Effects of Fecal Microbiome Transfer in Adolescents With Obesity

Key Points Question Will fecal microbiome transfer (FMT) lead to weight loss among adolescents with obesity? Findings In this randomized clinical trial of 87 adolescents with obesity, FMT alone did not lead to weight loss at 6 weeks. FMT alone was associated with a reduction in android-to-gynoid-fat ratio sustained for at least 26 weeks, particularly in female adolescents; changes in the overall gut microbiome composition; and a resolution of metabolic syndrome by 26 weeks in participants who had this undiagnosed condition at baseline. Meaning FMT alone is not an effective treatment for weight loss but may reduce visceral adiposity and improve health.

We will recruit 8 donors (4 males and 4 females), as recipients will only receive gut microbiome 95 from donors of the same sex. This is to enhance microbial variability and standardise the 96 treatment via gut microbiome transfer. Donors will be selected based on strict inclusion criteria 97 (

102
Eligible donors will be identified by word of mouth, the internal email system at the University of 103 Auckland, and social media networks. Potential donors will be given a detailed information sheet 104 about the study that includes a consent form. 105

106
To eliminate the risks of transmission of infectious diseases we will use screening procedures 107 equivalent to those used for blood donation in New Zealand 4 , and also screen donors for 108 potential faecal pathogens or multidrug-resistant organisms. As part of this regimen, all potential Given evidence that irritable bowel syndrome (IBS) may be related to the gut microbiome, it is 118 important to exclude potential donors who may have IBS. The Rome criteria are an accepted 119 clinical tool to identify individuals with IBS, but they are relatively insensitive so that strict 120 adherence to those criteria would potentially allow for individuals with mild IBS to donate 6 . 121 Therefore, we will screen for IBS using a conservative modification of the Rome criteria, where 122 we define a positive screen as having 3 or more episodes of abdominal pain per month as 123 described in part I of the criteria, as well as an additional symptom as defined in part II 7 . 124 125 Each donor is expected to produce a wet stool sample weighing 100-150 g. Our preliminary 126 laboratory data indicate that an average stool sample from a donor will generate sufficient gut 127 microbiome material for two same-sex recipients. Stool samples will be collected and 128 immediately processed for encapsulation. Capsules from each sample will be individually coded, 129 so that each recipient will receive an equal number of capsules (n=7) from each of the four same 130 sex donors. 131

Participants (recipients) 132 133
We aim to recruit 80 obese adolescents according to the inclusion and exclusion criteria 134 described in Table 2  Use of regular medications that may influence weight, metabolism, or the gut microbiome (including oral oestrogencontaining contraceptives, antidepressants, glucose-lowering drugs, diet drugs, as well as inhaled, topical, or oral steroids)  Consumption of probiotics  Type 1 diabetes, type 2 diabetes, or monogenic diabetes  Chronic diseases that could affect the primary outcome (other than obesity-related conditions)  Food allergies  Allergy to macrogol (active ingredient in the bowel preparation product)  Allergy to any over-the-counter medication  No antibiotic usage for three months prior to trial treatment 138 Eligible recipients will be recruited via social media, word of mouth, and paediatric endocrinology 139 clinics in Auckland. Potential recipients and caregivers will be given a detailed information sheet 140 about the study that includes a consent form. Consent will be obtained from recipients if they are 141 aged ≥16 years and from their parents if aged <16 years. Younger recipients will also be asked 142 to sign an assent form. All consent and/or assent will be obtained by the researchers prior to the recorded and kept in a secure folder and only accessible to the researchers, in order to protect 145 their confidentiality. 146 147

Randomisation, allocation, and blinding 148
149 Eligible participants will be randomised in a 1:1 ratio to either treatment or placebo group, 150 stratified by sex, using block randomisation with variable block sizes of 2 and 4 8 . Randomisation 151 sequences will be computer generated, and overseen by the biostatistician. Researchers and 152 participants will be blinded to capsule contents, both of which (placebo and gut microbiome) look 153 identical (white). 154

155
There are three steps in the blinding and allocation process. First, the independent research 156 nurse allocates the recipient to group A or B using the randomisation sequence. Second, the 157 placebo and treatment capsule packs each have a unique code (assigned by the technician who 158 encapsulated them). Lastly, the independent research nurse allocates the pack according to the 159 unique code associated with the randomisation sequence. 160

161
To maintain the integrity of the trial evaluation, statistical analyses will be performed at the 162 completion of the study after all trial data have been collected. The biostatistician will be blinded 163 to treatment allocation throughout the trial, as well as study investigators. Recipients will be 164 asked if they are able to identify the contents of capsules taken (i.e. placebo or gut microbiome) 165 at 6 weeks and 26 weeks. The effectiveness of treatment blinding will be assessed using the 166 Bang's blinding index 9 . Blinding success will be determined by the thresholds of Moroz et al. 10 : 167 unblinded (BBI ≥ 0.2); random guesses (−0.2 < BBI < 0.2); or opposite guesses (BBI ≤ −0.2). 168 169 Recipients will be unblinded in the case of any serious adverse events. These include on-going 170 gastrointestinal bleeding, severe vomiting and/or diarrhoea, treatment related systemic infection, 171 and treatment related severe allergic reaction, coma, collapse and death. Unblinding will be 172 done by an independent researcher who did not have any prior contact with the recipient, who 173 will be able to determine the individual's treatment allocation. 174 175

Study intervention 176
each recipient in the placebo group will ingest saline capsules, while those in the treatment 183 group will receive gut microbiome capsules. Each recipient will receive a total of 28 capsules 184 (approximately 14 ml of frozen microbial suspension or saline) administered over two 185 consecutive mornings under direct supervision from research staff, specifically 16 capsules in 186 the first morning and 12 capsules in the second morning. Recipients will be advised not to 187 change their diet, physical activity, and behaviour during the trial. A study with 32 recipients per group will have 80% power at 5% significance level (two-sided) to 214 detect a group difference of 0.19 in BMI SDS at 6 weeks after gut microbiome transfer. To 215 The timing of scheduled assessments is shown in Table 3. 229 230   Table 3. Timing of individual assessments in the Gut Bugs Trial. 231 Baseline 6 weeks 12 weeks 26 weeks

Clinic
Medical history and exam

Withdrawals 237 238
Any participant who withdraws from the trial will be contacted by the clinical team. Attempts will 239 be made to offer the participants the chance to provide crucial clinical information via follow up 240 text messages or emails. All reasons for withdrawal will be recorded in secure databases, and 241 will also be presented to the safety monitoring committee. All clinical assessments will be performed in accordance with Standard Operating Procedures 248 and documented in relevant worksheets (i.e. case report forms -CRF) (

253
These are checked for completeness of data by the clinical research team. All questionnaires 254 are checked for completeness. All paper CRFs, reports, questionnaires will be scanned and 255 saved to the Gut Bugs Trial shared drive in appropriately named folders. 256 257 All questionnaires (except dietary ones) will be scored using specifically designed Microsoft 258 Excel Spreadsheets for each questionnaire. Data will be entered once by the data entry person 259 and validated by another data entry person. A final check will be carried out by the clinical 260 research fellow.

262
All clinical data are to be entered into REDCap in accordance with the Data Management 263 Standard Operating Procedure. Data will be entered once by the data entry person, validated by 264 another data entry person, and finally checked and locked by the clinical research fellow. Below 265 are the step-by-step process for data checking, which will be reflected by the REDCap Status 266 stated below: 267 268 Incomplete Data entry is in progress and has been entered by first data entry person. Any 269 missing information should be highlighted and efforts to obtain that information 270 will be done by the clinical research team. 271 272

Unverified
Data entry has been checked by second data entry person who will change the 273 status from Incomplete to Unverified. 274 275

Complete
All available data has been entered and verified. 276 Data is ready for cleaning and monitoring. 277 The data monitoring team will change status to Complete. 278

279
All complete data will be regularly checked by the trial statistician to identify potential data entry 280 errors based on clinical plausibility. Dubious data points will be followed up and the respective 281 All data to be imported will be checked and cleaned by the trial statistician prior to importing into 293 REDCap. Each variable entered into REDCap will be given a unique name and the completed 294 codebook will be attached as an appendix to this document. Further details of data management 295 are included in the Appendix 1 under Data Management Standard Operating Procedure. 296 297 3.5 Safety monitoring and evaluation 299 300 An independent safety monitoring committee has been established to for the duration of the trial. 301 Participants' data will be monitored by the research team and the safety committee throughout 302 the study for any adverse events, in particular gastrointestinal symptoms and possible allergies. 303 Any possible adverse events are asked and recorded at 24 hours, 48 hours, 1 week, 3 weeks, 6 304 weeks, 12 weeks, and 26 weeks after the intervention in the adverse events worksheets. All 305 potential adverse events will be recorded in a secure database. Any serious adverse event that 306 is identified will be flagged and highlighted as soon as possible to the committee. If any 307 participant suffers harm as a result of trial participation, they will be eligible to apply for 308 compensation from the Accident Compensation Cooperation (ACC), which is a compulsory 309 insurance cover for personal injury for everyone in New Zealand 310 311

Outcome variables and definitions 312 313
The study's outcome variables are listed in Table 5. 314 315 Total body fat percentage and A/G ratios will be obtained from DXA scan reports performed on 328 the participants at these time points. The android region is defined as the area between the ribs 329 and the pelvis that is totally enclosed by the trunk region. The gynoid region includes the hips 330 and upper thighs and overlaps both the leg and trunk regions 13 . A/G ratio is calculated with the 331 formula android fat mass/ gynoid fat mass 13 . As we will be utilising 2 different types of DXA  Table 6. 346 Markers of inflammation will be assessed via uric acid (umol/L) and high-sensitivity C-reactive 364 protein (hsCRP) (mg/L). The normal reference ranges for uric acid and hsCRP are provided in 365 Clinic resting systolic and diastolic blood pressures will be measured at all assessments using 370 the same oscillometric digital blood pressure monitor (ri-champion ® N; Riester, Jungingen, 371 Germany) with an appropriately-sized cuff on the extended non-dominant arm. All 372 measurements will be recorded on each recipient while seated and after a 5-minute rest. Blood 373 pressure will be measured three times, and the median value calculated. The normal blood 374 pressure readings will be following the European Society of Hypertension guidelines for the 375 management of high blood pressure in children and adolescents 17 . 376

377
In addition, 24-hour ambulatory blood pressure monitoring will be performed at baseline and at 6 378 weeks, using an oscillometric device (Spacelabs OnTrak; Spacelabs Medical Inc, Redmond, 379 Washington, USA) on the non-dominant arm. Over a 24-hour period, blood pressure will be 380 measured every 20 minutes when the recipients are expected to be awake, and every 30 381 minutes when they are likely to be asleep (based on self-reported information). Recipients will be 382 asked to record the time they go to bed and the time they wake up over the period of monitoring, 383 so that waking and sleeping times can be more accurately identified. The average overall blood 384 pressure, awake blood pressure and the sleep blood pressure readings will be obtained. An 385 adequate blood pressure report needs at least 14 day readings and 7 night readings 18 . The 386 normal blood pressure readings will be following the European Society of Hypertension 387 guidelines for the management of high blood pressure in children and adolescents 17 . 388 389   The total score is a sum of all 3 sections. If more than 50% of the items are missing, the scores 427 should not be computed. 428 Using a small spatula, samples will be collected from three different areas of the stool (proximal, 447 middle, and distal) and inserted into specimen containers (Onelink). The specimen containers 448 will be immediately placed on ice and taken to the laboratory where they will be frozen and 449 stored at -80°C. DNA and RNA extraction will be completed within 5 days of donation. Time to 450 processing will be recorded. 451 452 Note that we will advise participants to try not to have a bowel movement in the morning prior to 453 their visit, having it in the clinic instead. For those participants who are unable to produce a stool 454 sample during their visit, they will be provided with a stool collection kit to take home and 455 detailed instructions on how to collect the stool sample. This kit is made up of: i) instructions on 456 how to use the stool collection kit; ii) specimen container; and iii) bedpan liner. Once the stool 457 has been collected in the home environment, the specimen container it should be immediately 458 placed into their home freezer, and kept there until it is delivered to the research team. 459 460 All extractions will be performed using Qiagen-AllPrep DNA/RNA mini kit®, due to variation in 461 extraction efficiencies with the different kits 36 . However, once the DNA or RNA is extracted and 462 archived, we will have a relatively stable record of the composition and activity of the flora. 463 464 Frozen faeces (~200 mg; weights will be recorded) will be subsampled from original faecal 465 samples. All DNA and RNA isolations will be performed in a disinfected class II hood at room 466 temperature. Briefly, stool samples will be incubated (10 min, room temperature) with vortexing 467 (30 sec every 2 minutes) and treated with RLT Plus buffer (1.2mL; Qiagen) and 12µL beta-468 mercaptoethanol (Sigma-Aldrich). Acid-washed glass beads [1 ml; ≤106 μm (-140 U.S. sieve) 469 (Sigma-Aldrich)] will be added to each sample and vortexed (10 min) on a TissueLyzer II and centrifuged (9000 rpm, 2 min, room temperature). The eluent will be added to an AllPrep 472 DNA (Qiagen) spin column and centrifuged (30 sec, 14000 rpm, room temperature). The eluent 473 and AllPrep DNA spin columns will be used for RNA and DNA extraction, respectively, according 474 to the manufacturer's instructions. Finally, DNA and RNA will be eluted with EB buffer and 475

Definitions of abnormal outcomes 391
RNase-free water, respectively, and aliquots stored at -80°C for downstream mixed omics 476 analysis. 477 478 A series of blank samples (sterile saline) will be extracted in parallel to sample extractions to 479 enable contamination testing. We will also extract ZymoBIOMICS™ Microbial Community 480 Standard I (Even, Cellular Mix; Catalog #D6300) to determine potential bias in the extraction 481 process. 482 483 For 16S amplicon sequencing, library preparation will be performed using an Illumina platform by 484 a commercial provider (to be determined) using standard protocols for the SV3-4 region. 485 Shotgun metagenomics sequencing will be performed by a commercial provider (to be 486 determined). 487 488 3.7 Diet, physical activity, and socioeconomic status 489 490 We will also be collecting data throughout the trial on lifestyle parameters that could possibly 491 affect treatment outcomes, namely physical activity levels, dietary intake and socioeconomic 492 status. Physical activity levels will be assessed via the IPAQ and ASAQ scores and the dietary 493 intake will be assessed via the NZAFFQ questionnaire. 494 moderate-intensity, and vigorous-intensity activities (work-related physical activity, transport-504 related physical activity, domestic and gardening activities and leisure time physical activity). 505 Total score will be a sum of all the domain specific scores. Activity-specific scores require 506 summation of all the scores for the specific type of activity across domains (total vigorous 507 activity, total moderate activity, total moderate and vigorous activity). We can also generate a

3-day food diary 563 564
This diary will also be used to describe all foods and fluids consumed over three days. 565 Recipients will be asked to describe all foods and fluids consumed in detail including brand 566 names, types of foods (e.g. low fat), and cooking methods. Quantities will be described using 567 standard household measures, as well as the information from food labels (where appropriate). 568 Recipients will be provided with standardized instructions for completing the dietary record by a 569 trained investigator, who will also review individual records with recipients to clarify errors, 570 omissions, questionable entries, or unclear descriptions. These dietary records will be entered 571 into FoodWorks software (v9.0, Xyris Software, Brisbane, Australia) by a trained investigator. 572 573

Socioeconomic status 574 575
This is evaluated via the New Zealand indices of multiple deprivation (IMD) scores 41 . There are 576 seven domains of deprivation; employment, income, crime, housing, health, education; and 577 geographical access. The overall IMD score is the combination of these seven domains. To 578 generate this scores, the participant's residential address is entered into qualtrics survey which 579 will then convert the information provided into the scores. 580 581

582
The CONSORT 2010 guidelines will be followed in reporting the main trial results 1,2 . Data 584 analyses will be performed in SAS v.9.4 (SAS Institute, Cary, NC, USA), SPSS v25 (IBM Corp, 585 Armonk, NY, USA), and/or Minitab v.16 (Pennsylvania State University, State College, PA, 586 USA). All statistical tests will be two-sided at p<0.05, with no adjustments for multiple 587 comparisons. Baseline demographics and clinical characteristics of recipients will be 588 summarised by randomisation group. The distribution of outcome measures will be first 589 evaluated using descriptive statistics. No interim analysis is planned for the trial. 590 591 4.1 Primary outcome analyses 592 593 Treatment evaluation will be performed on the principle of intention to treat (ITT), using data 594 collected from all randomised recipients. 595 596 A linear regression model will be used to assess the treatment effect between two groups on the 597 primary outcome (BMI SDS) at 6 weeks, adjusting for the baseline outcome value and sex (i.e. 598 stratification factor). Model-adjusted estimates and the differences between the two groups will 599 be calculated with 95% confidence intervals. 600 601 Missing data on the primary outcome will be imputed using multiple imputations, which create 602 multiple imputed datasets for the incomplete outcome variable that are analysed using same 603 regression models and combined for one inference. The Markov chain Monte Carlo (MCMC) 604 method will be used to produce the parameter estimates, assuming the data are from a 605 multivariate normal distribution and are missing at random. The SAS procedure, PROC MI, will 606 be used, and we plan to run 30 imputations to allow for both within and between imputation 607 variances. 608 collected from all randomised recipients on previously specified secondary outcomes. 644 Generalised linear mixed models will be used to evaluate the outcomes measured repeatedly 645 over time, using a link function appropriate to the distribution of the outcome variable. The fixed-646 effect model will include the value of the respective outcome at baseline, sex (stratification 647 factor), treatment group, visit, and the latter's interaction with treatment group. A random patient 648 effect will be considered in modelling to take into account the correlation between the data 649 collected from same participant. Model-adjusted group differences will be estimated at each 650 clinical assessment with 95% confidence intervals. The interaction effect between treatment 651 group and visit will be assessed. 652 5 MICROBIOME DATA ANALYSIS 653 654

Initial bioinformatics 655
The metagenomic sequencing data will be processed using bioBakery workflows using docker 657 images available at http://huttenhower.sph.harvard.edu/biobakery_workflows. Briefly, sequence 658 data quality control, including removal of any human reads, will be conducted using kneadData. 659 Taxonomic and functional profiles of the microbiome will be generated using MetaPhlAn v2.6 42 660 and HUMAnN2 43 , respectively. Additionally, strain level taxonomic profiling will be achieved by 661 SNP haplotype based profiling by StrainPhlAn software 44 . Quality control, read filtering, 662 trimming and dereplicatiion, read-pair joining, sequence denoising, chimera removal and 663 sequence table construction of 16S rRNA amplicon sequencing data will be conducted using 664 DADA2 R package 45 . Sequence taxonomies will be assigned using IDTAXA algorithm in 665 DECIPHER R package 46 . 666 667

Metagenomic assembly 668 669
Metaganomic reads of each sample will be assembled into contigs separately using MegaHIT 47 . 670 Open reading frames (ORFs) of the assembled contigs will be predicted using Prodigal 48 671 followed up by clustering the ORFs with >95% identity and 90% coverage into non-redundant 672 gene clusters by CD-HIT 49,50 . The resulting non-redundant gene catalogue will be merged using 673 existing gene catalogues to assure more comprehensive reference gene catalogue 51,52 . Gene 674 abundance will be determined by mapping reads from the metagenomic samples to the gene 675 catalogue using Burrows-Wheeler Aligner (BWA) 53 . The resulting gene abundance profiles will 676 be used to construct microbial pangenomes using MPSminer 54 . Similarity between a pair of 677 metagenomic strains (within the same species but in different samples) will be measured using 678 the percentage of shared genes in the smallest of the two genomes 55 . 679 and functional (metagenomic sequencing) profiles will be compared. Following comparisons for 686 both sexes separately (males and females) and all data as an aggregate (but by controlling sex 687 as a covariate) will be included: 688 1) Compare first post-treatment samples to look for consistent post-treatment 689 differences between the groups.
3) Compare microbial stability (Bray-Curtis dissimilarity and change in any individual 693 taxa) over the treatment between the groups. 694 4) Compare pre-and post-treatment similarities to donors (as an aggregate and 695 individually) between groups. 696

Engraftment analysis 698 699
We will use the SNP haplotypes from StrainPhlAn 44 to investigate microbial engraftment at the 700 strain level. We will first select a sequence similarity threshold for matching donor-recipient 701 strains by using the placebo group as a negative control; no strain transfer (or engraftment) 702 should be observed in placebo group. We will then treat all donor strains with higher than this 703 previously selected sequence similarity in recipient stool as successfully engrafted strains. We 704 will quantify engraftment per participant by counting the number of engrafted strains and by 705 measuring their total abundance in the recipient gut microbiome. We will then correlate these 706 engraftment measures to other clinical data, such as weight loss, quality of life and reported 707 post-treatment adverse effects. 708 The metagenomic assemblies will be used to confirm the results of SNP haplotype based 709 analysis above and to discover engraftment of genomes that are missing or not well-presented 710 in the reference databases. As above, the gene content similarity threshold for matching donor-711 recipient strains will be determined experimentally using the placebo group as a negative 712 control. The metagenomic assemblies will also be used to discover rare genes that are shared 713 between donor-recipient pairs and which may participate in the trophic cascade following the 714 FMT. 715 716

Identification of super-donor behavior 717 718
We will analyse the data to identify any evidence supporting the existence of super-donor(s) in 719 this study using the following tests: 720 1) Measure recipient shifts (in Bray-Curtis dissimilarity and Jaccard Index) towards all 721 donor microbiome profiles separately and compare these shifts using ANOVA.       *Indicates categorical (binary) outcomes, as defined in Table 6 769 [ ] Analysis of potential treatment effects on individual components may not be carried out in the absence of overall differences.