Resources & Evidence Contact us

PNG Conference: Probiotics: The case for routine prophylaxis in preterm infants

Results and outcomes described are anecdotal and specific to the institution and its protocol. For a list of published studies of EHMD, click here.

Learning Objectives:

  • To understand the dysbiosis in the pathogenesis of NEC and sepsis
  • To evaluate the risk vs benefits of probiotics in the care management of preterm infants
  • Discuss neonatal evidence-based medicine


Mark Underwood, MD
UC Davis, Chief of the Division of Neonatology


Hi, everyone. Welcome to the conference. My thanks to the folks at Prolacta for sponsoring this conference and particularly to Dr. Ram for the invitation to speak with you. My topic is probiotics: the case for routine prophylaxis in preterm infants.

I have no financial conflicts to disclose but I will be discussing the off-label use of human milk and probiotics in preterm infants.

So here's our outline. We'll start with some definitions, and then we'll talk about dysbiosis, particularly in the pathogenesis of NEC and sepsis. And then we'll talk about probiotics, and particularly their risks and benefits in preterm infants, and we'll finish with a quick discussion on neonatal evidence-based medicine.

So a few definitions: the word “microbiota” refers to the community of microbes that inhabit a specific niche―in our case, we're talking about the intestinal microbiota. “Colonization,” meaning new organisms moving into an anatomic location―in our case, colonization of the intestinal tract begins in utero, but really is most highly impacted by ruptured membranes, birth, and then initial feedings, antibiotics, and environment. And I'll make the distinction between vertical transmission, meaning from mother to baby versus horizontal transmission, which is from other folks and the environment to the baby. A “probiotic” is a dietary supplement that contains live healthy bacteria designed to improve health in some way, and a “prebiotic” as a dietary supplement that contains usually glycans, ideally, that are a food source for healthy bacteria, and not for pathogens. And then the word “dysbiosis” is kind of a new word. By dysbiosis, I mean an alteration in the microbiota of a given anatomic niche that's associated with disease. So here's an example: antibiotic exposure, we know, alters the intestinal microbiota, and that can lead to antibiotic-associated diarrhea, which is fairly common, or even Clostridium difficile colitis, which is fortunately less common, but can be devastated. For both of these disease processes, antibiotic-associated diarrhea and C. difficile colitis, association and causality are well-established.

Here's a schematic that sort of identifies an important association. So the key question really is, is intestinal dysbiosis associated with diseases that are either within the gut itself or systemic? And so if we start at the right of the slide, many interventions have been shown to alter the intestinal microbiota. This includes diet for adults and children. That means, for instance, that high-fat high-sugar western diet for babies, that means formula versus mom's milk, antibiotic exposure, delivery by C-section, and a wide variety of hygiene practices all have been shown to alter the intestinal microbiota, creating dysbiosis. And in this case, I mean an increase in pathogens like Enterobacteriaceae, which is the family of gram-negative organisms that include those well-known to neonatologist, such as E. coli, klebsiella, Pseudomonas, Proteus, serratia, etc, and decreases mutualists like bifidobacterium. So mutualists, or sometimes people call them commensals, are healthy bacteria that benefit the host. In this case, the baby. Though it changes in the intestinal microbiota or intestinal dysbiosis lead to increased intestinal permeability and increased systemic and local inflammation. And intestinal dysbiosis has been associated with a wide variety of diseases including those of the brain, lung, liver, kidney, immune system, and localized to the gut as listed on the far right, and so all of these disease processes have significant associations with intestinal dysbiosis.

So we're going to pause here and talk about human milk oligosaccharides. As you can see, on the left, human milk is mostly water, contains a variety of macro and micronutrients, plus some other bioactive molecules. When we look just at the macro and micronutrients, mostly lactose, a lot of fat. And then, in addition to protein, there's a pretty high component of what are called human milk oligosaccharides. These are small sugar molecules that interestingly are not digestible by the human gut. In other words, babies don't have the enzymes necessary to digest these oligosaccharides and break these linkages to use them as a food source. And so the interesting question is, of course, why is mom expending tremendous energy to create these complex molecules and put them into the milk when they have no nutritional value for the infant and the answer of course, is that these oligosaccharides are selectively consumed only by the microbes of the gut. And interestingly, not all microbes, but very few number of genera and species.

So, here's an example of how highly selective these human milk oligosaccharides can be. I'm showing you here two different bifidobacterial species or subspecies, on the left as bifidobacterium bifidum, which has a variety of membrane-bound glycosyl hydrolases, meaning on its membrane facing extracellular space are a variety of different enzymes that can break down most human milk oligosaccharides in an extracellular fashion, and then these monosaccharides are then transported into the cytoplasm of the bifidobacterium bifidum to be used as a food source. Bifidobacterium infantis, on the other hand, can import a variety of different human milk oligosaccharides into its cytoplasm, where it has all of the glycosyl hydrolases within its cytoplasm, to completely digest all of the human milk oligosaccharides structures. And so these human milk oligosaccharides are digestible by only a limited number of bacteria. And those bacteria use different mechanisms to transport and then digest them as a possible food source. So HMOs are a prebiotic, and they're highly selective. In other words, the mother is producing these sugar molecules as a way to shape the intestinal microbiota of her baby at great cost to herself.

Here's just a brief overview of how we think about the intestinal bacteria. So if we if you remember that there are nine bacterial phyla that inhabit the humans, and among those, the ones that are most commonly found are highlighted here. So in the intestine of the breastfed infant are a lot of actinobacteria. That's the phylum level. Bifidobacterium is the gen a genus. These are gram-positive organisms that are very effective consumers of human milk oligosaccharides. Also common in the gut of the breastfed infant are bacteroides. And these are gram-negative organisms that are also some of them are able to consume HMOs. I've also highlighted the Firmicutes, which are gram-positive organisms, like staph and strep. These are interesting because they're not able to consume human milk oligosaccharides. And the proteobacteria, which are the gram-negative organisms, the family that's most important in humans is Enterobacteriaceae. And as I mentioned previously, this includes a general like E. coli and klebsiella, and Pseudomonas. These are all non-HMO consumers, meaning they're not able to use human milk oligosaccharides as a as a food source.

So this is a study that we did in Bangladesh, this is 48, breastfed term infants. And at three different time points here six weeks, 11 weeks, and 15 weeks, we characterize their microbiota at the phylum level, and you can see they're dominated by actinobacteria. And when we looked at the genus level, it was all bifidobacterium. And particularly at the species level, almost all of this was bifidobacterium infantis. B. infantis, is an organism that really has an advantage over other microbes in its capacity to digest human milk. And these data are quite consistent with a study that was done over 100 years ago now, where an investigator looked at the feces of breastfed healthy infants. And he saw what he described, this is under the microscope, of course, as an almost pure culture, gram-positive if it or biofit bacteria, which later became known as the bifidobacteria.

And so this is just a summary from multiple different locations of what the fecal microbiota of a healthy breastfed infant looks like in different countries. If you look at the far left, that's our study that I mentioned on the last slide of Bangladesh. And you can see, these babies are dominated by bifidobacteria, with more than 70% of all the bacteria present in the feces in that category. If we look over to the far right, a second to the right, California, this is a study we did in Davis, California, among a group of very healthy moms who are non-smokers and breastfed their babies. And what you see here is a low number of bifidobacterial and a high number of proteobacteria, which is at the phylum level at the family level, it was Enterobacteriaceae. And the message here is that there's a wide degree of variation across the globe, and dysbiosis among healthy breastfed infants, even those not exposed to antibiotics is actually quite common.

This is a paper that we did just a couple of years ago. Where we looked at all the studies we could find over the past 100 years that described the microbiota of the healthy breastfed infant. So these are all studies of poop of breastfed babies, healthy breastfed babies. Now, of course, 100 years ago, they didn't have the sorts of tools we have today to identify a variety of bacteria. They were relying mostly on microscopy, and fecal pH and what would grow in culture. And so because those, those tools for identifying the fecal microbiota have changed, we focus just on fecal pH in this paper. Because pH hadn't the methods of measuring pH haven't changed in 100 years. And what we see on the figure on the right, is that the fecal pH of the healthy breastfed baby over the last 100 years has increased from about five to about six. And on the on the left, what I'm showing you here is which bacteria in the feces are associated with either an increase in fecal pH or a decrease in fecal pH. And what we see here is that the family bifidobacteriaceae tend to decrease fecal pH, whereas these others like Clostridium and bifidobacteriaceae, tend to increase the fecal pH. So in summary, what we're seeing over 100 years is an increase in fecal pH. In other words, the feces have become less, less acidotic, suggesting and consistent with a decrease in the fecal bifidobacteriaceae.

This is one of the best studies today looking at colonization of the preterm gut, particularly in very preterm babies. And so what this group in St Louis did was they collected 922 fecal samples from 58 preterm babies, and then they did something that only a neonatologist would do, they corrected each sample for gestational age. So a baby born at 24 weeks who had a poop specimen at 28 weeks. They called that a 28-week specimen. And what we can see here is that the bacilli, which includes the gram-positive Firmicutes, staph and strep, start out very high at 24 to 27 weeks, and then by 28 weeks have become much lower relative abundance, whereas clustered do just the opposite early in life, they're very low. And by about 34 weeks, the median number of clusters media has increased to almost 50% of the bacteria. The important slide is the bottom one in red, the proteobacteria tend to start low in the most premium return babies, but between about 27 weeks and about 33 weeks, proteobacteria which includes E. coli and klebsiella are the dominant organism in many very preterm babies.

So here's our take home lesson number one, the intestinal microbiota of the term infant is shaped by and co-evolved historically with lactation to protect the infant. So in other words, moms are expending a tremendous amount of energy to put these human milk oligosaccharides and other factors that shaped intestinal microbiota into her milk. Second point, intestinal dysbiosis is common and seems to be increasing particularly in developed countries and is associated with many inflammatory diseases. Third point, the very preterm infant is an evolutionarily new species. And so of course prior to the 70s and 80s, preterm infants rarely survive in order to reproduce. That's particularly true of babies less than 28 weeks. Final point, is there is no normal gut microbiota for the very preterm infant. In other words, dysbiosis as we define it, meaning an increase in the number of Enterobacteriaceae, and a decrease in the number of commensal organisms is universal in very preterm babies. In other words, it's extremely uncommon to find bifidobacteria and babies stools who are less than 28 weeks. On the other hand, they have lots of E. coli and klebsiella.

This is just a quick summary of one of the important points we know about necrotizing enterocolitis. So this is a nice study by the group in Canada, they looked at almost 17,000 infants with gestational age less than 33 weeks from 25 Canadian NICUs. The incidence of NEC was about 5%. The blue line won't surprise you. These are the gestational age or birth of the babies who develop NEC, and it's mostly babies born less than 32 weeks, look at NEC. The red line is the important one. This shows us the postmenstrual age at the onset of NEC, and you can see that somewhere between 27 and 33 weeks is when the highest incidence of necrotizing enterocolitis occurs. So if we line up those two studies that I've shown you previously, what we see is that the peak fecal proteobacterial level, or relative abundance, corresponds to the NEC, peak NEC, incidence of 27 to 28 weeks out all the way to 33 weeks, correct gestational age or postmenstrual age.

This is one of several studies, this one also by the group in St. Louis, looking at the feces of babies who have preterm babies, in this case less than 27 weeks at birth, who either developed NEC which is the cases on the left side of the screen, or did not develop NEC, which is the controls on the right side of the screen. And what we see among those babies who develop NEC, these are all from samples that were obtained before the baby's had NEC, we see this bloom of proteobacteria that occurs just prior to the onset of NEC. And we'll look at the controls. What we have here are matched samples. At first corrected gestational age. Of course, these babies never developed NEC and they didn't have a bloom or a high predominance of proteobacteria.

This is just a summary of six studies looking at the impact of Group B strep prophylaxis, on the intestinal or fecal microbiota of the infant. This, of course, includes mostly term infants. And what we see when a mom gets Group B strep prophylaxis, her baby has a higher relative abundance of proteobacteria, including the Enterobacteriaceae we've been talking about. And her baby has a lower relative abundance of the actinobacteria, particularly the bifidobacteriaceae and bifidobacteria general that we've been discussing.

This is just a summary of six studies that have showed the impact of antibiotic administration on the outcomes for very low birth weight infants. And what you can see that the overall take home message here is that the more days of antibiotics that a baby gets, particularly in the first two weeks of life, the higher the risk of later necrotizing enterocolitis, late-onset sepsis, and death. And so antibiotics clearly have an impact. This is not a course to suggest that we shouldn't be treating babies with sepsis with antibiotics. But it does suggest that antibiotic stewardship and careful use of empiric antibiotics is valuable in that it decreases the risk of necrotizing enterocolitis, late-onset sepsis, and death.

This is just two studies looking at the impact of acid blockade. So on the left side of the slide, we're seeing the impact of an h2 blocker on the fecal microbiota. So the black circles are babies who received an h2 blocker of some type. And you can see they're somewhat scattered, but the median relative abundance of gamma proteobacteria, these are the gram negatives that are associated with disease, is around 75%. The white circles are the babies who never got an h2 blocker. And while again, there's a big scatter, the relative the median relative abundance is about 30%. Quite a dramatic difference. And then on the right, we see a case control study 787 babies with NEC and over 2000 match controls. What they found was an increased risk of NEC with the use of an h2 blocker with an odds ratio of 1.7. And the average time from starting the h2 blocker to the onset of NEC was 19 days.

And this course is a terrific study done through the neonatal network where they looked at over 1000 ELBW babies, these are babies born at less than a kilo. And they looked at their intake of human milk in the first 14 days of life, and the risk of later developing necrotizing enterocolitis or death. And this is this is kind of an interesting slide. So the figure, note that the y-axis is the proportion of babies who never got NEC. And the x-axis is the age in days. And each line represents the percentage of human milk that they got in the first two weeks of life. So the 1.0 or top line are babies who receive nothing but mom's milk, and zero the bottom line, were babies who received nothing but formula in the first two weeks of life. And what we see here as a dose response, the more human milk they got in the first two weeks of life, the less likely they were to get NEC, the babies who got just mom's milk had about a 3% incidence, whereas the babies who got just formula around a 14% incidence. Quite a difference.

So here's take home lesson number two. Intestinal dysbiosis precedes and is associated with NEC. There's an increased risk of NEC in babies who receive antibiotics, particularly empiric antibiotics and acid blockers, and a decreased risk of babies who receive human milk especially in the first couple of weeks of life.

So now let's just talk about probiotics and NEC This is a meta-analysis of published studies of rodent models. The top ones are rats, the bottom two are mice. And what we see here is these studies all showed that if you gave a baby, she should be a rat pup, probiotics you decrease the incidence of NEC among those pups who receive probiotics after birth. There, of course, is a significant publication bias here, people tend not to publish papers where they don't see a difference. But all of these studies show the benefit of probiotics in preventing NEC and these rodent models.

This is a meta-analysis from 2018. Looking at all of the published studies to date, both randomized controlled trials and observational or cohort studies, looking at probiotic administration to premature infants, and we're looking at three outcomes NEC, death, and late-onset sepsis. And what we see here is that in in over 8000 babies in more than 20 randomized control trials, we have a decrease in NEC with a with a risk ratio of 0.57, a decrease in death with a risk ratio of 0.77, and a decrease in late-onset sepsis with a risk ratio of 0.88. I've also included all of the cohort studies published up to this point. And you can see over 10,000 babies in cohort studies with risk ratios for both NEC, death, and late-onset sepsis, that are quite similar to the randomized control trials. So here we have more than 20,000 premature babies who have been in studies of probiotic administration, either randomized control trials or observational cohort studies, showing a significant decrease in NEC, death, and late-onset sepsis. Investigators have also done meta-analyses of other outcomes in randomized controlled trials of probiotics in preterm infants. And what we see is on the top line, their time to full feedings was 1.5 days sooner for the babies who received the probiotics compared to the placebo infants. The other lines look at BPD using a definition at 28 days or 36 weeks corrected, and RLP and we see no difference in outcomes.

And here we're looking at weight gain and intraventricular hemorrhage. And we see no difference between the babies who got the probiotic and the babies who got the placebo.

This is another meta-analysis that looked at comparing just the randomized controlled trials. And the outcome here is either stage two or stage three NEC comparing a probiotic supplement to placebo. And you can see there's quite a wide variety here, some of them significant, some of them not significant. I put red arrows next to the probiotic products that had randomized control trials that showed a decrease risk of NEC; you can see there's a variety of different probiotic strains and species, some of which showed a protective effect and some of them did not.

So here's take home lesson number three, in randomized controlled trials of more than 8000 preterm infants, probiotics decrease the risk of NEC, death, and sepsis. Decreased time to full feeds and length of hospital stays but did not impact BPD, retinopathy of prematurity, intraventricular hemorrhage, weight gain, or neurodevelopment. Second take home lesson point here is the ideal strain combination and dose of probiotics remains unknown.

So we're left with the important question, looking at the risk versus benefits of routine prophylactic probiotics to preterm infants. The advantage to routine prophylaxis is decreased risk of death, NEC, sepsis, and decreased feeding intolerance. The downside or the risks are on the left side of the screen. The potential here for probiotics, sepsis, that means a baby who becomes ill, and we do a blood culture and it grows the very same strain that they've been ingesting, meaning that that probiotic species has translocated from the gut lumen into the bloodstream. There are reported cases for almost every probiotic strain. But these cases seem to be extremely rare. And when you look at the meta-analysis, the incidence of sepsis is actually lower among the babies who got probiotics than among the babies who got the placebo. Second risk here is contaminants. There are cases where probiotic products have become contaminated, particularly the famous one with a fungus and the baby developed a fungal sepsis and died. These case reports are well documented in the literature but seem to be extremely uncommon. I think it speaks to the necessity of having increased oversight into probiotic production in the US like is common in other countries. The next risk is errors in labeling. When we look at probiotic products that you can buy over the counter, it's uncommon to find other organisms in the probiotic that are not on the label. And then final point in terms of risk is we do have this uncertainty remaining regarding which strain or combination of strains and what dose is optimal.

So finally, let's just talk about evidence based neonatology. We pretend that we have lots of evidence to support all that we do. I've just put a couple of interesting things on the table here. On the left are common practices in neonatology where we have a strong evidence base. So top of the list here is surfactant decreases mortality. We have a wide variety of studies, comparing surfactant to placebo showing that the preterm babies with RDS who get surfactant to have decreased mortality. In the same category is strong evidence that human milk decreases the risk of necrotizing enterocolitis. That cooling modestly improved outcomes in hypoxic ischemic encephalopathy in developed countries. Interestingly, we don't have that data for low and middle income countries. And in the same category, probiotics decrease the risk of NEC and probiotics decrease the risk of death. On the right side of this table are common practices in NICUs, for which we don't have an evidence base. Diuretics for either prevention and treatment of BPD are common, but we really don't have strong evidence that it's helpful. Treating NEC stage two with stopping feedings gastric session and antibiotics is the standard of care. But interestingly, we have no evidence base that that improves outcomes. No one has ever done and nor will they ever do a trial where kids with documented NEC are randomized to either antibiotic or no antibiotic treatment. But we do know that NEC can be caused by viruses. And so if we did the study, we might find that the antibiotics are not particularly helpful. We also have an established approach to neonatal resuscitation with a ratio of three to one for ventilations and chest compressions. However, that's not based on any evidence and that it's superior to other sorts of ratios. Interestingly, Carsey challenge is common as a discharge criteria for the NICU. And yet, when you go looking for evidence to support that the car see challenges improve any outcome, I just don't find it. Finally, I've noted here that most NICUs have some kind of policy as to how many days free of bradycardia and deaturation episodes. A NICU requires in order to discharge and interestingly, we don't have a good evidence base to support any of those recommendations.

The summary here, of course, is that we do a lot of things in the NICU, some that we have evidence for some that we don't. We have more evidence for the prophylactic benefit of probiotics than many interventions, and stronger than a lot of things that are common in the NICU. Finally, I'll just note, this important statement that parents are not visitors in the NICU. This has become particularly important with the core Coronavirus outbreak. But I would advocate that parents are an essential part of the NICU team; that they need to be involved in the care of their babies and in decision-making. When we talk to parents about necrotizing enterocolitis, they want and deserve a frank discussion about NEC, human milk, and probiotics. So I would encourage you to include parents in the discussion as you decide whether or not to provide probiotic prophylaxis and whether or not to use a donor milk. I'll conclude by just suggesting that you view the NEC Society website. This is a wonderful, nonprofit organization that supports and includes parents in both research and understanding of necrotizing enterocolitis. They're having of course a wonderful NEC symposium, this will be their third opportunity to sponsor next opposing coming up in May and registration is open now.

Thanks very much for listening.