GASTROENTEROLOGY 1996;110:694–704
Sialosyl-Tn Antigen Is Prevalent and Precedes Dysplasia in Ulcerative Colitis: A Retrospective Case-Control Study STEVEN H. ITZKOWITZ,* ERIC YOUNG,* DEREK DUBOIS,* NOAM HARPAZ,‡ CAROL BODIAN,§ ANLI CHEN,* and DAVID B. SACHAR* *Gastrointestinal Research Laboratory, Departments of *Medicine, ‡Pathology, and §Biomathematical Sciences, Mount Sinai School of Medicine, City University of New York, New York, New York
Background & Aims: Expression of the carbohydrateassociated sialosyl-Tn (STn) antigen has been correlated with carcinogenesis in sporadic colon cancer. This retrospective study analyzed surveillance colonoscopy biopsy specimens to determine whether STn antigen could serve as a marker for neoplasia risk in patients with long-standing ulcerative colitis. Methods: Eleven patients who developed dysplasia or cancer and who had undergone at least three surveillance colonoscopies were matched with 11 controls who had not developed neoplasia. Sections from 969 available surveillance biopsy specimens were stained immunohistochemically with monoclonal antibody TKH2 and were interpreted blindly. Results: Compared with controls, patients expressed STn antigen more frequently in their biopsy specimens (44% patient biopsy specimens vs. 11% control biopsy specimens). Patients were also more likely to undergo colonoscopies with two or more colonic segments expressing STn antigen (73% vs. 22%), repeated STn antigen expression in two or more consecutive colonoscopies (91% vs. 27%), and STn antigen expression proximal to the hepatic flexure (64% vs. 0%). STn antigen appearance preceded the detection of neoplasia by 7 or fewer years, and its preferential expression in patients was not confounded by the degree of active inflammation. Conclusions: Repeated STn antigen expression in nondysplastic colonic mucosa may presage the development of neoplasia in long-standing ulcerative colitis and may be a useful adjunct to dysplasia during colonoscopic surveillance.
I
t is well established that patients with long-standing ulcerative colitis (UC), particularly those with extensive colitis of at least 8–10 years’ duration are at increased risk for developing colorectal neoplasms. Currently, the principal method of surveillance for the development of neoplasia is periodic colonoscopy with examination of multiple biopsy specimens for histological evidence of dysplasia and/or carcinoma. Although dysplasia is still the best available premalignant marker, some patients develop cancer without prior evidence of / 5e0a$$0004
02-09-96 13:22:16
gasas
dysplasia. Moreover, there can be difficulties with interpretation of dysplasia (both intraobserver and interobserver variation), especially in the presence of active inflammation. Therefore, other markers to complement dysplasia have been sought to facilitate clinical decision making. The mucin-associated sialosyl-Tn (STn) antigen is a promising marker in this regard. STn antigen expression, as detected immunohistochemically with monoclonal antibody TKH2, correlates well with the adenoma-carcinoma progression in sporadic colon cancer; it is rarely expressed in normal colonic mucosa or hyperplastic polyps but becomes increasingly expressed in adenomatous polyps and adenocarcinomas.1 – 4 Moreover, in colon cancer tissues, STn antigen expression has been associated with a poor prognosis, independent of tumor stage, DNA ploidy status, or histological type.5 In a previous study of patients with UC, we observed that STn antigen expression was increased in dysplastic and cancerous colonic mucosa.6 In a preliminary study, we reported that STn antigen expression in nondysplastic mucosa may be a marker of cancer risk in patients with UC.7 In that two-part study, STn antigen was observed in nondysplastic mucosa before the development of cancer in colonoscopic surveillance biopsy specimens from the segment of colon that developed cancer. However, this chronological analysis was performed only on the segment that developed cancer and not on surveillance biopsy specimens from other areas of the colon. In the second part of that study, the regional expression of STn antigen was evaluated, indicating that STn antigen was expressed not only in the area of the colon where carcinoma was present but also in nondysplastic mucosa remote from the cancer at the time of colectomy. However, the regional distribution of STn antigen before colectomy was not studied, and the regional expression of STn antiAbbreviation used in this paper: STn, sialosyl-Tn. 䉷 1996 by the American Gastroenterological Association 0016-5085/96/$3.00
WBS-Gastro
March 1996
STn AS A RISK FACTOR FOR NEOPLASIA IN UC 695
gen in colectomies without neoplasia was studied in only a few control patients. The purpose of the present study was to analyze the expression of STn antigen throughout the colon in a chronological fashion to determine both the timing and the regional distribution of STn antigen expression relative to neoplastic change. This aim was accomplished by a retrospective case-control analysis of STn antigen expression in surveillance colonoscopic biopsy specimens of patients with long-standing UC who either had or had not (controls) developed colon cancer.
Materials and Methods Clinical Material Among individuals with long-standing UC undergoing colonoscopic surveillance at the Mount Sinai School of Medicine, we selected 11 patients who developed dysplasia or cancer and who had undergone at least three prior surveillance colonoscopies and 11 controls who had not developed dysplasia or carcinoma during surveillance and who, therefore, have not undergone colectomy. Of the 11 patients, 10 underwent colectomy for dysplasia or carcinoma; 1 patient (no. 5) who developed dysplasia has not yet agreed to undergo colectomy. Except for 4 patients, the patients used herein have not been reported previously. For 3 of the patients, we previously reported results of their colectomy specimen,7 but because the focus of the present study is on the surveillance biopsy specimens preceding colectomy, there is no overlap of material. In the fourth patient, we reported on chronological STn antigen expression in surveillance biopsy specimens, but only in the segment that developed neoplasia. Controls were selected to match each corresponding case as closely as possible with respect to year of birth, age at onset of UC, duration of UC, extent of UC, number of years under surveillance, and total numbers of surveillance colonoscopy examinations as well as colonoscopic biopsy samples. Given the choice of controls, we tended to select those with somewhat younger disease onsets, longer disease durations, and more years of surveillance. None of the controls was used in our previous study. All tissues analyzed were surveillance colonoscopy biopsy specimens; colectomy specimens of the patients were not studied because the controls have not undergone colectomies for comparison. All available colonoscopic surveillance biopsy specimens were collected, yielding a total of 969 biopsy specimens examined. The patients in this study were under the care of several gastroenterologists, but all biopsy specimens were processed in standard fashion by the Department of Pathology, and one pathologist (N.H.) rendered the final histological diagnosis of dysplasia or cancer on all specimens. Owing to the retrospective nature of this study, we were not able to control the precise number of biopsy specimens per colonoscopy, but every segment of the colon was routinely sampled. The retrieval rate of specimens from the pathology archives exceeded 90%.
/ 5e0a$$0004
02-09-96 13:22:16
gasas
Colonic segments were defined as cecum, ascending colon, hepatic flexure, transverse colon, splenic flexure, descending colon, sigmoid colon, and rectum.
Immunohistochemical Staining All specimens were fixed in formalin, embedded in paraffin, and cut into 5-mm sections for immunohistochemical staining. Monoclonal antibody TKH2 (generously provided by Dr. Sen-itiroh Hakomori, Biomembrane Institute, Seattle, WA) was used to detect STn antigen.8 Monoclonal antibody TKH2 reacts with STn antigen that is not Oacetylated.9 – 11 Biotinylated rabbit anti-mouse immunoglobulin (Ig) G / IgA / IgM and streptavidin peroxidase conjugate were obtained from Zymed Laboratories (South San Francisco, CA). A standard immunohistochemical staining method using streptavidin-peroxidase was used, with all steps conducted at room temperature. Sections were deparaffinized and rehydrated with xylene and alcohol and washed in phosphatebuffered saline (PBS). The slides were then successively incubated (with two PBS washings between each step) in the following: 3% hydrogen peroxide in methanol for 10 minutes to reduce endogenous peroxidase; 5% normal rabbit serum in PBS for 20 minutes and then blotted without washing; monoclonal antibody TKH2 1:10 diluted hybridoma supernatant for 60 minutes; biotinylated rabbit antimouse IgG / IgA / IgM (1:100 in PBS) for 20 minutes; streptavidin-peroxidase conjugate (10 mg/mL in PBS) for 30 minutes; and 3,3ⴕ-diaminobenzidine for 10 minutes. The slides were then rinsed in tap water, counterstained with methyl green, dehydrated, and mounted. Staining controls. The TKH2 monoclonal antibody reacts with endothelial cells of the colonic mucosa, providing a useful internal positive control. A negative control slide was included in each staining procedure by substituting isotypematched normal mouse IgG1 for monoclonal antibody TKH2.
Interpretation of Slides Dysplasia was diagnosed by an experienced gastrointestinal pathologist (N.H.), using the standardized classification of the Inflammatory Bowel Disease–Dysplasia Morphology Study Group.12 The degree of inflammation was graded as follows: severe, mucosa with frank erosion or accompanied by ulcer exudate; moderate, neutrophil infiltration (cryptitis or crypt abscesses) of ú50% of the crypts in noneroded mucosa; mild, neutrophil infiltration of õ50% of the crypts in noneroded mucosa, neutrophils present in the lamina propria, or marginating neutrophils present in intramucosal capillaries; and inactive, lamina propria neutrophils rare or absent. STn antigen staining was interpreted by two observers (S.H.I. and E.Y.) without knowledge of clinicopathologic data. The number of crypts in a specimen showing any immunoreactive cells was expressed as a percentage of the total number of crypts in the specimen. Scoring was performed as follows: 0,
WBS-Gastro
696 ITZKOWITZ ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
Table 1. Characteristics of Patients and Controls
Patient 1 Control 1 Patient 2 Control 2 Patient 3 Control 3 Patient 4 Control 4 Patient 5 Control 5 Patient 6 Control 6 Patient 7 Control 7 Patient 8 Control 8 Patient 9 Control 9 Patient 10 Control 10 Patient 11 Control 11
Year of birth
Age at onset (yr)
Disease duration (yr)
Years of surveillance
No. of surveillance examinations
No. of biopsy specimens
1943 1951 1937 1946 1947 1939 1953 1953 1929 1936 1929 1945 1959 1963 1923 1935 1940 1936 1927 1933 1928 1931
33 24 27 15 27 34 12 18 55 42 26 14 14 9 23 13 30 34 29 28 37 25
13 18 26 30 13 21 25 23 9 15 36 31 18 22 43 45 22 23 35 32 28 36
6 8 9 11 7 12 10 13 7 10 3 15 8 7 5 6 6 13 5 13 10 14
3 4 7 5 6 5 11 7 6 6 3 5 6 4 5 3 6 5 6 6 7 8
24 33 58 40 38 45 73 59 50 53 25 35 42 41 33 24 49 41 44 47 67 48
28.5 { 11.5 23.3 { 10.5
24.4 { 10.7 26.9 { 8.8
6.9 { 2.2 11.1 { 3.0
6.0 { 2.1 5.3 { 1.4
45.7 { 15.9 42.4 { 9.7
Total patientsa Total controlsa a
Values are expressed as mean { SD.
°5% crypts positive; /, 6%–35% positive; //, 36%–65% positive; and ///, ¢66% positive. Statistical comparisons between patients and controls were performed by paired Student’s t test, Fisher’s Exact Test, or Mantel–Haenszel test where appropriate.
Results Matching of Patients and Controls All patients in this study had pancolitis defined histologically. As shown in Table 1, the patients and controls were closely matched with respect to year of birth, age at time of diagnosis of UC, duration of UC, number of years under surveillance, and number of surveillance colonoscopies. As of September 1995, none of the controls has been found to have developed dysplasia or carcinoma. Regional Distribution of STn Antigen Expression in Patients Versus Controls Table 2 shows the STn antigen expression for each colonoscopic examination on a patient-by-patient basis. Colectomy was performed in most of the patients within several months after the last colonoscopy. Overall percentage of STn antigen–positive biopsy specimens. For each of the 11 patient-control pairs, STn
antigen expression was more prevalent in patients than / 5e0a$$0004
02-09-96 13:22:16
gasas
controls (Table 2). Overall, STn antigen was expressed in 220 of 503 biopsy specimens (44%) from the patients compared with 53 of 466 biopsy specimens (11%) from the controls. The total STn antigen positivity ranged from 15% to 100% in the patients compared with 0%– 26% in the controls (P õ 0.005; paired Student’s t test). There was no instance of a control expressing STn antigen more frequently than its corresponding patient; in fact, 2 controls never expressed any STn antigen. Compared with the patients, STn antigen expression in the controls was found in fewer crypts per biopsy specimen and in fewer colonic segments (Table 2). Of the 503 biopsy specimens from the patients, 31 were neoplastic (containing either dysplasia or carcinoma) and 20 of these specimens (65%) expressed STn antigen (Table 3). Of the total 969 biopsy specimens, 42 were known to be from inflammatory polyps and 30 of these specimens (71%) were STn antigen positive. However, omitting the neoplastic and inflammatory polyp biopsy specimens from the analysis did not alter the significant differences in overall STn antigen expression between patients and controls. The expression of STn antigen in the neoplastic lesions is consistent with previous observations. However, the expression in inflammatory polyps is a new finding that, at present, is unexplained but that WBS-Gastro
Table 2. Expression of STn Antigen in Surveillance Colonoscopy Biopsy Specimens Cecum Patient 1 Colectomy 1989 1984 1983 Control 1 1993 1990 1987 1985 Patient 2 Colectomy 1990
Control 3 1994 1992 1989 1985 1982 Patient 4 Colectomy 1990 1989 1989 1987 1986 1985 1984 1983 1982 1981 1980 Control 4 1994 1992 1990 1988 1986 1984 1981
Hepatic flexure
Transverse colon
HGD ///LGD
LGD /// /// ////(/)
Splenic flexure
Descending colon
Sigmoid colon
Rectum
Total (%) 24/24 (100)
// // //
/ /// /
0 0 0 0
0 0
//
////// /// ///TA
// /// ///
/// ///
0 0
0 0 0 0
/// 0/0/// 0/0 0//
0 0 0// 0
HGD 0/(///)
HGD 0
/
/// /// // 5/33 (15)
0
0 0 0 0
0 0 / 21/58 (36)
0
1988 1987 0 1986 0 1984 0/(//) 1983 0/(0) 1981 0 Control 2 1991 0 1989 1984 0 1982 0 1979 0 Patient 3 Colectomy 1987 0 1987 1987 1987 0 1984 1980
Ascending colon
0
0
HGD 0
0/(0) 0 0 0 0
0/(0) / 0 0 0
0 / 0 0/(//) //
0 0 0 0 0
0 /// 0 0
0 /// / 0
0
0 0 0 0 0
0 0 0 0 0
0 0 0 0
0 0/0/0/0 // 0 0/0/0
0 0// // 0 0
HGD ///TVA /// /////// /// /// /// 0 5/40 (13) 0 / // 0 0 27/38 (71)
0
/ 0
0
0
0////
(///)
/////// (///) ///// /////// (///)
/
0
0
0
0
///
/ ///
/ 0 0 0 0
0 0 0
/// (///)
CA ///HGD ///////TA/HGD ///HGD //TA /// 0 /// 8/45 (18)
0 0 0 0
0 0 0 0 0
/ 0 0 0 0
0//// 0/0 0/0 0/0 0/0/0
0 0/0 0 0// 0///
/// 0 /// / 22/73 (30)
0 0 0 0 0
0
0 0 0 0 0
0 0 0 0
0 0
/ 0 0 0 0 0
/ 0 0 0 0 0
0 0TA
LGD 0// 0 0/////0 0/0 0 0 0/0TVA/HGD 0/0 0/0// 0
///// /// 0////(/) 0/0 (//0) ////0
CA 0CA 0 / ///
/// /// (//////)
// //
0 0 0 0/0 0 0// ///
0 0 0 0 0 (/) (/)
10/59 (17) 0 0 0 0 0 0
0 0 0 0
0 0 0 / 0 0
0 0 0 0 0 / 0
0 0 0 0 / 0 0
0 0/0 0/0 0// //0 0/0 //
(continued on following page)
/ 5e0a$$0004
02-09-96 13:22:16
gasas
WBS-Gastro
698 ITZKOWITZ ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
Table 2 (cont’d). Cecum Patient 5 1993 1992
/ ///
1990
//
Ascending colon
Hepatic flexure
Transverse colon
Splenic flexure
Descending colon
Sigmoid colon
Rectum
0 //LGD ///TVA /// 0TVA / // / 0
0 //
// ///
/// ///
(///) ///////
///
///
///
///
///
///
///
///
/
//(///) 0 /
/ 0 0
// // 0
// ///LGD //IND
// /// ///
0 0 0 0 0 0
0 0 0 0 0 0
0 0/0/0 0/0/0 0/0/0/0 0/0 0/0/0
0
0 0 0 0 0 0
41/50 (82)
1988 1987 1986 Control 5 1993 1992 1990 1987 1985 1983 Patient 6 Colectomy 1991
// / /
0
0
0
0
0
0
1990 1988 Control 6 1990 1987 1983 1979 1975 Patient 7 Colectomy 1991
0 0
0 0
0
0 0
/// //
//(///) (///)
/
///
0
0
/ 0 0
0 / 0 0
0 0 0
0/0/0 0/// 0 0/0 (//)
0 ///// 0/0 0/0 //
/ // 0
CA //IND
1989 1988 1986 1985 1983 Control 7 1994 1991 1989 1987 Patient 8 Colectomy 1988 1987 1985 1984 Control 8 1993 1991 1987 Patient 9 Colectomy 1991 1990 1989 1988 1987 1986
Total (%)
0
0/53 (0) 0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0
10/25 (40) 0//// ///
CA //CA 0
9/35 (26)
0 0 0
25/42 (60) 0LGD //
0/0IND
/LGD
//IND
///IND
/
/
/ 0 / 0
0 0 0 0
0 / / /
0 /
// // 0 // /
/ /// / 0LGD /
// /
/ // 0 0 0
0 0/0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0/0 0 0/0 0/0/0/0
0 0/0 0/0/0 0
0 0 (0)
CA 0 0 0 0/(/)
0 // 0 0
/HGD //0/0 0 /
0 0 0
/ 0 0
0 0 0
0/0 0/0/0 0/0
/
0/41 (0) 0 0 0 0 16/33 (48) HGD 0 0 /
0 / 0
//HGD ///0 / /
////(/) /// / / 3/24 (13)
0 0 0
0 0 0
/
/
(///////) 0 0/(0) 0
(0) 0 0/(///) /
CA /// 0 0 ///
/
///
//
14/49 (29) 0 0 0 /
0 0 0 0 / 0
0 0 0 0 // 0
0 0 0 0 0 0
0 0 0 0 /// 0
(continued on following page)
/ 5e0a$$0004
02-09-96 13:22:16
gasas
WBS-Gastro
March 1996
STn AS A RISK FACTOR FOR NEOPLASIA IN UC 699
Table 2 (cont’d). Cecum Control 9 1993 0 1991 0 1989 0 1986 0 1980 Patient 10 Colectomy 1991 0/(//) 1990 0/(/) 1989 0 1988 0 1987 0 1987 Control 10 1993 0 1991 0 1989 0 1984 0 1980 1980 0 Patient 11 Colectomy 1993 0 1992 0 1991 0 1990 0TVA / 1986 0 1985 1983 1980 Control 11 1992 1990 1988 1986 1984 1982 1980 1978
0 0 /
Ascending colon
Hepatic flexure
Transverse colon
Splenic flexure
Descending colon
0 0 0 0
0 0 0
0 0 0 0 0
0 0 0 0 0
0/0/0 0/0/0 0/0/0 0/0 0
Sigmoid colon
Rectum
Total (%) 4/41 (10)
0
0 //0 // 0////
/
10/44 (23) 0 0 0 0 0
/ 0 0 0 0
0 0 0 0 0
0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0
CA 0CA 0 0 0 ///0
0 0 ////0 0 / 0
// / 0 // ///
0 0 0/0
0 / 0 0
5/47 (12) 0/0/0 0/0/0 0/0 0/0 (///) ///0(//0)
(0/0)
0 0 0 0
CA //HGD (0) 0/(0) 0
0 0 0 0
0
0
//
0 0 /
0 0 ////
//0
0 0 ///
0 0 0 0 0 0
0/0 0/0 0/0 0/0 0 0 0 0/0
0
10/67 (15) 0 0 0TA 0 0 0 0 0
0 0 0 0
0 0 0 0
0TA 0 0 0 0
0 0 0 0
0 0 0 0
///
4/48 (8) 0
0 0
0 0
0 0 0 0
0 0 0 0 0
0
0/0 0 0 0 0/0 0
0 0 / / // /
NOTE. In some cases, more than one biopsy specimen per segment was taken, but the segment was scored positive if any one of the biopsy specimens was positive. Shaded areas show dysplastic lesions. Boxed areas show confirmed neoplasia in colectomy specimens. Parentheses indicate inflammatory polyps, and blanks indicate that no specimen was available. HGD, high-grade dysplasia (includes carcinoma in situ); IND, indefinite dysplasia; LGD, low-grade dysplasia; TA, tubular adenoma; TVA, tubulovillous adenoma.
might be related to inflammatory cytokine alteration of STn antigen expression. Segmental expression of STn antigen. Although 89% of the biopsy specimens from the control group lacked STn antigen expression, it was not uncommon to find a single biopsy specimen from an individual colonoscopy that was STn antigen positive. While the biological significance of only one STn antigen–positive biopsy specimen per colonoscopy is questionable, this / 5e0a$$0004
02-09-96 13:22:16
gasas
prompted a comparison of the frequency with which STn antigen occurred during colonoscopic examinations (Table 4). The finding of one or more STn antigen– positive biopsy specimens from an individual colonoscopy occurred in all of the patients and in 85% of their colonoscopy examinations. By comparison, one or more STn antigen–positive biopsy specimens occurred in 9 of the controls but in only 47% of their colonoscopy examinations. As the criterion for an STn antigen–posiWBS-Gastro
700 ITZKOWITZ ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
the expression of STn antigen was more widespread in nondysplastic mucosa from the patients than from the controls. It was rare to find STn antigen expression in proximal colonic biopsy specimens of the control group, despite histological evidence of pancolitis in all subjects (Table 2). In biopsy specimens proximal to the hepatic flexure, none of the controls but 7 of the patients expressed STn antigen on at least one occasion, and using the splenic flexure as a cutoff, only 4 of the controls compared with 10 of the patients expressed STn antigen proximally.
Table 3. STn Antigen Expression in Neoplasms and Inflammatory Polyps No. STn antigen positive/total no.
STn antigen positive (%)
1/2 3/3 4/6 4/5 8/15 20/31
50 100 67 80 53 65
24/33 6/9 30/42
73 67 71
Neoplasms Cancer High-grade dysplasia Low-grade dysplasia Indefinite dysplasia Adenomasa Total Inflammatory polypsb Patients Controls Total
STn antigen expression in the last colonoscopy. As yet another indication of the difference in re-
gional STn antigen expression between patients and controls, the last colonoscopy before colectomy for neoplasia in the patients was compared with the colonoscopy of the control that corresponded to similar disease duration (within 3 years). As shown in Table 5, the majority of patients expressed STn antigen in multiple segments throughout the colon, with 67% of patients expressing STn antigen in three or more colonic segments. In contrast, despite comparable disease duration, only 22% of the controls expressed STn antigen in two or more segments, and none of them expressed STn antigen in three or more colonic segments.
a A dysplastic lesion was considered an adenoma if it was pedunculated and the surrounding mucosa was not inflamed. Valid criteria for distinguishing between an adenoma and a dysplastic mass in UC are not established, and the data presented are not intended to emphasize any such distinction. b All inflammatory polyps showed moderate-severe inflammation.
tive colonoscopy examination was made more strict to include more widespread simultaneous expression of STn antigen, a sharper distinction occurred between patients and controls. All of the patients had at least some colonoscopy examinations with three or more STn antigen– positive segments, and all but 1 patient had colonoscopy examinations with four or more STn antigen–positive segments, whereas only 4 controls had colonoscopy examinations with three or more STn antigen–positive segments, and only 2 controls had colonoscopy examinations with four or more STn antigen–positive segments. Thus,
Timing of STn Antigen Expression Repeated STn antigen expression. Table 6 lists the number of patients and controls who showed consecutive STn antigen–positive colonoscopy examinations. For this analysis, a colonoscopy was considered STn antigen
Table 4. Number of STn Antigen–Positive Colonoscopy Examinations If one segment is STn antigen positive (%) Patient Pair 1 2 3 4 5 6 7 8 9 10 11 Colonoscopies Patients
3/3 5/6 3/3 7/9 6/6 3/3 6/6 4/4 4/6 5/6 5/8 51/60 11/11
(100) (84) (100) (78) (100) (100) (100) (100) (67) (84) (63) (85)
Control 4/4 2/5 3/5 4/7 0/6 3/5 0/4 2/3 3/5 2/6 4/8 27/58 9/11
(100) (40) (60) (57) (60) (67) (60) (33) (50) (47)a
If two or more segments are STn antigen positive (%) Patient 3/3 5/6 3/3 5/9 6/6 3/3 6/6 4/4 4/6 4/6 1/8 44/60 11/11
(100) (84) (100) (55) (100) (100) (100) (100) (67) (67) (13) (73)
Control 1/4 2/5 3/5 3/7 0/6 2/5 0/4 1/3 1/5 0/6 0/8 13/58 7/11
(25) (40) (60) (43) (40) (33) (20)
(22)a
If three or more segments are STn antigen positive (%) Patient 3/3 3/6 3/3 2/9 6/6 1/3 6/6 4/4 2/6 1/6 1/8 32/60 11/11
(100) (50) (100) (22) (100) (33) (100) (100) (33) (17) (13) (53)
P õ 0.0001; Mantel–Haenszel test.
a
/ 5e0a$$0004
02-09-96 13:22:16
gasas
WBS-Gastro
Control 0/4 1/5 1/5 3/7 0/6 1/5 0/4 0/3 0/5 0/6 0/8 6/58 4/11
(20) (20) (43) (20)
(10)a
If four or more segments are STn antigen positive ( %) Patient 3/3 1/6 1/3 1/9 6/6 1/3 4/6 3/4 1/6 0/6 1/8 22/60 10/11
(100) (17) (33) (11) (100) (33) (67) (75) (17) (13) (37)
Control 0/4 0/5 1/5 (20) 0/7 0/6 1/5 (20) 0/4 0/3 0/5 0/6 0/8 2/58 (3)a 2/11
March 1996
STn AS A RISK FACTOR FOR NEOPLASIA IN UC 701
Table 5. STn Antigen Expression in Last Colonoscopy STn antigen expression in last colonoscopya
No. of patients (%)
Table 7. STn Antigen Expression Relative to Inflammation
No. of controls (%)
Active inflammation Inactive
One or more STn antigen–positive colonic segment Two or more STn antigen–positive colonic segments Three or more STn antigen–positive colonic segments Four or more STn antigen–positive colonic segments
9/9 (100)
4/9 (44)b
8/9 (89)
2/9 (22)b
6/9 (67)
0/9 (0)
3/9 (33)
b
0/9 (0)b
a
For the patients, the last colonoscopy represents the colonoscopy just before colectomy. For the controls, the last colonoscopy represents the colonoscopic examination corresponding to comparable disease duration (within 3 years) as that of the patient’s last colonoscopy. Patient-control pair 5 was omitted because the patient has not undergone a colectomy, and patient-control pair 6 was omitted because the control did not achieve disease duration within 3 years of the patient. b P õ 0.05 for patients vs. controls; Fisher’s Exact Test.
Mild
Moderate
Severe
No. of STn antigen– positive biopsy specimens (%)a Patients 1/42 (2) 9/41 (22) 41/59 (69) 6/6 (100) Controls 7/156 (4) 0/37 (0) 5/30 (17) 0/5 (0) Significance levelb NS P ° 0.01 P ° 0.0001 P ° 0.002 STn antigen–positive biopsy specimen defined as ú5% positive crypts. b Significance level of difference in STn antigen–positive biopsy specimens between patients and controls using Fisher’s Exact Test. Multiple biopsy specimens from the same patient were treated as independent observations. a
in patients 10 and 11, carcinoma developed without consistent prior STn antigen expression in that region. Interval from first STn antigen expression to first detection of dysplasia. In all but two instances, STn anti-
positive if two or more segments expressed STn antigen, and consecutive examinations refers to any two colonoscopies in a row regardless of the interval between examinations. The vast majority of patients (91%) had at least two consecutive STn antigen–positive colonoscopy examinations, and 73% had three consecutive STn antigen–positive examinations. In contrast, only 3 of the controls (27%) showed repeated STn antigen expression on two consecutive examinations, and only 1 control had three consecutive STn antigen– positive colonoscopy examinations. In patients 1–9, repeated STn antigen expression occurred either in the segment that developed neoplasia or in the immediately adjacent segment (Table 2). However,
Table 6. Patients With Consecutive STn Antigen–Positive Colonoscopy Examinations
Two consecutive STn antigen–positive colonoscopy examinationsa Three consecutive STn antigen–positive colonoscopy examinations Four consecutive STn antigen–positive colonoscopy examinationsb
No. of patients ( %)
No. of controls (%)
10/11 (91)
3/11 (27)c
8/11 (73)
1/11 (9)c
2/8 (25)
0/8 (0)
a
STn antigen–positive colonoscopy examination defined as a colonoscopy specimen in which two or more segments expressed STn antigen. Consecutive indicates any two colonoscopies in a row regardless of interval between examinations. b Patient-control pairs 1, 6, and 8 were ineligible for this analysis because either the patient or the control underwent only three total examinations. c P õ 0.05 for patients vs. controls; Fisher’s Exact Test.
/ 5e0a$$0004
02-09-96 13:22:16
gasas
gen expression preceded the detection of dysplasia by 2–7 years (Table 2). In patients 1 and 5, the first available surveillance examination already showed dysplasia, and STn antigen was expressed simultaneously with dysplasia. Effect of Inflammation on STn Antigen Expression To explore the relationship, if any, between STn antigen expression and the degree of inflammation, a subset of biopsy specimens that had all been interpreted by the study pathologist (N.H.) were analyzed (Table 7). In the presence of inactive, quiescent inflammation, only 2% of biopsy specimens from patients and 4% from controls were STn antigen positive. With increasing degrees of active inflammation, the proportion of STn antigen–positive biopsy specimens increased but only in the patients. Thus, the percentage of STn antigen–positive biopsy specimens in the setting of moderate inflammation was 69% for the patients compared with only 17% from the controls, and in the setting of severe inflammation, all biopsy specimens from the patients but none from the controls were STn positive. Therefore, although STn antigen is more likely to be expressed in the presence of active inflammation than in quiescent disease, it is more closely related to the neoplastic potential of the colonic mucosa than to the degree of inflammation per se.
Discussion Little is known about STn antigen expression and colon carcinogenesis in inflammatory bowel disease. Monoclonal antibody TKH2 reacts strongly with small WBS-Gastro
702 ITZKOWITZ ET AL.
GASTROENTEROLOGY Vol. 110, No. 3
intestinal goblet cells but rarely reacts with goblet cells of the colon,4,9,13,14 suggesting that TKH2 reactivity by colonocytes is indicative of differentation along a small intestinal lineage.15 With malignant progression in UC, colonocytes undergo a process of small intestinalization, as shown not only by monoclonal antibody TKH2 reactivity but also with other markers of small intestinal enterocytes such as sucrase-isomaltase and small intestinal mucin antigen, a mucin antigen that is unrelated to STn antigen.16 – 18 Because the association between STn antigen expression and neoplastic progression in sporadic colon cancer has been well established,1–5 one might expect STn antigen to be a useful marker of malignant progression in UC-associated colonic neoplasia. However, formal analysis of this hypothesis is necessary because it is becoming clear that other molecular genetic alterations common to both sporadic and colitis-associated colon carcinogenesis can differ in frequency and time of appearance.15 In a previous study of UC tissues using another antiSTn antibody, the expression of STn antigen progressively increased in specimens showing inflammation alone, low-grade dysplasia, high-grade dysplasia, and carcinoma.6 That study, which purposely selected tissues from patients with neoplasia, prompted a further analysis of the colonic distribution and timing of STn antigen expression.7 This latter investigation disclosed that at the time of colectomy, STn antigen expression was prevalent in nondysplastic mucosa remote from the neoplasm, but few control colectomy specimens from neoplasia-free colons were studied. Moreover, in 6 of 7 patients with UCassociated cancer or high-grade dysplasia in that study, STn antigen was expressed in histologically normal (quiescent) mucosa of the neoplastic region several years before the detection of neoplasia, but the issue of STn antigen expression throughout the colon in the years before the detection of neoplasia was not addressed, and again, few controls were used. In the present study, we simulated the approach used for finding dysplasia during surveillance colonoscopy by analyzing STn antigen expression both throughout the colon anatomically and throughout the years chronologically. Another important new element of the current study was the careful selection of matched controls, thereby permitting a comparison of results between individuals who were known to have neoplasia and those who had not developed neoplasia despite comparable disease duration and intensity of surveillance. The retrospective study design has certain obvious limitations, including nonstandardized specimen collection at the time of colonoscopy, some difficulties with specimen retrieval from archives, and the possibility of controls converting to / 5e0a$$0004
02-09-96 13:22:16
gasas
patients (an issue that was considered in selecting controls). Nonetheless, the strengths of this design are the ability to study enough patients who have developed cancer and the certainty of the diagnosis and location of neoplasia at colectomy. Our results indicate that STn antigen was rarely expressed in colonic mucosa of patients with UC who had not developed neoplasia but was frequently expressed in nondysplastic mucosa of those who progressed to neoplasia. In every one of the case-control pairs, the percentage of STn antigen–positive surveillance biopsy specimens was greater in the patients than in the controls, and in fact, 2 controls never expressed STn antigen in any biopsy specimen, despite numerous repetitive sampling. In keeping with our previous finding of widespread STn antigen expression in colectomy specimens from cancer-bearing colons, we have now observed that even before the detection of neoplasia, STn antigen expression can be widespread throughout the colon in individuals who subsequently develop neoplasia. In fact, half of all colonoscopy examinations from the patients showed STn antigen expression in at least three colonic segments, compared with only 10% of colonoscopy examinations from the controls. In addition to manifesting widespread expression of STn antigen, the patients also showed repeated STn antigen expression in consecutive colonoscopy examinations. Thus, defining an STn antigen–positive colonoscopy examination as one with at least two segments positive for STn antigen, 91% of patients had two consecutive STn antigen–positive colonoscopy examinations and 73% had three consecutive STn antigen–positive examinations compared with values of 27% and 9%, respectively, in controls. Moreover, 9 of the 11 patients manifested STn antigen expression repeatedly in the segment of the colon that developed neoplasia or in the adjacent segment. The timing of STn antigen was such that it either occurred simultaneously with dysplasia (2 patients) or up to 7 years before dysplasia was detected (9 patients). This raises the question of whether individuals who are destined to develop neoplasia in UC constitutively express STn antigen with the onset of their colitis. It has been postulated that altered glycosylation of mucins in UC might predispose the epithelial cells to growth-modulating effects of luminal constituents, resulting in the development of neoplasia.19 Although our data do not permit an analysis of STn antigen expression during the earliest stages of colitis because surveillance colonoscopy is initiated only after 8–10 years of disease, one can compare the first available STn antigen expression in WBS-Gastro
March 1996
STn AS A RISK FACTOR FOR NEOPLASIA IN UC 703
patients and controls. Nine of the patients, but also 4 of the controls, expressed STn antigen in at least two segments at the time of the initial colonoscopy, suggesting that there does not seem to be a specific intrinsic glycosylation defect in the patients. No patient developed cancer without prior STn antigen expression, although in 2 patients the prior STn expression was only sporadic. Although this latter phenomenon may be caused by sampling error, it suggests that the absence of prior STn antigen expression does not preclude the development of neoplasia. The same limitation applies to dysplasia, which may be absent before the development of cancer. In this regard, it is noteworthy that 2 patients with long-standing UC (not included in this series) who developed cancer without any prior dysplasia, despite intensive surveillance, both showed prevalent STn antigen expression for several years before cancer detection.20 This observation suggests that STn antigen expression may by complementary to dysplasia as a marker of cancer risk. DNA aneuploidy is another biomarker that may be complementary to dysplasia in UC because it is prevalent and can precede dysplasia by several years.21–24 However, STn antigen expression has been found to be independent of aneuploidy in sporadic colon cancers and to carry its own independent prognostic effect.5 We are currently pursuing the relationship between STn antigen expression and DNA aneuploidy in prospectively acquired surveillance colonoscopy biopsy specimens of patients with UC.25 In the setting of inflammation, it is often difficult to distinguish between dysplasia and regenerative cytological changes.12 STn antigen was expressed in only 2%– 4% of biopsy specimens with quiescent, inactive inflammation, and although expression was increased in biopsy specimens with active inflammation, this increase was limited primarily to biopsy specimens from the patients rather than controls. Thus, the presence of inflammation itself does not obscure the interpretation of STn antigen expression. On the contrary, inflammatory changes seemed, if anything, to enhance the sensitivity of STn antigen expression as an indicator of cancer risk. In summary, the current observations strongly implicate STn antigen as a marker of colon cancer risk in longstanding UC. To be sure, patients in the control group occasionally showed STn antigen, but the expression was usually sparse, less reproducible over time, and rarely found in the proximal colon. The pattern of STn antigen expression in the patients suggests that STn antigen expression, particularly if it occurs in at least two colonic segments and repeatedly over two or more consecutive colonoscopies, may be a marker of a colon that is predis/ 5e0a$$0004
02-09-96 13:22:16
gasas
posed to neoplasia. Further prospective studies should help to define the potential utility of STn antigen in clinical management of patients with UC.
References 1. Itzkowitz SH, Yuan M, Montgomery CK, Kjeldsen T, Takahashi HK, Bigbee WL, Kim YS. Expression of Tn, sialosyl Tn, and T antigens in human colon cancer. Cancer Res 1989;49:197– 204. 2. Itzkowitz SH, Bloom EJ, Lau TS, Kim YS. Mucin-associated Tn and sialosyl-Tn antigen expression in colorectal polyps. Gut 1992;33:518–523. 3. Ørntoft TF, Harving N, Langkilde NC. O-linked mucin-type glycoproteins in normal and malignant colon mucosa: lack of T-antigen expression and accumulation of Tn and sialosyl-Tn antigens in carcinomas. Int J Cancer 1990;45:666–672. 4. Yonezawa S, Tachikawa T, Shin S, Sato E. Sialosyl-Tn antigen: its distribution in normal human tissues and expression in adenocarcinomas. Am J Clin Pathol 1992;98:167–174. 5. Itzkowitz SH, Bloom EJ, Kokal WA, Modin G, Hakomori S, Kim YS. Sialosyl-Tn: a novel mucin antigen associated with prognosis in colorectal cancer patients. Cancer 1990;66:1960–1966. 6. Thor A, Itzkowitz SH, Schlom J, Kim YS, Hanauer SB. Tumorassociated glycoprotein (TAG-72) expression in ulcerative colitis. Int J Cancer 1989;43:810–815. 7. Itzkowitz SH, Marshall A, Kornbluth A, Harpaz N, McHugh JBD, Ahnen D, Sachar DB. Sialosyl-Tn antigen: a marker of risk for colorectal neoplasia in patients with chronic ulcerative colitis. Gastroenterology 1995;109:490–497. 8. Kjeldsen T, Clausen H, Hirohashi S, Ogawa T, Iijima H, Hakomori S. Preparation and characterization of monoclonal antibodies directed to the tumor-associated O-linked sialosyl-2 r 6 a-N-acetylgalactosaminyl (sialosyl-Tn) epitope. Cancer Res 1988;48: 2214–2220. 9. Ogata S, Ho I, Chen A, Dubois D, Maklansky J, Singhal A, Hakomori S, Itzkowitz SH. Tumor-associated sialylated antigens are constitutively expressed in normal human colonic mucosa. Cancer Res 1995;55:1869–1874. 10. Jass JR, Allison LM, Edgar S. Monoclonal antibody TKH2 to the cancer-associated epitope sialosyl Tn shows cross-reactivity with variants of normal colorectal goblet cell mucin. Pathology 1994;26:418–422. 11. Jass JR, Roberton AM. Colorectal mucin histochemistry in health and disease: a critical review. Pathol Int 1994;44:487–504. 12. Riddell RH, Goldman H, Ransohoff D, Appelman HD, Fenoglio CM, Haggitt RC, A˚hren C, Correa P, Hamilton SR, Morson BC, Sommers SC, Yardley JH. Dysplasia in inflammatory bowel disease: standardized classification with provisional clinical applications. Hum Pathol 1983;14:931–966. 13. Iwata H, Itzkowitz SH, Werther JL, Hayashi K, Nakamura H, Ichinose M, Miki K, Tatematsu M. Expression of sialosyl-Tn in intestinal type cancer cells of human gastric cancers. Acta Pathol Jpn 1993;43:646–653. 14. Werther JL, Rivera-MacMurray S, Bruckner HB, Tatematsu M, Itzkowitz SH. Mucin-associated sialosyl-Tn antigen expression in gastric cancer correlates with an adverse outcome. Br J Cancer 1994;69:613–616. 15. Itzkowitz SH, Greenwald B, Meltzer SJ. Basic science review: colon carcinogenesis in inflammatory bowel disease. Inflam Bowel Dis 1995;1:142–158. 16. Hertzog PJ, Robinson HC, Ma J, Mackay IR, Linnane AW. Oncofetal expression of the human intestinal mucin glycoprotein antigens in gastrointestinal epithelium defined by monoclonal antibodies. Int J Cancer 1991;48:355–363. 17. Filipe MI, Sandey A, Ma J. Intestinal mucin antigens in ulcerative
WBS-Gastro
704 ITZKOWITZ ET AL.
18.
19. 20.
21.
22.
23.
GASTROENTEROLOGY Vol. 110, No. 3
colitis and their relationship with malignancy. Hum Pathol 1988;19:671–681. Andrews CW Jr, O’Hara CJ, Goldman H, Mercurio AM, Silverman ML, Steele GD Jr. Sucrase-isomaltase expression in chronic ulcerative colitis and dysplasia. Hum Pathol 1992;23:774–779. Rhodes JM. Colonic mucus and mucosal glycoproteins: the key to colitis and cancer? Gut 1989;30:1660–1666. DuBois D, Rosenberg IR, Ho I, Chen A, Itzkowitz SH. Expression of sialosyl-Tn (STn) in surveillance biopsies of patients with ulcerative colitis (UC)-associated colorectal cancer (CRC) but no prior dysplasia (abstr). Gastroenterology 1994;106:A1023. ´ st A˚. Comparative DNA Lo¨fberg R, Caspersson T, Tribukait B, O analyses in longstanding ulcerative colitis with aneuploidy. Gut 1989;30:1731–1736. ´ st A˚, Tribukait B. DNA aneuLo¨fberg R, Brostro¨m O, Karle´n P, O ploidy in ulcerative colitis: reproducibility, topographic distribution, and relation to dysplasia. Gastroenterology 1992;102: 1149–1154. Rubin CE, Haggitt RC, Burmer GC, Brentnall TA, Stevens AC, Levine DS, Dean PJ, Kimmey M, Perera DR, Rabinovitch PS. DNA aneuploidy in colonic biopsies predicts future development of dysplasia in ulcerative colitis. Gastroenterology 1992;103: 1611–1620.
/ 5e0a$$0004
02-09-96 13:22:16
gasas
24. Levine DS, Rabinovitch PS, Haggitt RC, Blount PL, Dean PJ, Rubin CE, Reid BJ. Distribution of aneuploid cell populations in ulcerative colitis with dysplasia or cancer. Gastroenterology 1991;101: 1198–1210. ¨ st A˚, Tribukait B, 25. Karle´n P, Brostro¨m O, Lo¨fberg R, Young E, O Itzkowitz S. Sialosyl-Tn (STn) antigen expression compared to dysplasia and DNA-aneuploidy in prospective colonoscopic surveillance biopsies from patients with longstanding ulcerative colitis (abstr). Gastroenterology 1995;108:A487.
Received August 18, 1995. Accepted October 18, 1995. Address requests for reprints to: Steven H. Itzkowitz, M.D., Gastrointestinal Division, Box 1069, Mount Sinai School of Medicine, One Gustave Levy Place, New York, New York 10029-6574. Fax: (212) 348-7428. e-mail:
[email protected]. Supported by Public Health Service grant RO1CA52491 from the National Cancer Institute, the Crohn’s and Colitis Foundation of America, and the Chemotherapy Foundation. Dr. Itzkowitz is a recipient of an Irma T. Hirschl Career Scientist Award. The authors thank Drs. Peter Rubin, Jerome Waye, Daniel Present, Irwin Gelernt, Isadore Kreel, Joel Bauer, Adrian Greenstein, and Jacob Walfish for providing access to the patient material.
WBS-Gastro
