Ann Surg Oncol DOI 10.1245/s10434-014-4236-y
ORIGINAL ARTICLE – UROLOGIC ONCOLOGY
Elevated Expression of N-Acetylgalactosaminyltransferase 10 Predicts Poor Survival and Early Recurrence of Patients with Clear-Cell Renal Cell Carcinoma Qian Wu, PhD1, Liu Yang, PhD1, Haiou Liu, PhD1, Weijuan Zhang, PhD2, Xu Le, MD, PhD3, and Jiejie Xu, MD, PhD1 1
Key Laboratory of Glycoconjugate Research, Ministry of Health, Department of Biochemistry and Molecular Biology, School of Basic Medical Sciences, Shanghai Medical College of Fudan University, Shanghai, China; 2Department of Immunology, School of Basic Medical Sciences, Shanghai Medical College of Fudan University, Shanghai, China; 3 Department of Urology, Zhongshan Hospital, Shanghai Medical College of Fudan University, Shanghai, China
ABSTRACT Purpose. The aim of this study was to evaluate the potential prognostic significance of N-acetylgalactosaminyltransferase 10 (GALNT10) in patients with clear-cell renal cell carcinoma (ccRCC) after surgical resection. Methods. We retrospectively enrolled 271 patients (202 in the training cohort and 69 in the validation cohort) with ccRCC undergoing nephrectomy at a single institution. Clinicopathologic features, overall survival (OS), and recurrence-free survival (RFS) were recorded. GALNT10 intensities were assessed by immunohistochemistry in the specimens of patients. The Kaplan–Meier method was applied to compare survival curves. Cox regression models were used to analyze the impact of prognostic factors on OS and RFS. Concordance index (C-index) was calculated to assess predictive accuracy. Results. In both cohorts, elevated GALNT10 expression in tumor tissues positively correlated with advanced TNM stage. High GALNT10 expression indicated poor survival and early recurrence of patients with ccRCC, particularly
Qian Wu, Liu Yang, and Haiou Liu have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1245/s10434-014-4236-y) contains supplementary material, which is available to authorized users. Ó Society of Surgical Oncology 2014 First Received: 16 June 2014 J. Xu, MD, PhD e-mail: [email protected]
with early-stage disease. After backward elimination, GALNT10 expression was identified as an independent adverse prognostic factor for survival and recurrence. The predictive accuracy of TNM, University of California Los Angeles Integrated Staging System, and stage, size, grade, and necrosis prognostic models was improved when GALNT10 expression was added. Conclusions. GALNT10 expression is a potential independent adverse prognostic biomarker for recurrence and survival of patients with ccRCC after nephrectomy.
Clear-cell renal cell carcinoma (ccRCC) comprises the major histological subtype of renal cell carcinoma (RCC), with approximately 40,000 patients being diagnosed annually in the US,1 and nearly 650,000 annually in China.2 Due to improved clinical diagnosis technology, such as the widespread use of abdominal imaging, more patients with ccRCC can be diagnosed at an early stage, and less than 10 % will die of cancer-related causes.3 However, nearly 30 % of patients diagnosed with localized ccRCC will undergo relapse and progression after nephrectomy, and patients with metastatic ccRCC will have a poor 5-year survival rate, ranging from 0 to 20 %.4 Previous studies showed that clinicopathological factors, such as TNM stage and Fuhrman’s nuclear grading, are independent prognostic factors.5 Eastern Cooperative Oncology Group performance status (ECOG-PS) is also used as an indicator of clinical outcome for patients with RCC. Although numerous molecular markers, including carbonic anhydrase IX (CA IX), vascular endothelial growth factor, and hypoxia-induced factor,6 and gene expression profiling 7 have been investigated, few have
Q. Wu et al.
helped identify new relevant prognostic factors. Integrated prognostic models have been used to predict clinical outcome; for example, the University of California Los Angeles Integrated Staging System (UISS) is applied when evaluating overall survival (OS) or cancer-specific survival (CSS) in RCC,8 and the Mayo Clinic stage, size, grade and necrosis (SSIGN) score more accurately predicts CSS in ccRCC.9,10 Owing to the fact that patients with similar clinical characteristics may have various outcomes, the identification of novel markers to improve the predictive accuracy of prognostic systems is urgently needed for ccRCC patients. Aberrant glycosylation is a well-known hallmark of tumors, and tumor-associated carbohydrate antigens (CAs) have been reported to be closely related to malignant transformation and disease progression in tumors.11 In mammals, mucin-type O-linked glycans constitute up to 80 % of the total cell-surface CAs.12 Mucin-type O-glycosylation occurs in the Golgi apparatus, and is initiated by UDP-N-acetyl-a-D-galactosamine:polypeptide N-acetylgalactosaminyltransferases (GALNTs), which comprise of 20 members, including GALNT1–GALNT14 and GALNTL1–GALNTL6.13 Recent studies have focused on the aberrant expression of GALNTs in different human tumors. For instance, GALNT3 has been revealed to be associated with disease aggressiveness of RCC,14 gastric carcinomas,15 lung adenocarcinoma,16 and pancreatic carcinomas,17 and GALNT6 may act as an independent indicator of breast cancer 18,19 and pancreatic cancer.20 However, few studies have been designed to explore the potential role of GALNT10 in human cancers. In this study, the expression of GALNT10 was analyzed by immunohistochemical (IHC) staining in ccRCC specimens, and its correlation with clinicopathologic features and clinical outcome, including endpoints of OS and RFS, were evaluated. In addition, incorporation of the UISS or SSIGN models with intratumoral GALNT10 expression could improve the predictive accuracy of these prognostic models. PATIENTS AND METHODS
histopathologically proven ccRCC, while exclusion criteria included other types of renal cancer as confirmed histopathologically, tumors with necrosis [80 %, and patients who died within the first month after surgery due to surgical complications. The database of all patients’ baseline clinical and pathological information and complete followup outcome data were recorded. TNM stage and Fuhrman grading were classified according to the TNM system of the 2010 American Joint Committee on Cancer (AJCC)21 and the 2004 World Health Organization (WHO) criteria,22 respectively. ECOG-PS score was assigned to each patient at the time of diagnosis. The UISS and SSIGN predictive models were applied to the patients. We calculated OS from the date of surgery to the date of death, and recurrence-free survival (RFS) was calculated from the date of recurrence to the date of death or the last follow-up from all causes. This study was approved by the hospital’s Ethics Committee, and informed consent was obtained from each patient. Tissue Microarray and Immunohistochemistry Tissue microarrays were established as previously described.22 Primary anti-GALNT10 antibody (diluted 1:600; Sigma-Aldrich, St Louis, MO, USA) was used for IHC staining. Before being employed, the specificity of the antibody was confirmed by IHC and Western blot with peptide competition. The staining intensity of each specimen was scored independently by two pathologists blinded to the clinicopathological data. A semi-quantitative Hscore for specimens was calculated by multiplying the staining intensities (0: negative; 1: weak staining; 2: moderate staining; 3: strong staining) by the area distributions (0–100 %), ranging from 0 to 300.23 For determination of the optimal cutoff score of the staining intensity applied to separate the patients into the ‘GALNT10 high’ and ‘GALNT10 low’ groups, we employed Xtile plot analysis using X-tile software v3.6.1 (Yale University, New Haven, CT, USA) to distinguish patients with poor or good prognosis to achieve the lowest p value and the highest hazard ratio (HR).24 The 50 % value was set as the cutoff point for tumoral GALNT10 density.
Patients Statistical Analysis Two independent cohorts comprising 271 consecutive patients who underwent curative nephrectomy at Zhongshan Hospital of Fudan University (Shanghai, China) were enrolled in this study. Specimens of the training cohort (n = 202) were obtained between 2003 and 2004, and specimens of the validation cohort (n = 69) were obtained in 2001. Inclusion criteria included patients with no history of previous anticancer therapy and other malignancies, bearing radical or partial nephrectomy, and
For statistical analysis, SPSS 21 (IBM Corporation, Armonk, NY, USA), MedCalc Software (version 11.4.2.0; MedCalc, Mariakerke, Belgium), and Stata 12.0 (StataCorp LP, College Station, TX, USA) were used. Numerical variables were analyzed using Student’s t test, while categorical variables were studied using the v2 or Fisher’s exact test. Survival (including OS and RFS) curves were established using the Kaplan–Meier method, and the
GALNT10 in ccRCC
difference between curves was compared using the logrank test. Number at risk was calculated at the beginning of each time period. The Cox proportional hazards regression model was applied to perform univariate and multivariate analysis, and those characteristics demonstrating statistical significance on OS and RFS in the univariate analysis were considered in the multivariate analysis. The Concordance index (C-index), which represents the area under the curve adapted for survival data, and Akaike information criterion (AIC) were evaluated to estimate and compare the predictive accuracy of multivariate models. All p values were two-sided and statistical significance was set at 0.05. RESULTS Immunohistochemical Findings GALNT10 expression was evaluated by IHC analysis in paired ccRCC specimens. As shown in electronic supplementary Fig. S1a, GALNT10 expression was observed to be higher in tumor tissues than in relevant peri-tumor tissues. The variable staining intensities are shown in electronic supplementary Fig. S1b. The average staining intensity score of peri-tumor tissues was 128, while the average score of the tumor tissues was 155 (p \ 0.001; electronic supplementary Fig. S1c). Analyzed by X-tile program to achieve a minimum p value, 150 was used as the optimal cutoff score to separate the training cohort specimens, while 140 was determined for the validation cohort. In addition, the percentage of patients with a high level of GALNT10 increased gradually, along with disease progression from TNM stages I–IV (electronic supplementary Fig. S1d). Association between GALNT10 Expression with Clinicopathological Features of Clear-Cell Renal Cell Carcinoma (ccRCC) The detailed characteristics of the two cohorts are shown in Table 1. Median follow-up was 88 months (range 11– 120 months) and 71 months (range 18–118 months) for the two cohorts, respectively. A total of 51 % (103 of 202) and 43.5 % (31 of 69) of tumor tissues were scored as having high GALNT10 staining density, respectively. Additionally, GALNT10 expression was significantly associated with Fuhrman grade (p = 0.002) in the training cohort, while no significance with any clinical characters was seen in the validation cohort, which may be due to the smaller size of patients. Furthermore, such heterogeneity may help to confirm that the predictor has universal applicability across the heterogeneous population of patients from different districts.
Prognostic Value of GALNT10 Intensity for Clinical Outcome of ccRCC Patients To further investigate the relation between GALNT10 expression and patient outcomes, we applied the Kaplan– Meier survival analysis to compare OS or RFS between GALNT10 low expression and GALNT10 high expression in the two cohorts. For RFS analysis,we excluded cases with synchronous nodal (N1) or distant metastasis (M1) at the time of surgery (n = 9 in the training cohort and n = 7 in the validation cohort). High GALNT10 was found to show poorer OS (p = 0.002 and p \ 0.001 for the training and validation cohorts, respectively; Fig. 1a, b) and RFS (p = 0.001 and p = 0.036 for the training and validation cohorts, respectively; Fig. 2a, b). We further investigated whether GALNT10 expression couldstratify patients with different TNM stage in an overall cohort of 271 patients. The dichotomized GALNT10 level was found to show strong prognostic relevance for TNM stage I patients, both for OS [HR 7.191, 95 % confidence interval (CI) 2.486–20.798, p \ 0.001; Fig. 1c] and RFS (HR 3.228, 95 % CI 1.550–6.728, p = 0.001; Fig. 2c). GALNT10 expression could also predict OS of a combination of TNM II, III, and IV patients (HR 2.106, 95 % CI 1.146–3.870, p = 0.021; Fig. 1d), while it failed to predict disease recurrence of T2–3N0M0 patients (HR 1.720, 95 % CI 0.950–3.115, p = 0.067; Fig. 2d). Univariate and multivariate analyses of the overall 271 patients were further performed. As shown in Table 2, univariate analysis revealed that GALNT10 was significantly associated with OS (HR 3.108, 95 % CI 1.879–5.141, p \ 0.001). GALNT10, together with TNM stage, Fuhrman grade, ECOG-PS, tumor size, and necrosis were estimated in multivariate analysis, and results demonstrated that GALNT10 remained an independent prognostic indicator for OS (HR 2.560, 95 % CI 1.540–4.256, p \ 0.001) as well as TNM stage (p \ 0.001), Fuhrman grade (p \ 0.001), and ECOGPS (p = 0.005) (Table 2). Univariate analysis for RFS demonstrated that a high level of GALNT10 was a risk factor for ccRCC (HR 2.341, 95 % CI 1.480–3.702, p \ 0.001). Multivariate analysis showed that T stage (p = 0.003), Fuhrman grade (p \ 0.001), ECOG-PS (p = 0.049), and GALNT10 (p = 0.002) remained as independent indicators of RFS in the overall cohorts (Table 2). Extension of Postoperative Prognostic Systems with GALNT10 Density To construct a more sensitive predictive model for outcomes of patients with ccRCC, we combined GALNT10 expression with TNM stage, and UISS or SSIGN score to testify their accuracy of survival conditions. Incorporation of GALNT10 increased the predictive value of these three models, namely when assessing OS: 0.758
Q. Wu et al. TABLE 1 Clinical characteristics and correlations with GALNT10 of two independent patient cohorts with clear-cell renal cell carcinoma Variable
N
%
Mean age (years)b
Low (n = 99)
High (n = 103)
55.3 ± 11.1
54.9 ± 12.3
Sex
p Valuea
N
%
0.812
Low (n = 38)
High (n = 31)
59.9 ± 14.1
62.2 ± 14.1
0.467
Male
140
69.3
71
69
Female
62
30.7
28
34
ECOG-PS 165
81.7
82
83
C1
37
18.3
17
20
4.4 ± 2.5
4.7 ± 2.9
Mean tumor size (cm)b T stage
50
72.5
29
21
19
27.5
9
10
57
82.6
34
23
12
17.4
4
8
4.9 ± 2.7
5.6 ± 2.9
0.096
0.207 0.340
128
63.4
67
61
45
65.2
26
19
2 3
19 52
9.4 25.7
10 20
9 32
13 8
18.9 11.6
6 5
7 3
4
3
1.5
2
1
3
4.3
1
2
0.968
0.449
N0
198
98.0
97
101
68
98.6
38
30
N1
4
2.0
2
2
1
1.4
0
1
M0
195
96.5
97
98
63
91.3
35
28
M1
7
3.5
2
5
6
8.7
3
3
122
60.4
66
56
45
65.2
25
20 5
M stage
0.271
TNM stage I
0.794
0.131
0.861
II
18
8.9
10
8
11
15.9
6
III
54
26.7
21
33
6
8.7
4
2
IV
8
4.0
2
6
7
10.2
3
4
Fuhrman grade
0.649
0.002
1
32
15.8
21
11
16
23.2
11
5
2 3
91 55
45.1 27.2
51 22
40 33
32 11
46.4 15.9
16 6
16 5
4
24
11.9
5
19
10
14.5
5
5
Necrosis
0.179
0.444
Absent
155
76.7
80
75
52
75.4
30
22
Present
47
23.3
19
28
17
24.6
8
9
UISS score
0.485
0.005
1
68
33.7
44
24
32
46.4
20
12
2
114
56.4
48
66
30
43.5
15
15
C3
20
9.9
7
13
7
10.1
3
4
50
72.5
28
22
SSIGN score 0–3
0.899
0.022 129
63.9
67
52
0.116 0.729
1
N stage
0.422 0.428
0.680
0
p Valuea
4–7
60
29.7
29
41
14
20.3
7
7
C8
13
6.4
3
10
5
7.2
3
2
Bold values are considered statistically significant (p \ 0.05) ECOG-PS Eastern Cooperative Oncology Group performance status, UISS University of California Los Angeles Integrated Staging System, SSIGN stage, size, grade, and necrosis a
A p value \0.05 is considered statistically significant
b
The results of continuous variables are expressed as mean ± SD
versus 0.684 for the TNM stage, 0.718 versus 0.673 for the UISS score, and 0.777 versus 0.734 for the SSIGN score cohort; and when assessing RFS: 0.695 versus 0.645 for the
TNM stage, 0.671 versus 0.624 for the UISS score, and 0.724 versus 0.695 for the SSIGN score (Table 3). The elevated tendency of the C-index, as well as the decreased
GALNT10 in ccRCC
Training cohort
a
Validation cohort
b
All cases
All cases 100
80 60 40 20
GALNT10 low GALNT10 high
Logrank test P= 0.002
Overall survival (%)
100
Overall survival (%)
FIG. 1 Analysis of OS considering GALNT10 expression in ccRCC tissues. Kaplan–Meier analysis of OS of patients in a the training cohort (n = 202) and b the validation cohort (n = 69), as well as c TNM stage I (n = 167), and d through TNM stages II–IV (n = 104), of an overall total cohort of 271 patients. p value was calculated using the log-rank test. OS overall survival, GALNT10 N-acetylgalactosaminyltransferase 10, ccRCC clear-cell renal cell carcinoma, HR hazard ratio, CI confidence interval
0
80 60 40 20
GALNT10 low GALNT10 high
0
24
48
72
96
120
0
24
Time after surgery (months) Number at risk GALNT10 low 99 GALNT10 high 103
c
48
72
96
120
Time after surgery (months)
97
91
76
60
0
98
87
65
50
0
Number at risk GALNT10 low 38 GALNT10 high 31
d
TNM stage I
38
31
25
18
0
29
24
14
7
0
TNM stage II+III+IV 100
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
Logrank test P< 0.001 HR: 7.191; 95%CI: 2.486-20.798
48
72
96
120
Overall survival (%)
100
Overall survival (%)
Logrank test P< 0.001
0
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
Time after surgery (months) Number at risk GALNT10 low 91 GALNT10 high 76
81
70
62
0
76
69
57
40
0
DISCUSSION In this present study, we depicted that the expression of GALNT10 varied in peri-tumor and tumor tissues in ccRCC. A higher level of GALNT10 was observed in tumor tissues and tended to positively correlate with TNM stage. Furthermore, high expression of GALNT10 means a poor outcome of OS and RFS. The dimidiate density of GALNT10 can be used to stratify TNM stage I and stage II ? III ? IV for OS, and T1N0M0 for RFS, in ccRCC. Univariate and multivariate regression analysis showed that a high level of GALNT10 is a significant independent indicator for predicting poor prognosis and early recurrence of ccRCC. In addition, incorporation of GALNT10 with postoperative prognostic models could improve predictive accuracy when estimating OS and RFS. In sum,
72
96
120
Time after surgery (months)
89
trend of AIC, suggest a better predictive accuracy. All these results suggested that a combination of GALNT10 and conventional prognostic models could generate better predictive systems for ccRCC patient outcomes.
48
Logrank test P= 0.021 HR: 2.106; 95%CI: 1.146-3.870
Number at risk GALNT10 low 46 GALNT10 high 58
46
41
31
16
0
51
42
22
17
0
analysis of GALNT10 expression may have a special function as an independent prognostic indicator. Altered glycan epitopes of cell surface proteins, accompanied with changed biological activity, might play an important role in cell–cell interaction and signal transduction, contributing to aberrant cell proliferation, migration and invasion behaviors.25 This often happens in conditions of abnormal expression or activity of glycosyltransferases. The evidence provided depicted that both Nglycosylation and O-glycosylation modification, such as glycosylation of the b1,6 N-acetyglucosaminyltransferase V (Mgat5) and GALNTs, were involved in numerous diseases,26 particularly in cancers.17,19,27–30 GALNT10 was first cloned and reported in 2002.31 Unlike most members of the GALNT family, GALNT10 shows glycoprotein glycosyltransferase activity, similar to GALNT7, the same subset of this whole family. The prior addition of N-acetylgalactosamine (GalNAc) to a synthetic peptide is necessary for catalyzing the transfer of sugar to a substrate by GALNT10.13 To date, limited studies have been conducted to investigate the function of GALNT10 in
Q. Wu et al.
b
Training cohort 100
All cases
80 60 40 20
GALNT10 low GALNT10 high
Logrank test P= 0.001
0 24
0
48
72
96
Recurrence-free survival (%)
Recurrence-free survival (%)
a
Validation cohort 100
All cases
80 60 40 20
GALNT10 low GALNT10 high
120
0
c 100
95
84
67
58
0
91
76
59
45
0
80 60 40
GALNT10 low GALNT10 high
0 0
Number at risk GALNT10 low 35 GALNT10 high 27
d
T1N0M0
20
24
Logrank test P= 0.001 HR: 3.228; 95%Cl: 1.550-60728
48
72
96
120
72
96
120
34
29
23
15
0
27
17
10
4
0
T2-3N0M0
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
48
Logrank test P= 0.067 HR: 1.720; 95%Cl: 0.950-3.115
72
96
120
Time after surgery (months)
89
80
64
59
0
74
63
50
36
0
human disease. Gao et al. revealed that GALNT10 had a positive correlation with tumor differentiation, and was a useful indicator in gastric cancer.32 We consistently found that a high level of GALNT10 was seen in tumor tissues, which indicated a poor outcome of both OS and RFS of ccRCC patients. However, the underlying mechanism of GALNT10 oncogenic functions in ccRCC progression remains unknown. A prior study showed that anepidermal growth factor receptor (EGFR) could be modified by GALNT2 in hepatocellular carcinomas.33 Galectin-3, an endogenous lectin, facilitates the interaction of MUC1 with EGFR, modulating the EGFR-mediated oncogenic signaling pathways in pancreatic cancer cells.34 Our former study revealed that Mgat5 mediated bisecting N-glycosylation on the EGFR, activated the EGFR/PAK1 pathway, and contributed to hepatoma cell resistance to anoikis.28 Collectively, further investigation needs to identify whether GALNT10 could alter glycosylation and activation of EGFR or other potential substrates in ccRCC. Currently, GALNT10 expression was reported to be significantly associated with OS and RFS. In an attempt to elaborate the predictive role of GALNT10 in the overall cohort of 271 ccRCC patients, subgroup analysis with
48
100
Time after surgery (months) Number at risk GALNT10 low 91 GALNT10 high 76
24
Time after surgery (months)
Recurrence-free survival (%)
Number at risk GALNT10 low 97 GALNT10 high 96
Logrank test P= 0.036
0
Time after surgery (months)
Recurrence-free survival (%)
FIG. 2 Analysis of RFS considering GALNT10 expression in ccRCC tissues. Kaplan–Meier analysis of RFS of patients in a the training cohort (n = 193) and b the validation cohort (n = 62), as well as c stage T1N0M0 (n = 167), and d stage T2– 3N0M0 (n = 88), of an overall cohort of 271 patients. Nine patients in the training cohort and seven patients in the validation cohort with tumor metastasis at the time of surgery were excluded from the analysis of tumor recurrence. p value was calculated using the logrank test. RFS recurrence-free survival, GALNT10 Nacetylgalactosaminyltransferase 10, ccRCC clear-cell renal cell carcinoma, HR hazard ratio, CI confidence interval
Number at risk GALNT10 low 41 GALNT10 high 47
40
33
26
14
0
44
30
19
13
0
respect to TNM stage demonstrated that dichotomized GALNT10 could stratify patients with early (stage I) and higher (stages II, III, and IV) stages for OS, but the HR of early-stage patients was greater than higher-stage patients (7.191 vs. 3.228), suggesting that GALNT10 may be a more important risk factor for early-stage patients. GALNT10 evaluation could only predict disease recurrence for early-stage (stage T1N0M0) patients, suggesting GALNT10 might be more valuable for early disease recurrence. Thus, ccRCC patients with high intratumoral GALNT10 density should have more aggressive therapies and require closer follow-up. Given that the patient cluster carrying higher-stage tumors is relatively small, further studies are required to determine whether GALNT10expression can be used to estimate outcome for patients in the higher stage of disease. We acknowledge several limitations to the current study. First, of the total number of patients enrolled there were insufficient cases with higher-stage tumor; thus, we could only draw the conclusion that GALNT10 is a predictive factor for outcome, particularly in early disease recurrence, for patients with early-stage tumor. Second, this was a retrospective, single-center study, and our
GALNT10 in ccRCC TABLE 2 Univariate and multivariate Cox regression analysis for overall survival and recurrence-free survival Variable
Univariate
Multivariate
HR (95 % CI)
a
p Value
HR (95 % CI)
p Valuea
Overall survival (n = 271) Age (years)b
1.020 (1.000–1.040)
0.048
Sex (female vs. male)
1.137 (0.688–1.880)
0.616
ECOG-PS (C1 vs. 0)
3.570 (2.244–5.678)
\ 0.001
2.085 (1.255–3.466)
Tumor size (cm)b
1.197 (1.118–1.281)
\0.001
1.043 (0.951–1.143)
0.370
TNM stage (II ? III ? IV vs. I)
3.761 (2.347–6.027)
\ 0.001
2.747 (1.571–4.803)
\ 0.001
Fuhrman grade (3 ? 4 vs. 1 ? 2) Necrosis (yes vs. no)
2.829 (1.793–4.464) 2.463 (1.548–3.917)
\0.001 \0.001
2.366 (1.488–3.763) 1.596 (0.992–2.568)
\0.001 0.054
GALNT10 (high vs. Low)
3.108 (1.879–5.141)
\0.001
2.560 (1.540–4.256)
\0.001
Age (years)a
1.013 (0.994–1.032)
0.192
Sex (female vs. male)
0.744 (0.472–1.172)
0.202
ECOG-PS (C1 vs. 0)
2.597 (1.600–4.216)
\0.001
1.693 (1.002–2.859)
0.049 0.322
0.005
Recurrence-free survival (n = 255)c
b
1.170 (1.093–1.253)
\0.001
1.048 (0.955–1.149)
T stage (2 ? 3 ? 4 vs. 1)
3.041 (1.956–4.725)
\0.001
2.434 (1.431–4.140)
0.003
Fuhrman grade (3 ? 4 vs. 1 ? 2)
2.696 (1.739–4.178)
\0.001
2.432 (1.555–3.804)
\0.001
Tumor size (cm)
Necrosis (yes vs. no)
1.945 (1.218–3.105)
0.005
1.312 (0.812–2.122)
0.268
GALNT10 (high vs. low)
2.341 (1.480–3.702)
\0.001
2.041 (1.286–3.240)
0.002
Bold values are considered statistically significant (p \ 0.05) HR hazard ratio, CI confidence interval, ECOG-PS Eastern Cooperative Oncology Group performance status, GALNT10 polypeptide Nacetylgalactosaminyl transferase 10, ccRCC clear-cell renal cell carcinoma a
A p value \0.05 is considered statistically significant
b
Factors were treated as continuous variables
c
For recurrence-free survival analysis, 16 patients with metastasis ccRCC were excluded
TABLE 3 Comparison of the prognostic accuracies of models for overall survival and recurrence-free survival Models
Overall survival (n = 271)
Recurrence-free survival (n = 255)a
C-index AIC
C-index
AIC
GALNT10
0.643
788.306 0.612
841.817
TNM stage
0.684
767.346 0.645
828.654
TNM stage ? GALNT10
0.758
750.214 0.695
818.489
UISS score
0.673
517.766 0.624
595.119
UISS score ? GALNT10
0.718
511.656 0.671
590.476
SSIGN score
0.734
742.598 0.695
814.941
SSIGN score ? GALNT10 0.777
731.715 0.724
808.781
An elevated C-index or a decreased AIC score mean a better prognosis C-index Harrell’s concordance index, AIC Akaike information criterion, GALNT10 polypeptide N-acetylgalactosaminyl transferase 10, UISS University of California Los Angeles Integrated Staging System, SSIGN stage, size, grade, and necrosis, ccRCC clear-cell renal cell carcinoma a
For recurrence-free survival analysis, 16 patients with metastasis ccRCC are excluded
findings should thus be replicated in other populations and larger cohorts to further validate our conclusions. Third, further calculations regarding the endpoint of CSS had to be omitted, and whether or not employing GALNT10 resulted in a well predictive value in other RCC subtypes still needs to be explored. Fourth, a prospective study enrolling more patients stratified by more detailed clinicopathologic characteristics and diverse histological subtypes could further validate our current findings. We are seeking to cooperate with other academic centers. In addition, the detailed mechanisms ofGALNT10 oncogenic functions in ccRCC need further laboratory and clinical investigation. ACKNOWLEDGMENT This work was supported by grants from the National Basic Research Program of China (2012CB822104), the National Key Projects for Infectious Diseases of China (2012ZX10002-012), the National Natural Science Foundation of China (31100629, 31270863, 81471621, 81472227), the Program for New Century Excellent Talents in University (NCET-13-0146), and the Shanghai Rising-Star Program (13QA1400300). All these study sponsors had no role in the study design, or in the collection, analysis, and interpretation of data. We thank Ms. Haiying Zeng
Q. Wu et al. (Department of Pathology, Zhongshan Hospital, Shanghai Medical College of Fudan University) for technical assistance. DISCLOSURE
The authors have declared no conflicts of interest.
REFERENCES 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63(1):11–30. 2. Yang L, Parkin DM, Ferlay J, Li L, Chen Y. Estimates of cancer incidence in China for 2000 and projections for 2005. Cancer Epidemiol Biomark Prev. 2005;14(1):243–250. 3. Joseph RW, Kapur P, Serie DJ, et al. Loss of BAP1 protein expression is an independent marker of poor prognosis in patients with low-risk clear cell renal cell carcinoma. Cancer. 2013;120(7): 1059–67. 4. Stewart GD, O’Mahony FC, Powles T, Riddick AC, Harrison DJ, Faratian D. What can molecular pathology contribute to the management of renal cell carcinoma? Nat Rev Urol. 2011;8(5): 255–65. 5. Lang H, Lindner V, de Fromont M, et al. Multicenter determination of optimal interobserver agreement using the Fuhrman grading system for renal cell carcinoma: assessment of 241 patients with [15-year follow-up. Cancer. 2005;103(3):625–9. 6. Zhao H, Ljungberg B, Grankvist K, Rasmuson T, Tibshirani R, Brooks JD. Gene expression profiling predicts survival in conventional renal cell carcinoma. PLoS Med. 2006;3(1):e13. 7. Sim SH, Messenger MP, Gregory WM, et al. Prognostic utility of pre-operative circulating osteopontin, carbonic anhydrase IX and CRP in renal cell carcinoma. Br J Cancer. 2012;107(7):1131–7. 8. Zisman A, Pantuck AJ, Dorey F, et al. Improved prognostication of renal cell carcinoma using an integrated staging system. J Clin Oncol. 2001;19(6):1649–57. 9. Frank I, Blute ML, Cheville JC, Lohse CM, Weaver AL, Zincke H. An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score. J Urol. 2002;168(6):2395–400. 10. Sun M, Shariat SF, Cheng C, et al. Prognostic factors and predictive models in renal cell carcinoma: a contemporary review. Eur Urol. 2011;60(4):644–61. 11. Brockhausen I. Pathways of O-glycan biosynthesis in cancer cells. Biochim Biophys Acta. 1999;1473(1):67–95. 12. Hollingsworth MA, Swanson BJ. Mucins in cancer: protection and control of the cell surface. Nat Rev Cancer. 2004;4(1):45–60. 13. Ten Hagen KG, Fritz TA, Tabak LA. All in the family: the UDPGalNAc:polypeptide N-acetylgalactosaminyltransferases. Glycobiology. 2003;13(1):1R–16R. 14. Kitada S, Yamada S, Kuma A, et al. Polypeptide N-acetylgalactosaminyl transferase 3 independently predicts high-grade tumours and poor prognosis in patients with renal cell carcinomas. Br J Cancer. 2013;109(2):472–81. 15. Ishikawa M, Kitayama J, Nariko H, Kohno K, Nagawa H. The expression pattern of UDP-N-acetyl-alpha-D-galactosamine: polypeptide N-acetylgalactosaminyl transferase-3 in early gastric carcinoma. J Surg Oncol. 2004;86(1):28–33. 16. Gu C, Oyama T, Osaki T, et al. Low expression of polypeptide GalNAc N-acetylgalactosaminyl transferase-3 in lung adenocarcinoma: impact on poor prognosis and early recurrence. Br J Cancer. 2004;90(2):436–42.
17. Yamamoto S, Nakamori S, Tsujie M, et al. Expression of uridine diphosphate N-acetyl-alpha-D-galactosamine: polypeptide N-acetylgalactosaminyl transferase 3 in adenocarcinoma of the pancreas. Pathobiology. 2004;71(1):12–8. 18. Berois N, Mazal D, Ubillos L, et al. UDP-N-acetyl-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase-6 as a new immunohistochemical breast cancer marker. J Histochem Cytochem. 2006;54(3):317–28. 19. Park JH, Nishidate T, Kijima K, et al. Critical roles of mucin 1 glycosylation by transactivated polypeptide N-acetylgalactosaminyltransferase 6 in mammary carcinogenesis. Cancer Res. 2010;70(7):2759–69. 20. Li Z, Yamada S, Inenaga S, et al. Polypeptide N-acetylgalactosaminyltransferase 6 expression in pancreatic cancer is an independent prognostic factor indicating better overall survival. Br J Cancer. 2011;104(12):1882–9. 21. Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471–4. 22. Zhu XD, Zhang JB, Zhuang PY, et al. High expression of macrophage colony-stimulating factor in peritumoral liver tissue is associated with poor survival after curative resection of hepatocellular carcinoma. J Clin Oncol. 2008;26(16):2707–16. 23. Detre S, Saclani Jotti G, Dowsett M. A ‘‘quickscore’’ method for immunohistochemical semiquantitation: validation for oestrogen receptor in breast carcinomas. J Clin Pathol. 1995;48(9):876–8. 24. Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bioinformatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004;10(21):7252–9. 25. Hakomori S. Glycosylation defining cancer malignancy: new wine in an old bottle. Proc Natl Acad Sci USA. 2002;99(16):10231–3. 26. Dube DH, Bertozzi CR. Glycans in cancer and inflammation– potential for therapeutics and diagnostics. Nat Rev Drug Discov. 2005;4(6):477–88. 27. Gaziel-Sovran A, Segura MF, Di Micco R, et al. miR-30b/30d regulation of GalNAc transferases enhances invasion and immunosuppression during metastasis. Cancer Cell. 2011;20(1):104–18. 28. Liu J, Liu H, Zhang W, et al. N-acetylglucosaminyltransferase V confers hepatoma cells with resistance to anoikis through EGFR/ PAK1 activation. Glycobiology. 2013;23(9):1097–109. 29. Onitsuka K, Shibao K, Nakayama Y, et al. Prognostic significance of UDP-N-acetyl-alpha-D-galactosamine:polypeptide Nacetylgalactosaminyltransferase-3 (GalNAc-T3) expression in patients with gastric carcinoma. Cancer Sci. 2003;94(1):32–6. 30. Taniuchi K, Cerny RL, Tanouchi A, et al. Overexpression of GalNAc-transferase GalNAc-T3 promotes pancreatic cancer cell growth. Oncogene. 2011;30(49):4843–54. 31. Cheng L, Tachibana K, Zhang Y, et al. Characterization of a novel human UDP-GalNAc transferase, pp-GalNAc-T10. FEBS Lett. 2002;531(2):115–21. 32. Gao Y, Liu Z, Feng J, et al. Expression pattern of polypeptide Nacetylgalactosaminyltransferase-10 in gastric carcinoma. Oncol Lett. 2012;5(1):113–6. 33. Wu YM, Liu CH, Hu RH, et al. Mucin glycosylating enzyme GALNT2 regulates the malignant character of hepatocellular carcinoma by modifying the EGF receptor. Cancer Res. 2012;71(23):7270–9. 34. Merlin J, Stechly L, de Beauce S, et al. Galectin-3 regulates MUC1 and EGFR cellular distribution and EGFR downstream pathways in pancreatic cancer cells. Oncogene. 2011;30(22):2514–25.
ORIGINAL ARTICLE – UROLOGIC ONCOLOGY
Elevated Expression of N-Acetylgalactosaminyltransferase 10 Predicts Poor Survival and Early Recurrence of Patients with Clear-Cell Renal Cell Carcinoma Qian Wu, PhD1, Liu Yang, PhD1, Haiou Liu, PhD1, Weijuan Zhang, PhD2, Xu Le, MD, PhD3, and Jiejie Xu, MD, PhD1 1
Key Laboratory of Glycoconjugate Research, Ministry of Health, Department of Biochemistry and Molecular Biology, School of Basic Medical Sciences, Shanghai Medical College of Fudan University, Shanghai, China; 2Department of Immunology, School of Basic Medical Sciences, Shanghai Medical College of Fudan University, Shanghai, China; 3 Department of Urology, Zhongshan Hospital, Shanghai Medical College of Fudan University, Shanghai, China
ABSTRACT Purpose. The aim of this study was to evaluate the potential prognostic significance of N-acetylgalactosaminyltransferase 10 (GALNT10) in patients with clear-cell renal cell carcinoma (ccRCC) after surgical resection. Methods. We retrospectively enrolled 271 patients (202 in the training cohort and 69 in the validation cohort) with ccRCC undergoing nephrectomy at a single institution. Clinicopathologic features, overall survival (OS), and recurrence-free survival (RFS) were recorded. GALNT10 intensities were assessed by immunohistochemistry in the specimens of patients. The Kaplan–Meier method was applied to compare survival curves. Cox regression models were used to analyze the impact of prognostic factors on OS and RFS. Concordance index (C-index) was calculated to assess predictive accuracy. Results. In both cohorts, elevated GALNT10 expression in tumor tissues positively correlated with advanced TNM stage. High GALNT10 expression indicated poor survival and early recurrence of patients with ccRCC, particularly
Qian Wu, Liu Yang, and Haiou Liu have contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1245/s10434-014-4236-y) contains supplementary material, which is available to authorized users. Ó Society of Surgical Oncology 2014 First Received: 16 June 2014 J. Xu, MD, PhD e-mail: [email protected]
with early-stage disease. After backward elimination, GALNT10 expression was identified as an independent adverse prognostic factor for survival and recurrence. The predictive accuracy of TNM, University of California Los Angeles Integrated Staging System, and stage, size, grade, and necrosis prognostic models was improved when GALNT10 expression was added. Conclusions. GALNT10 expression is a potential independent adverse prognostic biomarker for recurrence and survival of patients with ccRCC after nephrectomy.
Clear-cell renal cell carcinoma (ccRCC) comprises the major histological subtype of renal cell carcinoma (RCC), with approximately 40,000 patients being diagnosed annually in the US,1 and nearly 650,000 annually in China.2 Due to improved clinical diagnosis technology, such as the widespread use of abdominal imaging, more patients with ccRCC can be diagnosed at an early stage, and less than 10 % will die of cancer-related causes.3 However, nearly 30 % of patients diagnosed with localized ccRCC will undergo relapse and progression after nephrectomy, and patients with metastatic ccRCC will have a poor 5-year survival rate, ranging from 0 to 20 %.4 Previous studies showed that clinicopathological factors, such as TNM stage and Fuhrman’s nuclear grading, are independent prognostic factors.5 Eastern Cooperative Oncology Group performance status (ECOG-PS) is also used as an indicator of clinical outcome for patients with RCC. Although numerous molecular markers, including carbonic anhydrase IX (CA IX), vascular endothelial growth factor, and hypoxia-induced factor,6 and gene expression profiling 7 have been investigated, few have
Q. Wu et al.
helped identify new relevant prognostic factors. Integrated prognostic models have been used to predict clinical outcome; for example, the University of California Los Angeles Integrated Staging System (UISS) is applied when evaluating overall survival (OS) or cancer-specific survival (CSS) in RCC,8 and the Mayo Clinic stage, size, grade and necrosis (SSIGN) score more accurately predicts CSS in ccRCC.9,10 Owing to the fact that patients with similar clinical characteristics may have various outcomes, the identification of novel markers to improve the predictive accuracy of prognostic systems is urgently needed for ccRCC patients. Aberrant glycosylation is a well-known hallmark of tumors, and tumor-associated carbohydrate antigens (CAs) have been reported to be closely related to malignant transformation and disease progression in tumors.11 In mammals, mucin-type O-linked glycans constitute up to 80 % of the total cell-surface CAs.12 Mucin-type O-glycosylation occurs in the Golgi apparatus, and is initiated by UDP-N-acetyl-a-D-galactosamine:polypeptide N-acetylgalactosaminyltransferases (GALNTs), which comprise of 20 members, including GALNT1–GALNT14 and GALNTL1–GALNTL6.13 Recent studies have focused on the aberrant expression of GALNTs in different human tumors. For instance, GALNT3 has been revealed to be associated with disease aggressiveness of RCC,14 gastric carcinomas,15 lung adenocarcinoma,16 and pancreatic carcinomas,17 and GALNT6 may act as an independent indicator of breast cancer 18,19 and pancreatic cancer.20 However, few studies have been designed to explore the potential role of GALNT10 in human cancers. In this study, the expression of GALNT10 was analyzed by immunohistochemical (IHC) staining in ccRCC specimens, and its correlation with clinicopathologic features and clinical outcome, including endpoints of OS and RFS, were evaluated. In addition, incorporation of the UISS or SSIGN models with intratumoral GALNT10 expression could improve the predictive accuracy of these prognostic models. PATIENTS AND METHODS
histopathologically proven ccRCC, while exclusion criteria included other types of renal cancer as confirmed histopathologically, tumors with necrosis [80 %, and patients who died within the first month after surgery due to surgical complications. The database of all patients’ baseline clinical and pathological information and complete followup outcome data were recorded. TNM stage and Fuhrman grading were classified according to the TNM system of the 2010 American Joint Committee on Cancer (AJCC)21 and the 2004 World Health Organization (WHO) criteria,22 respectively. ECOG-PS score was assigned to each patient at the time of diagnosis. The UISS and SSIGN predictive models were applied to the patients. We calculated OS from the date of surgery to the date of death, and recurrence-free survival (RFS) was calculated from the date of recurrence to the date of death or the last follow-up from all causes. This study was approved by the hospital’s Ethics Committee, and informed consent was obtained from each patient. Tissue Microarray and Immunohistochemistry Tissue microarrays were established as previously described.22 Primary anti-GALNT10 antibody (diluted 1:600; Sigma-Aldrich, St Louis, MO, USA) was used for IHC staining. Before being employed, the specificity of the antibody was confirmed by IHC and Western blot with peptide competition. The staining intensity of each specimen was scored independently by two pathologists blinded to the clinicopathological data. A semi-quantitative Hscore for specimens was calculated by multiplying the staining intensities (0: negative; 1: weak staining; 2: moderate staining; 3: strong staining) by the area distributions (0–100 %), ranging from 0 to 300.23 For determination of the optimal cutoff score of the staining intensity applied to separate the patients into the ‘GALNT10 high’ and ‘GALNT10 low’ groups, we employed Xtile plot analysis using X-tile software v3.6.1 (Yale University, New Haven, CT, USA) to distinguish patients with poor or good prognosis to achieve the lowest p value and the highest hazard ratio (HR).24 The 50 % value was set as the cutoff point for tumoral GALNT10 density.
Patients Statistical Analysis Two independent cohorts comprising 271 consecutive patients who underwent curative nephrectomy at Zhongshan Hospital of Fudan University (Shanghai, China) were enrolled in this study. Specimens of the training cohort (n = 202) were obtained between 2003 and 2004, and specimens of the validation cohort (n = 69) were obtained in 2001. Inclusion criteria included patients with no history of previous anticancer therapy and other malignancies, bearing radical or partial nephrectomy, and
For statistical analysis, SPSS 21 (IBM Corporation, Armonk, NY, USA), MedCalc Software (version 11.4.2.0; MedCalc, Mariakerke, Belgium), and Stata 12.0 (StataCorp LP, College Station, TX, USA) were used. Numerical variables were analyzed using Student’s t test, while categorical variables were studied using the v2 or Fisher’s exact test. Survival (including OS and RFS) curves were established using the Kaplan–Meier method, and the
GALNT10 in ccRCC
difference between curves was compared using the logrank test. Number at risk was calculated at the beginning of each time period. The Cox proportional hazards regression model was applied to perform univariate and multivariate analysis, and those characteristics demonstrating statistical significance on OS and RFS in the univariate analysis were considered in the multivariate analysis. The Concordance index (C-index), which represents the area under the curve adapted for survival data, and Akaike information criterion (AIC) were evaluated to estimate and compare the predictive accuracy of multivariate models. All p values were two-sided and statistical significance was set at 0.05. RESULTS Immunohistochemical Findings GALNT10 expression was evaluated by IHC analysis in paired ccRCC specimens. As shown in electronic supplementary Fig. S1a, GALNT10 expression was observed to be higher in tumor tissues than in relevant peri-tumor tissues. The variable staining intensities are shown in electronic supplementary Fig. S1b. The average staining intensity score of peri-tumor tissues was 128, while the average score of the tumor tissues was 155 (p \ 0.001; electronic supplementary Fig. S1c). Analyzed by X-tile program to achieve a minimum p value, 150 was used as the optimal cutoff score to separate the training cohort specimens, while 140 was determined for the validation cohort. In addition, the percentage of patients with a high level of GALNT10 increased gradually, along with disease progression from TNM stages I–IV (electronic supplementary Fig. S1d). Association between GALNT10 Expression with Clinicopathological Features of Clear-Cell Renal Cell Carcinoma (ccRCC) The detailed characteristics of the two cohorts are shown in Table 1. Median follow-up was 88 months (range 11– 120 months) and 71 months (range 18–118 months) for the two cohorts, respectively. A total of 51 % (103 of 202) and 43.5 % (31 of 69) of tumor tissues were scored as having high GALNT10 staining density, respectively. Additionally, GALNT10 expression was significantly associated with Fuhrman grade (p = 0.002) in the training cohort, while no significance with any clinical characters was seen in the validation cohort, which may be due to the smaller size of patients. Furthermore, such heterogeneity may help to confirm that the predictor has universal applicability across the heterogeneous population of patients from different districts.
Prognostic Value of GALNT10 Intensity for Clinical Outcome of ccRCC Patients To further investigate the relation between GALNT10 expression and patient outcomes, we applied the Kaplan– Meier survival analysis to compare OS or RFS between GALNT10 low expression and GALNT10 high expression in the two cohorts. For RFS analysis,we excluded cases with synchronous nodal (N1) or distant metastasis (M1) at the time of surgery (n = 9 in the training cohort and n = 7 in the validation cohort). High GALNT10 was found to show poorer OS (p = 0.002 and p \ 0.001 for the training and validation cohorts, respectively; Fig. 1a, b) and RFS (p = 0.001 and p = 0.036 for the training and validation cohorts, respectively; Fig. 2a, b). We further investigated whether GALNT10 expression couldstratify patients with different TNM stage in an overall cohort of 271 patients. The dichotomized GALNT10 level was found to show strong prognostic relevance for TNM stage I patients, both for OS [HR 7.191, 95 % confidence interval (CI) 2.486–20.798, p \ 0.001; Fig. 1c] and RFS (HR 3.228, 95 % CI 1.550–6.728, p = 0.001; Fig. 2c). GALNT10 expression could also predict OS of a combination of TNM II, III, and IV patients (HR 2.106, 95 % CI 1.146–3.870, p = 0.021; Fig. 1d), while it failed to predict disease recurrence of T2–3N0M0 patients (HR 1.720, 95 % CI 0.950–3.115, p = 0.067; Fig. 2d). Univariate and multivariate analyses of the overall 271 patients were further performed. As shown in Table 2, univariate analysis revealed that GALNT10 was significantly associated with OS (HR 3.108, 95 % CI 1.879–5.141, p \ 0.001). GALNT10, together with TNM stage, Fuhrman grade, ECOG-PS, tumor size, and necrosis were estimated in multivariate analysis, and results demonstrated that GALNT10 remained an independent prognostic indicator for OS (HR 2.560, 95 % CI 1.540–4.256, p \ 0.001) as well as TNM stage (p \ 0.001), Fuhrman grade (p \ 0.001), and ECOGPS (p = 0.005) (Table 2). Univariate analysis for RFS demonstrated that a high level of GALNT10 was a risk factor for ccRCC (HR 2.341, 95 % CI 1.480–3.702, p \ 0.001). Multivariate analysis showed that T stage (p = 0.003), Fuhrman grade (p \ 0.001), ECOG-PS (p = 0.049), and GALNT10 (p = 0.002) remained as independent indicators of RFS in the overall cohorts (Table 2). Extension of Postoperative Prognostic Systems with GALNT10 Density To construct a more sensitive predictive model for outcomes of patients with ccRCC, we combined GALNT10 expression with TNM stage, and UISS or SSIGN score to testify their accuracy of survival conditions. Incorporation of GALNT10 increased the predictive value of these three models, namely when assessing OS: 0.758
Q. Wu et al. TABLE 1 Clinical characteristics and correlations with GALNT10 of two independent patient cohorts with clear-cell renal cell carcinoma Variable
N
%
Mean age (years)b
Low (n = 99)
High (n = 103)
55.3 ± 11.1
54.9 ± 12.3
Sex
p Valuea
N
%
0.812
Low (n = 38)
High (n = 31)
59.9 ± 14.1
62.2 ± 14.1
0.467
Male
140
69.3
71
69
Female
62
30.7
28
34
ECOG-PS 165
81.7
82
83
C1
37
18.3
17
20
4.4 ± 2.5
4.7 ± 2.9
Mean tumor size (cm)b T stage
50
72.5
29
21
19
27.5
9
10
57
82.6
34
23
12
17.4
4
8
4.9 ± 2.7
5.6 ± 2.9
0.096
0.207 0.340
128
63.4
67
61
45
65.2
26
19
2 3
19 52
9.4 25.7
10 20
9 32
13 8
18.9 11.6
6 5
7 3
4
3
1.5
2
1
3
4.3
1
2
0.968
0.449
N0
198
98.0
97
101
68
98.6
38
30
N1
4
2.0
2
2
1
1.4
0
1
M0
195
96.5
97
98
63
91.3
35
28
M1
7
3.5
2
5
6
8.7
3
3
122
60.4
66
56
45
65.2
25
20 5
M stage
0.271
TNM stage I
0.794
0.131
0.861
II
18
8.9
10
8
11
15.9
6
III
54
26.7
21
33
6
8.7
4
2
IV
8
4.0
2
6
7
10.2
3
4
Fuhrman grade
0.649
0.002
1
32
15.8
21
11
16
23.2
11
5
2 3
91 55
45.1 27.2
51 22
40 33
32 11
46.4 15.9
16 6
16 5
4
24
11.9
5
19
10
14.5
5
5
Necrosis
0.179
0.444
Absent
155
76.7
80
75
52
75.4
30
22
Present
47
23.3
19
28
17
24.6
8
9
UISS score
0.485
0.005
1
68
33.7
44
24
32
46.4
20
12
2
114
56.4
48
66
30
43.5
15
15
C3
20
9.9
7
13
7
10.1
3
4
50
72.5
28
22
SSIGN score 0–3
0.899
0.022 129
63.9
67
52
0.116 0.729
1
N stage
0.422 0.428
0.680
0
p Valuea
4–7
60
29.7
29
41
14
20.3
7
7
C8
13
6.4
3
10
5
7.2
3
2
Bold values are considered statistically significant (p \ 0.05) ECOG-PS Eastern Cooperative Oncology Group performance status, UISS University of California Los Angeles Integrated Staging System, SSIGN stage, size, grade, and necrosis a
A p value \0.05 is considered statistically significant
b
The results of continuous variables are expressed as mean ± SD
versus 0.684 for the TNM stage, 0.718 versus 0.673 for the UISS score, and 0.777 versus 0.734 for the SSIGN score cohort; and when assessing RFS: 0.695 versus 0.645 for the
TNM stage, 0.671 versus 0.624 for the UISS score, and 0.724 versus 0.695 for the SSIGN score (Table 3). The elevated tendency of the C-index, as well as the decreased
GALNT10 in ccRCC
Training cohort
a
Validation cohort
b
All cases
All cases 100
80 60 40 20
GALNT10 low GALNT10 high
Logrank test P= 0.002
Overall survival (%)
100
Overall survival (%)
FIG. 1 Analysis of OS considering GALNT10 expression in ccRCC tissues. Kaplan–Meier analysis of OS of patients in a the training cohort (n = 202) and b the validation cohort (n = 69), as well as c TNM stage I (n = 167), and d through TNM stages II–IV (n = 104), of an overall total cohort of 271 patients. p value was calculated using the log-rank test. OS overall survival, GALNT10 N-acetylgalactosaminyltransferase 10, ccRCC clear-cell renal cell carcinoma, HR hazard ratio, CI confidence interval
0
80 60 40 20
GALNT10 low GALNT10 high
0
24
48
72
96
120
0
24
Time after surgery (months) Number at risk GALNT10 low 99 GALNT10 high 103
c
48
72
96
120
Time after surgery (months)
97
91
76
60
0
98
87
65
50
0
Number at risk GALNT10 low 38 GALNT10 high 31
d
TNM stage I
38
31
25
18
0
29
24
14
7
0
TNM stage II+III+IV 100
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
Logrank test P< 0.001 HR: 7.191; 95%CI: 2.486-20.798
48
72
96
120
Overall survival (%)
100
Overall survival (%)
Logrank test P< 0.001
0
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
Time after surgery (months) Number at risk GALNT10 low 91 GALNT10 high 76
81
70
62
0
76
69
57
40
0
DISCUSSION In this present study, we depicted that the expression of GALNT10 varied in peri-tumor and tumor tissues in ccRCC. A higher level of GALNT10 was observed in tumor tissues and tended to positively correlate with TNM stage. Furthermore, high expression of GALNT10 means a poor outcome of OS and RFS. The dimidiate density of GALNT10 can be used to stratify TNM stage I and stage II ? III ? IV for OS, and T1N0M0 for RFS, in ccRCC. Univariate and multivariate regression analysis showed that a high level of GALNT10 is a significant independent indicator for predicting poor prognosis and early recurrence of ccRCC. In addition, incorporation of GALNT10 with postoperative prognostic models could improve predictive accuracy when estimating OS and RFS. In sum,
72
96
120
Time after surgery (months)
89
trend of AIC, suggest a better predictive accuracy. All these results suggested that a combination of GALNT10 and conventional prognostic models could generate better predictive systems for ccRCC patient outcomes.
48
Logrank test P= 0.021 HR: 2.106; 95%CI: 1.146-3.870
Number at risk GALNT10 low 46 GALNT10 high 58
46
41
31
16
0
51
42
22
17
0
analysis of GALNT10 expression may have a special function as an independent prognostic indicator. Altered glycan epitopes of cell surface proteins, accompanied with changed biological activity, might play an important role in cell–cell interaction and signal transduction, contributing to aberrant cell proliferation, migration and invasion behaviors.25 This often happens in conditions of abnormal expression or activity of glycosyltransferases. The evidence provided depicted that both Nglycosylation and O-glycosylation modification, such as glycosylation of the b1,6 N-acetyglucosaminyltransferase V (Mgat5) and GALNTs, were involved in numerous diseases,26 particularly in cancers.17,19,27–30 GALNT10 was first cloned and reported in 2002.31 Unlike most members of the GALNT family, GALNT10 shows glycoprotein glycosyltransferase activity, similar to GALNT7, the same subset of this whole family. The prior addition of N-acetylgalactosamine (GalNAc) to a synthetic peptide is necessary for catalyzing the transfer of sugar to a substrate by GALNT10.13 To date, limited studies have been conducted to investigate the function of GALNT10 in
Q. Wu et al.
b
Training cohort 100
All cases
80 60 40 20
GALNT10 low GALNT10 high
Logrank test P= 0.001
0 24
0
48
72
96
Recurrence-free survival (%)
Recurrence-free survival (%)
a
Validation cohort 100
All cases
80 60 40 20
GALNT10 low GALNT10 high
120
0
c 100
95
84
67
58
0
91
76
59
45
0
80 60 40
GALNT10 low GALNT10 high
0 0
Number at risk GALNT10 low 35 GALNT10 high 27
d
T1N0M0
20
24
Logrank test P= 0.001 HR: 3.228; 95%Cl: 1.550-60728
48
72
96
120
72
96
120
34
29
23
15
0
27
17
10
4
0
T2-3N0M0
80 60 40 20
GALNT10 low GALNT10 high
0 0
24
48
Logrank test P= 0.067 HR: 1.720; 95%Cl: 0.950-3.115
72
96
120
Time after surgery (months)
89
80
64
59
0
74
63
50
36
0
human disease. Gao et al. revealed that GALNT10 had a positive correlation with tumor differentiation, and was a useful indicator in gastric cancer.32 We consistently found that a high level of GALNT10 was seen in tumor tissues, which indicated a poor outcome of both OS and RFS of ccRCC patients. However, the underlying mechanism of GALNT10 oncogenic functions in ccRCC progression remains unknown. A prior study showed that anepidermal growth factor receptor (EGFR) could be modified by GALNT2 in hepatocellular carcinomas.33 Galectin-3, an endogenous lectin, facilitates the interaction of MUC1 with EGFR, modulating the EGFR-mediated oncogenic signaling pathways in pancreatic cancer cells.34 Our former study revealed that Mgat5 mediated bisecting N-glycosylation on the EGFR, activated the EGFR/PAK1 pathway, and contributed to hepatoma cell resistance to anoikis.28 Collectively, further investigation needs to identify whether GALNT10 could alter glycosylation and activation of EGFR or other potential substrates in ccRCC. Currently, GALNT10 expression was reported to be significantly associated with OS and RFS. In an attempt to elaborate the predictive role of GALNT10 in the overall cohort of 271 ccRCC patients, subgroup analysis with
48
100
Time after surgery (months) Number at risk GALNT10 low 91 GALNT10 high 76
24
Time after surgery (months)
Recurrence-free survival (%)
Number at risk GALNT10 low 97 GALNT10 high 96
Logrank test P= 0.036
0
Time after surgery (months)
Recurrence-free survival (%)
FIG. 2 Analysis of RFS considering GALNT10 expression in ccRCC tissues. Kaplan–Meier analysis of RFS of patients in a the training cohort (n = 193) and b the validation cohort (n = 62), as well as c stage T1N0M0 (n = 167), and d stage T2– 3N0M0 (n = 88), of an overall cohort of 271 patients. Nine patients in the training cohort and seven patients in the validation cohort with tumor metastasis at the time of surgery were excluded from the analysis of tumor recurrence. p value was calculated using the logrank test. RFS recurrence-free survival, GALNT10 Nacetylgalactosaminyltransferase 10, ccRCC clear-cell renal cell carcinoma, HR hazard ratio, CI confidence interval
Number at risk GALNT10 low 41 GALNT10 high 47
40
33
26
14
0
44
30
19
13
0
respect to TNM stage demonstrated that dichotomized GALNT10 could stratify patients with early (stage I) and higher (stages II, III, and IV) stages for OS, but the HR of early-stage patients was greater than higher-stage patients (7.191 vs. 3.228), suggesting that GALNT10 may be a more important risk factor for early-stage patients. GALNT10 evaluation could only predict disease recurrence for early-stage (stage T1N0M0) patients, suggesting GALNT10 might be more valuable for early disease recurrence. Thus, ccRCC patients with high intratumoral GALNT10 density should have more aggressive therapies and require closer follow-up. Given that the patient cluster carrying higher-stage tumors is relatively small, further studies are required to determine whether GALNT10expression can be used to estimate outcome for patients in the higher stage of disease. We acknowledge several limitations to the current study. First, of the total number of patients enrolled there were insufficient cases with higher-stage tumor; thus, we could only draw the conclusion that GALNT10 is a predictive factor for outcome, particularly in early disease recurrence, for patients with early-stage tumor. Second, this was a retrospective, single-center study, and our
GALNT10 in ccRCC TABLE 2 Univariate and multivariate Cox regression analysis for overall survival and recurrence-free survival Variable
Univariate
Multivariate
HR (95 % CI)
a
p Value
HR (95 % CI)
p Valuea
Overall survival (n = 271) Age (years)b
1.020 (1.000–1.040)
0.048
Sex (female vs. male)
1.137 (0.688–1.880)
0.616
ECOG-PS (C1 vs. 0)
3.570 (2.244–5.678)
\ 0.001
2.085 (1.255–3.466)
Tumor size (cm)b
1.197 (1.118–1.281)
\0.001
1.043 (0.951–1.143)
0.370
TNM stage (II ? III ? IV vs. I)
3.761 (2.347–6.027)
\ 0.001
2.747 (1.571–4.803)
\ 0.001
Fuhrman grade (3 ? 4 vs. 1 ? 2) Necrosis (yes vs. no)
2.829 (1.793–4.464) 2.463 (1.548–3.917)
\0.001 \0.001
2.366 (1.488–3.763) 1.596 (0.992–2.568)
\0.001 0.054
GALNT10 (high vs. Low)
3.108 (1.879–5.141)
\0.001
2.560 (1.540–4.256)
\0.001
Age (years)a
1.013 (0.994–1.032)
0.192
Sex (female vs. male)
0.744 (0.472–1.172)
0.202
ECOG-PS (C1 vs. 0)
2.597 (1.600–4.216)
\0.001
1.693 (1.002–2.859)
0.049 0.322
0.005
Recurrence-free survival (n = 255)c
b
1.170 (1.093–1.253)
\0.001
1.048 (0.955–1.149)
T stage (2 ? 3 ? 4 vs. 1)
3.041 (1.956–4.725)
\0.001
2.434 (1.431–4.140)
0.003
Fuhrman grade (3 ? 4 vs. 1 ? 2)
2.696 (1.739–4.178)
\0.001
2.432 (1.555–3.804)
\0.001
Tumor size (cm)
Necrosis (yes vs. no)
1.945 (1.218–3.105)
0.005
1.312 (0.812–2.122)
0.268
GALNT10 (high vs. low)
2.341 (1.480–3.702)
\0.001
2.041 (1.286–3.240)
0.002
Bold values are considered statistically significant (p \ 0.05) HR hazard ratio, CI confidence interval, ECOG-PS Eastern Cooperative Oncology Group performance status, GALNT10 polypeptide Nacetylgalactosaminyl transferase 10, ccRCC clear-cell renal cell carcinoma a
A p value \0.05 is considered statistically significant
b
Factors were treated as continuous variables
c
For recurrence-free survival analysis, 16 patients with metastasis ccRCC were excluded
TABLE 3 Comparison of the prognostic accuracies of models for overall survival and recurrence-free survival Models
Overall survival (n = 271)
Recurrence-free survival (n = 255)a
C-index AIC
C-index
AIC
GALNT10
0.643
788.306 0.612
841.817
TNM stage
0.684
767.346 0.645
828.654
TNM stage ? GALNT10
0.758
750.214 0.695
818.489
UISS score
0.673
517.766 0.624
595.119
UISS score ? GALNT10
0.718
511.656 0.671
590.476
SSIGN score
0.734
742.598 0.695
814.941
SSIGN score ? GALNT10 0.777
731.715 0.724
808.781
An elevated C-index or a decreased AIC score mean a better prognosis C-index Harrell’s concordance index, AIC Akaike information criterion, GALNT10 polypeptide N-acetylgalactosaminyl transferase 10, UISS University of California Los Angeles Integrated Staging System, SSIGN stage, size, grade, and necrosis, ccRCC clear-cell renal cell carcinoma a
For recurrence-free survival analysis, 16 patients with metastasis ccRCC are excluded
findings should thus be replicated in other populations and larger cohorts to further validate our conclusions. Third, further calculations regarding the endpoint of CSS had to be omitted, and whether or not employing GALNT10 resulted in a well predictive value in other RCC subtypes still needs to be explored. Fourth, a prospective study enrolling more patients stratified by more detailed clinicopathologic characteristics and diverse histological subtypes could further validate our current findings. We are seeking to cooperate with other academic centers. In addition, the detailed mechanisms ofGALNT10 oncogenic functions in ccRCC need further laboratory and clinical investigation. ACKNOWLEDGMENT This work was supported by grants from the National Basic Research Program of China (2012CB822104), the National Key Projects for Infectious Diseases of China (2012ZX10002-012), the National Natural Science Foundation of China (31100629, 31270863, 81471621, 81472227), the Program for New Century Excellent Talents in University (NCET-13-0146), and the Shanghai Rising-Star Program (13QA1400300). All these study sponsors had no role in the study design, or in the collection, analysis, and interpretation of data. We thank Ms. Haiying Zeng
Q. Wu et al. (Department of Pathology, Zhongshan Hospital, Shanghai Medical College of Fudan University) for technical assistance. DISCLOSURE
The authors have declared no conflicts of interest.
REFERENCES 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63(1):11–30. 2. Yang L, Parkin DM, Ferlay J, Li L, Chen Y. Estimates of cancer incidence in China for 2000 and projections for 2005. Cancer Epidemiol Biomark Prev. 2005;14(1):243–250. 3. Joseph RW, Kapur P, Serie DJ, et al. Loss of BAP1 protein expression is an independent marker of poor prognosis in patients with low-risk clear cell renal cell carcinoma. Cancer. 2013;120(7): 1059–67. 4. Stewart GD, O’Mahony FC, Powles T, Riddick AC, Harrison DJ, Faratian D. What can molecular pathology contribute to the management of renal cell carcinoma? Nat Rev Urol. 2011;8(5): 255–65. 5. Lang H, Lindner V, de Fromont M, et al. Multicenter determination of optimal interobserver agreement using the Fuhrman grading system for renal cell carcinoma: assessment of 241 patients with [15-year follow-up. Cancer. 2005;103(3):625–9. 6. Zhao H, Ljungberg B, Grankvist K, Rasmuson T, Tibshirani R, Brooks JD. Gene expression profiling predicts survival in conventional renal cell carcinoma. PLoS Med. 2006;3(1):e13. 7. Sim SH, Messenger MP, Gregory WM, et al. Prognostic utility of pre-operative circulating osteopontin, carbonic anhydrase IX and CRP in renal cell carcinoma. Br J Cancer. 2012;107(7):1131–7. 8. Zisman A, Pantuck AJ, Dorey F, et al. Improved prognostication of renal cell carcinoma using an integrated staging system. J Clin Oncol. 2001;19(6):1649–57. 9. Frank I, Blute ML, Cheville JC, Lohse CM, Weaver AL, Zincke H. An outcome prediction model for patients with clear cell renal cell carcinoma treated with radical nephrectomy based on tumor stage, size, grade and necrosis: the SSIGN score. J Urol. 2002;168(6):2395–400. 10. Sun M, Shariat SF, Cheng C, et al. Prognostic factors and predictive models in renal cell carcinoma: a contemporary review. Eur Urol. 2011;60(4):644–61. 11. Brockhausen I. Pathways of O-glycan biosynthesis in cancer cells. Biochim Biophys Acta. 1999;1473(1):67–95. 12. Hollingsworth MA, Swanson BJ. Mucins in cancer: protection and control of the cell surface. Nat Rev Cancer. 2004;4(1):45–60. 13. Ten Hagen KG, Fritz TA, Tabak LA. All in the family: the UDPGalNAc:polypeptide N-acetylgalactosaminyltransferases. Glycobiology. 2003;13(1):1R–16R. 14. Kitada S, Yamada S, Kuma A, et al. Polypeptide N-acetylgalactosaminyl transferase 3 independently predicts high-grade tumours and poor prognosis in patients with renal cell carcinomas. Br J Cancer. 2013;109(2):472–81. 15. Ishikawa M, Kitayama J, Nariko H, Kohno K, Nagawa H. The expression pattern of UDP-N-acetyl-alpha-D-galactosamine: polypeptide N-acetylgalactosaminyl transferase-3 in early gastric carcinoma. J Surg Oncol. 2004;86(1):28–33. 16. Gu C, Oyama T, Osaki T, et al. Low expression of polypeptide GalNAc N-acetylgalactosaminyl transferase-3 in lung adenocarcinoma: impact on poor prognosis and early recurrence. Br J Cancer. 2004;90(2):436–42.
17. Yamamoto S, Nakamori S, Tsujie M, et al. Expression of uridine diphosphate N-acetyl-alpha-D-galactosamine: polypeptide N-acetylgalactosaminyl transferase 3 in adenocarcinoma of the pancreas. Pathobiology. 2004;71(1):12–8. 18. Berois N, Mazal D, Ubillos L, et al. UDP-N-acetyl-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase-6 as a new immunohistochemical breast cancer marker. J Histochem Cytochem. 2006;54(3):317–28. 19. Park JH, Nishidate T, Kijima K, et al. Critical roles of mucin 1 glycosylation by transactivated polypeptide N-acetylgalactosaminyltransferase 6 in mammary carcinogenesis. Cancer Res. 2010;70(7):2759–69. 20. Li Z, Yamada S, Inenaga S, et al. Polypeptide N-acetylgalactosaminyltransferase 6 expression in pancreatic cancer is an independent prognostic factor indicating better overall survival. Br J Cancer. 2011;104(12):1882–9. 21. Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17(6):1471–4. 22. Zhu XD, Zhang JB, Zhuang PY, et al. High expression of macrophage colony-stimulating factor in peritumoral liver tissue is associated with poor survival after curative resection of hepatocellular carcinoma. J Clin Oncol. 2008;26(16):2707–16. 23. Detre S, Saclani Jotti G, Dowsett M. A ‘‘quickscore’’ method for immunohistochemical semiquantitation: validation for oestrogen receptor in breast carcinomas. J Clin Pathol. 1995;48(9):876–8. 24. Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bioinformatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004;10(21):7252–9. 25. Hakomori S. Glycosylation defining cancer malignancy: new wine in an old bottle. Proc Natl Acad Sci USA. 2002;99(16):10231–3. 26. Dube DH, Bertozzi CR. Glycans in cancer and inflammation– potential for therapeutics and diagnostics. Nat Rev Drug Discov. 2005;4(6):477–88. 27. Gaziel-Sovran A, Segura MF, Di Micco R, et al. miR-30b/30d regulation of GalNAc transferases enhances invasion and immunosuppression during metastasis. Cancer Cell. 2011;20(1):104–18. 28. Liu J, Liu H, Zhang W, et al. N-acetylglucosaminyltransferase V confers hepatoma cells with resistance to anoikis through EGFR/ PAK1 activation. Glycobiology. 2013;23(9):1097–109. 29. Onitsuka K, Shibao K, Nakayama Y, et al. Prognostic significance of UDP-N-acetyl-alpha-D-galactosamine:polypeptide Nacetylgalactosaminyltransferase-3 (GalNAc-T3) expression in patients with gastric carcinoma. Cancer Sci. 2003;94(1):32–6. 30. Taniuchi K, Cerny RL, Tanouchi A, et al. Overexpression of GalNAc-transferase GalNAc-T3 promotes pancreatic cancer cell growth. Oncogene. 2011;30(49):4843–54. 31. Cheng L, Tachibana K, Zhang Y, et al. Characterization of a novel human UDP-GalNAc transferase, pp-GalNAc-T10. FEBS Lett. 2002;531(2):115–21. 32. Gao Y, Liu Z, Feng J, et al. Expression pattern of polypeptide Nacetylgalactosaminyltransferase-10 in gastric carcinoma. Oncol Lett. 2012;5(1):113–6. 33. Wu YM, Liu CH, Hu RH, et al. Mucin glycosylating enzyme GALNT2 regulates the malignant character of hepatocellular carcinoma by modifying the EGF receptor. Cancer Res. 2012;71(23):7270–9. 34. Merlin J, Stechly L, de Beauce S, et al. Galectin-3 regulates MUC1 and EGFR cellular distribution and EGFR downstream pathways in pancreatic cancer cells. Oncogene. 2011;30(22):2514–25.
