Correlation of TP53 and MDM2 genotypes with response to therapy in sarcoma

Authors


Abstract

BACKGROUND:

Relatively few sarcomas harbor TP53 (tumor protein p53) mutations, but in many cases, amplification of MDM2 (murine double minute 2) effectively inactivate p53. The p53 pathway activity can also be affected by normal genetic variation.

METHODS:

The mutation status of TP53 and expression of MDM2, TP53, and their genetic variants SNP309 and R72P (Arg72Pro) were investigated in 125 sarcoma patient samples and 18 sarcoma cell lines. Association of the different genotypes and gene aberrations with chemotherapy response and survival, as well as response to MDM2 antagonists in vitro was evaluated.

RESULTS:

Twenty-two percent of the tumors had mutant TP53 and 20% MDM2 gene amplification. Patients with wild-type TP53 (TP53Wt) tumors had improved survival (P < .001) and TP53Wt was an independent prognostic factor (hazard ratio = 0.41; 95% confidence interval = 0.23-0.74; P = .03). Interestingly, there was a trend toward longer time to progression after chemotherapy for tumors with the apoptosis-prone p53 variant R72 (P = .07), which was strongest with doxorubicin/ifosfamide-based regimens (P = .01). Liposarcomas had low R72 frequency (33% versus 56%), but increased levels of MDM2 and MDM4 (51% and 11%, P < .001). MDM2 overexpression on a TP53Wt background predicted better response to MDM2 antagonist Nutlin-3a, irrespective of R72P or SNP309 status.

CONCLUSIONS:

Improved survival after chemotherapy was found in patients with TP53Wt tumors harboring the R72 variant. MDM2 overexpression in TP53Wt tumors predicted good response to MDM2 antagonists, irrespective of R72P or SNP309 status. Thus, detailed TP53 and MDM2 genotype analyses prior to systemic therapy are recommended. Cancer 2013. © 2012 American Cancer Society.

In addition to direct mutations, p53 can be inactivated by abrogation of other components of the signaling pathway or effector molecules that convey p53 downstream effects.1 In sarcomas, TP53 (tumor protein p53) mutations are infrequent,2 but in the wild-type TP53 (TP53Wt) tumors, p53 may be inactivated by amplification of the mouse double minute 2 (MDM2) gene (MDM2Ampl), in particular, in well-differentiated liposarcomas (WDLS).3-5 Complementary mechanisms have been suggested,6, 7 but it is still unclear whether and how the pathway is inactivated in the remaining tumors.

In addition to rare predisposition syndromes, such as Li-Fraumeni,8 more frequent normal genetic variation of TP53 and MDM2 is reported to influence cancer risk, progression and overall survival.9-12 A polymorphism replacing arginine (R72) with proline (P72) at position 72 in p53 (codon72 polymorphism, referred to as R72P) has been reported to affect p53 function.13 It has been shown that R72 increases the affinity of p53 to MDM2, affects its export from the nucleus, and provides increased apoptotic potential.14-16 The MDM2 polymorphism, SNP309, changes a base from T to G at nucleotide position 309 in the MDM2 promoter. This polymorphism is located in a binding site for the Sp1 transcription factor, and a G gives increased expression level of MDM2, hence reduced apoptotic response.17 The SNP309G may also give higher sensitivity to MDM2 antagonists such as Nutlin-3a.18 The reported frequencies of these polymorphisms vary (eg, 23%-67% for P72,19 with 58% in Norwegians,20 and 8%-58% for SNP309G, with 55% in Norwegians21). Here, we investigate whether these normal variants and somatic aberrations of TP53 or MDM2 in tumor material may affect disease outcome, and also how they may relate to Nutlin-3a responses in vitro.

MATERIALS AND METHODS

Tissue Samples

A total of 174 sarcoma patient samples were collected between 1980 and 2008. At time of primary diagnosis, all patients signed informed consent to use surplus diagnostic or surgical material in future investigations. The project was approved by the ethical committee of Southern Norway (Projects S-06133 and S-06134). Samples were collected immediately after surgery, cut into small pieces, frozen in liquid nitrogen, and stored at −80°C until use. Only malignant tumors confirmed by an experienced pathologist according to World Health Organization guidelines22 were used, categorized in Table 1. The clinical information was retrieved from the hospital's quality database (http://medinfo.net/medinsight/), and 125 untreated tumor samples were included in the analysis.

Table 1. Clinical and Treatment Characteristics of Study Population
Characteristicsn (N = 125)
  • Localization: Axial (deep trunk/abdomen/uterus) or extremity (arm/thigh/leg/thoracic wall). Tumor depth: superficial (ie, subcutaneous) or deep (ie, intramuscular/extramuscular/intra-abdominal/extracompartmental [if OS]). GIST: gastrointestinal stromal tumors. LMS: leiomyosarcoma. LS: liposarcoma (well-differentiated [n = 16]/myxoid [n = 8]/pleomorphic (n = 2)/dedifferentiated (n = 0)/unknown/other (n = 9)), MFH: malignant fibrous histiocytoma (includes myxofibrosarcoma). MPNST: malignant peripheral nerve sheath tumor. OS: osteosarcoma (includes one paraosteal). SS: synovial sarcoma. Others: epitheloid sarcoma/rhabdomyosarcoma/PNET/fibrous sarcoma/angiosarcoma/chondrosarcoma/unclassified.

  • a

    Stage system.45,46 Stage3: STS only. Stage3 OS (metastasis) grouped together with Stage 4 STS (distant spread). R0: wide. R1: marginal. R2: intralesional/gross tumor left.

  • b

    Chemotherapy received at least once during the entire treatment period (doxorubicin/ifosfamide/etoposide/cisplatin/methotrexate/vincristine/dacarbazine/cyclofosfamide/trofosfamide single drugs or combinations according to Scandinavian Sarcoma Group protocols).

Median age at diagnosis (range), y58 (8-87)
SexMale65
Female60
LocalizationAxial61
Extremities62
Unknown2
Primary tumor size0-10 cm38
>10 cm77
Unknown10
HistologyGIST5
LMS8
LS35
MFH31
MPNST8
OS17
SS8
Other13
GradeLow27
High92
DepthSuperficial30
Deep91
Unknown4
Stagea11
250
347
418
Unknown9
Resection of primary tumorR031
R154
R231
Unknown9
Local relapseYes34
No89
Unknown2
Median interval to first progression (range), m12 (0-345)
Median number of metastatic events (range)1 (0-11)
Radiotherapy primary tumorYes30
No94
Unknown1
Chemotherapy primary tumorYes17
No107
Unknown1
ChemotherapybYes43
No82

Cell Lines

We used 18 sarcoma cell lines; 9 osteosarcomas (OHS, MHM, KPD, IOR/OS14, IOR/OS15, IOR/MOS, U2OS [HTB96], SaOS-2 [HTB85], and OSA [SJSA-1 or CRL2098]) described in Ottaviano et al,23 4 liposarcomas (T449, T778, Fu-ddl-S1, and SW872 [HTB92]), and 1 rhabdomyosarcoma (RMS13) described in Müller et al.24 In addition, SA-4 (CRL7938) was obtained from ATCC (Manassas, Va), WLS-160 was given by Dr. Boven (Academic Hospital Vrije Universiteit, the Netherlands), and RMS4 and RMS28 by Dr. Look and Dr. Houghton (St Jude's Hospital, Memphis, Tenn). All cell lines were grown as described in Ohnstad et al.25

DNA Quantification by Real-Time Polymerase Chain Reaction

Frozen tissue was pulverized in liquid nitrogen and DNA isolated by phenol chloroform. Copy number analysis of MDM2 and MDM4 (also known as MDMX) was done by quantitative real-time polymerase chain reaction (PCR) using FastStartSYBRGreen (Roche). Oligonucleotide primers used were MDM2-Forward (5′-AAGCCAAACTG GAAAACTCAACAC-3′), MDM2-Reverse (5′-CAGGA ACATCAAAGCCCTCTTC-3′), MDM4-Forward (5′-C AGCAGGAGCAGCATATGGTATA-3′) and MDM4-Reverse (5′-GAAGCTCTGACGTCCCAGTAG-3′). ALB was used as internal reference, using the primers ALB-Forward (5′-TTTATTCACATCATTCTCTC-3′) and ALB-Reverse (5′-GAGTGAGATATGAGTTGAG-3′). Universal Human Reference DNA (Promega) was used as normal control. The fold-change was determined by the delta-Ct method. Fold-change < 0.5 was scored as loss, between 0.5 and 2 as normal, between 2 and 5 as gain, and amplification as > 5. Samples with PCR failure for target (n = 2) or reference (n = 7) were excluded from further analysis, 2 samples were missing.

Messenger RNA Quantification by Real-Time Reverse Transcriptase PCR

For expression analysis, total RNA was extracted using Trizol (Invitrogen) according to the manufacturer's instructions, and quantitative real-time reverse transcriptase PCR (RT-PCR) was performed using a High Capacity complementary DNA Archive Kit, TaqMan Gene Expression Assays and the ABI 7500 Real-Time PCR system (Applied Biosystems, Carlsbad, Calif). Assays Hs00234753_m1*, Hs00159092_m1*, and Hs01034249_m1* were used for MDM2, MDM4, and TP53, respectively. B2M (Hs99999907_m1) and TBP (Hs00427621_m1*) were used as references. Three normal controls were used (human uterus [AM7892], human adipose tissue [AM7956]; Ambion, Austin, Tex; and human bone total RNA). The expression levels were determined by the delta-Ct method. Gene expression was scored as low when fold-change relative to control average was < 0.5, as normal for 0.5 to 2, and as high for > 2.

DNA Sequence Analysis

TP53

TP53 somatic mutations and polymorphic variants (frequency of 1% or higher) in tumors and cell lines were scored by the AmpliChip p53 test (Roche/Affymetrix, in development), a DNA microarray-based sequence-variant scoring method.26 The PCR failed completely in 4 samples and partly in 7 samples (exon 3 in 6 samples, exon 7 in 1 sample). In cases of partial failure, sequence variations detected were registered, but obviously this could lead to underestimation of mutants and variants.

MDM2SNP309

The region surrounding SNP309 was amplified using flanking intronic primers, and the restriction enzyme MspA1I was used to cut the PCR products. The fragments were separated by capillary electrophoresis on a Megabase-1000 (GE Healthcare) and the variants indicated as TT, GG, and TG, respectively.27 Samples without detectable fragments (n = 7) were excluded.

Statistical Analysis

Association between genotypes and clinicopathologic factors was assessed using Pearson chi-squared test/Fisher exact test and independent 2-tailed t test. Time to recognized metastasis or sarcoma-related death was used as time to progression (TTP). Disease-specific survival (DSS) was time from day of diagnosis until death of disease or last follow-up. Survival probabilities were estimated using the Kaplan-Meier method and assessed by log-rank test. Cox regression analysis was used to adjust for potential confounding clinical variables (Table 1) and Bonferroni correction for multiple testing. Statistical analyses were performed using SPSS (version 16.0, SPSS Inc., Chicago, Ill), and P values < .05 were considered significant.

RESULTS

Patient Characteristics

Clinical characteristics of primary tumors and treatment-specific data are given in Table 1 (n = 125). A total of 114 samples were informative for SNP309 and MDM2/MDM4 copy number, 74 for expression (including the major subtypes: liposarcoma [LS], osteosarcoma [OS], leiomyosarcoma [LMS], malignant peripheral nerve sheath tumor [MPNST], and malignant fibrous histiocytoma [MFH]) and 121 for TP53 genotype. Complete results were available for 62 samples. The genotype frequencies are listed in Table 2.

Table 2. Genotype Frequencies
CharacteristicsNo. (n = 125)%
  • a

    rs 1042522,

  • b, a

    rs 2279744.

TP53R72a4536
P72a4939
mutation2722
unknown43
MDM2SNP309bTT5544
TG or GG4032
GG1915
unknown119
MDM2 copy numberdeletion11
normal2318
gain6552
amplification2520
 unknown119
MDM4 copy numberdeletion22
 normal9979
 gain97
 amplification43
 unknown119

P72 was most frequent in LS compared to all other subgroups (P = .05; Fig. 1A), particularly if WDLS or other/unknown LS (n = 10 and 8). Although R72 enrichment was noted in gastrointestinal stromal tumors (GIST), MPNST, and synovial sarcoma (SS), it was not significant. TP53Mut was detected in 27 samples (21.6%), of which 13 had the P72 variant. Fourteen cases had mutation in more than one position. The majority of mutations (23 of 29) occurred in exons 4 through 8, primarily exon 5 and 8, without codon preference. Seven mutations occurred outside the DNA-binding domain. Twelve of the TP53 mutations were not previously reported in sarcoma.19, 27 TP53Mut dominated in LMS (63%, P = .01, Fig. 1A), whereas the frequency of TP53Mut was low in LS (6%, P = .01). No TP53Mut was detected in GIST, MPNST or SS, but this was not significant (P > .2).

Figure 1.

Genotype distributions are shown in different histological subtypes. Bars represent percentages of (A) TP53Wt (R72, P72) and TP53Mut, (B) SNP309 (TT, TG or GG), (C) copy number level of MDM2, and (D) copy number level of MDM4. The chi-squared test was used to compare each histological subtype to the rest (reference category).

No significant differences in SNP309 distribution among histological subgroups were found (Fig. 1B). However, there was a tendency toward lower SNP309G frequency in SS (P = .06), and among LS, only WDLS displayed the GG genotype. MDM2Ampl was observed in 20% of samples and MDM4Ampl in 3%. As expected, the frequency of MDM2Ampl was high in LS (51%, 81% in WDLS, and 44% in other/unknown LSs) compared with the other subtypes (P < .001; Fig. 1C), and MDM4Ampl was only observed in myxoid or other LSs (11%; Fig. 1D).

Correlations Between Genotypes

The samples from major subtypes having a maximum of 2 missing data points were included in the analysis of patterns of genotypes/phenotypes. The results are visualized in a heat map in Figure 2. There was a strong correlation between MDM2 copy number and expression (P < .001). In general, TP53 mutation and MDM2 amplification was mutually exclusive with 2 exceptions: LS03 and MFH53. However, none of these expressed high level of MDM2. LS stands out with 93% TP53Wt samples, of which 80% had MDM2 overexpression. Notably, only 24% of TP53Wt samples were scored as TP53Norm for expression. MDM4 was also overexpressed in a substantial fraction of TP53Wt LS samples (52%, MDM4High), with a strong correlation with MDM2 expression (P < .001). No skewed distribution of R72P was observed in MDM2Ampl samples. However, almost all SNP309G in LS samples coexisted with MDM2 overexpression, but not conversely.

Figure 2.

Heat map visualizes the patterns of genotypes/phenotypes across the sarcoma subtypes. Patient tumor sample in rows (LS, liposarcoma; MFH, malignant fibrous histiocytoma; LMS, leiomyosarcoma; OS, osteosarcoma; MPNST, malignant peripheral nerve sheath tumor), characteristics in columns stratified as explained. Expression (exp) is fold-change relative to the average of 3 human reference RNAs. Copy number (cn) is fold-change relative to ALB (cn). Gray indicates genotype/phenotype was not available. Abbreviations: mut, mutant; wt, wild type.

The OS samples showed the same pattern of MDM4 overexpression in TP53Wt samples, where MDM2 was also overexpressed. P72 was observed together with TP53Mut in 67% of LMS. Except for the MFHs, of which about half had TP53Mut and almost all had MDM2Gain, no other specific patterns were found. No clear association of SNP309G and other genotypes/phenotypes could be seen.

Correlations With Clinical Data

Survival

At the time of final analysis, 71 patients were deceased due to sarcoma, 18 due to other disease, 8 of unknown cause, and 28 were still alive (1 with persistent sarcoma). The median follow-up was 71 months, DSS was 129 months, and TTP was 36 months.

No statistical differences were observed with respect to R72P polymorphism, but patients with TP53Mut had shorter DSS and TTP than those with TP53Wt (P < .001; Fig. 3A). Despite the importance of TP53 status for survival (hazard ratio[DSS] = 0.41, 95% confidence interval = 0.23-0.74, P = .03 and hazard ratio[TTP] = 0.43, 95% confidence interval = 0.25-0.73, P = .02), multivariate analysis revealed that axial localization increased the risk of sarcoma-related death 2-fold.

Figure 3.

Kaplan-Meyer plots depict (A-C) disease-specific survival and (D) time to progression in study population stratified by (A) TP53 status; R72, P72 or TP53Mut, (B) MDM2SNP309; TT or TG/GG, and (C,D) MDM2 expression level; low, normal or high. Survival probabilities were assessed by a log-rank test.

No statistical differences in DSS or TTP were detected for the SNP309 genotypes (Fig. 3B) or between histological subtypes, except in WDLS, which showed significantly better DSS and TTP than the other LSs. Both MDM2 copy number and expression levels correlated with prognosis: lower levels indicated shorter DSS and TTP (Fig. 3C,D). However, this was not significant in multivariate analysis, where absence of metastasis at diagnosis correlated with better DSS, and axial localization was associated with more than 3-fold increased risk of sarcoma-related death (hazard ratio = 3.64, 95% confidence interval = 1.76-7.53, P = .01). Notably, 77% of MDM2High tumors were LS and presence of WDLS was significantly associated with longer survival in univariate analysis. No significant differences in DSS or TTP were found with relation to MDM4 copy number or expression level.

Response to Chemotherapy

Complete information about radiotherapy was not available; hence analysis of treatment response was restricted to chemotherapy. A total of 43 patients received chemotherapy at least once: doxorubicin (30%), ifosfamide (26%) or a combination of these (28%) as regimen backbone. Doxorubicin was typically given at a dose of 50 to 75 mg/m2 and ifosfamide 4.5 to 9 g/m2 per cycle. Response to treatment was estimated as TTP. Median TTP in this selected cohort was 14 months.

Interestingly, there was a tendency toward longer TTP after chemotherapy with the apoptosis-prone R72 variant (compared with P72, P = .07; Fig. 4A). Patients with R72 responded significantly better to the doxorubicin/ifosfamide combination, although the number of patients was limited (P = .01; Fig. 4B). No significance was found for ifosfamide alone, doxorubicin alone, or other regimens, and no confounding factors were identified in univariate analysis. Hence, R72P status appears to affect chemotherapy response. Seven treated cases harbored P72 in combination with TP53Mut, but their longer TTP (compared to R72/TP53Mut, n = 2) was not statistically significant. There was no significant relation between SNP309G and TTP.

Figure 4.

Kaplan-Meyer plots depict time to progression for patients receiving (A) chemotherapy (regimen not specified) or (B) a doxorubicin/ifosfamide combination. Patients were stratified according to TP53 status; R72, P72, or TP53Mut.

Cell Lines

Response to Nutlin-3a

The cell line characteristics and genotypes are outlined in Table 3. One-third of the cell lines were TP53Mut, four with mutations not previously reported in sarcomas. All RMS cell lines were TP53Mut, whereas the majority of LS and OS cell lines were TP53Wt. None of the TP53Mut lines had SNP309G. All TP53Wt LS cell lines expressed high MDM2 level, whereas TP53Wt OS cell lines were either MDM2Ampl/MDM2High or MDM2Gain/MDM2Norm. In the latter cases, SNP309G was always present. Both variants of R72P were present in LS and OS cell lines, R72 only together with SNP309G in LS. As all LS were MDM2Ampl, no relation between R72 and MDM2Ampl could be observed. Similarly, no association was found in OS cell lines, and no significant relation with MDM4 was observed.

Table 3. Cell Line Characteristics
Cell LineHistologyPatient (Sex/Age)Origin/SiteTP53SNP 309MDM2 CnMDM2 ExpdMDM4 CnMDM4 ExpdTP53 ExpdNutlin-3a Response IC50 (μM)e
  • Abbreviations: ARMS, alveolar rhabdomyosarcoma; Cn, copy number; DDLS, dedifferentiated liposarcoma; Exp, expression; IC50, median inhibitory concentration; LR, local relapse; LS, liposarcoma, unspecified; Met, metastasis; mut, mutant; na, not available; nd, not done; OS, osteosarcoma, unspecified; PT, primary tumor; RMS, rhabdomyosarcoma, unspecified; WDLS, well-differentiated liposarcoma; wt, wild type.

  • a

    Tested and authenticated by fingerprinting.

  • b

    Multiple lesions in multiple bones.

  • d

    Expression numbers are: 1 (<0.5-fold change), 2 (0.5- to 2-fold change), 3 (2- to 5-fold change), 4 (>5-fold change).

  • e

    Values are means ± standard error of the mean for 3 to 5 experiments.24, 25

SA-4aLSM/nanawtTG4341 nd
WLS-160LSnanawtGG4nd4nd nd
SW872aLSM/36namutTTna3na3 >10
T449aWDLSF/68PT/retroperitoneumwtcTT4420 0.6 ± 0.1
T778aWDLSF/69LR/retroperitoneumwtcTT4421 0.7 ± 0.1
Fu-ddl-s1aDDLSM/61LR/retroperitoneumwtcTT4420 0.6 ± 0.2
OSAaOSM/19PT/femurwtTT442020.5 ± 0.1
SaOS-2aOSF/11NamutTT41300>10
U2OSaOSF/15PT/tibiawtTG313121.0 ± 0.5
KPDaOSM/8PT/femurwtGG31210>10
MHMaOSM/42PT/pelviswtcTT442011.9 ± 0.3
OHSOSM/14PT/femurbmutTT3121 nd
IOR/OS14OSM/13PT/nawtTG3120 nd
IOR/OS15aOSF/12PT/nawtcTT4241 nd
IOR/MOSaOSF/13PT/namutTG3120 nd
RMS4ARMSM/17PT/arm, lungmutTT2nd2nd nd
RMS13RMSM/17PT/bone marrowmutTT3131 >10
RMS28ARMSnaMet/axillary nodemutTT3nd2nd nd

We reported previously the response of 10 of the cell lines to Nutlin-3a.24, 25, 28 All cell lines sensitive to Nutlin-3a harboredTP53Wt together with MDM2Ampl or MDM2Gain (Table 3). Notably, the only cell line (scored by sequencing as TP53Wt) that did not respond was previously indicated as TP53Mut, due to failure to amplify exon 8 with the AmpliChip p53 assays, very low expression (fold-change of 0.01), and insensitivity to Nutlin-3a.24 No association between SNP309 or R72P and Nutlin-3a response was found in this panel.

DISCUSSION

Mutations in TP53 are less frequent and more heterogeneous in sarcomas than in other cancer types,29 and in addition to the amplification and overexpression of MDM2, other mechanisms, including MDM4 amplification, may be involved.4, 6, 7, 30 The hypothesis that MDM2Ampl and TP53Mut both provide inactivation of the p53 response has been challenged by the occasional report that the 2 aberrations may be found in the same tumor.31 We found such apparent double mutations in 2 samples, LS03 and MFH53 (Fig. 2). However, none of these expressed MDM2, thus confirming the hypothesis that MDM2Ampl and TP53Mut are mutually exclusive. Most likely, the amplification of the locus containing MDM2 was selected for by overexpression of other genes, such as CDK4 or HMGA2.32-34 Furthermore, contrary to common reports that TP53Mut leads to accumulation of p53 protein, we found low p53 expression in these samples (Fig. 2).

In line with previous reports,2, 5, 29 we found TP53Mut in only 22% of tumors and 39% of cell lines, with no codon preference. Fourteen of the mutations have not been previously reported in sarcoma. LS had only 6% TP53Mut. Thus, drug strategies that activate wild-type p53 may have potential in sarcomas, in particular liposarcomas.24, 35

Twenty percent of the patient samples had more than 5-fold amplification of MDM2, and significantly more in LS, primarily WDLS, confirming previous reports.36, 37 Furthermore, MDM4Ampl was found exclusively in the LS patient samples, strongly associated with MDM2Ampl, indicating that these aberrations are not mutually exclusive, and suggesting that therapeutic strategies that target both MDM2 and MDM4 (MDMX) may be particularly promising in LS patients.

Normal variation in the TP53 and MDM2 genes has also been reported to affect tumor properties.11, 38 The presence of SNP309G in the promoter may increase MDM2 expression.39 However, no overall enrichment of SNP309G was observed in this tumor material compared with that from healthy Norwegians, arguing against any predisposing effect,21 and no clear association between MDM2 expression level and SNP309 was found in this panel. The latter point is perhaps not surprising, because MDM2 expression is highly associated with gain or gene amplification, and there was no tendency for specific alleles to be overrepresented in tumors with altered MDM2 copy number.

As shown here, most published studies conclude that mutant TP53 is associated with poor outcome.40 Similar to loss of p53 function, overexpression of MDM2 could be a poor prognostic indicator, although published conclusions differ (reviewed in Onel and Cordon-Cardo41). We found association of low MDM2 (messenger RNA) expression with shorter DSS and TTP. It should be noted that MDM2 messenger RNA may not correlate with MDM2 protein,42 and also there are multiple splice variants complicating the interpretation of either. However, the relation to survival could not be verified in multivariate analysis, probably due to the strong influence of WDLS with high MDM2, but it is also noteworthy that low MDM2 level is associated with TP53Mut.

Presence of TP53 R72 has been associated with better induction of apoptosis.14 However, excluding LS, P72 was less frequent in this tumor material than in healthy Norwegians (30% versus 58%20), contradicting a protective role of R72. Interestingly, we found a tendency toward longer TTP after chemotherapy for patients with the R72 variant, as previously reported for squamous carcinomas.16 Subsequent subgroup analysis revealed that patients with R72 respond significantly better to doxorubicin/ifosfamide combination regimens than do those patients with the P72 variant. Although the number of patients was limited, the association was significant in multivariate analysis. The reduced chemotherapy response often observed in LS could be associated with lower prevalence of R72.

In addition, TP53Mut combined with R72 has previously been shown to reduce apoptosis after chemotherapy,43 because specific p53 mutants are capable of inhibiting p73, another possible mediator of cytotoxicity. In this series, the number of samples was too small to conclude, but patients with P72/TP53Mut showed longer TTP than their equivalent (R72/TP53Mut).

Finally, an important objective of this study was to investigate if TP53 or MDM2 polymorphism could modify sensitivity to MDM2 antagonists. In our hands, all cell lines with MDM2Ampl responded well to MDM2 antagonist Nutlin-3a, as long as TP53 was wild-type. Interestingly, one cell line (U2OS) only having MDM2Gain, but SNP309TG, also had satisfactory drug-response. In contrast, the TP53Wt OS cell line KPD was not sensitive to Nutlin-3a, despite being homozygous for the variant allele (SNP309GG). This did not appear to be due to inactivation by MDM4, as reported,44 because the MDM4 expression level was normal. However, TP53 expression level was very low. Recently, Knappskog et al reported that other SNPs (eg, SNP285 in breast and ovarian tumors) may interact with SNP309 and modulate the MDM2 phenotype.21 Thus, extended genotype examination may be necessary to guide future p53-activation therapy.

Conclusions

Patients with TP53 wild-type sarcomas and the R72 variant had improved survival and response to chemotherapy. If TP53 is mutated, presence of R72 may have the opposite effect. MDM2 overexpression predicts good response to MDM2 antagonists such as Nutlin-3a, but intact p53 pathway is essential. Detailed analysis of aberrations in the p53 pathways may help to predict tumor sensitivity and resistance to p53 activating therapy by MDM2 antagonists

Acknowledgements

We thank Dr. A.H. Pripp, Center for Biostatistics and Epidemiology, Oslo University Hospital, for statistical advice.

FUNDING SOURCES

This work was supported by the Norwegian Cancer Society and Donations to the Norwegian Radium Hospital.

CONFLICT OF INTEREST DISCLOSURES

Dr. LTV is employed by Hoffmann-La Roche, Incorporated. The other authors made no disclosure.

Ancillary