Trial tests pancreatic enzymes, bile acids in severely malnourished acutely ill children

A large international clinical trial has found that two biologically promising treatments—pancreatic digestive enzymes and the bile acid ursodeoxycholic acid—did not reduce deaths among acutely ill children with severe malnutrition. The randomized, double-blind study, conducted across Bangladesh, Kenya, Malawi and Uganda, was stopped early after interim analyses showed that neither treatment was likely to outperform placebo. The findings challenge hopes that repairing digestive and intestinal dysfunction with relatively simple add-on therapies could transform survival during the most dangerous phase of severe acute malnutrition.

Severe malnutrition affects approximately 12.2 million children under five worldwide, and children who become seriously ill while malnourished face a dramatically elevated risk of death. Standard hospital care includes specialized feeding, antibiotics and supportive treatment, yet mortality remains high. In the new trial, 71 of 429 enrolled children—16.6 percent—died within 60 days. Most deaths occurred rapidly: the median time to death was eight days, and 57 of the 71 deaths happened during the first 21 days, while the children were receiving the study products. The timing highlights the extreme physiological instability of this population and the difficulty of producing an immediate survival benefit with therapies aimed at digestion or the gut microbiome.

The trial was built around a 2 × 2 factorial design, allowing investigators to test two interventions separately and in combination without needing four entirely independent studies. Children aged between two and 59 months were randomly assigned to receive pancreatic enzymes or a matched placebo, and independently assigned to receive ursodeoxycholic acid or its placebo. This created four possible treatment combinations: pancreatic enzymes alone, bile acid alone, both treatments, or double placebo. The products were matched in appearance, volume and packaging, and caregivers, clinicians, investigators and outcome assessors remained unaware of treatment assignments until the database was locked. The children were followed for 60 days, with the primary endpoint defined as time to death from enrolment.

The scientific rationale was strong. Severe malnutrition is associated with exocrine pancreatic insufficiency, meaning that the pancreas may release too few enzymes to digest proteins, fats and carbohydrates efficiently. Pancreatic acinar cells normally produce enzymes such as trypsin, lipase and amylase. The liver and bile ducts, meanwhile, produce bile acids that help emulsify fats into micelles, structures that make lipids accessible to digestive enzymes and promote absorption in the small intestine. Pancreatic secretions and bile also have antimicrobial effects. They can influence which bacteria survive in the upper small intestine, where severe malnutrition is often accompanied by small-intestinal bacterial overgrowth and disruption of the normal microbial compartments.

That disruption may be more than a digestive inconvenience. In malnourished children, intestinal dysbiosis can contribute to inflammation, damage the gut barrier and impair growth. A weakened barrier may allow bacterial components or even viable bacteria to cross from the intestinal lumen into the bloodstream, potentially worsening systemic inflammation and sepsis. Bile acids can damage bacterial membranes, promote oxidative stress and affect microbial metabolism, while pancreatic secretions contain antibacterial molecules and peptides. Earlier observational work and a small, unblinded trial had suggested that pancreatic enzyme replacement might improve survival. Animal studies had also indicated that bile acid supplementation could improve intestinal integrity, reduce dysbiosis and protect against inflammation. The new trial was designed to determine whether those mechanisms translated into better outcomes in critically ill children.

Between June 2021 and October 2022, the researchers enrolled children admitted with acute non-traumatic illnesses and at least two danger signs of severe childhood illness. The children met World Health Organization criteria for severe malnutrition, including a weight-for-length or weight-for-height z-score below −3, a very low mid-upper-arm circumference, or nutritional oedema. The cohort had a median age of 10.4 months, and 55 percent were male. Their average mid-upper-arm circumference was 10.3 centimetres, while average weight-for-height and height-for-age scores were −3.44 and −3.61, respectively. Gastroenteritis was the most common diagnosis, followed by severe pneumonia, malaria and HIV infection. The pancreatic enzyme regimen supplied pancreatin containing lipase, amylase and protease at a target dose of 3,000 international units of lipase per kilogram twice daily. Ursodeoxycholic acid was given at 10 milligrams per kilogram twice daily, and both interventions continued for 21 days.

The results were unambiguous for pancreatic enzymes. Thirty-five of the 213 children assigned to enzymes died, compared with 36 of 216 assigned to placebo. The hazard ratio for death was 1.02, with a 95 percent confidence interval from 0.64 to 1.62, and the difference was not statistically significant. In practical terms, the mortality estimates were almost identical. Adjusting for the hospital site did not change the result. The bile acid comparison was less reassuring: 39 of 212 children assigned to ursodeoxycholic acid died, compared with 32 of 217 receiving placebo. Its hazard ratio was 1.27, with a 95 percent confidence interval from 0.79 to 2.02. This did not meet the threshold for statistical significance, but the point estimate leaned toward higher mortality rather than protection, prompting the investigators to raise safety concerns about empiric use during critical illness.

The trial’s adaptive design allowed the team to act on accumulating evidence. After 200 children had been enrolled, a prespecified Bayesian interim analysis found that both interventions met the criterion for being no worse than placebo, so both were retained. After 429 children had enrolled, a second analysis assessed whether either treatment was superior. Neither reached the required posterior probability of at least 80 percent. The independent Data Monitoring and Safety Board recommended stopping the trial for futility, and the Trial Steering Committee accepted that recommendation. This approach avoided exposing additional children to interventions that were unlikely to provide a meaningful survival advantage, while preserving the rigor of the randomized comparison.

Beyond mortality, the treatments produced little evidence of clinical benefit. Serious adverse events occurred in 138 children, with 77 events in the pancreatic enzyme group and 76 among those assigned to its placebo, and 79 in the bile acid group compared with 74 in the bile acid placebo group. Readmissions were also not reduced. Four children experienced grade 3 or 4 toxicity during the intervention period, although none of the events was judged attributable to the study products. Children assigned to bile acids spent fewer days receiving second-line antibiotics: 677 days in total compared with 832 among placebo recipients, corresponding to an incidence rate ratio of 0.78. That isolated finding was not accompanied by improvements in survival, hospital stay, sepsis, diarrhoea or third-line antibiotic use. Growth measures also failed to distinguish the treatment groups. Mid-upper-arm circumference, weight-for-height and weight-for-age generally improved by day 60, but enzyme and bile acid supplementation did not produce additional gains; height-for-age declined overall.

The researchers caution that the negative results do not prove pancreatic enzymes or ursodeoxycholic acid could never help any child with malnutrition. The study tested empiric treatment in children who were acutely ill, severely malnourished and often experiencing sepsis, dehydration, pneumonia, gastroenteritis or multiple forms of organ dysfunction. More than half of the deaths occurred within eight days, possibly before therapies intended to improve nutrient absorption, bacterial ecology or gut-barrier function could exert meaningful effects. The trial also did not measure pancreatic insufficiency directly, track changes in the microbiome, assess small-intestinal bacterial overgrowth or characterize the bile acid pool. Adherence after discharge was estimated from medication-bottle weight rather than confirmed biologically, and ranged approximately from 60 to 84 percent, leaving open the possibility that some children received less treatment than prescribed. Even so, the multinational design and rigorous placebo-controlled methods provide strong evidence against routinely adding either intervention to standard care for this critically ill population. The findings point instead toward integrated, phase-specific strategies combining medical, nutritional, social and health-system interventions rather than relying on a single digestive or microbiome-targeted therapy.