α-Internexin: a novel biomarker for pancreatic neuroendocrine tumor aggressiveness.

J C E M

O N L I N E

Hot Topics in Translational Endocrinology—Endocrine Research

␣-Internexin: A Novel Biomarker for Pancreatic Neuroendocrine Tumor Aggressiveness Bei Liu,* Laura H. Tang,* Zhaojun Liu,* Mei Mei, Run Yu, Deepti Dhall, Xin-Wei Qiao, Tai-Ping Zhang, Yu-Pei Zhao, Tong-Hua Liu, Yu Xiao, Jie Chen, Hong-Ding Xiang, Hai-Yan Wu, Chong-Mei Lu, Bin Lv, Ya-Ru Zhou, Ye Zhang, Dajun Deng, and Yuan-Jia Chen Departments of Gastroenterology (B.Li., M.M., X.-W.Q., H.-Y.W., C.-M.L., Y.-J.C.), Surgery (T.-P.Z., Y.-P.Z.), and Pathology (T.-H.L., Y.X., J.C.) and Key Laboratory of Endocrinology (Ministry of Health) (H.-D.X., Y.-J.C.), Department of Endocrinology, Peking Union Medical College Hospital, and Department of Biochemistry and Molecular Biology (Y.Z.), Institute of Basic Medical Sciences, Peking Union Medical College, Chinese Academy of Medical Sciences, Beijing 100730, China; Department of Pathology (L.H.T.), Memorial Sloan-Kettering Cancer Center, New York, New York 10065; Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education) (Z.L., D.De.), Department of Etiology, Peking University Cancer Hospital/Institute, Beijing 100142, China; Departments of Endocrinology (R.Y.) and Pathology (D.Dh.), Cedars-Sinai Medical Center, University of California, Los Angeles, California 90048; Department of Gastroenterology (B.Lv.), the First Affiliated Hospital of Zhejiang Chinese Medical University, Hangzhou 310006, China; and Department of Endocrinology (Y.-R.Z.), The Third Hospital of Hebei Medical University, Shijiazhuang 050051, China

Purpose: We aimed to test whether ␣-internexin could be a molecular biomarker of tumor aggressiveness and prognosis in pancreatic neuroendocrine tumors (PNETs). Patients and Methods: Using immunohistochemical staining and Western blot, we detected the expression of ␣-internexin in 350 tumors from 343 patients, of whom 257 were followed up. Methylation of ␣-internexin promoter was examined by bisulfite sequencing to identify the crucial region that determines gene expression. Methylation of gene promoter in tumors was quantitatively measured by denaturing high performance liquid chromatography (DHPLC). We correlated ␣-internexin expression with clinicopathological characteristics. Results: ␣-Internexin was expressed in 53% of 350 PNETs. The reduced expression of ␣-internexin was significantly associated with advanced stage (P ⬍ .0001), metastases (P ⬍ .0001), and recurrence (P ⫽ .003). ␣-Internexin expression was found in 57.1% of 212 surviving patients and in 17.1% of 35 deceased patients (P ⬍ .0001). Reduced expression of ␣-internexin was associated with shorter overall survival in PNET patients (log rank P ⬍ .0001) as well as in patients with noninsulinoma and nonfunctional (NF)-PNETs (log rank P ⫽ 0.0073 and P ⫽ 0.010, respectively). The crucial region of ␣-internexin promoter (⫺149 to ⫹96 nucleotides [nt]) was identified, and the hypomethylation of this area in PNETs was significantly associated with gene expression (P ⫽ .015). Conclusion: ␣-Internexin can be a useful prognostic biomarker for PNETs. (J Clin Endocrinol Metab 99: E786 –E795, 2014)

P

ancreatic neuroendocrine tumors (PNETs) are a group of uncommon tumors with distinct clinical syndromes, and most patients with PNETs present with ad-

vanced disease (1–7). The incidence and prevalence of PNETs have increased worldwide over the past 3 decades (8, 9). The molecular pathogenesis of sporadic PNETs is

ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received July 16, 2013. Accepted January 21, 2014. First Published Online January 31, 2014

* B.L., L.H.T., and Z.L. contributed equally to this work. Abbreviations: 5-aza-CdR, 5-aza-2⬘-deoxycytidine; DHPLC, denaturing high performance liquid chromatography; F, functional; IHC, immunohistochemical; NF, nonfunctional; nt, nucleotide; PNET, pancreatic neuroendocrine tumor.

E786

jcem.endojournals.org

J Clin Endocrinol Metab, May 2014, 99(5):E786 –E795

doi: 10.1210/jc.2013-2874

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 30 September 2014. at 05:18 For personal use only. No other uses without permission. . All rights reserved.

doi: 10.1210/jc.2013-2874

largely unknown (3, 4, 10), although inactivation of the MEN1 gene contributes to a subgroup of sporadic PNETs and intriguing advances in the genetic tumorigenesis have been on the horizon (3, 11–15). Insights into the molecular pathogenesis of PNETs, especially sporadic PNETs, are important for deciphering the genetic/epigenetic mechanisms underlying the tumorigenesis and for predicting the aggressiveness of PNETs. At present, there are only few reliable molecular biomarkers to predict the prognosis of the patients with PNETs and the outcomes in subsets of PNETs (1– 4, 16). If molecular biomarkers could be used to predict unfavorable prognosis in a patient with PNET, the patient would benefit from more aggressive antitumor therapy or more intensive care. Our recent proteomic study on insulinoma (data not published) showed that a cytoskeleton protein, ␣-internexin, was strongly expressed in insulinomas but not in their paired tissues. ␣-Internexin is mainly expressed in the nervous system (17, 18) and in neuroblastomas, which share some common features with neuroendocrine tumors (19). It has been reported that cytoskeletal proteins might play a role in the tumor initiation and progression (4, 20). Thus, we hypothesized that ␣-internexin could be expressed in PNETs, and it might be used to identify the tumor aggressiveness as well as to predict their prognosis. The purpose of this study were to demonstrate the expression of ␣-internexin in PNETs and to determine whether the protein could be used as a biomarker of tumor aggressiveness and prognosis. Moreover, the epigenetic mechanism determining ␣-internexin expression was investigated.


Table 1. Summary of Clinicopathological Features of PNET Patientsa Clinical Features Gender, n (%) Male Female Median age at surgery, y (range) Male Female PNET function type, n (%) F-PNET NF-PNET PNET subgroup, n (%) Insulinoma Noninsulinoma Metastasis, n (%) No metastasis Metastasis LN metastasis only Distant metastasis only LN and distant metastasis Grade, n (%) 1 2 3 Stage, n (%) I IIa IIb IIIa IIIb IV Follow-up information Available Not available Follow-up months, median (range) Disease-free survival (DFS), n (%) Alive with disease (AWD), n (%) Died of disease (tumor), n (%) Died of unknown cause, n (%) Survival with unknown status, n (%)b Clinicopathological features of tumors Primary tumor location, n (%) Pancreatic head Pancreatic body and/or tail Nonpancreas Tumor size, n (%) ⬍3 cm ⱖ3 cm Ki-67, n (%) ⬉2% ⬎2%

Patients and Methods Clinical characteristics and management From 343 patients with PNETs evaluated between 1988 and 2012 at Peking Union Medical College Hospital, Memorial Sloan-Kettering Cancer Center, and Cedars-Sinai Medical Center, 350 tumors were included in this study, which was approved by the Scientific Ethics Committee of each hospital. The diagnostic criteria for PNETs were reported previously (21, 22). Briefly, the tumors were mainly localized by computed tomography with contrast, magnetic resonance imaging, endoscopic ultrasound, and somatostatin receptor scintigraphy (21, 22). The 350 tumors and 145 paired pancreatic tissue specimens were studied (Table 1). Of 290 patients, 102 patients (35.2%) underwent enucleations; 77 patients (26.6%) had head, body, and tail resection; 59 patients (20.3%) had tail resection and splenectomy; and 52 (17.9%) underwent Whipple procedures. The pathological diagnosis was made by experienced pathologists. In the study, 257 patients were followed up from April 1989 until December 2013. The clinicopathological characteristics were analyzed and correlated with the experimental data.

E787

n ⴝ 343 153 (44.6) 190 (55.4) 51 (15– 85) 46 (17– 84) 199 (58.0) 144 (42.0) 160 (46.6) 183 (53.4) 238 (69.8) 103 (30.2) 28 (27.2) 33 (32.0) 22 (21.4) 150 (58.6) 97 (37.9) 9 (3.5) 96 (28.8) 100 (30.0) 44 (13.2) 2 (0.6) 35 (10.5) 56 (16.8) 257 (74.9) 86 (25.1) 61 (2–251) 175 (68.1) 35 (13.6) 35 (13.6) 10 (3.9) 2 (0.8) n ⴝ 350 131 (40.7) 187 (58.1) 4 (1.2) 176 (54.2) 149 (45.8) 195 (72.8) 73 (27.2)

a

Metastasis data were from 341 patients, grade data were from 256 patients, stage data were from 333 patients, primary location data were from 322 tumors, tumor size were from 325 tumors, and Ki-67 data were from 268 patients. b

Two patients were alive, but their disease status was unknown.

A total of 103 patients had either lymph node or distant metastases, whereas 238 patients did not have evidence of metastases. Ki-67 index was assessed in 268 tumors. We analyzed tumor grade in 256 tumors according to European Neuroendocrine Tumor Society (ENETS)/World Health Or-


E788

Liu et al

␣-Internexin in Pancreatic Endocrine Tumors

ganization guidelines (23) and analyzed stage in 333 patients according to the ENETS guidelines (23). Tumors 150 and 151, 207 and 208, 217 and 218, 281 and 282, 316 and 317, 340 and 341, 357 and 358 were from the same respective patients (Supplemental Table 1).


performance liquid chromatography (DHPLC) (Supplemental Methods 5) as previously described (24). DNA from the ␣-internexin gene unmethylated cell line PA1 and methylated cell line MGC803 was used as negative and positive control, respectively.

Statistical analysis Immunohistochemical staining, Western blot, and RT-PCR Sections of paraffin-embedded tumor tissue and paired pancreatic tissue were antigen-retrieved by microwave heating and stained with anti–␣-internexin (MAB5224) from Chemicon at a 1:100 dilution and rabbit antimouse IgG/horseradish peroxidase (catalog item 67834; Beijing Zhongshan Golden Bridge Biotechnology Co). The results were interpreted by 2 pathologists blinded to clinical data and patient outcome. The semiquantitative grading of immunohistochemical (IHC) results was summarized in Supplemental Methods 1 (published on The Endocrine Society’s Journals Online website at http://jcem.endojournals.org), similar to that in our previous report (22). Proteins extracted from fresh-frozen PNET tissues and the paired tissues were examined by Western blot with the anti–␣-internexin at 1:200 dilution and an enhanced chemiluminescence detection system (Amersham). ␤-Actin was used as an internal control. The expression of ␣-internexin was correlated with clinicopathological characteristics.

Identification of a crucial methylated region on the ␣-internexin promoter by bisulfite sequencing Fourteen human cancer cell lines (Supplemental Table 2) were cultured in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (HyClone). Genomic DNA isolated from these cell lines was bisulfite-modified by using EZ DNA methylation direct Kit (Zymo Research). The promoter of the ␣-internexin gene (Genbank accession no. NC_000010.9) was amplified by PCR focused on the ⫺787 to ⫹113-nucleotide (nt) region (⫹1 nt, transcription start site) including 60 CpG sites. PCR conditions and 4 CpG-free primer sets are summarized in Supplemental Methods 2 and Supplemental Figure 1. PCR products were sequenced as previously described (22). Protein was isolated for Western blot. Simultaneously, total RNA was isolated and 5 ␮g RNA was reverse-transcribed using Easy Script (TRANS) for RT-PCR. The cDNA was amplified with ␣-internexin primers (Supplemental Methods 3).

Reactivation of ␣-internexin expression by treatment with 5-aza-2ⴕ-deoxycytidine (5-aza-CdR) A375 cell line was treated with an inhibitor of DNA methyltransferase 5-aza-2⬘-deoxycytidine (5-aza-CdR) for 24, 48, and 72 hours, respectively, at various concentrations of 0␮M, 0.5␮M, 1␮M, and 2␮M. Dimethyl sulfoxide was used as control. The genomic DNA was isolated from these cells for bisulfite sequencing (Supplemental Figure 1 and Methods 2). The expression of ␣-internexin mRNA was tested by RT-PCR.

Quantitative detection of ␣-internexin methylation in tumors Genomic DNA isolated from 17 fresh-frozen tumors and 8 paired tissues was treated with bisulfite. The promoter region of ␣-internexin (⫺107 to ⫹62 nt) was amplified by PCR with the universal primers set 5 (Supplemental Figure 1 and Methods 4). PCR products were quantitatively analyzed by denaturing high

Significance was calculated using ␹2 test, Fisher’s exact test, and Mann-Whitney U test. Logistic regression was used for the multivariate analysis. A two-tailed test was used in all statistic analyses. The survival of patients was analyzed by the KaplanMeier method and Cox’s proportional hazards model. P ⱕ .05 was considered significant.

Results Clinicopathological characteristics We studied 350 PNETs from 343 patients. There were 147 NF-PNETs and 203 functional (F)-PNETs, including 162 insulinomas, 23 gastrinomas, 7 VIPomas, 10 glucagonomas, and 1 ACTHoma (Supplemental Table 1). Of the 343 patients, 257 patients (74.9%) were followed up; the median follow-up interval was 61 (range 2–251) months. Thirty-five patients died of the tumors, whereas 10 patients died of unknown reasons. Of the 212 surviving patients, 175 did not exhibit evidence of recurrent disease, 35 patients survived with progressive disease, and 2 patients were alive, but we could not ascertain whether they had disease or not. All tumors were well-differentiated, and most of these tumors (96.5%) were classified as grade 1 and grade 2, (Table 1 and Supplemental Table 1). Ki-67 index was assessed in 268 PNETs, and 195 tumors (72.8%) showed Ki-67 index ⬉2% (Table 1 and Supplemental Table 1). According to the ENETS guidelines, stage was assessed in 333 patients with PNETs. The clinicopathological features of each patient are listed in Supplemental Table 1 and summarized in Table 1. Expression of ␣-internexin in PNETs ␣-Internexin was expressed in 187 of 350 PNETs (53.4%) as detected by IHC and Western blot (Figure 1). In 203 F-PNETs, the expression of ␣-internexin was found in 137 tumors (67.5%), whereas the rate of protein expression in NF-PNETs was much lower than that in FPNETs (34% vs 67.5%, P ⫽ 5.78 ⫻ 10⫺10). ␣-Internexin was expressed in 121 of 162 insulinomas (74.7%), significantly higher than that (35.1%) in 188 noninsulinomas (P ⫽ 1.34 ⫻ 10⫺13). Correlation between ␣-internexin expression and tumor grading, staging, metastases, recurrence, and survival We correlated clinicopathological features with ␣-internexin expression (Supplemental Table 3). The loss of or


doi: 10.1210/jc.2013-2874


E789

who died of unknown reasons. We further analyzed the prognostic value of ␣-internexin in subtypes of PNET. The patients with PNETs showing reduced/lost expression of ␣-internexin had a shorter overall survival time (Figure 2A) (log-rank P ⫽ 7.3 ⫻ 10⫺5; Cox model 1: hazard ratio [HR] 4.285, 95% confidence interval [CI] 1.417–12.959, P ⫽ .010; Cox model 2: HR,5.750; 95% CI 1.937–17.07, P ⫽ .002; see Table 2) and a shorter disease-free survival time (Figure 2B) (log-rank P ⫽ .0057) than those with tumors expressing the protein, but a multivariable Cox model demonstrated that ␣-internexin was not an independent prognostic marker of disease-free survival (HR 1.547, 95% CI 0.705–3.397, P ⫽ .277). Noninsulinoma Lost/reduced expression of ␣-internexin was significantly associated with a shortened overall survival time (Figure 2C) (log-rank P ⫽ .0073; Cox model 1: HR 7.150, 95% CI 1.556 –32.845, P ⫽ .011; Cox model 2: HR 6.330, 95% CI 1.448 –27.675, P ⫽ .014; see Table 2) but not associated with disease-free survival (Figure 2D) (logrank P ⫽ .460).

Figure 1. Representative examples of ␣-internexin expression in PNETs and paired pancreatic tissues. A, top row left, hematoxylin and eosin staining of an insulinoma (upper part) and its paired pancreatic tissue (lower part); top row right, positive IHC staining of ␣-internexin in the tumoral tissue, with no expression of ␣-internexin in its paired pancreatic tissue; second row left, hematoxylin and eosin staining of an insulinoma; second row right, no expression of ␣-internexin in the insulinoma; third row left, hematoxylin and eosin staining of a glucagonoma; third row right, the glucagonoma with intermediate staining of ␣-internexin; bottom row left, a malignant NF-PNET without staining signals; bottom row right, a benign insulinoma with very strong ␣-internexin staining signals. Scale bar, 50 ␮m. B, Expression of ␣-internexin in 5 PNETs by Western blot (upper panel), and no expression of ␣-internexin in 6 normal pancreatic tissues (lower panel). SY5Y cell line was used as positive control. The number of tumors is identical to the number in Supplemental Table 1.

reduced ␣-internexin expression was significantly associated with age, recurrence, stage, metastasis, tumor size, death, and poor prognosis (ie, either death or survival with advanced disease) in PNETs but not correlated with gender, grade, Ki-67 index, or primary tumor location. (Supplemental Table 3). In the subtypes of PNETs, ␣-internexin expression was also associated with tumor stage, metastasis, and prognosis (Supplemental Table 3). The rate of ␣-internexin expression in patients alive without disease, patients alive with advanced disease, and deceased patients was 60%, 31.4%, and 17.1%, respectively, suggesting that ␣-internexin expression is correlated with favorable prognosis. We analyzed the survival of 247 patients who were followed up using Kaplan-Meier method, excluding 10 patients

Insulinoma Kaplan-Meier analysis showed that the expression of ␣-internexin was not significantly associated with overall survival or a disease-free survival in patients with insulinomas (Figure 2, E and F) log-rank P ⫽ .070 and P ⫽ .190, respectively). Functional PNETs The patients with F-PNETs showing lost/reduced expression of ␣-internexin had an unfavorable overall survival (Figure 2G) (log-rank P ⫽ .024; Cox model 1: HR 3.837, 95% CI 0.883–16.679, P ⫽ .073; Cox model 2: HR 3.490, 95% CI 0.825–14.768, P ⫽ .089; see Table 2). The patients with F-PNETs showing lost/reduced expression of ␣-internexin had an unfavorable disease-free survival than those with tumors expressing the protein (Figure 2H) (log-rank P ⫽ .012), but a multivariable Cox model showed that expression ␣-internexin was not an independent prognostic marker in F-PNETs (HR 1.365, 95% CI 0.401– 4.647, P ⫽ .619). Nonfunctional (NF) PNETs Patients with NF-PNETs showing lost/reduced ␣-internexin expression had a shortened overall survival (Figure 2I) (log-rank P ⫽ .010, Cox model 1: HR 7.605, 95% CI, 0.887– 65.226, P ⫽ .064; Cox model 2: HR 10.785, 95% CI, 1.344 – 86.564, P ⫽ .025; see Table 2). The expression of ␣-internexin did not correlate with disease-free survival in NF-PNETs patients (Figure 2J) (P ⫽ .280).


E790

Liu et al



Figure 2. Kaplan-Meier plots of patients with PNETs and subtypes of PNETs. A, C, E, G, and I, Kaplan-Meier plots of overall survival. B, D, F, H, and J, Kaplan-Meier plots of disease-free survival.

Identification of a crucial methylated region on the ␣-internexin promoter and its correlation with its expression in vivo and in vitro To identify the crucial CpG sites in the promoter of ␣-internexin, we analyzed the methylation status of the

promoter in 14 cancer cell lines with varying expression of ␣-internexin using bisulfite sequencings (Supplemental Figure 1). Four of the 14 cell lines showed expression of ␣-internexin in Western blot and RT-PCR (Figure 3, A and B). Bisulfite sequencing of the ⫺787 to ⫹113-nt region of


doi: 10.1210/jc.2013-2874

Table 2.


Multivariable Cox Models of Overall Survival Model 1

Variable All PNETs Age, y Gender Size, cm Ki-67 Stage INX expressiona Noninsulinomas Age, y Gender Size, cm Ki-67 Stage INX expression F-PNETs Age, y Gender Size, cm Ki-67 Stage INX expression NF-PNETs Age, y Gender Size, cm Ki-67 Stage INX expression a

E791

Model 2

HR (95% CI)

P Value

HR (95% CI)

P Value

0.998 (0.966 to 1.030) 0.501 (0.222 to 1.129) 0.941 (0.843 to 1.049) 6.810 (2.564 to 18.088) 2.279 (1.442 to 3.603) 4.285 (1.417 to 12.959)

.885 .095 .274 1.18 ⫻ 10⫺4 4.23 ⫻ 10⫺4 .010

0.995 (0.966 to 1.024) 0.586 (0.265 to 1.295) 1.007 (0.908 to 1.117) 9.348 (3.591 to 24.332)

.716 .186 .892 4.664 ⫻ 10⫺6

5.750 (1.937 to 17.070)

.002

0.998 (0.959 to 1.038) 0.400 (0.153 to 1.045) 0.920 (0.810 to 1.045) 4.460 (1.509 to 13.179) 2.184 (1.190 to 4.005) 7.150 (1.556 to 32.845)

.920 .061 .202 .007 .012 .011

0.993 (0.958 to 1.028) 0.652 (0.268 to 1.589) 0.985 (0.878 to 1.106) 6.621 (2.304 to 19.025)

.682 .347 .802 4.480 ⫻ 10⫺4

6.330 (1.448 to 27.675)

.014

0.984 (0.924 to 1.047) 0.860 (0.201 to 3.672) 1.275 (0.752 to 2.163) 7.392 (1.060 to 51.566) 2.892 (1.335 to 6.262) 3.837 (0.883 to 16.679)

.601 .838 .367 .044 .007 .073

0.983 (0.933 to 1.035) 1.150 (0.293 to 4.515) 1.448 (0.966 to 2.171) 7.533 (1.196 to 47.460)

.516 .841 .073 .032

3.490 (0.825 to 14.768)

.089

0.980 (0.934 to 1.028) 0.242 (0.072 to 0.817) 0.900 (0.777 to 1.044) 5.070 (1.379 to 18.649) 2.144 (1.037 to 4.433) 7.605 (0.887 to 65.226)

.410 .022 .163 .015 .039 .064

0.983 (0.942 to 1.025) 0.351 (0.115 to 1.068) 0.967 (0.850 to 1.099) 8.026 (2.428 to 26.538)

.426 .065 .602 .001

10.785 (1.344 to 86.564)

.025

INX expression means loss or reduced expression of ␣-internexin.

the ␣-internexin promoter showed that the crucial region (⫺107 to ⫹96 nt) was hypermethylated in 9 of 9 cell lines with silenced gene expression (Figure 3C, lower frame) but demethylated in 5 cell lines of which 4 cell lines had gene expression (Figure 3C, upper frame), indicating the methylation status of the 25 CpG sites within this region was crucially associated with the gene expression in vitro. To further confirm these findings, the A375 cell line without ␣-internexin expression was treated with an inhibitor of methyltransferase, 5-aza-CdR. RT-PCR showed subsequent ␣-internexin mRNA was reexpressed (Supplemental Figure 2). The demethylation of promoter was verified by bisulfite sequencing (Supplemental Figure 3). These data suggested that demethylation of the crucial region of ␣-internexin promoter was associated with gene expression in vitro. The methylation status of the ␣-internexin promoter was also quantitatively detected in 25 fresh-frozen specimens (17 PNETs and 8 paired tissues) by DHPLC (Figure 4). In 11 tumor samples that expressed the protein, the median value of methylation was 0.065 (0.012– 0.407), whereas it was 0.266 (0.042– 0.415) in 14 samples without ␣-internexin expression (P ⫽ .015), suggesting hy-

pomethylation of the ␣-internexin gene was associated with protein expression in vivo.

Discussion Recently, the ENETS and the International Union for Cancer Control/American Joint Cancer Committee /World Health Organization proposed 2 TNM (tumor, nodes, metastasis) staging systems for PNETs, which along with tumor grading were useful for predicting the prognosis of patients with PNETs (24). However, molecular biomarkers are still needed to predict the outcome of an individual patient with PNETs so that the patient would benefit from an individualized therapy. A number of molecular investigations on PNETs including profiling studies have provided insights into molecular pathogenesis of PNETs and identified several novel prognostic biomarkers, such as DAXX and ATRX genes, FGF13 mRNA, and miRNA-21 and LDH (15, 25–33). These markers could be potentially valuable for clinical implications. However, some of the findings (Cdk4 and phospho-Rb1 expression) were not associated with recur-


E792

Liu et al



Figure 3. The crucial CpG sites in regulating ␣-internexin expression. A, ␣-Internexin expression in human cancer cell lines identified by Western blot. Shown is expression of the protein in 4 of 14 cancer cell lines. The INS-1 cell line was used as positive control, and ␤actin used as internal control. B, RT-PCR confirming the ␣-internexin expression in some of the 14 cancer cell lines. Shown is expression of mRNA in the PA-1 cell line but not in MGC803 or A875 cell lines. ␤-Actin was used as internal control. C, Methylation status of 25 CpG sites in a crucial region was significantly correlated with protein or mRNA expression. Shown in upper black frame, the ⫺107 to ⫹96-nt region of the ␣-internexin promoter was demethylated in 5 human cancer cell lines. Four of them (M17, PA-1, SH-SY5Y, and SK-NSH) expressed ␣-internexin protein. In contrast, shown in the lower frame, the same region was hypermethylated in 9 of 9 cell lines without ␣-internexin expression (A875, A375, U251, MGC803, CL187, HCT15, RKO, SW620, and T84). Open circles represent unmethylated CpG sites; filled circles represent methylated CpG sites.

rence or survival (29, 30). Several prognostic biomarkers were mainly evaluated in NF-PNETs (15) or insulinomas (4, 22). Our present study demonstrated that ␣-internexin is extensively expressed in PNETs. PNETs are generally classified as NF-PNETs and F-PNETs, and as a unique type of PNET, insulinomas are quite different from the others, in view of their favorable prognosis (1– 6). Thus, we stratified the tumors into 4 categories, NF-PNETs, F-PNETs, noninsulinomas, and insulinomas. We found differential ␣-internexin expression in the subtypes of PNETs. The lost/reduced expression of ␣-internexin was significantly associated with metastasis and advanced stage not only in PNETs as a whole but also in some subtypes of PNET, suggesting that ␣-internexin could be a novel prognostic biomarker, particularly for overall survival. First, lost/reduced expression of ␣-internexin was significantly associated with recurrence and advanced stages in patients with PNETs. Second, the expression of ␣-internexin was significantly associated with overall survival in all of the PNET patients. Third, lost/reduced expression of ␣-internexin was associated with worse survival in patients with noninsulinomas and NF-PNETs. These findings demonstrate that the prognostic value of ␣-internexin might be subtype-dependent. Halfdanarson et al (6) and Missiaglia et al (25), among others, suggested that studies on PNETs should separate insulinomas from other PNETs with more aggressive behaviors. In the present study, we have validated the prognostic value of ␣-internexin in PNETs excluding insulinomas. In contrast, the ␣-internexin expression was not associated with disease-free survival in PNET patients.Recently,2studiesshowedthat expression of ␣-internexin was correlated with better overall survival in oligodendroglial tumors (32, 34). Their


doi: 10.1210/jc.2013-2874

Figure 4. Quantitatively detecting methylation of ␣-internexin by DHPLC. Shown is a high methylated peak of positive control MGC803 (black) and no methylated peak in negative control PA1 (green). Tumor samples 11T and 7T and the paired tissue sample 11N had relatively high methylated peak. Tumor samples 5T and 2T had a lower methylated peak; in contrast, their paired tissue samples 5N and 2N had relatively higher methylated peaks.

findings are consistent with our observations that ␣-internexin expression is a favorable prognostic marker. ␣-Internexin was expressed in human well-differentiated appendiceal and rectal carcinoid, but its clinical implications are unclear (35). If a patient harbors a tumor without or with reduced expression of ␣-internexin, the patient is likely to have an unfavorable prognosis and should be followed more stringently with more frequent imaging examinations and testing of serum levels of chromogranin A or gut peptides to screen for recurrence or metastasis. More aggressive antitumor treatment might be started earlier in this patient. In this study, patient 219 whose NF-PNET had reduced expression of ␣-internexin underwent resection of tumor. During the 6 years of follow-up, 2 small metastases were found in her liver. Another case, patient 273, whose glucagonoma exhibited complete loss of ␣-internexin expression underwent resection of the tumor in 2004, and 7 years later, the disease progressed with more than 20 metastases in the liver. These representative cases illustrate that assessing the expression of


E793

␣-internexin could be useful to individualize patient management. It was previously unclear whether ␣-internexin is epigenetically regulated and which region is crucial for gene expression. Usually, the methylation status of the proximal region within the gene promoter is correlated with gene silencing (36, 37); for example, our previous study confirmed this was true for MLH1 gene silencing in sporadic insulinomas (22). We find the promoter region (⫺107 to ⫹96) including 25 CpG sites is crucial for regulating ␣-internexin expression, and the hypomethylation of this region is significantly associated with gene expression in PNETs fresh tissues, suggesting the status of promoter methylation could play an important role in regulating ␣-internexin gene expression. The cytoskeleton proteins are involved in multiple cellular functions (38). They play an important role in cell signaling, proliferation, and differentiation, and their abnormalities are associated with more than 30 diseases (17, 38, 39). ␣-Internexin, an intermediate filament protein, is a component of the cytoskeleton. Interestingly, the expression of cytoskeleton proteins CK18 and CK19 correlated with aggressive behaviors in insulinomas (4), and intermediate filament proteins were involved in the development of pancreatic islet and interacted with menin, a crucial protein in PNET tumorigenesis (40, 41). Protein actinin-4, which regulated cell motility and tumor invasion through remodeling of the cytoskeleton (42), was associated with prognosis of pulmonary neuroendocrine tumors (43). Based on the above evidence, it is conceivable that reduced expression of ␣-internexin, a component of the cytoskeleton, might result in dedifferentiation in subgroups of PNETs and possibly promote tumor transformation and progression, contributing to the unfavorable prognosis in a subset of PNETs, although its specific underlying mechanism remains to be further investigated. Recently, in the series of articles concerning prognosis research strategy (44 – 46), the authors set out a framework of prognostic research, discussed the role of prognostic factors, and recommended a prospective study. One of our limitations is that the present study is retrospective, and potential variability in different hospitals might be present. Another limitation is the small number of patients who died of insulinomas (n ⫽ 6). In conclusion, we have provided solid evidence for expression of ␣-internexin protein in subtypes of PNETs as well as in PNETs as a whole, and the epigenetic regulation of the ␣-internexin gene for the first time. The strength of our study is that we studied one of the largest number of PNETs ever reported in a single study from China and the United States, the largest country in the east and the west, respectively, which allows us to obtain more reliable find-


E794

Liu et al


ings in these rare tumors. Loss of or reduced expression of ␣-internexin protein could be used as a potential prognostic marker in PNET patients, especially for overall survival in patients with noninsulinomas.

Acknowledgments We appreciate Tao Xu (Department of Epidemiology and Statistics, Institute of Basic Medical Sciences, Peking Union Medical College, Chinese Academy of Medical Sciences and School of Basic Medicine Sciences) for his help with the statistics. Address all correspondence and requests for reprints to: Dr Yuan-Jia Chen, Department of Gastroenterology, Peking Union Medical College Hospital, Peking Union Medical College, Chinese Academy of Medical Sciences, Beijing 100730, China. E-mail: [email protected]. This research was supported by Grants 30850002 and 81072054 from the National Natural Sciences Foundation of China (to Y.-J.C.), and also by the Mushett Family Foundation (to L.H.T.). Part of the study was orally presented at the Sixth Annual Conference of the European Neuroendocrine Tumor Society in 2009. Disclosure Summary: All authors declared no conflicts of interest.

References 1. de Wilde RF, Edil BH, Hruban RH, Maitra A. Well-differentiated pancreatic neuroendocrine tumors: from genetics to therapy. Nat Rev Gastroenterol Hepatol. 2012;9:199 –208. 2. Metz DC, Jensen RT. Gastrointestinal neuroendocrine tumors: pancreatic endocrine tumors. Gastroenterology. 2008;135:1469 – 1492. 3. Modlin IM, Oberg K, Chung DC, et al. Gastroenteropancreatic neuroendocrine tumours. Lancet Oncol. 2008;9:61–72. 4. Jonkers YM, Ramaekers FC, Speel EJ. Molecular alterations during insulinoma tumorigenesis. Biochim Biophys Acta. 2007;1775:313– 332. 5. Oberg K, Eriksson B. Endocrine tumours of the pancreas. Best Pract Res Clin Gastroenterol. 2005;19:753–781. 6. Halfdanarson TR, Rubin J, Farnell MB, Grant CS, Petersen GM. Pancreatic endocrine neoplasms: epidemiology and prognosis of pancreatic endocrine tumors. Endocr Relat Cancer. 2008;15:409 – 427. 7. Zikusoka MN, Kidd M, Eick G, Latich I, Modlin IM. The molecular genetics of gastroenteropancreatic neuroendocrine tumors. Cancer. 2005;104:2292–2309. 8. Rindi G, Wiedenmann B. Neuroendocrine neoplasms of the gut and pancreas: new insights. Nat Rev Endocrinol. 2012;8:54 – 64. 9. Yao JC, Hassan M, Phan A, et al. One hundred years after “carcinoid”: epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol. 2008;26: 3063–3072. 10. Cupisti K, Höppner W, Dotzenrath C, et al. Lack of MEN1 gene mutations in 27 sporadic insulinomas. Eur J Clin Invest. 2000;30: 325–329. 11. Capurso G, Festa S, Valente R, et al. Molecular pathology and genetics of pancreatic endocrine tumours. J Mol Endocrinol. 2012; 49:R37–R50.


12. Thakker RV. Multiple endocrine neoplasia type 1 (MEN1). Best Pract Res Clin Endocrinol Metab. 2010;24:355–370. 13. Wang EH, Ebrahimi SA, Wu AY, Kashefi C, Passaro E Jr, Sawicki MP. Mutation of the MENIN gene in sporadic pancreatic endocrine tumors. Cancer Res. 1998;58:4417– 4420. 14. Marx S, Spiegel AM, Skarulis MC, Doppman JL, Collins FS, Liotta LA. Multiple endocrine neoplasia type 1: clinical and genetic topics. Ann Intern Med. 1998;129:484 – 494. 15. Jiao Y, Shi C, Edil BH, et al. DAXX/ATRX, MEN1, and mTOR pathway genes are frequently altered in pancreatic neuroendocrine tumors. Science. 2011;331:1199 –1203. 16. Gumbs AA, Moore PS, Falconi M, et al. Review of the clinical, histological, and molecular aspects of pancreatic endocrine neoplasms. J Surg Oncol. 2002;81:45–53; discussion 54. 17. Omary MB, Coulombe PA, McLean WH. Intermediate filament proteins and their associated diseases. N Engl J Med. 2004;351: 2087–2100. 18. Rauch U, Klotz M, Maas-Omlor S, et al. Expression of intermediate filament proteins and neuronal markers in the human fetal gut. J Histochem Cytochem. 2006;54:39 – 46. 19. Foley J, Witte D, Chiu FC, Parysek LM. Expression of the neural intermediate filament proteins peripherin and neurofilament-66/␣internexin in neuroblastoma. Lab Invest. 1994;71:193–199. 20. Näthke I. Cytoskeleton out of the cupboard: colon cancer and cytoskeletal changes induced by loss of APC. Nat Rev Cancer. 2006; 6:967–974. 21. Chen YJ, Vortmeyer A, Zhuang Z, Huang S, Jensen RT. Loss of heterozygosity of chromosome 1q in gastrinomas: occurrence and prognostic significance. Cancer Res. 2003;63:817– 823. 22. Mei M, Deng D, Liu TH, et al. Clinical implications of microsatellite instability and MLH1 gene inactivation in sporadic insulinomas. J Clin Endocrinol Metab. 2009;94:3448 –3457. 23. Rindi G, Falconi M, Klersy C, et al. TNM staging of neoplasms of the endocrine pancreas: results from a large international cohort study. J Natl Cancer Inst. 2012;104:764 –777. 24. Deng D, Deng G, Smith MF, et al. Simultaneous detection of CpG methylation and single nucleotide polymorphism by denaturing high performance liquid chromatography. Nucleic Acids Res. 2002; 30:E13. 25. Missiaglia E, Dalai I, Barbi S, et al. Pancreatic endocrine tumors: expression profiling evidences a role for AKT-mTOR pathway. J Clin Oncol. 2010;28:245–255. 26. Nagano Y, Kim do H, Zhang L, et al. Allelic alterations in pancreatic endocrine tumors identified by genome-wide single nucleotide polymorphism analysis. Endocr Relat Cancer. 2007;14:483– 492. 27. Roldo C, Missiaglia E, Hagan JP, et al. MicroRNA expression abnormalities in pancreatic endocrine and acinar tumors are associated with distinctive pathologic features and clinical behavior. J Clin Oncol. 2006;24:4677– 4684. 28. Speisky D, Duces A, Bièche I, et al. Molecular profiling of pancreatic neuroendocrine tumors in sporadic and Von Hippel-Lindau patients. Clin Cancer Res. 2012;18:2838 –2849. 29. Tang LH, Contractor T, Clausen R, et al. Attenuation of the retinoblastoma pathway in pancreatic neuroendocrine tumors due to increased cdk4/cdk6. Clin Cancer Res. 2012;18:4612– 4620. 30. Lee J, Sung CO, Lee EJ, et al. Metastasis of neuroendocrine tumors are characterized by increased cell proliferation and reduced expression of the ATM gene. PLoS One. 2012;7:e34456. 31. Corbo V, Beghelli S, Bersani S, et al. Pancreatic endocrine tumours: mutational and immunohistochemical survey of protein kinases reveals alterations in targetable kinases in cancer cell lines and rare primaries. Ann Oncol. 2012;23:127–134. 32. Ducray F, Crinière E, Idbaih A, et al. ␣-Internexin expression identifies 1p19q codeleted gliomas. Neurology. 2009;72:156 –161. 33. Sorbye H, Welin S, Langer SW, et al. Predictive and prognostic factors for treatment and survival in 305 patients with advanced gastrointestinal neuroendocrine carcinoma (WHO G3): the NORDIC NEC study. Ann Oncol. 2013;24:152–160.


doi: 10.1210/jc.2013-2874


34. Mokhtari K, Ducray F, Kros JM, et al. ␣-Internexin expression predicts outcome in anaplastic oligodendroglial tumors and may positively impact the efficacy of chemotherapy: European organization for research and treatment of cancer trial 26951. Cancer. 2011;117: 3014 –3026. 35. Ishida M, Kushima R, Brevet M, Chatelain D, Okabe H. Co-expression of neuronal intermediate filaments, peripherin and alphainternexin in human well-differentiated endocrine neoplasms (carcinoid tumors) of the appendix. Mol Med Report. 2008;1:191–195. 36. Ushijima T. Detection and interpretation of altered methylation patterns in cancer cells. Nat Rev Cancer. 2005;5:223–231. 37. Deng G, Chen A, Hong J, Chae HS, Kim YS. Methylation of CpG in a small region of the hMLH1 promoter invariably correlates with the absence of gene expression. Cancer Res. 1999;59:2029 –2033. 38. Kim S, Coulombe PA. Emerging role for the cytoskeleton as an organizer and regulator of translation. Nat Rev Mol Cell Biol. 2010; 11:75– 81. 39. Herrmann H, Bär H, Kreplak L, Strelkov SV, Aebi U. Intermediate filaments: from cell architecture to nanomechanics. Nat Rev Mol Cell Biol. 2007;8:562–573. 40. Di Bella A, Regoli M, Nicoletti C, Ermini L, Fonzi L, Bertelli E. An appraisal of intermediate filament expression in adult and develop-

41.

42.

43.

44.

45.

46.

E795

ing pancreas: vimentin is expressed in ␣-cells of rat and mouse embryos. J Histochem Cytochem. 2009;57:577–586. Lopez-Egido J, Cunningham J, Berg M, Oberg K, Bongcam-Rudloff E, Gobl A. Menin’s interaction with glial fibrillary acidic protein and vimentin suggests a role for the intermediate filament network in regulating menin activity. Exp Cell Res. 2002;278:175–183. Honda K, Yamada T, Endo R, et al. Actinin-4, a novel actin-bundling protein associated with cell motility and cancer invasion. J Cell Biol. 1998;140:1383–1393. Miyanaga A, Honda K, Tsuta K, et al. Diagnostic and prognostic significance of the alternatively spliced ACTN4 variant in highgrade neuroendocrine pulmonary tumours. Ann Oncol. 2013;24: 84 –90. Hemingway H, Croft P, Perel P, et al. Prognosis Research Strategy (PROGRESS) 1: a framework for researching clinical outcomes. BMJ. 2013;346:e5595. Riley RD, Hayden JA, Steyerberg EW, et al. Prognosis Research Strategy (PROGRESS) 2: prognostic factor research. PLoS Med. 2013;10:e1001380. Steyerberg EW, Moons KG, van der Windt DA, et al. Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10:e1001381.

All members have access to The Endocrine Legacy – an online journal archive of all articles from 2011 back to Volume 1, Issue 1. www.endocrine.org/legacy


CUX1: a modulator of tumour aggressiveness in pancreatic neuroendocrine neoplasms.

A woman with metastatic pancreatic neuroendocrine tumor.

Chromogranin A is a reliable serum diagnostic biomarker for pancreatic neuroendocrine tumors but not for insulinomas.

AROS Is a Significant Biomarker for Tumor Aggressiveness in Non-cirrhotic Hepatocellular Carcinoma.

Pancreatic Neuroendocrine Tumor with Atypical Radiologic Presentation.

Hypoxia induced tumor metabolic switch contributes to pancreatic cancer aggressiveness.

Pancreatic body adenocarcinoma with neuroendocrine tumor characteristics: A case report.

Incidental Finding of Cystic Pancreatic Schwannoma Mimicking a Neuroendocrine Tumor.

Nonfunctional pancreatic neuroendocrine tumor masked as anemia: A case report.

Pancreatic neuroendocrine tumor in a child of 3.5 years old.

Tumor-associated macrophages are a useful biomarker to predict recurrence after surgical resection of nonfunctional pancreatic neuroendocrine tumors.

Vascular endothelial growth factor, a novel and highly accurate pancreatic fluid biomarker for serous pancreatic cysts.

Loss of stromal caveolin-1 expression: a novel tumor microenvironment biomarker that can predict poor clinical outcomes for pancreatic cancer.

TRIM29 as a novel biomarker in pancreatic adenocarcinoma.

Ectopic spleen and liver hemangioma mimicking metastatic pancreatic neuroendocrine tumor.

Hypercalcemia from metastatic pancreatic neuroendocrine tumor secreting 1,25-dihydroxyvitamin D.

Late onset asymptomatic pancreatic neuroendocrine tumor - A case report on the phenotypic expansion for MEN1.

Pancreatic neuroendocrine tumor with splenic vein tumor thrombus: A case report.

[Primary tumor resection of neuroendocrine pancreatic tumors with liver metastases].

Metastatic pancreatic neuroendocrine tumor with ectopic adrenocorticotropic hormone production.

24-Hydroxycholesterol participates in pancreatic neuroendocrine tumor development.

NKX6-1 Is a Novel Immunohistochemical Marker for Pancreatic and Duodenal Neuroendocrine Tumors.

FoxM1: a novel tumor biomarker of lung cancer.

Therapy for metastatic pancreatic neuroendocrine tumors.