Clinical and Biological Significance of CDK4 Amplification in Well-Differentiated and Dedifferentiated Liposarcomas

The CDK4 gene encodes a 33-kD protein that plays an important role in the regulation of the G₁-S transition of the cell cycle (1). CDK4 forms molecular complexes with members of the cyclin D family, such as cyclin D1 (CCND1). The phosphorylation of RB1 protein by the CDK4-CCND1 complex leads to the release of the E2F transcription factor, which up-regulates gene expression required for progression through the S-, G₂-, and M-phases (1). The activity of CDK4 is negatively regulated by p16INK4a, which prevents its binding to CCND1 (Fig. 1). p16INK4a is the product of the CDKN2A gene that also encodes p14ARF (2). Alterations of genes involved in the CCND1/CDK4/p16INK4a/RB1/E2F pathway play a crucial role in the pathogenesis of many tumor types (1, 3).

We and others have shown that well-differentiated/dedifferentiated liposarcoma (WDLPS/DDLPS) cells contain supernumerary ring or giant marker chromosomes composed of highly amplified sequences from the 12q14-15 chromosomal region (4, 5). Although the MDM2 and CDK4 genes were initially considered to be the two main targets of the 12q14-15 amplicon in WDLPS/DDLPS, we have recently shown (6) that only MDM2 (12q15) and HMGA2 (12q14.3) were consistently amplified in WDLPS/DDLPS. The CDK4 gene (12q14.1) belongs to a distinct inconsistent amplicon that is not present in about 10% of cases (6). The concomitant amplification of MDM2 and HMGA2 seems therefore to be the crucial event in WDLPS/DDLPS pathogenesis. However, considering the important function of CDK4 in the cell cycle, CDK4 gene amplification and overexpression very likely play an important role in the tumorigenic process. This may have clinical implications because several anticancer drugs belonging to the group of “CDK inhibitors” are currently in preclinical development or under clinical trials (7).

Further clarification of the role of CDK4 amplification in WDLPS/DDLPS is needed for a better understanding of the tumorigenesis of such tumors. We report here a study which aimed at determining whether the absence of CDK4 amplification in WDLPS/DDLPS is (a) correlated with specific clinicopathologic features, and (b) compensated by another genomic event involved in the CCND1/CDK4/p16INK4a/RB1/E2F pathway.

From 1991 to 2008, 45 cases of WDLPS/DDLPS, for which amplification of MDM2 but no amplification of CDK4 was detected (MDM2+/CDK4-), were retrieved from the respective databases of the Laboratory of Solid Tumor Genetics (Nice, France) and of the Department of Pathology of the Bergonié Institute (Bordeaux, France). The clinicopathologic characteristics of these cases were compared with those of a series of 143 WDLPS/DDLPS with amplification of both MDM2 and CDK4 (MDM2+/CDK4+) analyzed during the same period. In all the cases, the diagnosis of WDLPS/DDLPS was established according to the WHO Classification of Tumors (4). The MDM2/CDK4 status of all the cases was assessed by fluorescence in situ hybridization (FISH) and/or quantitative PCR as previously described (8).

FISH analysis was done in 30 cases (18 MDM2+/CDK4−, 12 MDM2+/CDK4+) with available formalin-fixed, paraffin-embedded blocks, in order to determine the genomic status of the CDKN2A (9p21.3), CCND1 (11q13.3), and RB1 (13q14.2) genes. Five-micrometer sections were deparaffinized, rehydrated, and incubated with pepsin by using the Histology FISH Accessory Kit (Dako) according to the manufacturer's instructions. Three FISH probes were used in this study: the p16/CEP9 Dual Color Probe (Abbott Molecular), and the bacterial artificial chromosome probes RP11-156B3 (11q13.3, CCND1) and RP11-839E5 (13q14.2, RB1). BAC clones from the Roswell Park Cancer Institute library were selected according to their location on the University of California Santa Cruz database (http://genome.ucsc.edu/; March 2006 release), were obtained from the Children's Hospital Oakland Research Institute (CHORI) (http://bacpac.chori.org/), and were prepared as probes for FISH analysis according to standard procedures. Microscopic analysis was done using a DM6000B microscope (Leica) and images were processed using the ISIS software (MetaSystems). At least 100 nuclei per slide were analyzed. Amplification was defined as the presence of ≥10 fluorescent signals per cell in >1% of cells. Deletion was defined as >50% nuclei with only one fluorescent signal.

DNA from frozen material of three cases of MDM2+/CDK4-WDLPS/DDLPS was extracted and processed for array comparative genomic hybridization (array-CGH) as previously described (9). The microarray used contained 3,342 sequence-validated BAC with an average resolution of 1 Mb (9).

Quantitative reverse transcription-PCR (qRT-PCR) was used to determine the expression levels of p16INK4a, p14ARF, CCND1, and RB1 in 16 WDLPS/DDLPS cases (7 MDM2+/CDK4-, 9 MDM2+/CDK4+) for which frozen material was available and in the 4 MDM2+/CDK4+ WDLPS/DDLPS cell lines 93449, 94778, 95T1000, and 98T1430 (6). Total RNAs were extracted from WDLPS/DDLPS and normal subcutaneous adipose tissue (NSAT) using either the RNeasy lipid tissue minikit (Qiagen; frozen fragments) or Trizol (Invitrogen; WDLPS/DDLPS cell lines) according to the manufacturer's protocol. The RNA samples were treated by DNA-free (Applied Biosystems). One microgram of total RNA was reverse-transcribed into cDNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems) and each qRT-PCR experiment was done twice in duplicate with the ABI PRISM 7300 Detection System using either FAM dyes (Applied Biosystems) for CCND1, RB1, and p14ARF or SYBRGreen dye for p16INK4a according to the manufacturer's protocol. RPLP0 (the large P0 subunit of the acidic ribosomal phosphoprotein) was used as endogenous control for normalization. QRT-PCR was done using the following TaqMan gene expression assays (Applied Biosystems): Hs99999189_m1 (p14ARF), Hs 00277039_m1 (CCND1), Hs 01078066_m1 (RB1), and Hs99999902_m1 (RPLP0). The reaction mix consisted of 10 μL of TaqMan master mix 2X, 1 μL of TaqMan gene expression mix, and 5 μL of 1/10 cDNA in a final volume of 20 μL. For SYBRGreen experiments, primer sequences were the following: RPLP0 forward: TGCATCAGTACCCCATTCTATCAT, RPLP0 reverse: AAGGTGTAATCCGTCTCCACAGA, p16INK4a forward: GGGGGCACCAGAGGCAGT, and p16INK4a reverse: GGTTGTGGCGGGGGCAGTT. The reaction mix consisted of 12.5 μL of SYBR Green PCR Master Mix (Applied Biosystems), 300 nmol/L forward and reverse primers and 5 μL of 1/10 cDNA in a final volume of 25 μL. Amplification of specific transcripts was confirmed by melting curve profiles generated at the end of the PCR program.

The PCR conditions were 2 min at 50°C and 10 min at 95°C, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. The comparative Ct (threshold cycle) method was used to achieve relative quantification of gene expression. The mRNA levels of genes of interest (R) were normalized to the mRNA levels of RPLP0: ΔCt = Ct_R - Ct_RPLP0. The relative amount of mRNA between controls (NSAT) and WDLPS/DDLPS was given by 2^−ΔΔCt where ΔΔCt = ΔCt _R of WDLPS/DDLPS - mean of ΔCt_R of controls (NSAT). The ΔCt values of the controls were homogenous.

Immunohistochemical staining was done on representative slides from each case, following the manufacturer's instructions with the following primary antibodies: CDK4 (clone DCS-31, Biosource International), P16 (CINtech P16-INK4A kit, clone E6H4, Dako North America, Inc.), and CYCLIND1 (clone SP4, Neomarkers). Appropriate positive and negative controls were used.

Descriptive statistics were used to show the distribution of variables in the population. Differences between groups were evaluated by χ² test or Fisher's exact test for categorical variables and t test for continuous variables.

The clinicopathologic characteristics of the patients are described in Table 1. In comparison with MDM2+/MDM2CDK4+ WDLPS/DDLPS, MDM2+/MDM2CDK4− WDLPS/DDLPS were more frequently low-grade lipoma-like lesions (64% versus 40%; P = 0.0039) and occurred in the majority of cases in the deep soft tissues of the extremities (71% versus 47%; P = 0.0045). They were very rarely located in the retroperitoneum (7%), whereas the MDM2+/CDK4+ tumors were retroperitoneal lesions in 35% of cases (P = 0.0002). Follow-up data were available for 56 patients with primary lesions at diagnosis (32 MDM2+/CDK4+, 24 MDM2+/CDK4−). The mean follow-up was 61 months for MDM2+/CDK4+ and 46 months MDM2+/CDK4− patients. The local recurrence rate was significantly higher in patients with MDM2+/MDM2CDK4+ tumors than in patients with MDM2+/MDM2CDK4− tumors (47% versus 12.5%; P = 0.0064). The distant recurrence rate was not statistically different between the two groups (12.5% versus 9.5%; P = 0.71). Ten deaths related to the disease were observed in the MDM2+/CDK4+ group and only one in the MDM2+/CDK4− group (31% versus 4%; P = 0.0116). In all the cases but three, deaths related to the disease were observed in patients with retroperitoneal DDLPS.

QRT-PCR analysis showed a strong overexpression of CDK4 in the MDM2+/CDK4+ cases analyzed whereas no significant overexpression was observed in the MDM2+/CDK4− cases (mean level compared with NSAT in MDM2+/CDK4+ WDLPS, 16.9; mean level compared with NSAT in MDM2+/CDK4− WDLPS, 1.76; P = 0,0005). Four cases of MDM2+/CDK4− WDLPS/DDLPS were also analyzed by immunohistochemistry for CDK4 expression. All the four cases were negative for CDK4. We also analyzed the expression status of CDK6 which encodes a protein that is also a kinase partner of CCND1. A trend for a higher expression of CDK6 was observed in MDM2+/CDK4− WDLPS/DDLPS (mean level compared with NSAT in MDM2+/CDK4− WDLPS, 7.6; range, 0.7-40.6) in comparison with MDM2+/CDK4+ WDLPS/DDLPS (mean level compared with NSAT in MDM2+/CDK4+ WDLPS/DDLPS, 1.9; range, 0.7-4.5). However, this trend in the series was only due to a single case of MDM2+/CDK4− WDLPS/DDLPS showing a strong overexpression of CDK6 (fold change 40.6). It was not statistically significant (P = 0.16).

Twenty-seven cases (10 MDM2+/CDK4+; 17 MDM2+/CDK4−) evaluated by FISH exhibited euploidy for chromosome 9 and CDKN2A. Heterozygous deletion of CDKN2A was found in only one case of WDLPS/DDLPS (MDM2+/CDK4+), and CDKN2A gain (3 to 5 copies) related to polysomy for chromosome 9 was observed in two cases (one MDM2+/CDK4+; one MDM2+/CDK4−).

QRT-PCR analysis showed overexpression of P16INK4a in all the 20 cases analyzed (13 MDM2+/CDK4+ cases including the case with heterozygous deletion of CDKN2A and 7 MDM2+/CDK4− cases; Fig. 2). The level of overexpression was significantly higher in cases with amplification of CDK4 (mean level compared with NSAT in MDM2+/CDK4+ WDLPS, 58.85; range: 6.5-135; mean level compared with NSAT in MDM2+/CDK4− WDLPS, 18.7; range: 4.1-60.9; P = 0.0043). Nineteen cases (7 MDM2+/CDK4+ WDLPS/DDLPS and 12 MDM2+/CDK4− WDLPS/DDLPS) were also analyzed by immunohistochemistry for P16 protein expression. A positive labeling for P16 was observed in all the cases but three (one MDM2+/CDK4+ WDLPS/DDLPS and two MDM2+/CDK4− WDLPS/DDLPS).

P14ARF was also overexpressed in all the cases analyzed (mean level compared with NSAT in MDM2+/CDK4+ WDLPS, 35.8; range, 10.8-84.9; mean level compared with NSAT in MDM2+/CDK4− WDLPS, 12.6; range, 6.7-22.9; P = 0.0064).

Twenty-nine cases (12 MDM2+/CDK4+, 17 MDM2+/CDK4−) evaluated by FISH exhibited disomy for CCND1. Gain (3 to 5 copies) of CCND1 was found in only one case of WDLPS/DDLPS (MDM2+/CDK4−). QRT-PCR analysis showed overexpression of CCND1 in 14 of 20 cases analyzed, independently of the CDK4 status (9 MDM2+/CDK4+, 6 MDM2+/CDK4−; 69% versus 86%; P = 0.42; mean level compared with NSAT in MDM2+/CDK4+ WDLPS/DDLPS, 7.3; range, 0.9-16.6; mean level compared with NSAT in MDM2+/CDK4− WDLPS/DDLPS, 4.9; range, 1.7-8.9; P = 0.13). A reduced expression of CCND1 was found in one case of MDM2+/CDK4+ WDLPS/DDLPS (fold change compared with NSAT, 0.33). Nineteen cases (7 MDM2+/CDK4+ WDLPS/DDLPS, 12 MDM2+/CDK4− WDLPS/DDLPS) were also analyzed by immunohistochemistry for CCND1 protein expression. A positive labeling was observed in 15 cases, a negative labeling in 3 cases (3 MDM2+/CDK4− WDLPS), and the result was not interpretable in 1 case.

Twenty-nine cases (12 MDM2+/CDK4+, 17 MDM2+/CDK4−) evaluated by FISH exhibited disomy for RB1. Gain (3 to 5 copies) of RB1 was found in only one case of WDLPS/DDLPS (MDM2+/CDK4−). QRT-PCR analysis showed a reduced expression (≥ 50%) of RB1 in 8 of the 20 cases analyzed independently of the CDK4 status (4 MDM2+/CDK4+, 4 MDM2+/CDK4−; 57% versus 31%; P = 0.25; mean level compared with NSAT in MDM2+/CDK4+ WDLPS/DDLPS, 0.32; range,0.3-2.9; mean level compared with NSAT in MDM2+/CDK4− WDLPS/DDLPS, 0.36; range, 0.18-1.7). In all the cases but one, reduced expression of RB1 and overexpression of CCND1 were mutually exclusive.

Three cases of MDM2+/CDK4− WDLPS/DDLPS were analyzed by array-CGH in order to detect specific genomic imbalances. The results are summarized in Table 2. Besides the 12q14-15 region, the sole recurrently amplified region was 1q21-23 in the three cases. No aberration of the CDKN2A, CCND1 and RB1 loci were detected in these three cases in accordance with the FISH results.

We recently reported that HMGA2 and MDM2 were always coamplified in WDLPS/DDLPS whereas CDK4 belonged to a distinct inconsistent amplicon (6). These results suggested that the amplification of CDK4 might not be as indispensable as the amplification of HMGA2 and MDM2 for WDLPS/DDLPS tumorigenesis.

In the present study, we found that WDLPS/DDLPS lacking CDK4 amplification were significantly associated with peculiar clinicopathologic features. Indeed, most of these lesions were low-grade lipoma-like WDLPS located in the limbs. Their rates of local recurrence and lethality were lower than those of MDM2+/CDK4+ lesions that were significantly more often retroperitoneal. Although a correlation between amplification/overexpression of CDK4 and a bad prognosis has been described in several tumor types (10–15), our results may also reflect the fact that free surgical margins are more easily obtained for lesions of the extremities than for retroperitoneal ones. Indeed, the disease-related mortality, typically associated with complications of multiple local recurrences, is higher for patients with retroperitoneal tumors than for those with peripheral tumors. Therefore, because our data are retrospective, the independent prognostic value of the CDK4 amplification remains to be confirmed in a prospective way.

The alternative occurrence of CDK4, p16INK4a, CCND1, or RB1 aberrations has been described in several tumor types (16–24). Indeed, the losses of p16INK4a or RB1 functions as well as overexpression of CCND1 represent alternative mechanisms for CDK4 amplification (25). In our study, however, we did not find any significant aberrations of the CDKN2A, CCND1, or RB1 genes that were confined to MDM2+/CDK4− WDLPS/DDLPS only. The CDKN2A gene encodes two tumor suppressor proteins, p16INK4a and p14ARF, by a mechanism of alternative splicing of the first exon. By using FISH, we did not find deletions of CDK2N2A. We observed a strong overexpression of P16INK4a and P14ARF in both MDM2+/CDK4+ and MDM2+/CDK4− WDLPS/DDLPS. Such an overexpression of tumor suppressor genes in malignant tumor cells is very intriguing. In this regard, it is interesting to note that P16INK4a overexpression was significantly higher in MDM2+/CDK4+ WDLPS/DDLPS than in MDM2+/CDK4− WDLPS/DDLPS. P16INK4a overexpression has been reported in several tumor types such as gliomas (26), myxoid liposarcomas (27), thyroid carcinomas (28), and cervical carcinomas (29). P16INK4a overexpression might be a secondary event to RB1 inactivation, because RB1 repressed P16INK4a at the transcriptional level (30, 31). In our series, RB1 expression was significantly reduced in 40% of the cases, independently of the CDK4 status. This suggests that in at least a subset of WDLPS/DDLPS, inactivation of RB1 and CDK4 overexpression may act as synergic or additional events to enhance the G₁-S transition and that the expressions of CDKN2A/P16 and RB are not always inversely related (32). An alternative explanation for P16INK4a overexpression in WDLPS/DDLPS might be related to the role of HMGA2. Although there are several lines of evidence indicating that HMGA2 should be considered an authentic oncogene (33), recent findings have shown that HMGA2 may also exhibit antioncogenic properties by playing a role in cellular senescence (34). Interestingly, Narita et al. have shown that high levels of HMGA2 expression induce P16INK4a expression in human diploid fibroblast and that the expression of MDM2 or CDK4 allowed such cells to bypass HMGA2-induced proliferation arrest (34). Therefore, the high level of P16INK4a in WDLPS/DDLPS might be related to the overexpression of HMGA2, the antiproliferative activity of which is circumvented by the consistent amplification of MDM2. Data about the respective expression of P14ARF and P16INK4a in tumor cells are scarce. The strong overexpression of both MDM2 and P14ARF in WDLPS/DDLPS suggests that p53 inactivation in WDLPS/DDLPS is the result of a complex regulation. Indeed, the physiologic function of p14ARF protein is to bind to the MDM2 protein and to inhibit its ubiquitin ligase activity, increasing the levels of the p53 protein (35). The expression of p14ARF is directly activated by E2F1 (36). Therefore, the overexpression of p14ARF in WDLPS/DDLPS may result from several oncogenic events that could deregulate the E2F1 activity, such as RB1 inactivation and CDK4 or CCND1 overexpression. We have shown that CCND1 is consistently overexpressed in WDLPS/DDLPS whatever the CDK4 status. Besides forming active complexes with CDK4 or CDK6 that promote cell cycle progression, CCND1 also functions as a transcriptional modulator by regulating the activity of several transcription factors and histone deacetylase (HDAC3; ref. 37). This latter activity that may play an important role in tumorigenesis is independent of CDK4. Although up-regulation of CCND1 protein in cancer frequently results from translational and/or posttranslational mechanisms (38), CCND1 mRNA is also overexpressed in several tumor types including nonadipose soft-tissue tumors (39, 40).

In order to obtain a global overview of genomic imbalances in MDM2+/CDK4− WDLPS/DDLPS, we analyzed three of these cases by array-CGH. As previously reported in MDM2+/CDK4+ WDLPS/DDLPS, two to five regions were coamplified with 12q14-15 in all the cases (41, 42). The sole recurrently region to be coamplified with 12q14-15 in MDM2+/CDK4− cases was 1q21-23. This aberration cannot be considered as specific to MDM2+/CDK4− WDLPS/DDLPS because it has also been described in MDM2+/CDK4+ WDLPS/DDLPS (41, 42). Although several genes located at 1q21-q23, including ATF6, DUSP12, COAS1, COAS2, COAS3, PRUNE, and FASLG, have previously been reported to be amplified in supernumerary ring and giant markers in WDLPS/DDLPS as well as in other sarcomas (42–45), the targets of the 1q21-23 amplicon, which are supposed to play an effective role in WDLPS/DDLPS tumorigenesis, are still unknown.

In summary, we have shown here that MDM2+/CDK4− WDLPS/DDLPS represent a distinct clinical subgroup with favorable prognostic features. Although CDKN2A deletion is a frequent event in human cancers, our results show that this aberration is not involved in the pathogenesis of WDLPS/DDLPS even in those lacking CDK4 amplification Moreover, we have shown a complex pattern of expression for CCND1, P16INK4a, P14ARF, and RB1 that is not dependent on the CDK4 status. These results underscore the complex coordinated regulation of the RB and p53 growth-control pathways in WDLPS/DDLPS. The knowledge of the way by which WDLPS/DDLPS cells regulate the mitogenic and antimitogenic signaling pathways during the cell cycle is of crucial importance for the design of new therapeutic strategies. Our results represent an important basis for further investigations with the aim of determining how the activities of the actors of the CCND1/CDK4/P16INK4a/RB1/E2F pathway are coordinated in WDLPS/DDLPS.

No potential conflicts of interest were disclosed.

We thank Dr. Marion Marty for her assistance with the immunohistochemistry experiments and Mrs. Stephanie Bonnafous for expert technical assistance.