Risk of peritonitis and fatal septicaemia and the need to defunction the low anastomosis

Br. J. Surg. 1991, Vol. 78, February, 196-1 98

N. D. Karanjia, A. P. Corder, P. J. Holdsworth and R. J. Heald

Coforectaf Research Unit, Basingstoke District Hospital, Basingstoke, Hampshire RG24 9NA, UK Correspondence to: Mr R. J. Heald

The aim of this stud-v was to investigate the need to dejunction the low anastomosis after anterior resection of the rectum with total mesorectal excision fo r rectal cancer. Two hundred consecutive patients (125 defunctioned, 75 non-defunctioned) undergoing low anterior resection for carcinoma were included in the study. Peritonitis requiring emergency laparotoni-v occurred in 8 per cent of the patients who did not have a defunctioning stoma compared with less than I per cent of those patients who had a defunctioning stoma (P < 0.01). There was no mortality related to closure of the stoma but seven patients developed a , faecal fistula and ten developed an incisional hernia. Despite current trends to avoid the defunctioning stoma, these results suggest that after total mesorectal excision the faecal stream should be temporarily diverted awa-v froin the anastomosis that is 6 cm or less f rom the anal uerge to protect against potentially life-threatening anastomotic leakage.

The Colorectal Research Unit a t Basingstoke has followed a consistent policy of regarding low anterior resection with total mesorectal excision as the operation of choice for rectal carcinoma. A 10-year audit of 200 consecutive anterior resections by one of the authors (R.J.H.) revealed a 3 per cent local recurrence rate in 160 operations classed as 'curative' and this suggests a high probability of real benefit from this technique both in terms of local control and long-term cure','. Abdominoperineal excisions have been substantially reduced and now account for 9.8 per cent of all cases'.

This paper addresses the adverse consequences of combining lower anastomosis with wider local clearance, a refractory problem of anastomotic leakage. Anxiety about this hazard has alternated with peer pressure against the use of the defunctioning colostomy3s4 to produce an inconsistent policy regarding faecal diversion. No attempt has been made to analyse the causes of anastomotic leakage, but rather the effect on the outcome of the decision to defunction has been studied.

Patients and methods Two hundred consecutive patients (1 14 men, 86 women; mean(s.d.) age 67( 12) years) who had undergone stapled low anterior resection for rectal carcinoma were studied with respect to anastomotic integrity. All operations were performed personally by R.J.H. or under his supervision. The technique for anterior resection in this series has previously been d e ~ c r i b e d ' . ~ . ~ . The splenic flexure was used for reconstruction in 157 (78.5 per cent) patients and the sigmoid colon in 43 (21.5 per cent). In 95 per cent of patients the anastomosis was constructed using a double purse-string technique; in the remainder a cross-stapling device was used with the EEA" (US Surgical Corporation, East Norwalk, Connecticut, USA) premium gun. In 168 (84.0 per cent) patients the anastomosis was 6 cm or less from the anal verge; in the remaining 32 (16.0 per cent) patients the anastomosis was more than 6 c m but less than 12cm from the anal verge. The mean(s.d.) anastomotic height in the entire series was 5.1(2.0) cm.

A group of 125 patients (75 men. 50 women; mean(s.d.1 age 67(11) years) had a defunctioning loop colostomy. All of these patients had a water-soluble contrast enema to check the integrity of the anastomosis before planned closure of the colostomy (usually 6-8 weeks after the original surgery).

Another group of 75 patients (39 men, 36 women; mean(s.d.) age 67(13) years) underwent stapled low anterior resection without a defunctioning stoma. These patients included only those considered by the surgeon to be at minimal risk from anastomotic disruption. They had good pelvic haemostasis, an excellent colonic blood supply, good

bowel preparation and were tit. This group did not have routine radiological assessment of anastomotic integrity, but patients did undergo iopamidol (Niopam", E. Merck Ltd.. Alton, UK) contrast studies if their clinical progress was imperfect.

The progress after operation of all patients was recorded using a computerized system of clinical audit. Particular reference was made to the incidence and consequences of anastomotic leakage in the two groups of patients studied.

Results are expressed as mean(s.d.) and compared between groups by using either the unpaired two-tailed t test or the 1' test where appropriate.

Results Defunctioned group (n = 125) In this group 118 patients had an anastomosis a t 6 cm or lower, and seven had an anastomosis higher than 6cm, up to a maximum of 12 cm from the anal verge. Overall the mean(s.d.) anastomotic height was 4.5(1,5) cm.

One patient (0.8 per cent) in this group, whose anastomosis was less than 6 c m from the anal verge, developed faecal peritonitis which necessitated an emergency laparotomy. At operation it was apparent that a faecal column persisted in a poorly prepared bowel proximal to the anastomosis. Peritoneal toilet and irrigation of the distal colon were undertaken. After an initially good recovery, the patient died from a cerebrovascular accident 30 days after this second laparotomy.

Twenty-one (16.8 per cent) patients in this group showed radiological evidence of anastomotic leakage. In 20 patients the anastomosis was 6 cm or less from the anal verge. In 15 patients the defect in the anastomosis remained asymptomatic and healing occurred after deferring colostomy closure for a further 1 4 months. The remaining six patients included one who healed rapidly after an examination under anaesthesia to remove a retrorectal concretion of stool, two in whom advancing local malignancy was thought to be contributing to the anastomotic disruption, and two who died before closure. One patient with obvious colonic ischaemia developed chronic pelvic sepsis, eventually requiring excision of the 'neorectum' 5 years after the initial operation.

One hundred and twenty stomas have been successfully closed with no mortality. However, closure of the loop colostomy was associated with seven faecal fistulae. all of which closed spontaneously. Ten patients have subsequently developed an incisional hernia.

196 0007-1 32319 1/02019&03 I99 I Butterworth-Heinemann Ltd

Defunctioning the low anastomosis: N. D. Karanjia et al.

Table I Chrorzologicul detai1,r q f f i i v patients who developedfrrecal peritoniris due to anristoniotic failure

Patient no. Sex

1 M - F 3 M 4 M 5 M

~. ~~

7

Dukes’ classification

B C C B A

Anastomotic height (cm)

6 3 3 4 5

Bowels opened (day)

3 7

10 5 4

Diarrhoea occurred (day)

4 10 25

6 5

Abdominal pain (day 1

4 12 25 6 5

Insidious bacteraemic shock

No Yes Yes Yes Yes

-

Delay to laparotomy ( h ) Outcome

< 1 Alive 36 Dead 12 Dead

5 Alive 8 Alive

Non-defunctioned group (n = 75) In this group 50 patients had an anastomosis at 6 cm or lower and 25 had an anastomosis that was higher than 6 cm and less than 12 cm. Overall the mean(s.d.) anastomotic height was 6.0(2.0) cm. The mean anastomotic height for the entire group was significantly greater than for the defunctioned group ( P < 0.05).

Although the exact incidence of anastomotic leakage is unknown in this group, 11 (15 per cent) patients developed symptoms which could be attributed to anastomotic leakage. In all 11 the anastomoses were end-to-end and were 6 cm or less from the anal verge.

In two of these patients no active intervention was necessary and their symptoms settled spontaneously. A further three patients required construction of a proximal colostomy; in two anastomotic disruption was found to be due to colonic ischaemia and in one patient a plastic drain was found to have eroded through the anastomosis. Of these three patients, two developed a rectovaginal fistula.

Six (8 per cent) of the patients who did not have a defunctioning stoma developed faecal peritonitis due to anastomotic failure and three of these patients had undergone non-curative resection due to extensive local disease. Comparison of the defunctioned and non-defunctioned groups of patients revealed that occurrence of faecal peritonitis was significantly greater in the patients who had not been defunctioned ( P <0.01). One of the six patients with peritonitis died from overwhelming sepsis before a laparotomy could be performed. The remaining five patients underwent a second laparotomy which confirmed the faecal peritonitis. Peritoneal toilet was performed, a defunctioning stoma was raised, and a distal colonic washout was carried out. The clinical features of these patients are given in Table 1. In each patient the blood supply of the colon appeared satisfactory. The anastomoses did not show evidence of gross disruption at reoperation except that leakage had occurred through a small defect producing general faecal peritonitis. It was not obvious in any of these cases that a retrorectal collection of blood or pus was the underlying cause of leakage from the anastomosis. In each patient convalescence had been smooth in the early days after operation. Each patient was initially thought to be progressing satisfactorily and each had an apparently normal bowel action. During subsequent days an attack of diarrhoea was followed by mild lower abdominal pain and insidious bacteraemic shock with minimal physical signs. This occurred between the fourth and 25th day after operation (Table 1 ) . The time from onset of pain to the second laparotomy varied from 3 to 36 h. Three patients with a delay of less than 8 h recovered completely and later had their stomas closed after radiological confirmation of an intact anastomosis. The two patients in whom there was a delay of 12 h or more died.

Discussion It is increasingly considered that a defunctioning colostomy may be safely dispensed with in low anterior r e ~ e c t i o n ~ , ~ . This is of benefit to the majority of patients who recover, but the

absence of a protective stoma still creates a potentially hazardous situation.

A low local recurrence rate is attributable mainly to total mesorectal excision and, to a lesser extent, the careful exclusion of exfoliated malignant cells from the operative Total mesorectal excision does, however, leave behind a large fixed cavity in the pelvis in which a haematoma or abscess may form. Furthermore, after total mesorectal excision even an anas- tomosis as low as 3 cm from the anal verge is effectively intraperitoneal. Much of the pain-sensitive peritoneum is removed from the posterior pelvis so that the development of bacteraemic shock may be insidious. Spreading peritonitis is all too easily missed after operation and anastomotic leakage should be considered in any case of unexplained deterioration. In the patients without a defunctioning stoma in the present study who developed leakage, anastomotic healing and postoperative recovery were both apparently proceeding normally, but peritonitis with minimal early symptoms and signs occurred in five patients.

In conventional and in higher anterior resections there is probably sufficient distensible rectum below the anastomosis to accommodate the arrival of a faecal bolus. Furthermore, the normal rectoanal inhibitory reflex still exists and may allow anal relaxation and escape of faeces. Many surgeons provide a ‘safety valve’ by stretching the anal sphincter, but we have avoided this because of concern about long-term incontinence and soiling. In the low anterior resection described in this series there is virtually no rectum to accommodate the contents of a peristaltic wave. The normal rectoanal inhibitory reflex may be absent in the early postoperative period and descending faeces might impinge on a tightly closed sphincter. Consequently, the anastomosis could be placed under stress by the progressive collection of faeces which may then escape through any areas of weakness in the anastomotic ring.

It is interesting that an attack of diarrhoea occurred in five out of six patients at the time the patient became unwell. Although it is not possible to determine whether an episode of diarrhoea was the cause or the effect of the anastomotic disruption, it is reasonable to speculate that intense peristaltic activity above a closed sphincter in combination with liquid stool provides the ideal circumstance for anastomotic leakage. It is precisely this set of circumstances which is prevented by defunctioning.

In this study, the patients who did not have a defunctioning stoma were those who were considered to be at minimal risk of anastomotic leakage. Despite this, peritonitis occurred in 8 per cent, ten times more often than in the defunctioned group. Thus it appears to be impossible to predict which patients are suitable for a single-stage procedure. The single case of peritonitis in the defunctioned group shows the need to lavage the colon distal to a defunctioning stoma; proper defunctioning must include such a washout if bowel preparation is less than perfect.

The consequences of leakage are much more serious when a colostomy has not been fashioned. Nearly half of the patients who developed peritonitis and septicaemia as a result of anastomotic leakage died, and anastomotic leakage caused almost half of the overall perioperative mortality rate in the

Br. J . Surg., Vol. 78, No. 1, February 197

Defunctioning the low anastomosis: N. D. Karanjia et al.

series (1.5 per cent of a total 3 5 per cent) including the only death after curative resection in a patient under 70 years of age. Since no mortality or life-threatening morbidity could be attributed in the present series to the construction or closure of a transverse loop colostomy, such a colostomy may be regarded as a sensible precaution when sphincter conservation and wide total mesorectal excision have been undertaken, and when a stapled anastomosis is constructed at 6 cm or less from the anal verge. 6.

2.

3.

4.

5 ,

References 1. Heald RJ, Ryall RDH. Recurrence and survival after total

Karanjia ND, Schache DJ, North WRS, Heald RJ. The close shave in anterior resection. Br J Surg 1990; 77: 510-12. Rosin RD. An obituary to the transverse loop colostomy. ('Gone with the wind'). J R Soc Med 1987; 80: 728-9. Fazio VW. Sump suction and irrigation of the presacral space. Dis Colon Rectum 1978; 21: 401-5. Lockhart-Mummery Sir H, Heald RJ, Hutchings RT. Anterior Resection of the Rectum. Weert, The Netherlands: Wolfe Medical Publications, 1983. Heald RJ, Husband EM, Ryall RDH. The mesorectum in rectal cancer surgery - clue to pelvic recurrence? Br J Surg 1982; 69: 613-16.

mesorectal excision for rectal cancer. Lancet 1986; ii: 1479-82. Paper accepted 6 August 1990

198 Br. J. Surg.,Vol.78, No. 2, February1991