How do probiotics prevent NEC?

We have reasons for thinking that bacteria may have a role in NEC and that the abnormal pattern of bowel colonisation seen in preterm infants is a risk factor,6 but we have a very incomplete understanding of its pathogenesis.7 We do not know if NEC is a single condition or a clinical entity with multiple causal pathways. We can only speculate on how probiotics might prevent NEC. We do not know if it is placement of large numbers of probiotic bacteria into the upper gastrointestinal tract that is important for probiotic efficacy or colonisation of discrete parts of the gastrointestinal tract. We do not know the ideal timing of the intervention, which microbial strains provide the benefits, or the importance of interactions with other preventive strategies for NEC, such as modified feeding practices.8 The current selection of probiotics used in clinical trials is not based on a clear understanding of pathways of pathogenesis or mechanisms of action.

Is any probiotic strain as good as any other?

A large number of different bacterial and fungal strains have been used in probiotic studies, usually in combinations. An example of one strain that has been as extensively used in RCTs as any other is Lactobacillus GG.9 Trials using Lactobacillus GG have not always shown benefit in the probiotic arm. An RCT comparing Lactobacillus GG with placebo in a paediatric intensive care unit (ICU) was stopped because of a non-significant trend towards an increased risk of infection in the probiotic group.10 A neonatal unit in Finland has used Lactobacillus GG prophylactically for very low-birth-weight infants since 1997; when the incidence of NEC was compared with other units, it was highest in the unit using prophylactic Lactobacillus GG.11 Bloodstream infection caused by Lactobacillus GG is well recognised,12 and blood stream infection caused by Lactobacillus GG has been reported in infants with short-gut syndrome.13 In the two studies that used Lactobacillus GG reported in the recent meta-analysis of probiotics for the prevention of NEC,1 the relative risk of blood culture positive for sepsis was (non-significantly) higher in the probiotic than placebo group,14 15 although all-cause mortality was reduced in the probiotic group. These reports do not preclude the use of Lactobacillus GG in infants but do suggest that Lactobacillus GG administration can be associated with adverse and beneficial outcomes and that Lactobacillus GG may not be the ideal probiotic for preterm infants.

Different probiotics are likely to have different degrees of benefit or detriment for different outcomes and may interact with each other differently in different combinations. Placebo-controlled trials should provide a clearer indication of which probiotics produce which effects than head-to-head comparisons; the published trials make little mention of the rationale for the organism or the combinations selected for testing.

Is a probiotic intervention a standardised intervention?

We do not know if the efficacy of probiotics requires administration of mixtures or single strains. The conclusions from the recently published meta-analysis1 have been questioned because the analysis required pooling of studies using different probiotics, mixtures, and doses in different populations and so may not be generalisable to probiotics as a class of interventions used in differing populations.16

The degree of variability in standardisation reflects the lack of availability of products that comply with pharmaceutical good manufacturing practice and the associated lack of regulatory guidance governing the use of probiotics as therapeutic agents. In Canada, the manufacture, packaging, labelling and importation for sale of probiotics are regulated by the Natural Health Directorate of Health Canada. To be sold in Canada, all Natural Health Products must have product licences, and the Canadian sites that manufacture, package, label and/or import Natural Health Products must have a site licence. The same group have produced a probiotic monograph that includes detailed information on acceptable health claims, doses, source materials and required risk information (http://www.hc-sc.gc.ca/fn-an/label-etiquet/claims-reclam/probiotics_qa-qr_probiotiques-eng.php). We have not been able to identify similar guidance or regulatory requirements in Europe or any international regulations governing the study and the use of probiotics as therapeutic agents. Widespread use of probiotics in vulnerable infants requires that there is standardisation of products, regulation of manufacturing, storage and distribution and a sufficiency of evidence of safety for the defined products chosen for use.

To what extent do probiotics cross-colonise infants, and does cross-colonisation have implications for neonatal outcome?

The routine administration of probiotics to any group of babies in the nursery has implications for all. Cross-colonisation of infants nursed in neonatal ICUs with potential pathogens is well recognised.17 18 Less is known about the extent to which probiotic strains cross-colonise infants in the neonatal ICU, but it is hard to believe that routine use of probiotics will not result in colonisation of some infants to whom probiotics have not been prescribed.

Early probiotic studies reported cross-colonisation of control group infants with probiotic strains.19 20 Unfortunately, the extent of colonisation within the control group(s) has not been reported in more recent studies even in those in which stool colonisation was investigated.21,–,23 This may reflect difficulties in clearly distinguishing probiotic strains from other colonising microflora. The consequences of cross-colonisation of infants in the nursery who are excluded from trials have not been reported; the reasons for exclusions from the published trials included those with severe asphyxia, with chromosomal abnormalities, with gastrointestinal or abdominal wall abnormalities and who were formula or breast fed (depending on the trial); infants were not generally exposed to the trial intervention until the start of enteral feeding. All of the excluded groups are probably at risk of probiotic colonisation either deliberately or inadvertently should probiotic use become routine use. Some groups of patients such as those with gastrointestinal diseases may be particularly at risk of adverse consequences associated with probiotic administration. Mortality (attributed to bowel ischaemia) in a recent Dutch study of the impact of enterally administered probiotics on infectious complications in adult patients with acute pancreatitis was twice as common in the probiotic group than the placebo group.24 Bloodstream infection caused by bifidobacteria had not been reported until a single case of bloodstream infection caused by Bifidobacterium breve was reported in an infant with omphalocele.25

It is suggested2 that while routinely giving probiotics to babies who would have been included in the published trials, we continue to undertake studies of those excluded babies at higher clinical risk; in routine practice, this would only be possible if great care was taken to avoid cross-colonisation (because of the transmissible nature of probiotics). Placebo-controlled trials can provide evidence of cross-colonisation and the effectiveness of measures taken to limit cross-colonisation.

What adaptations will probiotic strains undergo?

There is evidence that some types of early antibiotic exposures increase the risk of NEC?26 27 It is possible that colonisation with antibiotic-resistant microbes is an important element either in the pathogenesis of NEC or in the development of more severe stages of NEC. Colonisation of the small intestine with antibiotic-resistant bacteria would reduce with the ability of antibiotics to reduce the microbial load (and associated inflammatory stimuli) in the small intestine. We do not know if antibiotic susceptibility of the bowel flora is an important and independent determinant of the onset and clinical course of NEC. It is certainly plausible that degrees of small intestinal colonisation are important in the pathogenesis of NEC and that the earlier that there is a reduction in microbial numbers in the lumen of the small intestine in response to antibiotic therapy, the better the outcome.

Probiotic bacteria are usually relatively sensitive to β-lactam antibiotics,28 which are the most widely used group of antibiotics in neonatal practice.29 Some antibiotics (particularly some β-lactam antibiotics) reach high concentrations in the small bowel through biliary excretion. Is domination of the small intestine with antibiotic-sensitive probiotic bacteria an important element in probiotic efficacy? Probiotic bacteria can adapt to survive under adverse conditions.30 One study reported the acquisition of tolerance to bile salts by bifidobacteria associated with increasing resistance to antibiotics.31 When probiotics are enterally administered, large numbers reach the small intestine and mix with bile and the antibiotics excreted in the bile. There will be enormous selection pressure on probiotics to develop antibiotic tolerance or resistance. Probiotics will also be selected for their ability to colonise and persist in preterm infants. What will these adaptations be and what will be the consequences of these adaptations for continuing probiotic efficacy? Concurrent use of probiotics with antibiotics may well lead to the selection of antibiotic-resistant probiotic strains. It may be that these adapted strains will become dominant colonisers on the unit. What will the impact of that change have on infant outcomes? It is also possible that probiotic bacteria will acquire new genes from other microflora and that may also alter their propensity to cause disease.

These types of changes are unpredictable, and we have little previous experience on which to base reassurance. The situation becomes increasingly complex when we start to consider multiprobiotic interventions as have been used in many of the studies in preterm infants.

What are the ecological consequences likely to be?

Selective decontamination of the digestive tract (SDD) with antimicrobials reduces the risk of ventilator-associated pneumonia and death in adults admitted to ICUs,32 yet routine use of this intervention is limited to a minority of centres in the UK and few within the USA. Much of the reluctance to implement SDD relates to concerns about the selection of antibiotic-resistant strains. Despite the clear benefits of this intervention, it is still considered that there is a state of “clinical equipoise regarding this issue”33 because the benefits are balanced by concerns about the uncertain implications of selecting antibiotic-resistant strains in individuals and within ICUs. The selection of antibiotic resistance by an antibiotic intervention would not be surprising, yet despite the use of SDD for >10 years until a recent study from The Netherlands (where SDD is widely used), few studies have specifically investigated these effects. The Dutch study reported marked effects of SDD on intestinal colonisation with resistant gram-negative bacteria. The proportion of patients with resistant strains increased during the intervention and increased to higher levels than preintervention when the intervention was discontinued.34 An important message from this type of study is that ecological changes in an ICU will only be recognised if the microbial ecology is actively monitored. Recent probiotic studies have not reported the monitoring of microbiological or ecological consequences, such as changes in colonisation with antibiotic-resistant strains.

It is likely that ecological changes will take place as a consequence of probiotic interventions. Other microbes will adapt to the intervention. The bowel is probably the most active area for microbial genetic exchange in the human body.35 Relative to species with more pathogenic potential, not much is known about the genetics of probiotic bacteria. Hehemann et al36 recently reported that Japanese individuals carry bacteria that degrade seaweed. The porphyranase enzymes that are important in this process are the same as those found in bacteria that live on seaweed in the natural environment. The suggestion is that bacteria that colonise humans have acquired these genes from seaweed-colonising bacteria in transit through the human gastrointestinal tract. We do not know the extent to which probiotic bacteria can exchange or facilitate exchange of colonisation factors, virulence or antibiotic resistance genes with other bacteria, either resident or in transit, through the gastrointestinal tract. We do not know if probiotic bacteria will facilitate colonisation or stabilise colonisation with antibiotic-resistant strains. The possibility of adverse ecological consequences requires that we make efforts to monitor the ecological consequences of probiotic interventions.

What are the developmental consequences of probiotics?

The use of probiotics is a preventive not a therapeutic intervention, so it will be used in infants, most of whom will never experience the disease(s) that the intervention is designed to prevent. It is also an intervention that, unlike use of drugs such as steroids, has the potential to adapt and also to spread to others. Use of probiotics may, with experience, turn out to be a very safe and effective intervention, but we should be cautious and not assume safety on the basis of relatively short-term studies in selected groups of infants. “We should be vigilant about interfering with systems we poorly understand … There is little known about the effect of antibiotics on early patterns of microbial colonisation of newborn children, which might have important, long-lasting consequences for early human development”.37 The same can be said of the use of probiotics. Patterns of early mammalian colonisation may be important determinants of development and health later in life.38 The consequences of a sudden exposure of the preterm gut to a significant and continued microbial stimulus is likely to have important effects that have not so far been studied in clinical trials. While it is understandable that the focus of discussion on the effect of probiotics has emphasised the prevention of NEC, other direct and more widespread effects on the neonatal gut cannot be ignored. Interactions between bacteria and their animal hosts is a universal theme in developmental biology and may apply equally to the preterm infant. There is a substantial amount of evidence that bacteria, particularly gram-positive bacteria, have profound influences on the developing immune system and intestinal angiogenesis in different animal models.39 Studies of the effects of probiotic, colostrums and formula administration on caesarean-delivered preterm pigs (physiologically quite similar to preterm infants) found major physiological effects including increased intestinal weight, mucosa proportion, villus height, RNA integrity and brush-border aminopeptidase A and N activities in probiotic- and colostrum-fed piglets compared with those who were formula fed.40 This study illustrates the potential impact of probiotic feeding on early development and also the potential for interactions with other preventive interventions such as breast milk feeding.

Precaution in the use of probiotics in the preterm

More than 50 years ago, Staphylococcus aureus 502A was used to control outbreaks of S aureus infection in US neonatal units. Thirty-eight (5.9%) of 644 infants colonised with S aureus 502A in one unit in Dallas developed infection with the probiotic strain and one infant died: “The fatal case emphasises that bacteria of extremely low virulence may produce serious disease in compromised hosts”.41

The precautionary principle states that “when there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation”.42 This principle has been adapted and included within diverse areas of human decision making under conditions of uncertainty—for example, within frameworks for public health ethics.43 An important element within a precautionary approach is to identify and acknowledge areas of uncertainty and that we do not assume safety because of a lack of evidence of harm. A precautionary approach to the use of probiotics in preterm infants also requires that we monitor effects, particularly when potential changes are not predictable because of complexity, as is found in the biological interactions that take place between microbes and host in the human bowel.

NEC is one of the most devastating consequences of preterm birth. The recent meta-analysis strengthens the evidence that probiotic administration may have an important role in prevention. We would argue that there is an urgent need for more well-designed placebo-controlled clinical trials with adequate statistical power and sufficient duration of follow-up individually to provide clear clinical answers. Trials should use defined products produced to an agreed and regulated manufacturing standard. These trials should include the small growth-restricted infants at highest risk of NEC. Probiotics should be used in conjunction with processes and procedures to monitor the microbiological and ecological effects. The frequency and the outcomes of infants not included in the trial but at risk of colonisation should also be determined.