OUP user menu

MDM2 Copy Numbers in Well-Differentiated and Dedifferentiated Liposarcoma
Characterizing Progression to High-Grade Tumors

Patrick L. Ware MD, Anthony N. Snow MD, Maya Gvalani, Mark J. Pettenati PhD, Shadi A. Qasem MD
DOI: http://dx.doi.org/10.1309/AJCPLYU89XHSNHQO 334-341 First published online: 1 March 2014


Objectives: MDM2 gene amplification is associated with well-differentiated (WDL) and dedifferentiated liposarcomas (DDL). Using fluorescent in situ hybridization (FISH), we sought to characterize various patterns of MDM2 amplification among the morphologic spectrum of liposarcoma.

Methods: Forty-six cases of liposarcoma in various sites were examined and included 22 WDLs, 14 DLLs, and 10 negative control subjects.

Results: The MDM2 amplification ratio (MDM2/CEP12) was lower in WDL (10.2) compared with DDL (18.3) cases (P = .0000002). An amplification ratio of 16 showed optimal sensitivity (0.86) and specificity (0.96) as a cutoff point for progression to DDL. Borderline areas, defined as tumors with increased cellularity and atypia but with preserved lipomatous differentiation, showed a significantly higher MDM2 ratio (17.5; P = .0007) compared with WDL. Central (retroperitoneal and intra-abdominal) tumors also showed a significantly higher MDM2 ratio than peripheral ones (P = .02).

Conclusions: Differences in MDM2 amplification profiles among liposarcomas could help further define and predict progression to high-grade neoplasia.

Key Words:
  • MDM2
  • FISH
  • Well-differentiated liposarcoma
  • Dedifferentiated liposarcoma

With the advent of cytogenetic and molecular testing, a subset of malignant lipomatous neoplasms has been characterized with amplification involving the MDM2 gene on the 12q13–15 chromosomal sequence. Both well-differentiated liposarcoma (WDL) and dedifferentiated liposarcoma (DDL) have been found to harbor this genetic alteration, distinguishing them from other benign lipomas as well as malignant lipomatous and nonlipomatous neoplasms that do not show the same amplification. Clinically, this finding has led to a better distinction between benign lipomatous neoplasms and WDLs, especially in cases with subtle cytologic atypia on routine histology. Other genes, such as CDK4, HMGA2, and SAS that are also found within this chromosomal region, have been studied and linked to WDL and DDL.14

Fluorescent in situ hybridization (FISH) is a practical and convenient way to detect MDM2 amplification. FISH allows the MDM2 copy number to be counted in individual cells and a ratio to be calculated in reference to a standard centromeric probe (CEP12). Other methods of testing have also been applied through immunohistochemistry, polymerase chain reaction, and comparative genomic hybridization.1,58

Although MDM2 amplification has become a characteristic genetic marker for these tumors, the actual process of genetic progression from WDL to DDL remains unclear. Clinically, WDL and DDL have differing treatment regimens and mortality characteristics. Even though both WDL and DDL have distinctive histologic appearances, occasional cases show overlapping morphology and lack clear distinction. Being able to accurately distinguish these two entities and better define clinical progression from one to the other could provide more accurate information from which clinical decisions are made. Several studies have commented on the trend in variation of MDM2 copy numbers between WDL and DDL2,9,10; however, none of them have focused on studying this trend or using it for diagnostic purposes. The goal of this study is to determine if the amplification of MDM2 copy numbers is higher in DDL than in WDL using FISH analysis, to determine a potential cutoff point for progression to DDL, and to observe the patterns of this amplification.

Materials and Methods

Forty-six cases were chosen from our institution’s archives with original diagnoses of WDL (n = 22), DDL (n = 14), undifferentiated pleomorphic sarcoma (n = 1), myxoid liposarcomas (n = 2), pleomorphic liposarcoma (n = 1), myxoid sarcoma (n = 1), and various benign lipomatous neoplasms (n = 5). The cases were reviewed for appropriate diagnosis by a soft-tissue pathologist (S.A.Q.). Blocks were chosen and marked with tissue best representing the diagnosis. Four cases were considered to have low-grade dedifferentiation. Low-grade DDL was defined as a spindle cell neoplasm arising in conjunction with a WDL but lacking any lipomatous differentiation and showing low cellularity (approaching that of fibromatosis) with only mild atypia and low mitotic rate Image 1.11 Four cases had an additional block chosen with morphologic features representing “borderline areas,” defined as areas within the WDL with increased cellularity and atypia in which significant lipomatous differentiation was maintained (Image 1). Overall, a total of 50 samples from the 46 cases were included in the study.

Image 1

A, Well-differentiated liposarcoma (WDL) with cytologic atypia (H&E, ×200). B, Dedifferentiated liposarcoma showing high-grade spindle cell proliferation (H&E, ×100). C, Borderline histology characterized by increase in cellularity and atypia while maintaining significant lipomatous differentiation (H&E, ×100). D, Low-grade dedifferentiation with low cellularity and minimal atypia (H&E, ×100). E, WDL case with negative MDM2 amplification showing prominent sclerosis and abundant lipoblasts (H&E, ×100). F, Pleomorphic liposarcoma (control) showing a diffuse population of lipoblasts (H&E, ×100).

FISH Procedure

FISH was performed on 4-μm-thick sections of formalin-fixed, paraffin-embedded tissue. Each slide was baked overnight at 56°C and then deparaffinized using Hemo-De (Scientific Safety Solvents, Keller, TX) and 100% ethanol. The samples underwent pretreatment using the following protocol: 0.2 N hydrochloric acid (HCL) at room temperature (RT) for 20 to 25 minutes, deionized water (H2O) at RT for 3 minutes, pretreatment solution (sodium thiocyanate) at 80°C for 30 to 35 minutes, deionized H2O at RT for 3 minutes, and standard sodium citrate (SSC) buffer at RT for 5 minutes. Protein digestion was then performed using a protease solution of 25 mg pepsin/50 mL of 0.1 N HCL at 37°C for 45 to 60 minutes with an SSC buffer rinse for 5 minutes at RT. After pretreatment, the samples were evaluated for adequate digestion and placed again in the protease solution if further digestion was required. Using 0.5 μL of MDM2 probe (Abbott Molecular, Des Plaines, IL) and 0.5 μL of CEP12 (Abbott Molecular), a probe solution was made in a base of 7 μL LSI buffer (Abbott Molecular) and 2 μL of distilled H2O. The probe was applied to the tissue slides and coverslipped. Slides were hybridized by codenaturing at 74°C for 6 minutes and then fixing for 16 hours at 37°C. Slides were then washed in SSC/0.3% NP-40 solution at 73°C for 2 minutes and air dried, and 6-diamidino-2-phenylindole was applied.


The FISH slides were evaluated by two individuals, a FISH technologist (M.G.) and a pathologist (P.L.W.), both of whom were blinded to the original diagnosis. A total of 60 cells were counted from the area of interest on each slide. The MDM2 and CEP12 signals were averaged to obtain a total score for each sample. Using the averaged totals, a ratio of MDM2 copy number to CEP12 signal and an average MDM2 copy number per cell were calculated. Calculating the ratio was important to control for chromosome 12 polysomy. An amplification ratio of 2 or more has been shown to be diagnostic of amplification and malignancy.2 The average counts from both individuals were compared and a consensus was reached when the counts were significantly different.

Statistical Analysis

Means of subgroups, defined by distinguishing histologic characteristics, were compared using two-tailed Student t tests assuming unequal variances. This test is robust in this setting (low sample number and nonparametric data) meaning that significant P values (<.05) are highly likely to represent true differences in means. Descriptive statistics were calculated for each subgroup. Clinical sensitivity and specificity were calculated for multiple hypothetical threshold values used to distinguish between test groups. These values were also charted graphically to facilitate selection of optimal threshold values.


Fifty samples were reviewed from 43 patients (27 men and 16 women), with a mean age of 59 years (range, 22–86 years). Sites of involvement for all cases included retroperitoneum (n = 19), thigh (n = 8), abdomen (n = 4), scrotum/spermatic cord (n = 4), arm (n = 4), buttock (n = 2), chest wall (n = 1), pancreas (n = 1), back (n = 1), hip (n = 1), and axilla (n = 1). The average tumor size for all specimens was 15.9 cm. The average size for WDL was 18.7 and DDL was 16.7, and the average size of central (intra-abdominal) tumors was 20.5 cm and peripheral tumors was 13.8 cm Table 1.

View this table:
Table 1

Overall, MDM2 amplification was found in 34 of 36 WDL and DDL cases. Ninety-one percent (20/22) of WDLs and 100% (14/14) of the DDLs showed amplification Image 2. The mean MDM2 to CEP12 ratio was 10.2 (range, 1.4–17.2) in the WDL cases and 18.3 (range, 13.1–22.9) in the DDL cases (P = .0000002). The mean MDM2 copy number per cell was 20.9 (range, 3.9–29.9) in the WDL cases and 34.4 (range, 21.5–53.2) in the DDL cases (P = .00001) Table 2. Standard error of the means for each subgroup is provided in Figure 1, and a summary of the P values is provided in Table 3.

Image 2

A, Well-differentiated liposarcoma (WDL) with low amplification of MDM2. B, Dedifferentiated liposarcoma (DDL) with hypercellularity and high MDM2 amplification. C, Sample with borderline histology showing increased cellularity and moderately elevated MDM2 amplification. D, WDL with increased CEP12 signals per cells (polyploidy) as well as increased MDM2 amplification (fluorescence in situ hybridization with MDM2 probe [red] and CEP12 probe [green] at ×1000).

View this table:
Table 2
Figure 1

Mean MDM2 ratio scores comparing different subgroups with standard error of the means.

View this table:
Table 3

Multiple hypothetical threshold values were calculated using sensitivity and specificity rates comparing mean MDM2 amplification ratios between WDL and DDL. A graphical representation Figure 2 shows an MDM2 ratio threshold value of 16 as an optimum balance between sensitivity (0.86) and specificity (0.96).

Figure 2

Graphical representation comparing the threshold values of the mean MDM2 ratio scores between well-differentiated and dedifferentiated liposarcomas.

The borderline areas (n = 4) had a mean MDM2 ratio score of 17.5 (range, 14.3–19.0) and a mean MDM2 copy number per cell score of 31.4 (range, 29.4–33.5). Compared with mean WDL scores, borderline areas showed that differences in mean MDM2 ratio (P = .0007) and MDM2 copy number per cell (P = .00005) were statistically significant and were more within the DDL range. The four cases with low-grade dedifferentiation had a mean MDM2 ratio of 16.9 (range, 15.6–18.5) and a mean MDM2 copy number per cell of 29.7 (range, 25.7–34.3) (Table 2). The trend showed lower mean scores in low-grade dedifferentiation compared with DDL, but this difference was not found to be statistically significant. Two cases did not show MDM2 amplification (ratios of 1.4 and 1.8). Both were originally diagnosed as WDL and presented in a peripheral location. On review, the first case was located in the buttock (8.5 cm) and was felt to represent a sclerosing WDL. The second case was located in the thigh (12 cm) and was characterized by a fibrous background with areas of mature fat admixed with focal condensation of lipoblasts. Polysomy was detected on FISH in this case, but the overall ratio was normal. This case could potentially represent a maturing myxoid liposarcoma (Image 1).

Interestingly, WDL from peripheral sites (arm, buttock, spermatic cord, and thigh; n = 10) had a mean MDM2 ratio of 7.6 (range, 1.4–13.7), and WDL from central sites (retroperitoneum, abdomen; n = 12) had a mean MDM2 ratio of 12.2 (range, 7.7–17.2); the differences were found to be statistically significant (P = .02). Three WDL cases showed homogeneous polysomy with increased mean CEP signal (4.9/3.2/4.5) over baseline (Image 2). All of these cases had a low mean MDM2 ratio (1.4, 2.4, 4.4) and were seen in peripheral sites (thigh, thigh, and buttock). No cases of DDL showed significant polysomy. All control cases (n = 10) were negative for amplification (1.02–1.20) (Table 2).


The MDM2 gene has been found to regulate cell cycling by promoting degradation of the p53 protein. It is believed that upregulation of MDM2 within these tumors leads to inactivation of p53 and subsequent oncogenesis.2,3 MDM2 amplification detection has become a diagnostic tool for differentiating WDL and DDL from benign lipomatous neoplasms and high-grade sarcomas, respectively. Previous studies have found other gene amplifications within the chromosomal region 12q13–15, such as CDK4, HMGA2, and SAS, which also occur with increased frequency in WDL and DDL. Amplification of other chromosomal regions, such as 1p32, 6q23, and 2q14, has also been reported.3,7 Compared with other genetic markers, MDM2 amplification has been shown to have high sensitivity in characterizing WDL and DDL.1,2 Given the diagnostic utility of MDM2 amplification, we investigated the usefulness of FISH testing for better characterization of progression to DDL and studied the various amplification patterns present, with a focus on histology and location.

In our patient population, MDM2 amplification was found to be statistically significantly different between WDL and DDL using FISH. WDL was found to have lower mean scores for both MDM2/CEP12 ratio and MDM2 copy number per cell values compared with DDL. This finding is in line with MDM2 amplification ratio data in WDL and DDL literature, most recently reported by Miura et al9 and Weaver et al.2 This suggests that increased MDM2 amplification may be related to progression to DDL.

Histologically, the separation between WDL and DDL is usually distinct and abrupt; however, in evaluating both WDL and DDL specimens, one often encounters areas with overlapping histologic characteristics. We focused on areas that retained lipomatous differentiation but had increased cellularity. These foci, which conventionally would be classified as WDL, were termed borderline areas. Subset analysis showed that the difference in mean MDM2 ratios for borderline samples and for WDLs was statistically significant. It would seem likely that these areas are potential precursor architectures to dedifferentiation. Given the low number of cases evaluated (n = 4), further retrospective analysis and longitudinal studies are needed to see if this trend can be substantiated.

Low-grade DDL is a recognized morphologic subset of DDL, but its clinical significance has been a matter of debate.1113 In our study, four cases were classified as low-grade DDL. These tumors showed an overall lower ratio of MDM2 amplification compared with DDL, but the trend was not statistically significant.

Evaluation of using FISH preparations highlighted some patterns of MDM2 amplification. WDL typically had a heterogeneous amplification pattern. Amplification was identified not only in the pleomorphic atypical stromal cells but also in histologically normal-appearing adipocytes, as has been previously reported.2 Interestingly, the atypical stromal cells frequently showed significantly higher amplification of MDM2 but also increased CEP12 signals (indicating polysomy), with resulting MDM2 ratios falling within the expected range for WDL. DDL showed more homogeneous MDM2 amplification throughout the neoplastic cells with near-normal CEP12 signals. Cases with heterologous differentiation in the form of bone, cartilage, skeletal muscle, and smooth muscle also showed MDM2 amplification in these cellular elements, suggesting a common neoplastic origin (Image 2).

Peripherally located WDLs (extremities) were found to have significantly lower MDM2 amplification ratios than central (retroperitoneal/intra-abdominal) tumors. Compared with central WDLs, peripheral WDLs are known to have a more indolent clinical course and better overall survival.14 The fact that peripheral tumors have lower MDM2 amplification ratios could imply that WDL with lower ratios are biologically less likely to recur and dedifferentiate. On the other hand, it may suggest that retroperitoneal tumors have higher amplification because they are long-standing and are discovered later in the clinical course. Interestingly, central tumors were significantly larger than peripheral tumors in our study. Three cases of peripherally located WDLs also showed a homogenous polysomy pattern with an overall increase in CEP12 signals in most cells, resulting in a lower (marginally positive) MDM2 ratio. This identifies polysomy as a component and potential mechanism for MDM2 amplification within WDL. Polysomy was not observed in any WDL originating in deep locations or in DDL.

FISH has some limitations for determining MDM2 amplification. For instance, in cells with high amplification of the MDM2 gene, the accuracy was believed to be decreased because of signal overlap and clustering. This was more likely to occur with DDL because the increased signal in WDL was often accompanied by an increase in cell size and surface area, which allowed for spreading out of the signal and better accuracy. WDL also showed increased variability of MDM2 amplification in the same neoplasm. This is in agreement with Kashima et al,10 who noted that in rare instances this may lead to false-negative results and recommended that one may occasionally need to do the analysis on multiple slides.

Overall, the study shows that DDL characteristically has a higher MDM2 amplification ratio than WDL when evaluated using FISH. In addition, variability in location and histologic architecture in WDL and DDL may result in different MDM2 amplification profiles. This study provides an initial step to show how amplification profiles could provide an adjunct in characterizing progression to DDL. Correlation with other methods, such as polymerase chain reaction and comparative genomic hybridization, may help confirm our findings and more accurately quantify the degree of amplification. Further studies comparing not only morphology but also MDM2 copy numbers with clinical outcomes could help better define the boundaries between these two entities and the mechanism of progression to high-grade sarcoma. In addition, given the recent application of CDK4 inhibitors in the treatment of liposarcoma,15 studying amplification profiles for CDK4 could be a useful adjunct in characterizing the progression of these tumors and further guide management decisions.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
View Abstract