The diagnosis of low-grade dysplasia in Barrett’s esophagus and its implications for disease progression

THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2000 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.

Vol. 95, No. 12, 2000 ISSN 0002-9270/00/$20.00 PII S0002-9270(00)02141-9

ORIGINAL CONTRIBUTIONS

The Diagnosis of Low-Grade Dysplasia in Barrett’s Esophagus and Its Implications for Disease Progression Marek Skacel, M.D., Robert E. Petras, M.D., Terry L. Gramlich, M.D., Jessica E. Sigel, M.D., Joel E. Richter, M.D., and John R. Goldblum, M.D. Departments of Anatomic Pathology and Gastroenterology, The Cleveland Clinic Foundation, Cleveland, Ohio

OBJECTIVE: The reported risk of progression from low-grade dysplasia (LGD) to high-grade dysplasia (HGD) or carcinoma (CA) in Barrett’s esophagus varies. However, the validity of a diagnosis of LGD may be questioned because of interobserver variability. METHODS: A search of the Cleveland Clinic Foundation surgical pathology files between 1986 and 1997 yielded biopsy specimens from 43 patients with Barrett’s esophagus diagnosed and coded as LGD. Patients with concurrent or prior diagnoses of HGD or carcinoma were excluded. The LGD cases were randomized and blindly reviewed by three gastrointestinal (GI) pathologists along with cases originally diagnosed as Barrett’s esophagus without dysplasia (ND; n ⫽ 28), indefinite for dysplasia (IND; n ⫽ 14), or HGD (n ⫽ 15). Each pathologist classified every biopsy specimen as ND, IND, LGD, or HGD, and interobserver agreements were determined by kappa statistics (K). Follow-up data were available on 25 patients originally diagnosed with LGD. Progression was defined as a subsequent diagnosis of HGD or CA on esophageal biopsy or resection specimens. RESULTS: Agreement between two GI pathologists for a diagnosis of LGD was fair (K ⫽ 0.28) and poor (K ⫽ 0.21 and ⫺0.04). Individual GI pathologists agreed with the original diagnosis of LGD in 70%, 56%, and 16% of cases. The 25 patients with follow-up included 21 men and four women (mean age, 67 yr) with a mean follow-up of 26 months (range: 2– 84 months). Seven patients (28%) with follow-up developed HGD (five patients) or CA (two patients), 2– 43 months (median: 11 months) after a diagnosis of LGD. The individual GI pathologists’ diagnosis did not correlate with progression. However, when at least two GI pathologists agreed on LGD, there was a significant association with progression (seven of 17 patients, 41%, p ⫽ 0.04). When all three GI pathologists agreed on a diagnosis of LGD, four of five patients progressed (p ⫽ 0.012). In contrast, of the eight patients with follow-up and no agreement among GI pathologists for a diagnosis of LGD, none progressed.

CONCLUSIONS: A high degree of interobserver variability is seen in the histological diagnosis of Barrett’s esophagus– related LGD. Although the number of observations is low, a consensus diagnosis of LGD among GI pathologists suggests an increased risk of progression from LGD to HGD or carcinoma. (Am J Gastroenterol 2000;95:3383–3387. © 2000 by Am. Coll. of Gastroenterology)

INTRODUCTION Barrett’s esophagus, a condition in which the squamous mucosa of the esophagus is replaced by metaplastic columnar mucosa, has been linked to chronic gastroesophageal reflux disease. The association between Barrett’s esophagus and esophageal adenocarcinoma has been well known for at least 30 yr (1–3). In particular, specialized columnar epithelium (intestinal metaplasia), characterized by the presence of acid mucin– containing goblet cells, is the major risk factor for the development of dysplasia and adenocarcinoma (4). The recognition of specialized columnar epithelium in the tubular esophagus results in the placement of patients into an endoscopic surveillance program, the goal of which is to prevent progression to carcinoma or to detect carcinoma in an early and potentially curable phase (5–7). Dysplasia is the first step in the neoplastic process, and it has been regularly recognized in esophageal resection specimens both adjacent to and distant from Barrett’s-associated adenocarcinomas (8, 9). There is mounting evidence supporting high-grade dysplasia (HGD) as a marker of high cancer risk, in that 40 –70% of esophagi resected for HGD harbor an unsuspected adenocarcinoma (10 –13). However, much less is known about the natural history of low-grade dysplasia (LGD). One of the major detriments to determining the clinical significance of LGD is the high degree of interobserver variability in establishing this diagnosis (14, 15) and the variable protocols by which patients with LGD are followed. Thus, we evaluated a group of patients with Barrett’s esophagus in whom a diagnosis of LGD was made at our institution over a 12-yr period.

3384

Skacel et al.

MATERIALS AND METHODS Case Selection The files of the Department of Anatomic Pathology at the Cleveland Clinic Foundation were searched for cases of LGD arising in Barrett’s esophagus between 1986 and 1997. Barrett’s esophagus was defined as the presence of specialized columnar epithelium characterized by the presence of acid mucin– containing goblet cells arising anywhere in the tubular esophagus, regardless of length. Patients were excluded if they had prior or concurrent documentation of HGD or carcinoma. From this search, we identified biopsy specimens from 43 patients meeting the aforementioned criteria such that the specimens reviewed were that patient’s first diagnosis with LGD. In addition, biopsy specimens originally diagnosed as Barrett’s esophagus without dysplasia (ND; n ⫽ 28), Barrett’s esophagus with changes indefinite for dysplasia (ND; n ⫽ 14), and Barrett’s esophagus with HGD (n ⫽ 15) were also retrieved. These additional biopsies were added to the set solely for the purpose of randomization. All slides were then reviewed blindly by three experienced gastrointestinal (GI) pathologists (R.E.P., T.L.G., J.R.G., with 18, 11, and 7 yr of experience, respectively), and each was classified as ND, IND, LGD, or HGD using previously published criteria modified from the Inflammatory Bowel Disease Morphology Study Group (16). All biopsy specimens had been fixed in formalin (n ⫽ 7) or Hollande’s solution (n ⫽ 93). Interobserver Agreements Interobserver agreements between individual GI pathologists were determined by kappa statistics (K) (17, 18). Kappa coefficients are calculated taking into account both the observed agreement minus the extent of agreement that occurs by chance alone. Coefficients ⱕ0.20, 0.21– 0.40, 0.41– 0.60, 0.61– 0.80, and ⬎0.80 were designated as poor, fair, moderate, good, and very good interobserver agreement, respectively (19). Kappa values were calculated comparing the ratings of each individual GI pathologist against those of each of the other two GI pathologists. Clinical Follow-Up Follow-up data were available for 25 patients originally diagnosed with LGD. The gastroenterologists at our institution follow the endoscopic biopsy protocol proposed by Reid et al. (20), with four-quadrant biopsies taken using jumbo forceps at intervals of ⱕ2 cm throughout the length of the Barrett’s segment, with additional biopsies of any endoscopic lesions (13). Endoscopy with biopsy is repeated at 3– 6-month intervals until two consecutive negative interpretations are rendered, at which point the patient returns to yearly surveillance. If IND or LGD persists, surveillance at 3– 6-month intervals continues until dysplasia progresses. For the purposes of this study, progression was defined as a subsequent diagnosis of HGD or carcinoma on either biopsy or resection specimens. Progression was correlated with each of the following: original diagnosis of LGD, diagnosis of LGD established by each individual GI pathologist, and a

AJG – Vol. 95, No. 12, 2000

Table 1. Interobserver Agreement Among Pathologists for the Diagnosis of Low-Grade Dysplasia

All GI-1/GI-2 GI-1/GI-3 GI-2/GI-3

Observed Agreement (%)

Chance Agreement (%)

Kappa

Rating

45.7 51.2 55.8 30.2

34.5 32.2 44.7 33.2

0.17 0.28 0.20 ⫺0.04

Poor Fair Poor Poor

GI-1, 2, 3 ⫽ diagnosis of each individual gastrointestinal pathologist.

consensus diagnosis of LGD based on agreement of two out of three and of all three GI pathologists, respectively. The association between LGD diagnosis and subsequent progression to HGD or carcinoma was assessed using the ␹2 test.

RESULTS Interobserver Agreements A total of 43 cases originally diagnosed by 10 different surgical pathologists at our institution as LGD in Barrett’s esophagus met our inclusion criteria. The individual GI pathologists agreed with the original diagnosis of LGD in 70%, 56%, and 16% of cases, respectively. Agreement among the three combinations of two GI pathologists for a diagnosis of LGD was fair (K ⫽ 0.28) and poor (K ⫽ 0.20 and ⫺0.04; Table 1). Among the original diagnoses, there were 17 cases in which either of two GI pathologists involved in this study established the initial diagnosis of LGD (14 cases) or was involved as a consultant (three cases). Among these 17 cases, the GI pathologists agreed with their own original diagnosis in 65% of cases. Clinical Outcome (Table 2) The cohort of 25 patients with an original diagnosis of LGD with follow-up included 21 men and four women, with a mean age of 67 yr (range: 30 – 81 yr). In 13 patients, the diagnosis of LGD was made at the same time as the first diagnosis of Barrett’s esophagus (prevalent LGD). In the remaining 12 patients, LGD was diagnosed at a mean of 17 months (range: 3–70 months) after the initial diagnosis of Barrett’s esophagus (incident LGD). Mean length of follow-up was 26 months (range: 2– 84 months). Seven patients (28%) progressed to HGD (five patients) or carcinoma (two patients), 2– 43 months (median: 11 months) after the original diagnosis of LGD. Thus, of the 25 patients with follow-up information, five of 13 (38%) with prevalent LGD progressed, compared to two of 12 (17%) with incident LGD, a difference that was not statistically significant (p ⫽ 0.5). The five patients who progressed to HGD were diagnosed on follow-up biopsy 2, 2, 3, 6, and 7 months after the LGD diagnosis. Two of the five patients with a biopsy diagnosis of HGD underwent esophagectomy, both revealing HGD. Two patients with HGD chose to be followed with periodic endoscopic surveillance. Their follow-up biopsies have shown HGD up to 12 and 17 months after their first biopsy diagnosis of HGD, respectively. The last patient with a

AJG – December, 2000

Diagnosis of Dysplasia in Barrett’s Esophagus

3385

Table 2. Summary of 25 Patients With an Initial Diagnosis of Low-Grade Dysplasia, With Follow-Up Data Case 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Age (yr)

Sex

Time of LGD Dx.

Number of Agreements*

Surveillance Outcome

Follow-Up (months)

68 62 59 79 30 62 56 77 80 54 83 71 70 77 49 69 81 73 67 59 69 78 77 61 65

M M M M M M M F M M F M M M M M F M M M M M M F M

Incident Prevalent Incident Prevalent Prevalent Prevalent Prevalent Prevalent Prevalent Prevalent Incident Prevalent Incident Incident Prevalent Incident Incident Prevalent Prevalent Incident Incident Incident Prevalent Incident Prevalent

2 3 3 3 3 2 2 2 1 2 2 1 0 0 2 0 2 1 2 1 2 0 3 2 2

HGD CA CA HGD HGD HGD HGD No dysplasia No dysplasia No dysplasia No dysplasia Persistent LGD No dysplasia No dysplasia No dysplasia No dysplasia No dysplasia No dysplasia No dysplasia No dysplasia No dysplasia Indefinite No dysplasia Persistent LGD No dysplasia

2 43 16 6 3 2 7 84 11 43 4 6 42 54 60 33 19 60 3 40 30 8 20 27 18

* Number of GI pathologists who agreed with diagnosis of low-grade dysplasia. LGD ⫽ low-grade dysplasia; HGD ⫽ high-grade dysplasia; CA ⫽ carcinoma; M ⫽ male; F ⫽ female; LGD Dx. ⫽ original diagnosis of low-grade dysplasia.

biopsy diagnosis of HGD was determined to be a poor surgical candidate. Subsequent biopsies have repeatedly shown HGD up to 23 and 17 months after the initial diagnoses of LGD and HGD, respectively. Two patients had biopsy diagnoses of intramucosal carcinoma at 16 and 43 months after the initial diagnosis of LGD. Subsequent workup including esophageal ultrasonography and CT scan revealed that one of the patients had a high-stage tumor (T4N1M0). The second patient underwent esophagectomy, and an intramucosal carcinoma was confirmed in the resection specimen. Among the patients followed who did not progress (n ⫽ 18), surveillance biopsy specimens showed no dysplasia in 15 patients at a mean of 44.4 months (range: 8 – 84 months) and persistent LGD in two patients at 6 and 27 months. One patient had changes indefinite for dysplasia at follow-up 8 months after the original diagnosis of LGD. Correlation Between Pathological Agreement and Clinical Outcome Neither the original diagnosis nor that of the individual GI pathologists correlated with progression (␹2 test). However, when at least two GI pathologists agreed on a diagnosis of LGD, there was a significant association with progression (seven of 17 patients, 41%, p ⫽ 0.04; Fig. 1). Among the cases in which all three GI pathologists agreed on a diagnosis of LGD, four of five patients progressed (80%, p ⫽ 0.012). Of the eight patients with follow-up and no agreement among GI pathologists for a diagnosis of LGD, none progressed. Areas of

LGD from biopsy specimens with a different number of agreements on the diagnosis are depicted in Figures 2–5.

DISCUSSION Barrett’s esophagus is a premalignant condition. Once this diagnosis is established, patients are followed by periodic endoscopy with biopsy, the goal being to detect carcinoma in a curable stage or before it has a chance to develop. As of now, the histopathological identification of dysplasia

Figure 1. Progression to high-grade dysplasia/carcinoma in groups of patients with one, two, or three agreements on the diagnosis of low-grade dysplasia among GI pathologists (Kaplan-Meier curves).

3386

Skacel et al.

AJG – Vol. 95, No. 12, 2000

Figure 2. Focus of low-grade dysplasia in Barrett’s esophagus in a case in which all three GI pathologists agreed on the diagnosis. This patient progressed to invasive adenocarcinoma 16 months after this biopsy (case 3).

Figure 4. Low-grade dysplasia in Barrett’s esophagus in a case in which two of three GI pathologists agreed on the diagnosis. Highgrade dysplasia was found in a biopsy specimen 7 months after this diagnosis of low-grade dysplasia (case 7).

remains the best marker of increased cancer risk. Although morphological criteria for dysplasia have been established (8, 16), the risk of progression in patients who have a biopsy diagnosis of LGD is difficult to determine, in part because of the high degree of interobserver variability known to exist in establishing this diagnosis (15). There are very limited data published in the literature, mostly only in abstract form, correlating biopsy diagnosis of LGD with clinical outcome. Preliminary data reported by Weston et al. found that five of 62 patients (6%) with a diagnosis of Barrett’s esophagus–related LGD progressed to carcinoma (four patients) or HGD (one patient) at 3, 14, 16, 17, and 41 months after the diagnosis of LGD, respectively (21). Nine patients (16.4%) had persistent LGD during a mean follow-up of 29.2 months. The authors reported that 41 patients (74.5%) eventually “regressed” to no dysplasia during a mean followup of 22.1 months. The authors, however, did not report whether the diagnosis of LGD was validated by more than one

pathologist. In the study by Sontag et al. (also currently published only in abstract form), 33 of 848 patients (3.9%) with LGD progressed to HGD and 18 patients (2.1%) progressed to carcinoma (22). The vast majority of patients (94%) continued to have LGD documented on subsequent surveillance biopsies. In their study, the diagnosis of LGD was established by a single pathologist over an unknown period of time, and interobserver variability in this diagnostic category was not specifically addressed. The conclusions drawn from these cited studies could be questioned because of the high degree of interobserver variation known to exist for the diagnosis of LGD. Reid et al. found that the percentage of cases agreed upon by experienced GI pathologists was much lower at the low end of the histological spectrum (no dysplasia vs indefinite for dysplasia vs LGD) when compared to HGD and carcinoma (15). When grouping negative/indefinite for dysplasia/LGD, there was 85% and 87% agreement among pathologists on successive reviews when separating this category from HGD/carcinoma. However, there was only 58% and 61% agreement when

Figure 3. Focus of low-grade dysplasia in Barrett’s esophagus in a case in which all three GI pathologists agreed on the diagnosis. High-grade dysplasia was found in a biopsy specimen taken 6 months after this diagnosis of low-grade dysplasia (case 4).

Figure 5. Low-grade dysplasia in Barrett’s esophagus in a case in which two of three GI pathologists agreed on the diagnosis. This patient did not progress to high-grade dysplasia or carcinoma within a follow-up period of 43 months (case 10; arrow marks the surface).

AJG – December, 2000

the authors attempted to separate negative from indefinite for dysplasia/LGD. No attempt was made to distinguish indefinite from LGD. Importantly, this study did not account for agreement that could occur by chance alone. On the other hand, Montgomery et al. evaluated this problem using kappa statistics, similar to our study (23). According to their preliminary data, interobserver agreement in establishing the diagnosis of LGD was only fair (K ⫽ 0.32), supporting the results of our study. There are even fewer published studies evaluating the risk of progression from Barrett’s-related LGD that take into account interobserver variability. Preliminary data reported by Montgomery et al. (24) suggest that there is a definite risk of progression from LGD to HGD or carcinoma in those cases in which substantial agreement exists for the diagnosis of LGD among experienced GI pathologists. In their study, three of 19 cases (16%) with a “consensus diagnosis” of LGD (defined as agreement in ⱖ13 of 24 observations made by 12 pathologists on two separate occasions) progressed to either HGD or carcinoma. However, the nature of an interinstitutional study had limitations, including variable surveillance protocols and follow-up information. Nevertheless, their conclusions on the risk of progression from LGD to HGD or carcinoma are strikingly similar to ours. Similar to the studies of Reid et al. (15) and Montgomery et al. (23), we confirmed substantial interobserver variability in establishing the diagnosis of LGD, even among experienced GI pathologists. However, when agreement existed between two or among all three GI pathologists for the diagnosis of LGD, we found a significant correlation with histological progression, with seven of 25 patients (28%) progressing to HGD (five patients) or carcinoma (two patients). Although the number of observations is low, it seems that the risk of progression of LGD is higher than the literature suggests, especially when the diagnosis is confirmed by at least two experienced GI pathologists. We conclude that a high degree of interobserver variation is seen in the histological diagnosis of LGD, and this must be taken into account when assessing the risk of progression. A consensus diagnosis of LGD among GI pathologists suggests an increased risk of progression from LGD to HGD or carcinoma. Thus, it is advisable that biopsy specimens with suspected LGD be reviewed by at least two pathologists experienced with these specimens. Reprint requests and correspondence: John R. Goldblum, M.D., Cleveland Clinic Foundation, Department of Anatomic Pathology L25, Cleveland, OH 44195. Received May 12, 2000; accepted Aug. 15, 2000.

Diagnosis of Dysplasia in Barrett’s Esophagus

3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

21. 22.

23.

REFERENCES 1. Adler R. The lower esophagus lined by columnar epithelium: Its association with hiatal hernia, ulcer, stricture, and tumor. J Thorac Cardiovasc Surg 1963;45:13–32. 2. Hawe A, Payne WS, Weiland LH. Adenocarcinoma in the

24.

3387

columnar epithelial lined lower (Barrett’s) oesophagus. Thorax 1973;28:511– 4. Haggitt RC, Tryzelaar J, Ellis FH, et al. Adenocarcinoma complicating columnar epithelium-lined (Barrett’s) esophagus. Am J Clin Pathol 1978;70:1–5. Reid BJ, Weinstein WM. Barrett’s esophagus and adenocarcinoma. Annu Rev Med 1987;38:477–92. Hameeteman W, Tytgat GN, Houthoff HJ. Barrett’s esophagus: Development of dysplasia and adenocarcinoma. Gastroenterology 1989;96:1249 –56. Robertson CS, Mayberry JF, Nicholson DA. Value of endoscopic surveillance in the detection of neoplastic change in Barrett’s esophagus. Br J Surg 1988;75:760 –3. Reid BJ, Blount PL, Rubin CE. Predictors of progression to malignancy in Barrett’s esophagus: Endoscopic, histologic and flow cytometric follow-up of a cohort. Gastroenterology 1992;102:1212–9. Haggitt RC. Barrett’s esophagus, dysplasia, and carcinoma. Hum Pathol 1994;25:982–93. Spechler SJ, Goyal RK. Barrett’s esophagus. N Engl J Med 1986;315:362–71. Rice TW, Falk GW, Achkar E, et al. Surgical management of high-grade dysplasia in Barrett’s esophagus. Am J Gastroenterol 1993;88:1832– 6. Hamilton SR, Smith RR. The relationship between columnar epithelial dysplasia and invasive adenocarcinoma arising in Barrett’s esophagus. Am J Clin Pathol 1987;87:301–12. Lee RG. Dysplasia in Barrett’s esophagus. Am J Surg Pathol 1985;9:845–52. Falk GW, Rice TW, Goldblum JR, et al. Jumbo biopsy forceps protocol still misses unsuspected cancer in Barrett’s esophagus with high-grade dysplasia. Gastrointest Endosc 1998;49:170 – 6. Alikhan M, Khan A, Rahmani E, et al. Variable pathologic interpretation of columnar lined esophagus by general pathologists in community practice. Gastrointest Endosc 1999;50:23– 6. Reid BJ, Haggitt RC, Rubin CE, et al. Observer variation in the diagnosis of dysplasia in Barrett’s esophagus. Hum Pathol 1988;19:166 –78. Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory bowel disease: Standardized classification with provisional clinical implications. Hum Pathol 1983;14:931– 68. Schouten HJA. Statistical measurement of interobserver agreement, 1st ed. Utrecht: Elinkwijk, 1985. Fleiss JL. Statistical methods for rates and proportion, 2nd ed. New York: John Wiley & Sons, 1981. Altman DG. Practical statistics for medical research, 1st ed. New York: Chapman and Hall, 1991. Reid BJ, Weinstein WM, Lewin KJ. Endoscopic biopsies diagnose high-grade dysplasia or early operable adenocarcinoma in Barrett’s esophagus without grossly recognizable neoplastic lesions. Gastroenterology 1988;94:81–90. Weston AP, Sharma P, Topalowski M, et al. Low-grade dysplasia in Barrett’s esophagus: Variable fate during long-term prospective follow-up. Gastroenterology 1999;116:G1349. Sontag SJ, Schnell T, Chejfec G, et al. Barrett’s, low grade dysplasia and fear; yearly endoscopy is not justified: Surveillance every 2–3 years detects all cancers early. Gastroenterology 1999;116:G1381. Montgomery E, Bronner E, Goldblum J, et al. Diagnostic reproducibility of dysplasia in Barrett’s esophagus: A reaffirmation. Mod Pathol 2000;13:486A. Montgomery E, Bronner E, Goldblum J, et al. Dysplasia as a predictive marker for cancer in Barrett’s esophagus: A follow-up study based on consensus diagnoses in 151 cases from 10 institutions. Mod Pathol 2000;13:487A.