were histologically assessed according to the updated Sydney-classification, in 845 patients (mean age 55y) with H. pylori infection and no peptic ulcer or gross pathology were identified and further analyzed according to gastritis phenotypes, Prevalence of corpus-predominant gastritis (CPG), intestinal metaplasia (IM) and severe atrophy (SATR) were evaluated. Results: The age-adjusted prevalence of the gastritis phenotype was independent of the gender. Prevalence of histological characteristics considered for risk assessment are shown in the table. Conclusion: The prevalence of gastritis features at risk for cancer developmem increases with age. Among the investigated gastritis phenotypes 1M at any location as well as the combination of IM+CPG showed the highest increase being 4-fold for IM as well as for combinod CPG with IM at the age of 65 years versus less than 45 years These findings reinforce the necessity for the histological assessment even in subjects with a normal endoscopic appearance The age-dependent increase in prevalence n[ risk gastritis phenotypes however does not allow to estimate the individual risk for gastric cancer development - only the follow-up vail tell us.
CPG(%) IN(%) ,~ATR(%) CPG~IM(%)
845 154 241 109 64
114 68 79 88 0,9
l
II
120 125 117 67 33
15t 119 185 7.3 20
.........
199 151 251 151 6,5
166 229 392 127 127
W1330 The Correlation Between Lymph Node Metastasis and The Depth of Submueosal Invasion in Submueosal Colorectal Carcinoma in Japan-A Report from The Japanese Society for Cancer of The Colon and Rectuuru Kazuaki Kitajima, Jun Takeda, Shigehiko Fujii, Ko NagasakG Shingo Ishiguro, Tadakatzu Shiruoda, Akinori lwashita, Yo Kato, Yoichi Aiioka, Hidenobu Watanabe, Tetsuichirn Mutt, Takahiro Fujimori Background/Aim: It has been reported that [,vmph nodes metastasis occurs in only 83-i2% of patients with submucosal colorectal carcinoma. Most submucosal colorectal carcinoma without 1Dnph nodes metastasis would be considered suitable for endoscopic treatment. However, clear indications tor endoscopic treatment of submucosal colorectal carcinoma have not been establish. The Japanese Society for Cancer of the Colon and Rectmn has been attempting to determine the criteria tbr grading submucosaI colorectal carcinomas on the basis of the depth of submucosal invasion in order to clarify the indications for endoscopic treatment of submucosal colorectal carcinoma The aim of this study is to clarify the characteristic of submucosal colorectal carcinoma with lymph uode metastasis by using the criteria of the Japanese Society for Cancer of the Colon and Rectum. Methods: Eight hundred seventy seven submucosal colorectal carcinomas that have been resected surgically at 6 flcilines were analyzed. All cases were classified by the status of mnscularis mucosae on H.E stein into 3 types: type A- muscularis mucosae could be recognized clearly; type B- muscularis mucosae could not be recognized clearly but could be assumed; type C- mnscularis ruucosae could not be assumed. In type A and B, the depth of submucosal invasion was measured flora the Iower edge of muscularis mucosae or assumed muscularis mucosae to the deepest tnvasivr portion, in type C, it was measured from superficial of tmnor, but in the case o~ pedunculated type by macroscopic classification, it was measured from the level 2 line in Haggitt's classification. The case whose submucosal deepest invasive portion was limited to the upper level 2 was defined as head invasion. Results: Lymph node metastasis were observed in 92 of 877 submucosal colorectal carcinomas (I 0.4%), and they were not observed in cases with depth of invasion less than 1000 ~xm and head invasion. The rate of lymph nodes metastasis classified according to the depth of submucosaI invasion was as f01Iows: 11.0% (17/155) in 1000-2000 Ixm, 13559% (24/177) in 2000-3000 t~m, 13.563% (51/ 376) in over 3000 lain. Conclusion: These results suggest that submucosaI colorectal carcinomas with depth of invasion less than 1000 ixm and head invasion have no lymph node metastasis and these cases would be become indications for endoscopic treatment.
95 200 400 126 126
W1328 Identification of Osteopontin as a serum marker of ampullary carcinoma by gene expression profiling Tjarda Van Heek, Anirban Maitra, Jens Koopmann, Neff Fedarko, Alka Jain, Ayman Rahruan, Christine Iacobuzio-Donahue, Volkan Adsay, Raheela Ashlaq, Charles Yet, John L Cameron, Johan A. Offerhaus, Ralph Hruban, Karin D Berg, Michael Goggins Background: ,Gmpullary adenocarcinoma is an aggressive cancer with a poor prognosis. The area of this study was to use olignnucleotide microarrays to identify differentially expressed gettes in ampullary adenocarcinoma to better understand the biology of this disease and to identify potential markers for the early diagnosis of ampullary cancer. Materials & Methods: The Altymetrix Human Genome U133 GeneChip set (HG-U133A and HG-U133B) was used to obtain gene ex-pression profiles of 5 ampullary cancer and 10 normal duodenal samples. The Afiymetrix Data Mining Tool software was utilized to assess the genes that were overexpressed in ampullary carcinoma compared to normal duodenum. Results: Over 500 genes were identified as 3-tbld or greater e.xpressed in ampunary cancers compared to normal duodenum. The expression profiles of nine candidate overexpressed genes (osteopontin, PSCA, mesothetin, T1MP-1, mucin-1, mucin-5, fascin, heat shock protein 47, fibronectin 1) were confirmed by immunohistocbemistry or in situ hybridization on tissue micro arrays (TMA) containing 54 ampullan/ cancers. One of the genes ovarexpressed on the tissue micoarrays was osteopontin, which was expressed > 20-fold higher in ampullary adenocarcinoruas compared to normal duodenum by genechip analysis We therefore detelmined sernm osteopontin levels in ampuliary neoplasms and normal controls. Mean pre-operative serum osteopomin levels as measured by competitive ELtSA were 906 ng/ml (range 268 ng/ml) in patients with ampullary cancer, 867 ng/ml (range 160 ng/ml) in patients with an ampunary adenoma, and 204 ng/ml (range 65 ng/ml) in sera from age-matcbed healthy controls (P<0.O01). Using a cut-off of 2 standard deviations above the mean, sera from patients with ampullary neoplasms could be distinguished from healthy controls with a sensitivity of 100% and specificity of 95%. Conclusion: These results demonstrate that elevated serum osteopontin is a sensitive marker of ampullary adenoearcinoma and highlight the utility of gene expression profiling to identify" candidate tumor markers.
W1331 Nueleosomal DNA hypermethylation detected in sera of colon cancer patients as a marker of cancer surveillance Mikihiro Fujiya, Toshifumu Ashida, Kotaro Okamoto, Jun Sakamoto, Yubei Inaba, Yasuko Miyoshi, Atsuo Maemoto, Fnmika Orii, Atsumi Yasuda, Jim watan, Tokiyoshi Ayabe, Yusnke Saito, Yutaka Kohgo Background/Aim: For cancer screening, detection of tumor-specific DNA alteratioits in the sera has been considered to be highly specific, but the clinical application has been suspended because of the limitation of the sensitivity, and the heterogenity of the targeted mutation sites/types of the gene. DNA hypermethylation of cancer suppressor gene promotor, such as p16 promoter, is relatively common alteration observed in various cancers. However, the sensitivity of the detection of p16 hypermethylation in serum DNA has bee reported only 23-30% in colon cancer patieuts. To increase the sensitivity of the assay, we developed the novel method by collecting the nucleosonre DNA from the serum sample, whmh contains the de-acetylated histones tightly combined to methylated DNA sequences. Materials~lethods: Primary colorectal cancer tissues and serum samples were collected from 51 patients operated in our hospital. Comrol serum samples were obtained from 4 patients with adenomatons polyps and 10 healthy volunteers. Serum DNA samples were prepared by DNA absorbingresin. Nucleosome DNA in sera was prepared by immunoprecipitation with anti-histone H1 monoclonal antibody followed by DNA extraction, p16 promoter hypermethylation (p16 HM) in both DNA preparations were analyzed by methylation specific PCR (MSP). DNA samples isolated from colorectal tissues were also subjected to MSP. Results: p16 HM was detected in 26/51 tissue samples. In these 26 patients, p16 HM was detected in 22/26 (85%) by nudeosome DNA preparation, In contrast, only 16/26 (62%) of conventional serum DNA preparation showed p16 HM. Especially, in stage I1 colorectal cancers, sensitivity of MSP with necleosome DNA ( 11/15; 73 %) is signifcamly higher than that with conventional serum DNA preparation (6115; 43%). No false-positive result was obtained from control DNA samples. Conclusions: Nucleosome DNA preparation from the serum sample increased the sensitMty for detecting p16 HM rather than the conventional serum DNA preparation in colon cancer patients, especially in stage II patients with curable diseases. The overall sensitivity of the single gene analysis is still remained limited, however, combination of other hypermethylated genes to the targets of the nucleosome DNA analysis warrant the possible application for colon cancer surveillance.
W1329 Piru-10ncogene: a Novel Biomarker of Gastrointestinal Primary and Metastatic Adenocareinomas Simona Di Cart, Swaminathan Rajendiran~ Rajiv Dhir, EmilLa Carloni, Antonio Gasbarrini, Antonia Sepulveda Backgruund: Gastrointestinal (GI) carcinomas represent the second most common malignancies in the US, Pim-1 is a serine-threonine protein kinase. We recently identified pim-1 as a novel gene that is expressed in gastric carcinoma cells and found that it is induced by H. pylori, a known tnalor risk factor in gastric cancer development, Atins: To evaluate pim-1 as a diagnostic tool to assess the site of origin of GI adenocarcinomas and specifically to distinguish upper (,gastric and esophageal) from distal GI carcinomas (colorectal), based on evaluation of pim-1 expression in metastatic as well as in primary lesions. Methods: The expression of Pim-1 was evaluated in 64 adenocarcinomas (15 esophageal, 23 gastric and 26 colorectal) and non-neoplastic adjacent mucosa. L)anph node metastases were examined in five cases Fire-l was detected by mamunohistochemical stains of tbrmalin-fixed paraffin embedded tissue sections with a polycloual anti-pim-1 goat antibody (Santa Cruz). Three difterent scores were given depending on the percentage of positive tumor cells (Score 0: Pim-1 expression in <5% tumor cells; Score I: >5% <30% tumor cells; Score 2: >30% tumor cells). Resuhs: Plm-1 (score 2) was expressed in esophageal (9/15; 60%) and stomach (19/23; 82.6%) but not in colorectal adenocarcinomas (0/26) (p<0.0001). Normal gastric and colonic mucosa expressed low levels of piru-l, Of five cases of metastatic carcinoma in lymph nodes, the pim-1 immunoreactivity in the primary tumor was similar to the pattern in the metastatic loci. Conclusions: in this study we show that the pim-1 protooncogene is differentially expressed in stomach and esophageal carcinomas as compared to colorectal adenecarcinomas, making pim-1 an excellent biomarker to help identify" the site of origin of these tumors. Together, these fndings indicate that pim-1 is a novel biomarker for the differemial diagnosis of upper gastrointestinal adenocareinomas (esophageal and gastric) and lower gastrointestinal carcinomas (from the colorectum).
W1332 The Fate of Indefinite Dysplasia in Ulcerative Colitis Nimj Jani, Asher Kornbluth, Victoria Croog, Noam Harpaz, Steven ltzkowitz, Thomas U]lman Background: Little is known about the time-dependent risk of neoplastic progression in ulcerative colitis (UC) patients (pts) with mueosal changes interpreted as indefinitefor dysplasia (IND). Aim: To determine whether 1ND found at surveillance colonoscopy has a different rate of neoplastic progression than if no dysplasia is found. Methods: From the G[ Pathology database, we identified all pts who underwent surveillance colonoscopy for UC in 1996-7 and reviewed their medical records. We identified a cohort of pts with IND found at surveillance without prior history of neoplasia (fiat low-grade dysplasia (LGD), high-grade dysplasia (HGD), or cancer (CRC)). A comparable cohort with no dysp[asia (NOD) consisted of pts with a negative surveillance exam within two months of an IND subject, and with a similar number of previous negative exams. Cohorts were analyzed by lit~-table methods
A-649
AGA Abstracts
specifidties of 80%-95% relative to healthy control sera or sera from patients with bemgu colon disease. Classification trees were generated utilizing muhiple protein peaks in the mass range of 2-8 kDa. Sera from patients with adenomatous polyps were classified into both control and cancer groups. CONCLUSIONS: These initial results support the potential of this approach as an adjunct diagnostic tool for early detection of colon cancer, We are currently validating this system by processing a larger number of samples and also evaluating additional artificial intelligence/learning algorithms for the analysis of the profiling data. Supported by the National Cancer Institute Early Detection Research Network.
comparing progression to any grade of neoplasia and advanced neoplasia (HGD or CRC) Resuhs: We identitied 48 pts with IND and found 48 matched NoD controls. 13 of 48 1ND pts (27%) progressed to neoplasia: 9 LGD, 3 HGD and 1 CRC. Mean follow-up (f/u) was 54 months; only i pt progressed within 12 months of initial IND finding. In the NoD cohort 6 of 48 (13%) pts progressed: 1 HGD, 0 CRC; mean s was 53 months. On an actuarial basis, 5-yr progre.ssion to neoplasia was 30% for IND and 12% for NoD; this difference was signitkam (p= .04) by lite table analysts.(Figure) The difference in 5-yr progression to advanced neoplasia (12% and 0%) was not significant. Conclusions: Compared to patients who are free of dysplasia, a finding of 1ND was associated with a higher rate of progression to neoplasia. The optimal interval for surveillance examinations is not known, but our data support the practice of following patients with IND more closely.
W1335 Analysis of mRNA Expression Profiles in Colorectal Adenomas Using K-Nearest Neighbor Cluster Analysis Lawrence C, Lapointe, Graeme P. Young, Howard M. Chandler
Figure 1: Progression to Dysplasia
BACKGROUND: Several studies have explored gene expression profiles in cancer tissues using commerciall)' available DNA microarrays. Studies have also shown that cluster analysis techniques can be used to analyze gene expression profiles that are discriminating between normal tissues and colon cancer tissues. To be clinically useful as screening tools, these techniques must be directed toward discovering the earliest, treatable stages of clinical disease and not the late stage invasive cancers. AIM To determine if cluster analysis can be used to identify mRNA markers that are predictive for colorectal adenomas from quantitative expression profile data. METHOD Candidate mRNA markers for adenomas were identified by differential display and quantified using quantitative RT-PCR relative to normal tissues. The expression levels of these markers were measured in 71 tissue samples (21 normal, 20 tubular adenoma, 26 tubulovillous adenoma, and 4 villous adenoma.) Expression levels were analyzed using duster analysis based on the k-nearest neighbor (KNN) technique, This method classifies a given tissue (X) according to the class membership of the k tissues nearest to (X) in n-dimensional space described by the mZNA expression levels for a given set of markers. Tissues were considered unclassified if analysis of the k-nearest neighbors tissues failed to achieve a unanimous result, where k = 3 t0r this study'. The strength of each set of markers for predicting disease state was measured by comparing the chtster analysis solution to pathology results, RESULTS Cluster analysis showed that no combination of the markers achieved perfect classification of all 71 Normal and Adenozna tissues, However, near perfect discrimination (98.6%, 70/71) can be achieved for one set of three markers and six unique sets of four markers, in nearly all strong marker sets (6/7) the same normal tissue was incorrectly classified as adenomatous. CONCLUSIONS: Cluster analysis using the k-nearest neighbor technique is useful for identifying expression profiles that can correctly discriminate between adenomatous and nom~al tissues. Further, expression profiling using this technique may identify tiss-des that are judged 'normal' by histopathology but exhibit mRNA expression profiles that are predictive for precancerous genomic changes.
5,
~r [-I- - q,'l-~ Oorougt~ r-II*.,.P'r
.2,
'
~ II
r ,'~
-
A .......
I
I I
I
.t-
'~'~,o .,;j.
....... / -
.
.
.
.
+ 1lll
Moths
W1333 Fiberoptic In- Situ Detection of Dysplastic Polyps in Apc"~lW+Mice Using Autofluorescence Bhaskar Banerjee, Nicholas O. Davidson, Jeft~ey O. Henderson
W1336
There is intense interest in the potential applications of non-invasive cytological and spectroscopic methods m detect early' changes in the growth and malignant potential of luminal gastrointestinal tract tumors. One such method employs tissue autofluorescence. However, a significant limitation to the use of conventional tissue antofluorescence is its imprecision at longer wavelengths (450- 650 nm) and reliance upon complex cakulations. We have optimized a simpler method that uses anmfluorescence of shorter wavelengths, in the range of 320- 340 nm, to study celMar autofluorescence in dysplastic intestinal polyps in Apc~"e +mice, a robust model of adenomatous polyposis, using a small, portable fiberoptic device containing a Xenon lamp. Apc' ~"mice were sacrificed at 115 day's, the intestines washed and placed mucosal side up on a matte black surface. A two-way fiberopnc probe recorded autofinorescence and the intensities were normalized to the fluorescence intensity" from normal mucosa to derive an autotluorescence mtensity ratio (AIR). Three readings were recorded at each site. Serial recordings of autofluorescence were made at ~ 1 cm intervals along one intestinal segment that contained 3 polyps. Nl intestinal samples were microscopically confirmed adenomas A total of 69 pol)~s were studied in 11 mice; 63 were in the smaU intestine and 6 in the colon. The mean autofluorescence intensuy ratios -+ SD for small intestinal polyps was: 1.22 +_ 0.14 (p < 0.0001). For colonic polyps, the AIR -+ SD was 1 28 -+ 0.11 (p = 0.0016). The sensitMty was 96%. Measurements of AIR along the length of an intestinal segment correctly identified 3 polyps them the surrounding norrmal mucosa. The findings demonstrate that the mean AIR was significantly higher in polyps of the small intestine and colon compared to normal mucosa. Unlike other methods, the intensity of this autofluorescence band, which is produced by alterations in cellular tryptophan abundance, increases with neoplasia and requires simple measurements without complex spectral analysis. In conchision, tfbemptic measurement of autofluoreseence accurately detects in-situ intestinal dysplasia and may be adapted to real time detection and analysis of precursor lesion development.
Efficacy and feasibility of Colonoscopy Screening in increased risk subjects for Colorectal Cancer: our experience after 22 months of activity Vincenzo Matarese II, Alessandro Pezzoli, Lucio Trevisain, Gniseppina Marangoni, Viviana Cilala, Caterina Rizzo, Sergio Boccia, Sergio GnUini Aim: The district of Ferrara has one of the highest incidence of colorectal cancer(CRC) in Italy. Since January 2000 we have started a colonoscopy (C) screening program focused on first degree relatives of CRC patients. We now report the resuh after 22 months of this screening. Patients and Methods: Subjects included in the screening program aged between 45-75 yr with at least one first-degree relative with CRC. When C was refused, barium enema or fecal occult blood test were suggested. Demographic features are: 205 males and 254 females. Age range 57.3 years. 351 participants had one pasent, 104 one brother and 4 one son afi~cted by CRC. Of these participants 65(14.1%) had two first degree relatives atfected by CRC. Results: FromJanualy 2000 to October 2002, 459 subjects were interviewed In the screening we enrolled 451 subjects(98.2%); 8(1.7%) people refused the option screening, 16(3,5%) subjects chose to perform the barium enema and 10(2.2%) fecal occult blood test: all had negative findings. 425 agreed to undergo endoscopic examination(92.6%): 373 C have been already performed (52 are scheduled). Adenomas and carcinoma (Dukes A)were found in 88(24%) subjects and 3(0.8%) subjects respectively. Histological examination of the 123 lesions found (32.9%), showed: 32 hyperplastic polyps(26%), 60 tubular adenomas(48.8%), 21(171%) tubulovdlons (two of them surgically treated), 7(5.7%) with severe dysplasia and 3 adenocarcinoma(2.4%)Muhiple adenomas were found in 26(28.5%), and in 36(39.5%) the diameter was >1 cm. Cecum was reached in 85% of the endoscopic examinations. Sedation was used in only 21 colonoscopies(5.6%). One perforation related to polypectomy(0.3%) was observed. Conclusions: A C-based screening in selected high risk populations is well accepted(92.6% of attendance rate to C) even without sedation and relatively safe. Our results confirm a high prevalence of advanced neoplasm and early colon cancer in first degree relatives of CRC patients.
W1334 SELDI-TOF-MS Profiling of Serum for Early Detection of Colorectal Cancer Rsohard R, Drake, Upender Manne, Ban-Ling Adam, Christina Ahn, Lisa Cazares, O.J. ~mmes, Wdliam L. Bigbee, William E, Grizzle, David A Johnson
w1337 Detection of COX-2 Messenger RNA in Feces Is Beneficial for Coloreetal Cancer Screening Shigern Kanaoka, Ken-Ichi Yoshida, Mutsuhiro Ikuma, Naoyuki Mnira, Haruhiko Sugimura, Masayoshi Kajimum
PURPOSE: The etiology of colorectaI cancer (CRC) favors an early detection strategy, thus the objectwe of our study is to identi[} ~ a panel of multiple protein biortmrkers in patient serum for early detection of CRC using SELDI-TOF-MS proteomic technologies. METHODS: The SELDI-TOF-MS (Surface Enhanced Laser DesorptionBonization Time of Fhght Mass Spectrometry) ProteinChip from Ciphergen Biosystems Inc. allows differential capture and protein profiling of biological mixtures. Serum samples from patients v,4th sporadic colon cancer (pre-treatment, n = 20), health)' individuals and patients with benign colon diseases (n = 14), and patients with adenotamous polyps (n = 6) were processed for SELDI-TOFMS armlysis. Specimens were applied in duplicate to a Bioprocessor containing IMAC3copper ProteinChlp arrays. All loading and processing steps were automated using the Biomek 2000 robot Clustering and classification analyses were performed using the Ciphergen Biomarker Wizard and Biotnarker Patterns software packages, respectively. RESULTS: In the cross-validation analysis, colon cancer was predicted with sen~invities of 91-95% and
AGA
Abstracts
Backgrmmd: Fecal occult blood test (FOBT) has become an accepted technique of noninvasive screening for colorectal tumors, but lack of both sensitivity and specificity remains a problem, Early clinical studies with muhi-target DNA-hased stool assay suggest high sensitivity for colorectal cancer (CRC) while maintaining high specificity; however, clinical study with RNA-hased stool assay for colorectal cancer has not been reported. We aimed to develop RNA-based stool assay for CRC detection with high sensitivity and specificity, We picked COX-2 becanse it is overexpressed in most CRC, it theoretically provide an optimal molecular marker for detecting CRC Methods: Standard histological techniques were used to stage
A-650