, parental education, etc.). FRAs will collect anthropometry data by recording the child's weight using a digital scale with 2 g precision (Seca, model 728, Germany), length (using infantometer, Seca, model 416, Germany), and mid-upper arm circumference (MUAC) to the nearest millimeter (using a non-stretch tape). Collection of anthropometric data will subsequently occur at weekly intervals throughout the study.

Study Design: The study will be 8 weeks in duration and comprise 5 arms, with a minimum of 23 children per arm required to complete the study. Children will be randomly assigned to one of the five arms using a computer generated randomization scheme. Efforts will be made to balance arms with respect to gender. Mothers/primary caregivers will be be advised to maintain their child's current dietary and breastfeeding practices throughout the study. Following a 2-week 'run-in' phase with weekly fecal sample collection and a blood draw, treatment will be initiated. The treatment phase will be for 4 weeks, during which time each test product will be administered at the test center twice daily, separated by at least 6 hours. The amount of each serving actually ingested will be calculated by subtracting the leftover amount from the offered amount. Pre-weighed napkins will be provided; any food that is regurgitated, vomited or spilled will be swabbed, weighed and subtracted from the amount offered. Fecal samples will be collected weekly using established SOPs for rapid cryopreservation. A second blood draw will be performed at the end of the 4-week intervention period. There will be a 2- week post intervention phase during which 2 additional fecal samples will be collected (at the end of weeks 7 and 8) to assess changes in the microbiota that occur after cessation of treatment.

Children will be monitored daily by trained FRAs for any side effects/adverse events (e.g. rash, urticaria due to food allergy, or any significant changes in clinical status). If any adverse events are observed, children will be treated using appropriate management practices at Dhaka Hospital.

Fecal sample collection: Fecal samples will be collected weekly from each study participant during the course of the 8-week study, including during the 2 weeks prior to treatment, the four weeks of treatment, and the two weeks following cessation of treatment (11 samples/ child). Fecal samples will be used for culture-independent analysis of microbial composition using existing SOPs, and for subsequent culture-based initiatives to characterize growth promoting bacterial taxa (Raman et al., Science 2019, 365, eaau 4735; Gehrig et al., Science 2019; 365, eaau4732).

Blood sample collection: 5 mL of blood will be collected from each child immediately prior to initiating treatment, and just after the 4-week period of treatment is complete. Plasma will be prepared immediately and frozen at -80C prior to proteomics analysis as previously described (Gehrig et al., Science 2019, 365, eaau4732; Chen et al., N Engl J Med 2021; 384:1517-1528). Additionally, status of different blood biomarkers (hormonal regulators of appetite and satiety, micronutrient status, and systemic inflammation) will also be assessed using the samples collected before and after the intervention

Sample size estimate: The estimate is informed by fecal 16S rDNA datasets generated in a recent POC study in 12 to 18-month-old Bangladeshi children with moderate acute malnutrition (MAM) who had been treated with the ("reference") kitchen-prepared MDCF-2 formulation. This formulation repaired the microbiota of these children and produced an improvement in the rate of ponderal growth that was superior to that produced by the control Ready-to-Use Supplementary Food (RUSF; Chen et al., 2021). In this POC study, the investigators identified 209 bacterial taxa (Amplicon Sequence Variants; ASVs) that were present above 5 counts in at least 5% of the 939 fecal samples analyzed from 118 participants (Chen et al., 2021).

The investigators calculated 95% confidence intervals (CI; mean ± 1.96 standard error) for the change (delta) in abundance of each of the 209 bacterial taxa ASVs between day 0 (pre-intervention) and 4 weeks of intervention for the 59 children who completed treatment in the MDCF-2 group of the POC study. The investigators used a bootstrapping approach with ASV data from 5000 subsets of children from the MDCF-2 group and 5000 subsets of children from the RUSF group and identified 121 ASVs whose mean changes in abundance fell within the 95% CI of the mean changes for the MDCF-2 participants, but not in children that had received RUSF, with 80% power using a subset (sample) size of 20 children per group ("MDCF-2-responsive ASVs"). Therefore, we will aim to have a minimum of 20 children per arm complete this pre-POC study. Taking into account potential drop-outs (15%), a minimum of 23 children will be recruited per arm.

Analysis plan: To assess the comparability of microbiota repair elicited by alternative/shelf-stable test formulations of MDCF-2 and the reference kitchen-prepared formulation described in Chen et al (2021), the investigators sequence PCR amplicons generated from bacterial 16S rDNA genes present in DNA isolated from fecal samples collected from participants in the different study arms immediately prior to intervention (day 0) and at the end of intervention (day 28). ASV data will be processed using DADA2 (Callahan et al., 2016) and filtered as described in Chen et al. (2021). Using the same approach used for the POC study above, the investigators will calculate 95% confidence intervals of the mean change in abundance of each of the MDCF-2-responsive ASVs identified in this study, between day 0 and day 28, in children who received the reference MDCF-2 formulation, as well as the mean change for each of these ASVs in participants receiving the test formulations.

A test formulation will be considered comparable in efficacy to the reference kitchen-prepared MDCF-2 if >70% of the MDCF-2-responsive ASVs identified in this study have mean changes in abundance (day 0 to day 28) for the test formulation that fall within the 95% CI of their mean changes in abundance in the reference MDCF-2 arm over the same time period. Another embodiment of the analysis of microbial community repair elicited by the reference MDCF-2 and test formulations will involve a comparison of their respective effects after 4 weeks of treatment on the abundances of metagenome-assembled genomes (MAGs) identified by shotgun sequencing of the same fecal DNA samples used to produce 16S rDNA amplicons.

Exploratory outcomes will include the effects of the test formulations on features of the plasma and fecal proteomes, including MDCF-2-responsive proteins that are associated with various facets of healthy growth, CNS development and immune function, using approaches described in Chen et al., 2021.

Inclusion Criteria:

• Parent(s) willing to sign consent form
• Child aged 8-12 months
• WLZ score <-2 to -3 without bilateral pedal edema at the time of randomization
• Parent(s) willing to bring the child to the feeding centre according to the pre-defined schedule

Exclusion Criteria: Meeting any of the following criteria will exclude a subject from study participation -

1. Medical conditions: Malnourished children with complications requiring acute phase treatment in a hospital, children with tuberculosis (diagnosis based on WHO 2014 guidelines which have been incorporated in the national TB control guidelines of Bangladesh). The guidelines depend upon the following five diagnostic principles (three out of five should be positive):

   • Specific symptoms of TB
   • Specific signs of TB
   • Chest X-ray
   • Mantoux test
   • History of contact

2. Any congenital/acquired disorder affecting growth, i.e., known case of trisomy-21 or cerebral palsy; children on an exclusion diet for the treatment of persistent diarrhea; having known history of soy, peanut or milk protein allergy

3. Severe anemia (< 8 mg/dl)

4. Antibiotic use (within last 15 days before the onset of intervention)

5. Ongoing maternal antibiotic usage for breastfeeding infants

6. Receiving concurrent treatment for another condition