INTRODUCTION

For patients with soft tissue sarcoma (STS) in the extremities or trunk wall, survival is dependent largely on whether metastases develop. Distant spread, most often to the lungs, is seen in more than one third of all STS patients, and usually becomes clinically manifest within 2 years after the diagnosis of the primary lesion. Attempts to identify features that may predict metastases have been hampered by the inherent heterogeneity of STS, as well as by the relative rarity of these tumors. Thus, currently used prognostication systems rely on relatively crude clinical parameters, such as tumor size and depth, and histopathologic parameters such as morphologic subtype, grade, necrosis, and vascular invasion, all of which suffer from interobserver variability.^1-6 Lately, the potential value of various tumor characteristics at the DNA and protein levels have been studied, but, as discussed recently by Oliveira and Fletcher,⁷ most of the reported claims for novel prognostic markers either have been disputed by discordant results or have not been substantiated in independent patient series.

We have studied previously the prognostic significance of cytogenetic findings in a series of 467 benign and malignant soft tissue tumors, including 151 high-grade (grades 2 and 3 in a three-grade scale) sarcomas.⁸ In that study, clonal chromosome aberrations were categorized according to general karyotypic features (eg, degree of complexity and ploidy level) and rearrangements of specific chromosomal regions. These cytogenetic variables then were analyzed regarding time to metastasis, and selected variables were compared with established clinicopathologic predictors of metastasis development. Among the patients with high-grade sarcoma, 17 variables, in addition to grade and size, that were associated with increased risk of metastases were identified, and five of these emerged as independent cytogenetic predictors of adverse outcome: breakpoints in chromosome regions 1p1, 1q4, 14q1, and 17q2, and gain of regions 6p1/p2. Furthermore, an increasing effect on metastatic risk was seen with increasing number of the selected cytogenetic variables, even when different histopathologic types were studied separately.⁸

The present investigation was undertaken in order to investigate whether the previously observed clinicocytogenetic correlations could be reproduced in an independent set of high-grade STSs. For this purpose, we retrieved the clinicopathologic and cytogenetic data on 156 high-grade STSs located in the extremities or trunk wall, and studied the effect on metastasis risk for those cytogenetic variables that had been selected from our previously published series.

PATIENTS AND METHODS

RESULTS

DISCUSSION

In a previous analysis of 151 high-grade STSs with clonal chromosome aberrations, we investigated a large number (n = 268) of cytogenetic variables with regard to possible association with metastases.⁸ In the final analysis, when the relative impact of clinicopathologic and selected cytogenetic variables were compared with one another, malignancy grade, tumor size and depth, and five of the cytogenetic variables—breakpoints in chromosome regions 1p1, 1q4, 14q1, and 17q2 and gain of the short arm of chromosome 6—turned out to be associated independently with metastases. By necessity, however, the statistical screening approach used in that study will result in some false-positive correlations. False-negative results might have appeared due to low statistical power for rare cytogenetic aberrations, as well. In the present study, for which we had collected cytogenetic data on a new set of 156 high-grade STSs, we wanted to assess whether the previously identified correlations were reproducible. In order to make the two series more compatible, and to reduce an impact of tumor location on metastasis development, we decided to include only those tumors that were located in the extremites and trunk wall, yielding a total of 122 cases from the previous series and 156 in the new one. The patient characteristics (eg, regarding follow-up times and frequencies of metastases) were similar in the two series, the only obvious difference being the distribution of histopathologic subtypes of STS. To some extent this difference was due to the fact that the tumors in series 1 had all been part of separate studies within the CHAMP study group, focused on cytogenetic-morphologic correlations in selected histotypes (eg, adipocytic, fibroblastic, and spindle cell tumors). In contrast, the tumors in series 2 were more representative of a consecutive series of high-grade STS and had not been subject to detailed histopathologic re-evaluation. Thus, series 2 included a larger proportion of so-called malignant fibrous histiocytomas and unclassified pleomorphic sarcomas. Despite the differences between the two series, it could be noted that in the combined series, STSs showing myogenic differentiation were associated with increased risk for metastasis (27 of 47 v 97 of 231; P = .005), a finding that is in line with previous studies of pleomorphic STS.^13-15

As expected, grade and size turned out to be strong predictors of metastasis also in series 2. More interestingly, however, three of the 10 previously identified cytogenetic variables—breakpoint in 1p1 and gain of regions 1p1 and 6p1—turned out to provide additional prognostic information also in the new series (Table 3). Furthermore, in our previous study, five cytogenetic markers (ie, breakpoints in 1p1, 1q4, 14q1, and 17q, and gain of 6p1/p2) were statistically significant in addition to clinicopathologic parameters. In the present, combined series of the 278 sarcomas, two of these cytogenetic variables—breakpoint in region 1p1 and gain of region 6p1—remained statistically significant, in addition to grade and size, after the Cox regression analysis (Table 4). These two aberrations, present in 42 and 18 STSs, respectively, were distributed proportionally among the major histopathologic subtypes (data not shown). The numbers of cases in each morphologic subtype were too small to allow for statistical comparison of the seven of eight myogenic and 10 of 21 pleomorphic sarcomas with breakpoints in 1p1, and in three of three myogenic and six of nine pleomorphic sarcomas with gain of region 6p1. Furthermore, only two cases displayed both aberrations, and they were seen as part of highly complex karyotypes as well as together with well-known primary rearrangements, such as the t(X;18) in synovial sarcoma. Taken together, these features suggest that breaks in 1p1 and gain of 6p1 are not part of the same genetic pathway.

Although the results of the present study suggest strongly that structural rearrangements of the proximal part of chromosome arm 1p and gain of the proximal part of 6p are associated with the risk of metastases in patients with high grade STS, the molecular genetic consequences of these chromosomal rearrangements remain unclear. The resolution level we have used in these studies (ie, the chromosome region) is low—chromosome region 1p1 spans approximately 17 Mb and 6p1 14 Mb—and precludes speculations on whether specific genes might be affected by the rearrangements.

From a clinical point of view, the ideal prognostic marker should identify all patients, and exclusively those patients, who will develop metastases (ie, show 100% specificity and sensitivity). From the results of this and other studies, it seems safe to conclude, however, that it is unlikely that such a marker will ever appear for the heterogeneous group of tumors collectively referred to as STSs. The cytogenetic variables that remained significant, besides clinicopathologic parameters, were present in only 15% (breakpoint in 1p1) and 6% (gain of 6p1) of the cases (ie, in less than half of the patients who developed metastases). On the other hand, the same problem with specificity and/or sensitivity also applies to many of the clinical parameters that are widely used as prognostic marker in high-grade STS. For instance, in our study, a tumor size of ≥ 5 cm, which appeared as a strong independent predictor of metastases, was seen in 80% of the cases (ie, as often in tumors that did not metastasize as in those that did). Thus, when deciding on treatment strategies for patients with high-grade STS, one will have to take into account several parameters, among which the tumor karyotype may be a valuable adjunct.

Authors' Disclosures of Potential Conflicts of Interest

The authors indicated no potential conflicts of interest.

Author Contributions

GLOSSARY

Cytogenetic-morphologic correlations:

Studies that correlate cytogenetic findings in tumor cells to tumor morphology. Cytogenetically, the structure of the chromosome is analyzed for chromosomal aberrations using cytogenetic banding tech niques such as G-banding and molecular cytogenetics such as fluoresence in situ hybridization and comparative genomic hybridization. Tumor morphology is typically determined by a clinical pathologist using histochemical techniques.

Karyotype:

An organized chromosomal profile defining chromosomal arrangement and number. In a karyotype, chromosomes are photographically arranged and displayed in pairs, ordered by size. Chromosomal size, banding pattern, and centromere position are typically used as guides to determine chromosomal abnormalities, but improved resolution may be obtained by combining traditional banding techniques with genome-wide molecular cytogenetics such as multicolor fluorescence in situ hybridization (FISH) and locus-specific FISH.

Myogenic:

Pertaining to muscle tissue, myogenic may be related to any of the following: formation of muscle tissue, origination from myocytes or muscle tissue, or molecular signals pertinent to muscle tissues.

Pleomorphic:

Having the ability to exist in various different shapes and forms.