British Journal of Surgery 1997,84,525-528

Local recurrence of low rectal cancer after abdominoperineal anterior resection E. R U L L I E R , C. L A U R E N T , J . C A R L E S , J . S A R I C , P . M I C H E L * and M. P A R N E

Departments of Digestive Surgery and *Medical Statistics, University of Bordeaux, France Correspondence to: Dr E. RullieK Service de Chirurgie Digestive, H6pital Saint-Andri. 33 075 Bordeaux Ceder, France

and

X

Background The aim of this retrospective study was to compare the risk of local recurrence between two groups of patients with low rectal cancer treated by either abdominoperineal resection (APR) or anterior resection.

Methods From 1982 to 1992, 106 low rectal cancers (tumour situated 3-8 cm from the anal verge), of Dukes stage B and C were treated by curative surgery, 52 by APR and 54 by anterior resection. Mean follow-up was 60 months after APR and 50 months after anterior resection.

Results The local recurrence rate of low rectal cancer was 16 of 52 (31 per cent) after APR and 15 of 54 (28 per cent) after anterior resection. The risk of local recurrence in univariate and multivariate analysis was not associated with clinical and histological variables, nor with the type of intervention.

Conclusion Sphincter-saving resection can be performed for low rectal cancer without an increased risk of local recurrence.

Many retrospective studies have shown no increase in the frequency of local recurrence after anterior resection compared with abdominoperineal resection (APR) for rectal .cancer1-”. In spite of the difference in prognosis between low and high rectal cancers re orted by certain authors5-**’**”, the majority of s t ~ d i e s ~ ~ * ’ ~ * ’ ~ that have compared APR and anterior resection include all tumours between 3 and 20cm from the anal verge. Some studiesI0.l6 have evaluated the risk of local recurrence from mid-rectal cancer, but very few authors"^" have evaluated this risk for low rectal cancer only.

The aim of this study was to compare the risk of local recurrence between two homogeneous groups of patients who were treated for Dukes stage B and C low rectal cancer by APR or anterior resection.

Patients and methods From January 1982 to December 1992, 356 consecutive patients with rectal adenocarcinoma were treated surgically. In all, 159 had low rectal cancer, the lower edge of the tumour being situated between 3 and 8cm from the anal verge. The localization of the neoplasm was determined by endoanal clinical examination. From 1982 to 1986 the majority (87 per cent) of patients with low rectal cancer underwent APR because policy was to obtain a distal margin of clearance of the tumour of more than 5 cm. From 1987 to 1992 the distal margin of clearance was reduced to 2cm and the majority (71 per cent) of low rectal cancers were treated by anterior resection. When the sphincter- saving resection was performed for very low tumours, partial resection of the anal canal was performed along with the anterior resection (n = 9). The following were excluded from the study: tumours with a low risk of local recurrence and tumours with synchronous metastases (stages A and D of Dukes modified Gunderson and Sosin classificationlR), palliative resections (clinical or histological residual rectal tumour) and operative deaths (2.8 per cent). The study concerned itself with 106 Dukes stage B and C low rectal cancers, 52 treated by APR and 54 by anterior resection. Thirty-eight (73 per cent) of APRs and 44 (81 per cent) of anterior resections were performed by one surgeon (M.P.).

Paper accepted 12 July 1996

Factors analysed were patient age, sex, tumour size (greatest dimension), circumferential extent of the tumour, tumour height (distance between tumour caudal edge and the anal verge), Dukes stage (Gunderson-Sosin modification), histological grade (well, moderately or poorly differentiated), colloid component and whether patients had preoperative or postoperative radio- therapy. The different methods of anastomosis performed after anterior resection are shown in Table I. The mean(s.d.) distal margin of clearance evaluated after anterior resection on the fresh specimen was 2.7(0.7) cm.

Patients were followed according to clinical, laboratory (carcinoembryonic antigen level), endoscopic and radiological investigations (chest radiograph and liver ultrasonography or liver computed tomography (CT)) every 6 months. Pelvic CT and endoscopic or perineal biopsy was carried out when local recurrence was suspected. The results were analysed from 31 January 1995. For patients whose last consultation went back more than 6 months, the patient or general practitioner was contacted. Patients were then considered free from local recurrence if they had no recent sphincter disorder, no pelvic pain and no neurological or urinary symptoms.

Recurrence was considered local when it occurred within the pelvis with or without distal metastasis”. The date of local recurrence was taken as the date of diagnosis of local recurrence. All patients had at least 2 years follow-up. The mean(s.d.) follow-up was 60(41) months after APR and 50(24) months after anterior resection.

Statistical analysis The association between local recurrence and the type of intervention and other relevant variables was first assessed by univariate log rank tests. The risk of local recurrence according

Table 1 Method of anastomosis after 54 anterior resections

No.

Low colorectal anastomosis Manual 5 (9) Stapled 27 (50)

Stapled 10 (19) Coloanal anastomosis

Transanal manual 12 (22)

Values in parentheses are percentages

0 1997 Blackwell Science Ltd 525

526 E . R U L L I E R , C . L A U R E N T , J . C A R L E S , J . S A R I C , P. M I C H E L and M. P A R N E I X

to the type of intervention was studied using a Cox proportional hazards regression model. The independent variables were the type of intervention, radiotherapy and the variables associated with Iwal recurrence with P < 0.2 in the univariate analysis. Because the Dukes staging was found to be an effect modifier on the relation between the type of intervention and the local recurrence, two separate analyses were performed in the subsamples of stage B and C tumours.

Table 2 Clinical and therapeutic data of 106 patients treated for low rectal cancer

Results The distribution of variables according to the type of intervention is shown in Tables 2 and 3. This distribution was homogeneous for most of the variables in the two groups. The local recurrence rate for rectal cancer was 16 of 52 (31 per cent) after APR and 15 of 54 (28 per cent) after anterior resection (Table 4 ) . The univariate study showed that the risk of local recurrence was not influenced by the different variables (Table 5) . In particular, there was no significant difference in local

APR Anterior resection (n = 52) (n = 54)

Table 5 Association between local recurrence and clinical and histological variables on univariate analysis

~~

Age (years)* 63( 1 1) WI 1) Sex ratio (M : F) 35: 17 28 : 26 Size of tumour (cm)* 5.3( 1.9) 4.8( 1.9) circumferential tumours 12 14

Radiotherapy Height of tumour (cm)* 5.2( 1.9) 6.2( 1.5)

No. of patients 34 30 Mean (range) dose (Gy) 43.6 (30-72) 44.1 (24-61)

*Values are mean(s.d.). APR, abdominoperineal resection

Table 3 Histological data of I 0 6 low rectal tumours

APR Anterior resect ion ~~ ~~

Dukes stage* 12 (23) 16 (30) 17 (33) 12 (22)

Bl B? B, 2 (4) 2 (4)

3 (6) 3 (6) c, C; 3 (6) 0 (0)

Well differentiated 6 (12) 3 (6)

Poorly differentiated 6 (12) 2 (4)

15 (29) 21 (39) C*

Histological grade

Moderately differentiated 40 (77) 49 (91)

Colloid component Present 7(13) lO(19) Absent 45 (87) 44 (81)

Values in parentheses are percentages. APR. abdominoperineal resection. *Modified Gunderson-Sosinlx classification

Table 4 Local recurrence according to Dukes stage and operation

APR Anterior resection

Dukes stage B 1 of 12 5of 17

Total 7 of 31

B, B2 B3 1 o f 2

3 of 3 6of 15

Cl c2 C, 0 of 3 Total 9 of 21

Total 16 of 52

Dukes stage C

3of 16 30f 12 2 o f 2 8 of 30

Oof3 7 of 21

7 of 24 15 of 54

-

~

APR, abdominoperineal resection

Variable P*

Age ( 5 65 versus > 65 years) Sex (M versus F) Tumour size ( 5 5 versus > 5 cm) Circumferential tumour (Present versus absent) Tumour height ( 5 5 versus > 5 cm) Radiotherapy (Yes versus no) Dukes stage (B versus C) Histological grade (Well and moderately versus

poorly differentiated) Colloid component (Present versus absent) Intervention (APR versus anterior resection)

0.71 085 035 0.97 0.54 082 0.15 0.76

0.39 0.97

APR, abdominoperineal resection. *Log rank test

0 $ 0.4 0 5 0.2 0

0 20 40 60 80 100 120 140 160 Time after operation (months)

Fig. 1 Cumulative incidence of local recurrence according to type of intervention (Kaplan-Meier curve). 0, Abdomino- perineal resection; 0, anterior resection

1 .o

0.8 Q 5 0.6 al 0

a*---*. --QD ---. 4D.----Q

0.4 0 5 0.2 0

0 20 40 60 80 100 120 140 160 l7me after operation (months)

Fig. 2 Cumulative incidence of local recurrence according to tumour stage and type of intervention (Kaplan-Meier curve). Abdominoperineal resection: stage C (m) and stage B (0). Anterior resection: stage C (0) and stage B (0)

0 1997 Blackwell Science Ltd, British Journal of Surgety 1997,84,525-528

L O C A L R E C U R R E N C E O F R E C T A L C A N C E R 527

recurrence rate between APR and anterior resection (Fig. 1 ) . Multivariate analysis confirmed that there was no association between type of intervention and local recurrence in either of the two subsamples of stage B and C tumours (Fig. 2).

Discussion To claim that the risk of local recurrence of rectal cancer after anterior resection is no higher than that after APR remains today, as in the past, a considerable challenge. This is all the more true for low rectal tumours. It is difficult to compare two surgical techniques when one has a longer history than the other. APR or Miles' operation has been known" since 1908. Anterior resection was described in 1939 but has been largely developed since 1975 thanks to mechanical staplers2'. Technical progress and modifications of the oncological rules have allowed an increase in the number of anterior resections while the number of APRs performed has been decreasing.

Most teams with experience of rectal surgery have compared the risk of local recurrence between two groups: patients who had APR and those with anterior resection'-"*'"-'7 However, APR was almost always used for low rectal tumours and anterior resection for high rectal tumours. Several s t u d i e ~ ~ . ~ . ' ~ . ' ~ have shown that low rectal tumours have a worse prognosis than high because of the'risk of local recurrence. Of the 16 studies analysed comparing APR and anterior resection, 12 were carried out on all rectal tumours, irrespective of height'-''.l"-". The originality of the present work is that it concentrates on tumours of the lower rectum.

Among the numerous prognostic factors for rectal cancers, the most important is Dukes stage22. In certain studies6.' comparing APR and anterior resection, the number of stage A tumours is different between the two procedures. In another study5, the number of stage C tumours is significantly higher in one of the two groups. Sometimes Dukes stage D tumours have been included in the analysis6. In the present study, stage A and D tumours were excluded to increase the homogeneity of the two groups and separate analysis of stage B and C tumours was performed.

Williams et al." matched patients undergoing anterior resection and APR for prognostic factors such as age, sex, height of tumour and stage. A significant difference between the two groups appeared none the less for histological grade. There were 56 per cent poorly differentiated lesions in patients receiving APR and only 33 per cent in those given anterior resection. In the present study, even if the number of poorly differentiated tumours was greater for APR than anterior resection (12 versus 4 per cent, P = 0*15), histological prognostic factors (stage, grade and colloid component) were not associated with a difference in rate of local recurrence (Toble 5) .

Local infiltration of the tumour is not always identical in the groups that have been studied. Fixed tumours removed at APR and anterior resection represented respectively 34 and 22 per cent in the study of Williams and Johnstonno, and 26 and 17 per cent (P < 0.01) in the multivariate study of Phillips et all4 Fixed tumours, in particular those of the low rectum, have a worse prognosis. In the study of Konn et d5 the local recurrence rate after APR decreased from 25 to 13 per cent when low and fixed tumours were excluded. This increases the need to study the risk of local recurrence in a single segment of the rectum. Indeed, the bad 'anatomical

prognosis' of low rectal cancer could explain the misleadingly reassuring absence of increased local recurrence risk after anterior resection, if this risk is evaluated with all rectal cancers.

The number of women in the present study who underwent anterior resection is higher, although not significantly so (P = O e l ) , than the number who had APR. This sex ratio difference is frequent in s t ~ d i e s * ~ J ~ J ~ comparing APR and anterior resection, because the pelvic dissection is more difficult in men with the more frequently narrow pelvis. Nevertheless, only one group of authorsB has shown that the rectal cancer prognosis was better for women.

To increase the reliability of the results, it was decided first to focus the study on a homogeneous sample, i.e. tumours situated between 3 and 8 cm from the anal verge, of stages B and C. Despite this strict selection criteria, the two intervention groups were still different, in particular for tumour height (P = 0.003). It was therefore decided to use a different statistical approach, i.e. to assess the risk of local recurrence according to the type of intervention and the other prognostic factors.

A recent showed that the time to local recurrence was longer after APR than anterior resection, 32 versus 18 months, because it is easier to diagnose recurrence in the presence of a retained lumen after anterior resection. This delay in diagnosis of local recurrence could be a bias in the assessment of local recurrence between APR and anterior resection, unless the follow-up is very long. However, the follow-up for patients in certain studies-'~~.'~ is insufficient, sometimes only 24months. In this situation, when the analysis is carried out 2 years after the last inclusion, different author^^^^^" remind us that 80 per cent of local recurrences are diagnosed within 2 years of the tumour's excision. Williams et al." insist on the need to study the cumulative incidence of local recurrence over 5 years so as to correct for this lack of follow-up. This is what was done in the present study.

Tuscan0 et al." studied the local recurrence risk exclusively for low rectal tumours. They compared local recurrence in two historical groups: 11 APRs carried out between 1983 and 1986 and 13 anterior resections performed between 1986 and 1989. They concluded that local recurrence risk was no different in the two groups, but the number of patients in their study was small. On the contrary, multicentre s t ~ d i e s ~ . ' ~ . ' ~ can be more conclusive thanks to a higher number of patients. Still, in multicentre studies there often exists a degree of heterogeneity which puts the conclusions into question. Wolmark and Fisheri5 analysed the results of surgery versus surgery with radiotherapy in rectal cancer treatment. They reported a significant increase in local recurrence after anterior resection compared with APR, using the whole group of patients as the base. However, this increase disappears in the group of patients treated by surgery alone. Phillips et al.I4, in another multicentre study, found the rate of local recurrence was 12 per cent after APR and 18 per cent after anterior resection (P<O.O2). However, this was a heterogeneous group of tumours (rectal and rectosigmoid) and the local recurrence risk varied between 5 and 20 per cent according to the type of surgeon (there were 23 centres and 94 surgeons). The absence of details on the local recurrence rate relative to the two surgical procedures and to each team of surgeons prevents a conclusion with absolute certainty. In the present study of 106 patients

0 1997 Blackwell Science Ltd, Erirish Journal ofsurgery 1997, 84,525-528

528 E. R U L L I E R , C . L A U R E N T , J . C A R L E S , J . S A R I C , P . M I C H E L and M. P A R N E I X

with low rectal cancer, all were operated on according t o the same oncological principles, the majority (77 per cent) by one surgeon who was experienced in rectal surgery.

In conclusion, the rate of APR performed for rectal tumours situated between 3 and 8 cm from the anal verge fell from 87 t o 29 per cent when distal clearance was reduced t o 2 cm. This new policy with no increase in local recurrence rate between the two periods suggests that sphincter-saving resection for low rectal cancer is safe in our experience for Dukes stage B and C tumours. However, sphincter-saving resection must not be considered for all low rectal tumours. T h e choice between anterior resection and APR is influenced by the anatomy of the patient (narrow male pelvis, obesity, bulky prostate) and features of the tumour (distance from the anal verge, fixity). T h e difficulty of dissection must not lead us away from the need t o perform rectal resection with a minimum distal intramural clearance of 2 cm, a total mesorectal excision and clear lateral margins. Because most of the present patients were operated on by one surgeon, we think that the best chance of achieving sphincter-saving resection for low rectal cancer without increasing the risk of local recurrence is probably by having a highly experienced surgeon.

References Amato A, Pescatori M, Butti A. Local recurrence following abdominoperineal excision and anterior resection for rectal carcinoma. D b Colon Rectum 1991; 34: 317-22. Fick TE, Baeten CGMI, von Meyenfeldt MF, Obertop H. Recurrence and survival after abdominoperineal and low anterior resection for rectal cancer, without adjunctive therapy. Eur J Surg Oncol 1990; 16: 105-8. Gillen P, Peel ALG. Comparison of the mortality, morbidity and incidence of local recurrence in patients with rectal cancer treated by either stapled anterior resection or abdominoperineal resection. Br J Surg 1986; 73: 339-41. Heimann TM, Szporn A, Bolnick K, Aufses AH Jr. Local recurrence following surgical treatment of rectal cancer. Comparison of anterior and abdominoperineal resection. Dis Colon Rectum 1986; 2 9 862-4. Konn M, Morita T, Hada R et al. Survival and recurrence after low anterior resection and abdominoperineal resection for rectal cancer: the results of a long-term study with a review of the literature. Surg Today 1993; 23: 21-30. Luke M, Kirkegaard P, Lendorf A, Christiansen J. Pelvic recurrence rate after abdominoperineal resection and low anterior resection for rectal cancer before and after introduction of the stapling technique. World J Surg 1983; 7:

Neville R, Fielding LP, Amendola C. Local tumor recurrence after curative resection for rectal cancer. A ten-hospital review. Dis Colon Rectum 1987; 30: 12-17. Pilipshen SJ, Heilweil M, Quan SHQ, Sternberg SS, Enker

616-19.

WE. Patterns of pelvic recurrence following definitive resections of rectal cancer. Cancer 1984; 53: 1354-62.

9 Vlasak JW, Wagner D, Passaro E Jr, Wilson SE. Local recurrence after curative resection of rectal cancer: a comparison of low anterior resection and abdominoperineal resection. J Surg Oncol 1989; 41: 236-9.

10 Williams NS, Johnston D. Survival and recurrence after sphincter saving resection and abdorninoperineal resection for carcinoma of the middle third of the rectum. Br J Surg

1 1 Williams NS, Durdey P, Johnston D. The outcome following sphincter-saving resection and abdominoperineal resection for low rectal cancer. Br J Surg 1985; 72: 595-8.

12 Freedman LS, Macaskill P, Smith AN. Multivariate analysis of prognostic factors for operable rectal cancer. Lancer 1984;

13 Glenn F, McSherry CK. Carcinoma of the distal large bowel: 32-year review of 1026 cases. Ann Surg 1966; 163: 838-49.

14 Phillips RKS, Hittinger R, Blesovsky L, Fry JS, Fielding LP. Local recurrence following ‘curative’ surgery for large bowel cancer: I. The overall picture. Br J Surg 1984; 71: 12-16.

15 Wolmark N, Fisher B. An analysis of survival and treatment failure following abdominoperineal and sphincter-saving resection in Dukes’ B and C rectal carcinoma. A report of the NSABP clinical trials. National Surgical Adjuvant Breast and Bowel Project. Ann Surg 1986; 204: 480-9.

16 Graf W, Pahlman L, Enblad P, Glimelius B. Anterior versus abdominoperineal resections in the management of mid- rectal tumours. Acta Chirurgica Scandinmica 1990; 156:

17 Tuscano D, Catarci M, Saputelli A et al. Low anterior resection versus abdominoperineal excision: a comparison of local recurrence after curative surgery for ‘very low’ rectal cancer. Surg Today 1992; 22: 313-17.

18 Gunderson LL, Sosin H. Areas of failure found at reoperation (second or symptomatic look) following ‘curative surgery’ for adenocarcinoma of the rectum. Clinicopathologic correlation and implications for adjuvant therapy. Cancer

19 Marsh PJ, James RD, Schofield PF. Definition of local recurrence after surgery for rectal carcinoma. Br J Surg 1995;

20 Miles WE. A method of performing abdomino-perineal excision for carcinoma of the rectum and of the terminal portion of the pelvic colon. Lancet 1908; ii: 1812-13.

21 Fain SN, Patin CS, Morgenstern L. Use of a mechanical suturing apparatus in low colorectal anastomosis. Arch Surg

22 Rinnert-Congora S, Tartter PI. Multivariate analysis of recurrence after anterior resection for colorectal carcinoma. Am J Surg 1989; 157: 573-6.

23 Buhre LMD, Mulder NA, De Ruiter AJ, van Loon AJ, Verschueren RCJ. Effect of extent of anterior resection and sex on disease-free survival and local recurrence in patients with rectal cancer. Br J Surg 1994; 81: 1227-9.

24 Parneix M, Rullier E, Laurent C, el Kohen J, Dossou F, Dembele M. Rkcidives loco-regionales des cancers du rectum, deceptions et espoirs, i propos de 54 cas. Chirurgie 1994; 120: 219-26.

1984; 71: 278-82.

29: 733-6.

231-5.

1974; 34: 1278-92.

82: 465-8.

1975; 110: 1079-82.

0 1997 Blackwell Science Ltd, British Jounial of Surgery 1997,84,525-528