Trial design

There is emerging awareness that multi-arm phase II clinical trials offer a more efficient approach to identification of new therapies than parallel conventional trial designs, in which one agent is evaluated against a control. Our recent work on the pathophysiology of malnutrition enteropathy suggests that mucosal healing may be central to reducing mortality, but we do not yet know the efficacy of several plausible therapeutic approaches. In order to evaluate these four new potential therapies efficiently and quickly, they will be compared side-by-side using biomarkers of pathophysiology as endpoints. This design has the following advantages:

1. The four novel therapies will each be evaluated against standard care;

2. Allocation to all five treatment groups will run in parallel, so that all children will be randomised to any of the five groups at any point in time to avoid biasses due to seasonal variation;

3. The trial endpoints will evaluate a range of biomarkers that capture different domains of malnutrition enteropathy, to allow a comprehensive non-invasive assessment of mucosal healing;

4. A subgroup of children (those recruited in Lusaka) will have small intestinal endoscopy in a unit with a track record of safety in children;

5. Endoscopic biopsy will provide evaluation of healing and ensure that the selected therapy does not induce unwanted immune or other effects;

6. The overall trial design provides an efficient way of identifying one or more candidates for a future phase III trial.

Study setting and participant recruitment

The study will be conducted in hospitals in Lusaka, Zambia, and Harare, Zimbabwe, which have experience of several previous studies of severe acute malnutrition.11 20 24 Children hospitalised with SAM will be eligible for inclusion if they are aged 6 to 59 months, of either sex, are an inpatient in one of the children’s wards of the participating hospitals, have initiated transition (from F75 feed to F100 or RUTF), and clinically stable. SAM is defined using WHO definitions: weight-for-length/height Z-score of less than −3, or mid upper arm circumference of less than 11.5 cm and/or bilateral pedal oedema. During the Stabilisation Phase of treatment of SAM, readiness for transition is determined by the ability to finish F75 feed by mouth and reduction in oedema (if patient was admitted with oedematous SAM). Such a child is deemed to be ‘clinically stable’ and in general is alert, not requiring oxygen, not hypothermic, hypoglycaemic and is not in shock or dehydrated.

Children will be excluded if they have weight less than 5 kg, a neurological disability or oro-facial abnormality which would explain or partly explain poor feeding, haemoglobin concentration 6 g/dL or less, contraindication to any of the trial treatments (eg, allergy to cow’s milk protein) or any underlying condition, other than HIV, which in the opinion of the investigator would put the subject at undue risk of failing study completion or would interfere with analysis of study results. If the caregiver is unwilling to remain in hospital for the duration of the study treatment (14 days) they will not be recruited.

Interventions: investigational therapies

Colostrum is the first liquid secreted by the lactating breast, and breastfed children take it for approximately the first 3 days of life. It is similar to breast milk, but with higher protein content. Bovine colostrum is available as a high-protein bovine colostrum powder (Neovite) for use as a health-promoting nutritional supplement. Colostrum contains nutrients, immunoglobulins and growth factors, including epidermal growth factor and insulin-like growth factor 1 (IGF-1); it has been shown to reduce intestinal permeability in adults.25 Bovine colostrum is potentially allergenic in theory, though there is no evidence that cow’s milk products are harmful in children with SAM and the standard therapeutic feeds (F75 and F100 recommended by WHO) contain milk proteins.

N-acetyl glucosamine (GlcNAc) is a natural amino-sugar present on every cell surface. All breastfed children consume GlcNAc in human milk throughout lactation.26 Impaired glycosylation of glycosaminoglycans has been noted for many years in oedematous malnutrition, with reduced concentrations of glycosaminoglycans in blood, urine, kidney, brain and small intestine. Specific consequences of reduced heparan sulfate expression include gut epithelial leakiness with hypoalbuminaemia.27 GlcNAc administration has been demonstrated to restore the intestinal epithelial charged barrier in Crohn's disease.28 As with all sugars, GlcNAc may theoretically induce osmotic diarrhoea if not absorbed in the small intestine. Although this has not been observed clinically in animals or in older children, we will use a dose escalation schedule to minimise the chance of this happening.

Teduglutide is a long-acting form of GLP2 which has proven efficacy in intestinal failure, improving absorption and reducing the need for parenteral support.29 GLP2 is a hormone secreted by L cells in the terminal ileum, which drives epithelial repair and mediates intestinal adaptation by increased cellular proliferation and villus hypertrophy. Teduglutide will be given by subcutaneous injection (0.05 mg/kg/day) daily for 14 days. In a recently published description of tolerability of 12 weeks of teduglutide in children in the UK and the USA,30 vomiting was attributed to teduglutide in 10% of recipients and fever in 14%. No children had treatment withdrawal because of adverse events. Three adverse events were considered of special interest (intestinal obstruction, fluid overload and biliary derangements) but none were observed.

Budesonide is a corticosteroid which reduces inflammation in the gut but is then rapidly broken down in the liver which minimises systemic effects. Budesonide is standard therapy for Crohn’s disease, and can be used for refractory coeliac disease. Since malnutrition enteropathy is characterised by intestinal inflammation, with infiltration of activated T cells,31 an anti-inflammatory approach is rational. A prior trial of mesalazine in Kenya confirmed that an immuno-modulatory approach with mesalazine was safe in the setting of SAM,32 but targeting the small, rather than large, intestine with a more potent agent may be more effective. Budesonide may cause immunosuppression and other corticosteroid effects (oedema, hypertension, glucose intolerance, osteoporosis), but usually only after longer-term administration. It is the corticosteroid of choice for intestinal disorders because it causes fewer adverse events than prednisolone because of low systemic absorption and first-pass clearance by the liver. The dose will be tapered (1 mg orally three times daily for 7 days, then 1 mg twice daily for 4 days, then 0.5 mg twice daily for 3 days) to mitigate any possible effects of adrenal suppression.

Trial procedures

The trial schema is shown in figure 1. Children hospitalised with SAM will be enrolled from any of three hospitals (Lusaka Children’s Hospital, Lusaka, Zambia; Parirenyatwa Hospital, Harare, Zimbabwe; Harare Central Hospital, Harare, Zimbabwe) once they have completed the stabilisation phase of nutritional rehabilitation and are at the point of transition from F75 therapeutic milk (low-calorie feed) to RUTF or F100 milk (feeds richer in calories). Children will be randomised in an allocation ratio of (1:1:1:1:1) to either colostrum, GlcNAc, teduglutide, budesonide or Standard Care and followed for 14 days in hospital (table 1). All children will receive Standard Care according to WHO guidelines,9 and treatment with trial medications will commence after transition.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flowchart of recruitment into the TAME trial. GlcNAc,N-acetyl glucosamine; TAME, Therapeutic Approaches for Malnutrition Enteropathy.

Randomisation codes will be prepared in advance by the trial statistician, and treatment allocation will be revealed after enrolment by opening sealed envelopes held by the trial coordinators in each of the three study sites. Randomisation codes will be generated in permuted blocks and stratified by study site. Varying block sizes will be used; block size and distribution will not be disclosed to the study clinical teams. There will be no attempt at blinding as the interventions are readily distinguishable (eg, teduglutide is given as subcutaneous injection, so to mask all interventions would require administration of several placebo formulations). However, the primary endpoints are biomarkers so laboratory staff and data analysts will be blinded to randomised group.

Outcomes and endpoints

The primary endpoint for this trial will be measured on day 15 (allowable window 15 to 19 days) after initiating treatment by analysis of faecal biomarkers. Gut inflammation will be measured as a composite score of faecal myeloperoxidase, neopterin and alpha-1-antitrypsin, as described below. Secondary endpoints (table 2) will be measured by daily clinical review (for clinical endpoints) during the intervention period and by collection of biological specimens on day 15 (allowable window 15 to 19 days) after initiating treatment. As endoscopic biopsies will be collected from trial participants in Lusaka, we anticipate that at least 20 biopsy scores will be available for each trial arm. Few trials have used histological endpoints in SAM, so this component of the trial is comparatively novel and therefore exploratory.

Sample size and data analysis

Kosek et al 33 examined established markers of intestinal inflammation which can be used to obtain population-based measures of intestinal injury and altered intestinal function. They developed an environmental enteropathy (EE) activity score which is a composite score of three stool markers for mucosal inflammation. The composite score is derived from measurements of faecal myeloperoxidase (MPO), neopterin (NEO) and alpha-1-antitrypsin (AAT). These are stable compounds, relatively resistant to proteolysis and therefore suitable as biomarkers.

The weighted EE score is given in Equation 1:

Equation 1

We will use the principle of a weighted enteropathy score as our primary outcome. We will not trichotomise the variables as Kosek et al 33 did, but keep the score as a continuous composite variable. This is because we are interested in gauging the potential magnitude of the effect these treatments have, and applying a cut-off value to a continuous composite score could mask or discard information.34 35 The modified Enteropathy score is given in Equation 2, with modification of weighting of units to bring measurements into a comparable arithmetical range:

Equation 2

We will have a control arm, so can understand the effect of the treatment against a reference standard. The analysis will be per-protocol; no interim analysis is planned.

There are no available data from children with SAM to draw on for an understanding of the baseline variance of the EE biomarker score. The primary outcome response from each patient is the EE biomarker score at 15 days (window 15 to 19 days) after treatment initiation, adjusted for baseline values. We have assumed that the EE biomarker score is normally distributed with a common SD. To determine the sample size, we drew on two methods to establish an effect size. We assumed the difference would be larger than the likely inherent imprecision in the measurements of each outcome, and we have defined the magnitude of the effect on a standardised scale. To detect a medium/large effect of Cohen’s d effect size of 0.6, with 90% power and 90% confidence and a conservative correlation between baseline and follow-up estimate of 0.5, we will need a sample size of 36 per group across five groups to analyse with the analysis of covariance (ANCOVA) method (180 in total to be treated). We expect there to be approximately 10% loss to follow-up due to deaths, and 10% loss to follow-up due to other reasons, such as drug intolerance, withdrawal and missing specimens. Adjusting our sample size of 180 for 20% loss to follow-up, we therefore aim to randomise 225 patients in total (45 in each group).

We will use a mixed effect ANCOVA model to compare the environmental enteropathy activity score and secondary endpoints in each group against the control group, adjusting for several core covariates: sex (male/female), oedema (yes/no), HIV status (yes/no), diarrhoea (yes/no), breastfeeding (yes/no), baseline Weight for length z score (WLZ) scores (continuous), baseline biomarker/histology scores (continuous) and trial site. Treatment effects will be deemed statistically significant if the p value is less than or equal to 0.10 when compared with the control arm, for all four treatments. This less stringent cut-off has been chosen to reduce the likelihood of rejecting a potentially valuable treatment which might show benefit in a phase III trial. All trial committee members (see below) will contribute to a discussion at the end of the trial as to which, if any, of these treatment effects are most clinically significant.

We will not undertake any adjustments of the false-positive (type I) error rate, as the aim of this trial is to inform the treatment development process for this population, which would lead to a phase III trial if a degree of efficacy is observed. The general consensus is that adjustment for the type I error rate is not required in exploratory multi-arm multi-stage trials in phase II within the treatment development framework.36

If patients are discharged from hospital sooner than day 14, we will measure the primary endpoints as close as the discharge date as possible, and carry this observation forward as the primary outcome. Children who cannot submit a stool sample will be excluded from the primary analysis as the primary endpoint requires this.

Safety reporting

All adverse events will be reviewed for causality, expectedness and severity. Serious adverse events will be reported urgently to the Trial Management Group for evaluation, and all serious adverse reactions will be regarded as unexpected (ie, SUSARs). There are two adverse events of special interest for teduglutide: fluid overload and intestinal obstruction, as these have been reported in adults on long-term teduglutide. Complicated SAM has an inpatient mortality rate of around 15%, so we anticipate up to 45 deaths in the trial; however, all deaths, regardless of causality, will be reported to the local ethics committees in Zimbabwe and Zambia according to local reporting policies. The trial will be supervised by three committees: the Trial Management Group (TMG), the Data Monitoring and Ethics Committee (DMEC, which will meet every 6 months) and the Trial Steering Committee (TSC). The TMG and TSC will review the pattern of any Serious Adverse Events (SAEs) and Serious Unexpected Adverse Reactions (SUSARs) in real time, and are empowered to introduce urgent special measures if the profile of adverse events suggests this is required. The TSC and DMEC are composed of statisticians and paediatricians, and independent of the TMG.

Patient and public involvement

There was no patient or public involvement in the design of this trial.