A prospective comparison study for predicting circumferential rection margin between preoperative MRI and whole mount ctions in mid-rectal
cancer:Significance of different scan planes
Y .W.Kim a ,N.K.Kim a ,*,B.S.Min a ,H.Kim b ,J.Pyo b ,M.J.Kim c ,S.H.Cha c
a
Department of Surgery,Yoni University College of Medicine,134,Shinchon-Dong,Seodaemun-Ku,Seoul 120-752,South Korea
b
Department of Pathology,Yoni University College of Medicine,Seoul,South Korea c
Department of Diagnostic Radiology,Yoni University College of Medicine,Seoul,South Korea
Accepted 1May 2007Available online 18June 2007
Abstract
Aim :The aim of this study is to evaluate the accuracy of preoperative magnetic resonance imaging (MRI)in the prediction of circumfer-ential rection margin (CRM)and to determine whether each different MRI scan plane provides an accurate CRM asssment.
Method :Fifty-ven concutive patients with mid-rectal cancer were enrolled prospectively.The CRM measurement from each MRI plane according to tumor location was compared with CRM measurement on whole-mount ctions with the definition of threatened CRM as 2mm in distance.The difference in performance among the sagittal,axial and oblique MR images was analyzed by using receiver operating characteristic (ROC)curves (A z ).
Results :For anterior tumors (n ¼17),the A z of the sagittal,axial and oblique MR planes were 0.66,0.83and 0.79,respectively.For lateral tumors (n ¼17),the A z of the sagittal,axial and oblique MR planes were 0.53,0.66and 0.78,respectively.For posterior tumors (n ¼23),the A z of the sagittal,axial and oblique MR planes were 0.76,0.82and 0.97,respectively.
Conclusions :MRI provides an accurate prediction of preoperative CRM.There exist differences in diagnostic accuracy according to each different scan plane of MRI and tumor location within the rectum.Ó2007Elvier Ltd.All rights rerved.
Keywords:Rectal cancer;Magnetic resonance imaging;Circumferential rection margin;Whole-mount ctions;Scan plane
Introduction
Local recurrence is an important contributing factor to a patient’s prognosis following curative rection for rectal cancer.1After the introduction of total mesorectal excision (TME),which consists of complete removal of the rectum together with the mesorectum by preci disction along the mesorectal fascia,local recurrence rates have signifi-cantly decread to below 10%.2e 4TME is currently ac-cepted as the standard treatment in rectal cancer surgery,2,5however,local recurrence after TME still remains a signifi-cant problem associated with morbidity and mortality.
Even if TME is performed properly,circumferential rection margin (CRM)status is still a concern.CRM in-volvement is related to a high local recurrence and rves as
a powerful prognostic factor.6e 8Local recurrence due to CRM involvement can be further reduced by preoperative radiotherapy with or without chemotherapy than TME alo-ne.9e 13Thus it is important to identify patients who have a high risk of CRM involvement and are most likely to ben-efit from neoadjuvant therapy via accurate preoperative imaging studies.
Magnetic resonance imaging (MRI)has become a pre-ferred diagnostic tool for the preoperative asssment of CRM in patients with rectal cancer 14becau it can clearly visualize the outer margin of the tumor and the mesorectal fascia.15,16However,accurate prediction is not always pos-sible due to problems with partial voluming of MRI or the complexity of pelvic anatomy.17The rectum follows the sacrococcygeal curve and,at the anorectal junction,shows its concavity due to the puborectal sling.The anterior sur-face of the rectum is cloly related to the bladder,prostate,and minal vesicles in the male,and to the vaginal wall in
*Corresponding author.Tel.:þ82222282117;fax:þ8223138289.E-mail address:i.ac.kr (N.K.Kim).0748-7983/$-e front matter Ó2007Elvier Ltd.All rights rerved.
doi:10.1016/j.ejso.2007.05.001
Available online at
EJSO 34(2008)648e 654
the female.It is difficult to delineate the mesorectal fascia in the anterior portion of the rectum due to the relatively thin mesorectum.
To achieve a preci CRM prediction,a previous publi-cation suggested that MR images should be obtained by a standardized technique using various scan planes.18The sagittal,oblique coronal and oblique axial images as well as the standard axial image obtained perpendicular to the long axis of the rectum lead to accurate CRM prediction and better preoperative understanding of the mesorectal fascia and perirectal pelvic structures for performing TME on rectal cancer patients.
To our knowledge,there are no prospective reports which compare each CRM measurement taken on various scan planes parately with histologic CRM using whole-mount ctions.Therefore,the aim of this study is to evaluate the accuracy of preoperative MRI in predicting T stage,N stage, and CRM,and to determine whether each different MRI scan plane provides an accurate CRM asssment. Patients and methods
Patients
Between December2005and September2006,57con-
cutive patients(33males and24females;median age, 62(range,30e81)years)with biopsy proven primary rectal cancer were enrolled prospectively.Tumor location in rela-tion to the anal verge was preoperatively examined with rigid sigmoidoscopy and mid-rectal tumors(5e10cm from the anal verge)were lected.Patients were excluded if they had upper rectal tumor located above10cm from the anal verge or above the peritoneal reflection,or lower rectal tumor located less than5cm from the anal verge, or a T4lesion according to the TNM classification.19 Patients with a history of previous pelvic irradiation were also excluded.Patient characteristics are summarized in Table1.
Tumor asssment
T stage was recorded according to the TNM classifica-tion and N stage was recorded bad on whether the lymph node was involved by the tumor or not.CRM was defined by the shortest distance from the outermost margin of the tumor to the mesorectal fascia and a threatened CRM was defined as when the shortest distance was less than2mm.20 Magnetic resonance imaging methods
All patients underwent preoperative MRI with a1.5T whole-body system(Achieva,Philips Medical Systems, Best,The Netherlands)using four channel SENSE body coil(Philips Medical Systems).No bowel preparation,air insufflation,or antispasmodic agents were ud.
Axial T1-weighted conventional spin e echo images of the pelvis werefirst obtained using a24cmfield of view (FOV)in a5mm ction thickness,2mm interction gap,500e600/8e10(repetition time ms/echo time ms), a256Â192matrix,and one signal acquired.The sagittal, axial,oblique coronal,and oblique axial T2-weighted fast spin e echo images were quentially obtained using a24cm FOV in a3mm ction thickness,no interction gap,4000e6000/75e105(repetition time ms/echo time ms),a512Â256matrix,an echo train length of10e12, and the two signals were averaged.The oblique axial and oblique coronal images were obtained through45 angle clockwi rotation of the axial and coronal plane. Magnetic resonance imaging interpretation
One experienced gastrointestinal radiologist(CH Cha), with no knowledge of clinical and pathologic data,inter-preted each MR image on the PACS viewer(CentricityÔPACS solutions,GE Medical Systems,Milwaukee,WI, USA)regarding T stage,N stage,and CRM measurement.
All57tumors were divided into anterior,lateral and pos-terior groups within the rectum according to where the prin-cipal lesion of tumor to the rectal wall was located.For each tumor,CRM is measured on the sagittal,the axial,and either oblique axial or oblique coronal plane.CRM could not be accurately measured on both oblique axial and oblique cor-onal planes due to the pelvic concavity,therefore the shorter distance of the two CRMs,measured on the oblique axial and obliq识骨寻踪第八季
ue coronal plane,was chon(Fig.1A).
鼻头毛孔大怎么办Table1
Clinicopathological features of rectal cancer patients examined with pre-operative magnetic resonance imaging
Patients(n¼57) Median age(range)in years62(30e81)
Sex
Male33
Female24
Operation methods
Low anterior rection49 Hartmann’s operation1 Abdominoperineal rection2
Ultra-low anterior rection5
bdy
Median tumor distance from the anal verge
(range)in cm
8(5e10)
Tumor location within the circumference of the rectum
Anterior wall17
Lateral wall17
Posterior wall23
Depth of tumor invasion(T stage)
T17beyond the a
T210
T340
Tumor involvement of lymph nodeance
马达加斯加1下载
掉发吃什么好Negative25
Positive32
649
Y.W.Kim et al./EJSO34(2008)648e654
Surgery and gross examination
All patients underwent curative surgery by one surgeon (N.K.Kim)in accordance with the TME principle.The rec-tum and its surrounding mesorectal lymphovascular tissue were removed by sharp disction along the visceral pelvic fascia under direct visualization.2,5Optimal TME was de-fined if the specimen showed a good bulk of mesorectum with a smooth,lipoma-like surface with no or only minor defects/incisions (not deeper than 5mm).21Pathologic examination
Pathological specimen examination was performed by two pathologists (H.Kim and J.Pyo),using standardized whole-mount ctions via Quirke’s method.22After con-firming proper orientation,the rected specimen was thor-oughly examined with the naked-eye,marked with ink according to its direction,and fixed in 10%formalin for 48h,but not opened longitudinally on its front side.Serial ctio
ning of the specimen perpendicular to the longitudi-nal axis was carried out at 5mm intervals.The slices were embedded in liquid paraffin,ctioned,and stained with hematoxylin and eosin (H&E).The ctions were examined microscopically for T stage,N stage,and CRM measurement (Fig.1B,C).Statistical analysis
The accuracy,nsitivity,specificity,positive predictive value (PPV),and negative predictive value (NPV)were cal-culated for MRI to predict CRM measurement.The differ-ence in performance among the sagittal,axial and oblique MR images was analyzed by using receiver operating char-acteristic (ROC)curves (A z ).Statistical analysis was
买东西的英文
performed using SPSS software (version 12.0for Windows,Chicago,IL,USA).Results
Accuracy of magnetic resonance imaging predicting T and N stage
MRI correctly staged the depth of rectal wall invasion in 46patients,understaged it in two,and overstaged it in nine.Two T3tumors were understaged as T2by preoperative MRI.Six T1tumors were overstaged as five T2tumors and one T3tumor,and three T2tumors were overstaged as T3tumors.
MRI correctly staged N status in 40patients,overstaged it in one,and understaged it in 16.
Correlation of circumferential rection margin measurements between magnetic resonance imaging scan plane according to tumor location and whole-mount ctions
Tumor location was determined by its principal lesion to the rectal wall using the axial plane.Preoperative CRM was measured on the sagittal,axial,oblique planes and was compared with histologic CRM.The accuracy,nsitivity,specificity,PPV ,and NPV were prented for MRI in pre-dicting CRM measurement (Table 2).
Anterior tumors
For anterior tumors (n ¼17),MRI correctly predicted CRM for 12patients on the sagittal plane,14on the axial plane,and 14on the oblique plane.The A z values with 95%confidence interval (CI)of the sagittal,axial and obli-que plane were 0.66(95%CI:0.37,0.95),0.83(95%
CI:
Figure 1.T3mid-rectal cancer in a 69-year-old male.Measurement from the outermost margin of the tumor to mesorectal fascia is demonstrated.(A)On T2weighted oblique axial image,predicted circumferential rection margin is 1.17mm (arrow).(B)Ordered display of sliced specimen.Specimen orientation is marked with ink and the straightened rectum is fixed in buffered 10%formalin for 48h,and sliced perpendicular to the long axis at 5mm intervals.The shortest distance between tumor and mesorectal fascia is identified (arrow).(C)Corresponding histology finding.The slices are embedded in paraffin,c-tioned transverly,and stained with hematoxylin and eosin,and examined microscopically (H&E stain,Â12.5).Actual measured circumferential rection margin is 0.3mm (arrow).
650Y.W.Kim et al./EJSO 34(2008)648e 654
0.60,1.05),and 0.79(95%CI:0.53,1.04),respectively (Fig.2A).
Lateral tumors
For lateral tumors (n ¼17),MRI correctly predicted CRM for 10patients on the sagittal plane,12on the
axial plane,and 13on the oblique plane.The A z values with 95%CI of the sagittal,axial and oblique plane were 0.53
(95%CI:0.23,0.83),0.66(95%CI:0.37,0.95),and 0.78(95%CI:0.54,1.02),respectively (Fig.2B).
Posterior tumors
For posterior tumors (n ¼23),MRI correctly predicted CRM for 17patients on the sagittal plane,19on the axial plane,and 22on the oblique plane.The A z values with 95%CI of the sagittal,axial and oblique plane were 0.76
Table 2
Analysis of circumferential rection margin asssment between magnetic resonance imaging scan plane according to tumor location and whole-mount c-tions in 57patients Location Anterior Lateral Posterior MRI plane Sagittal Axial Oblique Sagittal Axial Oblique Sagittal Axial Oblique Accuracy (%)718282597176748396Sensitivity (%)8282913350838080100Specificity (%)508367738273728394PPV (%)759083406063445783NPV (%)
60
71
80
67
75
89九年级家长会发言稿
93
94
samui
100
MRI,magnetic resonance imaging;PPV ,positive predictive value;and NPV ,negative predictive
value.
Figure 2.Diagnostic performance reprented by receiver operating characteristic curves of the sagittal,axial,and oblique planes of magnetic resonance imaging for circumferential rection margin asssment.A z,corresponding area of receiver operating characteristic curves.(A)In anterior tumors,the A z value of the axial plane is the highest at 0.83.(B)In lateral tumors,the A z value of the oblique plane is the highest at 0.78.(C)In posterior tumor,the A z value of the oblique plane is the highest at 0.97.
651
Y.W.Kim et al./EJSO 34(2008)648e 654
(95%CI:0.52,1.00),0.82(95%CI:0.59,1.05),and0.97 (95%CI:0.91,1.04),respectively(Fig.2C). Correlation of circumferential rection margin measurements between magnetic resonance imaging
and whole-mount ctions when predictive margin involvement is defined as tumor within1,2or3mm of
the mesorectal fascia
The shortest CRM was lected among tho measured on the sagittal,axial,and oblique images irrespective of tumor location,which was applied to routine CRM mea-surement,and we altered the cutoff value of a CRM involve-ment at1,2,or3mm.The CRM measured on MR image using1,2or3mm CRM criteria was compared with histo-logic CRM using2mm criterion.The accuracy,nsitivity, specificity,PPV,and NPV are prented in Table3.
Using different cutoff values,MRI correctly predicted CRM for40patients with1mm criterion,49with2mm criterion,and39with3mm criterion.The A z values with 95%CI of1mm,2mm,and3mm criteria were0.65 (95%CI:0.49,0.80),0.84(95%CI:0.73,0.95),and0.58 (95%CI:0.43,0.73),respectively(Fig.3).
Discussion
Multidetector-row computed tomography(MDCT)with multiplanar(MPR)images has shown promising results of
preoperative staging for rectal cancer in recent years.How-ever,CT has its own limitation of poor tissue contrast as com-pared to MRI.23For this reason,further comparison studies are needed to determine if MDCT can compete MRI.24 MRI can clearly visualize the layers of the rectal wall and th
e mesorectal fascia which is important for asssing the pathologic staging and predicting CRM,and also re-veals an exact relationship between the rectal cancer and any adjacent pelvic organ through various scan planes in-cluding the sagittal,axial,oblique coronal,and oblique ax-ial planes.25e27Thus,we intend to evaluate the accuracy of preoperative MRI in predicting T stage,N stage and CRM, and to determine whether each different MRI scan plane provides an accurate CRM asssment.
The limitation of this study is that only one radiologist interpreted MR images.Accordingly,we could not asss inter-obrver variability.
Accuracy of MRI predicting T stage
Preoperative MRI accurately predicted T stage for46 patients.Two T3tumors were understaged as T2.Six T1 tumors were overstaged asfive T2and one T3,and three T2tumors were overstaged as T3.Previous studies reported that the overall accuracy rate for predicting T stage was 65e86%and most of the staging failures occurred in differ-entiating T2and borderline T3lesions due to desmoplastic reactions which make it difficult to distinguishfibrosis con-taining tumor cells fromfibrosis without tumor cells.15,16,28 Possible caus for overstaging of six T1tumors are stagin
g errors that occur when the signal intensity of T1tumor abuts the muscle layer and differentiating errors due in part to a limitation of MRI related to spatial resolution. Accuracy of MRI predicting N stage
MRI accurately predicted N stage for40patients,over-staged one,and understaged16.A lymph node with irregu-lar border characteristics or a mixed signal intensity was regarded as positive but a size criterion was not considered, and accuracy rates for N staging were reported between39%
Table3
Analysis of circumferential rection margin measurements between magnetic resonance imaging and whole-mount ctions when predictive margin involvement is defined as tumor within1,2or3mm of the mesorectal fascia
Criteria for CRM(%)1mm2mm3mm
Accuracy708249 Sensitivity419195 Specificity897720 PPV697143 NPV709388
CRM,circumferential rection margin;PPV,positive predictive value; and NPV,negative predictive
value.Figure3.Diagnostic performance reprented by receiver operating char-acteristic curves of1,2or3mm circumferential rection margin criteria compared with histologic circumferential rection margin using2mm criterion.The A z value of2mm criterion is the highest at0.84.A z,corre-sponding area of receiver operating characteristic curves.
652Y.W.Kim et al./EJSO34(2008)648e654
