Background
Small intestinal neuroendocrine tumor (SI-NET) arising from enterochromaffin cells is the most common type of gastrointestinal endocrine tumor. SI-NET is also termed midgut carcinoids, well-differentiated neuroendocrine tumor of the midgut, ileal carcinoid or neuroendocrine tumor of the midgut [
1,
2]. The tumors are usually slow-growing in nature [
3,
4]. SI-NETs are mostly sporadic, however, a few small families with a history of the disease have recently been characterized [
5,
6]. Many patients are asymptomatic and the disease can be indolent for many years and diagnosed incidentally. Approximately 20% of patients present with carcinoid syndrome, a clinical entity characterized by flushing, diarrhea, abdominal pain, cardiac valvular fibrosis and bronchial constriction [
7,
8]. The clinical picture of the carcinoid syndrome, which usually occur in patients with liver metastases, is due to excessive production and release of hormones and substances such as serotonin, tachykinins and prostaglandins [
9]. Metastases are first recognized in regional lymph nodes followed by the liver and less frequently in the ovaries or other distant sites [
3,
10]. Surgery is the first treatment of choice, however, in patients with metastatic disease surgical treatment seldom leads to cure [
11].
Several molecular cytogenetic studies have been performed with the aim of understanding the mechanisms of SI-NET development. Using conventional comparative genomic hybridization (CGH) we and others observed frequent copy number (CN) losses at 18q, 11q, 16q, and gains of 4p [
12,
13]. The most common aberration, loss of 18, was also identified by loss of heterozygosity (LOH) screening [
14] and has been validated in SI-NET by several other groups using array CGH (a-CGH) or single nucleotide polymorphism based arrays (SNP-arrays) [
5,
15‐
17]. These findings suggest the location of putative tumor suppressor gene(s) for SI-NET development in chromosome 18. In addition, the possibility of two or more genetically distinct groups in SI-NET has been suggested, namely those characterized by loss of chromosome 18 and a second group with gain of chromosomes 4, 5, 7 and 14 [
15,
16]. SI-NETs with gain of chromosome 14 were reported to have intact chromosome 18 and poor prognosis [
15]. However, so far no target gene or genes have been identified for the frequent genetic aberrations in SI-NETs.
To further refine and define regions of recurrent CN aberrations (CNAs) we applied a-CGH and genomic quantitative real-time PCR (qPCR) to a panel of SI-NETs from 32 patients and evaluated the findings to the clinical parameters and patient outcome. We also aimed to identify differences in CNA profiles between primary tumors and metastases.
Discussion
In this study we report frequent CN losses of chromosome 18 as well as of chromosomes 16q, 11q, 13 and 9 and gains of chromosomes 20, 14, 4, 5 and 7. These findings are in line with another Swedish study using a similar platform to identify CNAs in SI-NETs [
15], and using a parallel exome sequencing approach to find regions of recurrent losses and gains and mutated candidate genes therein [
26]. For some recurrent regions CNAs were more frequently observed in metastases than in primary tumors such as loss of 16q (33% vs. 6%) and gain of 7q (25% vs. 6%) (Table
4). The observations of more genetic aberrations in metastases compared to primary tumors indicate that genetic changes accumulate during tumor progression.
Table 4
Comparison of CNAs by a-CGH in recurrent regions in primary tumors and metastases
No of samples
| n = 18 | n = 12 | n = 30 |
Losses
| | | |
9pter-p13.2 | 2 / 18 (11%) | 1 / 12 (8%) | 3 / 30 (10%) |
9p13.1-11.2 | 3 / 18 (17%) | 2 / 12 (17%) | 5 / 30 17%) |
11q23.1-qter | 4 / 18 (22%) | 3 / 12 (25%) | 7 / 30 (23%) |
13q14.11-21.32 | 2 / 18 (11%) | 1 / 12 (8%) | 3 / 30 (10%) |
13q33.1 | 2 / 18 (11%) | 1 / 12 (8%) | 3 / 30 (10%) |
16q12.2-qter | 1 / 18 (6%) | 4 / 12 (33%) | 5 / 30 (17%) |
18pter-11.21 | 12 / 18 (67%) | 8 / 12 (67%) | 20 / 30 (67%) |
18q12.1-21.31 | 10 / 18 (56%) | 8 / 12 (67%) | 18 / 30 (60%) |
Gains
| | | |
4pter-qter | 4 / 18 (22%) | 4 / 12 (%) | 8 / 30 (27%) |
5pter-qter | 4 / 18 (22%) | 3 / 12 (25%) | 7 / 30 (23%) |
7p22.3 | 1 / 18 (6%) | 2 / 12 (17%) | 3 / 30 (10%) |
7p22.2-22.1 | 1 / 18 (6%) | 2 / 12 (17%) | 3 / 30 (10%) |
7q22.1 | 1 / 18 (6%) | 2 / 12 (17%) | 3 / 30 (10%) |
7q22.3-qter | 1 / 18 (6%) | 3 / 12 (25%) | 4 / 30 (13%) |
14q11.2 | 4 / 18 (22%) | 5 / 12 (42%) | 9 / 30 (30%) |
14q32.2-32.31 | 3 / 18 (17%) | 3 / 12 (25%) | 6 / 30 (20%) |
20pter-p11.21 | 4 / 18 (22%) | 4 / 12 (33%) | 8 / 30 (27%) |
20q11.1-11.21 | 5 / 18 (28%) | 2 / 12 (17%) | 7 / 30 (23%) |
20q12-qter | 6 / 18 (33%) | 5 / 12 (42%) | 11 / 30 (37%) |
Loss of chromosome 18 was the most prominent alteration observed in 70% of all tumors. This result is in agreement with previous studies and confirms the high frequency of chromosome 18 loss in SI-NETs [
12,
14,
15]. Loss of chromosome 18 was the only detectable recurrent CNA in four tumors of which three were primary, suggesting a potential role in tumor initiation. We identified two recurrent regions of losses of 18p and 18q and a 247 kb MOR of loss at the
EMILIN2 locus within the pter-11.21 interval which was verified by qRT-PCR.
EMILIN2, which is a component of extracellular matrix, suppresses the growth of cancer cells and has a role in cell survival and apoptosis [
27]. In addition, methylation of
EMILIN2 is associated with poor outcome in breast cancer [
28]. Another MOR of 2 Mb loss was observed at 18q22.1 in one tumor which overlapped with losses in 17 other tumors. Losses involving18q22.1 have been reported by several groups, and a 40 kb CN variation in 18q22.1 has been reported to be over-represented in SI-NET patients [
17]. Furthermore, this region corresponds to one of the three MOR of deletions (R2) described in familial SI-NET [
5]. This region encompasses the
CDH19 and
CDH7 genes, however, we could not confirm the loss of
CDH19 in tumor 16 with the small deletion in 18q22.1. This could be due to mixed tumor populations giving variable results for this particular tumor or represent an artifact finding. Mutation analyses of genes in the 18q22.1 region including
CDH7 and
CDH19 did not reveal any tumor specific mutations [
5], suggesting that other mechanisms of gene inactivation are operational.
Nine of the 30 tumors (30%) had intact chromosome 18 (Table
1). Three of these (20, 23, 27M) had only recurrent gains of chromosomes 14, two tumors (8, 29) were intact even for other recurrent alterations and the other tumors had gains of 14 together with other alterations such as gains of chromosomes 4 and 5 or losses involving chromosomes 3, 9, 11, 13 and 16. This finding is in agreement with previous reports and supports the suggestion of distinct genetic alterations and pathways in SI-NETs [
15,
16]. However, a mutually excluding relationship between loss of 18 and gain of 14 observed in another cohort [
15], was not revealed in our study. Furthermore we did not find an association between gain of 14 and poor survival, as reported in another cohort [
15].
Four of 12 (33%) metastases had loss at 16q as compared to 1/18 (6%) primary tumors confirming our previous observations by conventional CGH [
12], and suggesting a role in SI-NET progression. By contrast, another study reported that losses on 16q are more frequent in primary tumors than in metastases [
15]. This discrepancy could be due to the different type of metastases used in different studies. A possible role of chromosome 16 loss in tumor progression is supported by studies of other tumor types, e.g. advanced prostate cancer and relapses of Wilm’s tumor showing frequent LOH in 16q [
29]. Using qPCR we confirmed CNAs at the
CDH1 gene locus, located in the MOR of 561 kb at 16q22.1, in 60% of cases.
CDH1 is inactivated by mutation or promoter hypermethylation in e.g. gastric and breast cancers [
30,
31]. Inactivation of
CDH1 is associated with its dysfunction in cell-cell adhesion as well as triggering of cancer invasion and metastasis. We have also observed promoter hypermethylation of
CDH1 in SI-NETs (Fotouhi et al., unpublished data). Thus,
CDH1 could represent a potential candidate tumor suppressor gene in this region.
CN losses within 11q have been reported for many cancer types and have been associated with metastatic disease for example in pheochromocytoma [
32]. Loss of 11q is also linked to a high risk of relapse in neuroblastoma and poor clinical outcome in oral cancer [
33,
34]. In the present study, losses of 11q were frequently observed at almost similar frequencies in primary tumors (22%) and metastasis (25%).
In addition to gain of entire chromosome 20, a 1.4-4.2 Mb region at 20q13.33 was gained in about 33% of tumors. Similar alterations have been reported in e.g. digestive tract tumors and breast cancer [
35,
36], and have been correlated with lymph node metastasis in gastric cancer [
37]. Furthermore, the 20q13 region has been reported as the most commonly amplified region in cancer and 13 of the amplified genes were proposed as “cancer initiating genes” [
38]. Several cancer-related genes are located on 20q13.
TNFRSF6B, that may inhibit apoptosis and promote cell survival, is over-expressed in gastrointestinal tract tumors [
39], colorectal carcinoma [
40] and gastric cancer [
41]. Furthermore cases with gain of 20pter-p11.21 exhibited shorter overall survival, supporting a role of this alteration in aggressive SI-NET. However, since only 7 cases showed gain of this region, the association is based on a limited number of cases and firm conclusions would require analysis of additional cases.
Unsupervised clustering of all CNAs identified two distinct tumor groups (I and II) and 5 chromosomal clusters (a-e). Interestingly, gains of chromosomes 4, 5, 7 and 14 clustered together in tumor group II. This distinct genetic alteration in SI-NETs is in accordance with a previous study [
16]. Four metastases including two distant metastases (21, 27) which harbored losses of 16q, clustered together with loss of chromosome 11 (cluster-c). We investigated if different tumor groups and chromosomal clusters are associated with clinical variables or tumor types. As shown in Figure
3, clusters-d and b alterations (gains of 4, 5, 14 and 20) in tumor group II comprised a higher number of metastases (57%) than group I (32%). This finding suggests a possible role for gain of chromosomes 4, 5, 14 and 20 in progression of primary tumors to metastases in SI-NETs. In addition tumors with gain of 20pter-p11.21 were also associated with tumor group II (
P = 0.014) and the patients had a shorter overall survival (
P = 0.013). Female patients more frequently showed losses on chromosomes 11q and 16q and gains on 20p and 20q. Furthermore, gain on 7q and loss on chromosomes 9p and 18p were associated with younger age at diagnosis.
Competing interests
The authors declare that they have no competing interest.
Authors’ contributions
JH contributed to the study design, performed a-CGH experiments and data analysis, interpreted the results and drafted the manuscript. OF performed qPCR, clustering, statistical analysis and interpreted the results. LS performed a-CGH experiments. MK, AH, CL and JZ provided the clinical information. OF, LS, AH and JZ critically revised the manuscript. CL contributed to conception and study design, interpretation of data and substantially revised the manuscript. All authors read and approved the final manuscript.