Resection With Primary Anastomosis for Necrotizing Enterocolitis: A Contrasting View

By Arthur Cooper, Arthur J. Ross, III, James A. O'Neil l , Jr, and Louise Schnaufer

New York, New York and Philadelphia, Pennsylvania

�9 Resection w i th pr imary anastomosis is current ly being advocated for t rea tment of infants w i th necrot iz ing enterocol i t is. To determine whe the r our own data would suppor t such an approach, we rev iewed re t rospect ive ly our exper ience w i th this disease since 1974. Since that t ime, 173 infants have been admi t ted for t rea tment of advanced (surgical) disease in its acute phase, of whom 143 underwent resect ion for cure; the remainder e i ther underwent laparotomy w i th decompression (3), laparot- omy wi th drainage (31, laparotomy alone (141, died at operat ion (1), or could not be resusci tated suff iciently to w i ths tand operat ion (9). Excluded were pat ients who underwent operat ive repair of late s t r ic ture (6), all of whom survived w i th no morbid i ty . Among those resected for cure, 27 infants were careful ly selected by the operat - ing surgeon for t rea tment by means of resect ion w i th pr imary anastomosis, based on the l imited and apparent ly discrete nature of thei r disease; in three the procedure was combined w i th a decompressing enterostomy. In the major i ty of cases (14), the disease was found to involve mult ip le areas of intest ine, but was l imited to a part icular anatomic region, usually distal i leum and /o r ascending colon; in the remainder, i t was due to discrete ileal or je junal per forat ion or ulcer. Overall survival among those resected for cure was 65% (96/143). It was 48% (13/27) among those t reated by means of resect ion w i th pr imary anastomosis but 72% (83/116) among those who under- wen t resect ion w i th enterostomy. However , if the early years of the series (1974 to 1976) are excluded, a t ime when resect ion w i th en teros tomy had not yet become , establ ished as standard therapy, overal l survival was 77%, (77/100), 64% (9/14) among those anastomosed pr imari ly, and 79% (68/86) among those d iver ted or exter ior ized. Mor ta l i t y in all cases was due to associated sepsis; how-, ever, anastomot ic leaks were present in three of the seven nonsurvivors of pr imary anastomosis who were autopsied. There were no obvious dif ferences be tween the t w o groups of pat ients w i th regard to detai ls of presentat ion or management, or age or we igh t at bir th or operat ion. It appears f rom these data that wh i le pr imary anastomosis has been per formed far less f requent ly , and much more select ively, in recent years, pat ients t reated in this manner cer ta in ly fare no be t te r than infants t rea ted by o ther

From the Departments of Surgery, Babies" Hospital and Colum- bia University, College of Physicians and Surgeons, New York, and the Children's Hospital of Philadelphia and University of Pennsyl- vania School of Medicine, Philadelphia.

Presented at the 18th Annual Meeting of the American Pediatric Surgical Association, Hilton Head Island, South Carolina, May 6-9, 1987.

Address reprint requests to Arthur Cooper, MD, Department of Surgery, Babies' Hospital, Columbia-Presbyterian Medical Center, 3959 Broadway, New York, NY 10032.

�9 1988 by Grune & Stratton, Inc. 0022-3468/88/2301-0016503.00/0

methods. Indeed, it may actual ly jeopardize the survival of an infant who should, on the basis of the l imi ted and discrete nature of his disease, be expected to live. Thus, our exper ience does not suppor t resect ion w i th primary anastomosis as a superior mode of therapy. It also casts doubt on the rel iabi l i ty of surgical judgment in as.,~essing the t rue ex ten t of disease at the t ime of resect ion. �9 1988 by Grune & Stratton, Inc,

INDEX WORDS: Necrotizing enterocolitis.

S UCCESSFUL TREATMENT of the infant with advanced (surgical) necrotizing enterocolitis

(NEC) remains a distinct challenge to the pediatric surgeon. While increasing experience has led to improved survival, postoperative management may be complicated by considerable losses of salt and water if a proximal enterostomy is used. Chiefly for this reason, resection with primary anastomosis is again being advocated for treatment of infants with advanced NEC. To determine whether our own data would support such an approach, we reviewed retrospectively our entire surgical experience with this disease.

MATERIALS AND METHODS

The hospital charts, abdominal x-rays, and operative reports oral/ patients with advanced NEC treated at the Children's Hospital of Philadelphia between January of 1974 and December of 1986 constituted the basis for the study. Acute NEC was defined by the presence of pneumatosis intestinalis, portal venous gas, or pneumo- peritoneum on abdominal x-ray in an appropriate clinical setting. It was considered advanced when surgical exploration was mandated, and the diagnosis was subsequently confirmed by pathologic exami- nation. Excluded from the study were patients in whom a definite diagnosis of NEC could not be made, patients with discrete gastric perforations who otherwise had no evidence of NEC, patients whose disease required only medical therapy, and patients who underwent operative repair of late stricture.

A total of 172 patients were studied with regard to primary diagnosis, gestational age, birth weight, age and weight at onset of disease, presenting signs, roentgen findings, operative indications, operative (or nonoperative) management, operative technique, bac- teriology of intraabdominal infections, extent of disease, interval between onset of disease and operation, and ultimate survival. Data were then tabulated and, when appropriate, analyzed statistically using the Student's t-test or chi-square analysis.

Initial management of all patients with the disease consisted of complete bowel rest, nasogastric decompression with a large bore feeding tube, administration of intravenous antibiotics, and resusci- tation with intravenous fluids, fresh frozen plasma, and platelet concentrate, as necessary. Parenteral nutritional support via periph- eral veins was used in the majority of patients. The progress of the disease was monitored by means of serial abdominal examinations and interval hemoglobin, white blood and piatelet counts, and

64 Journal of Pediatric Surgery, Vo123, No 1 (January), 1988: pp 64-68

RESECTION WITH PRIMARY ANASTOMOSIS FOR NEC 65

Table 1. Operative Management

Resection 143 Decompression 3 Laparotomy only 14 Laparotomy and drainage 3 Died at operation 1 No operation 9

Total 173

abdominal x-rays obtained every 6 to 8 hours. Serial serum electro- lytes, blood glucose and arterial blood gases were performed as indicated.

Physical signs of generalized peritonitis, pneumoperitoneum, and/or overwhelming sepsis, as manifested by irreversible thrombo- cytopenia and uncorrectable metabolic acidosis, constituted absolute indications for operation. The presence of a sentinel loop on x-ray for 24 to 48 hours, and abdominal wall cellulitis, were relative indica- tions. In general, patients demonstrating signs of clinical deteriora- tion despite maximal support underwent operation.

RESULTS

Of the 172 consecutive infants admitted for treat- ment of advanced NEC in its acute phase, 143 under- went resection for cure. The remainder underwent either laparotomy with decompression, laparotomy with drainage, laparotomy alone, or could not be resuscitated sufficiently to withstand operation (Table 1). Among those resected for cure, 27 infants were carefully selected by the senior operating surgeon for treatment by means of resection with primary anasto- mosis, based on the limited and apparently discrete nature of the disease found at laparotomy. In three cases, the procedure was combined with a decompress- ing enterostomy (Table 2). In the majority of cases, the disease was found to involve multiple areas of intestine, but was limited to a particular anatomic region, usually distal ileum or ascending colon; in the remain- der it was due to discrete ileal or jejunal perforations or ulcers (Table 3).

Overall survival among those resected for cure was 65%. However, while survival was 72% in those who underwent resection with enterostomy or exterioriza- tion, it was only 48% in those treated by means of resection with primary anastomosis (Table 4). A simi- lar trend is noted even if the early years of the series, a time when resection with enterostomy had not yet become established as standard therapy (1974 to

Table 2. Operative Technique

Resection, enterostomy, oversew Resection, Mikulicz enterostomy Resection, enterostomy, mucous fistula Exteriorization of isolated perforation Resection, primary anastomosis

Total

57 20 32

7 27

143

Table 3. Extent of Disease v Survival

Survivors Nonsurvivors

Discrete perforations Jejunum and ileum 4 5

Gangrene with perforation Jejunum and ileum 4 3 Ileum and colon 1 2 Colon alone 1 0

Gangrene without perforation Jejunum and ileum 3 3 Ileum and colon 0 1 Colon alone 0 O

Total Perforations 10 10 Total Nonperforations 3 4

1976), are excluded, although this difference does not quite approach statistical significance.

Postmortem examinations were performed in seven of the 14 nonsurvivors. All anastomoses were patent, but leaks were present in two of the seven, while a third had a clearly gangrenous anastomosis that was dis- rupted during evisceration. In each of these cases, microscopic disease was present at the surgical mar- gins of resection. Extensive areas of transmural necro- sis were present in the four remaining cases and was thought to have been the cause of the sepsis that led to their deaths.

The severity of illness recorded in this series of patients, the vast majority of whom were not trans- ferred for definitive therapy until the disease had already progressed to a far advanced stage, may be appreciated from a review of admission laboratory data (Table 5). Both groups of infants were anemic and markedly acidotic, although thrombocytopenia was more common in nonsurvivors then survivors. However, there were no significant differences between the two groups of patients under study with regard to any of these parameters, details of presenta- tion or management, or age or weight at birth or operation (Table 6). In all but six, pneumoperitoneum was the primary indication for operation; the remain- der came to surgery as a result of clinical deterioration despite maximal supportive therapy.

Table 4. Survival Following Resection for Cure

Overall survival 65% (96/143) Early series (1/74 to 12/76) 44% (19/43) Later series (1/77 to 12/86) 77% (77/100)

Survival with primary anastomosis 48% ( 13/27)* Early series (1/74 to 12/76) 31% (4/13) Later series ( 1/77 to 12/86) 64% (9/14)

Survival with proximal enterostomy 72% (83/116) + Early series ( 1/74 to 12/76) 50% (15/30) Later series ( 1/77 to 12/86) 79% (68/86)

*X ~ = 3.04, p = .06

66 COOPER ET AL

Table 5. Laboratory Data on Admission

Survivors Nonsurvivors

Hemoglobin (g/dL) 12.0 • 4.0 13.4 • 4.1

White blood count ( • 103/p,L) 10.3 • 7,4 9.6 • 5.1

Platelet count (• IO~/#L) 230 + 218 145 • 119

Base deficit (mEq/L) 8,0 • 8.3 7.5 _+ 4.1

DISCUSSION

More than 20 years have passed since NEC was first recognized by Berdon and his colleagues as an endemic problem in the premature low birth weight infant, l This initial report prompted a number of other authors to publish their experience with this "new" disease, 2-4 and by the mid-1970s, a concensus had emerged regarding optimal surgical management of this condi- tion s based on the clinical evidence accumulated to date, as well as standard surgical principles with regard to management of intraabdominal infection. This approach recognized overwhelming sepsis related to gangrene to be as valid a criterion for surgical intervention as pneumoperitoneum; it also emphasized strict conservatism with regard to operative strategy. Primary anastomosis in the face of peritoneal contami- nation was considered too hazardous; resection with proximal enterostomy became the standard of care.

By the late 1970s and early 1980s, mortality rates had clearly begun to fall. 68 While there was no definite evidence that this was due to refinements in the surgical approach to the disease rather than general improvements in neonatal intensive care, a causal relationship was nonetheless inferred. The successful use of proximal enterostomy alone in infants with widespread gangrene, impending or actual, served to reinforce the concepts of fecal diversion and intestinal decompression as important to the treatment of NEC. 9'1~ However, the difficulties inherent in manag- ing the high output jejunal stoma, eg, excessive losses of salt, water, trace metals and other nutrients, late formation of strictures in the defunctionalized limb, need for secondary operations for stoma closure or revision, and skin breakdown adjacent to the stoma itself, ll'~2 not to mention increased length and cost of hospitalization, led some to reconsider the role of resection with primary anastomosis for selected infants with circumscribed disease.

Kiesewetter and others, in 1979, were among the first to report good results with this technique. Of

Table 6. Age and Weight at Birth and Onset

Survivors Nonsurvivors

Gestational age (wk) 32.7 _+ 3.3 31.3 • 3.9

Birth weight (kg) 1.68 • .68 1.49 § 0.81

Dayo fonse t (d) 11.8 _+ 11.9 10.3 • 13.9

Weight at onset (kg) 1.71 • .68 1.62 + O,77

"nine premature infants who underwent resection with primary anastomosis in the face of localized perfora- tion and peritonitis, eight survived. ''13 More recently, H a rb e rg and coworkers have advoca ted this approach. 14 However, of the 27 patients they reported in 1983, 24 came to operation only after perforation, and three came to operation only after development of an abdominal mass. They made no mention of gan- grene, diffuse or otherwise. Shortly thereafter , Pokorny and his colleagues also adopted this selective approach, 15 but while they reported an overall survival of 92% in infants undergoing resection with primary anastomosis v 62% in infants undergoing resection with enterostomy, ten of the 38 patients in the former group were infants who underwent operation for late strictures. If the latter are excluded, a trend toward survival for the low-risk inborn patients in their former group is still apparent, but falls short of achieving statistical significance. Sparnon and Kiely also now advocate the use of resection with primary anastomosis instead of resection with proximal enterostomy. ~6 However, their series is small and uncontrolled; more- over, as in Harberg's series, discrete intestinal perfora- tions predominated, as only two of their 17 patients appeared to have widespread gangrene.

The reasons for the success of these groups, all of whom have achieved excellent results, is not entirely clear. It is obviously tempting to attribute their success to careful patient selection. Kiesewetter and Pokorny both emphasized this approach for their patients, most of whom were inborn. However, Harberg and Sparnon both used a less selective approach, reporting the use of primary anastomosis in consecutive patients.

We cannot explain the discrepancies between our results and those of the forementioned authors, there- fore, unless it is on the basis of severity of illness. As our overall survival rate of 76% among infants resected for cure compares favorably with those not just of these groups, but with previous studies as well, it does not seem likely that our management was at fault. Thus, we are forced to conclude that our own patients, all of whom were outborn, had disease that was much more extensive at the time of presentation, as manifested by the fact that diffuse gangrene was a far more common finding in our own series than in any of the others. Perhaps this is because in our community of some 11 level 3 nurseries, but only two children's hospitals, transfer is often delayed until the disease has already progressed to a far advanced stage, as suggested by the marked acidosis and thrombocytopenia observed in our patients.

There is, of course, one aspect of our data that deserves special comment. Of the seven nonsurvivors who underwent postmortem examination, anastomotic

RESECTION WITH PRIMARY ANASTOMOSIS FOR NEC 67

breakdown was identified in three; this translates to an overall leak rate of 43% among nonsurvivors (22% if these results are extrapolated to include the entire series). Interestingly, a similar leak rate following primary anastomosis was reported by one of us (J.A.O.) more than a decade ago. 4 The most troubling aspect of the present data, however, is that in every case in which there was a leak, microscopic disease was present at the surgical margins of resection, despite the fact that the bowel appeared viable in the judgment of the senior operating surgeon.

Thus, even though we have used primary anastomo-

sis far less frequently and much more selectively in recent years, patients treated by means of resection with primary anastomosis have certainly fared no better than infants treated by other methods. Indeed, our data suggests that taking such an approach might actually have jeopardized the survival of infants who should, on the basis of the limited and discrete nature of their disease, have been expected to live. Thus, our experience does not support resection with primary anastomosis as a superior mode of therapy. It also casts doubt on the reliability of surgical judgment in assess- ing the true extent of disease at the time of resection.

REFERENCES

1. Berdon W, Grossman H, Baker DH, et al: Necrotizing entero- colitis in the premature infant. Radiology 83:879-887, 1964

2. Touloukian R J, Berdon WG, Amoury RA, et al: Surgical experience with necrotizing enterocolitis in the infant. J Pediatr Surg 2:389-401, 1967

3. Kosloske AM, Martin LW: Surgical complications of neonatal necrotizing enterocolitis. Arch Surg 107:223-228, 1973

4. O'Neill JA, Stahlman MT, Meng HC: Necrotizing enterocoli- tis in the newborn: Operative indications. Ann Surg 1982:274-279, 1975

5. Martin LW: Neonatal necrotizing enterocolitis: Surgical con- siderations, in Report of the 68th Ross Conference on Paediatric Research, Columbus, Ross Laboratories, pp 91-92, 1975

6. Bell M J, Ternberg JL, Feigin RD, et al: Neonatal necrotizing enterocolitis: Therapeutic decisions based upon clincial staging. Ann Surg 187:1-7, 1978

7. O'Neill JA, Holcomb GW Jr: Surgical experience with neo- natal necrotizing enterocolitis (NEC). Ann Surg 189:612-619, 1978

8. Kliegman RM, Fanaroff AA: Neonatal necrotizing enterocoli- tis: A nine-year experience. Am J Dis Child 135:608-611, 1981

9. Martin LW, Neblett WW: Early operation with intestinal

diversion for necrotizing enterocolitis. J Pediatr Surg 16:252-255, 1981

10. Firor HV: Use of high jejunostomy in extensive NEC. J Pediatr Surg 17:771-772, 1982

11. Rothstein FC, Halpin TC, Kliegman R J, et al: Importance of early ileostomy closure to prevent chronic salt and water losses after necrotizing enterocolitis. Pediatrics 70:249-253, 1982

12. Cogbill TH, Milikan JS: Reconstitution of intestinal con- tinuity after resection for neonatal necrotizing enterocolitis. Surg Gynecol Obstet 160:330-334, 1985

13. Kiesewetter WB, Taghizadeh F, Bower R J: Necrotizing enterocolitis: Is there a place for resection and primary anastomosis? J Pediatr Surg 14:36-363, 1979

14. Harberg F J, McGill CW, Saleem MM, et al: Resection with primary anastomosis for necrotizing enterocolitis. J Pediatr Surg 18:743-746, 1983

15. Pokorny W J, Garcia-Prats JA, Barry YN: Necrotizing en- terocolitis: Incidence, operative care, and outcome. J Pediatr Surg 21:1149-1154, 1986

16. Sparnon AR, Kiely EM: Resection and primary anastomosis for necrotizing enterocolitis. Pediatr Surg Int 2:101-104, 1987

Discuss ion

R. Touloukian (New Haven, CT): I share Dr Coo- pet 's concerns about primary anastomosis in patients with N E C with peritonitis and established sepsis, but the reverse side of the coin also merits a question and that is the fate of those patients with ileostomies or colostomies who are subsequently discharged from the hospital prior to restoration of the gastrointestinal continuity. What is your policy in so far as managing those patients are concerned? At our institution, we have been impressed by the risk of salt water imbal- ance and diarrhea in these patients and, of course, their lack of weight gain. We have subsequently performed very early closure of the ileostomies prior to their discharge from the newborn intensive care unit.

Charles McGill (Houston, TX): I agree that I do not understand the difference in our series results. Our cases tended to be referred early to our nursery, and so

we may have had less delay in getting them to the operating room. Our indications to operate seem to be the same.

I would like to point out that we have had only two leaks in our original report of 27 cases. Neither child died as a result of the leak. In our subsequent 19 cases we have had no leaks. We have not seen a single death attributable to a complication of doing a primary anastomosis in 46 cases now.

The application of primary anastomosis is selective in our hands also. We did not mention the frequency of this choice in our paper. In the last 33 patients we have operated on since the paper, we have done a primary anastomosis in 19. In ten we did resection with a stoma, and in four we simply opened and closed. That gives a 58% anastomosis rate. We have had three cases where more than one anastomosis was done. We feel

68 COOPER ET AL

confident that when we do a primary anastomosis it has a good rate of success. Our survival rate is still better than 85%.

William Pokorny (Houston, TX): Enterostomy in these small prematures is associated with serious com- plications. Distal stricture formation appears to be less following primary anastomosis. Extensive areas of transmural necrosis were present in 100% of the autop- sied cases following primary anastomosis in this series and was thought to be the cause of death in all patients. This suggests postoperative progression of the disease in all patients dying following primary anastomosis. Please comment on this since I have rarely seen it. In the past 10 years, nine of 14 patients undergoing anastomosis in this series have survived, so they have accentuated the worst of the two series to make their point. In our experience, there were no anastomotic leaks and three deaths among 28 patients who under- went resection and primary anastomosis for acute NEC. Why the difference? Resection and primary anastomosis continues to be the procedure of choice in nearly 50% of babies who have advanced NEC requir- ing surgical intervention by our hands.

B. O'Donnell (Dublin): NEC is a disappearing dis- ease. We survey 65,000 live births a year, 24,000 in Dublin, the largest maternity hospital in Europe. Last year, in the 24,000 live births, there were 18 cases of NEC; two of them came to surgery. This is a disap- pearing disease in the "developed" countries.

A. Cooper (closing): Dr. Touloukian, we too recog- nize the considerable salt and water losses among infants with high enterostomies. We also try to get them closed before the child leaves the hospital. Of those patients who underwent ileostomy, the vast majority did well. There was one long-term nonsurvi- vor who presented with sudden infant death presum-

ably following a gastrointestinal infection with asso- ciated dehydration. However, in the 110 or so who did not undergo primary anastomosis, this was the only mortality. Dr Pokorny, we did not observe a high incidence of distal stricture, probably well under 5%. Concerning our seven nonsurvivors who underwent autopsy, I am sorry if the manuscript implied necrosis was responsible for death. This was so only in the patients whose anastomoses did not leak; the other infants died from peritonitis due both to anastomotic leaks and progressive disease. Unfortunately, Dr Har- berg is not here to defend himself, but I did have the opportunity to speak with him earlier, and I think our joint conclusion was that most of his patients were referred early and were inborn, whereas virtually all of our patients were referred very late and were outborn. Thus, greater severity of disease appears to be the most likely explanation for the differences in our results. So far as the technical factors involved, I thought that "big" Texan sutures would have resulted in a higher leak rate but obviously it is our "t iny" Eastern sutures that have not done the job! Of course, the reason for analyzing the data as we did was to point out that in the later series the results did not reach statistical significance. However, there is still a clear trend that primary anastomosis is unsafe. Dr McGill, I would reiterate the comments I made to Dr Pokorny regard- ing the inborn v the outborn patients, and I am sorry if I implied that Dr Harberg's series was highly selected. I thought I had made the statement in the presentation that you had done it on a more or less unselected basis. Finally, Mr O'Donnell, we too believe that necrotizing enterocolitis is a disappearing disease, thank goodness, and hopefully it will soon disappear in our own "devel- oping" country as well!