PEDIATRICS Vol. 105 No. 3 March 2000, pp. 510-514

Magnetic Resonance Imaging of Intestinal Necrosis in Preterm Infants

Elia F. Maalouf, MRCP^*, Andrew Fagbemi, MRCP^*, Philip J. Duggan, MRCP^*, Subramanyam Jayanthi, MRCS; Serena J. Counsell, DCR^§, Helen J. Lewis, DCR^§, Alison M. Fletcher, DCR^§, Kokila Lakhoo, and FRCS; and A. David Edwards, FRCP^*

From the Departments of ^* Paediatrics and  Surgery and Gastroenterology; and ^§ Robert Steiner Magnetic Resonance Unit and Medical Research Council Clinical Science Centre, Imperial College School of Medicine, Hammersmith Hospital, London, England.

	ABSTRACT

Top Abstract Results Discussion Conclusion References

Background and Objective.  Noninvasive diagnosis of intestinal necrosis is important in planning surgery in preterm infants with necrotizing enterocolitis (NEC). We aimed to assess the potential of magnetic resonance imaging (MRI) for the diagnosis of intestinal necrosis.

Study Participants and Methods.  Abdominal MRI scans were performed in a group of preterm infants with suspected NEC and compared with surgical findings and to MRI results in a group of control infants. In addition, MRI was performed in 2 preterm infants with suspected NEC who did not require surgery.

Results.  Six infants with a median birth weight of 1220 g (range, 760-1770 g) and median gestational age at birth of 30 weeks (range, 28-34 weeks) were studied at a median postnatal age of 10 days (range, 4-19 days). Four infants had a bubble-like appearance in part of the intestinal wall, intramural gas, and an abnormal fluid level within bowel lumen. At surgery, NEC was found in 5 infants and sigmoid volvulus in 1. The site of the bubble-like appearance corresponded to the site of intestinal necrosis at surgery. Four control infants with a median birth weight of 1500 g (range, 730-2130 g) and a median gestational age of 31 weeks (range, 26-36 weeks) had abdominal MRI at a median postnatal age of 8 days (range, 4-70 days). None of the above findings were seen in any control infant. The bubble-like appearance was not seen in the 2 infants with suspected NEC who did not require surgery.

Conclusion.  Abdominal MRI allows the noninvasive diagnosis of bowel necrosis. This may aid the timing of surgical intervention in preterm infants with a clinical diagnosis of NEC.gangrene, ischemia, MRI, necrotizing enterocolitis.

Necrotizing enterocolitis (NEC) is the most common surgical emergency in preterm infants. It is generally believed to be secondary to several underlying factors including: intestinal injury, usually ischemic; bacterial colonization of the gut; and the presence of high osmolality solutions, usually formula milk, within the gut lumen. NEC usually involves the ileum and the proximal colon, but any part of the gastrointestinal tract may be affected.¹ The diagnosis of NEC is based on clinical findings (abdominal distension, lethargy, gastrointestinal bleeding, abdominal wall erythema, and acidosis) and plain abdominal radiographic criteria (dilated bowel loops, bowel-wall thickening, intramural gas, and portal venous gas).

Despite appropriate medical treatment, surgical intervention becomes necessary in 40% to 50% of cases of NEC. The ideal timing for surgery is after the development of intestinal necrosis but before the development of intestinal perforation.^1,2 Noninvasive diagnosis of intestinal necrosis is difficult^1,3 and no test has a high sensitivity.²

Preliminary experience with animal models^4-7 suggests that magnetic resonance imaging (MRI) may have a role in the diagnosis of bowel ischemia. The aim of this pilot study was therefore to describe abdominal MRI findings in NEC and to assess the potential role of MRI as a safe, noninvasive tool in the diagnosis of intestinal necrosis.

	STUDY PARTICIPANTS AND METHODS

Ethical approval for this study was given by the Hammersmith Hospitals Research Ethics Committee and informed parental consent was obtained in each case. Preterm infants who had clinical and radiologic signs suggestive of NEC and who subsequently underwent laparotomy were included in this study. The indications for surgical intervention were failure of medical treatment or intestinal perforation. Abdominal MRI was obtained before and after operation and compared with surgical findings. The severity of NEC was staged using clinical signs and criteria on plain abdominal radiographs (modified Bell criteria in which stage I is suspect, stage II is definite, and stage III is advanced NEC).⁸ The presence and the site of intestinal necrosis was determined at laparotomy.

A control group of preterm infants of similar gestational age (GA) and with no symptoms or signs suggestive of NEC were also studied for comparison. In addition, MRI was performed in 2 infants with suspected NEC who responded to conservative medical treatment and did not require surgery.

MRI Acquisition

Patients were scanned 2 to 3 hours before undergoing laparotomy and as soon as practical afterward. MRI was performed using a 1-Tesla neonatal magnetic resonance system (Oxford Magnet Technology, Oxford, UK/Picker, Cleveland, OH) located within the neonatal intensive care unit. The magnet has a 380-mm bore length, which allows good access to the infant for monitoring and ventilation during scanning. Full intensive care and monitoring were continued during scanning as previously described.⁹ Longitudinal relaxation time (T1)-weighted CSE and transverse relaxation time (T2)-weighted FSE images in the coronal and transverse planes were obtained in all infants. In addition, postcontrast T1-weighted CSE images were obtained in the coronal and transverse planes using intravenous dimeglumine gadopentetate (0.1 mmol/kg) as a contrast agent. Control infants were scanned once without the use of contrast agents. Infants with suspected NEC who did not require surgery were scanned within 48 hours of developing symptoms and signs suggestive of NEC. The sequence parameters for T1-weighted CSE and T2-weighted FSE are summarized in Table 1.

MRI Analysis

The images were analyzed before surgery, and results were compared with findings at laparotomy. Abnormal features recorded were the presence or absence of bowel-wall thickening, dilated bowel loops, intramural gas, abnormal fluid levels within bowel lumen, mesenteric edema, and bubble-like appearance in parts of the intestinal wall.

Bowel-wall thickening and bowel-loop dilatation were assessed subjectively in comparison to control infants. Intramural gas appeared as a line of low signal intensity within the bowel wall on T1- and T2-weighted images. Fluid levels within bowel lumen had low signal intensity on T1-weighted images and high signal intensity on T2-weighted images. Mesenteric edema had high signal intensity on T2-weighted images. Infants with a bubble-like appearance in parts of the intestinal wall had multiple small circular low-signal intensity lesions within the bowel wall giving a characteristic appearance on T2-weighted images. The site of bubble-like appearance of bowel was compared with the site of bowel necrosis observed at surgery with the help of detailed diagrams of operative findings.

	RESULTS

Top Abstract Results Discussion Conclusion References

Six infants suspected of having NEC with a median birth weight of 1220 g (range, 760-1770 g) and a median GA of 30 weeks (range, 28-34 weeks) were included in the study. Prelaparotomy MRI scans were acquired at a median postnatal age of 10 days (range, 4-19 days) and postlaparotomy MRI scans were acquired at a median of 8 days (range, 5-19 days) after surgery. Table 2 summarizes the clinical details and the timing of the MRI scans in all cases. Four control infants with a median birth weight of 1500 g (range, 730-2130 g) and a median GA of 31 weeks (range, 26-36 weeks) were scanned at a median postnatal age of 8 days (range, 4-70 days). Two infants with a birth weight of 715 and 2006 g and a GA of 25 and 36 weeks were scanned at a postnatal age of 49 and 4 days, respectively.

Radiograph Findings

On plain abdominal radiography, all 6 infants who underwent surgery had dilated bowel loops and intramural gas; 5 had thickening of the bowel wall (cases 1-5); 1 had air in the portal system (case 1); and 2 had evidence of free intraperitoneal gas (cases 2 and 6). Plain radiographs were not obtained in the control infants. The 2 infants with suspected NEC who did not require surgery had generalized bowel distension, dilated bowel loops, and intramural gas on plain abdominal radiographs.

Surgical Findings

The diagnosis of NEC was surgically confirmed in 5 of these infants (cases 1-5) but sigmoid volvulus with perforation was diagnosed intraoperatively in the sixth (Fig 1) (case 6). Four of the infants with NEC (cases 1-4) had nonviable necrotic bowel and required bowel resection and exteriorization. Bowel necrosis was subsequently confirmed on histologic examination. The fifth infant with NEC (case 5) had dusky but nonnecrotic bowel and did not require bowel resection or exteriorization. The infant with volvulus had healthy looking bowel. Table 2 summarizes the surgical findings.

View larger version (165K): [in this window] [in a new window]   Fig. 1.   Infant with volvulus born at 31 weeks' gestation and imaged at a postnatal age of 19 days (case 6). Transverse relaxation time-weighted FSE (repetition time, 3000 ms; echo time, 208 ms) transverse image showing dilated bowel loops but there is no bubble-like appearance of the bowel wall and no abnormal fluid level within bowel lumen.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Surgical diagnosis, preoperative and postoperative magnetic resonance imaging findings in infants who required surgery. Abbreviations: MRI, magnetic resonance imaging; pre, preoperative MRI findings; post, postoperative MRI findings.  indicates present.  indicates absent.

MRI Findings

Figure 2 summarizes pre- and postoperative MRI findings in the 6 infants who required surgery. Preoperative MRI was less motion artifacted in all infants with NEC compared with postoperative studies and control infants' MRI. All findings described were seen on both transverse and coronal planes.

On preoperative MRI dilated bowel loops, mesenteric edema and thick bowel wall were present in all 6 infants. In addition to the above findings, the 4 infants who required bowel resection had an abnormal fluid level within bowel lumen, intramural gas, and a bubble-like appearance in parts of the intestinal wall (Figs 3-5). The site of the bubble-like appearance on MRI corresponded well to the site of intestinal necrosis at surgery.

View larger version (90K): [in this window] [in a new window]   Fig. 3.   Infant with necrotizing enterocolitis born at 33 weeks' gestation and imaged at a postnatal age of 4 days (case 3). A, coronal transverse relaxation time-weighted fast spin echo (repetition time, 3000 ms; echo time, 208 ms) image showing the bubble-like appearance in the intestinal wall (arrows). B, transverse relaxation time-weighted fast spin echo (repetition time, 3000 ms; echo time, 208 ms) image showing intramural gas (short arrow) and an abnormal fluid level within bowel lumen (long arrow).

View larger version (85K): [in this window] [in a new window]   Fig. 4.   Infant with necrotizing enterocolitis born at 30 weeks' gestation and imaged at a postnatal age of 7 days (case 1). A, transverse relaxation time-weighted fast spin echo (repetition time, 3000 ms; echo time, 208 ms) image showing a bubble-like appearance within bowel wall (arrows). B, transverse relaxation time-weighted fast spin echo (repetition time, 3000 ms; echo time, 208 ms) image showing mesenteric edema (long arrow) and thickened bowel wall (short arrow).

View larger version (169K): [in this window] [in a new window]   Fig. 5.   Infant with necrotizing enterocolitis born at 34 weeks' gestation and imaged at a postnatal age of 14 days (case 4). A, transverse relaxation time-weighted fast spin echo (repetition time, 3000 ms; echo time, 208 ms) image showing a bubble-like appearance of the intestinal wall (arrow).

Postoperative MRI showed resolution of the abnormal fluid level within bowel lumen, the intramural gas, and the bubble-like appearance of the intestine in all 4 infants with necrotic bowel. Infants scanned between postoperative day 5 and day 9 still had persistence of either bowel-wall thickening, mesenteric edema, and/or dilated bowel loops. The 2 infants scanned on postoperative days 12 and 19 did not have any of the above findings. There was no correlation between bowel wall enhancement on postcontrast T1-weighted images and sites of bowel necrosis.

MRI scans obtained in control infants showed motion artifact secondary to bowel peristalsis. These infants did not have dilated bowel loops, bowel-wall thickening, abnormal fluid levels within bowel lumen, mesenteric edema, or any of the other findings described above.

MRI scans obtained in the 2 infants with suspected NEC who did not require surgery showed dilated and thickened bowel loops with mesenteric edema. There was no fluid level seen within the bowel lumen, no intramural gas, and no bubble-like appearance in any part of the intestine.

	DISCUSSION

Top Abstract Results Discussion Conclusion References

We have described the abdominal MRI findings in a small group of preterm infants with clinical suspicion of NEC who underwent surgery, compared those to a group of control infants and to 2 infants with NEC who were treated medically. We have showed that the imaging appearances and in particular the sites of abnormality were closely related to perioperative findings.

The noninvasive diagnosis of bowel necrosis, which is the main indication for surgical intervention in preterm infants with NEC in the absence of pneumoperitoneum, has proved to be difficult.^1-3 Plain abdominal radiographic features such as portal venous gas and fixed intestinal loops; and laboratory markers such as white blood cell count, immature to total neutrophil ratio, platelet count, and base excess have all been used to assist in the timing of surgery but none of these tests has a high sensitivity.^2,3

Abdominal ultrasound scanning with Doppler has been used to diagnose bowel necrosis without perforation but different reports give conflicting assessments of its value in differentiating necrotic from inflammatory bowel changes.^10,11

Computed tomography (CT) findings of thickened bowel wall with postcontrast enhancement, dilated fluid filled bowel, intramural gas, and intramural low attenuation zones have been described in adults with bowel ischemia^12-14 but the use of CT in preterm infants with NEC has not yet been reported. A specific CT finding for bowel necrosis in adults is intestinal pneumatosis which may be seen as cyst-like collections of gas in the bowel mucosa.¹³ The appearance of bubble-like collections of gas on CT have also been described in adults with Fournier's gangrene.¹⁵

The development of MRI in the abdomen has been impeded by motion artifact secondary to both respiratory and peristaltic motion. Infants with NEC or sigmoid volvulus tend to have an ileus as part of their illness and therefore the preoperative MRI obtained from study infants showed very little motion artifact.

MRI has been used in the diagnosis of early and late intestinal ischemia in several animal models.^4-7 High signal intensity on T2-weighted and isointense or slightly increased signal intensity on T1-weighted images was reported early (within 45 minutes) after an ischemic insult.^5,6 High signal intensity on both T1- and T2-weighted images was reported late (24 hours) after an ischemic insult.⁴ Postcontrast enhancement was reported to increase accuracy of diagnosis in the early phase⁷ but not the late phase.⁴ All animal models concentrated on measuring bowel wall signal intensity. A bubble-like appearance in the bowel wall was not reported. In our study there was no correlation between signal intensity on pre- and postcontrast scans and the site of necrotic bowel. The changes reported in animal studies may represent a relatively early phase of bowel ischemia compared with the in vivo progression of bowel necrosis in preterm infants.

NEC in preterm infants is not only related to bowel ischemia but several other factors including infection and feeding pattern are believed to contribute to the pathogenesis.¹ The characteristic pathologic finding in NEC is crepitant necrosis of the gut.¹ 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 gives the pathologic characteristic of crepitance. The bubble-like appearance has also been described on MRI in cases of Fournier's gangrene.^16,17

Abdominal MRI features, which were constantly associated with the surgical finding of bowel necrosis, were intramural gas, abnormal fluid level in bowel lumen, and bubble-like appearance in parts of the intestinal wall. The site of bubble-like appearance in the bowel wall on MRI corresponded to the site of necrotic bowel at surgery. The finding of bubble-like appearance of bowel wall on abdominal MRI in association with an abnormal fluid level in bowel lumen may therefore be diagnostic of bowel necrosis. Mesenteric edema and thickening of bowel wall were also seen in the case of NEC that did not have bowel necrosis, in the 2 infants with NEC who did not require surgery, and in the infant with volvulus. Intramural gas can be seen on plain abdominal radiographs in 85% of cases of NEC and is not indicative of bowel necrosis.¹ An abnormal fluid level within bowel lumen has been described as a CT finding in bowel ischemia in adults^12-14 but is not a specific sign of bowel necrosis.

	CONCLUSION

Top Abstract Results Discussion Conclusion References

In summary, dilatation of bowel loops, thickening of bowel wall, and mesenteric edema are nonspecific MRI findings and may be present in all infants with NEC. The additional findings of fluid level within the bowel lumen, intramural gas, and in particular bubble-like appearance in parts of the intestinal wall on MRI are highly suggestive of the presence of intestinal necrosis.

These results suggest that MRI is likely to be of value in diagnosing intestinal necrosis and determining the need for surgical intervention. MRI may also be of value in other intestinal conditions where gut necrosis is a feature. We suggest that further studies to determine the value of MRI in the diagnosis of intestinal necrosis are warranted.

	ACKNOWLEDGMENTS

This work was supported by Wellbeing, the Medical Research Council, the Garfield Weston Foundation, Picker International, and Oxford Magnet Technology.

We thank Professor G. M. Bydder for his support and his help in interpreting the magnetic resonance images.

	FOOTNOTES

Received for publication Feb 22, 1999; accepted Jun 15, 1999.

Reprint requests to (A.D.E.) Department of Paediatrics, Imperial College School of Medicine, Hammersmith Hospital Campus, Du Cane Road, London W12 ONN, England. E-mail: david.edwards{at}ic.uc.uk.

	ABBREVIATIONS

NEC, necrotizing enterocolitis; MRI, magnetic resonance imaging; GA, gestational age; T1, longitudinal relaxation time; T2, transverse relaxation time; CT, computed tomography; CSE, conventional spin echo; FSE, fast spin echo.

	REFERENCES

Top Abstract Results Discussion Conclusion References

1. Kosloske AM, Musemeche CA Necrotizing enterocolitis of the neonate. Clin Perinatol. 1989; 16:97-111 [Medline]
2. Kosloske AM Indications for operation in necrotizing enterocolitis revisited. J Pediatr Surg. 1994; 29:663-666 [CrossRef][Medline]
3. Gupta SK, Burke G, Herson VC Necrotizing enterocolitis: laboratory indicators of surgical disease. J Pediatr Surg. 1994; 29:1472-1475 [CrossRef][Medline]
4. Park A, Towner RA, Langer JC Diagnosis of persistent intestinal ischemia in the rabbit using proton magnetic resonance imaging. J Invest Surg. 1994; 7:485-492 [Medline]
5. Kaufman AJ, Tarr RW, Holburn GE, McCurdy M, Partain CL, James AE Jr Magnetic resonance imaging in ischemic bowel in rabbit model. Invest Radiol. 1988; 23:93-97 [CrossRef][Medline]
6. Elias DL, Nelson RC, Herbst MD, Zubowicz VN Magnetic resonance imaging for detection of arterial and venous occlusion in canine muscle flaps and bowel segments. Ann Surg. 1987; 206:624-627 [Medline]
7. Klein HM, Klosterhalfen B, Kinsel S, CT and MRI of experimentally induced mesenteric ischemia in a porcine model. J Comput Assist Tomogr. 1996; 20:254-261 [CrossRef][Medline]
8. Bell MJ, Ternberg JL, Feigin RD, Neonatal necrotising enterocolitis: therapeutic decisions based on clinical staging. Ann Surg. 1978; 187:1-7 [Medline]
9. Battin M, Maalouf EF, Counsell S, Physiological stability of preterm infants during magnetic resonance imaging. Early Hum Dev. 1998; 52:101-110 [CrossRef][Medline]
10. Siegel MJ, Friedland JA, Hildebolt CF Bowel wall thickening in children: differentiation with ultrasound. Radiology. 1997; 203:631-635 [Abstract/Free Full Text]
11. Fotter R, Sorantin E Diagnostic imaging in necrotising enterocolitis. Acta Paediatr Suppl. 1994; 396:41-44 [Medline]
12. Taourel PG, Deneuville M, Pardel JA, Regent D Acute mesenteric ischemia: diagnosis with contrast enhanced CT. Radiology. 1996; 199:632-636 [Abstract/Free Full Text]
13. Llauger CPJ, Puig J, Palmer J Computed tomographic findings in bowel ischemia. Gastrointest Radiol. 1989; 14:241-245 [CrossRef][Medline]
14. Frager D, Baer JW, Medwid SW, Rothpearl A, Bossart P Detection of intestinal ischemia in patients with acute small bowel obstruction due to adhesions or hernia. AJR Am J Roentgenol. 1996; 166:67-71 [Abstract/Free Full Text]
15. Amendola MA, Casillas J, Joseph R, Antun R, Galindez O Fournier's gangrene: CT findings. Abdom Imaging. 1994; 19:471-474 [CrossRef][Medline]
16. Okizuka H, Sugimura K, Yoshizako T Fournier's gangrene: diagnosis based on MR findings. AJR Am J Roentgenol. 1992; 158:1173-1174 [Medline]
17. Michel F, Fauchery A, Belhadji M, Couailler JF, Martin-Champetier R The contribution of imagery in the early diagnosis of Fournier's gangrene. Prog Urol. 1997; 7:471-475 [Medline]