We have used array CGH to analyze DNA copy number changes in a small panel of high-grade MPNSTs, in order to identify recurrent copy number alterations at high-resolution and thereby novel candidate oncogenes and/or tumor suppressor genes. After a recent pathological revision of our tumor collection, seven samples were classified as MPNSTs based on the current classification standard and included in the study. Although this is a small number of samples, the analyses revealed results of potential importance for this malignancy.
Considerably more gains than losses were observed in the seven MPNSTs. Only five of the 40 identified minimal recurrent regions of alteration were losses (see Figure and Table ), similar to previous observations by others [
7,
10,
13]. However, other studies have reported far more frequent losses than gains in MPNSTs as well [
4,
14]. Differences in the chromosomal regions altered in sporadic versus neurofibromatosis-associated MPNSTs have previously been demonstrated [
6,
9,
10,
13], but the only difference observed here was loss of 11q23.2-q23.3. This region was deleted in the three sporadic MPNSTs, but not in the four neurofibromatosis-associated MPNSTs. However, two of the neurofibromatosis-associated MPNSTs showed loss of other parts of chromosome 11, and the two other minimal recurrent regions of loss in chromosome 11, 11p13 and 11q22.3-q23.1, were deleted in both sporadic and neurofibromatosis-associated MPNSTs [see Additional file
2].
It has previously been reported that patients with neurofibromatosis-associated MPNSTs have a worse survival than patients with sporadic MPNSTs [
15,
16], but this has not been a consistent finding [
17,
18]. In this panel of tumors, the two longest surviving patients who showed no evidence of the MPNST tumor after 100 and 216 months, respectively, have neurofibromatosis (Table ).
The most frequent alterations observed in the seven MPNSTs were gains of regions in 1q, 8p, 9q and 17q, all present in five of the seven tumors analyzed. Within 1q, three minimal recurrent regions were identified. 1q24.1-q24.2 (3.7 Mb) and 1q24.3-q25.1 (3.5 Mb) were gained in five of seven samples, whereas 1q21.1 (1.2 Mb) was gained in three samples. Gain of regions in 1q has not frequently been reported in MPNSTs previously, but it is a recurrent finding in soft tissue sarcomas [
19,
20] and other malignancies. Since gene expression of six of the MPNSTs studied here has previously been analyzed using cDNA microarrays [
12], the expression level of genes located within these regions was investigated, but none of the genes present on the cDNA microarray were over-expressed compared to the median for soft tissue sarcomas (log
2 ratio > 1) in more than one sample. This was also the case for genes located within the minimal recurrent region in 9q34.11-q34.13 (1.0 Mb).
A minimal recurrent region in 8p23.1-p12 (22.9 Mb) was gained in five of the samples. Within this region, six genes showed increased expression in two or more of the samples [
12].
LOXL2 and
ZNF395 were over-expressed in three of the samples. The expression level of these two genes was in addition determined using quantitative real-time RT-PCR, and the expression level was compared to two benign tumors (Figure ).
LOXL2 showed more than 3-fold increased expression in three MPNSTs, whereas
ZNF395 showed approximately similar expression levels in the MPNSTs and the benign tumors. Increased expression of
LOXL2, a member of the lysyl oxidase family, has previously been shown in colon- and esophageal cancer [
21], and it has also been associated with breast cancer tumor grade [
22]. Thus,
LOXL2 may be a candidate target for the 8p23.1-p12 gain in MPNSTs.
Scattered high-level amplification (log
2 ratio > 1) and homozygous deletion (log
2 ratio < -1) were observed in some of the tumors [see Additional file
2]. Within 9p22.3-p21.2, homozygous deletion of the region harboring the gene
CDKN2A was observed in two samples, and heterozygous deletion in one sample [see Additional file
2]. Inactivation of this region is a frequent finding in MPNSTs [
23,
24], as well as other cancer types [
25]. The expression level of
CDKN2A was determined using quantitative real-time RT-PCR (Figure ), and the three samples with deletion of the region showed no or very low expression compared to the benign tumors, whereas the other three samples showed increased expression. Notably, the expression level of
CDKN2A was lower in the four patients with poor outcome than the two patients with disease-free survival (Figure ).
We observed that all five patients with poor outcome showed gain of the distal part of 17q, whereas the two patients with disease-free survival did not (Figure ). Several other studies have also reported that gain of 17q is frequent in MPNSTs [
6,
7,
9,
10,
13,
26], and this alteration has been associated with poor outcome [
10] and development of metastasis [
26]. These CGH studies have reported the region of gain to be from either 17q22 or 17q24 to 17q25/17qtel. The minimal recurrent region identified in this study, using microarrays, covered 17q23.2-q25.3 (qtel). This region was gained in five tumors, and four of these showed also gain of the proximal region (17q21.2-q23.2).
The expression level of genes located within the minimal recurrent region, as well as the proximal region where four of seven samples showed gain, was investigated. Ten genes showed increased expression relative to the median for soft tissue sarcomas in two or more of the six tumors analyzed [
12], although it should be noted that the cDNA microarray only contained probes for about half of the genes in the region (including novel and predicted genes). Three genes were over-expressed in three of the tumors;
ETV4,
HOXB7 and
BIRC5, and the expression levels of these genes, as well as
TOP2A, were in addition determined using quantitative real-time RT-PCR (Figure ). Of these four genes, only
BIRC5 is located within the minimal recurrent region of gain.
TOP2A,
ETV4 and
BIRC5 showed increased expression in all MPNST samples, in most cases more than 20-fold, whereas
HOXB7 showed approximately similar expression levels in the MPNSTs and the benign tumors. Even though only tumors from patients who died of the disease showed gain of this region, increased expression of these genes was also seen in tumors from the patients who showed no evidence of the disease. Thus, there were no correlations between the expression level of the genes and patient survival.
BIRC5, also known as survivin, is an inhibitor of apoptosis that has been shown to be highly expressed in the majority of cancers, including soft tissue sarcomas and osteosarcomas [
27,
28]. Increased expression of
BIRC5 has been associated with chemotherapy resistance, enhanced proliferation, increased tumor recurrence and shorter patient survival [
27].
BIRC5 is located within a 2 Mb region in 17q25 previously shown to be commonly amplified in MPNSTs [
26], and increased expression of
BIRC5 in MPNSTs compared to neurofibromas and benign schwannomas has been reported [
29,
30]. Hence, there is considerable evidence suggesting that
BIRC5 may be involved in MPNST tumorigenesis.
Increased expression of
TOP2A has been previously associated with poor cancer-specific survival in MPNSTs [
11], whereas increased expression of
ETV4, a member of the Ets family of transcription factors, has been associated with shorter patient survival in colorectal cancer [
31] and gastric cancer [
32]. Although these two genes were not located within the minimal recurrent region of gain, their associations with poor outcome in MPNSTs (for
TOP2A) and other malignancies suggest that increased expression of these genes may play a role in MPNST tumorigenesis as well.
Recently, it has been shown that expression of miRNAs can be deregulated in cancer, and that miRNAs may act as oncogenes and tumor suppressor genes [
33]. miRNAs are frequently found in regions of DNA copy number aberrations, and a general correlation between miRNA copy number and expression level has been reported [
34,
35]. In order to investigate if miRNAs may be candidate targets for the 17q gain, we analyzed the expression level of five miRNAs present in the minimal recurrent region (Figure ). Compared to the benign tumors, none of the miRNAs showed high expression levels, and no clear differences in the miRNA expression levels were seen between MPNSTs from patients with poor outcome and MPNSTs from patients who showed disease-free survival. Thus, no clear correlation could be found between DNA copy number and miRNA expression in 17q in our samples.
1,4
This article has been 
