Title Measurement of extent of spread of oesophageal squamouscarcinoma by serial sectioning

Author(s) Lam, KY; Ma, LT; Wong, J

Citation Journal Of Clinical Pathology, 1996, v. 49 n. 2, p. 124-129

Issued Date 1996

URL http://hdl.handle.net/10722/43598

Rights Creative Commons: Attribution 3.0 Hong Kong License

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by HKU Scholars Hub

14 Clin Pathol 1996;49:124-129

Measurement of extent of spread ofoesophageal squamous carcinoma by serialsectioning

K Y Lam, L T Ma,J Wong

Department ofPathology,University of HongKong, Queen MaryHospital, Hong KongK Y LamL T Ma

Department ofSurgeryJ Wong

Correspondence to:Dr Alfred King Ym Lam,Department of Pathology,Queen Mary Hospital,Polfulam Road,Hong Kong.Accepted for publication18 October 1995

AbstractObjectives-(I) To examine the prevalenceand extent of intramural metastasis insquamous cell carcinomas of the oeso-phagus so as to delineate the resectionmargins for these tumours; (2) to devisean appropriate method for measurementof these lesions which takes into accountof the contraction of the specimens afterresection.Methods-Oesophagectomy specimenswere prospectively collected from 96patients (87 males, nine females) withprimary oesophageal squamous cell car-cinoma over a two year period. The sizesofthe tumours were measured in situ, afterresection and after application of musclerelaxant (to regain their in situ length).The specimens were then serially sec-tioned for histological examination.Results-The sizes of the tumours meas-ured after application of muscle relaxantroughly corresponded to those measuredin situ. Intramural metastasis was ob-served in 26% of the cases. Sixty four percent (16 cases) of these were on the oralside, 72% (18 cases) on the gastic side,and 25% (nine cases) on both sides of thetumours. The most distant extent of in-tramural metastasis from the primarytumour was from 05 cm to 7-7 cm (mean =3-4 cm) on the oral side, and 05 to 9*5 cm(mean 4 cm) on the gastric aspect of thetumour. Intramural metastasis was seenonly in patients in whom the primary can-cer had deep muscle infiltration. Multipleneoplastic lesions could be detected in 33%of the patients. Both intramural meta-stasis and multiple neoplastic lesions wereassociated with extensive lymph node in-filtration. However, they had differenthistological features and extent of in-filtration.Conclusions-Intramural metastasis wasfrequently observed in oesophagealsquamous cell carcinoma. This impliesthat excision with wide margins should beconsidered for local control ofthe disease.(J7 Clin Pathol 1996;49:124-129)

Keywords: oesophagus, squamous carcinoma, intra-mural metastasis, multiple primaries.

Oesophageal cancers are tumours with a highmortality rate and a high incidence in orientalpopulations such as those of China and Hong

Kong.'2 About 90% of these cancers aresquamous cell carcinomas.3 The poor prognosisof these patients is partly a result of late pre-sentation of the lesions and difficulties inachieving curative resection of the tumours.Long term survival after excision of the oeso-phageal cancer is possible in only a small mi-nority of the patients. For instance, the fiveyear survival of patients with squamous cellcarcinoma after curative resection is 31 %, andafter palliative resection 15%, in Hong Kong.4Thus the primary objective of the treatment isto relieve the obstruction to allow the patientto swallow after the operation. Anastomoticrecurrence of the tumour will defeat the pur-pose of the resection. This is found in 6&1%of patients who have undergone resection ofoesophageal squamous cell carcinoma in ourlocality.5 Recurrence is closely related to themargin of clearance at operation, which is prin-cipally determined by the extent of intramuralspread of the primary tumour in the oeso-phagus. Intramural metastasis is also one ofthe most important prognostic factors for oeso-phageal carcinoma."' Therefore guidelineshave to be established from this known be-haviour of the tumour in order to decide theextent of the resection.We report a prospective study ofthe extent of

intramural metastasis ofoesophageal squamouscell carcinoma to define a safety margin forresection of the tumour. Since shrinkage ofspecimens occurs after resection, which mayaccount for the discrepancy claimed by thesurgeons and pathologists regarding the lengthof the margins,'2 we devised a method ofstretching the specimens so that the meas-urements are representative ofthe in situ values.

MethodsPatients with resectable primary oesophagealcarcinomas treated in Queen Mary Hospital inHong Kong from January 1984 to December1986 were evaluated prospectively. In caseswho had undergone Lewis-Tanner operations,measurement ofthe longitudinal length ofoeso-phageal tumours was made at operation beforethey were removed from the patients (in situ).Each of the resected fresh specimens wasopened along the border in such a way as toavoid cutting through the tumour, and it wasthen laid out flat. The lengths of the tumourswere again recorded. The specimens were thentaken fresh as soon as possible (within 20 min-utes) to the pathology department from the

124

Measurement of spread of oesophageal squamous carcinoma by serial sectioning

operating room by a research assistant. Themeasurements of the size of these tumourswere repeated immediately they reached thedepartment of pathology. The specimens werethen put into a preprepared solution for twominutes to allow them regaining their in situlengths. This solution was a mixture of 1 ml1% procaine hydrochloride (Novocain, Wm-throp), a muscle relaxant, with 500 ml 0 9%sodium chloride heated to 37°C. The speci-mens were pinned on a foam board by gentletraction and photographed in a fresh state forrecord keeping. The length of the tumours wasmeasured again.Three days after fixation in 10% formalin,

serial sectioning of the tumours in horizontaland longitudinal planes was done. The entirespecimen was then cut into slices longitudinallyat intervals of 0 5 cm in width with a razor andembedded in paraffin wax blocks. The blockswere labelled and numbered consecutivelyusing the centres of each tumour as the ref-erence point (fig 1). The sections from theparaffin wax blocks were processed with routinehaematoxylin and eosin staining and histo-logically examined.Only squamous cell carcinomas arising from

the oesophagus in Chinese patients were in-cluded in this study. Age and sex ofthe patientsand the location of the tumours were recorded.The tumours were classified into well, mod-erately, and poorly differentiated according tothe WHO classification.'3 Each of the sectionsfrom the primary oesophageal tumour was eval-uated for the depth of infiltration of the car-cinoma (in the mucosa, submucosa, circularmuscle, longitudinal muscle, and adventitia)through the wall of the oesophagus. Lymphnode metastases and tumour vascular per-meation were also identified. The presence andlocation of intramural metastasis were noted.We defined intramural metastasis as a meta-static lesion that is clearly separated from theprimary tumour and invading the oesophagealor gastric wall but not surrounded by endo-thelium. In each specimen, the greatest dis-tance ofintramural metastasis from the primarytumour was measured on each side (proximalor distal) of the primary tumour according tohistological examination of the slides from thelabelled blocks. In addition, the presence andnumber of multiple neoplastic lesions in theoesophagus were determined. Multiple neo-plastic lesions are defined as lesions (dysplasia,carcinoma in situ, or invasive carcinoma) thatare distant from one another by at least 1 cm.In each case, the largest sized tumour was takenas the primary lesion while the smaller ones asassociated lesions, or, if they were of the samesize, the tumour with deepest invasion was

Table 1 The four sets of measurements of the tumour and their relations

Range Mean (SD) Total length of tumour in(cm) (cm) the 46 specimens (cm)

In situ 10-12-5 6-28 (2 45) 289 (100%)*In theatre 1-G12-5 5-65 (2-14) 260 (90%)In resected specimen 1-1-12 5 5-67 (2-37) 261 (90%)After relaxation 1-3-12-5 6-41 (2 53) 295 (102%)

* The in situ measurement was used as the standard (100%) for comparison.

U4-

..l.IF

U1

0

si'tm~ 10 20 30 40Cli so 60 0 80 §0 bJ

Figure 1 The resected oesophagus was cut into slices by aserial blocking technique afterfixation. The tumour wascut into both longitudinal and horizontal sections. Theremaining specimen is divided into upperlproximal (U)and lowerldistal (L) proportion with the centre of tumouras the reference point. They were then cut horizontally at5mm intervals and labeUled consecutively, with thesmallest numbered block one nearest the tumour.

regarded as the primary lesion. The histologicalfeatures of intramural metastasis and multipleneoplastic lesions were both identified.

STATISTICSStatistical analysis was performed using Fish-er's exact test, X2 test, and Student's t test. Ap value of less than 0 05 was considered to besignificant.

ResultsThe measurements of the tumour length andthe relative proportion of shrinkage for the fourstates of the specimens (that is, in situ, in theoperation theatre after resection, and in thepathology department before and after ap-plication of muscle relaxant) in 46 cases ofoesophageal squamous cell carcinoma areshown in table 1. The lengths of the resectedtumours measured in the operation theatre and

125

Lam, Ma, Wong

Table 2 The clinicopathologicalfeatures of the patients with resected oesophagealsquamous cell carcinomas

Male Female Whole group(n = 87) (n = 9) (n = 96)

Age range, years 44 to 86 52 to 74 44 to 86Mean age (SD), years 61 (8) 63 (7) 61 (8)Range of the size of the tumour, cm 1-3 to 1-5 2 2 to 9 7 1-3 to 12-5Size of the tumour, mean (SD), cm 6-1 (2 4) 6 3 (2-8) 6-1 (2-4)Differentiation of the tumour, n (%)

Well 31 (36%) 4 (44%) 35 (36%)Moderate 40 (46%) 4 (44%) 44 (46%)Poor 16 (18%) 1 (12%) 17 (18%)

Site of the main lesion, n (%)Upper 7 (8%) 1 (11%) 8 (8%)Middle 57 (66%) 6 (67%) 63 (66%)Lower 23 (26%) 2 (22%) 25 (26%)

Table 3 Clinicopathological features ofpatients with and without intramural metastasis

Intramural metastasis

Positive cases (n 25) Negative cases (n = 71)

Age range, years 44 to 86 45 to 78Mean age (SD), years 61 (9) 62 (8)Male/female 24/63 1/8Range of size of the tumour, cm 4-3 to 8-5 1-3 to 12 5Size of the tumour, mean (SD), cm 6-3 (1-5) 6-0 (2 7)Differentiation of the tumour, n (%)

Well 11 (44%) 24 (34%)Moderate 10 (40%) 34 (48%)Poor 4 (16%) 13 (18%)

Site of the main lesion, n (%)Upper 2 (8%) 6 (8%)Middle 17 (68%) 46 (65%)Lower 6 (24%) 19 (27%)

in the pathology department were nearly thesame. However, the lengths measured in thisstate showed 10% shrinkage when comparedwith those obtained by in situ measurements.The tumour lengths measured after applicationofthe muscle relaxant corresponded reasonablyto the in situ values (102% of the in situmeasurements). The measurements after ap-plication of the relaxant were thus used in theother parts of the study.Ninety six patients (87 males, nine females)

with resected primary oesophageal carcinomaswere included in the study. The clinico-pathological features of these cases are shownin table 2. The mean age of the patients was

Figure 2 Intramural metastases in the oesophageal wall. The lesions present as nodulesand are covered by intact epithelium. Scanty or no significant inflammation is present.(Haematoxylin and eosin, original magnification x 115.)

61 years (range 44 to 86) and the mean size ofthe main tumour was 6 1 cm (ranging from 1 3to 12 -5 cm). Thirty six per cent ofthe squamouscell carcinomas (35 cases) were well differ-entiated, 46% moderately differentiated (44cases), and 18% poorly differentiated (17cases). Most of the cases were found in themiddle portion of the oesophagus (8% in theupper, 66% in the middle, and 26% in the lowerportion). There was no significant differencebetween males and females with regard to theage distribution (p = 042, z= 0-8, t test) of thepatients, or the size (p=0-84, z=0 2, t test),differentiation (p = 0-25, Fisher exact test), andsite (p = 0 99, Fisher exact test) ofthe tumours.

Intramural metastasis was found in 26% ofthe patients (25 cases; 24 males and one female)after serial sectioning of the specimens. Theclinicopathological features of the cases withand without intramural metastasis are given intable 3. The male to female ratio was similarin both groups of patients (p=0 44, Fisher'sexact test). There was no significant differencebetween the patients with and without in-tramural metastases with regard to age (p =062, z=0 49, t test), size (p=0 5, z=0 68,t test), site (p=0 96, X2=0 88, df=2), anddifferentiation (p = 0-66, x2 = 0X83, df= 2) ofthe tumours.

Intramural metastases usually consisted ofnodular masses in the submucosa or themuscles of the specimens (fig 2). Scanty or noinflammatory cells were present around thelesions. Metastases could be seen in the oralor the gastric side of the primary oesophagealcancer or in both. Sixty four per cent of thepatients (16 cases) in this study had intramuralmetastasis on the oral side, 72% on the gastricside (18 cases), and 25% on both sides (ninecases). Intramural metastases on the oral sideof the specimens were separated from the maintumour by from 0 5 to 7-7 cm (mean = 3*4 cm,median = 2 5 cm, SD = 2 2 cm), while those onthe gastric side were from 0 5 to 9 5 cm away(mean = 4 cm, median = 2 5 cm, SD = 3 4 cm).No significant difference was identified in theextent of infiltration of intramural metastaseson either the oral or the gastic side (p =0054.z = 0 62, t test).

Multiple neoplastic lesions were detected in33% of the patients (32 cases; 30 males andtwo females). The lesions were usually flat,diffuse or ulcerated, and with a prominentinflammatory reaction around them. Of these32 cases, 22% (seven cases) were invasive car-cinomas, 19% (six cases) were carconoma insitu, and 59% (19 cases) were dysplastic le-sions. The proximal extent of multiple neo-plastic lesions from the primary tumours was14-5 cm while the distal extent was 13-1 cm.The clinicopathological features of patientswith and without multiple neoplastic lesionsare given in table 4. There was no significantsex and age difference in patients with andwithout multiple neoplastic lesions (p=0 7,Fisher's exact test for sex; p = 0 58, z = 0 56 forage by t test). The size of the primary tumoursin patients with multiple neoplastic lesions wasnot different from that from patients withsingle oesophageal carcinomas (p = 084, z = 0-2,

126

Measurement of spread of oesophageal squamous carcinoma by serial sectioning

Table 4 Clinicopathological features ofpatients with and without multiple neoplasticlesions

Multiple neoplastic lesions

Positive cases (n = 32) Negative cases (n= 64)

Age range, years 48 to 74 44 to 86Mean age (SD), years 62 (8) 61 (9)Male/female 30/2 57/7

Range of size of the tumour, cm 2 to 10 1-3 to 12-5

Size of the tumour, mean (SD), cm 6-2 (2-1) 6-1 (2-6)Intramural metastases, n

Positive cases 11 21Negative cases 14 50

t test). Furthermore, there was no relationbetween the presence of intramural metastasisand multiple neoplastic lesions (p = 0-19, X2=

1 73, df= 1).Lymph node metastases were present in 64%

of the 96 patients with oesophageal squamouscell carcinoma. They were identified in 80%of patients with intramural metastasis (20 of25 cases) and in 58% of patients without in-tramural metastasis (41 out of 71 cases). Thusthere was a significant difference in the fre-quency of lymph node infiltration in patientswith and without intramural metastasis (p=0.047, X2 = 3.95, df= 1). In patients with mul-tiple neoplastic lesions, 78% (25 cases) hadlymph node infiltration by carcinoma, while inpatients without multiple neoplastic lesions,56% (36 cases) had lymph node metastases.The association between lymph nodes meta-stases and multiple neoplastic lesions (p = 0 04,x2 =4 41, df= 1) was significant.

Vascular permeation by tumours was seen in17% of the patients (16 cases). It was oftenseen at the base of the tumour but could oc-

casionally be identified at a distance from theprimary tumour (at a maximum distance of5 cm from the main tumour). The conditionwas observed in half the cases with intramuralmetastasis (nine of 18 cases) and in 10% ofpatients without intramural metastasis (sevenof 71 cases). Therefore vascular permeationby tumours was found to be associated withintramural metastasis (p<0*01, x2 =1569). Itwas further observed that vascular permeationwas found in 28% of cases with multiple neo-

plastic lesions (nine cases) and in 11% of

patients without multiple neoplastic lesions(seven cases). There was significant associationbetween vascular permeation by tumours andmultiple neoplastic lesions (p = 0 03, X2= 454,df= 1). Lymph node metastases were not re-lated to vascular permeation by tumours (p=0-044, Fisher's exact test).The various clinicopathological features as-

sociated with the depth of oesophageal car-cinomas are given in table 5. Early oesophagealsquamous cell carcinomas were present in threecases in which the carcinoma invaded up tothe submucosa. There was no difference inage (p=O096, z=0-052, t test) or sex (p=0 9, Fisher's exact test) between tumours withtransmural infiltration and those confined tothe wall of the oesophagus. On the other hand,there was significant relation between the depthoftumour infiltration and the size ofthe tumour(p=0 009, z=2-6, t test). Although vascularpermeation was only present in those cases

where there was at least infiltration in the lon-gitudinal muscle layer of the oesophagus, vas-

cular permeation by tumours was not relatedto the depth of infiltration (p = 0-6, Fisher'sexact test). On the other hand, lymph nodemetastases were significantly correlated with a

depth of tumour infiltration (p = 0-02, X2 =5-18, df=1). Intramural metastasis was alsoseen only in patients with deep muscle (lon-gitudinal muscle) infiltration. However, therewas no significant relation between the level ofinfiltration by the primary tumour and in-tramural metastasis (p = 0 07, Fisher's exacttest).

DiscussionIntramural metastasis in oesophageal cancers

was first reported by Watson in 1933,6 but sincethen little attention has been paid to it in theEnglish literature except for a few reports.7'13The incidence of intramural metastasis ob-served by these research groups varied greatlyfrom 9% to 35%. The incidence detected inthe current study (26%) by serial sections ofthe entire surgically resected oesophagus alsofell within this range. It is thus clear that in-tramural metastasis is frequently observed inoesophageal squamous cell carcinomas.

Table 5 The relation of depth of infiltration of carcinoma in the wall of oesophagus to other clinicopathological factors

Level of involvement of the oesophageal wall by carcinoma

Mucosa Submucosa Circular muscle Longitudinal muscle Adventitia

Total number of cases 0 3 5 17 71Male/female 0/0 3/0 3/2 16/1 65/6

Age range, years - 49 to 71 53 to 76 48 to 77 44 to 86Mean age (SD), years - 61 (11) 63 (4) 61 (8) 61 (9)Size range (cm) - 2-3 to 2-6 1-3 to 8-5 2 to 10 7 2-1 to 12-5Mean size, (SD), cm - 2-4 (0 2) 3-8 (2-8) 5-8 (2-4) 6-5 (2 3)Lymph node infiltration, n

Positive cases 0 0 2 8 51Negative cases 0 3 3 9 20

Vascular permeation by tumour, nPositive cases 0 0 0 3 13Negative cases 0 3 5 14 58

Intramural tumour metastasis, nPositive cases 0 0 0 3 22Negative cases 0 3 5 14 49

127

Lam, Ma, Wong

There is no consensus so far on whetherintramural metastases are more commonly ob-served on the proximal (oral) side or the distal(gastric) side of the primary lesion. In theprevious reports, as well as in our present study,the number of cases with intramural metastaseson the proximal and the distal side is almostthe same. However, most studies (except thatof Kato and his colleagues9) have shown atendency for there to be more metastases onthe distal side. Nevertheless, these metastasescan occur on both sides of the primary tumourand the published reports do not indicate astrong predilection for direction of spread.Lymph node infiltration, which we identified

in 80% of patients with intramural metastasisin this study, was significantly associated withintramural metastasis. The high frequency oflymph node metastases in the presence of in-tramural metastasis has also been noted inother studies."'0 These findings concur withthe theory that intramural metastases may ori-ginate in the intramural lymphatic spread ofthecarcinoma. This also suggests that intramuralmetastases may be a secondary focus of theinvolved lymph nodes, in addition to the prim-ary carcinoma. On the other hand, vascularpermeation was seen in 50% of patients withintramural metastasis. Although vascular per-meation was associated with intramural meta-stasis in our study, its frequency in patientswith intramural metastases was not as high asthat of lymph nodes metastases.

Intramural metastases was not present inearly oesophageal squamous cell carcinomaswhich we examined in this study. They wereonly found when the tumours had invadedinto the longitudinal muscle. However, nosignificant association was noted betweenthe depth of invasion and the presence ofintramural metastasis. The other clinico-pathologic variables like age, sex of the patient,and location, size, and site of the primary tu-mours were not related to the presence ofintramural metastases. Furthermore, there wasno relation between multiple neoplastic lesionsand intramural metastasis.Anastomotic recurrence has been reported

to be related to the length of the resectionmargin from the primary oesophageal car-cinoma.5 Careful attention must therefore bepaid to the choice ofmargins during operationson carcinoma of the oesophagus. In addition,shrinkage of the oesophageal tissue may occurafter operation."2 In this study, we propose amethod-using muscle relaxant on the freshtissue-to regain the in situ lengths of thetumours so as to obtain representative meas-urements of the extent of intramural meta-stases. However, the lengths of the tumoursobtained by this method appear to be slightlygreater than the in situ length (102% of the insitu length, see table 1), probably because ofrelaxation of the numerous myofibroblasts inthe tumour stroma. The second drawback inthis investigation lies in the confidence limitsof the measurement of the extent of intramuralmetastasis. This is no better than 0 5 cm, asthe blocks were taken at 0-5 cm intervals.Nevertheless, the most distant extent of the

intramural metastases found in this study was7*7 cm proximal to the primary tumour and9*5 cm distal. The results obtained from thepresent study are comparable with the clinicalfindings reported previously in our hospital.5In that study, a proximal resection margin ofless than 5 cm measured in situ was found tohave a 20% risk of anastomotic recurrence, anda margin of between 5 and 10 cm had only an8% risk. Thus these findings indicate that longresection margins are necessary for completeclearance of oesophageal carcinomas. This iseasier to achieve on the distal margin ratherthan on the proximal one since a longerproximal margin would need a pharyngo-laryngectomy and the voice is not preserved.

Multiple neoplastic lesions are common inpatients with oesophageal cancer.'415 This maybe related to the impact of external factors likesmoking and alcohol intake on carcinogenesisin this group of tumours. In the present study,multiple neoplastic lesions were detected in33% of patients with oesophageal squamouscell carcinoma. As in the study by Kuwano etal,'5 we found no significant difference in sexor age of our patients or the size of the primarylesion between patients with single tumoursand those with multiple neoplastic lesions inthe oesophagus. On the other hand, lymphnode metastases and vascular permeation weremore common in patients with multiple neo-plastic lesions. The proximal extent of the le-sions from the primary tumour was 14-5 cmwhile the distal extent was 13 1 cm. Thus whenmultiple neoplastic lesions were found in theoesophagus it was difficult to achieve a clearproximal margin. Microscopically, multipleneoplastic lesions are flat, diffuse, or ulcerated,with prominent inflammatory reaction in thetumour stroma, in contrast to intramural meta-stases which are usually nodular masses withintact overlying epithelium, with scanty or noinflammatory cells. Thus multiple neoplasticlesions are different from intramural metastasesin the histological features and the extent ofinvolvement in the oesophagus.

In summary, intramural metastasis are fre-quently observed in oesophageal squamous cellcarcinomas. Recurrent and possibly persistentdisease is the rule with surgical treatment alone.In the light of this, it is important to havewide resection margins as well as adjuvanttreatments to control the local and systemicdisease process in the treatment of oesophagealcarcinoma.

1 Parkin DM, Laara E, Muir CS. Estimates of the world-wide frequency of sixteen major cancers in 1980. Int JICancer 1988;41: 184-97.

2 Moses FM. Squamous cell carcinoma of the esophagus.Nature history, incidence, etiology, and complications.Gastroenteml Clin North Am 1991;20:703-16.

3 Lam KY, Loke SL, Ma LT. Histochemistry of mucin se-creting components in mucoepidermoid and adeno-squamous carcinoma of the oesophagus. J Clin Pathol1993;46: 1011-5.

4 Law SYK, Fok M, Cheng SWK, Wong J. A comparison ofoutcome after resection for squamous cell carcinomasand adenocarcinomas of the esophagus and cardia. SurgGynecol Obstet 1992;175:107-12.

5 Tam PC, Siu KF, Cheung HC, Ma L, Wong J. Localrecurrences after subtotal esophagectomy for squamouscell carcinoma. Ann Surg 1987;205:189-94.

6 Watson WL. Carcinoma of the esophagus. Surg GynecolObstet 1933;56:884-97.

128

Measurement of spread of oesophageal squamous carcinoma by serial sectioning

7 Burgess HM, Baggenstoss AH, Moersch HJ, Clagett OT.Carcinoma of the esophagus: a clinicopathologic study.Surg Clin North Am 1951;31:965-76.

8 Takubo K, Sasajima K, Yamashita K, Tanaka Y, Fujita K.Prognostic significance ofintramural metastasis in patientswith esophageal carcinoma. Cancer 1990;65:1816-9.

9 Kato H, Tachimori Y, Watanabe H, et al. Intramural meta-stasis of thoracic esophageal carcinoma. IntJ Cancer 1992;50:49-52.

10 Maeta M, Kondo A, Shibata S, Yamashiro H, MurakamiA, Kaibara N. Esophageal cancer associated with multiplecancerous lesions: clinicopathologic comparisons betweenmultiple primary and intramural metastatic lesions. Gastro-enterol 3pn 1993;28: 187-92.

11 Fukuzumi N. Intramural spread of esophageal carcinoma.Surg Pathol 1994;5:269-94.

12 Siu KF, Cheung HC, Wong J. Shrinkage of the esophagusafter resection for carcinoma. Ann Surg 1986;23:173-6.

13 Watanabe H, Jass JR, Sobin LH. Histological classificationof oesophageal tumours. In: Histological yping of oeso-phageal and gastric tumours, 2nd edn. Berlin: Springer-Verlag, 1990:13.

14 Mizobuchi S, Kato H, Tachimori Y, Watanabe H, Yama-guchi H, Itabashi M. Multiple primary carcinoma of theoesophagus. Surg Oncol 1993;2:249-53.

15 Kuwano H, Ohno S, Matsuda H, Mori M, Sugimachi K.Serial histologic evaluation of multiple primary squamouscell carcinomas ofthe esophagus. Cancer 1988;61:1635-8.

129