Antoine Italiano is a Visiting Research Fellow from the Bergonie Institute, Bordeaux, France.
The p53 and phosphoinositide-3-kinase, catalytic, alpha polypeptide/v-akt murine thymoma viral oncogene homolog/mechanistic target of rapamycin (PIK3CA/AKT/mTOR) pathways frequently are altered in sarcoma with complex genomics, such as leiomyosarcoma (LMS) or undifferentiated pleomorphic sarcoma (UPS). The scale of genetic abnormalities in these pathways remains unknown in angiosarcoma (AS).
The authors investigated the status of critical genes involved in the p53 and PIK3CA/AKT/mTOR pathways in a series of 62 AS.
The mutation and deletion rates of tumor protein 53 (TP53) were 4% and 0%, respectively. Overexpression of p53 was detected by immunohistochemistry in 49% of patients and was associated with inferior disease-free survival. Although p14 inactivation or overexpression of the human murine double minute homolog (HDM2) were frequent in LMS and UPS and could substitute for TP53 mutation or deletion, such alterations were rare in angiosarcomas. Phosphorylated ribosomal protein S6 kinase (p-S6K) and/or phosphorylated eukaryotic translation initiation factor 4E binding protein 1 (p-4eBP1) overexpression was observed in 42% of patients, suggesting frequent activation of the PIK3CA/AKT/mTOR pathway in angiosarcomas. Activation was not related to intragenic deletion of phosphatase and tensin homolog (PTEN), an aberration that is frequent in LMS and UPS but absent in angiosarcomas.
Angiosarcomas (AS) represent a rare (<2%) subgroup of soft tissue sarcomas characterized by an aggressive clinical behavior that occur not only in different anatomic locations but also in distinct clinical settings, such as de novo (primary) AS or after radiation therapy or chronic lymphedema (secondary AS).1 Radical surgery and adjuvant radiotherapy when indicated represent the cornerstone of treatment for patients with localized disease. However, despite receiving adequate locoregional treatment, ≥50% of patients will develop metastatic relapse and will die of disease.2 To date, the genetic aberrations involved in AS genesis are poorly understood. Until recently, data were limited to case reports or small case series indicating a complex pattern of numerical and structural genetic aberrations.3-11 We and others have recently identified v-myc myelocytomatosis viral oncogene homolog (MYC) amplification as a recurrent genetic aberration in secondary AS,12, 13 but no specific recurrent event has been reported in primary AS to date.
Alterations of tumor protein 53 (TP53) gene and the phosphatase and tensin homolog (PTEN) are among the most frequent genetic events in human cancer, and the role of their combined inactivation in tumorigenesis has been well examined in several contexts, including lymphoma, prostate cancer, glioblastoma, and medulloblastoma.14-20 It is noteworthy that the p53 and phosphoinositide-3-kinase, catalytic, alpha polypeptide/v-akt murine thymoma viral oncogene homolog/mechanistic target of rapamycin (PIK3CA/AKT/mTOR) pathways frequently are altered in sarcoma with complex karyotypes. It also was demonstrated recently that mutations and/or deletions of the TP53 gene occur in approximately 80% of leiomyosarcomas (LMS), 70% of undifferentiated pleomorphic sarcomas (UPS), and 60% of pleomorphic liposarcomas.21-23 Large deletions on chromosome 10, including the tumor suppressor gene PTEN, also have been observed in approximately 40% to 50% of LMS and 20% to 25% of UPS.24, 25 Moreover, in vitro and in vivo studies25, 26 suggest a crucial role of the PIK3CA/AKT/mTOR pathway in LMS genesis. Because of its rarity, studies regarding the incidence of p53 and PIK3CA/AKT/mTOR pathway aberrations in AS have used heterogeneous methods and were based on single case reports or small case series.27-35 Therefore, we investigated the status of key genes involved in these pathways in a series of primary and secondary AS by sequencing, array comparative genomic hybridization (CGH), and immunohistochemistry (IHC).
MATERIALS AND METHODS
Patient and Samples
Sixty-two tumors (30 primary AS and 32 secondary AS) were included in this study on the basis of availability of tumor material for genetic studies. The clinicopathologic characteristics are listed in Table 1. Diagnoses were established according to the World Health Organization Classification of Tumors.1 This study was approved by the MSKCC institutional review board (IRB protocol 02-060).
Table 1. Clinical and Pathologic Characteristics of Patients With Angiosarcoma
Genomic DNA was isolated from frozen tumor tissue by phenol/chloroform extraction, and quality was confirmed by spectrophotometry and electrophoresis. Array CGH analysis was performed as described previously.36 Briefly, tumor and reference DNA samples were labeled with cyanine-3-dUTP and cyanine-5-dUTP, respectively, with random priming (Agilent Enzymatic Labeling Kit; Agilent Technologies, Santa Clara, Calif). The reference sample comprised a pool of DNA from multiple clinically healthy donors (Promega, Madison, Wis) of the same sex as the AS patient. The labeled probes were combined and hybridized to 180,000-feature, genome-wide oligonucleotide CGH array (design 052252; Agilent Technologies). Arrays were scanned at 3-μm resolution using an Agilent G2565CA scanner. Image data were processed with Feature Extraction (version 10.10) and Genomic Workbench (version 6.5; Agilent Technologies). Data were filtered to exclude probes that exhibited nonuniform hybridization or signal saturation and were normalized using a centralization algorithm with a threshold of 6. The Aberration Detection Method 2 (ADM2) algorithm was used to define DNA copy number aberrations using a “3 probes minimum” filter and a threshold of 6 with a fuzzy zero correction, and genomic gain and loss were defined as log2 test:reference DNA signal intensity >0.2 and <−0.2, respectively. A genomic copy number amplification was defined as a region within which the log2 ratio of tumor DNA to reference DNA exceeded 2.0.
One microgram of total RNA was reverse-transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Carlsbad, Calif) at 25°C for 10 minutes, 37°C for 120 minutes, 85°C for 5 minutes, and holding at 4°C. Next, 20 ng/μL of resultant cDNA were used in quantitative polymerase chain reaction (PCR) analysis using a 7500 Real-Time PCR System (Applied Biosystems) and predesigned TaqMan ABI Gene expression Assays (Hs_00234753_m1 for murine double minute human homolog [HDM2]; Hs_99999189_m1 for protein 14/cyclin-dependent kinase inhibitor 2A [P14/CDKN2A]). Amplification was carried out at 95°C for 10 minutes and for 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. To calculate the efficiency of the PCR and to assess the sensitivity of each assay, we also performed a 5-point standard curve (80 ng/μL, 26.67 ng/μL, 8.88 ng/μL, 2.96 ng/μL, and 0.98 ng/μL). Triplicate CT values were averaged, and amounts of target were interpolated from the standard curves and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (reference gene).
The mutational status of TP53 (exons 2-11), PIK3CA (exons 3, 10, and 20), v-raf murine sarcoma viral oncogene homolog B1 (BRAF) (exon 15), and neuroblastoma v-ras oncogene homolog (NRAS) (exons 2 and 3) was assessed by direct sequencing of genomic DNA. Protocols and primers are available on request. Sequence analysis was performed with a Sequence Scanner (version 1.0; Applied Biosystems).
Enhanced Mismatch Mutation Detection Analysis and Copy Number Analysis by Quantitative Multiplex Polymerase Chain Reaction
A screening of genomic variants of PTEN was performed using enhanced mismatch mutation detection analysis (EMMA), an electrophoretic heteroduplex analysis based on the use of a matrix that increases the electrophoretic mobility between homoduplex and heteroduplex.37 Primers were designed to include flanking intronic sequences (at least 50 base pairs from splices sites) and to avoid known polymorphism and prevent mispriming (primer sequences are available upon request). PTEN coding sequences were divided into 9 amplicons and were analyzed in 3 multiplex PCRs. EMMA was performed as described previously using an EMMA polymer (Fluigent, Paris, France) according to the manufacturer's instructions.38 Amplicons that had abnormal profiles were sequenced and analyzed on the ABI PRISM 3130XL Genetic Analyzer using the Collection and Sequence Analysis software package (Applied Biosystems, Courtaboeuf, France). Nucleotide position was indicated on the basis of the coding sequence NM_000314.4 and according to the recommended guidelines (available at: http://www.emqn.org/emqn.php and www.hgvs.org/mutnomen; [Accessed September 30, 2011]).
Quantitative multiplex PCR was carried out on 200 ng of tumor DNA samples, which were measured on a NanoDrop spectrophotometer (ND-1000; Labtech International, Palaiseau, France). All primers were designed to include flanking intronic sequences and to prevent detection of the processed pseudogene (PTEN1) that presents 98.6% homology to the cDNA of PTEN (primer sequences are available upon request). Amplicon sizes ranged from 111 to 347 base pairs. Exons of the PTEN gene were analyzed using 2 assays that consisted of 4 and 5 PTEN amplicons, 1 amplicon of 2 genes proximal to PTEN (3′-phosphoadenosine 5′-phosphosulfate synthase 2 [PAPSS2] and p53-regulated DNA replication inhibitor [KILLIN]), and amplicons of the mutS homolog 2 (MSH2) and breast cancer 1 (BRCA1) genes used were as internal controls for copy number analysis. Multiplex PCRs were performed with the Qiagen Multiplex PCR kit according to the manufacturer's recommendations (Qiagen, Courtaboeuf, France) and 1 μL of 10X primer mix (final concentration, 1-10 μmol/L of each primer), 1 of which carried a 6-FAM label. Eight percent dimethyl sulfoxide was added for the assay that included PTEN exon 1 and KILLIN amplicons. Amplification was performed as recommended by the manufacturer (Qiagen), and 24 cycles were performed for quantitative PCR. The DNA fragments generated by quantitative multiplex PCR were separated by capillary electrophoresis on the ABI 3130XL Genetic Analyzer (Applied Biosystems). Data were analyzed with GeneMapper software (version 4.0; Applied Biosystems), and copy number variation was evaluated for each amplicon by calculating the relative peak height for each sample with respect to the mean value of the relative peak height observed in all samples.
IHC for p53 (DO7, 1:800 dilution; Dako, Carpinteria, Calif), HDM2 (clone 1F2, 1:50 dilution; Calbiochem, San Diego, Calif), p14 (clone 4C6/4, 1:1000 dilution; Cell Signaling Technology, Danvers, Mass), phosphorylated ribosomal protein S6 kinase (p-S6K) (clone Ser235/236, 1:50 dilution; Cell Signaling Technology), phosphorylated eukaryotic translation initiation factor 4E binding protein 1 (p-4eBP1) (clone Thr37/46, 1:500 dilution; Cell Signaling Technology), and p44/phosphorylated extracellular signal-related kinase (p44/p-ERK) (clone Thr202/Tyr204, 1:50 dilution; Cell Signaling Technology) was performed according to each manufacturer's recommendations. Immunoreactivity was scored as the percentage of nuclear staining per 10 high-power fields in several areas, regardless of staining intensity. A 20% cutoff value for the detection of positive nuclear reactivity was selected for all antibodies.39, 40
Descriptive statistics were used to indicate the distribution of variables in the population. Differences between groups were evaluated using chi-square tests or Fisher exact tests for categorical variables and t tests for continuous variables. The statistical analyses of baseline demographics and clinical outcomes were based on all data available up to the cutoff date (August 31, 2011). The Kaplan-Meier method was used to estimate disease-free survival, which was defined as the time after diagnosis during which patients had no evidence of disease.
TP53 Gene Alterations Are Very Rare in Angiosarcomas
It has been reported that TP53 deletions are frequent events in sarcomas with complex karyotypes, especially LMS and UPS. To assess the frequency of this aberration in AS, we analyzed 18 tumors (6 primary AS and 12 secondary AS) by array CGH analyses, none of which exhibited copy number loss of the TP53 gene. Next, we analyzed the mutational status of TP53 by sequencing exons 2 through 11 on genomic DNA from 52 tumors. We observed a TP53 mutation in only 2 tumors (4%) (A138S on exon 5 and R238Q exon 7), which were not present in the corresponding germline DNA from the same patient. Both tumors with TP53 mutation occurred in patients who developed a secondary breast AS in a previously irradiated field (tumors AS10 and AS17). Both tumors carried MYC amplification, and 1 had a kinase domain receptor (KDR) A1065T exon 24 mutation (tumor AS17), as previously reported.41 Moreover, we observed an R72P polymorphism in TP53 exon 4 in 30 tumors (58%), including 19 secondary AS and 11 primary AS. P53 protein expression was then assessed by IHC in 46 tumors. Twenty-three tumors (50%) were negative for p53 expression, and 23 (50%) were positive (including the 2 tumors with TP53 mutation). There was no significant difference in p53 IHC status between primary AS and secondary AS. Moreover, there was no correlation between the presence of the R72P polymorphism and protein expression.
p53 Protein Expression Correlates With Disease-Free Survival in Angiosarcoma
Because p53 protein status assessed by IHC previously was associated with outcome in several tumor types, including sarcomas with complex genomics, we decided to analyze its correlation with clinical behavior in AS. We observed that p53-positive tumors (defined as >20% nuclear positivity) were associated significantly with worse disease-free survival (3.4 months [95% CI, 0.8-6 months] vs 14.9 months [95% CI, 2.5-27.4 months]; P = .002) (Fig. 1).
p14 Inactivation or HDM2 Overexpression Does Not Substitute for TP53 Inactivation in Angiosarcoma
Because p14 is implicated in p53 degradation by proteasome through its interaction with HDM2, we analyzed p14 and HDM2 copy number alterations by array CGH and their expression levels (real-time PCR, IHC) to identify the potential loss of p14 expression or HDM2 overexpression, which may compensate for a lack of TP53 inactivation. However, no deletion of the CDKN2A locus encoding for P14 or HDM2 amplification was observed by array CGH. Real-time reverse transcriptase-PCR was performed on 19 tumors (10 secondary AS and 9 primary AS) and did not reveal any significant difference in P14 or HDM2 expression between primary and secondary AS or between AS and other vascular tumors (data not shown). IHC was performed for 43 tumors. P14 and HDM2 expression was present in 36 of 40 tumors (89%) and in 6 of 40 tumors (15%), respectively, with interpretable results. We observed a concordance between the presence/absence of p14 overexpression and the absence/presence of HDM2 expression in 81% of tumors.
No PTEN Intragenic Mutations or Deletions Are Detected in Angiosarcoma
PTEN deletion is a frequent event in sarcomas with complex karyotypes. To assess the incidence of this genomic aberration in AS, we analyzed a series of 26 tumors using quantitative multiplex PCR to detect gross genomic rearrangements and using array CGH to detect smaller deletions. Six tumors revealed duplication of all exons in PTEN, whereas 8 tumors revealed deletion of all exons. No tumor with intragenic rearrangement or deletion was identified. These results indicate the presence of a large genomic imbalance involving chromosome 10 but do not favor PTEN as a target gene. Next, we used EMMA to detect point mutations that may inactivate the function of PTEN and substitute the absence of a specific deletion. We did not observe any PTEN mutations in this series of AS. We analyzed expression of the PTEN gene using published expression data from a series of 21 AS and 3 other vascular tumors previously studied on the U133A Affymetrix platform (Affymetrix, Inc., Santa Clara, Calif).40 Although the expression of PTEN was relatively low, it was uniform across the samples and was similar between AS and other vascular tumors.
The PIK3CA/AKT/mTOR Pathway Is Activated in a Subset of Angiosarcomas
Previous studies suggested a crucial role of the PIK3CA/AKT/mTOR pathway in the malignant transformation of endothelial cells. Although PTEN did not appear to be specifically altered in AS, we hypothesized that the PIK3CA/AKT/mTOR pathway was activated through alternative mechanisms in at least a subset of AS. Therefore, we performed further immunohistochemical analyses to evaluate the expression of p-S6K and p-4eBP1, 2 downstream targets of mTOR complex 1 (mTORC1). We also evaluated activation of the extracellular signal-related kinase (ERK) pathway because of its multiple interactions with the PIK3CA/AKT/mTOR pathway and its potential involvement in the phosphorylation of S6 ribosomal protein. We observed that 17 of 40 AS (42%) were positive for pS6K and/or p-4eBP1 (Fig. 2). We also observed that 12 of 39 tumors (31%) were positive for p44/p-ERK. Eleven of 17 tumors that were positive for p-S6K and/or p-4eBP1 also were positive for p44/p-ERK (Pearson correlation coefficient, 0.6; P = .02). Next, we investigated the incidence of point mutations in PIK3CA by screening this gene for 13 common hot-spot mutations in exons 3, 10, and 20 in 33 tumors (22 secondary AS and 11 primary AS). We observed a recurrent truncating frame-shift mutation in exon 10 of the PI3KCA gene (S553fs) in 13 tumors (39%), including 5 primary AS and 8 secondary AS. This mutation was then confirmed by PCR using mutant-specific allele exon 10 primers (forward, 5′-CAGTTACTATTCTGTGACTGGTGT-3′; reverse, 5′-AGATTTTCTATGGACCACAGG-3′) (Fig. 3). To further explore the biologic significance of this unusual mutation, we sequenced cDNA from 3 mutated tumors that had available RNA. However, no mutation was identified, suggesting that the mutant PIK3CA transcript undergoes nonsense-mediated decay and, thus, is not detectable at the cDNA level (Fig. 3). Moreover, there was no correlation between PI3KCA mutational status and the p-S6K and/or p-4eBP1 IHC results.
NRAS and BRAF Are Not Mutated in Angiosarcomas
Given the high proportion of AS that was positive for p44/ERK IHC in our series and the potential role of RAS dysregulation in abnormal vascular growth control, we hypothesized that a subset of AS would be associated with mutation of the BRAF or NRAS gene. We screened BRAF exon 15 and NRAS exons 2 and 3 in 52 AS samples; however, no mutations were identified at these hot spots.
Recent studies investigating the TP53 pathway in large series of sarcomas with complex genomics (LMS, UPS, and pleomorphic liposarcomas) have demonstrated that TP53 alterations are highly recurrent in patients with these tumors, and up to 80% of patients exhibit at least 1 TP53 alteration (mutation and/or deletion).21-23 Our results indicating a deletion rate of 0% and a mutation rate of 4% indicate that AS displays a different TP53 alteration profile than the majority of other sarcomas with complex karyotypes. Our findings are in agreement with previous smaller scale reports suggesting that TP53 mutations are uncommon in AS.28, 30
It is noteworthy that a polymorphism in exon 4 (a proline to arginine substitution at codon 72 [P72R]) was detected in 58% of the analyzed tumors. Unlike the deleterious tumor-associated P53 mutations, which occur in approximately 98% of tumors in the DNA-binding-domain of the protein, single nucleotide polymorphisms are expected to be phenotypically silent. The R72P polymorphism occurs in the polyproline domain, which lies between the N-terminal transactivating domain and the DNA-binding domain of p53. This polymorphism has been associated with a modulation of the proapoptotic potential of p53 and cancer susceptibility.42 However, the level of evidence is poor because of important methodological issues, and several meta-analyses failed to demonstrate a risk between the R72P polymorphism and any cancer.42 It also is noteworthy that a high incidence of this polymorphism recently was reported in patients who had atypical vascular lesions that occurred after radiotherapy.35 In our series, we observed this single nucleotide polymorphism in both primary and secondary AS, suggesting the absence of any link between this polymorphism and the etiology of AS.
The rarity of deleterious TP53 mutations or deletions in AS does not necessarily suggest that the TP53 pathway is intact and functional in these tumors. p14 acts on the TP53 pathway by interacting with HDM2, an E3 ubiquitin ligase that targets p53 for degradation by the proteasome. This results in a stabilization and activation of p53. Loss of p14 expression and/or HDM2 overexpression is frequent in sarcomas, and it has been demonstrated that such loss substitutes for p53 inactivation in tumors without TP53 gene aberrations. Our results indicate that these genetic aberrations are not commonly involved in AS genesis. Despite these results, we observed that a high proportion of AS displayed p53 overexpression. Normal p53 protein has an extremely short half-life, and it is not detected by IHC. However, after genotoxic stress, the p53 protein level increases rapidly because of stabilization through post-transcriptional mechanisms. In contrast, mutation of the TP53 gene results in nuclear accumulation of the abnormal protein, which has a longer half-life and can be detected by IHC. On the basis of the assumption that only stable (presumably mutated) protein will be expressed, TP53 IHC often is used as a surrogate to predict the presence of TP53 mutations. In our study, however, we observed a high level of discordance between the IHC results and the sequencing results. Although the 2 tumors that harbored TP53 mutations were identified as positive on immunostaining, 21 tumors that harbored wild-type TP53 also were stained positive. Such discrepancies were reported previously in several studies, suggesting that IHC is not a reliable tool for TP53 mutation screening.43, 44 One hypothesis is that this overexpression is the result of specific oncogenic stresses leading to stabilization of the wild-type protein. In this instance, we would expect to observe an improved outcome as the result of increased tumor apoptosis, but this was not the case. A second hypothesis is that the nuclear accumulation of p53 depends on mechanisms different from mutation and reflects functional changes. It is noteworthy that p53 overexpression by IHC has been identified as a strong prognostic factor in many human cancers, including soft tissue sarcomas.43, 44 Our data indicate that this biomarker also has prognostic value in AS.
We also demonstrated that, in contrast to other sarcomas with complex genomics, PTEN is neither deleted nor mutated in AS. Moreover, our expression data indicating similar levels of PTEN expression across the all AS and other vascular tumors suggest that epigenetic silencing or transcriptional defect of PTEN is unlikely. All together, our results are in agreement with previous reports indicating that the constitutional, biallelic inactivation of PTEN is associated with vascular malformation but not with malignant vascular tumors.45
Despite the lack of PTEN alteration in AS, we identified activation of the PIK3CA/AKT/mTOR pathway in a significant proportion of tumors. Several studies previously demonstrated a crucial role of PIK3CA/AKT/mTOR signaling in endothelial cell homeostasis.46 Moreover, it has been demonstrated that embryonal fibroblasts transformed with the PI3KCA gene induce AS at the site of injection in chickens.47, 48 It is noteworthy that we observed a recurrent frameshift mutation that affected the helical domain of PI3KCA in approximately 40% of tumors. The same mutation recently was reported at a low incidence rate in 1 patient with large B-cell lymphoma49 and in 1 patient with synovial sarcoma.50 Our results indicating that this mutation is not detectable at the cDNA level strongly suggest that this aberration does not have functional consequences. Therefore, further studies are needed to identify the mechanisms involved in PIK3CA/AKT/mTOR activation in AS. Indeed, the patterns of regulation of the PIK3CA/AKT/mTOR pathway in cancer cells are very complex and involve cross-talk with several other pathways. For instance, it is now recognized that PI3KCA and mitogen-activated protein kinase (MAPK) signals converge upstream and downstream of mTORC1.51 It is noteworthy that we observed a good correlation between the activation of these 2 pathways in our series, and the majority of tumors that were positive for p-S6K and/or p-4eBP1 also were positive for p44/p-ERK. Overall, our findings may have therapeutic implications, specifically that PI3KCA and/or mTOR antagonists represent a promising approach in the systemic treatment of AS, as suggested previously by a recent in vitro study and preliminary clinical data.52, 53 These data also suggest that, along with the inhibition of vascular endothelial growth factor receptors, the inhibition of MAPK signaling may be a relevant approach in AS, as suggested by preclinical data,54 and also may contribute to the recently described clinical activity of multitarget tyrosine kinase inhibitors, such as sorafenib in AS.55 Moreover, synergy is suggested with the combination of mTOR and MAPK inhibitors.51
In summary, the current results demonstrate that AS represents a distinct subgroup among sarcomas with complex karyotypes. Although TP53 mutation and PTEN deletion are frequent in LMS and UPS, these aberrations are not involved in the pathogenesis of AS. However, AS is characterized by a complex pattern of alterations of the TP53 and PIK3CA/AKT/mTOR pathways that does not depend on MYC amplification status. These results underscore the complex, coordinated regulation of the survival and growth-control pathways in AS. Our results represent an important basis for further investigations with the objective of determining how the activities of these pathways are coordinated in AS and for the design of new therapeutic strategies.
We thank Myriam Ardo Dia and Delphine Lafon for their helpful technical assistance and Milagros Soto for editorial assistance.
Note Added in Proof
This work was supported by grants PO1 CA047179-15A2 (C.R.A.) and P50 CA 140146-01 (C.R.A.); Cycle for Survival (C.R.A.); Angiosarcoma Awareness (C.R.A.); North Carolina State University Canine Cancer Genomics Fund; the Fulbright Program-French American Commission (A.I.); and Fondation Monahan, French National Cancer Institute: Soutien pour la Formation àla Recherche Translationnelle en Cancérologie, 2010 (A.I.).