Enhanced CD24 Expression in Colorectal Cancer Correlates with Prognostic Factors

Transcription

1 The Korean Journal of Pathology 2006; 40: Enhanced CD24 Exression in Colorectal Cancer Correlates with Prognostic Factors Yoon-La Choi 5 Yan Hua Xuan 1,7 Sang-Jeon Lee 2 Seon Mee Park 3 Wun Jae Kim 4 Hee-Jin Kim 6 Seok-Hyung Kim 1 1 Deartment of Pathology, 2 Deartment of Surgery, 3 Deartment of Internal Medicine, 4 Deartment of Urology, Chungbuk National University College of Medicine, Cheongju; 5 Deartment of Pathology, 6 Deartment of Laboratory Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea; 7 Deartment of Pathology, Yanbian University College of Medicine, Yanji, China Received : January 3, 2006 Acceted : March 14, 2006 Corresonding Author Seok-Hyung Kim, M.D. Deartment of Pathology, College of Medicine, Chungbuk National University, 62 Gaesin-dong, Heungduk-gu, Cheongju , Korea Tel: Fax: latoshkim@chungbuk.ac.kr Background : CD24 was originally described as a B cell-secific marker, however its aberrant exression in various solid tumors has recently been reorted. Our objective was to determine the attern and extent of the CD24 exression in colorectal cancer and its related lesions, and to clarify its correlation with clinico-athological arameters and esecially those associated with atients rognoses. Methods : A total of 307 colorectal cancers and the related lesions (150 carcinomas, 30 high-grade adenomas, 49 low-grade adenomas, 41 hyerlastic olys, and 37 normal colorectal eithelia) were immunohistochemically analyzed by treating CD24 monoclonal antibody onto tissue embedded araffin blocks. Results : CD24 exression was very rarely observed in the normal eithelia, hyerlastic olys, and low-grade adenomas; however, in high-grade adenomas, the CD24 exression was shown to be mildly increased in the cytolasm (13.3%). In carcinomas, the CD24 exression was increased substantially in both the membrane (38.0%) and the cytolasm (44.7%). The exression of CD24 in the membrane was ositively correlated with tumor size (<0.01). The CD24 exression in the cytolasm was ositively correlated with several unfavorable arameters, including a larger tumor size (<0.01), a higher tumor grade (<0.01), a higher rate of tumor invasion (<0.05), and a higher TNM stage (<0.05). Conclusion : High levels of CD24 exression in the membrane and cytolasm were characteristic in colorectal cancer, and the cytolasmic CD24 exression was correlated with several unfavorable clinical arameters. Key Words : CD24; Colorectal neolasm; Immunohistochemistry CD24 consists of a small rotein core that contains 27 amino acids. It is extensively glycosylated and then bound to the membrane via a hoshatidylinositol anchor. 1,2 CD24 was originally described as a B cell-secific marker that was exressed at the early stages of B cell develoment. Subsequent studies have demonstrated a high degree of CD24 exression on neutrohils and renal tubular eithelial cells, and several studies have also shown that CD24 can be exressed on several solid (non-hematooietic) tumors including small cell carcinoma, 3 neuroblastoma, 4 renal cell carcinoma, 5 ovary serous adenocarcinoma, 6 and rostate adenocarcinoma. 7,8 Colorectal carcinoma ranks the second most common cause of cancer deaths in Westernized countries. With an estimated 134,000 new cases er year and about 55,000 deaths er year occurring in association with this condition, the disease is resonsible for 10% of all-cancer related deaths in the United States. 9 There are only a limited number of reorts on the CD24 exression in colorectal cancer, and there have been no comrehensive studies designed to determine the clinical significance of the CD24 exression in the context of colorectal cancer. Akashi et al. have demonstrated the CD24 exression in a few colorectal cancer cell lines with using RT-PCR. 10 Nestl et al. have demonstrated the exression of CD24 mrna in 10 cases of colorectal carcinoma via in situ hybridization. 11 Saito et al. have revealed an increased CD24 exression in two colon cancer secimens, with using suression subtractive hybridization (SSH). 12 However, these findings were not confirmed at the rotein level. In addition, all of these studies were conducted on an extremely small scale, 103

2 104 Yoon-La Choi Yan Hua Xuan Sang-Jeon Lee, et al. and the results were not correlated with the clinico-athological findings. On the other hand, the significance of CD24 exression has recently been investigated in several hematological and non-hematological malignancies other than colon cancer, in terms of both tumor rogression and the ultimate rognoses. 13 In cases of ovarian cancer, high levels of cytolasmic CD24 exression were correlated with unfavorable clinical arameters and oor rognoses. 6,14 In addition, the uregulation of CD24 is considered to be an indeendent rognostic marker in the cases of lung, 15 breast, 16,17 and rostate cancer. 7,8 In this regard, we conducted a comrehensive study of CD24 exression at the rotein level via erforming immunohistochemistry in cases of colorectal cancer and its recancerous lesions to determine the correlation between the CD24 exression and a set of clinico-athologic arameters including the atient survival rates and survival durations. Our results demonstrated that aberrant and inaroriate CD24 exression in cases of colorectal cancer was significantly associated with the rogression of cancer. MATERIALS AND METHODS Patients, tissue samles, and reagents We investigated 307 cases of colorectal carcinoma and its related lesions, and these were obtained from the surgical athology files ket by the Deartment of Pathology of the Chungbuk National University Hosital. The criteria for inclusion were as follows: the histoathologic diagnosis of colorectal lesions, the availability of clinical follow-u data, and the availability of araffin-embedded tissue secimens. The selected cases included 150 cases of adenocarcinoma, 30 cases of high-grade adenoma, 49 cases of low-grade adenoma, 41 cases of hyerlastic olys, and 37 cases of normal colorectal eithelium. All of the cancer atients had undergone surgical oerations, and none had received either chemotheray or radiotheray before their surgical resections. The athology slides were reviewed in order to analyze the athologic arameters, including tumor size, the histological grade, deth of invasion, and the resence of nodal metastasis. The 150 colorectal carcinomas [atient age range=27-87 years; average age=64 years; 75 female and 75 male cases] encomassed 90 early cases (TNM stage I=24, TNM stage II=66), and 60 advanced cases (TNM stage III=44, TNM stage IV= 16). Of these cases, 28 cases (18.8%) were classified as well-differentiated adenocarcinomas, 107 cases (72.5%) as moderatelydifferentiated, and 13 cases (8.7%) as oorly-differentiated adenocarcinomas. TNM staging was assessed according to the staging system established by the American Joint Committee on Cancer (AJCC). 18 The tissue samles of benign lesions including the adenomas and hyerlastic olys, were mostly obtained from the indeendent atients and artly from the tissues in the vicinity of the main malignant tumor. The tissue samles of normal colorectal eithelium were obtained from tissues in the vicinity of main malignant tumor. We emloyed tissue microarray slides for the urose of erforming effective detection. For rearation of these slides, we unched tissue cores (3.0 mm in diameter) from the original blocks and then inserted them into the new araffin blocks (each of which contained 30 holes) that were reared to accet the tissue cores. The serially-sectioned slides were then roduced. Each tissue microarray slide (1 3 inches) held 30 secimens, which allowed us to analyze 30 secimens simultaneously with a minimum of variation during the rocess. Each secimen was round in shae, and it was 3.0 mm in diameter, this rovided a sufficient amount of tissue for the histoathologic analysis. All the archival materials had been routinely fixed in 10% neutral-buffered formalin and then embedded in araffin. Four micrometer-sections were reared on silane-coated slides (Sigma, St Louis, MO, USA). The immuno kits were urchased from DAKO, Inc. (Glostru, Denmark). The immunohistochemical rocedure The tissue sections in the microslides were dearaffinized with xylene, hydrated in series of graded alcohol solutions and they were immersed in 3% H 2 O 2 in order to quench the endogenous eroxidase activity. The sections were then microwaved in 40 mm Borate buffer (H 8.3) sulemented with 1 mm EDTA and 1 mm NaCl for 20 min for enhancing antigen retrieval. 19,20 After antigen retrieval, avidin and biotin were consecutively alied to the slides in order to revent any endogenous biotinrelated background. 21 The sections were then incubated with 1:100 diluted anti-cd24 monoclonal antibody (Ab-2, clone 24C02, Neomarkers, Fremont, CA) for 60 min; this was followed by three successive rinses with washing buffer and then further incubation with biotinylated goat anti-mouse Abs (DAKO) for an additional 20 min. After rinsing, the tissue sections were incubated with HRP-conjugated stretavidin (DAKO) for 20 min at room temerature. The slides were washed and the chromogen was develoed for 5 min with liquid 3,3 -diaminoben-

3 CD24 in Colorectal Tumors 105 zidine (DiNonA, Seoul, Korea). The slides were counter-stained with Meyer s hematoxylin, dehydrated, and then mounted with Canada balsam for examination. We used distilled water with % tween 20 as a rinsing solution. 22 Evaluation of results of immunohistochemical The scoring method of Sinicroe et al. (1995) 23 was alied for the evaluation of both the intensity of the immunohistochemical and the roortion of the stained eithelial cells. The membranous and nuclear s were indeendently analyzed. The intensity was subclassified as follows: 1, weak; 2, moderate; or 3, strong. The ositive cells were quantified as a ercentage of the total number of eithelial cells, and the results were assigned to one of five categories: 0, <5%; 1, 5-25%; 2, 26-50%; 3, 51-75%; and 4, >75%. The ercentage of ositivity of the tumor cells and the intensity were A B C then multilied in order to generate the immunoreactive score (IS) for each of the tumor secimens. The ositiviy of the immuno was determined when the IS was one or more. Each lesion was searately examined and scored by two athologists (Y.L.C & S.H.K), and those cases with discreant scores were discussed to obtain a consensus. Statistical analysis Statistical analyses were conducted with Fisher s exact tests, Pearson s 2 tests, ANOVA, Mann-Whitney tests, Tukey s HSD, and Duncan s test as a ost hoc test. With regard to the survival analysis, we analyzed 150 colorectal carcinoma atients via Kalan-Meier analyses. We used log rank tests in order to comare the different survival curves. Univariate and multivariate survival analyses were then conducted with using the Cox regression model. The models were adjusted for age and gender of the atients, the location, grade, and stage of the tumors, the chemotheray, and the membranous and cytolasmic exression levels of CD24. values less than 0.05 were regarded to be statistically significant. All statistical analyses were erformed using the SPSS software (SPSS, Chicago, USA). RESULTS D E F Fig. 1. Immunohistochemical of the microarray unches from the samles of colorectal adenocarcinoma and its related lesions. a, early stage adenocarcinoma; b, advanced stage adenocarcinoma; c, high grade adenoma; d, low grade adenoma; e, hyerlastic oly; f, normal colorectal mucosa. Diffuse and strong exression of CD24 was observed in adenocarcinoma (A & B). Focal and weak was observed in high-grade adenoma (C). No exression was noted in low-grade adenoma (D), hyerlastic oly (E), and normal colorectal mucosa (F). CD24 exression atterns in colorectal carcinoma and its related lesions In the normal colorectal mucosa and recancerous lesions, we found that the CD24 exression was either comletely absent or very low (Table 1). The CD24 membranous mostly showed aical localization rather than basolateral or circumferential attern, and no significant differences were evident for the membranous among the benign lesions (Table 1, Fig. 1, 3). No significant difference was also observed in the Table 1. Exression of CD24 in the colorectal carcinoma and its related lesions Mean of IS of CD24 Immuno Diagnosis No. P Membrane Cytolasmic Percentage of CD24 ositive cases Membrane Cytolasmic P Carcinoma ± ±2.15 < /150 (38.0%) 67/150 (44.7%) <0.001 Adenoma High grade ±8 7±6 1/30 (3.3%) 4/30 (13.3%) Adenoma Low grade ±4 0.02±4 1/49 (2.0%) 1/49 (2.0%) Hyerlasia /41 (0%) 0/41 (0%) Normal ±6 0.05±3 1/36 (2.7%) 2/36 (5.4%) IS, Immuno Score; P, robability.

4 106 Yoon-La Choi Yan Hua Xuan Sang-Jeon Lee, et al. CD24 cytolasmic among the benign lesions excet for the high grade adenoma. The exression rate was significantly higher in the high-grade adenoma than in the other benign lesions (<0.05) (Table 1, Fig. 1). Both the membranous and cytolasmic CD24 levels were significantly higher in the carcinoma than were those in any of the other benign lesions, including the recancerous lesions, in term s of both the average intensity (mean of the IS) and the exression rate (<0.001) (Table 1). Correlation of the CD24 exression with the clinico-athological arameters A C Fig. 2. Immunohistochemical of colorectal adenocarcinoma (high-ower view). a, adenocarcinoma TNM stage I; b, adenocarcinoma TNM stage II; c, adenocarcinoma TNM stage III; d, adenocarcinoma TNM stage IV. Diffuse and strong CD24 exression is seen in all cases of adenocarcinomas (A-D). Strong membranous CD24 along the aical surface is observed in earlier stages of adenocarcinoma (A & B). However, in advanced stage, membranous was determined to be much weaker than in earlier stage (C & D). B D Neither the membranous nor the cytolasmic CD24 exression was correlated with the atients age, gender, lymh node metastasis, or the location of main tumor (Table 2). The average intensity in both the membrane and the cytolasm was determined to be significantly higher in the cases with large tumor size (<0.05). Both the membranous and cytolasmic CD24 exressions were clearly associated with the grades of the tumors (<0.005); however, the details of their atterns were different. The membranous CD24 was significantly higher in the grade 2 (moderately differentiated) tumors than in grade 1 (well differentiated) tumors and in the grade 3 (oorly differentiated) tumors. By contrast, the cytolasmic increased roortionally with the grade of the tumor. Membranous CD24 exression was not found to be significantly correlated with direct tumor invasion or the TNM stage (clinical stage), whereas the cytolasmic exression was found to be increased in the cases with a higher degree of tumor invasion and All stage CD24m(+) CD24m(-) = A All stage CD24c(-) CD24c(+) = B Overall CD24m(-) Stage I & II CD24m(+) =889 Stage I & II CD24c(-) CD24c(+) =816 Overall CD24m(-) Stage III CD24m(+) = CD24c(-) Stage III CD24c(+) =678 Overall Stage IV =753 CD24m(+) CD24m(-) C Stage IV =407 CD24c(+) CD24c(-) D Fig. 3. Univariate survival analysis of CD24 exression in 150 cases of colorectal adenocarcinoma. Neither cytolasmic (A) nor membranous (B) showed a statistically significant correlation with rognosis. However, when membranous exression was analyzed searately in each grou of cancer stage (C), membranous was significantly associated with better rognosis in atients with stage III. On the other hand, cytolasmic was not associated with atients survival when analyzed searately by cancer stage (D).

5 CD24 in Colorectal Tumors 107 Table 2. Correlation of CD24 membranous and cytolasmic exression with clinicoathological features in colorectal carcinomas Variables No. Membrane Mean of IS Cytolasmic Membrane CD24(+) No. (%) Cytolasmic Age 60 years ± ± (39.6%) (49.0%) 24 <60 years ±2.07 2± (33.3%) 19 (35.2%) Sex Female 75 0± ± (40.0%) (46.7%) 11 Male ± ± (34.7%) 31 (41.3%) Tumor size <6.0 cm 96 7±1.88 <0.05 1±1.44 < (31.3%) (39.6%) cm ± ± (46.2%) 26 (50.0%) Node metastasis Negative ± ± (41.1%) (38.9%) 25 Positive 55 9± ± (3%) 29 (52.7%) Location Rectum 93 6± ± (33.3%) (41.9%) 35 Non-rectum ± ± (42.1%) 26 (45.6%) Grade I 28 5±4 < ±7 < (1%) < (14.3%) <0.005 II ± ± (44.9%) 55 (51.4%) III 13 2± ± (38.5%) 7 (53.8%) Tumor invasion I 6 7± ±0.00 < (16.7%) 8 0 (0 %) <0.05 II ±2.93 9± (31.6%) 7 (36.8%) III ± ± (38.8%) 55 (45.5%) IV 3 0± ± (33.3%) 3 (100%) TNM staging I ± ±1.28 < (29.2%) 8 6 (25.0%) <0.05 II ± ± (43.9%) 27 (4%) III 44 7±1.49 9± (27.3%) 21 (47.7%) IV ± ± (50.0%) 12 (75.0%) an advanced TNM stage; this finding was statistically significant (<0.05) (Table 2, Fig. 2). Univariate and multivariate survival analyses for the colorectal cancer atients The univariate analysis using the Kalan-Meier method revealed that gender, TNM stage, tumor grade, and chemotheray were associated with the atients survival (Table 3). The correlation of the membranous and cytolasmic CD24 exressions with the overall survival rate of the colorectal cancer atients was also analyzed by using the Kalan-Meier method (Table 3, Fig. 3A, B). However, neither the membranous nor cytolasmic CD24 exression were found to be associated with the atients survival with any statistical significance. We further analyzed the effect of the CD24 exression on the overall survival rate according to the TNM stage. The CD24 exression in the membrane (CD24m) was significantly correlated with a better rognosis for the atients with stage III but not for the atients with other stages (Fig. 3C), whereas the CD24 exression in the cytolasm (CD24c) was not associated with the atients survival in any stages (Fig. 3D). We also erformed univariate and multivariate analyses in order to assess the relative contribution of the imortant clinicoathologic arameters, as well as the CD24 exression, on the survival of the atients with using the Cox-roortional hazards regression model (Table 4). Secifically, gender, age, tumor location, chemotheray, tumor grade, and TNM stage were included as the clinico-athologic arameters. Among these, tumor grade, TNM stage, and the membranous CD24 exression were selected as indeendent rognostic factors. The membranous CD24 exression (CD24(m)) was significantly associated with a better rognoses (=0.01) (The ratio of risk=67), while the cytolasmic CD24 exression (CD24(c)) was not an indeendent rognostic factor (Table 4). We also searately erformed the univariate and multivariate analyses for each clinical stage searately (Table 5). Secifically, gender and chemotheray were clearly associated with the atients rognosis in stage III (Table 5). The

6 108 Yoon-La Choi Yan Hua Xuan Sang-Jeon Lee, et al. Table 3. Univariate analysis of the overall survival of colorectal cancer atients using Kalan-Meier survival analysis Variables No. of atients (%) 1-year 3-year membranous CD24 exression (CD24(m)) was not associated with the atients rognosis in any of the stages (Table 5). The cytolasmic CD24 exression (CD24(c)) was correlated with the atients better rognosis in stage IV (RR=0.048, =0.017) but not in other stages (Table 5). DISCUSSION - value Sex Male Female Age < TNM stage I-II III IV <.0001 Tumor size < Location Colon Rectum Tumor grade I II III <.0001 Chemotheray No Yes CD24 (m) < CD24 (c) < CD24 as a marker of rogressions in colorectal neolasm In this study, we showed that the cytolasmic and membranous exressions of CD24 were rare in the normal, benign, and remalignant lesions, but we also noted an abrut rise in the rates of both exressions in cases of colorectal cancer. In addition, cytolasmic CD24 was ositively correlated with the cancer stage. Considering the adenoma-carcinoma sequence in colorectal carcinogenesis and the restriction of the CD24 exression to the carcinoma, CD24 can be considered to be a marker of rogression during the colorectal carcinogenesis, and CD24 can be used as a diagnostic marker to differentiate the carcinoma from benign recursor lesions including high grade adenoma. The develoment of carcinoma from adenomatous lesions is referred to as the adenoma-carcinoma sequence and this has been well documented for colorectal carcinoma. 24 The functional imlication of the CD24 exression in terms of carcinogenesis needs to be elucidated. Is the CD24 exression a rognostic factor in colorectal cancer? Considering the results of the univariate and multivariate survival analyses, it is unlikely that the cytolasmic CD24 exression is a rognostic factor for colorectal cancer desite its clear correlation with several rognosis-related clinico-athological arameters. The ositive correlation of the cytolasmic CD24 exression with a better rognosis for stage IV cancer atients is shown in Table 5, but its significance remains to be determined. However, the rognostic significance of membranous CD24 is more comlicated. According to the result of the Kalan-Meier survival analysis, the membranous CD24 exression seemed to be clearly correlated with better survival for the tage III atients (Fig. 4C). Multivariate analysis using the Cox-roortional hazards regression model including almost all the rognostic factor as covariates revealed that the membranous CD24 exression was an indeendent rognostic factor (Table 4). However, when the survival analysis was done searately for each grou of cancer stages and adjusted with several covariates via the Cox-roortional hazards regression model, the CD24 membranous exression was not associated with the atients rognosis in any of the cancer stages (Table 5). Collectively, it is still questionable whether the membranous CD24 exression is an indeendent rognostic factor for colorectal cancer. Why was the CD24 cytolasmic exression not associated with atients survival desite its clear correlation with rognosis-related clinico-athological arameters? The cytolasmic CD24 exression was obviously correlated with a high tumor grade, aggressive tumor invasion, and an advanced stage (Table 2). The question then is: why was the cytolasmic CD24 exression was not associated with the atients survival?. One of the ossible reasons for this discreancy can be a confounding factor such as the different treatments after surgery, such as chemotheray. However, since almost all covari-

7 CD24 in Colorectal Tumors 109 Table 4. Univariate and multivariate analyses for rognostic variables of overall survival of colorectal cancer atients using Cox-roortional hazards regression Predictors Univariate Ratio of risk (95% CI) value Multivariate Ratio of risk (95% CI) value Sex Male 0 0 Female 59 (10-06) 25 ( ) Age ( ) ( ) Location Rectum 1 1 Colon 50 ( ) 13 ( ) Chemotheray Not done 1 1 Done ( ) 10 ( ) Tumor grade I 0 0 II ( ) ( ) III ( ) ( ) TNM staging I 0 0 II ( ) (61-343) III ( ) ( ) IV ( ) ( ) CD24(m) Negative 0 0 Positive 39 (80-38) 67 (71-88) CD24(c) Negative 0 0 Positive ( ) ( ) h(t/x)=h0(t)ex( 0+ 1sex+ 2age+ 2location+ 4grade+ 5chemoTx+ 6stage+ 7CD24). ates, including chemotheray, that can affect the atients survival were controlled and adjusted for by using Cox-roortional hazard regression model, it is unlikely that this discreancy was caused by the confounding factors such as chemotheray. The essential reason for this discreancy may be the absence of correlation with lymh nodal metastasis rather than the confounding factors. The cytolasmic CD24 exression was correlated with direct tumor invasion and differentiation of tumor cells rather than lymh nodal metastasis (Table 2). Considering that most of the atients in this study underwent comlete excision, nodal metastasis might have exerted a greater influence on atients rognosis rather than that of direct tumor invasion. The relationshi between nodal metastasis and the atients survival has been suggested in other studies. In breast cancer, the CD24 exression did not correlate with direct tumor invasion or tumor cell differentiation; however, it was associated with nodal metastasis and shortened overall survival for the atients suffering with breast cancer. 13 Comarison of our results with those in revious studies Significant rates of CD24 ositivity have been reorted for many common human tumors. Furthermore, for several tyes of tumor entities, higher rates of CD24 exression or CD24 ositivity have been significantly associated with a shorter survival of atients. 13 And CD24 exression has been reorted in colorectal cancer, however, the clinical significance of the CD24 exression varies considerably deending on the tyes of tissues. According to Weichert et al., CD24 correlates with atients survival only when the CD24 is highly exressed. In cases of ancreatic cancer and heatocellular carcinoma, there was no correlation between the exression of CD24 and the attenuation of the overall survival of the atients, 28,29 and our results were in line with these findings. There have been few reorts that have evaluated the membranous and cytolasmic CD24 exressions searately to determine their clinical significance, and these limited studies reorted no

8 110 Yoon-La Choi Yan Hua Xuan Sang-Jeon Lee, et al. Table 5. Univariate and multivariate analysis of the overall survival of atients with stage I-II, III, and IV colorectal cancer using Coxroortional hazards regression Stage I & II Univariate RR (95% CI) Multivariate RR (95% CI) Stage IV Multivariate RR (95% CI) Predictors Univariate RR (95% CI) Univariate RR (95% CI) Stage III Multivariate RR (95% CI) Sex Male Female 74 (60-39 (12-18 (67-07 (14-32 ( ( ) 2.822) 50) 25) 2.563) 5.506) Age (04-39 ( ( ( ( ( ) 2.513) 2.754) 3.268) 156) ) Location Rectum Colon 61 ( ( (23-86 (46-75 (31-32 ( ) 1.282) 1.517) 1.911) 2.867) 2.787) Chemotheray Not done Done ( ( ( ( ( ( ) ) 47) 02) 3.876) ) CD24(m) Negative Positive 62 (04-30 (90-43 ( ( (92-34 ( ) 2.085) 49) 1.838) 1.648) 1.557) CD24(c) Negative Positive ( (63-76 ( (29-75 ( ( ) 4.298) 2.115) 2.854) 1.884) 78) h(t/x)=h0(t)ex( 0+ 1sex+ 2age+ 3location+ 4chemoTx+ 5CD24). aarent relationshi between the membranous CD24 exression and favorable rognoses. Kristiansen et al. 14 have shown that erforming immuno of ovarian cancer revealed two distinct modalities; membranous immunoreactivity and an additional cytolasmic. In this study, membranous CD24 exression was found to be absent in the normal ovarian surface eithelium, but it was frequently observed in the cases of invasive carcinoma, with no correlation to the clinicoathological arameters. However, the cytolasmic CD24 exression was correlated with an attenuation of atient survival on both univariate and multivariate analyses, although there was no correlation with tumor grade, the T stage, or the M status. 14 Our results were generally in accordance with these findings with regard to the significance of the cytolasmic and membranous CD24 exression. In summary, we showed that CD24 is abundantly exressed in colorectal cancer tissues, and that the cytolasmic and membranous overexression of CD24 can be a secific marker of disease rogression from the remalignant tumor hase to the develoment of carcinoma. The cytolasmic CD24 exression was associated with a variety of clinicoathological arameters related to oor rognoses. Although the biological function of CD24 and its clinical significance are not comletely understood, our results clearly demonstrated the diagnostic and theraeutic usefulness of the CD24 rotein in the context of colorectal cancer. ACKNOWLEDGEMENT This work was suorted by a Chungbuk National University Research grant. We wish to thank Jung Sun Lee and Mee-Young Shim for their technical assistance in the rearation of the manuscrit. REFERENCES 1. Pirruccello SJ, LeBien TW. The human B cell-associated antigen CD24

9 CD24 in Colorectal Tumors 111 is a single chain sialoglycorotein. J Immunol 1986; 136: Fischer GF, Majdic O, Gadd S, Kna W. Signal transduction in lymhocytic and myeloid cells via CD24, a new member of hoshoinositol-anchored membrane molecules. J Immunol 1990; 144: Jackson D, Waibel R, Weber E, Bell J, Stahel RA. CD24, a signal-transducing molecules exressed on human B cells, is a major surface antigen on small cell lung carcinomas. Cancer Res 1992; 52: Mechtersheimer G, Barth T, Ludwig R, Staudter M, Moller P. Differential exression of leukocyte differentiation antigens in small round blue cell sarcomas. Cancer 1993; 71: Droz D, Zachar D, Charbit L, Gogusev J, Chretein Y, Iris L. Exression of the human nehron differentiation molecules in renal cell carcinomas. Am J Pathol 1990; 137: Choi YL, Kim SH, Shin YK, et al. Cytolasmic CD24 exression in advanced ovarian serous borderline tumors. Gynecol Oncol 2005; 97: Kristiansen G, Pilarsky C, Pervan J, et al. CD24 exression is a significant redictor of PSA relase and oor rognosis in low grade or organ confined rostate cancer. Prostate 2004; 58: Kristiansen G, Pilarsky C, Wissmann C, et al. Exression rofiling of microdissected matched rostate cancer samles reveals CD1666/ MEMD and CD24 as new rognostic markers for atient survival. J Pathol 2005; 205: Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, CA Cancer J Clin 1998; 48: Akashi T, Shirasawa T, Hirokawa K. Gene exression of CD24 core olyetide molecule in normal rat tissues and human tumor cell lines. Virchows Arch 1994; 425: Nestl A, Von Stein OD, Zatloukal K, et al. Gene exression atterns associated with the metastatic henotye in rodent and human tumors. Cancer Res 2001; 61: Saito A, Fujii G, Sato Y, Gotoh M, Sakamoto M, Toda G, Hirohashi S. Detection of genes exressed in rimary colon cancers by in situ hybridization: overexression of RACK1. J Clin Pathol 2002; 55: Kristiansen G, Sammar M, Altevogt P. Tumour biological asects of CD24, a mucin-like adhesion molecule. J Mol Histol 2004; 35: Kristiansen G, Denkert C, Schluns K, Dahl E, Pilarsky C, Hautmann S. CD24 is exressed in ovarian cancer and is a new indeendent rognostic marker of atient survival. Am J Pathol 2002; 161: Kristiansen G, Schluns K, Yongwei Y, Denkert C, Dietel M, Petersen I. CD24 is an indeendent rognostic marker of survival in nonsmall cell lung cancer atients. Br J Cancer 2003; 88: Fogel M, Friederichs J, Zeller Y, et al. CD24 is a marker for human breast carcinoma. Cancer Lett 1999; 143: Kristiansen G, Winzer KJ, Mayordomo E, et al. CD24 exression is a new rognostic marker in breast cancer. Clin Cancer Res 2003; 9: Greene FL, Page DL, Fleming ID, et al. American Joint Committee on Cancer Staging Manual 6th ed. Sringer 2002; Kim SH, Kook MC, Song HG. Otimal conditions for the retrieval of CD4 and CD8 antigens in formalin-fixed, araffin-embedded tissues. J Mol Histol 2004; 35: Kim SH, Kook MC, Shin YK, Park SH, Song HG. Evaluation of antigen retrieval buffer systems. J Mol Histol 2004; 35: Kim SH, Jung KC, Shin YK, et al. The enhanced reactivity of endogenous biotin-like molecules by antigen retrieval rocedures and signal amlification with tyramine. Histochem J 2002; 34: Kim SH, Shin YK, Lee KM, Lee JS, Yun JH, Lee SM. An imroved rotocol of biotinylated tyramine-based immunohistochemistry minimizing nonsecific background. J Histochem Cytochem 2003; 51: Sinicroe FA, Ruan SB, Cleary KR, Stehens LC, Lee JJ, Levin B. bcl-2 and 53 oncorotein exression during colorectal tumorigenesis. Cancer Res 1995; 55: Cho KR, Vogelstein B. Genetic alterations in adenoma-carcinoma sequence. Cancer 1992; 70: Lim SC, Oh SH. The role of CD24 in various human eithelial neolasias. Pathol Res Pract 2005; 201: Lee SK, Lim SC, Kim KJ. The role of CD24 in colonic adenocarcinoma. J Korean Soc Coloroctal 2005; 21: Weichert W, Denkert C, Burkhardt M, et al. Cytolasimic CD24 exression in colorectal cancer indeendently correlates with shortened atient survival. Clin Cancer Res 2005; 11: Jacob J, Bellach J, Grutzmann R, et al. Exression of CD24 in adenocarcinomas of the ancreas correlates wit higher tumor grades. Pancreatology 2004; 4: Huang LR, Hsu HC. Cloning and exression of CD24 gene in human heatocellular carcinoma: otential early tumor marker gene correlates with 53 mutaion and tumor differentiation. Cancer Res 1995; 55: