PEDIATRICS Vol. 106 No. 5 November 2000, pp. 1170

The Role of Magnetic Resonance Imaging in Necrotizing Enterocolitis

To the Editor.

The treatment of necrotizing enterocolitis (NEC) is a continuing challenge for neonatalogists and pediatric surgeons. Although many infants are treated successfully without operation, others require surgical intervention for intestinal perforation or necrosis. Determining which infants with NEC require operation is a diagnostic challenge. In a recent issue of Pediatrics, Maalouf et al conclude, "Magnetic resonance imaging (MRI) is likely to be of value in diagnosing intestinal necrosis and determining the need for surgical intervention."¹

Although the authors have chosen a crucial subject for study and have raised some interesting questions, they have not presented data sufficient to support their conclusions. The major problems of this study include: selection of a study group inappropriate to answer the research question, selection of a control group inappropriate to the research question, retrospective analyses of the results of the MRI scans by unblinded radiologists already aware of the clinical findings, and a sample size far too small to allow results to be generalized.

The authors defined the entry criteria for the study to include patients with clinical and radiographic signs of severe NEC who subsequently underwent operation for NEC. All 6 of the study patients already had a combination of clinical and radiographic abnormalities mandating operation before the MRI scan was obtained. Each had extensive pneumatosis on plain films and at least 2 had intraperitoneal gas.

The authors describe the MRI findings to include dilated bowel loops, mesenteric edema, and thick bowel wall. Each of these findings was present on the plain radiographs, and MRI was not needed to see them. The authors then describe what they believe to be the pivotal finding in their study: "bubble-like appearance." "Bubble-like appearance" is the term used by the authors to describe the MRI appearance of intramural gas or what is commonly called pneumatosis. The authors state in the discussion that "the bubble-like appearance seen on abdominal MRI in areas of necrotic bowel may be explained by the presence of gas bubbles within the bowel wall, which give the pathologic appearance of crepitance." They argue that this MRI finding is the key to early identification of necrotic bowel.

I would suggest that they have only demonstrated with MRI what the plain film already showed in all 6 patientsgas in the bowel wall. The fact that pneumatosis in these 6 patients correlated with bowel necrosis means little. All 6 patients already had clinical and radiographic evidence of bowel necrosis requiring operation. That is how they gained entry into the study in the first place. It is no surprise that the area of pneumatosis in these 6 patients correlated with the area of bowel necrosis.

The relevant and critical question is whether pneumatosis seen on MRI in the absence of obvious clinical indications for operation would correlate with bowel necrosis. That question is not addressed by the data in the study. This leads to the issue of the inappropriate control group. The authors selected 4 premature infants of similar gestational age to the study patients to serve as controls. MRI scans did not reveal dilated bowel loops, mesenteric edema, thickened bowel loops, etc. The "bubble-like appearance" was also absent. This is not unexpected because these infants did not have NEC. The authors do not suggest that MRI be used to distinguish infants with NEC from those without it. Rather, they suggest MRI may distinguish those with NEC requiring operation from those with NEC not requiring operation. An appropriate control group would include infants with severe NEC in whom bowel necrosis was absent. One wonders what the plain films looked like in the control infants in this study. "Plain radiographs were not obtained in the control infants."

The other primary concern with the study's design is the lack of blinding of the MRI radiologist to the patient's clinical status and course. The study's intent was to do MRI scans a few hours before the operation and compare the MRI findings with the operative findings. It is not at all clear how these studies would have been interpreted in the absence of surgical findings and how this interpretation might compare with findings in patients with severe NEC who did not require operation. It is difficult to argue that findings identified in retrospect with previous knowledge of the true pathology would have been identified prospectively as predicting that pathology.

In addition to the above concerns, it is important to remember that the sample size of this study is very small. It includes 6 study patients and 4 controls. It would be inappropriate to generalize findings from this small group to care of neonates with NEC.

So finally, what harm can come from this report of a novel approach to a challenging clinical problem? Plenty. Currently, many operations for NEC are done in the neonatal intensive care unit because these patients are believed to be too unstable for transport. Several reports have suggested that the trip to the operating room is at least as hazardous as the operation itself.^2,3 In our institution, the journey to MRI is far more tortuous and lengthy than the trip to the operating room. Although I applaud the authors for completing all of these studies with no reported harm to the patients, I do not take lightly any suggestion that a critically ill premature infant should be transported to the MRI scanner. I would argue that considerable harm may come from such an approach. The data regarding the value of the MRI would need to be exceedingly compelling to justify the transport.

Furthermore, the authors have not shown data regarding the MRI appearance of patients with pneumatosis who do not have necrosis. We know this is common because most infants with pneumatosis recover without operation. What if these patients have a "bubble-like appearance" as well.² Acceptance of the author's data could lead to unnecessary operation in patients who would otherwise recover with only medical therapy. Finally, one wonders whether all infants with perforated NEC would have this "bubble-like appearance." If not, then its absence could lead one to delay necessary operation.

I applaud the authors for taking on a problem that has remained unsolved for some time. I agree with them that further study is indicated. I would argue, however, that currently available evidence does not suggest that MRI has a role in the management of infants with NEC.

R. Lawrence Moss
Stanford University School of Medicine
Packard Children's Hospital
Pediatric Surgery
Palo Alto, CA 94304

REFERENCES

1. Maalouf EF, Fagbemi A, Duggan PJ, Magnetic resonance imaging of intestinal necrosis in preterm infants. Pediatrics. 2000; 105:510-514 [Abstract/Free Full Text]
2. Frawley G, Bayley G, Chondros P Laparotomy for nectorizing enterocolitis: intensive care nursery compared with operating theater. J Paediatr Child Health. 1999; 35:291-295 [Medline]
3. Gavilanes AW, Heineman E, Herpers MJ, Use of neonatal intensive care unit as a safe place for neonatal surgery. Arch Dis Child Fetal Neonatal Ed. 1997; 76:F5153

In Reply.

We are grateful for Dr Moss's attention and the generous and serious tone of his letter. In essence, while praising our efforts, he censures us for not publishing a formal analysis of the sensitivity and specificity of magnetic resonance imaging (MRI) in infants suspected of having necrotizing enterocolitis (NEC) and for not reporting a detailed comparison of MRI with abdominal radiographs.

Our article was never intended to do this. The stated aim of this pilot study was "to describe abdominal MRI findings in NEC and to assess the potential role of MRI" (italics added). We hoped that by bringing what we believe are the first MR images of proven NEC to the attention of the pediatric community we might stimulate other groups to consider research in this field. We concluded our article by saying, "We suggest that further studies to determine the value of MRI in the diagnosis of intestinal necrosis are warranted."

We did not even attempt to assess the value of MRI as a clinical test in this study of 6 infants. Still less did we suggest that seriously ill infants with suspected NEC be taken on perilous journeys to distant and unfamiliar MRI departments. Our infants were examined using a MRI specifically designed for studying sick preterm infants and located within the neonatal intensive care unit. We did not assume that this resource is currently available in every institution.

Naturally, we are now undertaking the additional study required to determine the value of MRI as a diagnostic test in a larger group of infants. In this study we calculated the sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios for the test, comparing it to radiograph studies. It includes the results he requests from infants with MRI signs of necrosis who were not taken to surgery immediately and those without pneumatosis on abdominal radiograph. These data will allow the community to make a better judgment of the value or otherwise of MR to their research or practice, and may help to alleviate some of Dr Moss's misgivings. I hope Dr Moss will forgive us for not preempting the completion of this work by giving interim results here, beyond that they are encouraging.

We are grateful to Dr Moss for reemphasizing the potential risks of a challenge in clinical practice based on early research results in a very small number of infantsit is a warning that bears frequent repetition. We agree that it is not yet appropriate to regard abdominal MRI as essential for infants with suspected NEC. Indeed, although we found that MRI is likely to be of value in diagnosing intestinal necrosis, we will not use it to influence our own clinical decisions until we have developed the evidence base to make this appropriate.

David Edwards
Department of Paediatrics
Division of Paediatrics, Obstetrics and Gynaecology
Imperial College of Science Technology and Medicine
Hammersmith Hospital Campus
London W12 ONN England