To the authors' knowledge, little is known regarding the alterations of G1-S checkpoint and their significance in chordoma, a rare bone tumor. The authors investigated the clinicopathologic relevance of cell cycle abnormalities in chordoma.
The expression levels of p53, murine double minute 2 (MDM2), retinoblastoma protein (pRb), cyclin D1, p16INK4a, and p27Kip1 were investigated using immunohistochemical techniques; p53 mutations were studied by polymerase chain reaction (PCR)-single-strand conformation polymorphism, and mdm2 amplification was analyzed using real-time quantitative PCR. The results were compared with clinicopathologic parameters in 101 lesions.
Approximately 10–45% of primary tumors presented alterations of p53, MDM2, cyclin D1, and pRb proteins; most tumors lacked expression of p16INK4a and p27Kip1. Alterations of p53, MDM2, cyclin D1, and pRb proteins were found to have cooperative effects on both higher proliferative ability (MIB-1 labeling index [LI]) and increased nuclear pleomorphism, a previously described prognostic indicator for patients with chordoma. Multivariate analyses revealed that, among these alterations, p53 overexpression was the only independent factor for higher MIB-1 LI. At the genetic level, mdm2 gene amplification was detected in 15.4% of the lesions but did not correlate with MDM2 overexpression or other clinicopathologic parameters. No p53 mutations were detected in the current series. Survival analysis revealed that p53 overexpression, but no other cell cycle alterations, was associated with a reduced overall survival.
Chordoma is a rare malignant bone tumor that originates from the remnants of the embryonic notochord. Conventional chordoma, which is the most common subtype, is regarded as a slow-growing tumor with relatively low malignant potential.
Multiple genetic changes in oncogenes and tumor suppressor genes occur during oncogenesis.1 The genes that act at the G1-S checkpoint represent some of the most important targets in molecular tumorigenesis,2, 3 and the disruption of the checkpoints leads to an abnormal proliferation of tumor cells. P53 is known as a tumor suppressor gene, and mutations of the p53 gene constitute the most common alteration in human malignancies. The mouse double minute 2 (mdm2) gene is a negative regulator of p53 function, and its amplification or overexpression presumably results in a loss of p53 function.4 The normal function of cyclin D1, coupled with cyclin-dependent kinase 4 (CDK4), is to phosphorylate and subsequently inactivate retinoblastoma protein (pRb),5 which suppresses cell growth. CDK inhibitors also negatively regulate the action of cyclin/CDK complexes, preventing cell cycle progression.6
To our knowledge, aberrations in the G1-S checkpoint largely are unknown in chordoma. In the current study, we investigated both the p53 and Rb pathways in conventional chordoma to determine whether abnormalities at the G1 checkpoint play a role in oncogenesis, proliferation, and progression and whether they can be prognostic indicators.
MATERIALS AND METHODS
In total, 101 conventional chordomas (70 primary lesions and 31 recurrent lesions) were used for the light microscopic study: Eighty-eight lesions (57 patients) were obtained from the Department of Pathology at Nordstadt Medical Center, and 13 lesions (13 patients) were obtained from the Department of Pathology at Otto-von-Guericke University Magdeburg, Magdeburg, Germany. Seventy-one lesions (occurring in 46 patients) were located in the skull base, and 30 lesions (occurring in 24 patients) were located in nonskull base regions. Patients ranged in age from 11–80 years (median, 49 yrs), and patients with primary nonskull base chordomas (NSBCs) were significantly older (mean age, 57.0 yrs) than patients with primary skull base chordomas (SBCs) (mean age, 41.9 yrs) (P = 0.0001). All patients provided informed consent, and the study was approved by the local ethics committee.
Histologic sections obtained at biopsy or from surgically resected specimens routinely were stained with hematoxylin and eosin for diagnostic purposes. The studies were performed using sections from 10% formalin-fixed, paraffin-embedded tissues, highlighting the representative areas of the tumor.
Light Microscopic Study
We subclassified chordomas into two groups: the nonsolid subtype, which has a classic cord-like structure, and the solid subtype, which consists mainly of a diffuse proliferation of tumor cells. Nuclear pleomorphism, mitosis, and apoptosis were evaluated as positive or negative.
The expression of cell cycle-related proteins was detected using the following antibodies: anti-p53 (DO-1, 1:50 dilution [Oncogene Research Products, Cambridge, MA] and ab4060; rabbit polyclonal, 1:50 dilution [Abcam, Cambridge, MA]), anti-MDM2 (IF2, 1:100 dilution [Oncogene Research Products]), anti-cyclin D1 (M7155, 1:100 dilution [Dako Corporation, Hamburg, Germany]), anti-pRb (polyclonal, 1:15 dilution [BioGenex, San Ramon, CA]), anti-p16INK4a (JC8, 1:100 dilution [Quartett, Berlin, Germany]), anti-p27kip1 (SX53G8, 1:25 dilution [Dako Corporation]), and anti-Ki67 (MIB-1, 1:100 dilution [Dako Corporation]). The sections were retrieved antigenetically, and were treated with a primary antibody, followed by staining with an avidin-biotin-peroxidase complex (Immunotech, Marseille, France) or an alkaline phosphatase detection kit (Vector Laboratories, Burlingame, CA). Expression of p16INK4a and p27Kip1 was examined in 57 lesions along with the expression of other proteins and the MIB-1 labeling index (LI) in 101 lesions. In our previous study, some of the lesions were analyzed for MIB-1 LI.7
Evaluation of Staining
The MIB-1-positive cells were counted in well labeled areas, as determined by scanning at low magnification. The MIB-1 LI was determined as follows: 1) per 1000 tumor cells in the selected fields at × 400 magnification or 2) per all tumor cells in 10 fields using the same magnification if there were fewer than 1000 cells.
Sections that were stained immunohistochemically with p53, MDM2, cyclin D1, p16INK4a, and p27Kip1 were graded according to the ratio of positive cells: 0: no positive cells; 1: < 5% positive cells; 2: < 20% positive cells; and 3: > 20% positive cells. The grading was obtained as a consensus between two pathologists (T.N. and D.K.). No clinicopathologic information was available before the specimens were reviewed. The expression levels of p16INK4a and p27Kip1 were considered positive at Grade ≥ 1, and the overexpression of p53 and cyclin D1 was considered positive at Grade 2 or 3. In this study, MDM2 generally was expressed at lower levels compared with other malignancies, most likely due to tumor specificity. However, statistical analysis revealed a close association of MDM2 expression (Grade ≥ 1) with other variables, including nuclear pleomorphism and the MIB-1 LI; therefore, expression ≥ Grade 1 was treated as overexpression. Sections stained with pRb were graded according to the ratio of positive cells: 0: no positive cells; 1: < 50% positive cells; and 2: > 50% positive cells. Similar to the study by Cote et al.,8 who reported possible pRb overexpression in bladder carcinoma, and other investigators,9–11 who reported a correlation between increased pRb expression and cell proliferation in human malignancies, our initial statistical analysis revealed a correlation between an elevated pRb level (Grade 2) and the MIB-1 LI (Table 1). By contrast, lesions that lost pRb expression completely (Grade 0) did not present an increased MIB-1 LI. Consequently, Grade 2 pRb expression was considered altered expression.
Table 1. The MIB-1 Labeling Index According to Retinoblastoma Protein Expression in Chordoma
Genomic DNA was purified from 52 samples of chordoma, including 37 SBCs and 15 NSBCs. Five 10-μm paraffin block sections were cut and mounted on glass slides. The specimens were microdissected to obtain representative samples of tumor tissue. The standard proteinase K digestion/phenol-chloroform preparation was employed.
Polymerase Chain Reaction-Single Strand Conformation Polymorphism for p53 Gene Mutation
Exons 4–8 of the p53 gene were analyzed using the polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP) technique. A methodologic description and primer sequences were provided previously in detail.12 In brief, for SSCP analysis, an aliquot of the PCR product (7 μL), was denatured in formamide buffer at 95 °C for 5 minutes, chilled on ice, and loaded onto ultrathin mutation detection enhancement gels (AT Biochem, Malvern PA). Gels were stained using a modified silver staining protocol that was introduced by Goldman and Merril.13 PCR fragments that demonstrated mobility shifts of their single strands were sequenced directly on an automated fluorescence sequencer (ALF Express; Pharmacia Biotech) using the T7 sequenase protocol.
mdm2 Gene Amplification
LightCycler-based PCR assay
For amplicon detection, the LightCycler DNA Master Hybridization Probes Kit was used as described by the manufacturer (Roche Diagnostics, Mannheim, Germany). Briefly, along with the 2 primers, 2 different oligonucleotides hybridize to an internal sequence of mdm2 (GenBank U28935) or to the internal housekeeping phenylalaninhydroxylase gene (PAH; GenBank AF 404777). Primers and oligonucleotide probes were chosen using the TIB MOLBIOL computer program (Berlin, Germany; available at URL: www.TIBMOLBIOL.de/oligo_ag.html) to ensure their total gene specificity.
PAH primer sequences were sense (PAH-S; position 244–264) 5′-CCA TgC CAC TgA gAA CTC TCT-3′ and antisense (PAH-A; position 415–394) 5′-TCT TAA gCT gCT ggg TAT TGT C-3′, amplifying a 171-base pair fragment. The mdm2 primer sequences were sense (mdm2-ex2F; position 231–250) 5′-CTg TgT TCA gTg gCg ATT gg-3′and antisense (mdm2-ex3R; position 486–466) 5′-Tgg AAT CTg Tgg ggT TAC-3′, amplifying a 255-base pair fragment.
One of the 2 gene-specific hybridization probes was labeled at the 5′ end with LightCycler Red (LC Red) fluorophore and 3′-phosphorylated to avoid extension: LC Red 640 5′-ACA TgT CTg TAC CTA CTg Atg ggg gCT g p-3′ for mdm2 (position 440–466) and LC Red 640 5′-ATA CCT Cgg CCC TTC TCA gTT CgC p-3′ for PAH (position 337–360). The second probe was 3′-labeled with fluorescein: 5′-TTT TTC CTT gTA ggC AAA TgT gCA ATA CC X-3′ for mdm2 (position 410–438) and 5′-gTT TTg gTC TTA gAA CTT TgC TgC CAC X-3′ for PAH (position 309–335). Only in the case of adjacent hybridization of both probes did fluorescence energy transfer between both fluorophores occur. Fluorescence was monitored once each cycle at the end of the annealing phase, with increased fluorescence related to product accumulation.
The 20-μL PCR reaction for mdm2 and PAH amplification contained 200 ng of tumor DNA, the 2 hybridization probes (4 pmol of the fluorescein-labeled probe and 2 pmol of the LC Red 640 probe), 2-μL LC-DNA Master Hybridization Probe Mix, 25mM MgCl2 (2.4 μL for mdm2 and 1.4 μL for PAH), and 10 pmol each of the sense and antisense primers. PCR conditions for mdm2 were as follows: a first denaturation step (95 °C for 30 sec) followed by 45 cycles of denaturation (95 °C for 0 sec), annealing (62 °C for 45 sec), and extension (72 °C for 13 sec), and a final cooling step (40 °C for 30 sec). PCR conditions for PAH were as follows: a first denaturation step (95 °C for 30 sec) followed by 45 cycles of denaturation (95 °C for 0 sec), annealing (54 °C for 30 sec), and extension (72 °C for 7 sec), and a final cooling step (40 °C for 30 sec).
Standard curve creation
Serially diluted plasmid samples (from 107 to 104mdm2 copy numbers and from 108 to 105 PAH copy numbers) were used as external standards in each run. To ensure identical amplification efficiency for the creation of the plasmids, PCR fragments of mdm2 and PAH were cloned directly into pCR®2.1-TOPO® vector (TOPO™ Cloning Kit; Invitrogen, Carlsbad, CA). Finally, plasmids were sequenced using M13 forward primer according to the manufacturer's instructions (Taq DyeDeoxy™ Terminator Cycle Sequencing Kit; Applied Biosystems, Warrington, United Kingdom). Separate standard curves for mdm2 and PAH were included in each PCR run. In all samples, the correlation coefficient for the curves was − 1. Based on this linear correlation, the copy numbers were calculated from the standard curve and by crossing points of each sample.
Interpretation of LightCycler data
Relative mdm2 amplification was calculated by dividing the amount of mdm2 signals by the amount of the endogenous housekeeping gene PAH signals, × 10. This ratio is an mdm2 value normalized to amplification of the PAH gene. Amplification of the PAH housekeeping gene serves as a control for DNA quality, as a PCR amplification inhibitor, and as a reference for relative quantification. For positive and negative controls, we used an osteosarcoma cell line (OsACl; obtained from The American TypeCulture Collection, Rockville, MD) with amplified mdm2 gene and placenta cells without mdm2 amplification.
The average of ratios of all the runs for the placenta (mean ratio, 0.1 ± 0.15) and the positive cell line (mean ratio, 1.5 ± 0.24) was calculated. Approximately 20% of the distance between both mean ratios was defined as a cutoff value (ratio, 0.3) (i.e., ratios ≤ 0.3 were considered negative and ratios > 0.3 were considered positive). To exclude PCR contamination, we used sterile water instead of DNA for one capillary. Each run was performed twice.
The correlation between cell cycle alterations and patient age or MIB-1 LI was examined by the Mann–Whitney U test. The Fisher exact test was used for the association of cell cycle aberrations with other clinicopathologic parameters. Multivariate analysis based on the logistic model was then performed to search for predictors of higher MIB-1 LI and increased nuclear pleomorphism. For the multivariate analysis of MIB-1 LI, lesions were subdivided into higher or lower MIB-1 LI groups by taking the median as a cut-off value. Cooperative effects exerted by cell cycle abnormalities were estimated using a Spearman correlation rank test. The effects of cell cycle alterations on survival were tested using Kaplan–Meier survival plots and were analyzed using the log-rank test.
Incidence of Alterations of Cell Cycle-Related Proteins
With regard to immunostaining for p53, there was a correlation between the results achieved with DO-1 and Abcam (P < 0.0001) (Table 2); we adopted the former as representative data for p53 overexpression. Analysis of primary lesions revealed p53 overexpression in 14 SBCs (30.4%) and in 5 NSBCs (20.8%); MDM2 overexpression in 4 SBCs (8.7%) and in 4 NSBCs (16.7%); and cyclin D1 overexpression in 12 SBCs (26.1%) and in 12 NSBCs (50.0%) NSBCs. Possible pRb overexpression8 was detected in 17 SBCs (37.0%) and in 15 NSBCs (62.5%). Few chordomas expressed p27Kip1 or p16INK4a.
Alterations in cyclin D1 and pRb were seen more often in recurrent SBCs than in primary SBCs (P = 0.003, 0.082). These alterations occurred with greater frequency in primary NSBCs compared with primary SBCs (cyclin D1, P = 0.064; pRb, P = 0.048).
We succeeded in amplifying and screening mutations of exons 5, 7, and 8 of the p53 gene in 37 samples; mutations of exon 6 in 30 samples; and mutations of exon 4 in 16 samples. SSCP analysis and subsequent DNA sequencing revealed no mutations. A mdm2 amplification was detected in 8 samples (15.4%), including 6 SBCs and 2 NSBCs.
Correlation between Cell Cycle Aberrations
p53 overexpression was correlated significantly with the overexpression of MDM2 (P = 0.002) and cyclin D1 (P = 0.002). However, MDM2 overexpression was not associated with mdm2 amplification. There was a trend toward an association between the expression of cyclin D1 and pRb (P = 0.074), between the expression of cyclin D1 and p27Kip1 (P = 0.063), and between the expression of pRb and p27Kip1 (P = 0.063).
Effects of Cell Cycle Alterations on Clinicopathologic Parameters
Table 3 summarizes the correlation between cell cycle aberrations and MIB-1 LI or nuclear pleomorphism in primary lesions. The MIB-1 LI was correlated with overexpression of p53 (P < 0.0001), MDM2 (P = 0.0004), and cyclin D1 (P = 0.007). Among these alterations, overexpression of p53 (P = 0.027) was the only independent factor that indicated a higher MIB-1 LI. Nuclear pleomorphism was associated significantly with the overexpression of p53 (P = 0.007) and MDM2 (P = 0.007), although neither association was a statistically significant, independent factor (P = 0.055, 0.076) (Table 4).
Table 3. The MIB-1 Labeling Index According to Cell Cycle Alterations in Primary Chordoma
The accumulation of p53, MDM2, cyclin D1, and pRb was compared with the MIB-1 LI and nuclear pleomorphism in primary lesions (Table 5). Lesions with no alterations presented an average MIB-1 LI of 0.8, and lesions with 4 alterations revealed an index of 8.1. There was a correlation between the number of altered proteins and the MIB-1 LI (P < 0.0001). Similarly, there was a significant association between the number of altered proteins and increased nuclear pleomorphism (P = 0.010).
Table 5. Effects of Accumulation of Cell Cycle Alterations on the MIB-1 Labeling Index and Nuclear Pleomorphism in Primary Chordoma
No. of altered proteins (p53, MDM2, cyclin D1, and pRb)
In primary lesions, patients who had pRb alterations tended to be older (mean age, 51.3 yrs) than patients who did not have pRb alterations (mean age, 43.6 yrs; P = 0.054). However, no correlation was observed between the alterations in other proteins or mdm2 amplification and patient age, histologic subtype, mitosis, or apoptosis.
Regarding the correlation between cell cycle alterations and the MIB-1 LI in recurrent lesions, only p53 overexpression and MDM2 overexpression were associated with higher MIB-1 LI (P = 0.033 and P = 0.090, respectively). No correlation was detected between cell cycle alterations and nuclear pleomorphism in patients with recurrent lesions.
Treatment and Prognosis
Brief clinical information regarding treatment was available for 56 patients. The lesions initially were excised in these patients. Follow-up information was available for 40 patients who had primary lesions. Twenty-nine of these patients had SBC, and 11 patients had NSBC (range, 0–185 months). The 5-year survival rate was 38.9% for patients with p53 overexpression and 79.4% for patients without p53 overexpression (Fig. 1). The log-rank test revealed a significant difference in survival between these 2 groups (P = 0.027). Neither anatomic site, age, gender, aberrations in other proteins (MDM2, cyclin D1, pRb, and p27Kip1) nor mdm2 amplification was found to correlate with clinical outcome.
One of the most important checkpoints in oncogenesis is the restriction point in the late G1 stage. Abnormalities in the p53 and Rb pathways lead to cell cycle deregulation, thus contributing to tumorigenesis and progression in many human malignancies. However, to our knowledge, the significance of cell cycle aberrations largely remains unknown in chordoma. Bergh et al.14 found no p53 overexpression in 38 conventional chordomas. Conversely, Pallini et al.15 reported that approximately 40% of chordomas overexpressed p53 protein. However, the reason for this striking difference, which lay in the ratio of p53 overexpression, is unclear; it may have been caused by the different antibodies adopted. In the current study, we used 2 different antibodies and found p53 overexpression in approximately 30% of lesions, but we found no p53 mutations. This discrepancy between results at the gene and protein levels may be caused by possible p53 overexpression without p53 mutation; p53 overexpression can be caused by accumulation of p53 protein due to the hyperphosphorylation by ataxia telangiectasia mutated (ATM),16–18 or it can be induced by inactivation of MDM2 for certain reasons, including binding to p14 (p19)ARF19–21 or phosphorylation by ATM,22 leading to stabilization of p53. Furthermore, undetectable missense mutations of p53 gene may be present in the initial screening using PCR-SSCP analysis; p53 mutations can be seen outside exons 4–8, which were searched in the current study. Based on these points, further investigations will be needed to clarify the mechanism of p53 overexpression and the discrepancy between p53 overexpression and mutation in chordoma.
We detected mdm2 gene amplification in 15% of chordomas and MDM2 overexpression in 13%; however, there were no correlations between them. Such a discordance, as reported previously in some malignancies,23–25 can be caused by mechanisms that induce MDM2 overexpression without mdm2 amplification (e.g. mutation of p14ARF),26 overexpression of MDMX (an MDM2 homologue),27, 28 and enhanced translation or rearrangement of mdm2.25, 29 Conversely, monoclonal antibodies may fail to detect MDM2 protein because of the existence of its multiple isoforms.30
Cooperative effects exerted by p53 and MDM2 overexpression were reported by several investigators.23, 31, 32 Cordon-Cardo et al.23 indicated that this may be due to some activities in mutant p53-MDM2 complexes. Otherwise, the p53-independent role of MDM233, 34 may cooperate with p53 alteration. Conversely, cooperative effects of p53 and pRb alterations also were reported previously.35 Because alterations of cyclin D1 and pRb are based on their own genetic abnormalities and occur independently, they seem to work synergetically. In our current study, alterations of p53, MDM2, and cyclin D1 were correlated with the MIB-1 LI and with nuclear pleomorphism in primary chordomas. Furthermore, there were cooperative effects exerted by alterations of p53, MDM2, cyclin D1, and pRb on both the MIB-1 LI and nuclear pleomorphism. This fact indicates that the accumulation of cell cycle alterations leads to abnormal proliferation and progression in chordoma. It is interesting to note that our multivariate analyses indicated that, among these alterations, p53 overexpression was the only independent factor for higher proliferative ability and was a nearly significant factor for nuclear pleomorphism, suggesting the pivotal role of p53 overexpression in these 2 events.
There was a correlation between p53 and cyclin D1 overexpression in chordoma. This finding indicates that cyclin D1 may be overexpressed by mutated p53 through down-regulation of p21waf1. Correlations between other cell cycle alterations observed in the current study remain undetermined; however, such aberrations may occur in a hot spot of chromosomal instability or mutations. CDK inhibitors negatively regulate cell proliferation. Infrequent expression of p16INK4a and p27Kip1 may play an important role in the formation of chordoma.
In the current study, pRb alteration was associated with patient age, although the association did not reach statistical significance. In addition, alterations of cyclin D1 and pRb were observed more frequently in primary NSBC than in primary SBC and occurred more often in recurrent SBC than in primary SBC. In general, patients with NSBC were older than patients with SBC, which was confirmed in our series. Furthermore, considering the fact that many NSBCs develop as giant masses and that patients with NSBC often show symptoms for several years before diagnosis, the interval between the beginning of tumorigenesis and the date of sampling is expected to be longer in patients with NSBC than in patients with SBC. Consequently, abnormalities of cyclin D1 and pRb are considered late tumorigenic events in chordoma.
Abnormalities of the G1-S checkpoint were associated with prognosis in some bone36, 37 and soft tissue31, 38 sarcomas. In particular, Taubert et al.38 reported a prognostic relevance of p53 alterations in sarcomas. In the current series, overexpression of p53 and MDM2 was associated with increased nuclear pleomorphism, which was an unfavorable prognostic indicator in our previous study.7 Furthermore, although patients with p53 overexpression,accounted for only a subset of the population, they were found to have a shorter survival compared with patients who had no p53 overexpression. These observations, based on the preliminary data, indicate that alterations of the p53-pathway (in particular, p53 overexpression) are associated with a worse prognosis in patients with chordoma.
In conclusion, abnormalities of G1 checkpoint proteins were considered to contribute cooperatively to proliferation and progression in chordoma. Alterations in the Rb pathway may be late events in chordoma. Although to our knowledge the mechanism of p53 overexpression remains unclear in chordoma, our preliminary data indicate that increased immunohistochemical staining of p53 protein is associated with a worse prognosis and decreased survival in patients with chordoma.
The authors thank Ms. Carola Kuegler, Ms. Claudia Miethke, Ms. Nadine Wiest, Ms. Simone Staeck, Ms. Hiltraud Scharfenort, and Ms. Antje Schinlauer (Department of Pathology, Magdeburg University) for their technical assistance; Dr. Tomoo Iwakuma (Department of Molecular Genetics, The University of Texas M. D. Anderson Cancer Center, Houston, TX) for helpful discussions; and Mr. Bernd Wuesthoff for editing the article.