Multicentre analysis of oncological and survival outcomes following anastomotic leakage after rectal...
Transcript of Multicentre analysis of oncological and survival outcomes following anastomotic leakage after rectal...
Original article
Multicentre analysis of oncological and survival outcomesfollowing anastomotic leakage after rectal cancer surgery
M. den Dulk1, C. A. M. Marijnen3, L. Collette4, H. Putter2, L. Pahlman5, J. Folkesson5,J.-F. Bosset6, C. Rodel7, K. Bujko8 and C. J. H. van de Velde1
Departments of 1Surgery and 2Medical Statistics, Leiden University Medical Center, Leiden, The Netherlands, 3Department of Radiotherapy,Netherlands Cancer Institute-Antoni van Leeuwenhoek Hospital, Amsterdam, 4Department of Statistics, European Organization for Research andTreatment of Cancer Data Center, Brussels, Belgium, 5Department of Surgery, Uppsala University Hospital, Uppsala, Sweden, 6Department ofRadiation Therapy, Besancon University Hospital, Besancon, France, 7Department of Radiation Therapy, University of Frankfurt, Frankfurt am Main,Germany, and 8Department of Radiotherapy, Maria Sklodowska-Curie Memorial Cancer Centre, Warsaw, PolandCorrespondence to: Professor C. J. H. van de Velde, Leiden University Medical Center, Department of Surgery, K6-R, PO Box 9600, 2300 RC Leiden, TheNetherlands (e-mail: [email protected])
Background: The association between diverting stomas and symptomatic anastomotic leakage afterrectal cancer surgery was studied, as well as the impact of leakage on local recurrence, distant metastasis,and disease-free, overall and cancer-specific survival.Methods: Data from the Swedish Rectal Cancer Trial, Dutch TME trial, CAO/ARO/AIO-94 trial,EORTC 22921 trial and Polish Rectal Cancer Trial were pooled (n = 5187). All eligible patients withoutdistant metastases at the time of low anterior resection were selected (n = 2726); overall survival wasstudied in patients aged 75 years or less (n = 2480). Multivariable models were used to study theassociation between diverting stomas and anastomotic leakage, and between leakage and recurrence orsurvival.Results: Some 9·7 per cent of patients were diagnosed with a symptomatic anastomotic leak; divertingstomas were negatively associated with leakage (11·6 per cent without and 7·8 per cent with a stoma;P = 0·002). Anastomotic leakage was negatively associated with overall survival in the multivariableanalysis (hazard ratio (HR) 1·29 (95 per cent confidence interval 1·02 to 1·63); P = 0·034), but not withcancer-specific survival (HR 1·12 (0·83 to 1·52); P = 0·466).Conclusion: Diverting stomas were associated with less symptomatic anastomotic leakage. Oncologicaloutcome was not significantly influenced by leakage, but overall survival was reduced.
Paper accepted 21 April 2009Published online in Wiley InterScience (www.bjs.co.uk). DOI: 10.1002/bjs.6694
Introduction
Surgery is the cornerstone in the treatment of rectal cancer.Widespread propagation of standardized total mesorectalexcision (TME) has improved overall survival1,2. However,TME may be associated with an increased risk ofdeveloping anastomotic leakage3 with attendant morbidityand mortality in the postoperative period4,5. Leaksmay be associated with decreased local control6–11
and survival7,12,13. Therefore, the rate of (symptomatic)
The Editors have satisfied themselves that all authors have contributedsignificantly to this publication
anastomotic leakage has been considered as one of thequality indicators of surgical performance14.
Studies to identify risk factors for anastomotic problemsand methods to reduce symptomatic leaks are clearlyimportant15,16. At the end of last century, two smallrandomized trials tested the hypothesis that a divertingstoma reduces the incidence of anastomotic leakage17,18.Although both trials showed fewer anastomotic leaks withstoma use, the difference was not statistically significant. Alarger randomized trial concluded that a diverting stomasignificantly reduced the risk of symptomatic anastomoticleakage19.
In this study, the data from five large Europeanrandomized clinical trials were pooled to study the
Copyright 2009 British Journal of Surgery Society Ltd British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
Anastomotic leak after rectal cancer surgery 1067
association between the creation of a diverting stoma andthe rate of symptomatic leakage after a (low) anteriorresection for rectal cancer. In addition, the impact ofanastomotic leakage on the rate of local recurrence,distant metastasis, disease-free survival, overall survivaland cancer-specific survival was investigated.
Methods
Patient and treatment variables of the following five trialswere pooled: Swedish Rectal Cancer Trial20, Dutch TMEtrial21, German CAO/ARO/AIO-94 trial22, EuropeanOrganization for Research and Treatment of Cancer(EORTC) 22 921 trial23 and the Polish Rectal CancerTrial24. The period of inclusion, randomization arms andnumber of included patients are shown in Table 1. Fromthis pooled database of treatment variables, all eligiblepatients treated with a low anterior resection and withoutdistant metastases at time of surgery were selected. Inthe Swedish Rectal Cancer Trial no data on stomas wereavailable, although stomas in that trial were rarely used asvery high anastomoses were usually created. The patientsfrom that trial were thus excluded from all analyses relatedto stomas. The fifth edition of the tumour node metastasis(TNM) classification of malignant tumours was used todetermine the TNM stage25. To control the analyses ofoverall survival, disease-free survival and cancer-specificsurvival for different age limits allowed in the various trials,those analyses were restricted to patients aged 75 years orless.
In the included trials, only symptomatic anastomoticleakages were documented. Anastomotic leakage wasdefined as clinically apparent leakage such as faecaldischarge from pelvic drain or abdominal wound,or radiologically, endoscopically or surgically provenanastomotic leakage in symptomatic patients such as thosewith peritonitis.
Endpoints, variables and statistical analysis
The χ2 test was used for comparisons of categoricalvariables. Univariable and multivariable logistic regressionanalyses were performed with the following variables tostudy their association with anastomotic leakage: sex,age, distance of tumour from anal verge, TNM stageand presence of a stoma. The multivariable analysis wasadjusted for trial and randomization arms.
To study the effects of anastomotic leakage on localrecurrence, distant metastasis, overall survival, disease-free survival and cancer-specific survival, Cox regressionanalyses were used, stratified for trial and randomization
Table 1 Period of inclusion, randomization arms and number ofpatients per trial for all patients included in the five trials
Trial Period Randomization n
Swedish Rectal 1987–1990 Preop. 5 × 5 Gy RT 1180Cancer Trial Surgery alone
Dutch TME trial 1996–1999 Preop. 5 × 5 Gy RTwith TME
1861
TME alone
German 1995–2002 Preop. CRT 823CAO/ARO/AIO-94 trial
Postop. CRT
EORTC 22921 trial 1993–2003 Preop. 45 Gy RT 1011Preop. CRTPreop. 45 Gy RT and
postop. CTPreop. CRT and
postop. CT
Polish Rectal CancerTrial
1999–2002 Preop. 5 × 5 Gy RTwith TME
312
Preop. CRT with TME
Total 5187
RT, radiotherapy; TME, total mesorectal excision; CRT,chemoradiotherapy; EORTC, European Organization for Research andTreatment of Cancer; CT, chemotherapy.
arms. The following confounders were first studied byunivariable analysis: sex, age, distance of tumour from analverge, TNM stage and circumferential resection margin(CRM) involvement. Variables with P ≤ 0·100 were thenentered in the multivariable Cox regression models. Apositive CRM was defined as microscopic or macroscopictumour in the resection margin (not available in theSwedish Rectal Cancer Trial). Time to local recurrence,distant metastases and overall survival were calculated as thetime from surgery to local recurrence, distant metastasesand death respectively. For overall survival, the analyseswere performed first for all selected patients and then witha landmark selection excluding all patients who died within90 days after surgery to correct for short-term mortalityassociated with anastomotic leakage itself. Disease-freesurvival, defined as time from surgery to first event oflocal recurrence, distant metastases or death, and cancer-specific survival, defined as time from surgery to death dueto rectal cancer, were studied only using the landmarkselection excluding patients with 90-day postoperativemortality. The probability of local recurrence is reportedas cumulative incidences, with death as the competingrisk; cancer-specific survival is reported as 1 - cumulativeincidence, with death from causes other than rectal canceras the competing risk26.
Data were analysed with the statistical package SPSS
version 14.0 for Windows (SPSS, Chicago, Illinois,
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
1068 M. den Dulk, C. A. M. Marijnen, L. Collette, H. Putter, L. Pahlman, J. Folkesson, J.-F. Bosset, C. Rodel, K. Bujko and C. J. H. van de Velde
USA). A two-sided P value ≤ 0·050 was considered tobe statistically significant.
Results
In total, 5187 patients were included in the Swedish RectalCancer Trial, Dutch TME trial, German CAO/ARO/AIO-94 trial, EORTC 22 921 trial and the Polish Rectal CancerTrial. Reasons for exclusion and number of patients areshown in Fig. 1. Of 1962 patients who had a procedureother than low anterior resection, 1749 were treated withan abdominoperineal resection. For the analyses, 2726patients (52·6 per cent) were included. Patient and diseasecharacteristics of these patients are shown in Table 2. Themedian follow-up was 5·9 (range 0·2–14·9) years. Overall,disease-free and cancer-specific survival was studied in 2480of these 2726 patients, who were aged 75 years or less.
Anastomotic leakage
In total, 264 patients (9·7 per cent) were diagnosedwith anastomotic leakage. No information on stomaconstruction was available for the Swedish Rectal CancerTrial (n = 430). Therefore, these patients were excludedin the analyses relating to stomas, for which 2296patients were studied. In 1226 patients (53·4 per cent) astoma was constructed; in 1067 patients (46·5 per cent) nostoma was created; status was unknown for three patients(0·1 per cent). Symptomatic anastomotic leakage occurredin 124 patients (11·6 per cent) without a stoma, whereas itwas diagnosed in 96 patients (7·8 per cent) with a stoma(P = 0·002).
Table 3 shows the results of the univariable andmultivariable analyses for risk factors associated withanastomotic leakage. From the univariable analyses, bothsex and the presence of a diverting stoma were selected forentry in the multivariable analysis (P ≤ 0·100). Trial andtreatment arms were entered in the analysis as adjustment.Female sex and the presence of a diverting stoma wereboth independently associated with a reduced chance ofdeveloping symptomatic anastomotic leakage.
The anastomotic leakage rates per trial and random-ization arm are shown in Table 4. In none of the trialswas a significant difference found between the randomizedtreatment arms.
Of patients without anastomotic leakage, 1·3 per cent(33 of 2446) died within 30 days of surgery, whereasthe 30-day mortality rate after anastomotic leakage was5·7 per cent (15 of 263) (P < 0·001). For one patientwith anastomotic leakage, no details on death status wereavailable.
Included int the five trialsn = 5187
Eligible patientsn = 5063
No distant metastases attime of surgery
n = 4688
Low anterior resectionn = 2726
Age ≤ 75 yearsn = 2480
Age > 75 yearsn = 246
Ineligible patientsn = 124
Distant metastases at timeof surgery
n = 375
Procedure other than lowanterior resection
n = 1962
Fig. 1 Flow diagram of selected and excluded patients
Anastomotic leakage and local recurrence
Anastomotic leakage was not associated with localrecurrence in the univariable analysis and therefore notentered in the multivariable analysis; the 5-year localrecurrence rate was 8·8 (95 per cent confidence interval(c.i.) 7·6 to 10·0) per cent for patients without anastomoticleakage and 12·0 (95 per cent c.i. 7·4 to 16·5) per cent forthose with a leak (P = 0·103). The cumulative incidence oflocal recurrence with death as competing risk for patientswith and without anastomotic leakage is depicted in Fig. 2a.
Anastomotic leakage and distant metastasis
The univariable analysis for the association betweenanastomotic leakage and distant metastases was not
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
Anastomotic leak after rectal cancer surgery 1069
Table 2 Patient and tumour characteristics of the patientpopulation after selection of all eligible patients without distantmetastases at the time of surgery who underwent low anteriorresection
No. of patients(n = 2726)
SexF 1018 (37·3)M 1708 (62·7)
Age (years)
≤ 60 1008 (37·0)61–70 1007 (36·9)> 70 711 (26·1)
Trial
Swedish Rectal Cancer Trial 430 (15·8)Dutch TME trial 1132 (41·5)German CAO/ARO/AIO-94 trial 495 (18·2)EORTC 22921 trial 502 (18·4)Polish Rectal Cancer Trial 167 (6·1)
Distance of tumour from anal verge (cm)
≥ 5·0 2197 (80·6)< 5·0 500 (18·3)Unknown 29 (1·1)
TNM stage
0/I 951 (34·9)II 804 (29·5)III 954 (35·0)Unknown 17 (0·6)
CRM involvement
No 2070 (75·9)Yes 87 (3·2)Unknown 569 (20·9)
Stoma*
No 1067 (46·5)Yes 1226 (53·4)Unknown 3 (0·1)
Anastomotic leak
No 2452 (89·9)Yes 264 (9·7)Unknown 10 (0·4)
*Excludes 430 patients from the Swedish Rectal Cancer Trial for whomno data on stoma construction were available. TME, total mesorectalexcision; TNM, tumour node metastasis; CRM, circumferential resectionmargin. Values in parentheses are percentages.
significant; the rate of distant metastasis at 5 years was 25·6(95 per cent c.i. 23·7 to 27·3) per cent and 27·5 (95 per centc.i. 21·4 to 33·6) per cent respectively for patients withoutand with anastomotic leakage (P = 0·480). Therefore,no multivariable analysis with anastomotic leakage wasperformed for distant metastases.
Table 3 Univariable and multivariable logistic regression analysisof risk factors associated with anastomotic leakage
Univariable analysis Multivariable analysis
Odds ratio P Odds ratio P
Sex 0·002 0·002F 1·00 1·00M 1·56 (1·18, 2·07) 1·64 (1·20, 2·24)
Age (years) 0·956 —≤ 60 1·0061–70 1·00 (0·74, 1·34) 0·975> 70 0·95 (0·69, 1·32) 0·780
Distance of tumourfrom anal verge(cm)
0·949 —
≥ 5·0 1·00< 5·0 0·99 (0·71, 1·38)
TNM stage 0·608 —0/I 1·00II 1·14 (0·83, 1·57) 0·418III 1·15 (0·85, 1·57) 0·362
Stoma 0·002 0·001No 1·00 1·00Yes 0·65 (0·49, 0·85) 0·62 (0·47, 0·82)
Values in parentheses are 95 per cent confidence intervals. TNM, tumournode metastasis.
Anastomotic leakage and overall survival
First, the analyses were performed with all selectedpatients. Anastomotic leakage was significantly associatedwith a worse overall survival rate in both the univariableanalysis (hazard ratio (HR) 1·49 (95 per cent c.i. 1·20 to1·84); P < 0·001) and the multivariable analysis (HR 1·48(95 per cent c.i. 1·19 to 1·83); P < 0·001). The 5-yearoverall survival rate was 74·4 (95 per cent c.i. 72·4 to 76·4)per cent for patients without anastomotic leakage comparedwith 66·4 (95 per cent c.i. 60·1 to 72·7) per cent for thosewith leakage (P < 0·001).
Table 5 shows the results of both the univariable andmultivariable analyses for risk factors associated withoverall survival, excluding patients who died within 90 daysof surgery (n = 52). The 5-year overall survival rate was75·5 (95 per cent c.i. 73·4 to 77·4) per cent for patientswithout anastomotic leakage versus 71·5 (95 per cent c.i.62·2 to 77·8) per cent for those with a leak (P = 0·030).Male sex, age above 70 years, advanced TNM stageand postoperative anastomotic leakage were associatedwith diminished overall survival in both univariable andmultivariable analyses. Kaplan–Meier curves for overallsurvival are shown for all patients in Fig. 2b and excludingpatients who died in the first 90 postoperative days inFig. 2c.
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
1070 M. den Dulk, C. A. M. Marijnen, L. Collette, H. Putter, L. Pahlman, J. Folkesson, J.-F. Bosset, C. Rodel, K. Bujko and C. J. H. van de Velde
Table 4 Anastomotic leak rates and univariable logistic regression analysis for anastomotic leak per trial and randomization arm
Univariable analysis
n Anastomotic leak rate (%)* Odds ratio† P
Swedish Rectal Cancer Trial 0·283Surgery only 209 18 (8·6) 1·005 × 5 Gy RT + surgery 221 26 (11·8) 1·41 (0·75, 2·67)
TME trial‡ 0·418TME alone 578 65 (11·2) 1·005 × 5 Gy RT + TME 553 54 (9·8) 0·85 (0·58, 1·25)
CAO/ARO/AIO-94 trial§ 0·609Preop. CRT 241 39 (16·2) 1·00Postop. CRT 248 36 (14·5) 0·88 (0·54, 1·44)
EORTC 22921 trial —Preop. RT 122 0 (0·0) NEPreop. CRT 125 0 (0·0) NEPreop. RT + postop. CT 122 4 (3·3) NEPreop. CRT + postop. CT 133 4 (3·0) NE
Polish Rectal Cancer Trial¶ 0·657Preop. CRT 81 8 (10) 1·00Preop. 5 × 5 Gy RT 83 10 (12) 1·25 (0·47, 3·35)
Values in parentheses are *percentages and †95 per cent confidence intervals. Leakage unknown for ‡one, §six and ¶three patients. Odds ratios were notestimable (NE) for the European Organization for Research and Treatment of Cancer (EORTC) trial owing to the small number of patients withanastomotic leakage. RT, radiotherapy; TME, total mesorectal excision; CRT, chemoradiotherapy. Because of differences in trial design and datacollection, anastomotic leak rates are not comparable between trials.
Table 5 Univariate and multivariable Cox regression analysis for overall survival excluding patients who died within 90 days of surgery
Univariable analysis Multivariable analysis
n Hazard ratio P§ Hazard ratio P¶
Sex < 0·001 < 0·001F 902 1·00 1·00M 1526 1·43 (1·23, 1·67) 1·33 (1·14, 1·56)
Age (years) < 0·001 < 0·001≤ 60 997 1·00 1·0061–70 984 1·16 (0·98, 1·38) 0·084 1·23 (1·04, 1·46) 0·016> 70 447 1·86 (1·54, 2·25) < 0·001 2·06 (1·70, 2·49) < 0·001
Distance of tumour from anal verge (cm)* 0·466 —≥ 5·0 1939 1·00< 5·0 464 1·08 (0·88, 1·32)
TNM stage† < 0·001 < 0·0010/I 845 1·00 1·00II 712 2·11 (1·70, 2·63) < 0·001 2·08 (1·67, 2·26) < 0·001III 858 3·93 (3·21, 4·81) < 0·001 4·02 (3·28, 4·92) < 0·001
CRM involvement 0·045 0·704No 1848 1·00 1·00Yes 81 1·63 (1·11, 2·39) 0·013 1·17 (0·79, 1·72) 0·442Unknown 499 1·09 (0·76, 1·56) 0·651 0·94 (0·64, 1·40) 0·774
Anastomotic leakage‡ 0·030 0·034No 2199 1·00 1·00Yes 220 1·29 (1·02, 1·63) 1·29 (1·02, 1·63)
Values in parentheses are 95 per cent confidence intervals. Survival unknown for *25, †13 and ‡nine patients. TNM, tumour node metastasis; CRM,circumferential resection margin.
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
Anastomotic leak after rectal cancer surgery 1071
1·0
0·8
0·6
0·4
0·2
0
Loca
l rec
urre
nce
No. at riskNo leakLeak
2452264
2018193
1496148 84
894
2 4 6
Time after surgery (years)
a Local recurrence
1·0
0·8
0·6
0·4
0·2
0
Ove
rall
surv
ival
(al
l pat
ient
s)
No. at riskNo leakLeak
2233237
1913190
1409148 83
837
2 4 6
Time after surgery (years)
b Overall survival for all patients
1·0
0·8
0·6
0·4
0·2
0
Ove
rall
surv
ival
(exc
ludi
ng 9
0–da
y de
aths
)
No. at riskNo leakLeak
2199220
1913190
1409148 83
837
2 4 6
Time after surgery (years)
c Overall survival excluding 90–day deaths
1·0
0·8
0·6
0·4
0·2
Dis
ease
-fre
e su
rviv
al
0
No. at riskNo leakLeak
2199220
1675165
1238128 75
770
2 4 6
Time after surgery (years)
d Disease-free survival
1·0
0·8
0·6
0·4
0·2
0
Can
cer-
spec
ific
surv
ival
No. at riskNo leakLeak
2199220
1849183
1355143 78
790
2 4 6
Time after surgery (years)
e Cancer-specific survival
No anastomoticleakAnastomoticleak
Fig. 2 a Local recurrence, b overall survival for all patients, c overall survival after exclusion of patients who died within 90 days ofsurgery, d disease-free survival and e cancer-specific survival, shown as cumulative incidence (a), Kaplan–Meier survival (b–d) and1 – cumulative incidence (e) curves for patients with and without anastomotic leakage. a P = 0·103, b P < 0·001, c P = 0·030, dP = 0·033, e P = 0·466 (univariate Cox regression analysis)
Anastomotic leakage, stomas and overall survival
When the analyses for overall survival were repeated withthe two variables of anastomotic leakage and stomas inthe model, both were significantly associated with a worseoverall survival (data not shown). However, no statisticalsignificant interaction between anastomotic leakage andstomas could be demonstrated (P = 0·255). Patients witha stoma had an increased risk of death (multivariablemodel: HR 1·24 (95 per cent c.i. 1·04 to 1·48); P = 0·015).Fig. 3a shows Kaplan–Meier curves for overall survivalseparately for patients with/without anastomotic leakageand with/without stomas. Fig. 3b shows the curves for
overall survival excluding patients who died within 90 daysof surgery. The difference between Fig. 3a and Fig. 3bis due to early postoperative mortality. Patients withoutanastomotic leakage and without a stoma fared betterthan the other three groups in the long term. Forpatients with no anastomotic leakage and no stoma, withno leakage and with a stoma, with anastomotic leakageand without a stoma, and with anastomotic leakage andwith a stoma, the 90-day mortality rate was 1·3, 1·9,8·9 and 5·8 per cent respectively. The difference in 90-day postoperative mortality was significant only betweenpatients with and those without anastomotic leakage(P < 0·001).
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
1072 M. den Dulk, C. A. M. Marijnen, L. Collette, H. Putter, L. Pahlman, J. Folkesson, J.-F. Bosset, C. Rodel, K. Bujko and C. J. H. van de Velde
0
No. at riskNo leak, no stomaNo leak, with stomaLeak, no stomaLeak, with stoma
102611286
8608678970
7706407252
5463513726
295
1·0
0·8
0·6
0·4
0·2
Ove
rall
surv
ival
(al
l pat
ient
s)
2 4 6
Time after surgery (years)
a Overall survival for all patients
0
No. at riskNo leak, no stomaNo leak, with stomaLeak, no stomaLeak, with stoma
100710281
8498678970
7706407252
5463513726
295
1·0
0·8
0·6
0·4
0·2
Ove
rall
surv
ival
(exc
ludi
ng 9
0–da
y de
aths
)
2 4 6
Time after surgery (years)
b Overall survival excluding 90–day deaths
No anastomotic leak,no stomaNo anastomotic leak,with stomaAnastomotic leak,no stomaAnastomotic leak,with stoma
Fig. 3 Kaplan–Meier overall survival curves for patientswith/without anastomotic leakage and with/without a stoma: a allpatients and b after exclusion of patients who died within 90 daysof surgery. a P < 0·001, b P = 0·022 (univariate Cox regressionanalysis)
Anastomotic leakage and disease-free andcancer-specific survival
Anastomotic leakage was associated with a worse disease-free survival rate: HR 1·26 (95 per cent c.i. 1·02 to1·56); P = 0·033) in the univariable analysis and HR 1·24(95 per cent c.i. 1·01 to 1·56); P = 0·040) when adjusted forsex, age and TNM stage. The disease-free survival curveis shown in Fig. 2d. The 5-year disease-free survival ratewas 66·9 (95 per cent c.i. 64·9 to 68·9) per cent for patientswithout anastomotic leakage and 60·6 (95 per cent c.i. 53·7to 67·5) per cent for those with leakage (P = 0·033). Theestimates of the cumulative incidence for cancer-relatedmortality with death from causes other than rectal canceras competing risk are shown in Fig. 2e. No significantassociation was found between cancer-specific survival andanastomotic leakage (HR 1·12 (95 per cent c.i. 0·83 to1·52); P = 0·466); the 5-year cancer-specific survival ratewas 80·6 (95 per cent c.i. 78·8 to 82·4) per cent for patientswithout and 79·5 (95 per cent c.i. 73·6 to 85·4) per cent forpatients with anastomotic leakage (P = 0·466).
Discussion
In this study, patient data from five large randomizedEuropean trials for rectal cancer were pooled. Although thedecision to create a stoma was left to the discretion of thesurgeon, and each individual trial was not designed to studyanastomotic leakage, the present results are interestingowing to the large number of patients included from severalEuropean countries with a long and well documentedfollow-up. Anastomotic leakage was significantly reducedwhen a diverting stoma was created. Interestingly, leakswere associated with decreased disease-free and overallsurvival rates, but oncological outcome measures (localrecurrence, distant metastases and cancer-specific survival)were not affected.
Apart from the early consequences after a leak, suchas sepsis-related death, anastomotic failure has beenreported to be associated with decreased local control6–11
and survival7,12,13. However, the association betweenanastomotic leakage and local control cannot be confirmedin all studies: in a population-based cohort study inNorway (1958 patients), anastomotic leakage did notresult in an increased local recurrence rate27. In thepresent study, anastomotic leakage was associated withboth reduced disease-free survival and overall survivalrates. Interestingly, when excluding early postoperativemortality, overall survival in the groups with and withoutanastomotic leakage was very similar in the first 4 years.After 4 years, however, overall survival decreased inpatients who had a symptomatic leak. In the present
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
Anastomotic leak after rectal cancer surgery 1073
analysis, no association was found between anastomoticleakage and cancer-specific survival, although in otherstudies such an association was demonstrated7,12,13.Apparently, patients in the pooled database who developedanastomotic leakage had a higher chance of dying thanthose without anastomotic leakage, but mainly owing tocauses other rather than rectal cancer.
The observed consequences of anastomotic leak-age – early and late morbidity and mortality – stress theimportance of decreasing the incidence of (symptomatic)anastomotic leakage. One of the options is to create adiverting stoma. Recently, Matthiessen and colleagues19
performed a randomized trial in 234 patients who under-went a low anterior resection. Patients were randomizedbetween a diverting loop stoma and no stoma. In this studyit was found that a diverting stoma decreased the rate ofsymptomatic anastomotic leakage. Huser and co-workers28
did a systematic review and meta-analysis of 27 retrospec-tive and four randomized clinical trials on the role of adiverting stoma in low rectal cancer surgery. They con-cluded that a diverting stoma reduces the rate of clinicallyrelevant anastomotic leakage and is thus recommended inoperations for low rectal cancer. Nevertheless, it should notbe forgotten that stoma closure is also associated with mor-bidity and mortality29,30. In addition, one in five divertingstomas is never closed31.
In the present analysis, patients without leakage andwithout a stoma had a better survival than those withno leakage and with a stoma. As the pooled studiesdid not randomize between stoma and no stoma (thedecision to create a stoma was left to the discretionof the surgeon), there is probably a selection bias here.However, this reflects daily clinical practice and it may bepossible that patients with a stoma had more co-morbiditythan those without a stoma. Even so, patients with astoma had less symptomatic leakage, and postoperativemortality after anastomotic leakage tends to be lower witha stoma (5·8 versus 8·9 per cent), although this was notstatistically significant. Owing to the above-mentionedbias, the question of whether the presence of a stoma (asan isolated variable) might improve overall survival cannotbe answered by this study.
Many observational studies have examined the asso-ciation between preoperative treatment and anastomoticleakage. In national population-based studies in Swedenand Norway, preoperative radiotherapy was found to beassociated with anastomotic leakage27,32. Similarly, in acase–control study using the Swedish Cancer Registry,preoperative radiotherapy was found to be a risk factorfor anastomotic leakage15. In randomized trials, how-ever, there is no association between anastomotic leak and
short-course radiotherapy16,33. Owing to different treat-ment protocols and other variance, anastomotic leakagerates cannot be fairly compared across trials, althoughcomparison within each trial is valid. In none of the fiverandomized trials discussed here was there a significantdifference in anastomotic leak rate due to preoperativetreatment, but trials are notorious for not necessarilyreflecting real practice. Indeed, based on the real-lifeobservational studies15,27,32, there are clearly other (con-founding) factors that affect the selection of patients forpreoperative radiotherapy contribute to the higher risk ofleak.
Anastomotic leakage cannot be avoided but its conse-quences can be limited by a diverting stoma28,34. Apartfrom a diverting stoma, some have found that placementof a pelvic drain can limit the consequences of anasto-motic leakage16, although others could not find such anassociation35. Nevertheless, prompt diagnosis and treat-ment of anastomotic leakage is necessary to limit morbid-ity and mortality. Standardized postoperative surveillanceresults in early identification of and reduced mortality fromsymptomatic anastomotic leakage4.
Acknowledgements
M.d.D. is supported by a Quality Assurance Fellowshipof the European Society of Surgical Oncology. Theauthors are grateful to all institutes that participated in theSwedish Rectal Cancer Trial, Dutch TME trial, GermanCAO/ARO/AIO-94 trial, EORTC 22921 trial and thePolish Rectal Cancer Trial. The authors declare no conflictof interest.
References
1 Wibe A, Møller B, Norstein J, Carlsen E, Wiig JN, Heald RJet al. A national strategic change in treatment policy for rectalcancer – implementation of total mesorectal excision asroutine treatment in Norway. A national audit. Dis ColonRectum 2002; 45: 857–866.
2 Kapiteijn E, Putter H, van de Velde CJ, Cooperativeinvestigators of the Dutch ColoRectal Cancer Group. Impactof the introduction and training of total mesorectal excisionon recurrence and survival in rectal cancer in theNetherlands. Br J Surg 2002; 89: 1142–1149.
3 Carlsen E, Schlichting E, Guldvog I, Johnson E, Heald RJ.Effect of the introduction of total mesorectal excision for thetreatment of rectal cancer. Br J Surg 1998; 85: 526–529.
4 den Dulk M, Noter SL, Hendriks ER, Brouwers MA, van derVlies CH, Oostenbroek RJ et al. Improved diagnosis andtreatment of anastomotic leakage after colorectal surgery.Eur J Surg Oncol 2009; 35: 420–426.
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
1074 M. den Dulk, C. A. M. Marijnen, L. Collette, H. Putter, L. Pahlman, J. Folkesson, J.-F. Bosset, C. Rodel, K. Bujko and C. J. H. van de Velde
5 Hallbook O, Sjodahl R. Anastomotic leakage and functionaloutcome after anterior resection of the rectum. Br J Surg1996; 83: 60–62.
6 Branagan G, Finnis D, Wessex Colorectal Cancer AuditWorking Group. Prognosis after anastomotic leakage incolorectal surgery. Dis Colon Rectum 2005; 48: 1021–1026.
7 Law WL, Choi HK, Lee YM, Ho JW, Seto CL.Anastomotic leakage is associated with poor long-termoutcome in patients after curative colorectal resection formalignancy. J Gastrointest Surg 2007; 11: 8–15.
8 Petersen S, Freitag M, Hellmich G, Ludwig K. Anastomoticleakage: impact on local recurrence and survival in surgery ofcolorectal cancer. Int J Colorectal Dis 1998; 13: 160–163.
9 Bell SW, Walker KG, Rickard MJ, Sinclair G, Dent OF,Chapuis PH et al. Anastomotic leakage after curative anteriorresection results in a higher prevalence of local recurrence.Br J Surg 2003; 90: 1261–1266.
10 Jung SH, Yu CS, Choi PW, Kim DD, Park IJ, Kim HC et al.Risk factors and oncologic impact of anastomotic leakageafter rectal cancer surgery. Dis Colon Rectum 2008; 51:902–908.
11 Ptok H, Marusch F, Meyer F, Schubert D, Gastinger I,Lippert H. Impact of anastomotic leakage on oncologicaloutcome after rectal cancer resection. Br J Surg 2007; 94:1548–1554.
12 McArdle CS, McMillan DC, Hole DJ. Impact of anastomoticleakage on long-term survival of patients undergoing curativeresection for colorectal cancer. Br J Surg 2005; 92:1150–1154.
13 Walker KG, Bell SW, Rickard MJ, Mehanna D, Dent OF,Chapuis PH et al. Anastomotic leakage is predictive ofdiminished survival after potentially curative resection forcolorectal cancer. Ann Surg 2004; 240: 255–259.
14 Bittner R, Burghardt J, Gross E, Grundmann RT,Hermanek P, Isbert C et al. [Quality indicators for diagnosticand therapy of rectal carcinoma.] Zentralbl Chir 2007; 132:85–94.
15 Jestin P, Pahlman L, Gunnarsson U. Risk factors foranastomotic leakage after rectal cancer surgery: acase–control study. Colorectal Dis 2008; 10: 715–721.
16 Peeters KC, Tollenaar RA, Marijnen CA, KleinKranenbarg E, Steup WH, Wiggers T et al. Risk factors foranastomotic failure after total mesorectal excision of rectalcancer. Br J Surg 2005; 92: 211–216.
17 Graffner H, Fredlund P, Olsson SA, Oscarson J,Petersson BG. Protective colostomy in low anterior resectionof the rectum using the EEA stapling instrument. Arandomized study. Dis Colon Rectum 1983; 26: 87–90.
18 Pakkastie TE, Ovaska JT, Pekkala ES, Luukkonen PE,Jarvinen HJ. A randomised study of colostomies in lowcolorectal anastomoses. Eur J Surg 1997; 163: 929–933.
19 Matthiessen P, Hallbook O, Rutegard J, Simert G,Sjodahl R. Defunctioning stoma reduces symptomaticanastomotic leakage after low anterior resection of therectum for cancer: a randomized multicenter trial. Ann Surg2007; 246: 207–214.
20 Swedish Rectal Cancer Trial participants. Improved survivalwith preoperative radiotherapy in resectable rectal cancer.N Engl J Med 1997; 336: 980–987.
21 Kapiteijn E, Marijnen CA, Nagtegaal ID, Putter H,Steup WH, Wiggers T et al. Preoperative radiotherapycombined with total mesorectal excision for resectable rectalcancer. N Engl J Med 2001; 345: 638–646.
22 Sauer R, Becker H, Hohenberger W, Rodel C, Wittekind C,Fietkau R et al. Preoperative versus postoperativechemoradiotherapy for rectal cancer. N Engl J Med 2004;351: 1731–1740.
23 Bosset JF, Collette L, Calais G, Mineur L, Maingon P,Radosevic-Jelic L et al. Chemotherapy with preoperativeradiotherapy in rectal cancer. N Engl J Med 2006; 355:1114–1123.
24 Bujko K, Nowacki MP, Nasierowska-Guttmejer A,Michalski W, Bebenek M, Pudelko M et al. Sphincterpreservation following preoperative radiotherapy for rectalcancer: report of a randomised trial comparing short-termradiotherapy vs. conventionally fractionatedradiochemotherapy. Radiother Oncol 2004; 72: 15–24.
25 Sobin LH, Wittekind C. TNM Classification of MalignantTumors (5th edn). John Wiley: New York, 1997.
26 Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics:competing risks and multi-state models. Stat Med 2007; 26:2389–2430.
27 Eriksen MT, Wibe A, Norstein J, Haffner J, Wiig JN;Norwegian Rectal Cancer Group. Anastomotic leakagefollowing routine mesorectal excision for rectal cancer in anational cohort of patients. Colorectal Dis 2005; 7: 51–57.
28 Huser N, Michalski CW, Erkan M, Schuster T,Rosenberg R, Kleeff J et al. Systematic review andmeta-analysis of the role of defunctioning stoma in low rectalcancer surgery. Ann Surg 2008; 248: 52–60.
29 Bakx R, Busch OR, Bemelman WA, Veldink GJ, Slors JF,van Lanschot JJ. Morbidity of temporary loop ileostomies.Dig Surg 2004; 21: 277–281.
30 Duchesne JC, Wang YZ, Weintraub SL, Boyle M, Hunt JP.Stoma complications: a multivariate analysis. Am Surg 2002;68: 961–966.
31 den Dulk M, Smit M, Peeters KC, Kranenbarg EM,Rutten HJ, Wiggers T et al. A multivariate analysis oflimiting factors for stoma reversal in patients with rectalcancer entered into the total mesorectal excision (TME) trial:a retrospective study. Lancet Oncol 2007; 8:297–303.
32 Matthiessen P, Hallbook O, Andersson M, Rutegard J,Sjodahl R. Risk factors for anastomotic leakage after anteriorresection of the rectum. Colorectal Dis 2004; 6:462–469.
33 Swedish Rectal Cancer Trial participants. Initial report froma Swedish multicentre study examining the role ofpreoperative irradiation in the treatment of patients withresectable rectal carcinoma. Br J Surg 1993; 80:1333–1336.
34 Gastinger I, Marusch F, Steinert R, Wolff S, Koeckerling F,
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd
Anastomotic leak after rectal cancer surgery 1075
Lippert H. Protective defunctioning stoma in low anteriorresection for rectal carcinoma. Br J Surg 2005; 92:1137–1142.
35 Merad F, Hay JM, Fingerhut A, Yahchouchi E, Laborde Y,
Pelissier E et al. Is prophylactic pelvic drainage useful afterelective rectal or anal anastomosis? A multicenter controlledrandomized trial. French Association for Surgical Research.Surgery 1999; 125: 529–535.
25 years ago
Local recurrence following ‘curative’ surgery for large bowel cancer: I. The overallpicture
One of the earliest indicators that specialisation improves outcomes in cancer surgery.
‘‘Approximately 38 000 men and women in England and Wales develop a carcinoma of thecolon or rectum each year, half of whom (19 000) can be expected to undergo a curativeremoval of their tumour and survive. Subsequently, 10–15 per cent (2000–3000) maydevelop a local recurrence. . . we have in this paper examined those factors which influencesubsequent development of local recurrence. . .
Between 1976 and 1980 the data on 4228 patients with a histologically provenadenocarcinoma of the large bowel were collected. A local recurrence has been found in 309patients (14 per cent). . .
[There was] a wide range of local recurrence between individual Consultant operators.Selecting those 20 Consultant Surgeons who had entered 30 or more patients (range:31–101) surviving a curative resection performed by the Consultant himself, the incidence oflocal recurrence was: < 5 per cent, 3; 5–10 per cent, 7; 10–15 per cent, 3; 15–20 per cent, 6;> 20 per cent, 1; (P < 0·05, d.f. = 19).
. . . We must assume that overall technical expertise is an essential ingredient for the bestresults and emphasizes the special requirements for those who wish to practise colon andrectal Surgery.’’
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.(DOI: 10·1002/bjs.1800710104)
All BJS articles from volume 1 issue 1 are now available online at www.bjs.co.uk
Copyright 2009 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of Surgery 2009; 96: 1066–1075Published by John Wiley & Sons Ltd