ORIGINAL ARTICLE

Changing patterns of recurrence after treatmentfor colorectal cancer

Cameron F. E. Platell

Accepted: 15 March 2007 / Published online: 29 March 2007# Springer-Verlag 2007

AbstractBackground The management of patients with colorectalcancer has changed appreciably over the last 16 years. Theaims of this study were to compare the rates and patterns ofdisease recurrence over the last 10 years with a historicalcontrol group.Materials and methods Data was obtained from a prospec-tive database that had recorded all patients presenting withcolorectal cancer from 1996 to 2006. This data wascompared with a retrospective data set that included allpatients treated with colorectal cancer at the same institu-tion from 1989 to 1995. The Kaplan–Meier technique wasused to calculate the 5 year recurrence and local recurrencerates for the two groups.Results There were 710 patients in the study group and 475patients in the control group. There were more patients withrectal cancer and stage I cancer in the study group. Whencomparing the study group vs the control group, there wasan increase in the time to recurrence (2.1 vs 1.6 years, n.s.)and a decrease in the 5 year recurrence rate for patientsundergoing curative resections (17% [95% CI 12%–20%]vs 42% [95% CI 36%–49%], p<0.001). These changeswere noted for both colon (16% vs 34%, p<0.001) andrectal cancers (18% vs 50%, p<0.001). There was also adecrease in local recurrence in patients with rectal cancer(8.8% [95% CI 4.5%–13.1%] vs 33.6% [95% CI 23.6%–43.6%], p<0.001).Conclusions Within this institution, there has been asignificant trend during the last 16 years towards reduced

disease recurrence, both local and metastatic, and aprolongation in the time to develop recurrence.

Keywords Colon cancer . Rectal cancer . Recurrence .

Local recurrence . Survival . Lateral margins . Metastases .

Liver metastases . Lungmetastases .

Cox proportional hazards model

Introduction

Over the last two decades there have been a number ofnotable advances in the management of patients withcolorectal cancer [1]. These advances include refinementsin surgical technique, the introduction of both adjuvant andpalliative chemotherapy and radiotherapy and improve-ments in imaging with the introduction of computedtomography (CT), magnetic resonance imaging (MRI) andpositron emission tomography (PET) scanners [1]. There-fore, it is relevant to reflect on patient outcomes over thisperiod to see if the advances in management have changedthe patterns of recurrence in patients with colorectal cancer.

The aims of this study were to compare the rates andpatterns of disease recurrence in patients with colorectalcancer managed during the period 1996 to 2006 with ahistorical control group managed at the same institutionfrom 1989 to 1995.

Materials and methods

The study group consisted of patients entered into aprospective database maintained by the Colorectal SurgicalUnit within a public, tertiary referral, teaching hospital.Only patients with biopsy proven adenocarcinoma of thecolon and rectum were included. This database commenced

Int J Colorectal Dis (2007) 22:1223–1231DOI 10.1007/s00384-007-0306-y

C. F. E. Platell (*)St. John of God Colorectal Cancer Centre,212/25 McCourt Street,Subiaco, Perth, WA 6008, Australiae-mail: cplatell@cyllene.uwa.edu.au

in February 1996 and was regularly updated. The controlgroup represented patients diagnosed and managed withcolorectal cancer at the same institution between 1989 and1995. This data was collected and analysed retrospectivelyin 1997. The database was linked to the Cancer Registry ofWestern Australia to retrieve missing data and verify therecorded date and cause of death. The majority of patientswho died did not undergo autopsies, so we could notaccurately record recurrence in patients who died fromcauses other than cancer.

The TNM staging system for colorectal cancer was usedin this study. All patients with rectal cancers underwentimaging of their livers and lung fields before surgery.Staging in patients undergoing local excisions only wasbased upon the findings of endoanal ultrasound.

All the endpoints were defined before the collection ofdata. These were the same endpoints that were used indefining data for the historical control group. The surgicalprocedure was termed curative if there was macroscopicremoval of the primary tumour with histological assessmentshowing clear margins (R0 resection) and where there wasno obvious metastatic disease. If isolated, synchronousmetastatic disease was completely excised in conjunctionwith the primary resection, then the procedure was alsotermed curative. A palliative procedure was where obviousmacroscopic tumour remained after surgery (R2 resection)or if there was a positive surgical margin (R1 resection, i.e.tumour within 1 mm of the resection line). The presence ofunresectable metastases also deemed a procedure palliative.

Recurrence was defined as the development of eitherbiopsy proven or radiological evidence (including apositive PET scan) of tumour regrowth after the initialsurgery. For rectal cancers, local recurrence was defined ascancer detected in the pelvis regardless of whether newmetastases were found elsewhere. Isolated local recurrencewas defined as cancer detected in the pelvis only. For coloncancers, local recurrence was defined as cancer detectedwithin the local anatomical region of the primary resection.

Further definitions included:

– The rectum commences in the area where the taeniacoli of the sigmoid colon coalesce into a uniform outerlongitudinal muscular wall. At colonoscopy, it includesthe area up to 18 cm from the anal verge.

– American Society of Anaesthesia (ASA) score of 1 to 5used as a measure of a patient’s general condition.

– Ultra-low anterior resection implies the partial resectionof the rectum with an anastomosis constructed belowthe peritoneal reflection and within 6 cm from the analverge. All such procedures were performed with theintent of completing a total mesorectal excision, and allwere covered with a diverting loop ileostomy.

– TEM means transanal endoscopic microsurgery.

– Adjuvant chemotherapy implies that the patient under-goes a course of therapy after surgery and that theycomplete at least four of six cycles.

– Adjuvant radiotherapy implies that the patient under-goes a course of therapy before or after surgery andwhere they complete the full course.

All patients in the study group (1996–2006) were regularlyreviewed in a surgical clinic. Patients with colon cancers wereseen at least every 6 months for the first 2 years and once ayear thereafter. Patients with rectal cancers were seen every3 months for 2 years and then every 6 months for 5 years. Inaddition, they underwent a colonoscopy at 12 months post-treatment, then every 3 years and had an annual abdominalultrasound for the first 5 years. CEA blood tests wereperformed every 6 months for 5 years. The follow-up in thecontrol group (1989–1995) was variable and dependent on theindividual surgeon.

The data was imported into the SPSS 11.0 (Macintoshversion) for statistical analysis. The mean, standard deviationand range were used as descriptive statistics. The Wilcoxon–Mann–Whitney test was used to assess the differences ininterval measurements between the two groups. The chi-squared test was used to assess for differences in nominal data.The Kaplan–Meier product limit estimate of survival wasused to calculate the recurrence functions. All recurrence rateswere calculated to 5 years with 95% confidence intervals. Thefollow-up time was from the date of histologic diagnosis ofcolorectal cancer until either recurrence developed or theyremained alive as of January 27, 2006. If a patient died, thenthe follow-up time terminated at that date. The log rank testwas used to compare recurrence curves.

The Cox proportional hazards model was used to define therelationship between a number of covariates and recurrence ofcancer. All the variables were categorical, apart from age(years), which was analyzed as a continuous variable. Allvariables were entered into the analysis using the forced-entrymethod. Significance was defined as the probability of a type1 error of less than 5%. The results are presented as therelative risk (RR) and 95% confidence intervals (95% CI).The likelihood statistic was used as an overall test of the Coxregression model.

The prospective database was approved by the FremantleHospital executive and the Health Department of WesternAustralia as part of a quality of care initiative.

Results

There were 475 patients in the control group and 710patients in the study group. The demographics for thesepatients are detailed in Table 1. The age and sex distribution

1224 Int J Colorectal Dis (2007) 22:1223–1231

were similar, as was the mean follow-up time. There was asignificantly higher proportion of rectal cancers (53% vs34%, p<0.001) and stage I colorectal cancers (23% vs12%, p<0.001) in the study group. These differences instaging were noted for both colon and rectal cancers(Table 2). Although it should be noted that 19% of the stageI rectal cancers in the study group received pre-operativeradiotherapy (45 Gy) that may have resulted in down-staging. The surgical procedures performed on patients withrectal cancer are detailed in Table 2. Patients in the studygroup were more likely to have a local excision (13% vs6%, p=0.025) and less likely to have an abdominoperinealresection (12% vs 33%, p<0.001). A similar proportion ofpatients in both groups underwent curative resections(76.6% controls vs 75.9% study) (Table 3). Patients in thestudy group were also more likely to receive adjuvant

chemotherapy (36% vs 4%, p<0.001) and radiotherapy(35% vs 9% of rectal cancers, p<0.001). In the studygroup, 27 (3.9%) patients underwent no procedure and 27(3.9%) patients only had a de-functioning stoma with noattempt at resecting the cancer. This data was not availablefor the control group.

The 5 year cancer recurrence rate was 53.9% in thecontrol group and 37.6% in the study group (Table 4). Forpatients having curative surgery, the recurrence rate wassignificantly higher in the control group for both coloncancers (33.6% vs 16%, p<0.001) and rectal cancers(50.1% vs 17.8%, p<0.001), (Table 4, Figs. 1 and 2). Themean time to recurrence tended to be longer in the studygroup (2.1 vs 1.6 years, n.s.). The longest observed time todevelop recurrence was 7.7 years. A comparison betweenthe two groups showed a significant trend towards reducedrecurrence rates in the study group for stages I, II and IIIdisease (Table 4).

The anatomical distribution for recurrence differed be-tween the two groups (Table 3). In the control group, thecommon sites for metastatic disease were the liver (71%),followed by the peritoneal cavity (24%). In the study group,the common sites for metastatic disease were the liver(38%), followed by the lungs (32%) and the peritonealcavity (22%). In the study group, there was a trend for lungmetastases to develop in patients with cancers in the lowerthird of the rectum (82% of lung metastases developed inthis group). In this group of patients, the cancers meanheight above the anal verge was only 6.5+3.9 cm vs amean of 8.9+4.1 cm for all rectal cancers (n.s.).

A Cox hazard regression was performed on patients in thestudy group with colorectal cancer who had undergonecurative surgery to determine independent predictors of

Table 3 Cancer recurrence in patients undergoing curative surgery

Control group(1989–1995)

Study group(1996–2006)

Curative resection 364 (76.6%) 539 (75.9%)Adjuvant chemotherapy 13 (4%) 194 (36%)Adjuvant radiotherapy(rectal cancers only)

12/127 (9%) 97/278 (35%)

Time to recurrence (years) 1.6±1.1 (0.4–5.6) 2.1±1.6 (0.2–7.7)Time to local recurrence(years)

1.6±1.0 (0.5–4.5) 1.9±1.4 (0.4–5.6)

Metastases 94 (26%) 63 (12%)Local recurrence 49 (13%) 16 (3%)Isolated local recurrence 28 (8%) 5 (1%)Location of metastasesLiver 82 (71%) 33 (38%)Lung 5 (4%) 28 (32%)Peritoneal 28 (24%) 19 (22%)Bone – 4 (5%)CNS – 2 (2%)Adrenal – 2 (2%)

Table 1 Basic data for the control group and study group

Control group(1989–1995)

Study group(1996–2006)

Number 475 710Age (years) 69±12 (32–95) 68±12 (29–94)Male:female ratio 1.1:1 1.2:1Follow-up (years) 3.3 (0.4–9.7) 3.1 (0.3–9.9)DistributionRight colon 157 (33%) 189 (27%)Left colon 154 (32%) 139 (20%)Rectum 164 (34%) 382 (53%)*

*p<0.001

Table 2 Staging information and procedures performed

Control group(1989–1995)

Study group(1996–2006)

Colon cancer stagingI 27 (9%) 63 (19%)*II 114 (37%) 106 (32%)III 96 (31%) 88 (27%)IV 71 (23%) 67 (20%)Unknown 3 (1%) 4 (1%)Rectal cancer stagingI 30 (18%) 134 (35%)**II 46 (28%) 67 (18%)III 49 (30%) 76 (20%)IV 36 (22%) 98 (26%)Unknown 3 (1%) 7 (2%)Rectal surgeryAnterior resection 77 (47%) 245 (64%)Abdominoperineal 54 (33%) 44 (12%)**Local excision 10 (6%) 50 (13%)*Hartmann’s procedure 8 (5%) 24 (6%)Other 15 (9%) 19 (5%)

*p<0.05, chi-squared test.**p<0.001, chi-squared test.

Int J Colorectal Dis (2007) 22:1223–1231 1225

cancer recurrence (n=539). The covariates evaluated in-cluded age, sex, ASA score, rectal location, stage, positiveresection margin, radiotherapy, chemotherapy, emergencysurgery, anastomotic leak and abscess formation. Indepen-dent predictors for recurrence included stage (e.g. stage IIIRR 6.366, 95% CI 2.011–20.155, p=0.009) and the use ofchemotherapy (RR 2.286, 95% CI 1.259–4.151, p=0.007).

The 5 year local recurrence rate for patients with rectalcancer undergoing resection was 33.6% for the control groupand 8.8% for the study group (Table 4, Fig. 3). The 5 yearrate for the development of isolated local recurrence aftercurative resections was 20.7% in the control group and3.1% in the study group (Table 4). The mean time todevelop local recurrence was 1.6 years in the control group

Table 4 A comparison of the recurrence rates in patients undergoing surgery for colon and rectal cancer

Groups Control group (1989–1995) Study group (1996–2005)5 year recurrence rate (95% CI) (n) 5 year recurrence rate (95% CI) (n)

All patients 53.9% (48.4–59.4) (475) 37.6% (32.0–40.6) (710)Curative resection (R0) 42.4% (36.1–48.7) (364) 17.0% (11.7–19.5) (533)*Stage I 21.1% (6.8–35.4) (57) 6.4% (0–8.3) (165)*Stage II 27.9% (18.9–36.9) (160) 14.6% (7.2–19.0) (203)*Stage III 59.9% (49.9–69.9) (146) 31.5% (22.4–39.6) (169)*Rectal cancer curative resection 50.1% (39.6–60.6) (127) 17.8% (10.2–21.2) (281)*Colon cancer curative resection 33.6% (26.8–40.4) (237) 16.0% (9.6–20.6) (251)*Rectal cancer local recurrence (R0, R1, R2) 33.6% (23.6–43.6) (164) 8.8% (4.5–13.1) (389)*Colon cancer local recurrence (R0) 10.2% (5.5–14.9) (237) 2.6% (0.2–5.0) (251)*

*p<0.001 for log rank comparisons between study group and control group.

Recurrence - Colon Cancer

FOLLOWUP (years)

6543210

One

Min

us C

umul

ativ

e Su

rviv

al

1 . 0

0 .9

0 .8

0 .7

0 .6

0 .5

0 .4

0 .3

0 .2

0 .1

0 .0

GROUP

Control

Study

Control 237 186 151 111 91 68 45

Study 251 198 167 135 104 81 57

Fig. 1 Recurrence after curativeresections for colon cancerscomparing the study group(1996–2006) to the controlgroup (1989–1995)

1226 Int J Colorectal Dis (2007) 22:1223–1231

vs 1.9 years in the study group (n.s.). A similar trend wasnoted for local recurrence after surgery for colon cancer(Table 4).

An analysis was performed to look at various rates oflocal recurrence after surgery for rectal cancer in the studygroup (Table 5). The rates of local recurrence varied widelyaccording to the definition. A majority of patients whodeveloped local recurrence did so in association withsystemic recurrence. It was noted that R1 resections had ahigher recurrence rate when compared with R0 resections.With regards to surgical procedures, local excision wasassociated with the highest rate of local recurrence (27.4%,n=46), followed by abdominoperineal resection (12.5%, n=42). An R0 ultra-low anterior resection was associated withthe lowest rate for local recurrence (2.4%, n=169). Therecurrence rates per stage were stage I, 8.4%; stage II, 2.9%and stage III, 11.7%. If patients undergoing local excisionsare removed from the analysis, then the local recurrencerates per stage were stage I, 0%; stage II, 1.2% (95% CI 0–3.6) and stage III, 10.2% (95% CI 1.4–19.0).

Local recurrence was highest in patients undergoinglocal excisions. In general, this sub-group was of an olderage (mean age 75 vs 67 years for the remainder, p<0.001)

and had poorer general health (65% were ASA 3 or 4 vs29% in the remainder, p<0.001) when compared with thestudy group in general. The staging in this group was stageI, 85% (T staging, T1=63% and T2=22%); stage II, 9%;stage III, 2% and stage IV, 4%. The majority of theseprocedures were performed using the TEM equipment(91%) and 20% received post-operative radiotherapy. Ofthese patients, six developed local recurrence; in five ofthese patients, the recurrence developed in the mesorectumabove the site of the previous local excision.

A Cox hazard regression was performed to determine thosecovariates that were independent predictors for the develop-ment of local recurrence in patients in the study group who hadundergone surgery for rectal cancer. The covariates included:age, sex, anastomotic leak, pelvic abscess, radiotherapy,chemotherapy, emergency surgery, ASA score, cancer stage(I to IV), type of surgery (local excision vs ultra-low anteriorresection vs abdominoperineal resection) and resection status(R0 vs R1 vs R2). Those factors that were significantlyassociated with local recurrence were local excision (RR46.77, 95% CI 5.60–380.93, p=0.005), cancer stage (stageIII RR 13.58, 95% CI 1.72–107.49, p=0.026) and resectionstatus (R1 RR 66.00, 95% CI 4.87–894.61, p=0.002).

Recurrence - Rectal Cancer

FOLLOWUP (years)

6543210

One

Min

us C

umul

ativ

e S

urvi

val

1 . 0

0 .9

0 .8

0 .7

0 .6

0 .5

0 .4

0 .3

0 .2

0 .1

0 .0

GROUP

Control

Study

Control 127 106 72 52 39 27 22

Study 281 229 183 135 108

77 51

Fig. 2 Recurrence after curativeresections for rectal cancerscomparing the study group(1996–2006) to the controlgroup (1989–1995)

Int J Colorectal Dis (2007) 22:1223–1231 1227

Of the 71 patients who developed recurrence in the studygroup, 30 (42%) proceeded to have further surgery. Theseprocedures included hepatectomy (n=7), laparotomy+ inter-nal bypass (n=8), lobectomy (n=4), abdominoperinealresection (n=3), right hemicolectomy (n=2), pelvic exen-teration (n=1) and other procedures (n=4). In those patientswho underwent an attempted curative re-operation, the5 year overall survival rate was 24.6% (95% CI 9.9% to39.3%, (n=18).

Discussion

The results of this study have shown a significant trendover the last 16 years towards reduced recurrence inpatients treated for colorectal cancer. These results wereconsistent irrespective of the stage of the cancer and itslocation. This study has also highlighted a change in thepattern of recurrences with later presentations and anincrease in the diagnosis of metastatic disease outside ofthe liver, especially in the lung. It is difficult to ascribethese improvements in outcomes to one single factor. Morelikely, they have resulted from a number of management

changes. These include the widespread application ofadjuvant therapies. Both chemotherapy and radiotherapyare now accepted standards in the management of colonand rectal cancers [1]. Yet the improvements in survival werealso seen in patients with stage I disease. Such improvementsare more likely related to surgical technique and perioperativecare [2]. There are now a number of reviews highlighting theimportance of the individual surgeon to patient’s outcome inthis area [1, 3]. The potential benefits of improved surgicaltechnique may be reflected in a more accurate anatomicaldissection, reduced requirements for blood transfusions andminimized septic complications. Such factors are recognisedas influencing cancer survival [4].

The improvements in recurrence may also reflect betterquality staging and histological review [5]. More accuratestaging through the detection of metastases or nodal diseasecould create an apparent improvement in results. The currentavailability of a variety of imaging modalities (i.e. CT, MRIor PET scanners) has improved the perioperative staging ofpatients and has also improved the detection of recurrentdisease at an earlier stage [1]. Within this state, there hasbeen a major improvement in histopathological reporting ofspecimens. Pathologists now have a standardized reporting

Local Recurrence - Rectal Cancer

FOLLOWUP (years)

6543210

One

Min

us C

umul

ativ

e Su

rviv

al

1 . 0

0 .9

0 .8

0 .7

0 .6

0 .5

0 .4

0 .3

0 .2

0 .1

0 .0

GROUP

Control

Study

Control 127 106 72 52 39 27 22

Study 281 229 183 135 108 77 51

Fig. 3 Local recurrence aftercurative resections (R0) for rec-tal cancers comparing the studygroup (1996–2006) to the con-trol group (1989–1995)

1228 Int J Colorectal Dis (2007) 22:1223–1231

procedure. Improved staging will tend to select out poorprognostic patients, leading to apparent stage migration. Asthe multivariate analysis confirmed, tumour stage is still themost important determinant for recurrence.

It was of interest to note the trend towards laterdevelopment of recurrence and the increase in diagnosisof lung metastases. This could be explained by severalfactors. Firstly, if patients are being diagnosed at an earlierstage, then their recurrence will be apparently delayed inwhat is termed as “lead time bias”. Secondly, the improve-ments in adjuvant and palliative therapies may have animpact on delaying the onset of recurrence. The increase inthe diagnosis of lung metastases was an interesting findingin this study and these metastases were usually associatedwith cancers in the lower third of the rectum. This mayrepresent a vascular metastatic spread that is directlyentering the systemic circulation and bypassing the moreusual portal circulation. Alternatively, adjuvant therapy maybe more effective at treating hepatic metastases andtherefore allow for the development of both lung metastasesand metastases at other sites. Nonetheless, the increaseddiagnosis of lung metastases may simply reflect increaseduse of imaging such as CT scans.

Local recurrence after treatment for rectal cancer wasshown to improve significantly over the study period. Yetone of the difficulties in reporting local recurrence is how itis defined [6, 7]. Dent et al. [6] highlighted the numerousinadequacies in documentation and methodology in thepublished literature on local recurrence. It is not surprisingthat rates of between 3% and 30% are reported. The authorsrecommended that local recurrence should be defined as“tumour detection in the pelvis regardless of whether newmetastases are found elsewhere”. They concluded that theoverall Kaplan–Meier 5-year rate should be calculatedalong with 95% confidence intervals. Other factors caninfluence the rate of local recurrence, including how wellpatients are followed-up.

There appears to be a consensus that local recurrencerates should be less than 10% after surgery for rectal cancer[1, 7–14]. Yet the results of this study show that localrecurrence rates varied widely according to how the resultswere analyzed. An initial calculation was performed thatincluded all patients who underwent surgery for rectalcancer (palliative and curative, R0, R1 and R2 resections).This analysis allowed for an unambiguous overview oflocal recurrence rates. Then a number of subsets wereanalysed that included curative resections (R0), resectionswith positive margins (R1) and various types of surgicalprocedures. Presenting information like this can allow formeaningful comparisons between papers. Clearly, undergo-ing a curative ultra-low anterior resection was associatedwith the lowest local recurrence rate of 2.4%, yetpublishing a result like this in isolation is misleading.

The majority of patients identified as having localrecurrence in this paper also had systemic metastases. Thisis in contrast to other studies. In the Dutch radiotherapy trialon patients undergoing surgery for resectable rectal cancers[10], more than half of the local recurrences were isolateddisease (52%). Heald and colleagues [8] have noted a fairlyeven distribution for local recurrence between isolatedrecurrence and recurrence plus metastases (1.4% vs 2.2%compared with 3.1% vs 5.7% in this study). The findings ofthis study may represent better quality staging in the patientspresenting with local recurrence because these patientswould routinely undergo a thoraco-abdominal CT scan andPET scans (from 2003) to stage their disease. Suchexaminations are more likely to detect metastatic disease [1].

Bill Heald has done more to focus attention on localrecurrence rates after surgery for rectal cancer than anyother surgeon [8, 10, 11]. In particular, he has championedthe efficacy of good quality surgery over any adjuvanttherapies [11]. It is instructive to try to compare some ofthese results with those observed in this study. Heald andcolleagues reported local recurrence rates in patientsundergoing anterior resection of 2% for Dukes A, 4% forDukes B and 7.5% for Dukes C [11]. This is without the

Table 5 Local recurrence in patients in the study group (1996–2006)undergoing surgery for rectal cancer

Rectal cancer 5 year localrecurrence rate(95% CI) (n)

5 year isolated localrecurrence rate(95% CI) (n)

ResectionR0 7.1% (3.4–10.8)

(305)2.4% (0–5.1) (305)

R1 38.1% (2.4–73.8)(37)

StageI 8.4% (1.7–10.1)

(101)II 2.9% (0–7.0)(103)III 11.7% (2.5–20.9)

(76)Surgical proceduresLocal excision 27.4% (2.7–47.2)

(46)Ultra-low anterior resectionR0 2.4% (0–5.3)

(169)0% (169)

R0 and R1 5.0% (0.9–9.1)(177)

1.4% (0–4.1) (177)

Abdominoperineal resectionR0 and R1 12.5% (0–27.2)

(42)Curative majorresection (R0)

2.7% (0.4–5.0)(238)

–

Int J Colorectal Dis (2007) 22:1223–1231 1229

use of any adjuvant therapies. Systemic recurrence devel-oped in 37% of Dukes C patients. Bokey et al. [15] reportedon local recurrence rates for rectal cancer after curativemajor resections and no adjuvant therapy. The rates were2.5% for Dukes A, 6.5% for Dukes B and 22.6% for DukesC. In this study, if we consider patients undergoing curativeresections (excluding local excisions), the recurrence rateswere 0% for stage I, 2.4% for stage II and 10.2% for stageIII. The systemic recurrence rate for stage III rectal cancerswas 34.5% (95% CI 20.8–48.2) and is only marginallybetter than the rate published by Heald of 37%. This isdespite the use of adjuvant therapies.

The presence of a positive surgical margin (R1) was animportant predictor for recurrence in patients with rectalcancer. The rate of R1 resections relate to two importantfactors, the size of the primary cancer and the quality ofthe resection. Birbeck et al. [16] identified positive cir-cumferential margins as a strong predictor for recurrenceand survival. In this study, 11% of patients in the studygroup who had had a macroscopic complete excision oftheir tumour were found to have positive margins. In acomparable study that used the same definition for apositive lateral margin, Birbeck et al. [16] found an overallrate of 28.2% for R1 resections. This rate varied betweensurgeons from 15% to 30%. Chapuis et al. [17], using thesomewhat narrower definition of “transected tumour”,found positive lateral margins in 8% of rectal cancerpatients.

Patients undergoing local excisions for rectal cancer had apoor outcome. The 5 year recurrence rate of 27% was highwhen compared with rates of between 12% and 27%published in the literature [1]. Patients were considered forlocal excisions if their cancers were T1 and N0 (on endoanalultrasound), well or moderately differentiated on biopsy andless than 2.5 cm in size. In general, patients were counselledto have major surgery if otherwise fit and well. However,patients chose to undergo local excision for various reasons.These included being unfit for major surgery or that theywanted to avoid major surgery or a permanent stoma. Asmall group of patients chose local excision as a palliativeprocedure. Pre-operative staging remains a problem in thisgroup and all patients who developed local recurrence did soin lymph nodes more proximal to the site of the originaltumour, probably indicating occult nodal metastases at thetime of diagnosis.

Similar poor results were also noted in patients undergoingabdominoperineal excisions. This has also been a point ofrecent discussion and has lead to a re-evaluation of thetechnique for such resections [18]. This study identified a highproportion of R1 resections (16/42, 38%) in the abdominoper-ineal specimens. This is despite 72% (31/42) of these patientsreceiving pre-operative radiotherapy for locally advanceddisease. This compares with 46% (82/177) of patients

undergoing ultra-low anterior resections who required radio-therapy, of which only 4.5% (8/177) were R1 resections. Inpatients undergoing abdominoperineal resections, almost all ofthe positive margins (14/16) were extending out along thepelvic floor to the lateral pelvic sidewall. This leaves littleanatomical room for wider excisions. Possibly, the anatomicalpathways by which these distal cancers are spreading aredifferent to more proximal tumours, leading to difficulty inattaining an R0 resection status.

One aspect of following-up patients after treatment forcolorectal cancer is to detect recurrent disease [1]. Yetdespite the enormous effort that is involved in the regularreview of high volumes of patients, the success indiagnosing recurrence was disappointing. Out of 71 patientsdetected with recurrence, only 18 proceeded to have acurative resection and their 5 year survival rate was 25%. Inparticular, of the 4 patients undergoing lobectomies, allhave developed further recurrent disease within 18 months.This is despite PET scanning showing isolated diseasebefore attempted curative resection. These results aresimilar to those reported from large series on outcomesafter resection of colorectal pulmonary metastases [19].Relapse rates of around 70% are noted. Despite moreencouraging results for the resection of liver metastases, itwas of interest to note in this study how few patients areultimately found to be suitable for such resections [20].

In conclusion, this study observed significant improve-ments in rates of recurrence after treatment for cancers of thecolon and rectum. It also noted a delay in the time torecurrence and an increase in the diagnosis of metastaticdisease outside of the liver. The reporting of local recurrenceremains an important issue. To allow for meaningfulcomparisons of local recurrence rates between series, wepropose a number of simple guidelines. These include a cleardefinition of the rectum and local recurrence, the use of theKaplan–Meier technique to calculate the 5 year rates plus 95%confidence intervals and finally, to calculate an overall rate forall rectal cancers undergoing resection before consideringsubset analysis that may include curative resections or majorresections only.

References

1. Australian Cancer Network Colorectal Cancer Guidelines Revi-sion Committee (2005) Clinical practice guidelines for theprevention, early detection and management of colorectal cancer(CRC), 2nd edn. The Cancer Council Australia and AustralianCancer Network, Sydney http://www.cancer.org.au/content.cfm?randid=408243 (Accessed Jun 2006)

2. Platell C, Lim D, Tajudeen N, Tan J, Wong K (2003) Doessurgical sub-specialization influence survival in patients withcolorectal cancer? World J Gastroenterol 9(5):961–964

3. Bokey EL, Chapuis PH, Dent OF, Mander BJ, Bissett IP,Newland RC (2003) Surgical technique and survival in patients

1230 Int J Colorectal Dis (2007) 22:1223–1231

having a curative resection for colon cancer. Dis Colon Rectum46(7):860–866

4. Platell C (1998) The influence of cardiopulmonary bypass oncolorectal cancer. Dis Colon Rectum 41:1371–1375

5. Quirke P, Durdey P, Dixon MF, Williams NS (1986) Localrecurrence of rectal adenocarcinoma due to inadequate surgicalresection. Lancet 2:996–999

6. Dent OF, Chapuis PH, Bokey EL, Newland RC (2001) Method-ology and reporting in studies of local recurrence after curativeexcision of the rectum for cancer. Br J Surg 88:1476–1480

7. Goudet P, Roy P, Arveux I, Cougard P, Faivre J (1997) Population-based study of the treatment and prognosis of carcinoma of therectum. Br J Surg 84(11):1546–1550

8. Moore E, Heald RJ, Cecil TD, Sharpe Gd, Sexton R, Moran BJ(2005) Almost all five year disease free survivors are cured followingrectal cancer surgery, but longer term follow-up detects some localand systemic recurrences. Colorectal Dis 7:403–405

9. Caricato M, Borzomati D, Ausania F, Valeri S, Rosignoli A,Coppola R (2006) Prognostic factors after surgery for locallyrecurrent rectal cancer: an overview. Eur J Surg Oncol 32:126–132

10. Hermanek P, Heald RJ (2004) Pre-operative radiotherapy forrectal carcinoma? Has the case really been made for short coursepre-operative radiotherapy if surgical standards for rectal carcino-ma are optimal? Colorectal Dis 6:10–14

11. Cecil TD, Sexton R, Moran BJ, Heald RJ (2004) Total mesorectalexcision results in low local recurrence rates in lymph node-positive rectal cancer. Dis Colon Rectum 97(7):1145–1149

12. Das P, Skibber JM, Rodriguez-Bigas MA, Feig BW, Chang GJ,Hoff PM et al (2006) Clinical and pathological predictors oflocoregional recurrence, distant metastasis, and overall survival in

patients treated with chemoradiation and mesorectal excision forrectal cancer. Am J Clin Oncol 29(3):219–224

13. Marijnen CA, Nagtegaal ID, Kapiteijn E, Kranenbarg EK,Noordijk EM, van Krieken JH, van de Velde CJ, Leer JW,Cooperative Investigators of the Dutch Colerectal Cancer Group(2003) Radiotherapy does not compensate for positive resectionmargins in rectal cancer patients: report of a multicenterrandomized trial. Int J Radiat Oncol Biol Phys 57(4):1311–1320

14. Rothenberger DA (2004) Invited commentary. Dis Colon Rectum47(7):1149–1150

15. Bokey EL, Ojerskog B, Chapuis PH, Dent OF, Newland RC,Sinclair G (1999) Local recurrence after curative excision of therectum for cancer without adjuvant therapy: role of totalanatomical dissection. Br J Surg 86(9):1164–1170

16. Birbeck KF, Macklin CP, Tiffin NJ, Parsons W, Dixon MF,Mapstone NP, Abbott CR, Scott N, Finan P, Johnston D, Quirke P(2002) Rates of circumferential resection margin involvementvary between surgeons and predict outcomes in rectal cancersurgery. Ann Surg 234(4):449–457

17. Chapuis PH, Lin BP, Chan C, Dent OF, Bokey EL (2006) Riskfactors for tumour present in a circumferential line of resectionafter excision of rectal cancer. Br J Surg 93:860–865

18. Daniels IR, Strassburg J, Moran BJ (2006) The need for futuresurgical low rectal cancer studies. Colorectal Dis 8(Suppl3):25–29

19. Ashley AC, Deschamps C, Alberts SR (2006) Impact ofprognostic factors on clinical outcome after resection of colorectalpulmonary metastases. Clin Colorectal Cancer 6(1):32–37

20. Lewis AM, Martin RC (2006) The treatment of hepatic metasasesin colorectal carcinoma. Am Surg 72(6):466–473

Int J Colorectal Dis (2007) 22:1223–1231 1231