Original Contributions Distribution of Dysplasias and Early Invasive Carcinoma in Barrett’s Esophagus JOSEPH E. McARDLE, MD, KLAUS J. LEWIN, MD, GAYLE RANDALL, MD, AND WILFRED WEINSTEIN, MD To

assess

the anatomic relationships

between

areas of dysplasia

and “early” carcinoma, we evaluated histologically cosal surfaces of seven esophagectomy grade dysplasia

specimens resected for high-

or early invasive (intramucosal

and submucosal)

carcinoma. We developed

surface area maps and assessed the var-

ious degrees

or carcinoma

of dysplasia

or early invasive (intrarnucosal and suh~nucosal) carci1101na. The specimens were dissected so that the entire mucosal surface was examined. Maps werca constructed to anatomically locate areas of dysplasia and carcinoma. A number of topologic analyses were then made to further delineate the anatomic relationships between foci

the entire mu-

at 10 equidistant

points.

Our analysis shows an equal likelihood of high-grade dysplasia and/ or early invasive carcinoma Barrett’s epiihelium. rett’s epithelium

occurring

throughout

MATERIALS

a

sequence. The amount of dysplastic epithelium

we found no association between the extent of dysplasia

and the likelihood current standard

of finding carcinoma. This study supports of practice for clinical surveillance

with Barrett’s esophagus

by uniformly

distributed

the

of patients

endoscopic

bi-

opsy of the complete length. In addition, the presence of any degree of dysplasia ,may be an indication for close clinical follow-up.

PATHOL23:479-482.

the adjacent

histology.

AND METHODS

Sevenrsophagectomy specimens were r,evlcwed from patients undergoing resection for rndoscopicallv proven highgrade dysplasia or carcinoma. in which histologic evaluation showed either high-gi-ade dysplasia or earlv invasive (intraniucosal and suhniucosal) carcinoma. The endoscopies and biopsies were done according to xi ongoing surveiliance protocol in patients with Rarrett’s esophagus.” Six of the seven patients repot-ted here had dysplasia andi/ot- lc,lr-cinema diagnosed elsewhere and were refet-red for second opinion and surgeiv. One of the patients came from our formal surveillance progr&. Clinical and endoscopic information is provided in ‘fable 1 Specimens were fixed in Rouin’s xolut ion for a period of Z-1 to 48 hours prior to dissectioll. ‘fhe tissue blocks of the entire spec,imen were submitted front longitudinal strips of’ esophagus. botmd proximally by squamous epithelium and distally by gastric. epithelium. PI-oximal ntargins of each piece of tissue were marked with Indin ink so that c’ompcments of eac.h strip were correctly oriented with regard to proximal and distal ends. Three serial sections from each block were rt‘viewed, including one alcian blue (pH 2.5)-stained slide. Each c.ase was reviewed by three of the :authorh (J.E.M., K,j.I,., and W.W.). Barrett’s mucosa was defined by the presence of goblet cells or alcian blue-positive columnar surface cells. The c lassifcation of dysplasia was lnade using the criteria modified from Reid et al,!’ consisting of a ~OLII-tiered classification system: (I) negative for dysplasia, (2) low-grade dysplasia or indefinite for dysplasia, (3) high-grade dysplasia, and (4) intramucosal and submucosal carcinoma (“early” carcinoma). High-grade dvsplasia, as commonly defined and previously reported by us:!’ constitutes a severe degree of cytologic and architectural atypia, and differs from intrarnucosal carcinoma by the lack of invasion through the basement membrane. Agreement as to the severity of dysplasia or the presence of intramucosal carcinoma was achieved bv a consensus of the reviewing aurhol-s. Three hundrrd fifty-;ight histologic slides from the se\‘en esophagectomy specimens ‘were examined. Sections from slides corresponding to longitudinal strips were then reassembled and the length of various histologic patterns was directly measured from the slides. ‘l‘hesr measurements wer-e used to generate specimen diagrams (Fig 1) using DeTX). signel 2.0 software (Micrografx Inc, Richardson, In addition, a number of calculations were made from the measurements; these are presented in T2able 9. The average

appears related to the surface area of Barrett’s epithelium present. However,

and

Foci of carcinoma appear within fields of Bar-

and adjacent to areas of dysplasia, supporting

dysplasia-carcinoma

of carcitlonla

the length of

HUM

Copyright CQ1992 by W.B. Saunders Company

;I number of’ studies have documented the association of’ esophageal columnar metaplasia (Barrett’s esophagus 1with malignancy. I-” It is now generally agreed that patients with Barrett’s esophagus are at a significantly increased risk of developing carcinoma. In attempts to identify al-risk mucosa, a few studies have at tempted to determine specific histopathologk changes associated with malignancy. ‘N?Although some studies have generated representations of the mucosal surfaces 01 esophal:ectomies for Barrett’s esophagus with dysplasia/carc,inonia,“.“’ no study has systematically evaluated anatomic :tn(i spatial relationships within Barrett’s epithelia showing dysplasia and early carcinoma by means of direc.1 measurements from specimens. This information potentially has important ramifications both in terms of improving the surveillance of patients with Barrett’s esophagus and of understanding the pathophysioloCgy of the development of malignancy in Barrett’s esophagus. To assess this question, we reviewed seven esophagectomv specimens resected for high-grade dysplasia

_____---

479

HUMAN PATHOLOGY

TABLE Casr No.

sex

Aw

WI

Volume 23, No. 5 (May 1992)

1. Endoscopic Correlation

RESULTS Clinical

and endoscopic

1. Two of the patients

Kicqs\ Kewlr~

Endowq~ic I.esion

length of Barrett’s epithelium per specimen (cm) was deter-mined by the sum of the lengths of Barrett’s epithelium perstrip divided by the number of strips per case. The total surface area of Barrett’s epitheliurn per case (cm’) was determined h) the product of the average length of Barrett’s epithelium, the number of strips per case, and 0.3 (the approximate thickness in centimeters of embedded tissue). Similar calculations were made for surface area showing any degree of dysplasia or carcinoma, high-grade dysplasia and carcinoma, and carcinoma. In addition, for each strip the histologic pattern at intervals of 10% of the length of Barrett’s epithelium was recorded. The percentages of various histologic patterns present at each point from all seven esophagectomy specimens. irrespective of length of Barrett’s mucosa, is presented graphically in Fig 2. Location 1 indicates the histology at the proximal-most extent of Barrett’s epithelium; the remaining points indicate the histology at equidistant increasingly distal points, including the gastric mucosaljunction (location 11). Areas showing mucosal ulceration, either primary or secondary to healing biopsy site. were excluded from the calculations.

Table

and Patient Data

is indicated in (nos. 1 and 2) have been

information

reported elsewhere.” Of the seven patients, tive were men and two were women. The average age at esoph-

agectomy was 7 1 years. Endoscopic abnormalities ranged from no lesion identified (four cases) to either white plaques or mucosal irregularities. There was good correlation between preoperative biopsy diagnosis and esophagectomy diagnosis in five cases. The remaining two cases showed high-grade dysplasia on biopsy and proved to have intramucosal carcinoma in the esophagectomy specimen. Measurements from esophagectomy specimens are represented in Table 2. The average length of Barrett’s mucosa measured from tissue sections ranged from 1.!I to 7.3 cm. The total surface areas occupied by Barrett’s mucosa as well as various other histologic patterns are also indicated. The amount of surface area occupied by any degree of dysplasia or carcinoma appears to be associated with the amount of Barrett’s epithelium. For example, cases with the greatest or least amount of Harrett’s mucosa (cases no. 6 and 7, respectively) also tended to show the greatest or least amount of any degree of dysplasia or carcinoma. There does not appear to he an association, however, between the amount of any degree

FIGURE

1.

ogy maps.

I

squamousmueos.

m

Bsrrstrr,

-

Barr.10,

low gradwlndefinite dysplsasla

Barr&s,

hlgh grade dyrplask

m

no dyrphds

Csrclnomr

480

Specimen histol-

DYSPLASIA

IN BARRETT’S

ESOPHAGUS

(McArdle et al)

of dysplasia

or carcinoma and the amount of either highgrade dysplasia and carcinoma or carcinoma alone. Ikphagectomy specimens are presented graphicallv in Fig 1 Of the six cases showing foci of carcinoma, fcju;. are seen adjacent to areas of high-grade dysplasia. The remaining two show foci of carcinoma adjacent to areas of low-bT”dc or indefinite dysplasia. No case shows carcinoma arising in an area without some degree of dysplasia. Foci of carcinoma generally appear to be loc.,ited within a field of dysplastic Barrett’s epithelium as opposed to occurring at a proximal or distal margin. In addition. in strips showing carcinoma within a field of dysplasia, foci of carcinoma are generally separated from low-grade dysplasia by an intervening focus of highgrade dysplasia. The analysis of the location of histologic patterns tar combined strips for all cases, irrespective of length, i\ presented graphically in Fig 2. The location of carcinoma a’ppears fairly uniform throughout the distribution of Barrett’s epithelium, being present at all loc.ations with a slightly higher incidence within central portions. The presence and amount of carcinoma also mirror the presence and amount of high-grade dysplasia, being slightly more common in central portions of the esophagectonty specimens. This relationship does not hold true for low-&T&de or indefinite dysplasias that have ;I higher incidence at proximal and distal portions of the specknens. The proportion of locations showing no dysplasia appears fairly uniform throughout the length of Barretl:‘s q)ithelium.

FIGURE 2. Histologic appearance of esoph’clgeal epithelium at 10% intervals of the length of Barrett’s mucosa.

Ilormalities. ‘The correlation betweet) preoperative and postoperative diagnosis was generally good. However, when errors were made, they were the result of preoperative biopsies underestimating the severity of disease. There appears to be an association between the amount of Barrett’s mucosa and the amount of any degree of dysplasia or carcinoma in our specimens. Howc\‘er. the surface area of high-grade dysplasia and carc,inoma, or carcinoma alone, does not appear to he associated with the surface area of dysplastic Barrett’s epithelium. Whether this represents a true causal relationship or is an artifact of patient sektion ancl small number requires further investigation. The majority of foci of carcinoma appear adjacent to foci of high-grade dysplasia. ,411 foci of carc~inoma appear adjacent to some degree of dysplasia. In addition, all foci of carcinoma occur within a field of Barrett’s epithelium as opposed to occurring at a proximal ordistal end. These observations support the common conception of a dysplasia-carcinoma sequence. The location of carcinoma appears, fairly uniform throughout the length of Barrett’s epithelium, and generally mirrors the location of high-grade dysplasia. Foci of carcinoma do not appear clustered either proximally or distally, suggesting that differences in the local environment. such as inflammation secondary to acid reflux, may not play a primary role in malignant transformation. hgain, whether this represents a true phenornenotl or selection artifact requires additional studv.

DISCUSSION (&es reviewed in this series include only esophagectomy speckens showing high-grade dysplasia or t,arly (intram~~cosal and suhmucosal) carcinoma. This ~t’as done to minimize the gross distortion of’ the specimens generally present with more deeply invasive lesions. An additional feature of invasive carcinomas is mucosal extension of the tumor, altering the native histologic relationships. Preoperative endoscopic findings from our cases t onfirm previous reports of the difficulty in the diagnosis (bf carcinoma from endoscopic appearances.” Fifty per( t‘nt of our cases that showed intramucosal carcinoma in the esophagectomy specimen showed no mucosal ah-

TABLE

2.

Surface

Area

of Barrett’s

Esophagus

and Dysplasia

I

ti

IO.ti

L) r

ti. I ?.li li.5 .5.7 7.3 1.!J

I0.W

7.”

11.M I!)..:, 17.1 24.O!I t?.ti!l

I 1.-lti I ti.2ti 15.6 L”J.tj‘,. 5.28

.4 .-I Ci 7 Ahhreviatiotw

I S;D/HGD/(:A.

any degree

of dysplasia

and/or

3.7 L’ 0.75 I .?!, x.x!2 I.2 Yl.55 ?.til

Ill.17

or carcinoma;

481

HGD/CA,

Carcinoma

high-grade

dysplasia

or carcinoma;

0 0.39 O.!J!) ?.85 0.7.5 0.“7 _ 1 .ti5 (:I\, carunwna

HUMAN PATHOLOGY

Volume 23, No. 5 (May 1992)

Our findings support the current clinical standard of practice for endoscopic monitoring of patients with Barrett’s esophagus. In those patients, biopsies are taken of any visible lesions and, at 2-cm intervals, two to four at each level, proximally from grossly normal gastric mucosa, through the length of metaplastic epithelium, to grossly normal esophageal squamous mucosa. High-grade dysplasias and “early” carcinomas do not show a characteristic endoscopic appearance, do not seem to occur with increased likelihood either proximally or distally, and do not seem to be associated with length of Barrett’s esophagus. Therefore, thorough sampling of the entire length is necessary for early detection in both long and shorter segments of Barrett’s esophagus.

REFERENCES 1. SpechlerSJ, 315:362-371.

Goyal

RK: Barrett’s

esophagus.

N hngl J Med

1%6

482

2. Spechler SJ, Robhins AH. Ruhins HB, et al: kirnocarcinomd Barrett’s esophagus: An overrated risk? Gastroenterolob~ 87:927933, 1984 3. Cameron AJ, Ott BJ. Payne WS: The incidence of adenocarcinema in columnar-lined (Barrett’s) esophagus. N Engl J Med 3 13: 857-859, 1985 4. Haggitt RC, Reid BJ, Rahinovitch PS, et al: Barrett’s esophagus, correlation between mucin histochemisrry, flow cytometry and histologic diagnosis for predicting increased cancer risk. Am J Pathol 131:53-61, 1988 5. Rubio CA, Riddell RH: Mucosal cysts in Barrett’s mucow with dysplasia. APMIS 97:?97-301, 1989 6. Rubio CA, Riddell RH: Musculo-fibroua anomaly in Barrett’s mucosa with dysplasia. Am J Surg Pathol 12:885-889, 1988 7. Rubio CA. Riddell RH: Atypical mitoses in dyspfasias of the Barrett’s mucosa. Pathol Res Pratt 184: l-5. 1989 8. Hamilton SR, Smith RR: The relationship between columna epithelial dysplasia and invasive adenocarrinoma arising in Harrett’s esophabws. Am J Clin Pathol 87:301-312, 1987 9. Reid BJ, Weinstein WM, Lewin KJ, et al: Endoscopic hiopsy can detect high-grade dysplasia or ear-ly adenocarc-inoma in Barrett‘5 esophabws without grossly recognizable neoplasric lesions. (;astrc)enterology 94:8 l-90, 1988 10. Rabinovitch PS, Reid BJ, Haggitt R