Pediatric Short Bowel Syndrome: Redefining Predictors of Success

Abstract

A review of 80 pediatric short bowel syndrome patients demonstrated that cholestasis and age-adjusted bowel length, but not the ileocecal valve, are the main predictors of mortality. The combination of bowel length <10% of predicted and cholestasis results in such high mortality that early referral for intestinal transplantation should be considered.

Infants and children with short bowel syndrome (SBS) suffer extensive morbidity and mortality.^1–5 SBS is generally defined as the loss (either congenital or acquired) of an extensive length of the small intestine resulting in an inadequate absorption of enteral nutrients.⁶ These children are often dependent on parenteral nutrition (PN) for long periods of time, which leads to high rates of sepsis, cholestasis, and mortality.⁷ Prior to the advent of PN, of course, SBS carried a dismal prognosis.⁸ After dependence on PN for a variable period of time, some pediatric SBS patients demonstrate bowel adaptation and are eventually able to wean off PN.^9–12 One early study suggested that survival in SBS required at least 15 cm of small intestine and an intact ileocecal valve (ICV) or at least 40 cm in the absence of the ICV.⁹ More recently, however, these criteria have been called into question, making it difficult to determine which infants will survive, or wean off PN, without an intestinal transplant.^7,10,11

Risk factors thought to contribute to the high morbidity and mortality of SBS include the length of remaining small bowel, sepsis, loss of the ICV, and development of PN-associated cholestasis.^12–15 Unfortunately, a major limitation in conclusively identifying these risk factors is the relative infrequency of SBS. Most previous reports on children with SBS have been limited by relatively small numbers of patients or have not used adequate statistical methods to determine which factors influence survival and the ability to wean off PN. The current study was designed to use rigorous statistical analysis of a large sample of infants with SBS to determine which factors influence SBS outcomes. In the primary analysis, we examined predictors of survival in SBS; and in the secondary analysis, we examined predictors of weaning off PN.

METHODS

RESULTS

DISCUSSION

SBS in infants and children is a complex and highly morbid condition. These patients suffer repeated episodes of sepsis, dehydration, and metabolic derangements and often require multiple hospital readmissions.¹³ PN constantly demands exacting attention to detail and carries the dreaded risk of PN-associated liver disease.

Multiple studies in the literature note similar survival rates to our study.^22–24 In a recent large population-based study, the risk of death from SBS and its complications in children was 37.5%.²⁴ However, somewhat disappointingly, mortality rates reported here are not appreciably different from rates reported over 30 years ago.⁹ It is possible that our mortality rates are affected by the inclusion of extremely premature and critically ill infants not previously considered viable and thus not included in older reports.

The current study has found that key predictors of mortality in SBS include cholestasis (conjugated bilirubin ≥2.5 mg/dL) and percentage of small bowel length. On the other hand, the number of septic episodes, presence of the ICV, and etiology of SBS were not predictive of mortality. Our data show that percentage of small bowel length (percentage of the expected length at a given gestational age), rather than the absolute length of remaining bowel, accurately predicts survival of neonates with SBS. Although the importance of remnant small bowel length has long been recognized by a number of investigators in both pediatric and adult SBS,^9,22–25 rapid elongation of the small bowel during the latter half of gestation has confounded attempts at correlating absolute small bowel length with survival in infants with SBS. This may account for the discrepancy between our findings and other reports which suggest that small intestinal length does not predict survival.²⁶ We found that a small bowel length ≥10% of expected for a given gestational age¹⁷ was highly predictive of survival. The correlation is not absolute, of course, but 10% of bowel length is a strongly predictive cutoff point.

Conjugated bilirubin (≥2.5 mg/dL) is a very significant predictor of survival. This is supported by previous data.^12,26–28 In our analysis, a conjugated bilirubin level ≥2.5 mg/dL provided the optimal sensitivity (90.9%) and specificity (62.5%) for predicting mortality.

We found that presence of the ICV and percentage of small bowel length are the primary predictors of weaning PN. There was a significantly increased likelihood that patients who retained the ICV would be weaned from PN. The ileum has greater adaptability after bowel resection than more proximal intestinal segments, which may explain its importance in weaning off PN.^10,29–31 Alternatively, absorptive function of the remaining colon may contribute to adaptation after small bowel loss.³⁰ It was not possible to distinguish the effects of reduced colonic length versus loss of the ICV, due to the strong correlation between these 2 covariates for obvious anatomic reasons. Clearly, the ICV will also delay small bowel transit and may improve absorption.^31,32 Although an improvement in the ability to wean PN in patients with greater small intestinal length has been reported,^25,33 our data provide additional support for this concept.

It was interesting to note that the only factor correlating with the development of cholestasis was percentage of predicted small bowel length. Terra et al³⁴ suggested that small bowel remnants less than 50 cm correlated with catheter sepsis and bacterial translocation. As both of these latter complications may exacerbate cholestasis, reduced bowel length may contribute indirectly to development of cholestasis.

The striking incidence of failure to thrive in our SBS population was concerning. It suggests that we may not be adequately supporting our SBS infants, although overfeeding is also fraught with complications. In a smaller study of 16 pediatric patients on long-term PN, failure to thrive was a significant problem (12 of the 16) and may have been attributable to a lack of adequate protein supplementation.³⁵

In summary, cholestasis (conjugated bilirubin ≥2.5 mg/dL) and small bowel length <10% of expected length are the main predictors of death in this patient population. In addition, weaning of the SBS patient off PN was examined. For this secondary outcome, presence of the ICV and small bowel length ≥10% of expected length were significant predictors. These data may improve clinicians’ ability to assess the need for intestinal transplantation earlier in the course of SBS.

Discussions

Dr. Kathryn D. Anderson (San Marino, California): I have a follow-up of about 28 years of a patient who I operated on who had 20 cm of small bowel with no ileocecal bowel and also lost most of his colon. He weaned off TPN, obviously survived. Just in the last year his bowel is simply wearing out. It is not long enough, and he is facing now (25-plus years later) a small bowel transplant. Have you any really long-term data on these patients of yours?

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): I don't have anyone with quite that long a follow-up. My experience is more limited. But our clinic certainly has continued to bring in patients through Dr. Coran for the past 30-odd years, and what is evident is that these patients, once they are diagnosed with short bowel syndrome, even if they wean from parenteral nutrition, have life-long issues that they carry with them. I am often amazed how many nutritional anddigestive problems they carry with them that are not pickedup or sufficiently diagnosed. Your particular problem, though, I haven't encountered.

Dr. Robert J. Touloukian (New Haven, Connecticut): Dr. Teitelbaum and his colleagues in this elegant study on pediatric short bowel syndrome have used statistical analysis of over a hundred of their cases to redefine predictors of successful treatment. In doing so, they have actually reconfirmed the work of Douglas Wilmore, a member of this Association, who showed in 1972, literally only 3 years following the introduction of total parenteral nutrition, that the criteria for survival for full-term infants having major intestinal resection was 15 cm with the ileocecal valve present or 5% of estimated total intestinal length for full-term infants and 40 cm or 13% of residual bowel length in the absence of the ileocecal valve.

As has been pointed out by the authors, absolute residual length measurements are no longer a useful predictor of survival since nearly half of all pediatric short bowel patients are infants with operations for necrotizing enterocolitis having a birth weight of 1,500 grams or less and a gestation ratebelow 30 weeks, a situation which was unheard of in the early1970s.

I have a couple of questions regarding the inclusion criteria used in the study. The patients in the study were accrued from 1977 to the present time where treatment modalities, including safety issues with TPN, the choice of antibiotics and ventilator management to name just a few, are known to be highly significant predictors of survival for intact nonoperated babies with necrotizing enterocolitis. A breakdown of the statistical analysis for an early and a late group in a 26-year retrospective study would be vital to determine which other time-related factors may influence outcome.

The second question is: What definition is used for short bowel syndrome as the indicator for inclusion in the study? The authors state in their manuscript that the definition of short bowel syndrome varies widely amongst clinicians. But I believe it is more than a “extensive loss of the small bowel resulting in an inadequate absorption of enteral nutrients.” In a retrospective study such as this one, a minimum time of TPN support, such as three months for example, should be used as the entry point for further study of outcome. I would ask the authors what minimum time requirement for TPN was used in their initial definition of short bowel syndrome? Despite enormous advances in neonatology and the surgical care of small infants requiring major bowel resection, cholestatic liver disease is not much better understood now than shortly following the introduction of TPN.

Dr. Teitelbaum correctly points out that elevated conjugated bilirubin above 2.5 milligrams percent is a significant predictor of mortality. What was of interest to me is that prematurity alone is not an independent variable for predicting death or severe cholestasis in this study, a finding I had not previously thought to be true.

What we have learned is that reconstituting the residual bowel remnants to maximize length by early take-down and closure of the stomas and ensuring against surgical complications by meticulous technique in the constructing of the anastomoses gives the best chance to prevent TPN cholestasis. I would ask the authors if complications related to the wound and anastomosis and the time when the GI tract was finally reconstituted were studied and did that influence outcome? As surgeons we continue to be the principal variable in outcome by understanding the timing of operation, the choice of the procedure, the judgment on how much bowel to resect, and how rigorous we are in attempting to preserve the ileocecal valve.

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): Would there be a difference in outcomes if you look at some of the patients that were treated early versus late? As I said, I think, in the talk, 90% of our patients were less than a year of age when diagnosed with short bowel syndrome. So the bulk of our patients really came over the past 10 years. We did have some older patients that were born as early as 1977 and that did skew a few of our patients to a little bit older range. There is not a specific difference in survival in these patients treated earlier versus late.

I think what is even more dramatic, as you alluded to Dr. Wilmore's study in 1972, the survival rate in that group of patients is actually not significantly different than our survival of 27%. In some ways it is sort of sad that we have not really progressed that much in our overall mortality rate, but we are taking on a much more complex group of patients than in the early ’70s.

Our definition of short bowel syndrome was 70% of small bowel length. When you mentioned the duration of TPN, it is important to keep in mind that we included all patients with short bowel syndrome globally, and so there were a few patients that were on for less than 3 months that actually died secondary to TPN-related cholestasis that rapidly progressed early in their course.

Finally, we did not look at specific wound complications and whether they greatly impacted the overall care. I would say parenthetically that these kids are very delicate, infants in particular, and there is sort of a race against time in these particular patients to get continuity of their bowel intact, close their ileostomies and have them start utilizing their gut to prevent them from developing TPN-associated cholestasis. And certainly having a wound infection and the like would significantly impact adversely those particular patients.

Dr. Marshall Z. Schwartz (Philadelphia, Pennsylvania): Iwould like to congratulate Dr. Teitelbaum and his colleagues for addressing what is a major issue in the pediatric age group. The morbidity from short bowel syndrome in the pediatric age group is significant with frequent hospitalizations mostly related to the complications of parenteral nutrition and the catheters that we use to deliver the nutrition. In addition, as Dr. Teitelbaum has alluded to, the cost of this management is significant. The numbers that Dr. Teitelbaum stated are probably rather low, because if you extrapolate these costs out for a few years you are well into millions of dollars per infant and increasing as long as they survive.

This study was initiated to assess the factors that impact mortality in patients with short bowel syndrome and especially those associated with parenteral nutrition. The investigators concluded that cholestasis was the most significant correlate. But in the presentation we heard that bowel length also is a significant factor in survival. I believe that it is an important finding that, based on postnatal age and gestational age, that there is a correlation with the percent of lost intestine.

I have the following questions: I was a little surprised that the diagnosis was not a predictive factor, and I am particularly referring to gastroschisis. In my experience, infants that have had an in utero vascular occlusion related to gastroschisis resulting in loss of most of their midgut, that the residual bowel does not work very well regardless of how long it is. Did you specifically look at that issue?

Secondly. The mortality was based on cholestasis. However, are you referring to cholestasis or is this fibrosis as a result of cholestasis? Many of these infants do not get into trouble with liver disease until 6 months of age, but they may have had cholestasis much earlier.

Finally, you have done extensive work on the relationship of cholestasis to short bowel syndrome. From that work, have you altered how you manage infants with cholestasis on parenteral nutrition and has that had an impact on lowering their liver enzymes or direct bilirubin?

One last comment. In the end, you had the same conclusion, that is, to preserve the ileocecal valve and to try to retain as much bowel length as possible despite apparent lack of a relationship with cholestasis. I wonder if you might comment about this further.

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): I think going into the analysis I did have a preconceived notion that my gastroschisis patients would do worse just for the reasons you mentioned: it is an ischemic-related event. Nevertheless, when we did the analysis that did not work. As we have talked about before, perhaps with a larger group, with multiple centers contributing their data, it may be beneficial in getting at that particular question better. Even with a hundred patients, that may still not be enough to answer that particular question.

With regard to mortality, clearly cholestasis may complicate the course of a patient and the hospital may predispose them to infections or other complications. But it is truly the progression of cholestasis to liver fibrosis that will result in these patients’ deaths. And that is, unfortunately, not always in a predictable fashion. There are some patients that may take several months to develop that particular course and there are some patients that actually progress much more quickly.

You asked how we could treat cholestasis, or how we have changed our approach in treating this. And I don't have any good answers. This coming month a paper is coming out from our group where we have tried to look at the use of cholecystokinin to prevent the development of cholestasis in these particular patients. And despite looking at almost 250 patients, we failed to find efficacy in using that particular hormone.

We have also tried eliminating or decreasing the amount of lipids in our patients. Although I do think it is somewhat radical I have seen some fairly dramatic declines in hyperbilirubinemia in patients treated with a course of steroids. That has to be used pretty cautiously because obviously that has an immunosuppressive effect on these patients who are predisposed to infections.

Dr. Paul R. Schloerb (Kansas City, Kansas): My interest is in the type of parenteral nutrition used. Several years ago we surveyed TPN usage in 103 academic medical centers and found that about three-quarters of them were using excessive amounts of glucose. Based upon one pediatric patient of Dr. William Buntain, a pediatric surgeon in Kansas at the time, she was born with little small bowel and large bowel and had a Martin procedure with a side-to-side anastomosis. The child did very well for 14 months using an amount of glucose which was 12 grams per kilogram calculated body cell mass per day. Using this formula over a period of several years, we have not observed the fatty liver complications so often associated with TPN. So I would like to ask: What were the constituents of your parenteral nutrition solution?

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): I think that everybody has speculated on one particular substrate in parenteral nutrition that may be a cause of cholestasis. And there isn't any firm answer. I think clearly limiting the amount of glucose or dextrose that one administers is important only in that you obviously do not want to overfeed these patients. Because overfeeding them is going to result in hepatic steatosis, which will worsen one's liver condition.

At the beginning of our series we actually treated our patients with an adult amino acid-type formulation and toward the end, the latter 7 to 10 years, we utilized a pediatric formulation which has taurine supplemented in it. And in a separate study we were able to demonstrate quite accurately that just the supplementation of taurine in these patients was able to significantly decrease the incidence of TPN-related cholestasis.

Dr. Jean C. Emond (New York, New York): This is very elegant both as an organizational system for caring for patients with gut failure as well as a very fine statistical analysis. I wanted to ask you, though, about the issue of integration of all this care with transplant candidacy. We heard yesterday about a hundred multivisceral transplants. Many of us suspect that a lot of those transplant patients had livers that could recover or had preventable liver disease with optimal care, while some of those intestines would have probably been okay.

What is your philosophy about how to integrate transplant care with intestinal failure care? Twenty years ago biliary atresia care was separated into transplant camp and the Kasai camp and now we have come together. Have you developed an integrated approach that involves evaluation for intestine or liver transplant?

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): That is a very good question. I don't have a precise answer. Obviously, with every decision you make there is a risk and benefit that you have to weigh, which is why I showed that one slide that showed the relative risk of combining two bad hits—a child cholestasis plus very reduced small bowel length. I think you can make a poor judgment both ways attimes. And I have certainly miscalled and not sent a patientfor an intestinal transplant soon enough and had that childexpire.

Conversely, once you generally send a patient to a transplant center, it is set that the child will be transplanted. I do think that has changed somewhat, and many centers are much more cautious now. So I have tended to be a little bit more aggressive than I have maybe several years ago because I do feel that the outcomes are improved. But I think that this particular type of data may help make that decision and help one sort that out.

Dr. J. Alex Haller, Jr. (Baltimore, Maryland): Dr. Teitelbaum, you have now given us a lot of new information; and in keeping with our evidence-based surgery approach, I would extend some of the other questions to ask you if you are in any way changing your immediate decision in the operation for the initial episode that is causing the short bowel syndrome.

Many of us over the years have had to close newborn babies with midgut volvulus where all the bowel is dead, then talking with parents and having a second look perhaps, and that decision was made more or less for us. But nevertheless, now we have your new data. And if I understand correctly, fully 61% of these patients with short bowel syndrome had their short bowel condition determined before 2 weeks of age. Clearly a decision could have been made that these babies were, in a sense, miscarriages at that point, since they were also largely premature.

Have you included that approach into your management of these patients? We all know there are things worse than death when you are a newborn baby. Since we have a long track record of making this kind of decision for extensive midgut volvulus; and as we all know, the gastroschisis baby has actually also midgut volvulus likely as a part of the short bowel syndrome, have you changed your management in keeping with this information rather than prolonging the lives of children who are going to be costly to themselves, not to speak to the cost to the family?

Dr. Daniel H. Teitelbaum (Ann Arbor, Michigan): I think that is an excellent question. Because clearly for one to really attain a successful outcome even if that child needs a transplant, that child needs to be alive and surviving for some period of time. It may not be so much that I would change my approach, but probably have a more realistic outlook to some of our patients that have virtually no small bowel at all. In some of those patients, not doing anything may be the wiser choice in that regard. That is clearly a judgment call and rather difficult to make except on a case-by-case basis, but I think it is an excellent point.

Footnotes

REFERENCES