Endosonography for rectal carcinoma : Preoperative TNM staging compared to histology

Transreccal endosonography was performed preoperatively in 58 
patients with rectal carcinoma. The results of endosonography were compared 
to the histology of resected specimens according to the new (1987) TNM 
(tumour, nodes, metastases) classification. Endosonography was accurate in the 
staging of tumour categories except with T2 carcinomas because of concomitant 
peritoneal abscesses, inflammation or tissue changes after irradiation therapy. 
The overall accuracy of endosonography was 81 %. Overstaging occurred in 17% 
and understaging in 2%. The accuracy of endosonography for staging regional 
lymph nodes was 74%, sensitivity was 95% and specificity 61 %. Endosonography 
was not accurate in the staging of distant metastases due to the limited penetration 
depth of ultrasound used. In conclusion, endosonography will become the 
standard for staging rectal carcinoma.

Ence qui couchc !'extension aux ganglions lymphariques, la precision de la classification atteignait 74%, la sensibilite 95% et la specificite 61 %.A cause de la capacite de penetration limitee des ultrasons utilises, l'endosonographie erait imprecise Jans la classification des metastases a distance.En conclusion, l'endosonographie est en passe de devenir la technique standard de classification des cancers du rectum.nosing rectal carcinoma.The submucosal extent of the tumour, however, cannot be assessed.Even computed tomography is not accurate for staging rectal carcinoma because of inability to image the individual layers of the rectal wall (1,2).Endoscopic and nonoptic sonography, generally known as endosonography, were developed to improve the diagnostic value of ultrasound by directly approaching the target lesion via the gastrointestinal lumen with a high frequency ultrasonic beam (3)(4)(5)(6)(7)(8).This technique has been reported to be accurate for clinical TNM (tumour, nodes, metastases) staging of gastrointestinal carcinoma (9)(10)(11)(12)(13).

Academic Medical
Recently, there has been a revision of the T categories and stage grouping of the TNM classification, which now permits a direct translation to Duke's classification 04-17).The N classification has been revised to account for the lymph nodes as well as their location.These changes were made based on data from the Erlangen Tumor Registry ( 16).The aim of th is study was to assess the accuracy and limitations of transrectal endosonography in clinical TNM staging of rectal carcinoma according to the new (1987) TNM classification.

MA TERLALS AND METHODS
Between March 1984 and April 1989 endosonography was performed preoperatively in 58 patients with rectal carcinoma proven by endoscopic biopsy.There were 40 males and 18 females with ages ranging from 26 to 90 years (average 66).These examinations were performed within four weeks before surgery.The results of endosonographic images an<l histology of resected specimens were staged according to the new ( 1987) TNM classification.Instruments: The author has been routinely using the rigid Atoka ASU-59 and a flexible Atoka prototype ASU-57 to examine the rectal carcinomas (Fig- ure l ).For rectosigmoid colon, the author has been using the side-viewing 10 MHz echoendoscope (EUM2) or the forward-viewing echocoloscope (AXF-EUM2).The latter can be more readily maneuvred endoscopically because of its forward-viewing optics (Figure 2).This radial scanner has a sector of approximately 300° because the biopsy channel adjacent to the transducer hampers the transmission of ultrasound.The area under the biopsy channel cannot be seen due to the total reflection of the ultrasonic beam.The recently available Olympus echoendoscope can be attached with a water-filled balloon at the transducer (Figure 3 ).This makes imaging of colorectal abnormalities much easier.The specifications of these instruments are summarized in Table l.Investigation technique: The technique of investigation is compatible with rectosigmoidoscopy in patients lying in the left lateral decubitus position after phosphate enema.Rectal digital examination is obligatory to assess the local anatomy and co dilate the sphincter and muscle prior to insertion of the instrument.The nonoptic instrument is blindly inserted as deeply as possible.Thereafter, the instrument should carefully be withdrawn until abnormalities are imaged sonographically.By filling the balloon with water the polypoid or exophytic configuration of tumours can be clearly visualized.The method of investigation with the echoendoscope is similar to the examination of the stomach for gastric carcinomas.
The tumour should be visualized endoscopically.The echoprobe is positioned adjacent to the tumour.Thereafter, the balloon or the colorectal lumen is filled with water to produce adequate transmission for ultrasound.Whenever possible, the instrument should be passed beyond the lesion into the proximal colonic segments to visualize lymph node abnormalities and to determine the proximal noninfiltrated area.This is important co give accurate information for localization of resection margins.The interpretation of normal and pathological structures is based on the results of previous studies.Endosonographic criteria for new TNM staging are given in Table 2.
Criteria for assessing lymph node metastases are as follows.Lymph nodes with a hypoechoic pattern and clearly delineated boundaries are suspicious of malignancy.Direct extension of mural abnormalities into adjacent lymph nodes is highly suspicious of malignancy.Lymph nodes with a hyperechoic pattern and indistinctly demarcated boundaries are indicative of bcnignancy.
Staging of distant metastasis with endosonography was excluded because

There were 14 false positives In the NO class and one false negative In the NI class. ' Patients with Inaccurate staging according to the separate NI and N2 definitions of metastasis. Values given ore numbers of patients. (For explanations of N categories, see Tobie 2)
lymp h node metas t as is a long t he suprarectal blood vesse ls, liver metastas is a nc.l periton ea l c.l issem ina t ion could not be imaged.T his was exp lainec.lby the limi ted pe netration depth of ultrasounc.l and the c.l ifficult anatomical route to reach the target of inte rest.

RESULTS
T able 3 summarizes the results of endosonography and histology in assessing the depth of t umour infiltration.The depth of tumour infiltration was assessed in 58 patients.
T l cc1 rcinoma was correctly diagnosed in five of seven patienu.. Overstaging occurred in two patie nts due to peritumoral infiltration .
T2 carcinoma was correctly diagnosed in seven of 15 patients (Figure 4).Overstaging occurred in e ight patients c.lue to perircc t al abscesses (th ree pc1t ients) , peritumoral inflam mation (four patients) or preoperat ive irrad iation with destruction of peritumoral structures (one patient).T3 carcinoma was correctly diagnosed in all 34 patients (Figure 5).T4 carcinoma was correctly diagnosed in one of two patients.Understaging occurred in one patient because penetration into the dorsal wall of the vagina was not imaged.
Table 4 summarizes the results of endosonography and histology in assessing reg ional lymph nodes.The number of lymph nodes in one resected specimen varies from one to 17 with a total of 33 7 (average per resected specimen six).Metastases (N 1 and NZ) were found in 114 lymph nodes (34%).Nonmetastatic tumours (diameter range 2 to 18 mm) were correctly diagnosed in 22 of 36 patients.Incorrect diagnoses were made in the remaining 14 patie nts.Lymph node metastases (diameter range 6 to 22 mm) were correctly diagnosed in 21 of 22 patients.False negative diagnoses were made in the remaining patient due to granulomatous inflammation.However, accurate staging accord ing ro the separate N 1 and NZ definitions of metastasis was done in only 15 patients: Nl metastasis was correctly staged in eight of 13 patients and NZ in seven of nine patients.N l metastasis was erroneously classified as NZ in four patients and NZ as N 1 in two.The overall accuracy of endosonography was 74%, scns1t1v1ty 95% and specificity 61 %.The positive predictive value was 60% and negative was 96%.

DISCUSSION
Endosonography is accurate in the assessment of tumour category becau~e of the ability to image the depth of tumour infiltration and transition between normal and pathological wall structure.A close correlation between endosonography findings and histology can be demonstrated.Various sections of endosonography images are crucial fo r the assessment of maximal depth and extent of carcinomatous infiltration.This is essential for comparing the clinical T category with the pathological T category.Overstaging may occur because of peritumoral inflammation, which cannot be distinguished from carcino mato us infiltration on ultrasound.Understaging may occur due to severe stenosis, which cannot be passed with the instrument.
Endosonography is more accurate for determining metastatic involvement than for identifying nonmetastatic lymph nodes.Distinction between a micrometastatic and a benign lymph node or between an inflammatory enlarged lymph node and lymph node metastasis cannot be made based on ultrasound alone.Therefore, false positive and false negative diagnoses may occu~.
Stage grouping is a combination of T, N and M categories.Therefore, stage grouping can be assessed if additional cranscutaneous ultrasound or computed tomography is incorporated.Incorrect diagnosis of each category, however, may lead to erroneous stage grouping.This has also been reported in the staging of upper gastro intestina l carcinomas (9)(10)(11)(12)(13).
In conclusion, the author believes that endosonography will become the standard procedure for staging rectal carcinomas.The important information for the surgeon is the delineation of the tumour free regio n , bo th proximal and distal to the primary site.Moreover, lymph nodes adjacent to and distant from the edge of tumour should be carefully examined.With this info rmation, radical tumour resection and lymph node dissection can be planned.In cases of nonsurgical treatment with laser photocoagulation or irradiation, the depth of tumour infiltration before and after treatment should be carefully measured.In this manner, documentation after therapy can be done.

FigureFigure 2 )
Figure I) A rigid Aloka ultrasonic instrument ( ASU-57) with an echoprobe ( e) attached at the tip of the rigid shaft.w Water channel for filling with water the balloon attached at the transducer

Figure 4 )Figure 5 )
Figure 4) A Endosonogram of a hypoechoic rectal carcinoma ( t) with penetration into the muscularis propria ( mp) localized ventrally adjacent to the prostate gland (pr).B Corresponding endoscopy showing a polypoid tumour

TABLE 1 Technical data of various colorectal echoendoscopes
All echoprobes ore mechanical sector or radial scanning (I B(f or 36(J>).Only the Olympus AXF-EUM2 llos copoblllty for endoscopicol/y guided puncture or biopsy

TABLE 2 Endosonographic criteria for 1987 TNM staging of rectal carc inoma
CAN J 0ASTR0ENTEROL V OL 4 No 9 DEC EMHER l 990