Optimisation, validation and field applicability of a 13C-sucrose breath test to assess intestinal function in environmental enteropathy among children in resource poor settings: Study protocol for a prospective study in Bangladesh, India, Kenya, Jamaica, Peru and Zambia

Introduction Environmental enteropathy (EE) is suspected to be a cause of growth faltering in children with sustained exposure to enteric pathogens, typically in resource-limited settings. A major hindrance to EE research is the lack of sensitive, non-invasive biomarkers. Current biomarkers measure intestinal permeability and inflammation, but not the functional capacity of the gut. Australian researchers have demonstrated proof of concept for an EE breath test based on using naturally 13 C-enriched sucrose, derived from maize, to assay intestinal sucrase activity, a digestive enzyme that is impaired in villus blunting. Here, we describe a coordinated research project to optimise, validate and evaluate the usability of a breath test protocol based on highly enriched 13 C-sucrose to quantify physiological dysfunction in EE in relevant target populations. Methods and analysis We use the 13 C-sucrose breath test (13 C-SBT) to evaluate intestinal sucrase activity in two phases. First, an optimisation and validation phase will (1) confirm that a 13 C-SBT using highly enriched sucrose tracers reports similar information to the naturally enriched 13 C-SBT; (2) examine the dose-response relationship of the test to an intestinal sucrase inhibitor; (3) validate the 13 C-SBT in paediatric coeliac disease (4) validate the highly enriched 13 C-SBT against EE defined by biopsy in adults and (5) validate the 13 C-SBT against EE defined by the urinary lactulose:rhamnose ratio (LR) among children in Peru. Second, a cross-sectional study will be conducted in six resource-limited countries (Bangladesh, India, Jamaica, Kenya, Peru and Zambia) to test the usability of the optimised 13 C-SBT to assess EE among 600 children aged 12-15 months old. Ethics and dissemination Ethical approval will be obtained from each participating study site. By working as a consortium, the test, if shown to be informative of EE, will demonstrate strong evidence for utility across diverse, low-income and middle-income country paediatric populations. Trial registration number NCT04109352; Pre-results. Our coordinated study design is summarised in table 1. Features of the coordinated research projects that make up the study protocol EE, environmental enteropathy; n/a, not available; SES, socioeconomic status. In the first phase of the study, our overall objective is to optimise and validate a protocol for a non-invasive stable isotope test based on an enriched sucrose substrate (13C-SBT), among adults and children. To accomplish this, we will complete five coordinated studies. We will first optimise the 13C-SBT protocol and second, validate the 13C-SBT in successive adult and paediatric populations. We will then establish analytical validity (technical test performance), clinical validity (the test’s ability to accurately and reliably identify a disorder of interest) and field usability (assessment of the test in the actual context where it would be used).42 43 The objective of the first study is to establish that the highly enriched sucrose tracers for an 13C-SBT report similar information in comparison to the original naturally enriched 13C-SBT. We will conduct a cross-over study of 20 adults using three commercially available sucrose tracers (13C6 fructose; 13C6 glucose and 13C12 sucrose). We will also determine whether the addition of unlabelled carrier sucrose is necessary to replicate the original ‘flooding dose’ approach reported by Ritchie et al.34 To assess gut permeability, participants will also receive coadministration of 5 g lactulose, 1 g rhamnose, 0.5 g xylose, 0.2 g 3-O-methyl-D-glucose and 5 g sucralose dissolved in water. The objective of the second study is to characterise the dose response of the 13C-SBT in response to three different doses of the intestinal sucrase inhibitor acarbose. A randomised cross-over trial of 20 adults will be conducted. Breath 13CO2 will be collected serially for 4–6 hours and 13CO2 recovery compared across treatments. Both studies 1 and 2 will be conducted in Glasgow, UK (University of Glasgow). While the ideal ‘gold standard’ would be to breath test and biopsy children who are identified as having EE, from a logistical perspective, this study design is infeasible in the resource constrained environments where EE is prevalent. Therefore, the objective of the third study is to examine whether the 13C-SBT, optimised according to the protocol established in study 1, varies between children with clinically diagnosed coeliac disease presenting with gastrointestinal symptoms to outpatient clinics (n=20) versus healthy coeliac controls (n=20) and healthy non-coeliac controls (n=20) (Adelaide, Australia, Flinders University). In this study, active disease is defined as positive serology for specific IgA antibodies in patients on a gluten-containing diet, or in patients undertaking a gluten challenge, followed by endoscopic evaluation and histological examination of duodenal biopsy for characteristic features of coeliac disease (villus atrophy, crypt hyperplasia and mucosal inflammation), with a Marsh III classification considered coeliac positive. The degree of villous atrophy will be correlated to the patients 13C-SBT result. Additionally, tissue biopsies, collected from inflamed and normal sections of small bowel, will be assayed ex vivo for SI activity and correlated with the 13C-SBT. The non-coeliac control group includes children presenting for endoscopy for non-coeliac disease, including abdominal pain and reflux disease, and do not present with any small intestinal pathology. The objective of the fourth study is to validate the 13C-SBT against intestinal sucrase activity and villus atrophy among adults with and without EE. Both villus atrophy and intestinal sucrase activity are measured using biopsy. A case–control study design (n=20 cases from a high-risk enteropathy setting and n=20 controls) will be used. (Lusaka, Zambia, University of Zambia) The objective of the fifth study is to correlate the 13C-SBT with an established, non-invasive biomarkers of EE (urinary LR ratio) and one secondary biomarker of EE (plasma kyurenine:tryptophan ratio) among children under 2 years of age from a high-risk enteropathy setting (n=30). (Iquitos, Peru, Asociación Benefica Proyectos de Informática, Salud, Medicina, y Agricultura (A.B. PRISMA) and the University of Virginia). A substudy will remeasure the 13C-SBT at 7 days to document test reproducibility. In the second phase, we will conduct a multisite study in six resource-limited countries to determine the field usability of the optimised and validated 13C-SBT in diagnosing EE in children aged 12–15 months. The specific primary objectives are to assess the relationship between the 13C-SBT and the LR ratio among children 12–15 months of age, and to assess the relationship between the 13C-SBT and child stunting. Our secondary objectives are to assess the relationship between the 13C-SBT and secondary biomarkers of EE. We will also conduct exploratory analyses to characterise the relationship between the 13C-SBT and child sex, SES, dietary diversity and household food security. This six study sites are: Dhaka, Bangladesh (the International Centre for Diarrhoeal Disease Research); Bangalore, India (St. John’s Research Institute, St. John’s National Academy of Health Sciences); Kingston, Jamaica (The Tropical Metabolism Research Unit of the Caribbean Institute for Health Research, University of West Indies); Kakamega, Kenya (Masinde Muliro University of Science and Technology); Iquitos, Peru44 (Asociación Benefica PRISMA and the University of Virginia) and Ndola, Zambia (Tropical Disease Research Centre). These sites represent a range of epidemiological contexts which enhances the cross-context applicability of study results. Each site will enrol 100 infants between 12 and 15 months of age. This range was selected because it is within the window of infant growth faltering that implies clinical or public health relevance but is also old enough to reduce the influence of breastfeeding on LR performance and to permit a several hour fast during initial assessment of the test.45 46 At each site, 90 children will be recruited from areas deemed high risks for EE, due to a lack of improved water and sanitation infrastructure or because the known prevalence of stunting is relatively elevated. Ten relatively high SES infants from a nearby community will also be enrolled. All children will be recruited and enrolled through convenience sampling, either at the community level (if the study site has previously censused the community) or through child clinic visits. Exclusion criteria include the presence of Severe Acute Malnutrition (weight for height z-score ≤−3 SD), HIV positive status, any chronic illness medical or surgical contributing to growth retardation, or weight-for-height z-score more than +2 SD. Study procedures for each participating child are outlined in figure 1 and described in detail in online supplemental appendix 1. In brief, the 13C-SBT will be assessed in each child at one time point, as well as a 2-hour urinary LR test, an assessment of weight, length and body composition using the deuterium dilution technique (either saliva or urine), and a fasting plasma sample for the assessment of additional EE biomarkers. After 3 months, the height and weight measurement will be repeated. Each site will use the same harmonised study protocols for all data collection. Information about household SES, household food security (using the household food insecurity access scale), and child dietary diversity will also be asked of each caretaker using standardised instruments designed and validated for use across low-income and-middle income countries.47–49 Information about child morbidity and the consumption of C4 foods will also be collected using standardised questions. In certain sites with high rates of cell phone coverage, ancillary morbidity data will also be collected by phone every 2 weeks, although this will not be used in the primary or secondary analysis. Key study data are summarised in online supplemental appendix 2, and study forms are provided in online supplemental appendix 3. Flow diagram for phase 2 coordinated study protocol shown here is the timeline of participant activities. Darker grey boxes represent core study activities, while light grey boxes indicate activities that some, but not all, study sites will undertake. Primary and secondary study aims are based on core activities. bmjopen-2019-035841supp001.pdf bmjopen-2019-035841supp002.pdf bmjopen-2019-035841supp003.pdf All data will be digitised by site via double-data entry and will be managed in accordance with institutional norms and local ethical committee approvals. Each site will confidentially maintain personal information about enrolled participants as necessary for study administration. Deidentified data will be shared with the University of Michigan where centralised data consistency checks will be performed, and inconsistencies will be communicated back to each site for resolution. After this process is complete, pooled analyses will be performed. 13C breath tests can be summarised in several ways.50 Based on expert opinion, cumulative per cent of dose recovered at 90 min post administration (cPDR90), and time to 50% recovery (T50), will be taken as our primary measures of the 13C-SBT. Other metrics for summarising the 13C-SBT test, at both specific time points and overall, will also be considered. LR ratio: To assess the relationship between the 13C-SBT and EE, our case definition for EE, and primary outcome measure, will be based on the LR ratio, which is the most widely accepted, non-invasive test of EE. Because LR ratios vary by analytical platform26 and may also be influenced by test administration procedures, cut-offs for ‘EE’ will be defined based on the empirical distribution of the data. To establish cut-offs for LR, the distribution of this variable among high-SES children (pooled across sites) will be examined (eg, LR ratios above the 90th percentile for upper SES children will be regarded as ‘elevated’). If insufficient numbers of higher-SES children are recruited, or if a subset of higher-SES children cannot be recruited across all sites, cut-off values for LR will be determined based on internal study percentiles (eg, below vs above the median) or on cut-offs from the literature.51 If the outcomes of the phase 1 studies support evidence for a LR cut-off based on the extent of villous atrophy, this will also be considered. Anthropometry: We will use WHO growth standards to define length for age Z-score (LAZ), weight for age Z-score (WAZ) and weight for length Z-score. Stunting, underweight and wasting will be defined based on WHO growth standards (≤ −2 Z-scores length for age, weight for age and weight for length).52 Our primary outcome measures are Secondary outcome measures Other prespecified outcome measures are We will assess the coefficient of variation and correlation coefficient between repeated cumulative percent of dose recovered at 90 min postadministration on separate SBT tests administered 1 week apart done on the same child (Peru site only). We will determine if significant associations exist between 13C-SBT measured as cPDR90/T50 and the Water Assets Maternal Education and Income (WAMI index). The WAMI index is a previously validated composite index of environmental variables to create an index that expresses the socioeconomic and physical environment in diverse geographical contexts.47 The sample size for the phase 2 study was predetermined (N=10 upper SES and 90 lower SES children per site), power calculations were conducted. Calculations were based on previously reported Pearson’s correlations of 0.67 (95% CI 0.42 to 0.82) between the naturally enriched 13C-SBT and LR.34 The current test, using highly enriched sucrose, is expected to be more sensitive than the original test, so sample size estimates are regarded as conservative. Assuming the Pearson’s correlation between the enriched 13C-SBT and LR is similar, 90 children per site is sufficient to estimate the correlation between the two tests with an SD of 0.58 (95% CI 0.55 to 0.79).53 The minimal detectable correlation within each site, would be 0.31.54 Between sites, we estimate statistical power to detect meaningful differences between children with and without EE, based on a cut-off of the LR ratio. The proportion of children who will be classified as having EE relative to this these cut-offs is unknown, so the estimated detectable difference was calculated across a range of values (10%–50%). We estimate that differences in the 13C-SBT on the order of 0.60 standard deviations (50% prevalence of EE) to 0.99 SD (10% prevalence of EE) will be detectable with 80% power. For comparison, Ritchie observed differences in the 13C-SBT cumulative percentage of dose recovered at 90 min between healthy Aboriginal and non-Aboriginal children on the order of ~0.84 SD, and differences between Aboriginal children with and without acute diarrhoea on the order of ~0.92 SD. Power calculations were also performed to assess the relationship between the 13C-SBT and child stunting based on the known prevalence of stunting in each site (table 2). Power calculations are based on the primary comparison of stunted to non-stunted children Table 2 shown here are estimated detectable differences in the 13C-SBT between stunted and non-stunted children, based on the estimated prevalence of stunting in specifically proposed study communities (estimates of stunting prevalence provided by study community). The percentage of variability in the 13C-SBT based on site is unknown, so a range of design effects (1.0–0.5) are provided. *Asterisks refer to the overall sample size adjusted for the design effect. 13C-SBT, 13C-sucrose breath test. All analyses will be stratified by site and then pooled. Analysis stratified by site will be limited to bivariable comparisons, and pooled data will be used to construct multivariable models, using either fixed or random effects to account for site. We will examine the relationship between the 13C-SBT and LR both continuously, and as dichotomous variables. Continuous analyses will include calculation of both site-specific and pooled correlation coefficients (to provide direct comparison to Ritchie34 and regression models will be developed where the dependent variables will be log-transformed LR, and the independent variable will be cPDR90 and T50. For dichotomous analyses, Receiver Operating Characteristic (ROC) curves will be used to calculate the sensitivity and specificity of be cPDR90 and T50 cut-offs to predict relatively elevated LR test results. We will also examine the association between the 13C-SBT and lactulose and rhamnose excretion individually. To characterise the relationship between the 13C-SBT test and child anthropometry, we will compare be cPDR90 and T50 values for the 13C-SBT and concurrently measured LAZ and WAZ. We will also consider fat mass and fat-free mass. Nutritional status will be analysed both continuously and will dichotomised (ie, into stunted and non-stunted). T-tests will be used to compare be cPDR90 and T50 between the stunted and non-stunted groups within sites, and simple and multivariable linear regression models including random effects for country membership, where the dependent variables will be LAZ, WAZ and the independent variable will be 13C-SBT. In addition, we will consider adjustment for factors such as the age, sex, breastfeeding status and recent illness history of the child. Our first secondary objective is to assess the relationship between the 13C-SBT and secondary biomarkers of EE. This activity will be contingent on the availability of these biomarker results from a sufficient number of infants across the study sites. Following the prior approachs,27 55 the relationship between EE biomarkers will be explored and EE scores will be generated via principal components analysis, partial least squares regression, or other variable reduction techniques, and comparisons between be cPDR90 and T50 and these scores will be examined similarly to LR. Following previous approaches, we will examine the association between the 13C-SBT and subsequent change in WAZ, and LAZ,25 56 57 which enhances the comparably of our results to those of other studies. We will again consider adjustment for factors that may influence child growth trajectory such as the age, gender, breastfeeding status, and recent illness history of the children in pooled models only. Finally, we will conduct exploratory analyses. To assess the reproducibility of the 13C-SBT we will examine the coefficient of variation and correlation coefficient between repeated tests from the same child (Peru site only). We will also examine the relationship between the 13C-SBT and child sex, SES, dietary diversity and household food security. Regression models will be developed where the dependent variable will be the result of the 13C-SBT test (transformed if necessary) and the independent variables will include factors that may be associated with the infant gut function, including breastfeeding, dietary diversity, age, sex, food security, history of recent illness and SES scores.47 If between-site differences in 13C-SBT are observed, the models will include site-level random intercepts. No patient involvement. Each study protocol has been approved (Zambia, Australia, Peru, Bangladesh, India, UK, Zambia, Kenya) or is pending approval (Jamaica) by the institutional review board or boards relevant to that study site. Written informed consent will be obtained from the participant themselves, and/or the legal guardian of each participant, by members of each local study team. In both the phase 1 and the coordinated phase 2 studies, each study site will use a unique consent form, reflective of both of core study activities and any site-specific activities also being performed. We will publish and disseminate our results once the project is complete. By conducting the field usability phase of our study across six countries, our test, if shown to be informative of EE, will demonstrate strong evidence for utility across diverse, low-income and middle-income country paediatric populations.