Novel gastrointestinal tools (GI Tools) for evaluating gut functional capacity in adults with environmental enteropathy in Zambia and Zimbabwe: A cross-sectional study protocol

Study aims

Aim 1: Evaluate a novel portfolio of stable isotope-based tools for evaluating carbohydrate and protein digestion and amino acid absorption.

Aim 2: Test a novel non-invasive fluorescence sensor-based tool for real-time evaluation of gut permeability.

Aim 3: Characterise the microbiome and metabolome across well characterised and novel physiological sites (duodenal aspirate, duodenal mucus, duodenal tissue, stool, plasma, urine).

Aim 4: Quantify systemic and intestinal inflammation using established biomarkers of EE, immune cell phenotyping, epithelial PRR expression, and in vitro epithelial barrier models.

Aim 5: Determine if antibiotics, given intravenously, permeate into the gut lumen in sufficient quantities to modulate the microbiota.

All of these will be correlated with histological mucosal morphometry, lactulose permeation, and established biomarkers of EE.

Study strengths and limitations

Study outcomes

The expected outcomes for each aim are shown in Table 1.

Study design

This cross-sectional study will be conducted in adults in Zambia using a range of stable isotope labelled amino acids and fluorescent tracers,^1,30 systemic immune assessments, EE biomarker analysis, microbiome and metabolome analysis, enzyme functional capacity, proteomics and metaproteomic analysis, barrier integrity and function capacity analysis using cell lines, antibiotic reverse permeation, duodenal morphometry, and duodenal expression of PRR to identify potential new tools for measuring gut functional capacity in Zambian adults with EE.

In a sub-study in Zimbabwean adults, we will explore the extent to which antibiotics leak from systemic circulation into the gut, where they may modulate the microbiota.³¹ The study will be carried out in Harare, where sampling of caecal luminal fluid collected at colonoscopy will be used to quantify reverse permeation of intravenously administered benzylpenicillin from the systemic circulation into the gut lumen. Future work will focus on refinement and simplification of these techniques to make them suitable for use in children.

Study population

The initial study will be conducted in 80 adults from a population in Lusaka, Zambia, where our previous studies have demonstrated that EE is virtually ubiquitous in adults and children.³² Additionally, 20 controls living in high socioeconomic status (SES) residential areas (Rhodes Park, Longacres, Kabulonga, Ibex Hill, Chelston, Avondale, Sunningdale, Roma, Chamba Valley, Lilayi, New Kasama and others) within Lusaka will be included. A further 20 patients attending for routine colonoscopy (for diagnostic purposes) in Harare, Zimbabwe will be enrolled for the antibiotic permeation study. As these participants will already be booked for colonoscopy they will be drawn from higher socio-economic groups.

Sample size estimation

As there are no, or very limited, data on these measurements in this population, the study was powered based on previous studies using stable isotope breath tests and permeability studies. Previous data from an enteropathy vs. healthy child cohort and ¹³C-sucrose breath tests³³ has suggested that calculating the cumulative ¹³CO₂ excretion, % relative to the dose, at 90 minutes post-intake best represents the villus integrity and the absorption capacity of only the small intestine. Preliminary data have shown that in a group of adults from Misisi with demonstrable sucrase-isomaltase (SI) expression in biopsies,³⁴ the mean cumulative % of dose excreted as ¹³CO₂ at 90 minutes is 20.29% compared to 16.02% in the group without SI expression, with SD of 5.29% – resulting in an effect size (ES) of 0.81.³⁴ With α = 0.05, power = (1-β) = 0.80 we estimate we need 20 samples per group for achieving 80% power with a one-tailed test. We will recruit 30 participants into the definitive studies, allowing for potential dropouts and incomplete sample sets. To estimate sample size for permeability studies we used data on circulating LPS concentrations in stunted children which were mean 481 EU/ml (SD 408) in Lusaka, compared to controls (mean 192, SD 113). This results in an effect size of 0.96, with α = 0.05 and power = 0.80 and we would require 18 in each group for a two-tailed test. The overall sample size was set at 80 adults to allow for sufficient power to cover a range of studies in adults and children.

Participant recruitment

For adults from Misisi compound, consent will begin with an invitation delivered door-to-door, while social media approaches, like Facebook and/or Instagram will be used for the controls. All household members eligible to participate will be invited to group discussions. The study group discussion meetings will include the following:

Potential participants who are unable to make any of the focus group meetings will be invited to the field clinic to discuss the information sheet and study procedures with a member of the study team. All potential participants will be invited to provide written informed consent in individual interviews with a member of the study team afterwards. They will be invited to ask further questions or express concerns. If happy to proceed, they will be asked to sign or thumbprint the consent form, which will be read out to them in the language of their choice if they are unable to read. Consenting and enrolment will occur at least 48 hours after they have attended a focus group meeting or study explanation visit. Only participants who give consent will undergo screening and enrolment into the study. However, if for any reason during the study, the participant decides to withdraw, they will be allowed to exit the study but will not be replaced. In the event of withdrawal, participants will have a choice whether their samples can be retained or must be destroyed.

Inclusion and exclusion criteria

This study will include and exclude participants following the criteria shown in Table 2.

Screening and initial assessment

Participants will be consented for HIV testing as part of the consent process, before enrolment. Consenting for HIV testing and communication of results will be performed by an experienced and fully trained team of research staff and counsellors. HIV counselling and testing have been performed by this team both as part of standard medical care in the Misisi clinic and as part of the routine research process in our recent studies for over 15 years.^10,35 Likewise, HIV counselling and testing are standard of care in Zimbabwean health care facilities. HIV test results will be given in person and confidence in the clinic of recruitment. Participants found to be HIV seropositive will be allowed to discuss their results and management in more detail. With their consent, they will be referred to HIV specialists for further investigation and management including anti-retroviral therapy per Ministry of Health guidelines and procedures in both Zambia and Zimbabwe. Test results remain confidential and will not be shared with anyone outside of the study team without the participant’s explicit consent.

Anthropometry

All anthropometry measures will be taken by qualified research nurses trained in anthropometry. Weight will be measured using a calibrated scale, standing height using a height scale, and mid-upper arm circumference (MUAC) using a MUAC tape. Grip strength will be measured using a Takei Grip-D dynamometer, waist and hip circumferences using a measuring tape, and lean and fat mass using a BodyStat^® 1500 Impedance instrument (BodyStat^®, Douglas, Isle of Man).

Dietary assessment

A multi-pass 24-hour dietary recall will be used to assess participants on day 2. All participants will be asked about the food they may have had in the last 24 hours which will be documented (Described in detail elsewhere.³⁶ While this 24-hour period may not be representative of the participants’ habitual intake, it was intended primarily to help interpretation of the metabolic profiling and thus would be most useful prior to the collection of samples for metabolic analysis.

Day 1

This will include the final consent interview, administration of a questionnaire by the study nurse, clinical examination, and anthropometry (including weight, height, impedance, and deuterium approaches to body composition). Urine and stool collection devices will also be given.

Day 2

Participants will bring urine and stool samples with them. Before administration of the solutions, a saliva sample, 9 mL of blood will be collected in lithium heparin, 2 mL in EDTA and 4.5 mL in citrated blood collection tubes, respectively, by venepuncture. A drink (prepared as shown in Table 3) of ²H₂O (deuterated water, 1 g ²H₂O per kg body weight) containing ¹³C-spirulina protein (1.25 mg kg⁻¹) and (²H₈)-Phe (0.035 mg kg⁻¹ in 200 mL clean water) will then be given at, or close to, 08:00 followed by breath sample collection using a straw and Exetainer tube (Labco, UK) every 20 mins for the first 1 h and every 30 mins for the remaining 3 h. A saliva sample will be collected at 180 mins, and a urine sample collection at 180 mins (3 h). A second drink of ²H₂O (1 g ²H₂O per kg body weight) will be given at 180 mins (3 h). During these procedures, a multi-pass 24-hour dietary recall will be administered by the study nurse. At the end of the procedures, the participant will be asked to fast after midnight and given stool and urine collection devices in readiness for day 3 procedures.

Day 3

Following an overnight fast, an intravenous cannula will be introduced into both arms and 5 mL of blood will be collected into a lithium heparin tube via cannula. The fingertip probe will be clamped gently onto a fingertip, and continuous fluorescence recordings will begin. An intravenous bolus of ²H₅-phenylalanine (0.6 mmol kg⁻¹) and 5,5,5-²H₃-leucine (1.2 mmol kg⁻¹) in 10 mL normal saline filtered through a 0.2 mm syringe filter (solution C) will be given over 2 mins, and an oral bolus of ¹³C₆-leucine (1.2 mmol kg⁻¹), ¹³C₆-phenylalanine (0.6 mmol kg⁻¹), 5 g lactulose, 1 g rhamnose, 0.5 g D-xylose, 0.2 g 3-O methyl-D-glucose, 200 mg fluorescein, 1.667 mg L-arginine-(guanidineimino-¹⁵N₂) HCl and 1.0 g glycyl sarcosine (solution E) given. A continuous intravenous infusion of ²H₅-phenylalanine (0.15 mmol kg⁻¹) and ²H₃-leucine (0.3 mmol kg⁻¹) (solution D) will be commenced, at 5 mL h⁻¹, and sips of ¹³C₆-phenylalanine (0.15 mmol kg⁻¹) and ¹³C₆-leucine (0.3 mmol kg⁻¹) (solution F) administered as sips every 20 mins (prepared as shown in Table 3). Breath samples will be collected every 20 mins for the first 1 h and every 30 mins for the remaining 3 h. Blood samples will be collected into lithium heparin collection tubes via the intravenous cannula: 2 mL every 20 mins for the first 1 h and every 30 mins for the remaining 3 h, and 5mL for the last sample. Urine will be collected up to 180 mins (3 h) while saliva collected every 60 mins for 3 h. After 240 mins (4 h) the participant will be transferred to the endoscopy room where upper gastrointestinal endoscopy will be performed using doses of sedation with midazolam and pethidine selected by the endoscopist. Oximetry will be used throughout the procedure. During endoscopy, up to 10 mL of duodenal aspirate will be collected with clean tubing. Using the same tubing, 20 mL 0.9% of sterile saline will be used to collect duodenal lavage. A brushing will be collected using an endoscopic cytology brush, and then the brush will be cut off into 1000 μL PBS. For histology, 3 biopsies will be collected into normal saline using 2.8 mm biopsy forceps, orientated on cellulose acetate paper (Sartorius, Gottingen, Germany) in the endoscopy room and placed in formalin thereafter. An additional 5 biopsies will be collected into 3 cryovials (2+2+1). All samples collected during endoscopy, except for the 3 histology biopsies, will be snap-frozen in liquid nitrogen immediately after collection. At the end of day 3, the participant will be given urine and stool collection devices for the next day.

Day 4

Only stool and urine samples will be collected.

Permeability analysis

The permeability will be measured by the appearance in the blood of fluorescein ingested orally and through use of a multi-agent urinary recovery assay. Fluorescein will be purchased from Sigma-Aldrich (46960-100G-F) and has a similar molecular size to lactulose (molecular weight of fluorescein = 332 g/mol, molecular weight of lactulose = 342 g/mol) and similar molecular radius (lactulose, 0.42 nm; fluorescein 0.5 nm). We have previously demonstrated that the rate and degree of permeation of fluorescein vary under different permeability conditions³⁸ and in patients with intestinal disorders.³⁹ Importantly, following permeation into the blood, fluorescein fluorescence is quantifiable using a fingertip probe similar to a pulse oximeter probe.⁴⁰ Hence, transcutaneous measurement (performed using a fingertip probe) of fluorescence from orally ingested fluorescein will be used to provide rapid, non-invasive assessment of permeability.

To complement these novel, non-invasive, fluorescent measurements, we will also assess permeability using the urinary recovery of five orally ingested sugar and dipeptide molecules (lactulose, rhamnose, D-xylose, 3-O methyl-D-glucose and glycyl sarcosine). Urinary recovery of these molecules will be quantified using mass spectrometry thereby permitting assessment of more traditional permeability measures (e.g. urinary lactulose recovery and lactulose: rhamnose ratio) as well as quantification of the permeation of multiple agents with different molecular weights. Together, these two approaches allow a comprehensive, minimally invasive assessment of intestinal permeability.

Stable isotopes analysis

The non-radioactive stable isotope labelled amino acids are used to trace the digestion and absorption of protein (day 2) and bidirectional transmucosal amino acid flux (day 3). The labelling strategy is chosen to minimise isotopic crosstalk in the analysis and maximise the metabolic pools that can be sampled ethically. These include breath (¹³CO₂), plasma (¹³C / ²H amino acids), intestinal aspirates (¹³C / ²H amino acids), and urine (¹³C / ²H amino acids). Breath samples will be analysed using a Thermo Scientific™ Delta Ray™ Isotope Ratio Infrared Spectrometer (Thermo Fisher Scientific, Bremen, Germany) in the TROPGAN laboratory (breath ¹³CO₂) and ¹³C & ²H labelled amino acids analysed by a Agilent 1290 Infinity LC coupled to a 6560 triple quadrupole mass spectrometer (Agilent Technologies Inc., Cheadle, UK) in SUERC (¹³C/²H amino acid isotopologues).

Metabolic phenotyping

Urine, stool, plasma, and duodenal aspirates will be analysed by Nuclear Magnetic Resonance spectroscopy (NMR). High-throughput ¹H NMR spectra will be acquired using a 600MHz Bruker Avance III™ HD NMR spectrometer equipped with a 5 mm BBI Z-gradient probe, high-order shims, and automated tuning and matching (Bruker Biospin, Rheinstetten, Germany). Samples will be analysed in automation using standard pulse sequences with water suppression. For urine, stool, and duodenal aspirates, two experiments will be acquired: (1D) ¹H NMR (noesygppr1d, standard Bruker pulse program), and (2D) ¹H−¹H J-resolved (J-Res) experiment (jresgpprqf ). (1D) ¹H CPMG (Carr–Purcell–Meiboom–Gill, cpmgpr1d) will also be acquired for plasma only. Each 1D spectrum will be automatically phased and baseline corrected, digitized, and imported into MATLAB for preprocessing and statistical modelling.

Microbiome analysis

Stool samples, gastric and duodenal aspirates and intestinal biopsies will undergo nucleic acid extraction, library construction and whole metagenome shotgun sequencing at Imperial College London to identify taxonomic and functional composition of multi-site gastrointestinal microbiomes comparing standard short-read sequencing approaches (Illumina HiSeq™ (Illumina, San Diego, CA, USA)) with novel long-read approaches (MinION (Oxford Nanopore Technologies Ltd., Oxford, UK) and PacBio Sequel^® (Pacific Biosciences Inc., Menlo Park, CA, USA)). An additional qPCR step will be performed to quantify the absolute abundance of bacteria by determining the total number of 16S rRNA gene copies. These results will also be compared across sequencing platforms. These sequencing efforts will be duplicated in parallel work in Lusaka to determine the ease of application to an African laboratory. Additional characterisation of microbial uptake of amino acids and conversion of L to D amino acids will also be carried out. Raw sequencing data will be analysed via established bioinformatics pipelines including DADA2 (for 16S rRNA genes), EPI2ME and MetaPhlAn3 (compositional), HUMAnN3 (functional) as with other data, these results will be compared with duodenal biopsy histology, lactulose permeation data, and biomarkers.

To evaluate microbial uptake of amino acids, the ratio of ¹³C & ²H labelled amino acids in bacterial cells derived from duodenal aspirates will be measured using mass spectrometry. This ratio will serve as an indicator for microbial usage of amino acids pathological leaked into the gut lumen from systemic circulation. In addition, a quantitative ultra-high pressure liquid chromatography mass spectrometry (UPLC–MS/MS) assay will be used to measure the concentrations of L- and D-amino acids in duodenal aspirates. The concentration of labelled D-amino acids will serve as an indicator of microbial consumption and usage of total amino acids content in the duodenum.

Histology and immunohistochemistry analysis

Formalin fixed biopsies will then be embedded in wax. Sections (4 μm) will be stained using haematoxylin and eosin (H&E) and scanned on an Olympus VS-120 scanning microscope. Morphometry will be performed to generate measurements of villus height and crypt depth as has been our standard practice for many years and on which several publications are based.^8,41,42 Recently a histological scoring system for EE has been developed in which we were involved, and this will be applied to generate scores for villus architectural change, Brunner’s gland penetration, Paneth cell depletion, Goblet cell depletion, intra-epithelial lymphocytosis, and epithelial abnormalities.⁴³ The expression of PRRs will be assessed in separate histological sections of antibody-labelled duodenal biopsies via immunofluorescence microscopy.

EE biomarker analysis

To characterise the sample sets fully, giving context to observed changes in the composition of the microbiota, biomarkers of EE (shown in Table 5) will be quantified in plasma and stool aliquots via ELISA.

Cell culture analysis

To evaluate the capacity of circulating microbial products in plasma and microbial PAMP content in stool to activate innate immune signalling (immunogenicity; an indicator of microbial translocation), plasma and faecal water aliquots from each participant will be co-cultured with THP-1 Dual reporter cells for NFkB and IRF3 (monocytic cell line; obtained from Invivogen), which provide a global indicator of PAMPs that can trigger immune cell activation via Toll like receptors, a major pathogen recognition receptors of the innate immune system. We will then screen samples for more specific PRR signalling activity using PRR-specific HEK293 reporter cells (obtained from Invivogen), focusing initially on TLR4 (the receptor for bacterial LPS/endotoxin) and TLR5 (receptor for bacterial flagellin), but expanding to and other PRRs if indicated by microbial metagenomics/microbiome analyses. For PRR-specific reporter HEK293 assays, we will conduct parallel cultures with HEK293 lacking the receptor of interest (HEK293-null cells) as a negative control.

To evaluate the functional effect of PAMPs and immune mediators present in plasma and stool on gut epithelial barrier integrity (transepithelial resistance, TEER), permeability, PRR and tight junction expression, and epithelial cell activation, we will add aliquots of plasma/faecal water separately to the basal/apical side of in vitro models of the gut epithelial barrier. For the model epithelium we will use Caco-2 (cell line derived from the human colonic epithelium; obtained from American Type Culture Collection, ATCC, cultured in transwell inserts so that they have an apical side (equivalent to the epithelial side facing the gut lumen) and a basal side (equivalent to the epithelial side facing the lamina propria),⁴⁴ to recapitulate the polarity of the gut epithelium. PRR-specific inhibitors or antagonists and/or antibodies and filtered faecal water will be used in parallel experiments to attempt to disaggregate the effects of the specific PAMPs present in the stool samples. We will use these data to explore the relationship between immunogenicity (from THP-1 and HEK293 reporter assays) and functional impact of the sample on the epithelial barrier in vitro (from Caco-2 transwell models). We hypothesise that samples from participants with more circulating immunogenic PAMPs will be associated with greater loss of barrier function in vitro.

Antibiotic analysis

In the 20 patients attending for routine colonoscopy (for diagnostic purposes) in Harare, using luminal fluid, antibiotic analysis will be carried out by NMR or SUERC by LC-MS.

Statistical analysis plan

With the complexity of the data set from the techniques described above, the biostatistical framework will be generated in a data-dependent manner. Data from novel tools will be integrated with mucosal morphometry, lactulose permeation, and biomarkers in blood and stool. All data will be tested for normality using a Shapiro-Wilk test and transformed accordingly (normalisation, scaling, etc.). Both parametric and non-parametric tests will be used and both univariate and multivariate analysis will be performed on all datasets generated, as appropriate. For univariate tests, to control for type-I errors, multiple testing will be corrected for by applying the Hommel correction (for Family-Wise Error Rate) or Benjamini-Hochberg (for False Discovery Rate) as appropriate for mildly (positively) correlated variables, or Benjamini-Yekutieli or Storey-Tibshirani (both for False Discovery Rate) methods for highly (positively and inversely) correlated variables. For multivariate tests, data will first be split into training (modelling) and test (evaluation) sets prior to any procedures on the data such as centering, scaling, or other transformations. The predictive classification models will be calculated only on the training portions and evaluated (accuracy and/or F1-score) on the test portions, with hyperparameter tuning using cross-validation on the training data. When small sample sizes do not allow setting aside a completely independent test set, repeated cross-validation will be used to ensure the robustness of the models by calculating a multitude of models (e.g. 1,000 individual models). The evaluation of performance is estimated across all models on the portion of the data that was set aside (test) in each partition. Where algorithms use random initializations of parameters, for each analysis the starting random number seed is saved to allow replication. All data generated will be compared with all duodenal morphometry data which is the most direct measure of EE severity; these comparisons will allow for validation of new real-team assessment tools and established laboratory techniques for sensitivity to differences in EE severity within the cohort.

Public involvement

Dissemination will include the participants and their families, together with Community Advisory Boards, the University of Zambia School of Medicine, and the University of Zimbabwe Faculty of Medicine and Health Sciences. We will also disseminate results to the Zambian National Health Research Authority and the Medical Research Council of Zimbabwe, and the relevant ethics committees in all the African partner countries, both in written form and by taking opportunities of local research dissemination meetings. This is established practice in all the research groups who will be collaborating on the work proposed. All publications will be open access.

Data management

Processed data and extracted variables from all sites will be stored on OneDrive (hosted at Imperial College London) with secure access for all researchers involved in the study. One copy of the raw data will be stored at the institution that generated it, and another copy will be stored in the same cloud storage as the processed data. All databases will be archived with related questionnaires and any related dictionaries in soft copy on the same drive(s). All data records and materials will be archived appropriately (in accordance with ICH GCP Guidelines) and retained for 5 years according to the Sponsor’s, QMUL’s, requirements. Data with identifiers will not be stored, for example, personal identifiers on the front sheet of all clinical trial records will be detached prior to archiving and will not be entered on any database. These identifiers will be kept as hard-copy files in locked cabinets in study offices.

Confidentiality

Data security and confidentiality will be in line with General Data Protection Regulations, HRA Guidance, International Conference on Harmonisation, Guideline for Good Clinical Practice E6, Revision 2, and Queen Mary University of London Data Protection principles and policies. Only authorised study personnel will have access to study files and data. Any data that will be shared with other workers in the field will be anonymised. Identifiable personal data will not be used during analysis or when publishing results. The information that is collected will be treated confidentially. The information entered on separate forms only contains a participant number and the initials and only the data considered important for the study. Research staff are aware of confidentiality of personal data and meeting the requirements of the General Data Protection Regulation (GDPR). Personal information will only be used to contact the patient via letter or telephone after informed consent has been obtained. The only link between personalised data and anonymised data will be a register kept in hard copy in a locked cupboard by the compliance manager. Data will be maintained at an appropriate vigilance (RAID 1) level following current institutional advice. Data access will be restricted to only authorised personnel with individual passwords.

Ethical considerations

The protocol has been approved by the National Heath Research Agency (REF NHREB00001/02/2022, dated 2^nd February 2022) and the University of Zambia Biomedical Research Ethics Committee (reference number 2291-2021, dated 20^th December 2021). The Zimbabwean studies have been approved by the Medical Research Council of Zimbabwe (MRCZ/A/3065, dated 31 July 2023). The work will be carried out in full concordance with the principles of the Declaration of Helsinki, and all participants will give written consent. Queen Mary Ethics of Research Committee issued a letter of no objection (QMERC22.060, dated 22^nd February 2020).