DNA Sequence Copy Number Changes in Gastrointestinal Stromal Tumors: Tumor Progression and Prognostic Significance

INTRODUCTION

GISTs 3 , classified previously as smooth muscle tumors, constitute the most important group of primary mesenchymal tumors of the gastrointestinal tract. These tumors occur at all levels of the gastrointestinal tract and usually present between the sixth and eighth decades. Immunohistochemically, GISTs are usually positive for KIT (CD117) and CD34, variably positive for smooth muscle actin, and usually negative for desmin, in contrast to true smooth muscle tumors. GISTs are negative for S100 protein, in contrast to schwannomas (1, 2) .

Clinically and pathologically, GISTs represent a spectrum of tumors including benign and malignant variants, the latter of which are generally identified by mitotic activity (>1 mitosis/10 high-power field). In some cases, the prediction of biological potential is difficult, because larger tumors with lower mitotic activity may also occasionally metastasize.

CGH enables screening of entire tumor genomes for gains and losses of DNA copy number and consequent mapping of aberrations to chromosomal subregions (reviewed in Refs. 3 and 4 ). Recently, we reported using CGH that loss of genetic material at chromosome arm 14q is the most frequently occurring aberration in both benign and malignant primary GISTs (5) . The DNA copy number changes seen in GISTs were not detected in leiomyomas and leiomyosarcomas (6) . Therefore, the immunophenotypic characteristics and the genetic profile of GISTs have clearly placed it as a separate tumor entity different from other mesenchymal tumors of the gastrointestinal tract. However, the prognostic evaluation of GISTs remained a difficult issue, requiring a complex multiparametric approach.

This study was performed to analyze the CGH changes in benign and malignant primary GISTs and to investigate whether progressive DNA copy number changes occur in recurrent and metastatic GISTs. We also investigated whether any DNA copy number change has prognostic significance for GISTs.

MATERIALS AND METHODS

RESULTS

Changes in DNA copy numbers were detected in all samples. DNA copy number changes were seen in all chromosomes in malignant primary and metastatic GISTs, whereas fewer chromosomes were involved in benign GISTs (Figs. 1 and 2 ⇓ ⇓ ). Benign GISTs had a mean of 2.6 aberrations/case (losses:gains, 5:1) and contained significantly fewer DNA copy number changes and fewer gains than malignant primary and metastatic GISTs (P < 0.01). High-level amplifications were not seen in benign GISTs. In malignant primary GISTs, the mean number of aberrations/case was 7.5 (losses:gains, 1.6:1), whereas the mean was 9 in metastatic GISTs (losses:gains, 1.8:1). The CGH profiles in metastatic GISTs and malignant primary tumors were similar, but metastatic GISTs showed a larger number of high-level amplifications. There was no correlation between the location and any specific DNA copy number changes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Summary of DNA copy number gains and losses detected by CGH in 60 GIST patients. Each bar represents one patient, regardless of the number of samples available, and the minimal common overlapping regions are presented. Gains are on the right hand side, losses on the left. Green broken lines, benign tumors; green continuous lines, malignant primary tumors that had no metastases samples. Blue lines, the aberration was seen in the malignant primary tumor and in at least one metastatic sample from the same patient. Red lines, changes that were seen on metastatic samples but not in their primary tumors for the same patient.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Comparative frequencies of losses (left) and gains (right) detected in 24 benign GISTs (green), 36 malignant primary GISTs (blue), and 35 metastatic GISTs (red). Chromosomal numbers are shown in the middle, and the minimal common overlapping regions, when applicable, are added to each bar.

Changes that were frequently detected in all GISTs with no statistical differences between benign, malignant primary, and metastatic tumors included losses in chromosome arms 1p (51%), 14q (74%), and 22q (53%).

Gains and high-level amplifications at 8q and 17q were significantly more often detected in metastatic GISTs (57 and 43%) than in benign GISTs (8 and 0%; P < 0.001) and malignant primary GISTs (33 and 25%; P < 0.05). Gains and high-level amplifications at 20q were only seen in malignant primary and metastatic GISTs (P < 0.01), and gains at 5p were not detected in benign GISTs (P < 0.01). Losses in 9p were never seen in benign tumors (P < 0.001), and they were more frequent in metastatic GISTs than in malignant primary tumors (63 and 36%; P < 0.05). The losses in 13q were less frequent in benign GISTs than in malignant primary (P < 0.05) and metastatic (P < 0.01) GISTs. In addition, several other changes were seen more frequently in malignant primary and metastatic GISTs than in benign GISTs. The details of DNA copy number changes are shown in Table 1 ⇓ and Fig. 1 ⇓ . Fig. 2 ⇓ shows the relative frequencies of the aberrations in all GISTs, and Table 2 ⇓ summarizes the significant DNA copy number changes.

DISCUSSION

On the basis of a large series of clinically benign and malignant GISTs and their metastases, we investigated the possible value of CGH to identify genetic markers of tumor behavior. The observation that benign tumors, almost exclusively, contain losses rather than gains, whereas malignant primary GISTs and their metastases contain gains and high-level amplifications, presents an apparently clear criterion for categorization of these tumors. Accordingly, losses are more likely related to the development of GISTs, whereas accumulation of additional genetic alterations, especially gains/amplicons, is required for malignant transformation and metastatic behavior in GISTs. Absence of gains correlated with benign GISTs with verified good prognosis, whereas presence of gains was more frequent in malignant primary GISTs and their metastases (P < 0.01). Therefore, absence of gains can be considered as a good prognostic parameter, which can be used as a new complementary diagnostic criterion for characterization of GISTs.

Gains and high-level amplifications at 5p, 8q, 17q, and 20q and losses in 9p, 13q, 15q, and 19q correlated with malignant primary GISTs and metastatic GISTs. Moreover, the gains at 8q, 17q, and 20q and the losses in 9p were more frequent in metastatic GISTs than in malignant primary GISTs (P < 0.05). Therefore, clones with this genetic make-up have a potential capability of developing metastases.

Losses in 1p, 14q, and 22q seen in benign GISTs seem to be unique because they have been reported rarely in other tumors at frequencies as high as observed in GISTs (9) . These changes were frequently seen in benign GISTs, indicating that they may be early events in GIST development. The fact that these losses were maintained with tumor progression suggests that they are required for the maintenance of GIST phenotype. The present extended study also supports the notion that these changes may be unique for GIST development. Recently, we reported high-resolution deletion mapping of chromosome 14 in GISTs and identified two possible tumor suppressor loci in 14q11.2 and 14q23 (10) .

The gains/high-level amplifications at 5p, 8q, 17q, and 20q as well as the losses in 9p, 13q, 15q, and 19q were detected in many malignant GISTs and their metastases. Such changes have also been reported in several other tumors, such as carcinomas of lung, breast, and ovary, and in squamous cell carcinomas of the head with variable frequencies (reviewed in Refs. 3 and 4 ). Gains and high-level amplifications at 8q22–q24 and 17q21–qter have been implicated as poor prognostic indicators in breast cancer, ovarian cancer, osteosarcomas, and neuroblastomas (11–14) . The gains and high-level amplifications at 8q and 20q correlate with invasiveness in breast cancer (15, 16) .

Although these chromosomal regions are known to contain several oncogenes, such as CMYC at 8q and ERBB2 at 17q, the possibility of yet undiscovered oncogenes cannot be ruled out (14) . At 20q, several genes have been implicated in breast cancer, and the region is known to harbor specific amplified genes (AIB1, AIB3, and AIB4; Ref. 17 ). Several candidate genes are located at 20q, e.g., the PTP1B/PTPN1 gene (20q12), which is involved in growth regulation, and the MYBL2 gene (20q13), which plays an important role in cell cycle progression. Moreover, the human cellular apoptosis susceptibility (CAS) gene has been mapped to this same region (18) . There are no known oncogenes that map to 5p.

Losses in 9p were associated with recurrences and unfavorable outcome in squamous cell carcinoma of the head and neck (19) and in astrocytic tumors (20) . Losses in 13q12–q13 are associated with poor prognosis in familial and sporadic breast cancer (21) . Losses in 15q have been observed rarely upon CGH. LOH studies have suggested the presence of a putative tumor suppressor gene in 15q in small cell lung cancer (22) . However, there are no data related to its potential prognostic significance or identified tumor suppressor genes.

In summary, our results show that genetic changes can be used as a complementary diagnostic tool for the prognostication of GISTs and for the differentiation between benign and malignant tumors. The increased number of changes and/or increased number of gains correlate with malignant behavior. Furthermore, the genetic changes seen as gains at 5p, 8q, 17q, and 20q and losses in 9p, 13q, 15q, and 19q are new prognostic parameters for GISTs.