- Top of page
- MATERIALS AND METHODS
For the detection of chromosome translocations/chimeric genes and specific genetic abnormalities in soft tissue tumors, we conducted fluorescence in situ hybridization (FISH) analysis on 280 cases of soft tissue and other tumors using formalin-fixed paraffin-embedded tissue sections. The detection rate of the FISH split-signal was 84% (129/154 cases) for the translocation-associated soft tissue tumors, such as Ewing's sarcoma/primitive neuroectodermal tumor, synovial sarcoma, alveolar rhabdomyosarcoma, myxoid liposarcoma, clear cell sarcoma and so forth. Positive split-signals from EWSR1, SS18 and FOXO1A probes were detected in 3% (2/64) of various histological types of carcinoma, lymphoma, melanoma, meningioma and soft tissue tumors. In FISH using the INI1/CEP22 probe, the INI1 deletion signal was detected in 100% (9/9) of epithelioid sarcoma. In well-differentiated and dedifferentiated liposarcomas, detection of MDM2 amplification signals in FISH using the MDM2/CEP12 probe were both as high as 85% (11/13) and 100% (13/13), respectively. In other adipocytic and non-adipocytic tumors requiring differentiation from these types, detection was only 13% (5/39), and CEP12 polysomy was frequently detected. As these results demonstrate the high sensitivity and specificity of FISH, we concluded FISH to be a useful pathological diagnostic adjunct for definite and differential diagnosis of soft tissue tumors.
- Top of page
- MATERIALS AND METHODS
Having conducted FISH analysis on a soft tissue tumor group with chimeric genes derived from chromosome translocations, as well as on well-differentiated and dedifferentiated liposarcoma in which the MDM2 gene is amplified, and on epithelioid sarcoma in which the INI1 gene is deleted, we detected a high prevalence (83% to 100%) of genetic abnormalities in tumors other than extraskeletal myxoid chondrosarcoma.
In particular, among the soft tissue tumor group with chimeric genes derived from chromosome translocations, split-signals of specific genes were detected at high rates in Ewing's sarcoma/PNET, clear cell sarcoma of soft parts, desmoplastic small round cell tumor, angiomatoid fibrous histiocytoma, synovial sarcoma, myxoid liposarcoma, and low grade fibromyxoid sarcoma, which is in accordance with previous reports on FISH analysis.5,7,9–15
Among these tumor groups, FISH analysis has revealed a high prevalence of EWSR1 gene rearrangement in extraskeletal myxoid chondrosarcoma.16 Conducting FISH on a mixture of new cases as well as on those previously reported8 revealed rearrangement of the EWSR1 gene in 76% (16/21) of cases and rearrangement of the NR4A3 gene in 71% (15/21) cases. Given that the size of the NR4A3 probe is smaller than that of the EWSR1 probe, the signal intensity arising from the NR4A3 probe was slightly less than that of the EWSR1 probe. For this reason, in this study, the split-signal detection rate of NR4A3 was slightly lower than that of EWSR1. In addition, the split-signal detection rate in extraskeletal myxoid chondrosarcoma was found to be slightly lower than that in other histological types. The presence of a novel chimeric gene and weakly established diagnostic criteria for this tumor were estimated reasons for this.
Similarly for alveolar rhabdomyosarcoma, FISH analysis with the FOXO1A probe on a mixture of new cases as well as on those previously reported6 revealed a FOXO1A split-signal in 94% (16/17) of cases and increased FOXO1A centromeric region copy numbers in 65% (11/17) of cases. Previous studies have reported the same genome copy number increase in alveolar rhabdomyosarcoma.17 Protein coded by the PAX3-/PAX7-FOXO1A gene generated by chromosome translocation is a strong transcription factor, and tumor cell proliferation has been shown to be enhanced upon the expression of the PAX3-FOXO1A gene.18 This suggests that PAX3-/PAX7-FOXO1A expression and the increase in copy number mediate the mechanisms of tumor formation and progression.
In accordance with previous reports,19,20 the PDGFB split-signal was detected by FISH with a high prevalence (100%, 13/13 cases) in DFSP. Furthermore, an increased PDGFB gene centromeric region copy number (median green/red signal ratio of 1.8) was detected in 85% (11/13) of cases. Various reports have been made on excess signals being generated due to an increase in chromosome translocation-induced COL1A1-PDGFB gene copy number,19,20 and our results matched with these findings. Abbott et al. showed that by using a dual-color split-signal probe in DFSP-FS, the median COL1A1-PDGFB gene copy number in the fibrosarcoma region (2.8) was higher when compared with the DFSP region (1.7).19 On the other hand, differences in median PDGFB gene centromeric region copy numbers were not observed in this study, as values were 2.0 in normal DFSP, 2.4 in the DFSP part of DFSP-FS, and 1.6 in the fibrosarcoma part of DFSP-FS. These results are in accordance with FISH studies using dual-color, dual-fusion probes.21,22 The present findings suggest that a tumor-forming mechanism, other than COL1A1-PDGFB gene increase, mediates fibrosarcomatous transformation of DFSP.
The TFE3 split-signal was detected in 100% (9/9) of ASPS cases, which was a high detection rate similar to a past report.23 As to immunohistochemical status of TFE3 expression, 95% (18/19) of ASPS demonstrated strong (3+) nuclear staining, and one (5%) demonstrated moderate (2+) nuclear staining.24 In the present study, we determined that cases with one fused and one missing green signal were those of unbalanced chromosome reciprocal translocation, where a fused signal is green (indicating the TFE3 gene centromeric region) and red (indicating the telomeric region). Unbalanced chromosome reciprocal translocation has been known as a characteristic of ASPS.25 As it is difficult to determine such signal patterns as split-signals, care is needed in the analysis of ASPS FISH signals. In addition, previous studies have shown that excess signals have been detected as a result of multiplication of the X chromosome short arm telomeric region, which is fused to ASPSCR1 and located on chromosome 17 due to chromosome interchange.23 We observed many different types of signal patterns in our FISH analysis. Therefore in addition to balanced and unbalanced chromosome reciprocal translocation, partial multiplication or deletion of the X chromosome or chromosome number abnormalities might occur in ASPS.
As FUS-DDIT3 gene formation from chromosome translocation t(12;16)(q13;p11) or EWSR1-DDIT3 gene formation from chromosome translocation t(12;22)(q13;q12) are known in myxoid liposarcoma,1FUS/EWSR1-DDIT3 fusion can also be assessed using FUS and EWSR1 probes. In a previous report, the DDIT3, FUS and EWSR1 split-signal was detected in 100% (18/18), 94.4% (17/18) and 0% (0/18) of myxoid liposarcoma cases, respectively.15
In this study, we conducted FISH analysis with EWSR1 and SS18 probes on various histological types of tumors requiring differential diagnosis from Ewing's sarcoma/PNET and synovial sarcoma. Similarly, the FOXO1A probe was used to conduct FISH analysis on embryonal rhabdomyosarcoma, which is often difficult to distinguish from alveolar rhabdomyosarcoma. As reported previously,6 the split-signal of the FOXO1A gene was undetected in embryonal rhabdomyosarcoma. EWSR1 and SS18 gene rearrangements were undetected in not only soft tissue tumors but also carcinoma, malignant lymphoma, malignant melanoma, and meningioma; thus, high specificity was shown. Among these, the EWSR1 split-signal was detected in both myoepithelial carcinoma and malignant lymphoma (lymphoblastic lymphoma) as previously reported.26,27 Therefore, care is required for differentiating these histological types.
Epithelioid sarcoma is classified into the typical distal-type that arises in the limbs of young adults and the proximal-type that arises mainly in the groin and pelvis of middle-aged adults. In comparison to the distal-type, the proximal-type has been reported to be more malignant, leading to a poorer outcome.28 On the other hand, a recent report showed no major difference in the degree of malignancy between these two types.29 Modena et al. demonstrated by FISH analysis of proximal-type (six cases) and distal-type epithelioid sarcomas (five cases) that INI1 gene deletion is found only in proximal-type epithelioid sarcoma (five cases).4 Immunohistochemistry has been used to show frequent loss of the INI1 protein expression in MRTs, and proximal-type and distal-type epithelioid sarcomas.29,30 In our FISH analysis, INI1 gene deletion was observed in 100% of proximal-type (five cases) and distal-type (four cases) epithelioid sarcomas, which supports the INI1 protein loss result, 95% (61/64) of proximal-type and 91% (58/64) of distal-type, obtained by immunohistochemistry.30 On the basis of FISH conducted by Modena et al., the signal pattern of INI1 deletion was mostly homozygous deletion.4 In our study, heterozygous deletion (when INI1 : CEP22 ratio = 1:2 or 1:1) was detected in 28.4% of proximal-type and 24% of distal-type cases, showing no major differences. On the other hand, homozygous deletion (when INI1 : CEP22 ratio = 0:1 or 0:2) in distal-type was 60.5%, which was higher in comparison to the 37.8% in the proximal-type epithelioid sarcomas. We will need to increase the number of cases to further analyze these genetic abnormalities such as INI1 gene deletions, which are characteristic of proximal-type and distal-type epithelioid sarcomas.
Supernumerary ring or giant rod marker chromosomes are observed in well-differentiated and dedifferentiated liposarcoma, and these characteristic chromosomes consist of amplified sequences of the 12q13-15 region containing MDM2, CDK4, HMGA2, and SAS genes.3,31 Weaver et al. detected MDM2 gene amplification in 100% of well-differentiated liposarcoma (13 cases) and dedifferentiated liposarcoma (14 cases) by FISH.32 On the other hand, the MDM2 gene amplification frequency shown by Sirvent et al. was 94% (30/32 cases) in well-differentiated liposarcoma and 100% (8/8 cases) in dedifferentiated liposarcoma.33 In our FISH analysis, MDM2 gene amplification was detected in 85% (11/13) of well-differentiated liposarcoma cases and 100% (13/13) of dedifferentiated liposarcoma cases. In agreement with the reports of Sirvent et al.,33 our results showed a lower detection rate of MDM2 gene amplification in well-differentiated liposarcoma when compared with dedifferentiated liposarcoma. Furthermore, in agreement with the reports of Weaver et al.,32 the mean MDM2/CEP12 ratio was lower in well-differentiated liposarcoma (11.47) compared with dedifferentiated liposarcoma (18.59). Weaver et al. did not detect CEP12 polysomy in well-differentiated liposarcoma,32 although we found CEP12 polysomy in 15% (2/13) of well-differentiated liposarcoma cases. It has been known clinically that dedifferentiated liposarcoma has a poorer outcome in comparison to well-differentiated liposarcoma.34 The correlation between the MDM2 gene status and the biological behavior of well-differentiated and dedifferentiated liposarcomas needs to be examined in further studies.
Immunohistochemically, MDM2 was expressed in 100% (44/44) of well-differentiated and 95.1% (58/61) of dedifferentiated liposarcoma cases.35 We have experienced several cases of well-differentiated liposarcoma with negative MDM2 immunostaining but MDM2 gene amplification. In these situations, the use of FISH for the detection of MDM2 amplification is frequently conclusive.
In terms of histopathology, there are many cases where differentiation of well-differentiated liposarcoma from lipoma, spindle cell lipoma, and other liposarcomas (myxoid, pleomorphic, and spindle cell) becomes problematic. In lipoma, chromosome translocation-induced chimeric genes are formed from the HMGA2 gene that is present within the 12q13-15 region. Known partner genes include LPP (3q27-28), NFIB (9p22), CXCR7 (2q37), EBF1 (5q33), and LFHP (13q12).31 In spindle cell lipoma, which is a subtype of lipoma, deletion of 16q, partial deletion of 13q, and monosomy of chromosome 13 have been detected.31 There are no known characteristic chromosome structures or copy number abnormalities in pleomorphic liposarcoma, where chromosome and genetic abnormalities in this type are highly variant and complex.31,36 Spindle cell liposarcoma is classified as a subtype of well-differentiated liposarcoma, although because of the lack of MDM2 and CDK4 gene amplification, the possibility of being a subtype of a different histological type has been suggested.37 In our FISH analysis, MDM2 gene amplification was detected in 33% (1/3) of pleomorphic liposarcoma cases, although the mean MDM2/CEP12 ratio (2.2) was lower in comparison to well-differentiated (11.47) and dedifferentiated (18.59) liposarcomas. MDM2 gene amplification was undetected (0/21 cases) in myxoid liposarcoma (nine cases), spindle cell liposarcoma (three cases), lipoma (six cases), and spindle cell lipoma (three cases). In addition, CEP12 polysomy was detected in a 100% (3/3) of pleomorphic liposarcoma cases, though it was as low as 5% (2/38 cases) in other liposarcomas (well-differentiated, dedifferentiated, myxoid, and spindle cell). CEP12 polysomy has not been reported previously in lipoma, though we detected this in 17% (1/6) of lipoma cases and 33% (1/3) of spindle cell lipoma cases.
Myxofibrosarcoma and pleomorphic MFH, which require differential diagnosis from the dedifferentiated part of dedifferentiated liposarcoma, both have characteristic genetic abnormalities. There are no specific chromosome structures or copy number abnormalities in these histological types, although they carry abnormalities such as increase, decrease, and high level amplification of various chromosome regions, which leads to highly complex karyotypes.36 In our FISH analysis, MDM2 gene amplification was detected in 33% (4/12) of myxofibrosarcoma cases, although the mean MDM2/CEP12 ratio (4.28) was lower when compared with dedifferentiated liposarcoma (18.59). MDM2 gene amplification was undetected in pleomorphic MFH (0/3 cases). In previous studies, MDM2 gene amplification was undetected in myxofibrosarcoma (one case), although it was found in 40% (4/10) of pleomorphic MFH cases.32 Furthermore, the detection rate of CEP12 polysomy in dedifferentiated liposarcoma was 0% (0/13 cases) in the present study, though this was high in myxofibrosarcoma (75%; 9/12 cases) and pleomorphic MFH (100%; 3/3 cases).
Regarding the pleomorphic liposarcoma and myxofibrosarcoma cases with MDM2 amplification, unusual myxoid MFH-like, mixed or homologously dedifferentiated subtypes of liposarcoma should be included into differential diagnosis.38–40 However, these cases occurred in the extremities, different from the retroperitoneum typical of dedifferentiated liposarcoma, and lacked well-differentiated liposarcoma components morphologically.
In conclusion, this study demonstrates the high sensitivity and specificity of FISH for detecting chromosomal and genetic abnormalities (the formation of chimeric genes by chromosome translocation and the deletion or amplification of specific genes) specific to soft tissue tumors. FISH is therefore a useful pathological diagnostic adjunct for analyzing formalin-fixed paraffin-embedded tissue sections of soft tissue tumors. Moreover, definite diagnosis could be obtained by identifying FISH signal patterns characteristic to each histological type.