J Cancer Res Clin Oncol (2014) 140:109–116 DOI 10.1007/s00432-013-1536-8
ORIGINAL PAPER
Low CADM2 expression predicts high recurrence risk of hepatocellular carcinoma patients after hepatectomy Sen Yang · Hong‑Li Yan · Qi‑Fei Tao · Sheng‑Xian Yuan · Guan‑Nan Tang · Yuan Yang · Li‑Li Wang · Yi‑Liang Zhang · Shu‑Han Sun · Wei‑Ping Zhou
Received: 12 June 2013 / Accepted: 4 October 2013 / Published online: 17 November 2013 © Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To investigate the expression and clinical significance of CADM2 in hepatocellular carcinomas (HCC). Methods The level of expression of CADM2 mRNA was assessed in frozen tumor specimens and adjacent noncancerous tissues from 30 HCC patients by real-time PCR. The protein level was determined by immunohistochemistry on a tissue microarray containing tumor and adjacent noncancerous tissues from 234 HCC patients. Clinicopathological characteristics associated analysis was performed through SPSS18. Results CADM2 was strikingly down regulated in HCC. CADM2 expression was associated with differentiation (P = 0.000), serum alpha-fetoprotein (P = 0.003), vascular invasion (P = 0.001), and hepatitis B surface antigen (HBsAg, P = 0.038). Furthermore, patients with low CADM2 expression had significantly poorer recurrencefree survival (RFS) (40.8 and 34.2 % vs. 56.3 and 50.1 % in 3- and 5-year RFS, respectively, P = 0.005). Subgroup Sen Yang, Hong-Li Yan, and Qi-Fei Tao have contributed equally to this study. Electronic supplementary material The online version of this article (doi:10.1007/s00432-013-1536-8) contains supplementary material, which is available to authorized users. S. Yang · Q.-F. Tao · S.-X. Yuan · Y. Yang · L.-L. Wang · W.-P. Zhou (*) The Third Department of Hepatic Surgery, Eastern Hepatobiliary Surgery Hospital, Second Military Medical University, 225 Changhai Road, Shanghai 200438, China e-mail: [email protected] H.-L. Yan · G.-N. Tang · Y.-L. Zhang · S.-H. Sun (*) Department of Medical Genetics, Second Military Medical University, 800 Xiangyin Road, Shanghai 200433, China e-mail: [email protected]
analysis revealed that the difference in RFS between groups with low- and high-CADM2 expression still existed among patients belonging to stage 0 or A of BCLC staging system (P = 0.008), patients with tumor ≤5 cm in size (P = 0.013), and alpha-fetoprotein-negative patients (P = 0.003). Moreover, low expression was more frequently observed in the early recurrence group (within 2 years, P = 0.007). Further multivariate Cox regression analysis indicated that CADM2 expression level, tumor size, tumor number, vascular invasion, HBsAg were independent risk factors for HCC recurrence. Conclusion CADM2 serves as a novel predictor of RFS in HCC patients after curative resection. Keywords CADM2 · Liver neoplasms · Relapse · Immunohistochemistry · Clinical significance
Introduction Hepatocellular carcinoma (HCC) is one of the most popular malignant carcinomas with high mortality worldwide and has higher morbidity and mortality in China because of the popularity of HBV-induced hepatitis (Ferlay et al. 2010; ElSerag and Rudolph 2007). Nowadays surgical resection is still the main method of therapy; however, it does not have favorable effect with 5-year overall survival rate about 50 %. High recurrence rate after hepatectomy contributes greatly to the poor prognosis with 5-year recurrence rate of about 70 % (Takayama 2011; Fong et al. 1999; Grazi et al. 2001). Many factors such as tumors’ biological characters, stage of the tumor, the condition of the liver, and the treatment procedure are associated with recurrence (Ju et al. 2009; Wu et al. 2012; Huang et al. 2013). Genes’ expression levels that reflect on tumors’ biological characters were reported to
13
110
have the ability to predict HCC recurrence risk (Mann et al. 2007; Hoshida et al. 2008; Hu et al. 2009). These genes can be used as biomarkers to guide the usage of appropriate treatments to prevent recurrence. However, few biomarkers have been identified to predict HCC recurrence risk. Recent studies indicated that cell adhesion molecules (CADMs), a newly identified protein family, might serve as tumor suppressors. Most CADMs belong to an immunoglobulin superfamily, the members of which express three extracellular immunoglobulin-like loops, a transmembrane region, and an intracellular domain (Takai et al. 2003; Murakami 2005). They were found to be lost or at strikingly low levels in many kinds of malignant carcinomas. For example, CADM1 (also known as Necl-2, TSLC1, cynCAM1, IGSF4A) has been reported to be reduced in lung cancer (Kuramochi et al. 2001), prostate cancer (Fukuhara et al. 2002), esophageal cancer (Ito et al. 2003), and breast cancer (Takahashi et al. 2012). And several articles have reported that CADM3 (also known as Necl-1, TSLL1, cynCAM3, IGSF4B) and CADM4 (also known as Necl-4, TSLL2, cynCAM4, IGSF4C) also functions as tumor suppressor in various types of cancer cells (Williams et al. 2006; Raveh et al. 2009; Nagata et al. 2012). However, CADM2 (also known as Necl-3, cynCAM2, IGSF4D) has not been well characterized in cancer (Chang et al. 2010), and there have been no reports to date regarding the expression and clinical significance of CADM2 in HCC. Our previous HCC expression profile genechip indicated CADM2 expression was strikingly reduced in tumor tissues (unpublished data). Therefore, we further investigated the expression and clinical significance of CADM2 in HCC using high-throughput tissue microarray technology and rich HCC case resource in Eastern Hepatobiliary Surgery Hospital (Shanghai China). We found that CADM2 is strikingly down regulated in HCC and can serve as a novel predictor for recurrence-free survival (RFS) of HCC patients after hepatectomy.
Materials and methods Patients and clinical sample Under the inclusion and exclusion criteria: (a) with frozen tumor tissue and paired noncancerous liver tissue stored (for mRNA examination) or with paraffin tissue blocks stored (for tissue microarray construction) in our hospital; (b) underwent curative resection during June and July 2011 (for mRNA examination) or between May 2004 and June 2006 (for tissue microarray construction); (c) distinctive pathologic diagnosis of HCC; (d) without anticancer treatment and distant metastasis before surgery, 300 patients were identified as the base population for mRNA examination,
13
J Cancer Res Clin Oncol (2014) 140:109–116
and 1,020 patients were identified for tissue microarray construction. Finally, 30 patients were randomly selected from its corresponding base population using random number for mRNA examination. Similarly 242 patients were selected for tissue microarray construction. The adjacent noncancerous tissues were isolated at least 2 cm away from the tumor tissue border. For the cohort of 242 HCC patients, complete clinicopathological and follow-up data are available. These patients’ clinicopathological data are presented in Table 1. Tumor differentiation was defined according to Edmondson’s grading system (I, welldifferentiated; II, moderately differentiated; III, poorly differentiated; IV, undifferentiated) (Edmondson and Steiner 1954). Micrometastases were defined as tumors adjacent to the border of the main tumor that could be observed only under the microscope. The Barcelona Clinic Liver Cancer staging system (BCLC) ranked hepatocellular carcinomas in five stages, ranging from 0 (very early stage) to 5 (terminal stage) (Llovet et al. 1999). The day of surgery was considered day 0 in calculating postoperative survival. Followup ranged from 1 to 80 months. The study was approved by the institutional review board of the Eastern Hepatobiliary Surgery Hospital, and all patients provided written informed consent to participate in the study. Any information contained in this article cannot identify the patients. Examination of the mRNA expression level of CADM2 Total RNA of the 30 patients’ tumor tissues and adjacent noncancerous tissues was isolated using TRIzol reagent (Invitrogen CA), and first strand cDNA was generated using random primers and M-MLV reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. Real-time PCR was performed using StepOne™ Real-Time PCR System (Applied Biosystems, Foster City, USA) and SYBR® Green (Takara, Dalian, P. R. China). The CADM2-specific primers were 5′-CCTCAATGCCACCCCTCAG-3′ (forward) and 5′-TTCTCCGCCATCCTTTGTCC-3′ (reverse), and the GAPDH-specific primers were 5′-CGGATTTGGTCGTATTGGG-3′ (forward) and 5′-CTGGAAGATGGTGATGGGATT-3′ (reverse). PCR condition was 95 °C for 5 min, followed by 40 cycles at 95 °C for 15 s, 58 °C for 15 s, 72 °C for 20 s. Gene expression in each sample was normalized to relative to that of GAPDH mRNA. Tissue microarray construction and immunohistochemistry This tissue microarray which was constructed by Shanghai Biochip Corporation (China) contains 242 patients’ tumor tissues and adjacent noncancerous tissues. Each tissue has two repetitions on the tissue microarray. The tissue microarray slides were stained using EnVision system (Dako Danmark)
111
J Cancer Res Clin Oncol (2014) 140:109–116 Table 1 Association between CADM2 protein level and clinicopathological characteristics χ2
Clinical characteristics
CADM2 Lowa
Higha
All cases Gentle Male Female Age ≥50 5 cm ≤5 cm Tumor number 1 2 3 4 Micrometastases No Yes Vascular invasion Negative Positive microscopically Positive macroscopically HBsAg Negative Positive HBeAg Negative Positive BCLC stage 0 A B C
46 98
46 44
7 137
18 72
62 82
41 49
117 24 2 1
73 13 3 1
34 110
31 59
48 77 19
51 35 4
12 132
16 74
108 36
75 15
8 101 16
10 64 12
19
4
P value
0.853
0.409
2.975
0.105
8.528
0.003
13.302
0.000
0.140
0.787
1.266
0.737
3.240
0.098
13.902
0.001
4.690
0.038
2.257
0.146
6.772
0.080
Italic values are statistically significant at (P 20 ng/ml a
Patients whose adjacent noncancerous tissues’ immunohistochemistry score minus tumor tissues’ immunohistochemistry score is less than 1 belonged to high-expression group, and the other patients belonged to low-expression group
Fig. 1 Expression of CADM2 in HCC tumor tissues and adjacent noncancerous tissues. a mRNA level of CADM2 in 30 HCC patients’ tumor tissues and adjacent tumor tissues. Each bar shows the mRNA level ratio of adjacent noncancerous tissue to tumor tissue (N/T). b Immunohistochemistry score distribution of the 234 HCC patients’ tumor tissues (T) and adjacent noncancerous tissues (N). Numbers at the bar refer to the case number of patients within the immunohistochemistry score range. c Distribution of difference between tumors’ and paired noncancerous tissues’ immunohistochemistry score. Numbers at the bar refer to the case number of patients within the specific range. Difference means adjacent noncancerous tissues’ immunohistochemistry score minus paired tumor tissues’ immunohistochemistry score (N–T)
according to the manufacturer’ instruction with a rabbit polyclonal antibody as the primary antibody (Abnover Taiwan) (diluted 1:100). Analyzing the tissue microarray was performed by two experienced researchers. Immunohistochemistry score was mainly depended on the intensity. No staining is scored 0, weak staining is scored 1, moderate staining is scored 2, strong staining is scored 3, and strikingly strong staining is scored 4. Representative pictures of respective semiquantitative score are shown in Fig. S1. Each tissue’s
13
112
J Cancer Res Clin Oncol (2014) 140:109–116
Fig. 2 Representative pictures of the immunohistochemistry of CADM2. The upper four pictures are from HCC tumors, and the lower four pictures are from their paired adjacent noncancerous tissues,
respectively. The left four pictures belong to patient 1, and the right four pictures belong to patient 2. The magnification of a/e/c/g is ×50, while b/d/f/h are their corresponding pictures with magnification ×100
four scoring results including two repetitions analyzed by two researchers were finally averaged.
patients through real-time PCR. The result indicated that CADM2 mRNA level in tumor tissues was significantly lower than in adjacent noncancerous tissues (Fig. 1a). We further examined the protein level of CADM2 through immunohistochemistry on a tissue microarray including 234 valid HCC patients’ tumor tissues and adjacent noncancerous tissues (8 patients were excluded for their incomplete tissue cores on the tissue microarray). We found that tumors had significantly weaker immunohistochemistry staining than adjacent noncancerous tissues (Figs. 1b, c, 2). Further analysis revealed 61.5 % patients (144 patients) showed lower CADM2 protein level in tumor tissues than in adjacent noncancerous tissues (the difference of immunohistochemistry score ≥1), and 24.8 % patients (58 patients) showed significantly lower CADM2 protein level in tumor tissues than in adjacent noncancerous tissues (the difference of immunohistochemistry score ≥2) (Fig. 1c).
Statistical analysis All statistical analyses were performed by SPSS18. The χ2 test was used to evaluate the association between CADM2 expression level and clinicopathological characteristics. The cumulative survival probability was evaluated using the Kaplan–Meier method, and the significance of differences was assessed using the log-rank test. To determine independent prognostic factors, the Cox multivariate regression analysis was used. A P
ORIGINAL PAPER
Low CADM2 expression predicts high recurrence risk of hepatocellular carcinoma patients after hepatectomy Sen Yang · Hong‑Li Yan · Qi‑Fei Tao · Sheng‑Xian Yuan · Guan‑Nan Tang · Yuan Yang · Li‑Li Wang · Yi‑Liang Zhang · Shu‑Han Sun · Wei‑Ping Zhou
Received: 12 June 2013 / Accepted: 4 October 2013 / Published online: 17 November 2013 © Springer-Verlag Berlin Heidelberg 2013
Abstract Purpose To investigate the expression and clinical significance of CADM2 in hepatocellular carcinomas (HCC). Methods The level of expression of CADM2 mRNA was assessed in frozen tumor specimens and adjacent noncancerous tissues from 30 HCC patients by real-time PCR. The protein level was determined by immunohistochemistry on a tissue microarray containing tumor and adjacent noncancerous tissues from 234 HCC patients. Clinicopathological characteristics associated analysis was performed through SPSS18. Results CADM2 was strikingly down regulated in HCC. CADM2 expression was associated with differentiation (P = 0.000), serum alpha-fetoprotein (P = 0.003), vascular invasion (P = 0.001), and hepatitis B surface antigen (HBsAg, P = 0.038). Furthermore, patients with low CADM2 expression had significantly poorer recurrencefree survival (RFS) (40.8 and 34.2 % vs. 56.3 and 50.1 % in 3- and 5-year RFS, respectively, P = 0.005). Subgroup Sen Yang, Hong-Li Yan, and Qi-Fei Tao have contributed equally to this study. Electronic supplementary material The online version of this article (doi:10.1007/s00432-013-1536-8) contains supplementary material, which is available to authorized users. S. Yang · Q.-F. Tao · S.-X. Yuan · Y. Yang · L.-L. Wang · W.-P. Zhou (*) The Third Department of Hepatic Surgery, Eastern Hepatobiliary Surgery Hospital, Second Military Medical University, 225 Changhai Road, Shanghai 200438, China e-mail: [email protected] H.-L. Yan · G.-N. Tang · Y.-L. Zhang · S.-H. Sun (*) Department of Medical Genetics, Second Military Medical University, 800 Xiangyin Road, Shanghai 200433, China e-mail: [email protected]
analysis revealed that the difference in RFS between groups with low- and high-CADM2 expression still existed among patients belonging to stage 0 or A of BCLC staging system (P = 0.008), patients with tumor ≤5 cm in size (P = 0.013), and alpha-fetoprotein-negative patients (P = 0.003). Moreover, low expression was more frequently observed in the early recurrence group (within 2 years, P = 0.007). Further multivariate Cox regression analysis indicated that CADM2 expression level, tumor size, tumor number, vascular invasion, HBsAg were independent risk factors for HCC recurrence. Conclusion CADM2 serves as a novel predictor of RFS in HCC patients after curative resection. Keywords CADM2 · Liver neoplasms · Relapse · Immunohistochemistry · Clinical significance
Introduction Hepatocellular carcinoma (HCC) is one of the most popular malignant carcinomas with high mortality worldwide and has higher morbidity and mortality in China because of the popularity of HBV-induced hepatitis (Ferlay et al. 2010; ElSerag and Rudolph 2007). Nowadays surgical resection is still the main method of therapy; however, it does not have favorable effect with 5-year overall survival rate about 50 %. High recurrence rate after hepatectomy contributes greatly to the poor prognosis with 5-year recurrence rate of about 70 % (Takayama 2011; Fong et al. 1999; Grazi et al. 2001). Many factors such as tumors’ biological characters, stage of the tumor, the condition of the liver, and the treatment procedure are associated with recurrence (Ju et al. 2009; Wu et al. 2012; Huang et al. 2013). Genes’ expression levels that reflect on tumors’ biological characters were reported to
13
110
have the ability to predict HCC recurrence risk (Mann et al. 2007; Hoshida et al. 2008; Hu et al. 2009). These genes can be used as biomarkers to guide the usage of appropriate treatments to prevent recurrence. However, few biomarkers have been identified to predict HCC recurrence risk. Recent studies indicated that cell adhesion molecules (CADMs), a newly identified protein family, might serve as tumor suppressors. Most CADMs belong to an immunoglobulin superfamily, the members of which express three extracellular immunoglobulin-like loops, a transmembrane region, and an intracellular domain (Takai et al. 2003; Murakami 2005). They were found to be lost or at strikingly low levels in many kinds of malignant carcinomas. For example, CADM1 (also known as Necl-2, TSLC1, cynCAM1, IGSF4A) has been reported to be reduced in lung cancer (Kuramochi et al. 2001), prostate cancer (Fukuhara et al. 2002), esophageal cancer (Ito et al. 2003), and breast cancer (Takahashi et al. 2012). And several articles have reported that CADM3 (also known as Necl-1, TSLL1, cynCAM3, IGSF4B) and CADM4 (also known as Necl-4, TSLL2, cynCAM4, IGSF4C) also functions as tumor suppressor in various types of cancer cells (Williams et al. 2006; Raveh et al. 2009; Nagata et al. 2012). However, CADM2 (also known as Necl-3, cynCAM2, IGSF4D) has not been well characterized in cancer (Chang et al. 2010), and there have been no reports to date regarding the expression and clinical significance of CADM2 in HCC. Our previous HCC expression profile genechip indicated CADM2 expression was strikingly reduced in tumor tissues (unpublished data). Therefore, we further investigated the expression and clinical significance of CADM2 in HCC using high-throughput tissue microarray technology and rich HCC case resource in Eastern Hepatobiliary Surgery Hospital (Shanghai China). We found that CADM2 is strikingly down regulated in HCC and can serve as a novel predictor for recurrence-free survival (RFS) of HCC patients after hepatectomy.
Materials and methods Patients and clinical sample Under the inclusion and exclusion criteria: (a) with frozen tumor tissue and paired noncancerous liver tissue stored (for mRNA examination) or with paraffin tissue blocks stored (for tissue microarray construction) in our hospital; (b) underwent curative resection during June and July 2011 (for mRNA examination) or between May 2004 and June 2006 (for tissue microarray construction); (c) distinctive pathologic diagnosis of HCC; (d) without anticancer treatment and distant metastasis before surgery, 300 patients were identified as the base population for mRNA examination,
13
J Cancer Res Clin Oncol (2014) 140:109–116
and 1,020 patients were identified for tissue microarray construction. Finally, 30 patients were randomly selected from its corresponding base population using random number for mRNA examination. Similarly 242 patients were selected for tissue microarray construction. The adjacent noncancerous tissues were isolated at least 2 cm away from the tumor tissue border. For the cohort of 242 HCC patients, complete clinicopathological and follow-up data are available. These patients’ clinicopathological data are presented in Table 1. Tumor differentiation was defined according to Edmondson’s grading system (I, welldifferentiated; II, moderately differentiated; III, poorly differentiated; IV, undifferentiated) (Edmondson and Steiner 1954). Micrometastases were defined as tumors adjacent to the border of the main tumor that could be observed only under the microscope. The Barcelona Clinic Liver Cancer staging system (BCLC) ranked hepatocellular carcinomas in five stages, ranging from 0 (very early stage) to 5 (terminal stage) (Llovet et al. 1999). The day of surgery was considered day 0 in calculating postoperative survival. Followup ranged from 1 to 80 months. The study was approved by the institutional review board of the Eastern Hepatobiliary Surgery Hospital, and all patients provided written informed consent to participate in the study. Any information contained in this article cannot identify the patients. Examination of the mRNA expression level of CADM2 Total RNA of the 30 patients’ tumor tissues and adjacent noncancerous tissues was isolated using TRIzol reagent (Invitrogen CA), and first strand cDNA was generated using random primers and M-MLV reverse transcriptase (Invitrogen) according to the manufacturer’s instructions. Real-time PCR was performed using StepOne™ Real-Time PCR System (Applied Biosystems, Foster City, USA) and SYBR® Green (Takara, Dalian, P. R. China). The CADM2-specific primers were 5′-CCTCAATGCCACCCCTCAG-3′ (forward) and 5′-TTCTCCGCCATCCTTTGTCC-3′ (reverse), and the GAPDH-specific primers were 5′-CGGATTTGGTCGTATTGGG-3′ (forward) and 5′-CTGGAAGATGGTGATGGGATT-3′ (reverse). PCR condition was 95 °C for 5 min, followed by 40 cycles at 95 °C for 15 s, 58 °C for 15 s, 72 °C for 20 s. Gene expression in each sample was normalized to relative to that of GAPDH mRNA. Tissue microarray construction and immunohistochemistry This tissue microarray which was constructed by Shanghai Biochip Corporation (China) contains 242 patients’ tumor tissues and adjacent noncancerous tissues. Each tissue has two repetitions on the tissue microarray. The tissue microarray slides were stained using EnVision system (Dako Danmark)
111
J Cancer Res Clin Oncol (2014) 140:109–116 Table 1 Association between CADM2 protein level and clinicopathological characteristics χ2
Clinical characteristics
CADM2 Lowa
Higha
All cases Gentle Male Female Age ≥50 5 cm ≤5 cm Tumor number 1 2 3 4 Micrometastases No Yes Vascular invasion Negative Positive microscopically Positive macroscopically HBsAg Negative Positive HBeAg Negative Positive BCLC stage 0 A B C
46 98
46 44
7 137
18 72
62 82
41 49
117 24 2 1
73 13 3 1
34 110
31 59
48 77 19
51 35 4
12 132
16 74
108 36
75 15
8 101 16
10 64 12
19
4
P value
0.853
0.409
2.975
0.105
8.528
0.003
13.302
0.000
0.140
0.787
1.266
0.737
3.240
0.098
13.902
0.001
4.690
0.038
2.257
0.146
6.772
0.080
Italic values are statistically significant at (P 20 ng/ml a
Patients whose adjacent noncancerous tissues’ immunohistochemistry score minus tumor tissues’ immunohistochemistry score is less than 1 belonged to high-expression group, and the other patients belonged to low-expression group
Fig. 1 Expression of CADM2 in HCC tumor tissues and adjacent noncancerous tissues. a mRNA level of CADM2 in 30 HCC patients’ tumor tissues and adjacent tumor tissues. Each bar shows the mRNA level ratio of adjacent noncancerous tissue to tumor tissue (N/T). b Immunohistochemistry score distribution of the 234 HCC patients’ tumor tissues (T) and adjacent noncancerous tissues (N). Numbers at the bar refer to the case number of patients within the immunohistochemistry score range. c Distribution of difference between tumors’ and paired noncancerous tissues’ immunohistochemistry score. Numbers at the bar refer to the case number of patients within the specific range. Difference means adjacent noncancerous tissues’ immunohistochemistry score minus paired tumor tissues’ immunohistochemistry score (N–T)
according to the manufacturer’ instruction with a rabbit polyclonal antibody as the primary antibody (Abnover Taiwan) (diluted 1:100). Analyzing the tissue microarray was performed by two experienced researchers. Immunohistochemistry score was mainly depended on the intensity. No staining is scored 0, weak staining is scored 1, moderate staining is scored 2, strong staining is scored 3, and strikingly strong staining is scored 4. Representative pictures of respective semiquantitative score are shown in Fig. S1. Each tissue’s
13
112
J Cancer Res Clin Oncol (2014) 140:109–116
Fig. 2 Representative pictures of the immunohistochemistry of CADM2. The upper four pictures are from HCC tumors, and the lower four pictures are from their paired adjacent noncancerous tissues,
respectively. The left four pictures belong to patient 1, and the right four pictures belong to patient 2. The magnification of a/e/c/g is ×50, while b/d/f/h are their corresponding pictures with magnification ×100
four scoring results including two repetitions analyzed by two researchers were finally averaged.
patients through real-time PCR. The result indicated that CADM2 mRNA level in tumor tissues was significantly lower than in adjacent noncancerous tissues (Fig. 1a). We further examined the protein level of CADM2 through immunohistochemistry on a tissue microarray including 234 valid HCC patients’ tumor tissues and adjacent noncancerous tissues (8 patients were excluded for their incomplete tissue cores on the tissue microarray). We found that tumors had significantly weaker immunohistochemistry staining than adjacent noncancerous tissues (Figs. 1b, c, 2). Further analysis revealed 61.5 % patients (144 patients) showed lower CADM2 protein level in tumor tissues than in adjacent noncancerous tissues (the difference of immunohistochemistry score ≥1), and 24.8 % patients (58 patients) showed significantly lower CADM2 protein level in tumor tissues than in adjacent noncancerous tissues (the difference of immunohistochemistry score ≥2) (Fig. 1c).
Statistical analysis All statistical analyses were performed by SPSS18. The χ2 test was used to evaluate the association between CADM2 expression level and clinicopathological characteristics. The cumulative survival probability was evaluated using the Kaplan–Meier method, and the significance of differences was assessed using the log-rank test. To determine independent prognostic factors, the Cox multivariate regression analysis was used. A P
