mdm2 mRNA level is a prognostic factor in soft tissue sarcoma.

BACKGROUND
The oncogenic properties of murine double minute-2 (mdm2) protein over-expression, which mostly results from the interaction with the tumor suppressor p53, are well described and their negative impacts on the prognosis of affected patients is well characterized. However, clinical relevance of mdm2 mRNA expression is poorly investigated.


MATERIALS AND METHODS
In this study, 65 soft tissue sarcoma (STS) samples were analyzed for mdm2 mRNA expression by a quantitative reverse transcription polymerase chain reaction (RT-PCR) approach using available validated ready-to-use assays based on the TaqMan technology (PE Applied Biosystems, Weiterstadt, Germany). Mdm2 data were correlated to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression calculated from the same sample.


RESULTS
For patients with a mdm2/GAPDH mRNA ratio below 50 zmol/amol the survival was strikingly reduced in comparison to patients with a ratio of > or =50 (p = 0.0241). Multivariate Cox analysis showed that the difference in prognosis for patients with tumor stage 2 and 3 became even more pronounced between patients with a ratio of <50 zmol/amol and patients with a ratio of > or =50 (p = 0.0041; RR = 5.6). To test if the group with an mdm2 mRNA expression > or =50 is homogenous concerning the prognosis, the group was divided into three subgroups with values of 50 to <100, 100 to <500 and > or =500. The subgroup with values of 100 to <500 showed the best prognosis (p = 0.0164); whereas, the one with values of 50 to <100 showed the worst prognosis in this group and, in between, was the one with values of > or =500. After omitting patients of stage 1 and 4, the subgroup with values of 100 to <500 showed an even more striking best prognosis (p = 0.0015); the other subgroups remained in the same sequence. The risk of tumor-related death over 5 years was most conspicuous in patients with mdm2 mRNA expression <50 than in those with ratios of 100 to <500 displaying a 13.3-fold higher risk. In a comparison between mdm2 mRNA levels and P53 protein expression or p53 mutational status, no relationship was found.


CONCLUSIONS
In our study, the mdm2 mRNA level appears to be an independent prognostic factor for STS patients, marking its role in STS genesis and as a potential factor for gene therapeutical approaches.


Introduction
The mdm2 gene was originally isolated by virtue of its amplification in a tumorigenic derivative of NIH-3T3 cells (1,2). Beside localiza-Abstract Background: The oncogenic properties of murine double minute-2 (mdm2) protein over-expression, which mostly results from the interaction with the tumor suppressor p53, are well described and their negative impacts on the prognosis of affected patients is well characterized. However, clinical relevance of mdm2 mRNA expression is poorly investigated. Materials and Methods: In this study, 65 soft tissue sarcoma (STS) samples were analyzed for mdm2 mRNA expression by a quantitative reverse transcription polymerase chain reaction (RT-PCR) approach using available validated ready-to-use assays based on the TaqMan® technology (PE Applied Biosystems, Weiterstadt, Germany). Mdm2 data were correlated to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression calculated from the same sample. Results: For patients with a mdm2/GAPDH mRNA ratio below 50 zmol/amol the survival was strikingly reduced in comparison to patients with a ratio of у50 (p ϭ 0.0241). Multivariate Cox analysis showed that the difference in prognosis for patients with tumor stage 2 and 3 became even more pronounced between patients with a ratio of Ͻ50 zmol/amol and patients with a ratio of у50 (p ϭ 0.0041; RR ϭ 5.6). To test if the group with an mdm2 mRNA expression у50 is homogenous concerning the prognosis, the group was divided into three subgroups with values of 50 to Ͻ100, 100 to Ͻ500 and у500. The subgroup with values of 100 to Ͻ500 showed the best prognosis (p ϭ 0.0164); whereas, the one with values of 50 to Ͻ100 showed the worst prognosis in this group and, in between, was the one with values of у500. After omitting patients of stage 1 and 4, the subgroup with values of 100 to Ͻ500 showed an even more striking best prognosis (p ϭ 0.0015); the other subgroups remained in the same sequence. The risk of tumor-related death over 5 years was most conspicuous in patients with mdm2 mRNA expression Ͻ50 than in those with ratios of 100 to Ͻ500 displaying a 13.3-fold higher risk. In a comparison between mdm2 mRNA levels and P53 protein expression or p53 mutational status, no relationship was found. Conclusions: In our study, the mdm2 mRNA level appears to be an independent prognostic factor for STS patients, marking its role in STS genesis and as a potential factor for gene therapeutical approaches. Molecular Medicine 6(1): [50][51][52][53][54][55][56][57][58][59]2000 Molecular Medicine tion of the human mdm2 gene to chromosome 12q13-14, mdm2 gene amplifications were detected in a variety of human malignancies and particularly in sarcomas (3-8). The oncogenic properties of the human mdm2 gene product have been attributed mostly to its interaction with the tumor suppressor gene p53 (9). Mdm2 promotes inactivation of p53 by its rapid degradation, inhibition of p53-mediated apoptosis/ growth arrest and by masking the transactivation domain of p53, thus, impairing the interaction with the transcriptional machinery (3, [10][11][12][13][14][15][16][17][18][19]. The effect of mdm2 can be modulated by other proteins as RB-1 or p19 ARF (20,21). On the other hand, mdm2 can act independently from p53, for example, it interacts with transcription factors of the E2F-family and the human TATA-binding protein (22,23), inhibits Rb growth regulatory function (24), contributes to tumorigenesis in p53 Ϫ/Ϫ mammary epithelial cells (25), mediates TGF-ͱ1 resistance (26), and inhibits the G 0 /G 1 -S-phase transition in normal human diploid cells (27). In numerous tumor cell lines and malignant tumors, particularly sarcomas, the human MDM2 protein overexpression is a characteristic feature (28) which can be correlated to poor prognosis (7,29). However, overexpression may occur independently of gene amplification and might correlate with an increased transcription and/ or a different translation efficiency of human mdm2 transcripts (30)(31)(32)(33)(34)(35). Occurrence of different mRNA transcript levels and splice products in malignant tumors is well described (31,36,37), but only recently was an impact on tumor behavior and prognosis uncovered (38,39). Comparably less is known about mdm2-mRNA expression in sarcomas, beside correlating it to mdm2 gene amplification (7, 35,40). Therefore, the aim of this study was to investigate the level of human mdm2 transcription, if there is a relationship to p53 mutational state or P53 protein expression, and if the mdm2 mRNA level has a prognostic impact for soft tissue sarcoma (STS) patients.

Automated mdm2 and GAPDH Transcript Analysis by Quantitative Fluorescence PCR
Two commercially available, validated polymerase chain reaction (PCR) assays for quanti-tation of murine double minute-2 (mdm2) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene transcripts (ZeptoQuant Nukleinsäure-Diagnostika, Leipzig, Germany) were used in our laboratory (41). The mdm2 assay detects all but the shortest splice variant of mdm2 mRNA, previously described by Sigalas et al. (38). Briefly, conventional 96-well bases were loaded with an 8-well ready-to-use standard DNA strip either coated with eight different amounts of reference-DNA for quantitation mdm2 or GAPDH transcripts, respectively, foreward and reverse primer, and the TaqMan® probe. The double labeled probes were either 5'-labeled with the fluorescent reporter dye 6carboxyfluorescein (FAM) for detection of mdm2 or 2,7-dimethoxy-4,5-dichloro-6-carboxyfluorescein (JOE) for GAPDH, and the common 3'fluorescent quencher dye 6-carboxytetramethylrhodamine (TAMRA) in order to generate the respective mdm2 or GAPDH standard reference curves for each run. The remaining free base positions were loaded with the required number of sample tubes containing just the respective TaqMan® oligonucleotide sets. Reaction premixes containing PCR buffer, which was supplemented with the passive fluorescence dye 6-carboxy-tetramethyl-rhodamin (ROX), dNTPs, and 1.25 U of AmpliTaq® "GOLD" (PE Applied Biosystems) were assembled according to the manufacturers instructions. Aliquots of the mixes were added to each reaction tube by using a BIOMEK® 2000 laboratory automation workstation (Beckman Instruments Inc., Fullerton, CA, U.S.A.). Sample and standard reactions (final volume of 50 Ȑl) differed only by the addition of 2 Ȑl aliquots of the analyzed cDNA sample. PCR amplification and detection was performed with an ABI PRISM® 7700 Sequence Detection System (PE Applied Biosystems). Sample cDNA amounts were calculated from data obtained with the simultaneously amplified reference DNA strips. Mdm-2 data were correlated to GAPDH cDNA (zeptomoles [zmol, 10 Ϫ21 ] mdm2 mRNA per attomole [amol, 10 Ϫ18 ] GAPDH mRNA) calculated from the same cDNA sample.
p53 Mutational Analysis The tumor samples were examined for mutations in the p53 gene by nonradioactive PCR-SSCP-sequencing. DNA was isolated from frozen tumor samples and the p53 gene (exons 4 to 9) was amplified in PCR reactions as de-scribed previously (42). In an SSCP-pre-screen for mutations, PCR products were investigated in 6 or 10% (PAA)-ready-made gels (Novex, Heidelberg, Germany) for abnormal single strand DNA shifts and striking cases were cycle-sequenced on an ABI 373 using the Dye Terminator Kit (PE Applied Biosystems).

Statistical Analysis
Prognostic analysis began with a descriptive presentation of the cumulative survival functions according to the Kaplan-Meier method and a univariate evaluation of prognostic differences with a log-rank test. The Cox regression model, which was used to estimate the effect of mdm2 mRNA expression on prognosis, was adjusted to stage, localization and type of surgical resection. A probability (p) of Ͻ 0.05 was defined as significant and a relative risk (RR) was calculated. The statistic analyses were carried out using software from SPSS Inc. (SPSS 8.0). The cut points for mdm2 mRNA expression levels represent prognostic relevant thresholds, which were set after continously sliding cut points in increment values of 10.

mdm2 mRNA Expression
We investigated 65 STS samples for human mdm2 mRNA expression using the ABI PRISM® 7700 Sequence Detection System (PE Applied Biosystems) and quantitative ready-touse fluorescence PCR assays. The assays, which were by automation using mostly robotic workstations supporting the 96-well format, allowed precise and reproducible data recovery combined with excellent sensitivity at low zeptomole detection levels. Dynamic ranges of usually 4-6 logs without prior sample dilution (up to Ͼ10 6 copies per tube) were easily achieved. The slopes of the individual reference curves were very close to the mean slopes (SD Ϯ 2-5%), at simultaneous correlation coefficients of the calculated linear fits usually Ͼ0.99 (Fig. 1).
The correspondence of data with an inhouse competitive GAPDH PCR protocol was found to be 99% (41). Human mdm2 cDNA amounts were calculated from an external reference curve obtained with eight known amounts of mdm-2 standard reference DNA, calibrated as described earlier (43). Data were normalized to the number of GAPDH transcripts measured in the same cDNA sample, which were reported to be constant in several heterogenous tumors, tissues and cell lines (44)(45)(46)(47). Thus, the calculated ratios of both cDNAs reflected the initial ratios of the mRNAs in the sample.
At first, a threshold for the ratio of mdm2/ GAPDH-mRNA was determined after cut point sliding. A value of 50 (zmol mdm2/amol GAPDH mRNA) as the most relevant cut point. In 15 STS samples, the ratio mdm2/GAPDH-mRNA was below 50; whereas, the other 50 samples showed values у50, with a maximum of 7571. In a Kaplan-Meier curve, patients with mdm2 mRNA values below 50 showed a highly decreased average survival time (18 months), compared with the patients with mdm2 mRNA values у50 (Ͼ60 months). In an univariate log rank test, a significantly better survival for patients exhibiting mdm2 expression values у50 (p ϭ 0 .0241) was found.
For a multivariate analysis, a Cox regression model was applied to examine whether mdm2 mRNA expression was a prognostic factor independent of other known risk factors, such as tumor stage, kind of tumor resection and localization. As shown in Table 1, mdm2 mRNA expression (Ͻ50 versus у50) did not seem to be an independent prognostic factor (p ϭ 0.1355). But after excluding patients with tumors of stage 1 and 4, the prognostic relevance became clearly visible (p ϭ 0.0041, RRϭ5.6) ( Table 1, Fig. 2). This can be reasoned by the fact that survival of both patient groups might be rather independent of mdm2 mRNA levels (i.e. patients with stage 1 tumors had mostly a good survival; 4 out of 5 patients with a ratio у50 survived); whereas, for patients with stage 4 tumors, poor prognosis is determined by the occurring manifested metastases. Patients with stage 2 and 3 tumors are of especially high interest in clinical practice, because their tumors may have a similar histo-morphological appearance; whereas, tumor behavior and prognosis can differ dramatically. When primary tumors and relapses were investigated, no difference in the mdm2 mRNA content was found (data not shown).
Next, we investigated if elevated human mdm2 transcript levels may be generally correlated with better survival or if this patient group might consist of several subpopulations with a diverging prognosis. When the group of patients with a mdm2/GAPDH-mRNA ratio у50 was subdivided into three groups (50 to Ͻ100; 100 to Ͻ500; and у500; Table 2), the group with a ratio of 100 to Ͻ500 was distinguished by significantly better survival, compared with the reference group showing a ratio of Ͻ50 (p ϭ 0.0164). The remaining two groups still showed a better survival than the reference group, but no statistical significance (p ϭ 0.1981 and p ϭ 0.2003) was found. Again, excluding patients with stage 1 and 4 tumors, the group with a ratio mdm2/GAPDH-mRNA of 100 to Ͻ500 showed the best survival (76% of patients alive) at an increased significance level (p ϭ 0.0015; Fig. 3), compared with the reference group (42% of patients alive). Most striking was that the risk of tumor-related death was 13.3-fold increased in the patients with a mdm2 mRNA expression Ͻ50, comparison with those with a ratio of 100 to Ͻ500 (Table 2).

P53 and MDM2 Protein Expression
P53 protein expression was detected by Western blots analysis and normalized to actin expression. A gross differentiation between four P53 expression groups divided into zero, modest, marked or strong expression was made. The group with zero and with strong expression showed the highest number of cases, which was not surprising since tumors may have lost p53 or may show abnormally high P53 protein expression levels. However, no relationship between the mdm2 RNA level and the detectable P53 protein expression, independent of considering tumors stage was observed (Table 3). In previous studies, a significant relationship between MDM2 protein expression, detectable immunohistochemistry, and a poor prognosis was found (29). However, the MDM2 expression, detected immunohistochemically or in Western blots, did not correlate with either the mdm2 mRNA level or survival (data not shown).

p53 Mutational Analysis
Fifty-five out of 65 STS samples were investigated for p53 mutations (exons 4-9). In eight  2 and 3; n ϭ 52). The threshold for the ratio of mdm2/GAPDH mRNA was set at a value of 50, after threshold sliding in Cox regression analyses. Curves for patients with an mdm2 mRNA level below 50 (Ϫ) and ജ50 (…) are significantly different (p ϭ 0.0041). A relative risk of 5.6 of tumor-related death is associated with a mdm2 mRNA level below 50, compared with a higher mdm2 mRNA expression level.
cases, a p53 mutation was detected as previously described (48). Independent of the type of mutation, all patients but one showed a higher mdm2 mRNA expression (Ͼ50), four cases were in the range of 50 to Ͻ100, and three in the range of 100 to Ͻ500. Concerning survival in the group with a mdm2 mRNA expression of 50 to Ͻ100, two out of four patients died and, in the group with 100 to Ͻ500, one out of three patients died (Table 4). This followed the observed trend that patients with mdm2 mRNA expression 100 to Ͻ500 had an increased overall survival. But we suggest, that rather the type of p53 mutation, i.e., patients with non-frameshift mutations had a poorer survival than those with frame-shift mutations, seems to correlate with prognosis (48).

Discussion
Our results show that human mdm2 mRNA expression, measured with an automated quantitative RT-PCR protocol, is an independent molecular prognostic factor for STS patients. In a Kaplan Meier test, the survival rate of patients with a mdm2/GAPDH-mRNA ratio below 50 was strikingly reduced (average, 18 months survival time), compared with patients with a ratio у50 (Ͼ60 months average survival time) (p ϭ 0.0241). In a multivariate Cox model confined to patients with tumor stage 2 and 3, the difference in prognosis became even more pronounced between patients with a ratio of Ͻ50 and patients with a ratio of у50 (p ϭ 0.0041; RR ϭ 5.6). The result that diminished, rather than increased, human mdm2 mRNA expression correlates with a poor survival is somewhat surpris- The group of patients with a mdm2/GAPDH-mRNA ratio у50 was further subdivided into three groups, 50 to Ͻ100; 100 to Ͻ500 and у500. The group with a ratio of 100 to Ͻ500 showed the best prognosis, compared with the reference group characterized by a ratio of Ͻ50 (p ϭ 0.0015). The other two groups still showed a better survival than the reference group (p ϭ 0.1981 and p ϭ 0.2003, respectively). The relative risk of tumor-related death was 13.3 times higher in the patients with a mdm2 mRNA expression Ͻ50, compared with those with a ratio of 100 to Ͻ500.  (49). Comparable with our data, a higher mdm2 mRNA expression was associated with a better prognosis in patients with ovarial carcinomas. On the other hand, there are several reports that describe significantly higher levels of mdm2 mRNA expression that are associated with an unfavourable prognosis in (AML) patients (30), with a subset of aggressive breast tumors (50) and with high invasiveness of hepatocellular carcinomas (51). Furthermore, it is well known that many malignant tumors, and particularly sarcomas, are characterized by elevated levels of mdm2 protein (28), which can be associated with a poor survival for sarcoma patients (7,29,52). However, our data suggest that a protein overexpression does not neccessarily correlate with increased mdm2 mRNA expression. Therefore, it would be of high interest to find out if increased MDM2 protein expression is based on increased mRNA stabilization rather than on upregulation of transcription. There are several reports showing RNA stabilization as a major reason for increased mRNA levels, for example the heat shock protein hsp 70, insulinlike growth factor binding protein IGFB-BP3 and waf-1 (53)(54)(55). Noticeably, the latter two genes are like mdm2 target genes of p53 and, for waf-1, mRNA stabilization, can be p53dependent and p53-independent (53). We did not find any relationship between mdm2 mRNA level and P53 protein expression. However, the relationship between P53 protein level and mdm2 mRNA expression is controversial. On one hand, there are several reports showing no relationship (6,37,56,57). On the other hand, some studies concluded that tumors with p53 mutations may have an decreased mdm2 mRNA expression or that tumors without p53 mutations were characterized by higher mdm2 mRNA expression (40,51). In our study, seven out of eight tumor samples with  none  17  17  19  11  15  17  18  1  moderate  11  11  12  11  11  11  12  1  marked  13  13  15  14  13  12  13  3  strong  14  13  1 0  14  13  11  18  3  total  15  14  26  10  12  11 21 8 p53 mutations did not show a striking decrease of mdm2 mRNA levels (50 to Ͻ100 and 100 to Ͻ500; Table 4). One possible explanation could be that one remaining wild type p53 allele is capable of maintaining mdm2 gene activity (58). Alternatively, mdm2 mRNA expression can occur independently from p53 (26,37,59). Furthermore, tumor cell lines with high levels of transcriptionally inactive p53 only might be unable to induce the expression of the MDM2 protein (60). Recently, a correlation between alternatively spliced mdm2 transcripts (missing the p53-binding site) and stabilized wild type p53 protein could be shown in glioblastoma cells, but its significance still remains unclear (61). Although we could detect short alternatively spliced forms of the mdm2 mRNA in STS, the correlation between the occurrence of spliced transcripts and the p53 gene status (wild-type/ mutant) and the P53 protein expression level remains unclear (Bartel et al., submitted). However, the relationship between mdm2 mRNA level and P53 protein expression is not yet clear and needs further investigation (28). A comparable negative result was obtained by the attempt to correlate MDM2 protein expression previously evaluated by immunohistochemistry (52) or by Western blot analysis (data not shown) to mdm2 mRNA expression level. Furthermore, neither P53 nor MDM2 expression could be correlated to prognosis of the investigated 65 STS patients in a multivariate Cox regression analysis, which might be due to the relatively small number of patients. Nevertheless, it seems neccessary to stress here that only alterations in the gene, transcript or/and translational level may result in an oncogenic potential of mdm2. mdm2 normally acts as a cell cycle regulator (27), co-transcriptional factor (22,23) and a cellular regulator of several P53 protein functions (16,62,63). We suggest that moderately increased mdm2 transcript levels (100 to Ͻ500) might be related to a normal mdm2 function rather than a reduced (Ͻ50) or an extreme high level (Ͼ500). In summary, the human mdm2 RNA level appears as an independent prognostic factor for STS patients, especially the potential role of mdm2 alterations in STS tumorigenesis, in STS diagnosis and as an target for gene therapeutical approaches.