C

Tdvtufarpat(gT

Rd

Int. J. Radiation Oncology Biol. Phys., Vol. 65, No. 2, pp. 445–451, 2006Copyright © 2006 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/06/$–see front matter

doi:10.1016/j.ijrobp.2005.12.027

LINICAL INVESTIGATION Rectum

MRI IDENTIFIED PROGNOSTIC FEATURES OF TUMORS IN DISTALSIGMOID, RECTOSIGMOID, AND UPPER RECTUM: TREATMENT WITH

RADIOTHERAPY AND CHEMOTHERAPY

SARAH BURTON, M.R.C.S.,* GINA BROWN, F.R.C.R.,† IAN DANIELS, F.R.C.S.,‡

ANDY NORMAN, B.SC.,§ IAN SWIFT, F.R.C.S.,* MUTI ABULAFI, F.R.C.S.,*ANDY WOTHERSPOON, M.R.C.PATH.,� AND DIANA TAIT, F.R.C.R.¶

*Division of Colorectal Surgery, Mayday University Hospital, Croydon, UK; †Academic Department of Radiology, andDepartments of §Medical Statistics and Computing, �Histopathology, and ¶Clinical Oncology, Royal Marsden Hospital, Sutton,

Surrey, UK; ‡Division of Colorectal Surgery, Epsom and St. Helier NHS Trust, Epsom, UK

Purpose: Neoadjuvant therapy is traditionally reserved for locally advanced mid and low rectal cancers. Intumors above this level, the need for adjuvant treatment is based on poor histopathologic features, but thisapproach has potential disadvantages. The aim of this study was to determine whether magnetic resonanceimaging (MRI) could accurately stage tumors of the distal sigmoid, rectosigmoid, and upper rectum and helpdirect preoperative treatment.Materials and Methods: A total of 75 patients with distal sigmoid, rectosigmoid, and upper rectal tumors wereassessed preoperatively by MRI. If tumor extended beyond the planned surgical resection plane, chemoradio-therapy was offered.Results: Of the 75 patients, 57 (76%) underwent primary surgery. Agreement between the MRI prognosis andhistopathologic findings was 84% (95% confidence interval [CI], 72.6–92.7%). The other 18 patients underwentneoadjuvant chemoradiotherapy for poor prognostic features with predicted surgical resection margin involve-ment. The histopathologic examination confirmed tumor downstaging in 9 of the 18 patients who underwentchemoradiotherapy. The 3-year survival rate in the good prognosis group (91%; 95% CI, 77.1–97.3%) was notsignificantly different from that of the chemoradiotherapy group (81.4%; 95% CI, 52.4–93.6%). The poorprognosis group undergoing primary surgery had significantly worse survival (62.2%; 95% CI, 30.3–82.8%,p < 0.03).Conclusion: Our findings indicate that tumors of the distal sigmoid, rectosigmoid, and upper rectum can bestaged accurately using high spatial resolution MRI and that those with poor prognostic disease may benefit frompreoperative therapy. © 2006 Elsevier Inc.

MRI staging, Colorectal cancer, Neoadjuvant chemotherapy, Preoperative radiotherapy.

sp

tt(icitpietI

m

INTRODUCTION

he definitions of rectal cancer vary, but the most acceptedefinition is that of tumor arising within 15 cm of the analerge as measured by rigid sigmoidoscopy (1). Many publica-ions have subdivided the site of rectal cancer into thirds—pper (�10–15 cm), middle (5–10 cm), and low (�5 cm)—or the purposes of analysis (2, 3). Such subdivision hasllowed the identification of different prognostic groups ofectal cancer patients. Low rectal cancers appear to have aoorer prognosis, which may reflect the difficulty in achievingnegative circumferential resection margin (CRM) with such

umors (4). The Dutch Total Mesorectal Excision SurgeryTME) trial concluded that upper third rectal cancers did notain significant benefit from short-course radiotherapy (5).his has translated into the belief that tumors of the distal

Reprint requests to: Gina Brown, F.R.C.R., Department ofadiology, Downs Road, Sutton, Surrey SM2 5PT, United King-

om. Tel: (�44) 208-661-3964; Fax: (�44) 208-661-3506; E- A

445

igmoid, rectosigmoid, and upper rectum do not benefit from areoperative treatment strategy.

In the management of such tumors, it is standard practiceo offer these patients only adjuvant postoperative therapy ifhe histologic assessment identifies poor prognostic features6). However, this approach has some disadvantages, withnherent delays in beginning therapy owing to surgicalomplications and patient convalescence (6). This can resultn poor compliance and completion rates in the postopera-ive period (7). The preoperative identification of knownoor prognostic features such as circumferential marginnvolvement, peritoneal involvement, nodal disease, andxtramural vascular invasion (EMV) (8–10) could renderhese patients eligible for preoperative treatment strategies.n tumors of the distal sigmoid, rectosigmoid, and upper

ail: gina.brown@rmh.nhs.ukReceived Sept 15, 2005, and in revised form Dec 15, 2005.

ccepted for publication Dec 15, 2005.

ra(tnosvrpr

mpTaddtp

pr9loi�b

wcprBcf

M

tuGsiaiT1apaw2o

oom

cdbotwAo

C

(

URD

446 I. J. Radiation Oncology ● Biology ● Physics Volume 65, Number 2, 2006

ectum, the adjacent small bowel is the critical normal tissuend its tolerance prohibits treatment in the adjuvant setting11) of a potentially large target volume, namely the peri-oneal cavity. However, justification exists for consideringeoadjuvant radiotherapy in primarily unresectable tumorsf the distal sigmoid, rectosigmoid, and upper rectum. Highpatial resolution magnetic resonance imaging (MRI) isery accurate in assessing such poor prognostic factors inectal cancer patients (12). We investigated extending itsotential use in distal sigmoid, rectosigmoid, and upperectal tumors.

METHODS AND MATERIALS

Data were collected retrospectively on all upper rectal, rectosig-oid, and distal sigmoid cancer patients referred for MRI after

resenting to our unit between January 1999 and December 2002.he local ethics review board of the participating institutionspproved the study. None of the patients included in this study hadistant metastatic disease at presentation. The tumor site wasefined prospectively according to the distance from the anal vergeo the tumor’s lowest limit as measured by MRI in the sagittallane (Table 1).Our definitions were based on our interpretation of the current

ublished data. A study to determine the level of the peritonealeflection found the mean anterior height from the anal verge to becm in women and 9.7 cm in men (13). We therefore defined the

ower limit of upper rectal cancers as 10 cm (i.e., above the levelf the peritoneal reflection). In 2000, U.S. guidelines were issuedndicating that a cancer should only be classified as rectal if it was

12 cm from the anal verge on rigid proctoscopy (14). This wasased on evidence that the local recurrence rate in tumors �12 cm

Table 1. Definitions of tumor site

Tumor site Distance from anal verge (cm)

pper rectum 10 to �12ectosigmoid �12 to �15istal sigmoid �15

Table 2. Treatment allocation of rectal cancextended to tumors of distal sigm

Prognostic group MRI f

Good T1-T2, T3 �5 mm, N0-CRM negative

Bad T3 �5 mm, T4, N2, EMCRM negative

T4 invading adjacent orCRM positive

Abbreviations: EMV � extramural invasionOur local staging stratification was based on

cancer patients with T3 tumors �5 mm depth oT3 �5 mm spread (24).

Peritoneal involvement is a known poor pro26).

Increasing percentage of nodal disease redu

EMV has been shown to be an independent poor

as 9.6% compared with 30.1% and 30.7% for mid and low rectalancers, respectively, and was more consistent with a colonicattern of local failure (15). Hence, we used �12 cm to define ourectosigmoid group. The Association of Coloproctologists (Greatritain and Ireland) have defined rectal cancer as any tumor �15m. Therefore, we used �15 cm as our definition of the lower limitor distal sigmoid cancers.

RI techniqueAll patients underwent preoperative MRI of the pelvis according

o the protocol described below. Only T2-weighted sequences weresed. MRI was performed on a 1.5-T scanner (Siemens, Erlangen,ermany) with a four-element pelvic phased array wrap-around

urface coil at any of the involved hospitals. All patients weremaged in the supine position. Neither intravenous antiperistalticgents nor contrast agents were administered. A coronal localizingmage was obtained to select the axial and sagittal images with a

2-weighted fast spin echo (FSE) sequence (TR �3000 ms, TE28 ms, echo-train length [ETL] 16, 5 mm thickness, four signalverages, scan duration 3–5 min). The sagittal images were used tolan 3-mm oblique high spatial resolution axial images. The im-ges were acquired in a plane orthogonal to the tumor and rectalall using a T2-weighted FSE sequence (TR �3000, TE 128 ms,56 � 256 matrix, ETL 16, field of view 16–18 cm). Dependingn the tumor length, the scan duration was between 6 and 12 min.All MRI scans were prospectively reviewed by a single radiol-

gist for T stage, depth of extramural invasion, N stage, presencef extramural venous invasion, and CRM or surgical resectionargin status.A preoperative treatment strategy of 12 weeks of neoadjuvant

hemoradiotherapy was only offered to patients with MRI-pre-icted positive CRMs or surgical resection margins (i.e., tumoreyond the planned resection plane—bladder, pelvic side walls, orther adjacent organ), and these patients were enrolled in phase IIrials (16). The presence of other poor prognostic features on MRIas reported but did not change the patient’s planned treatment.ll the remaining patients proceeded with the traditional approachf primary surgery (17–21) (Table 2).

hemoradiotherapy techniquePatients received a protracted venous infusion of 5-fluorouracil

300 mg/m2 body surface area/d for 12 weeks) with intravenous

ording to MRI features (these criteria wereectosigmoid, and upper rectum)

s Treatment strategy

EMV, potentially Primary surgery

sent, potentially Primary surgery

d/or potentially Preoperativechemoradiotherapy

� circumferential resection margin.nce of 5-year survival rates of 85% for rectalmural spread vs. 54% survival for tumors with

factor for both rectal and colon cancers (25,

erall survival in colorectal cancer (27).

ers accoid, r

eature

N1, no

V pre

gans an

; CRMevide

f extra

gnostic

ces ov

prognostic factor in colorectal cancer (8, 28).

b6dr5psClew

H

tCuscteNi

ae

pw

S

ptbMd

m(arddwbctwpm(

(wwt

M

ftIm(6agm

URDT

TTT

w

NN

GPT

7

447Preoperative treatment of distal sigmoid and rectosigmoid cancers ● S. BURTON et al.

olus injection of mitomycin C (7 mg/m2 body surface area everyweeks). Starting on Week 13, the 5-fluorouracil dose was re-

uced to 200 mg/m2 body surface area/d, and concomitant pelvicadiotherapy (45 Gy in 25 fractions) was begun, followed by a.4–9-Gy boost to the tumor bed (16). Patients were treated in therone position with a full bladder. Both maneuvers have beenhown to decrease the volume of small bowel. Patients underwentT treatment planning, using diagnostic MRI scans to aid in

ocalization of the target volume. The small bowel was maximallyxcluded using customized blocking techniques (22–24). Surgeryas planned for 6 weeks after chemoradiotherapy.

istopathologic techniqueAll resections were performed with curative intent, and his-

opathologic examinations were performed according to the Royalollege of Pathologists guidelines (25). The CRM was defined inpper rectal cancers as positive if tumor was within 1 mm of theurgical resection margin. In cases of an apparent histopathologi-ally complete response after chemoradiotherapy, multiple blockshroughout the whole rectum were obtained, and all cases consid-red for postoperative therapy were reviewed centrally. The T stage,

stage, presence of EMV, CRM status, and depth of extramuralnvasion were documented.

Data were collected on prognostic features identified on MRInd correlated with the prognostic features identified on histologicxamination.

Cancer-specific survival data were collected for the differentrognostic groups. The date of death or last follow-up appointmentere used as the endpoints for the survival curves.

tatistical analysisThe kappa test was used to compare each MRI-predicted poor

rognostic feature with the histopathologic findings to determinehe accuracy. We calculated the 95% confidence intervals using theinomial distribution. Survival was analyzed using the Kaplan-eier methods, and the log–rank test was used to detect any

ifferences between the survival curves.

Table 3. Treatment allocation and tumor site

Tumor sitePatients

(n)Surgeryalone (n)

Chemoradiotherapy(n)

pper third rectal 25 (33) 16 (64) 9 (36)ectosigmoid 24 (32) 19 (79) 5 (21)istal sigmoid 26 (35) 22 (85) 4 (15)otal 75 (100) 57 (76) 18 (24)

Data in parentheses are percentages.

Table 4. Correlation of MRI predicted T stage with finalhistologic findings in patients undergoing primary surgery

MRI predicted T stage

Final histologic T stage

T1-T2 T3 T4

1-T2 14 8 03 5 18 64 0 2 4

Agreement 63%; weighted kappa with quadratic (Fleiss–Cohen)

eights 0.59. t

RESULTS

A total of 75 patients were included in the study, with aedian age of 65 years (range, 37–86 years); 41 patients

54%) were men. The tumors were evenly distributedmong the three sites, with 25 (33%) defined as upper thirdectal tumors, 24 (32%) as rectosigmoid, and 26 (35%) asistal sigmoid tumors (Table 3). On the basis of the MRI-etermined criteria (Table 1), 18 of the 75 patients under-ent neoadjuvant chemoradiotherapy to downstage the tumorefore surgical resection. The remaining 57 patients pro-eeded to primary surgical resection. The proportion ofumors undergoing each therapy differed according to site,ith 16 (64%) of 25 upper third rectal tumors undergoingrimary surgery compared with 19 (79%) of 24 rectosig-oid tumors and 23 (85%) of 26 distal sigmoid tumors

Table 3).The median lymph node count after resection was 12

range, 6–38). Of 18 patients in the primary surgery groupith poor prognostic disease on histologic examination, 17ere offered postoperative chemotherapy, and only 2 pa-

ients refused further treatment.

RI agreementThe agreement between MRI and the histologic findings

or each prognostic variable was determined by correlatinghe MRI prediction against the final histologic outcome.n the primary surgery patients (n � 57), the T stage agree-ent was 63%, with a weighted kappa agreement of 0.596

Table 4). Nodal staging by MRI criteria was accurate in5% of cases (37 of 57; Table 5). However, in the surgery-lone patients, if the prognostic features were grouped intoood or bad according to the criteria in Table 2, the agree-ent of the MRI prediction of prognosis was 84.5% (range,

Table 5. Correlation of MRI-predicted N stage with finalhistologic findings in patients undergoing primary surgery

MRI

Histopathologic finding

Node negative Node positive

ode negative 23 9ode positive 11 14

Agreement 65%; kappa 0.28.

Table 6. Correlation of MRI-predicted prognosis with finalhistologic prognosis in patients undergoing primary surgery

MRI-predicted prognosis

Final histologic prognosis

Good Poor Totals

ood 35 5 40oor 4 13 17otals 39 18 57

Agreement for MRI prediction of prognosis 84% (95% CI,2.6–92.7%, kappa 0.63); sensitivity 90%, specificity 72%, posi-

ive predictive value 88%, and negative predictive value 76%.

7T

mcOwtth9s

E

tiptCwa

5TfhrwMtr

igadsfr

S

csn(u

gssvc

ipihaacp

AP

8

Ffme

SSSST

448 I. J. Radiation Oncology ● Biology ● Physics Volume 65, Number 2, 2006

2.6–92.7%), with a kappa agreement of 0.64 (Table 6).he sensitivity was 90%, with a specificity of 74%.The MRI prediction of the presence or absence of extra-ural vascular invasion was accurate in 41 (72%) of 57

ases, with a negative predictive value of 84% (Table 7).f the 57 patients, 3 had MRI-predicted involved CRMs,ith only 1 case confirmed on histologic examination in

he primary surgery group. The original MRI reports forhese patients suggested that the margins were not at risk;ence, their allocation to primary surgery. MRI showed a6% (55 of 57) accuracy for correctly identifying the CRMtatus.

ffect of chemoradiotherapyEighteen patients underwent neoadjuvant chemoradio-

herapy. All 18 patients had potentially involved CRMs ornvasion of adjacent organs as determined by MRI androceeded to chemoradiotherapy to downstage or downsizeheir tumor (Fig. 1). Of the 18 cases, 15 were predicted to beRM positive on MRI. Of these, the tumor of 13 patientsas successfully downstaged by chemoradiotherapy to neg-

tive margins as confirmed by histologic examination, with

Table 7. Correlation of MRI-predicted EMV with finalhistologic findings in patients undergoing primary surgery

MRI-predicted EMV

Final histologic EMV

Absent Present

bsent 31 6resent 10 10

Agreement 72% (41/57), kappa 0.354, positive predictive value4%.

Pts undergoing chemoradiotherapy

18

MRI predicted CRM+ve 15

Histological CRM negative

3

Histological CRM positive

2

Goodprognosis

2

Badprognosis

6

Bad prognosis

2

MRI predicted organ invasion

3

Histological CRM negative

13

Goodprognosis

7

Badprognosis

1

ig. 1. Actual treatment allocation according to MRI prognosticactors and outcomes on final histologic examination. MRI �agnetic resonance imaging; Pts � patients; CRM � circumfer-

ntial resection margin; �ve � positive.

patients having a complete pathologic response (Table 8).he remaining 8 patients demonstrated fibrosis extending

rom the tumor edge to the circumferential margins onistologic examination, consistent with tumor shrinkage inesponse to the neoadjuvant treatment (Fig. 2). All 3 casesith posttreatment persistent adjacent organ invasion onRI were confirmed as such on histopathologic examina-

ion. The 2 patients with positive margins were in theectosigmoid group.

Of the 18 patients with a predicted poor prognosis onnitial MRI who underwent chemoradiotherapy before sur-ery, the disease of 9 (50%) was successfully converted togood prognosis (Figs. 1 and 3). The effect of chemora-

iotherapy on the International Union Against Cancer stagehowed that the number of Stage III tumors was reducedrom 16 (89%) to 7 (39%), with a complete pathologicesponse in 28% (Table 8).

urvivalThe 3-year cancer-specific survival rate was 92% (95%

onfidence interval [CI], 77.1–97.3%) for the good progno-is tumors, 62.2% (95% CI, 30.3–82.8%) for the bad prog-osis tumors that underwent primary surgery, and 81.4%95% CI, 52.4–93.6%) for the poor prognosis tumors thatnderwent neoadjuvant therapy before surgery (Fig. 4).The difference in the 3-year survival rate between the

ood prognosis group and bad prognosis group undergoingurgery alone was statistically significant (p � 0.03). Noignificant difference could be demonstrated between sur-ival in the good prognosis group and the neoadjuvanthemoradiotherapy group.

DISCUSSION

In rectal cancer, neoadjuvant chemotherapy is increas-ngly being combined with radiotherapy for the dual pur-ose of enhancing radiosensitivity of the tumor and reduc-ng local recurrence (6, 16). Such neoadjuvant therapiesave yet to be explored in the region of the rectosigmoidnd sigmoid colon. However, local recurrence rates as greats 30% in distal sigmoid, rectosigmoid, and upper rectalancers have been reported (26) and therefore support aroposal for neoadjuvant therapies.

Table 8. Downstaging effect of chemoradiotherapy

UICC stage

PretreatmentMRI-predicted

stage (n)

Posttreatmentpathologicstage (n)

tage 0, T0N0 0 5tage I, T1-T2N0 0 0tage II, T3-T4N0 2 6tage III, any T, N1-N2 16 7otal 18 18

Abbreviation: UICC � International Union Against Cancer.

The avoidance of radiotherapy in tumors of the distal

sridaiatcgccae

r

aTttsrhdms

orPst

449Preoperative treatment of distal sigmoid and rectosigmoid cancers ● S. BURTON et al.

igmoid, rectosigmoid, and upper rectum is based on soundadiotherapeutic principles concerning normal tissue toxic-ty issues and the aim of encompassing an anatomicallyefined compartment, relating to patterns of tumor spread,t the target volume. These two principles are criticallynterconnected, because the size of the target volume willffect the volume of normal tissue irradiated. However, thearget volume can be confined to the area of surgical diffi-ulty and does not necessarily need to encompass the re-ional lymph nodes. Sufficient evidence has shown an in-reased incidence of bowel toxicity in the postoperativeompared with the preoperative setting, making this a moreppealing proposition than irradiating an ill-defined postop-rative area at risk (11, 27).

Our study has shown that preoperative MRI can accu-ately stratify tumors of the distal sigmoid, rectosigmoid,

Fig. 2. Evidence of upper third rectal tumor downstaginchemoradiotherapy. (a) Tumor extending to mesorectalPreoperative chemoradiotherapy was given to enable tumtreatment showed extensive low signal intensity whercorresponding to fibrosis (arrow). (c) Surgical specimentumor (arrow). (Figure appears in color online.)

Fig. 3. (a) Example of locally advanced distal sigmoid cT2-weighted fast spin echo images (TR 4000, TE 101, ecbladder dome. On the basis of these magnetic resonancechemoradiotherapy. (b) Example of locally advanced dOblique coronal T -weighted fast spin echo images (TR 4

2

chemoradiotherapy showing marked tumor regression. Tumor

nd upper rectum into prognostic groups. Agreement of theand N stage of the distal sigmoid, rectosigmoid, and upper

hird rectal tumors was 63% and 65%, respectively. Al-hough the T and N stage of the more proximal rectal andigmoid tumors using MRI is less accurate than the 85–90%ate achieved with MRI of rectal cancers alone (12), weave shown that by identifying either locally extensiveisease or disease encroaching on the potential resectionargin, we can distinguish between good and poor progno-

is tumors with 84.5% accuracy.Magnetic resonance imaging prediction of the presence

r absence of EMV, a known poor prognostic factor inectal and colon cancer (9, 10), was correct in 72% of cases.athologic studies have shown that EMV predicts for poorurvival and development of systemic failure (9, 10). Thus,he accurate prediction of this feature may be of benefit in

involved margin to negative margin with neoadjuvantand thus to potential surgical resection margin (arrow).ression from potential resection margins. (b) Scans afteror had been, with regression away from margin andrmed fibrosis at circumferential margins but no viable

efore chemoradiotherapy. Consecutive oblique coronalin length 11) through extensive sigmoid tumor invadinging findings, patient was offered intensive preoperativegmoid tumor in same patient after chemoradiotherapy.E 101, echo train length 11) through sigmoid tumor after

g fromfasciaor rege tum

confi

ancer bho traimag

istal si000, T

was more readily resectable.

fr

rttjtat

rMfclMpwofiswocnptd

5r7opcpuwpdfgsa

aaibntrt

re appe

450 I. J. Radiation Oncology ● Biology ● Physics Volume 65, Number 2, 2006

uture preoperative stratification of high-risk patients withectosigmoid and distal sigmoid cancer.

Tumors of the distal sigmoid, rectosigmoid, and upperectum are more difficult to stage than mid and lower rectalumors, possibly because of the technical challenge of ob-aining true axial images through the tortuous rectosigmoidunction. Future developments in software, particularlyhree-dimensional T2-weighted high spatial resolution im-ges, may improve visualization of these areas to increasehe accuracy of T and N staging.

The CRM status is used as a surrogate marker for localecurrence and poor survival in rectal cancers (28). Using

RI, our predicted CRM positive rate was 20% (15 of 75)or tumors above the peritoneal reflection. However, afterhemoradiotherapy in selected high-risk patients, our histo-ogic CRM positive rate was reduced to 4% (3 of 75). Also,

RI was 96% accurate in predicting the margin status in therimary surgery group. One limitation of our study was thate have relied on the extent of fibrosis to indicate theriginal tumor extent. Although it could not be proved, thebrotic stroma extending to the CRMs in the treated casesupports the view that the reduction in positive CRM rateas probably an effect of chemoradiotherapy rather thanverstaging by MRI. The International Union Against Can-er TNM stage was decreased in 9 of 16 patients undergoingeoadjuvant therapy, with 5 patients having a completeathologic response. This translated into a point estimate ofhe 3-year survival rate for patients with poor prognosis

0

20

40

60

80

100

0 1 2Time since in

Pro

babi

lity

of s

urvi

val (

%)

Good

Bad (No CRT)

Bad (CRT)

Fig. 4. Cancer-specific survival curves for good prognneoadjuvant therapy. CRT � chemoradiotherapy. (Figu

isease undergoing chemoradiotherapy of 81.4% (95% CI, m

REFEREN

ACPGBI; 2001.

2.4–93.6%), not significantly different from the survivalate of patients with good prognosis disease (91%; 95% CI,7.1–97.3%). However, the difference in the point estimatef the 3-year survival between those with good and badrognosis disease undergoing primary surgery was signifi-ant (p � 0.03), even though most bad prognosis tumoratients proceeded to postoperative chemotherapy. It seemsnlikely that additional survival benefit could be achievedith neoadjuvant therapy in the good prognosis group ofatients. However, such therapy appears to have made aifference in the survival of patients with poor prognosticeatures with initially unresectable disease, and this sug-ests that a similar benefit could potentially be demon-trated in those with tumors above the peritoneal reflectionnd resectable poor prognosis disease.

CONCLUSION

Our results indicate that routine MRI-based preoperativessessment of tumors in the distal sigmoid, rectosigmoid,nd upper rectum may be an effective method for identify-ng patients with high-risk features. Whether such patientsenefit from preoperative identification and treatment usingeoadjuvant chemoradiotherapy needs to be tested prospec-ively. Although some centers already treat upper rectal andectosigmoid tumors with neoadjuvant chemoradiotherapy,he potential to extend this approach to distal sigmoid tu-

3 4 5esentation (years)

mors, bad prognosis tumors, and tumors treated withars in color online.)

ors is an exciting prospect.

CES

1. Association of Coloproctology of Great Britain and Ireland.Guidelines for the management of colorectal cancer. London:

2. Marijnen CA, Kapiteijn E, van de Velde CJ, et al. Acute sideeffects and complications after short-term preoperative radio-

itial pr

osis tu

therapy combined with total mesorectal excision in primary

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

451Preoperative treatment of distal sigmoid and rectosigmoid cancers ● S. BURTON et al.

rectal cancer: Report of a multicenter randomized trial. J ClinOncol 2002;20:817–825.

3. Nagtegaal ID, van de Velde CJ, van der Worp E, et al.Macroscopic evaluation of rectal cancer resection specimen:Clinical significance of the pathologist in quality control.J Clin Oncol 2002;20:1729–1734.

4. Bonadeo FA, Vaccaro CA, Benati ML, et al. Rectal cancer:Local recurrence after surgery without radiotherapy. Dis Co-lon Rectum 2001;44:374–379.

5. Kapiteijn E, Marijnen CA, Nagtegaal ID, et al. Preoperativeradiotherapy combined with total mesorectal excision for re-sectable rectal cancer. N Engl J Med 2001;345:638–646.

6. Chawla AK, Kachnic LA, Clark JW, et al. Combined modalitytherapy for rectal and colon cancer. Semin Oncol 2003;30:101–112.

7. Pahlman L, Glimelius B. Pre- or postoperative radiotherapy inrectal and rectosigmoid carcinoma: Report from a randomizedmulticenter trial. Ann Surg 1990;211:187–195.

8. Mulcahy HE, Toner M, Patchett SE, et al. Identifying stage Bcolorectal cancer patients at high risk of tumor recurrence anddeath. Dis Colon Rectum 1997;40:326–331.

9. Shepherd NA, Saraga EP, Love SB, et al. Prognostic factors incolonic cancer. Histopathology 1989;14:613–620.

0. Newland RC, Dent OF, Lyttle MN, et al. Pathologic determi-nants of survival associated with colorectal cancer with lymphnode metastases: A multivariate analysis of 579 patients.Cancer 1994;73:2076–2082.

1. Frykholm GJ, Glimelius B, Pahlman L. Preoperative or post-operative irradiation in adenocarcinoma of the rectum: Finaltreatment results of a randomized trial and an evaluation oflate secondary effects. Dis Colon Rectum 1993;36:564–572.

2. Brown G, Radcliffe AG, Newcombe RG, et al. Preoperativeassessment of prognostic factors in rectal cancer using high-resolution magnetic resonance imaging. Br J Surg 2003;90:355–364.

3. Narajian MM, Belzer GE, Cogbill TH, et al. Determination ofthe peritoneal reflection using intraoperative proctoscopy. DisColon Rectum 2004;47:2080–2085.

4. Nelson H, Petrelli N, Carlin A, et al. Guidelines 2000 forcolon and rectal cancer surgery. J Natl Cancer Inst 2001;93:583–596.

5. Pilipshen SJ, Heilweil M, Quan SH, et al. Patterns of pelvicrecurrence following definitive resections of rectal cancer.Cancer 1984;53:1354–1362.

6. Chau I, Allen M, Cunningham D, et al. Neoadjuvant systemic

fluorouracil and mitomycin C prior to synchronous chemora-

diation is an effective strategy in locally advanced rectalcancer. Br J Cancer 2003;88:1017–1024.

7. Merkel S, Mansmann U, Papadopoulos T, et al. The prognos-tic inhomogeneity of colorectal carcinomas stage III: A pro-posal for subdivision of stage III. Cancer 2001;92:2754–2759.

8. Shepherd NA, Baxter KJ, Love SB. Influence of local perito-neal involvement on pelvic recurrence and prognosis in rectalcancer. J Clin Pathol 1995;48:849–855.

9. Shepherd NA, Baxter KJ, Love SB. The prognostic impor-tance of peritoneal involvement in colonic cancer: A prospec-tive evaluation. Gastroenterology 1997;112:1096–1102.

0. Stocchi L, Nelson H, Sargent DJ, et al. Impact of surgical andpathologic variables in rectal cancer: A United States commu-nity and cooperative group report. J Clin Oncol 2001;19:3895–3902.

1. Conroy T, Guillemin F, Stines J, et al. [Diagnosis and stagingof colorectal cancers]. Rev Prat 1994;44:2706–2713.

2. Gallagher MJ, Brereton HD, Rostock RA, et al. A prospectivestudy of treatment techniques to minimize the volume ofpelvic small bowel with reduction of acute and late effectsassociated with pelvic irradiation. Int J Radiat Oncol BiolPhys 1986;12:1565–1573.

3. Shanahan TG, Mehta MP, Bertelrud KL, et al. Minimizationof small bowel volume within treatment fields utilizing cus-tomized “belly boards.” Int J Radiat Oncol Biol Phys 1990;19:469–476.

4. Koelbl O, Richter S, Flentje M. Influence of patient position-ing on dose–volume histogram and normal tissue complica-tion probability for small bowel and bladder in patients re-ceiving pelvic irradiation: A prospective study using a 3Dplanning system and a radiobiological model. Int J RadiatOncol Biol Phys 1999;45:1193–1198.

5. Quirke P, Williams GT. Minimum data set for colorectalcancer histopathology reports. London: Royal College of Pa-thologists; 1998.

6. Willett C, Tepper JE, Cohen A, et al. Local failure followingcurative resection of colonic adenocarcinoma. Int J RadiatOncol Biol Phys 1984;10:645–651.

7. Sauer R, Becker H, Hohenberger W, et al. Preoperative versuspostoperative chemoradiotherapy for rectal cancer. N EnglJ Med 2004;351:1731–1740.

8. Nagtegaal ID, Marijnen CA, Kranenbarg EK, et al. Circum-ferential margin involvement is still an important predictor oflocal recurrence in rectal carcinoma: Not one millimeter buttwo millimeters is the limit. Am J Surg Pathol 2002;26:350–

357.