Context.—Defects in mismatch repair proteins have been identified in Lynch syndrome–associated liposarcomas, as well as in rare sporadic sarcomas. However, it is unclear if mismatch repair defects have a role in sarcoma tumorigenesis. Microsatellite instability is a surrogate marker of mismatch repair defects.
Objectives.—To determine whether sporadic dedifferentiated liposarcomas display microsatellite instability and, if so, to evaluate whether such instability differs between the lipogenic and nonlipogenic components of these tumors.
Design.—The diagnoses of conventional dedifferentiated liposarcoma were confirmed by a combination of morphologic, immunophenotypic, and molecular studies. Standard fluorescence-based polymerase chain reaction, including 5 mononucleotide microsatellite markers (BAT25, BAT26, NR21, NR24, and MONO27), as well as 2 pentanucleotide repeat markers (Penta C and Penta D), was used to test for instability and loss of heterozygosity.
Results.—We demonstrated only a single case (1 of 43) with microsatellite instability at one mononucleotide marker. No sarcomas showed high-level microsatellite instability. However, loss of heterozygosity at the pentanucleotide markers was observed in 8 of 43 cases. The presence of loss of heterozygosity was overrepresented in the nonlipogenic (dedifferentiated) components compared with the paired lipogenic (well differentiated) components.
Conclusions.—Mismatch repair defects do not contribute to sporadic dedifferentiated liposarcoma tumorigenesis. Whether the observed loss of heterozygosity drives tumorigenesis in liposarcoma, for example by affecting tumor suppressor or cell cycle regulator genes, remains to be determined.
Germline mutations in DNA mismatch repair genes, including MLH1, MSH2, MSH6, and PMS2, are seen in Lynch syndrome (hereditary nonpolyposis colon cancer syndrome) and can be detected experimentally by microsatellite instability (MSI). Defects in mismatch repair frequently result in colorectal and endometrial adenocarcinoma.1,2 However, other malignancies have been demonstrated to arise less frequently in Lynch syndrome and include tumors of the ovary, biliary tract, upper gastrointestinal tract, and urothelium, as well as soft tissue sarcomas.1–6 Although various sarcomas are associated with Lynch syndrome, including leiomyosarcoma and undifferentiated pleomorphic sarcoma, liposarcoma seems to be the most common.2–6 In patients with Lynch syndrome, all of these sarcoma subtypes have mutations in mismatch repair (MMR) genes. Furthermore, a 2003 study6 showed increased rates of sarcoma in a PMS2-deficient mouse model. Taken together, these findings suggest that defective DNA repair mechanisms may contribute to sarcoma pathogenesis. The role of mismatch repair defects in sporadic sarcomas is poorly understood, although studies5,7,8 so far suggest that such defects seem to be rare.
Liposarcoma, specifically the well-differentiated and dedifferentiated subgroup, represents the most commonly occurring sporadic (ie, not Lynch syndrome associated) soft tissue sarcoma. Dedifferentiated liposarcoma is defined by the presentation, either synchronous or metachronous, of well-differentiated (ie, lipogenic) and dedifferentiated (ie, nonlipogenic) components.9 Both components share a reproducible amplification of chromosome subregion 12q13–15.10–14 Our group recently reported that the matched pairs of lipogenic and nonlipogenic components of dedifferentiated liposarcoma are genetically quite similar.13 Therefore, large chromosomal gains and losses likely take place early during tumorigenesis. The mechanisms of progression from a lipogenic to a nonlipogenic sarcoma remain poorly characterized but may depend on smaller genetic changes.
Although MSI, resulting from mismatch repair defects, has been rarely reported in a heterogeneous group of sporadic sarcomas,8,15 its presence in liposarcoma specifically has not been well characterized. Consequently, the role of MSI, if any, in sporadic liposarcoma tumorigenesis and/or progression is unknown. The present study evaluates the presence of MSI in a large, well-categorized group of sporadic dedifferentiated liposarcomas arising from well-differentiated precursors.
MATERIALS AND METHODS
After approval of this study by our institution's committee on human research, the departmental archives were searched for examples of dedifferentiated liposarcoma. We retrieved 43 tumors from the Department of Pathology database with adequate normal and lesional paraffin-embedded tissue for microdissection.
The diagnosis of well-differentiated liposarcoma required the presence of a lipogenic component and a synchronous or metachronous nonlipogenic component.9 The dedifferentiated components were graded according to the current World Health Organization criteria,9 with pleomorphic sarcomas resembling undifferentiated sarcoma (so-called malignant fibrous histiocytoma) considered high grade and uniform spindle cells with mild nuclear atypia, often in a fascicular pattern, considered low grade. As defined herein, the pattern of low-grade dedifferentiation is indistinguishable from “cellular” well-differentiated liposarcoma.9,16
We have previously reported on 16 of these cases as demonstrating the 12q13–15 amplification.13 The remaining 27 cases showed overexpression of MDM2 and CDK4 by immunohistochemistry. All tumors were sporadic, with no documented history of Lynch syndrome.
Microdissection and MSI Testing
From each tumor, the best-preserved nonlipogenic, dedifferentiated component and normal tissue were identified on hematoxylin-eosin–stained slides. Manual microdissection was then performed on unstained slides from formalin-fixed, paraffin-embedded tissue for DNA extraction as previously described4,17,18 and as typically performed at our institution for MSI testing. The presence of mismatch repair defects was detected using standard fluorescence-based polymerase chain reaction with the Promega MSI Analysis System, version 1.2 (Promega, Madison, Wisconsin) kit, including 5 mononucleotide microsatellite markers (BAT25, BAT26, NR21, NR24, and MONO27), as well as 2 pentanucleotide repeat markers (Penta C and Penta D). Although the pentanucleotide markers are not evaluated for MSI, most individuals are heterozygous at these loci, allowing for detection of loss of heterozygosity (LoH). The mononucleotide repeats are not used in assessment of LoH because most individuals are homozygous at these loci. In cases where MSI or LoH was present, the well-differentiated component was also tested for both MSI and LoH using the same methods. Microsatellite instability was subcategorized into the following 3 groups: microsatellite stable, defined as the absence of MSI in all 5 mononucleotide markers; MSI-low, defined as MSI at one marker; and MSI-high, defined as MSI at 2 or more loci.19 Loss of heterozygosity was defined as an apparent loss of 1 of 2 alleles in the lesional tissue compared with normal tissue.
Clinicopathologic and Genomic Features
Most tumors (36 of 43; 84%) showed a lipoma-like liposarcoma (Figure, A) juxtaposed to a high-grade pleomorphic or spindle cell sarcoma in the nonlipogenic component (Figure, B through D). The remaining 7 tumors showed low-grade histology, as defined by the World Health Organization, with uniform spindle cells with mild atypia in a fascicular pattern.9 All of these latter tumors averaged fewer than 5 mitoses per 10 high-power fields so they may also be considered so-called cellular well-differentiated liposarcoma according to the criteria by Evans.16 None of the tumors demonstrated the recently described “homologous” dedifferentiated phenotype.20,21 The most common site was the retroperitoneum (25 of 43; 58%). Based on our inclusion criteria, all dedifferentiated liposarcomas demonstrated 12q13–15 amplification (Figure, E) and/or MDM2 and CDK4 overexpression by immunohistochemistry (data not shown).
Microsatellite instability data are summarized in Table 1, with detailed results showing individual cases and markers given in Table 2. Forty-two of 43 cases (98%) of dedifferentiated liposarcoma were microsatellite stable. All paired lipogenic components separately tested (n = 9) were also microsatellite stable. One case (2%) of dedifferentiated liposarcoma demonstrated MSI at a single locus (NR24) and therefore was classified as MSI-low; this same case also had LoH at Penta D. The case consisted of a high-grade dedifferentiated liposarcoma, 6.5 cm, primary to the trunk, which developed in a 57-year-old man who at initial presentation had metastatic disease to the lung. The paired lipogenic component of this tumor was MSI stable.
Loss of Heterozygosity
Overall, LoH was observed in 8 of 43 dedifferentiated liposarcomas (19%) (Table 1). Loss of heterozygosity was identified at Penta C (3 of 43; 7%), Penta D (3 of 43; 7%), or both (2 of 43; 5%). The dedifferentiated liposarcomas with LoH showed predominantly (7 of 8; 88%) high-grade pleomorphic histology, but the presence of LoH did not correlate with grade (P = .80) or tumor location (P = .40). In 4 tumors (50%), LoH was present only in the nonlipogenic component but was absent in the matched lipogenic component (Table 2).
The association between Lynch syndrome and liposarcomas, as well as rare previous reports of sporadic sarcomas with MSI, raises the possibility that defective mismatch repair genes have roles in the pathogenesis of liposarcoma.2,5,8,15 However, no previous studies specifically addressing MSI in sporadic liposarcomas have been reported to date. In addition, while dedifferentiated liposarcomas are genetically complex, no known reproducible genomic changes distinguish the lipogenic and nonlipogenic components of these tumors.13 Our study sought to determine if MSI was present in sporadic cases of dedifferentiated liposarcoma and, if so, to evaluate whether these genetic alterations varied between paired well-differentiated (lipogenic) and dedifferentiated (nonlipogenic) components.
Our results demonstrated largely the absence of MSI, with only one tumor categorized as MSI-low. Therefore, mismatch repair defects likely do not contribute to sporadic dedifferentiated liposarcoma tumorigenesis. The low incidence of MSI, as hypothesized by Suwa et al,15 suggests that in liposarcoma MSI is a secondary effect of overall genomic loss and alterations, which may affect mismatch repair genes rather than having a causative role in tumorigenesis. The previously observed association between Lynch syndrome and liposarcoma may represent tumorigenesis mechanisms involving mismatch repair defects unique to these patients.
Prior studies examining MSI in sarcomas have focused on morphologically or genetically (ie, pleomorphic undifferentiated sarcoma) heterogeneous groups of sarcomas. Ericson et al8 examined 209 soft tissue sarcomas with pleomorphic sarcoma-like histology for defects in mismatch repair using both immunohistochemistry (antibodies against MLH1, MSH2, and MSH6) and polymerase chain reaction and demonstrated that rare cases (2 of 209 tumors) have MSI. Suwa et al15 demonstrated that 3 of 39 sarcomas had MSI, including 1 of 7 liposarcomas. The latter study used different microsatellite loci than are typically used for Lynch syndrome screening and lacked the mononucleotide repeats that have been shown to be more sensitive and specific for detection of defective mismatch repair22 ; LoH at the same loci was noted when present. Our study is the first to date to comprehensively examine a large number of dedifferentiated liposarcomas with paired well-differentiated components and more definitively excludes MSI as a tumorigenic mechanism in sporadic liposarcomas.
The present study also demonstrated LoH in a subset (8 of 43) of liposarcomas and showed that LoH occurred more frequently in the dedifferentiated component than in their paired well-differentiated component. Using the same cohort of liposarcoma cases as in the present study, Horvai et al13 identified complex genetic changes, including confirmation of 12q amplification in both dedifferentiated liposarcoma components. However, no genetic changes were identified that consistently distinguish lipogenic and nonlipogenic components of dedifferentiated liposarcomas. Prior studies using other methods have sought to determine the changes that lead to possible progression in liposarcoma. These include studies23–25 focused specifically on LoH at the retinoblastoma (Rb) gene, which was identified more frequently in dedifferentiated liposarcomas (54%–100%), with less frequent (0%–8.7%) concurrent loss in paired well-differentiated components and no loss in tumors demonstrating only well-differentiated liposarcoma morphology. Given the increased frequency of LoH in dedifferentiated liposarcoma and the role of Rb in cycle cell regulation, it is hypothesized that Rb LoH may contribute to the progression from well-differentiated to differentiated liposarcoma. Suwa et al15 also identified a single case of liposarcoma with LoH at 2 separate MSI loci. Loss of heterozygosity at other gene loci has not previously been described in liposarcoma, but it could also contribute to tumorigenesis if occurring at loci where genetic alterations would influence tumor suppressors or cycle cell regulatory genes.
The Penta C and Penta D microsatellites are located on chromosomes 9p and 21q, respectively.26 Our prior work has not shown that these loci are consistently altered in dedifferentiated liposarcoma.13 Thus, it is unclear whether LoH at Penta C and Penta D represents recurring mutations driving tumorigenesis or passenger mutations. Compared with other pleomorphic sarcomas, dedifferentiated liposarcoma shows few genetic changes, suggesting that early tumorigenesis does not require marked chromosomal instability.14,27 The dedifferentiated components of liposarcomas, based on the relative enrichment of LoH, may suggest a more generalized genomic unrest during progression of this subgroup of tumors.
In summary, MSI and LoH are rare events in sporadic dedifferentiated liposarcomas. Given that MSI-low only occurred in a single case, mismatch repair defects likely do not contribute to sporadic dedifferentiated liposarcoma tumorigenesis. The results demonstrate that LoH at either of 2 loci is present in a subset of dedifferentiated liposarcoma and less frequently manifests in their paired well-differentiated tumor. Therefore, LoH may contribute to tumorigenesis or may represent passenger mutations during tumorigenesis.
This research was supported by the Residents' Teaching and Research Endowments from the Department of Pathology, University of California, San Francisco (Dr Davis). We thank Martin Powers, MD, for critical review of the manuscript.
The authors have no relevant financial interest in the products or companies described in this article.
Presented in part at the 101st Annual Meeting of the United States and Canadian Academy of Pathology; March 19, 2012; Vancouver, British Columbia, Canada.