Translational Research in the Gynecologic Oncology Group: Evaluation of Ovarian Cancer Markers, Profiles and Novel Therapies

Cancer Biology

The Cancer Genome Atlas (TCGA) project in ovarian cancer is beginning to release data that is confirming that epithelial ovarian cancers (EOCs) exhibit extensive molecular heterogeneity with alterations in numerous pathways including oncogenes, tumor suppressors, cell cycle regulation and DNA repair. EOCs exhibit aneuploidy, chromosomal alterations, genomic instability, mutations, amplifications, overepxression, amplifications, silencing, modifications, splicing and epigenetic mechanisms as well as natural and induced sequence variations. Genomic and epigenetic alterations not only drive tumorigenesis, invasion, metastasis and disease progression, but also affect which patients will or won’t respond to specific treatments or experience adverse events. Among the hundreds of defects and alterations observed within the tumors in individual patients with EOC, most are likely passengers while only a fraction are species-specific operators (drivers). Identification of the casual drivers in individual EOC patients will enable us to design more effective marker-driven clinical trials that select the right drugs for the right patients.

Although the molecular classification of EOC clearly represents a landmark advance for women with a diagnosis of EOC or increased risk of this disease, the more challenging work lies before us. Effective treatments with long-term clinical benefit will not only require sustained inactivation or re-control of the critical drivers of tumorigenesis operating in a particular cancer patient but must also anticipate and counteract natural feedback loops, redundant and divergent genes and pathways as well as innate and acquired resistance mechanisms that are differentially induced in select EOC patients. This includes drug efflux, metabolism, detoxification, clearance along with DNA repair pathways, expression, post-translational modifications, silencing, alternative splicing, isoform-switching and epithelial-mesenchymal transitioning, for example. It is also becoming clear that a number of the molecular defects and mechanisms operative in EOC vary more by cell type and grade than by disease site. A number of clinical trials are already focusing eligibility criteria on select histologies within or across disease sites.

Signal Transduction

Tumorigenesis, invasion, metastasis and disease progression can be controlled of membrane-bound, cytoplasmic and nuclear receptors that can be activated by ligands. Following activation, receptors dimerize or oligiomerize and undergo conformational changes, autophosphorylation and phosphorylation of signaling molecules that ultimately regulate transcription, translation, and post-translational modifications as well as processes affecting cell proliferation, maturation, contact, adhesion, migration, invasion, survival, resistance, and the production and secretion of growth factors, cytokines, chemokines and soluble receptors (Figure 1A and 1B). These autocrine, paracrine and systemic factors then affect different cells in the tumor microenvironment and distant sites thus further regulating cancer progression, angiogenesis, vasculogenesis, permeability, immune function as well as the efficacy and toxicities associated with cancer treatments.

Inappropriate receptor activation promotes tumorigenesis and can be induced by a number of mechanisms including overexpression of autocrine and paracrine factors. Receptors can also be mutated causing constitutive activation in the absence of ligand binding, or be overexpressed via gene amplification, transcriptional activation or post-transcriptional mechanisms which typically require ligand availability and binding for activation. Cross talk between different receptor super families can also activate receptors by a ligand-independent mechanism. Cancer progression, invasion and metastasis are promoted by the amplification, mutation or overexpression of signaling molecules downstream from receptors. Various cell types within the tumor microenvironment can be induced to secrete pro-inflammatory factors that stimulate the vasculature to recruit leukocytes to the tumor. After activation, these tumorassociated leukocytes can release factors that recruit more inflammatory cells and stimulate angiogenesis and neovasculogenesis to sustain tumor growth, promote disease progression, and facilitate tumor invasion and metastasis. Schematics are provided for p53 (Figure 1C), cell cycle regulation (Figure 2A), effects of genotoxic stress (Figure 2B), nucleotide excision repair (Figure 2C) and BRCA1 (Figure 2D) as these pathways are the subject of a number of TR studies conducted by the GOG.

Cancer Therapeutics

Insights into the molecular and biochemical mechanisms operative in cancer development, progression and metastasis have uncovered a wide array of molecules in tumor cells and/or the tumor microenvironment including stromal cells, endothelial cells, endothelial precursor cells, pericytes, and immune cells that can be targeted therapeutically. Among these agents are the molecular targeting therapies that inhibit receptor tyrosine kinases, non-receptor tyrosine kinases, serine/threonine kinases, transferases, proteases as well as other enzymes, processes and/or pathways. Some of the molecular targeting therapies are selective inhibitors while others are dual inhibitors or multiple inhibitors (Table 1). Figures 1A and 1B provide a few examples of molecular targeting agents that can inhibit epidermal growth factor (EGF), EGF receptor (EGFR)^ErbB1/Her1, ErbB2^Her2, vascular endothelial growth factor-A (VEGF-A), VEGF receptor (VEGFR) or downstream signaling molecules. A number of these agents are being evaluated in human EOC and specifically in GOG clinical trials as illustrated in Figures 1A and 1B. In addition to the molecular targeting agents, there is an arsenal of traditional cytotoxic anticancer drugs (Table 2). Alkylating agents and microtuble inhibitors have been particular effective in EOC. Despite high initial response rates to first-line treatment and re-challenge with platinum agents and taxanes, about 30% of women with advanced stage EOC fail to respond to initial platinum-taxane based chemotherapy and 5-year survival remains below 40% for women with advanced stage EOC who underwent surgical staging and cytoreduction.

GOG 111 Protocol

GOG 111 was a Cancer Therapy Evaluation Program (CTEP)-sponsored, randomized phase III protocol by McGuire and colleagues that showed improvements in response rate (p=0.01), progression-free survival (PFS, p<0.001) and overall survival (OS; p<0.001) following intravenous paclitaxel and cisplatin compared with intravenous cyclophosphamide and cisplatin in women with previously-untreated, histologically-confirmed, suboptimal stage III EOC who underwent surgical staging and had >1 cm residual disease or stage IV EOC (Table 3) [1]. Birrer and colleagues initiated a series of retrospective studies to evaluate the prognostic relevance of a panel of markers including cell cycle regulators [2], the p53 tumor suppressor gene [3] and several proto-oncogenes including ErbB2^Her2 [4] and cMYC [5] in archival formalinfixed and paraffin-embedded (FFPE) primary tumor specimens from women with advanced stage who participated in the GOG 111 protocol.

High cyclin E protein expression, defined as >40% cyclin E positive tumor cells, was associated with a shorter median survival (29 versus 35 months) and an increased risk of death (hazard ratio [HR]=1.4, 95% confidence interval [CI]=1.0–2.1, p=0.05) [2]. This association was most notable in women with stage III disease (HR=1.7, 95% CI-1.1–2.6, p=0.03), serous histology (HR=1.8, 95% CI=1.2–2.8, p=0.01) and non-measurable disease (HR=2.4, 95% CI=1.4–4.3, p<0.01) and those randomly allocated to paclitaxel and cisplatin (HR=1.8, 95% CI=1.1–2.9) [2]. Amplification of cyclin E, detected by fluorescence in situ hybridization (FISH), was shown to be associated with high verse low cyclin E expression (p<0.006) [2]. Investigations are currently underway to evaluate the clinical utility of tumor expression of p27, cyclin D1 and p57 in the GOG 111 cohort (Table 3). Overexpression of p53, defined as ≥10% tumor cells exhibiting nuclear staining using the N-terminal DO-7 antibody (Figure 1C), was associated with GOG performance status (p=0.018) and grade (p=0.003), but not with PFS or OS [3]. ErbB2^Her2 amplification, defined by FISH as >2 or >4 copies of ErbB2^Her2/chromosome 17, was a rare event in EOC and was not associated with clinical characteristics, tumor characteristics or any measure of outcome including PFS or OS [4]. In addition, cMYC amplification, defined by FISH as ≥1.5 or ≥2 copies cMYC/chromosome 8, was not associated with clinical characteristics, tumor characteristics, PFS or OS [5]. Polysomy 8 was observed in 22 patients without cMYC amplification and 3 with cMYC amplification, and was associated with age and measurable disease status, but not other clinical covariates or outcomes [5].

GOG 114/132 Protocols

Berchuck and colleagues at Duke University Medical Center initiated a series of retrospective studies to evaluate the prognostic relevance of a panel of tumor suppressors, angiogenic markers, cell cycle regulators, transcriptional regulators and DNA repair proteins in frozen and archival FFPE primary tumor specimens available from women with advanced stage EOC who participated in the GOG 114 or the GOG 132 protocol. GOG 114 was a CTEP-sponsored, intergroup, randomized phase III trial with the Southwestern Oncology Group (Protocol 9227) and the Eastern Cooperative Oncology Group (Protocol GO114) by Markman and colleagues which showed improvements in PFS (p=0.01) and OS (p=0.05) with high-dose intravenous carboplatin followed by intravenous paclitaxel and intraperitoneal cisplatin compared with intravenous paclitaxel and cisplatin in women with previously-untreated, histologically-confirmed, optimally-resected, stage III EOC who underwent surgical staging and had <1 cm residual disease (Table 3) [6]. GOG 132 was a CTEP-sponsored, randomized phase III trial by Muggia and colleagues which demonstrated that inferior response rates (p<0.001) and PFS (p<0.001) but similar OS were observed with paclitaxel monotherapy compared with either cisplatin monotherapy or the paclitaxel and cisplatin combination in women with previously-untreated, histologically-confirmed, suboptimally-resected stage III and stage IV EOC who underwent surgical staging and had >1 cm residual disease (Table 3) [7].

A mutation in exons 2 to 11 of the multifunctional tumor suppressor, p53, was associated with non-mucinous or clear cell histologies (p=0.018) and a short-term reduction in the risk of disease progression (HR=0.4, 95% CI=0.2–0.8, p=0.014) and death (HR=0.3, 95% CI=0.1–0.8, p=0.014) [8]. These striking risk reductions were time-dependent and eventually disappeared around three years following initiation of primary treatment [8]. Overexpression of p53, defined as tumors with any detectable p53 immunostaining using the N-terminal DO-1 antibody (Figure 1C), was observed in 55 patients (100%) with only missense mutation(s), seven patients (32%) with truncation mutations, and eight patients (40%) lacking a mutation in exons 2 to 11, and was associated with tumor grade (p0.018) but was not associated with PFS or OS [8]. Maspin, another tumor suppressor, was not detected by immunoblot analysis in 19 (28%) of the frozen primary tumors tested, and non-detectable maspin was associated with suboptimally-debulked disease (p=0.034) and an increased risk of disease progression (HR=1.89, 95% CI=1.04–3.45, p=0.038) and death (HR=1.99, 95% CI=1.07–3.69, p=0.030) [9]. Follow up studies are underway to determine if loss of maspin expression was associated, at least in part, to methylation-induced epigenic silencing [10]. Associations were observed between categorized immunoblot expression of the pro-angiogenic factor, VEGF-A, and p53 overexpression (p=0.022), VEGFR-1 and either race (p=0.027) or histologic subtype (p=0.007), and thrombospondin-1 (an angiogenic inhibitor and promoter) and either PFS (HR=2.19, 95% CI=1.29–3.71, p=0.004) or OS (HR=1.93, 95% CI=1.12–3.32, p=0.018) [11]. High CD105 microvessel density (CD105-MVD), defined as ≥19.25 CD105 positive vessels per high density field, was associated with increased risk of disease progression (HR=1.873, 95% CI=1.102–3.184, p=0.020) but not death, whereas CD31-MVD, defined as ≥24.25 CD31-positive vessels per high power field, was not associated with PFS or OS [12]. In the GOG 114/132 cohort, none of the cancers exhibited homozygous deletions in p16, but loss of immuno-expression of p16 was associated with wild-type versus mutant p53 (p=0.03) and Rb expression (p<0.001) [13]. Investigations are currently underway to determine the optimal combination of G₁ stimulators (cyclin D1, cyclin E, cdk4, Ki67) and inhibitors (p16, pRb, p27, p14) with clinical factors that predicts PFS and OS in the GOG 114/132 cohort [14]. Relative immunoblot expression of the p63 isoform lacking the transactivation domain, ΔNp63α (a homolog of p53), was associated with debulking status (p=0.023), relative expression of VEGF-A (p=0.045), and an increased risk of disease progression (HR=1.483; 95% CI=1.060–2.076; p=0.021) but not with p53 status or survival [15]. Panasci and colleagues are currently examining the prognostic relevance of x-ray repair cross-complementing protein group 3 (XRCC3), a member of RecA / RAD51-related protein family that interacts with RAD51C and is involved in DNA repair and homologous recombination to maintain chromosome stability, in the GOG 114/132 cohort.

GOG 148 Protocol

GOG 148 (Table 3) was a CTEP-sponsored, serum marker protocol by Burger and colleagues which demonstrated that among women who participated in a randomized phase III protocol for early stage EOC (GOG 95 or 157) or advanced stage EOC (GOG 111, 114, 132, 152 or 162) and had either low or high CA 125 levels, those with high sTNFR-I and low sTNFR-II levels had the lowest risk, patients with low sTNFR-I and sTNFR-II or high sTNFR-I and sTNFR-II levels had an intermediate risk, and patients with low sTNFR-I levels and high sTNFR-II levels had the highest risk of disease progression [16]. The prognostic value of serial assessment of these soluble death receptors in women with low and high CA125 is currently under investigation.

GOG 157 Protocol

GOG 157 was a CTEP-sponsored, randomized phase III trial by Bell and colleagues which demonstrated that a 6 versus 3 cycle regimen of paclitaxel and carboplatin resulted in significantly more frequent grade 3 or 4 neurotoxicity and anemia, and statistically similar PFS and OS in women with previously untreated, histologically-confirmed, completely-resected stage IA grade 3 (or clear cell tumors), stage IB grade 3 (or clear cell tumors), stage IC or stage II EOC who underwent optimal surgical staging (Table 3) [17]. Retrospective studies were also initiated to evaluate the prognostic relevance of p53 [3], cell cycle regulators [18] and angiogenic markers [19] in archival FFPE primary tumors from women who participated in the GOG 157 protocol. Overexpression of p53, defined as ≥10% tumor cells exhibiting nuclear staining using the DO-7 antibody (Figure 1C), was observed in 51% (73/143) of the GOG 157 early stage EOCs and was associated with worse PFS (logrank test: p=0.013), a 2-fold higher risk of disease progression (95% CI=1.15–3.63; p=0.015), and a similar risk of death, but was not a statistically significant independent prognostic factor for PFS (HR=1.81, 95% CI=0.99–3.30, p=0.052) [3]. Investigations are underway to examine the prognostic relevance of the IHC expression of cyclin E and p27 [18] as well as the angiogenic markers: thombospondin-1, angiopoietin-1, SPARC, VEGFR-2 (Flk-1), TIE-2 and VEGF-A [19] in the GOG 157 cohort.

GOG 158 Protocol

GOG 158 was a CTEP-sponsored, randomized phase III trial by Ozol and colleagues which demonstrated that the experimental combination of a 3-hour infusion of paclitaxel followed by carboplatin was less toxic, easier to administer and not inferior in terms of PFS and OS to the control regimen with a 24-hour infusion of paclitaxel followed by cisplatin in women with previously-untreated, histologically-confirmed, optimal-resected stage III EOC who underwent adequate surgical staging and had <1 cm residual disease (Table 3) [20]. The presence of detectable versus undetectable platinum DNA adducts was associated with longer median OS (60.3 versus 36.3 months; p=0.029) and a reduced risk of death (HR=0.607, 95% CI=0.385–0.958, p=0.032) for women with detectable versus undetectable adducts, but was not associated with PFS or an independent prognostic factor for OS [21]. ERCC1 mRNA expression, categorized as positive versus negative, in post-treatment peripheral blood leukocytes was not associated with either PFS or OS [21].

GOG 175 Protocol

GOG 175 was a CTEP-sponsored, intergroup randomized phase III trial with the Southwest Oncology Group by Mannel and colleagues that will compare PFS, OS and the frequency of adverse events in women with previously-untreated, histologically-confirmed, completely-resected stage IA grade 3 (or clear cell), stage IB grade 3 (or clear cell), stage IC or stage II EOC who were treated with intravenous paclitaxel and carboplatin every 3 weeks for 3 cycles followed by a 1-hour intravenous infusion of 40 mg/m² every week for 24 weeks versus observation for 24 weeks (Table 3). Archival tumor specimens recovered from the GOG 175 virtual tissue bank underwent array comparative genome hybridization (aCGH) analysis to compare copy number aberrations (CNAs) in DNA repair genes including the Fanconi anemia complementation group (FANC) and RAD51 families with the rest of the genome [22,23].

GOG 172/182 Protocols

A series of TR studies are leveraging specimens and resources from the GOG 172 and 182 protocols. GOG 172 was a CTEP-sponsored, randomized phase III trial by Armstrong and colleagues which demonstrated that the intraperitoneal versus the intravenous cisplatin and paclitaxel regimen resulted in improvements in PFS (p=0.05) and OS (p=0.03) with 5.5 and 15.9 month longer median PFS and OS, respectively, worse adverse effects (p≤0.001) and quality of life before cycle 4 and 6-weeks after treatment completion, but not 1-year after treatment completion in women with previously-untreated, histologically-confirmed, optimally-resected, stage III EOC who underwent adequate surgical staging and had <1 cm residual disease (Table 3) [24]. GOG 182 was a five arm, CTEP-sponsored, international intergroup phase III randomized trial with Australia and New Zealand GOG, Medical Research Council United Kingdom; Istituto Mario Negri, Southwest Oncology Group, Eastern Cooperative Oncology Group, North Central Cancer Treatment Group, Cancer and Leukemia Group B, National Surgical Adjuvant Breast and Bowel Project and Radiation Oncology Group by Bookman and colleagues which demonstrated that addition of a third cytotoxic agent in a triplet or sequential doublet regimen following optimal or suboptimal cytoreductive surgery for the treatment of advanced stage EOC or PPC provided no added benefit to PFS or OS compared with standard paclitaxel and carboplatin in women with previously-untreated, histologically-confirmed stage III or stage IV EOC or PPC with either optimally-resected disease (≤1 cm residual disease) or suboptimally-resected disease (>1 cm residual disease) following initial surgery (Table 3) [25].

Thus far, mutations in BRCA1 were identified in DNA extracted from a buffy coat specimen from 16 (5%) of GOG 172 patients [26]. A thorough evaluation of the type and distribution of mutations and common variations observed in BRCA1 and associations with clinical outcome in the GOG 172 cohort are currently underway. BRCA1 promoter methylation was observed in specific CpG sites in sporadic EOC including women who participated in GOG 172 and transcript expression of BRCA1 by RT-PCR was significantly lower in women with a methylated compared with an unmethylated BRCA1 promoter [27]. Low IHC expression of BRCA1, defined as <10% positive tumor cells, was associated with advanced stage (p<0.001), serous histology (p<0.001), better PFS (p=0.03) and OS (p=0.006), and a reduced risk of disease progression (HR=0.64, 95% CI=0.42–0.96) and death (HR=0.51, 95% CI=0.32–0.83) [28]. Results of additional IHC studies of BRCA1 by route of administration in GOG 172 cases will be presented at the 2010 Society of Gynecologic Oncologist (SGO) Meeting. Denaturing high-performance liquid chromatography, sequence analysis, and single nucleotide polymorphism genotyping by pyrosequencing for the CHEK2 gene demonstrated that variations in CHEK2 do not appear to make a significant contribution to the pathogenesis of sporadic EOC in the United States [29]. In the GOG 172 cohort, the codon 118 polymorphism in ERCC1 was not significantly associated with disease progression or death whereas the C/A or A/A versus C/C genotypes in C8092A in ERCC1 were associated with 6- and 17-month shorter median PFS and OS, respectively, and an increased risk of disease progression (HR=1.44, 95% CI=1.06–1.94, p=0.018) and death (HR=1.50, 95% CI=1.07–2.09, p=0.018) [30]. Subset analysis stratified by treatment regimen demonstrated a distinct PFS and OS advantage for women with the C/C compared with either the C/A or A/A genotypes in C8092A in ERCC1 in women randomly allocated to the IP treatment arm [30]. The associations between codon 118 and C8092A polymorphisms in ERCC1 and PFS and OS in the GOG 182 cohort are under active investigation [31]. Results of an IHC study of ERCC1 in GOG 172 will be presented at the 2010 SGO Meeting. Additional studies are also underway to evaluate the relationship between common polymorphisms in DNA repair genes (BRCA1 and BRCA2 [32] as well as XRCC1), efflux pumps (ABCB1, ABCC2 and ABCG2 [33]) and detoxicification enzyme (GSTρ1) and measures of clinical outcome including PFS, OS and common severe adverse effects, and to perform genome-wide association studies in the GOG 172 and 182 cohorts.

GOG 198 Protocol

GOG 198 was a CTEP-sponsored, randomized phase III trial by Hurteau and colleagues that evaluated oral daily thalidomide versus tamoxifen and the prognostic relevance of serum VEGF in women with biochemical-recurrent epithelial ovarian, fallopian tube or primary peritoneal cancer (Table 3). Investigations are also underway to evaluate the serial changes in VEGF-A during treatment and other angiogenic markers and cytokines in this setting.

GOG 218 Protocol

GOG 218, sponsored by CTEP and Genetech, is a three-arm, randomized, doubleblinded, placebo-controlled phase III trial by Burger and colleagues that will evaluate paclitaxel, carboplatin and placebo plus placebo maintenance versus paclitaxel, carboplatin and bevacizumab plus placebo maintenance versus paclitaxel, carboplatin and bevacizumab plus bevacizumab in women with previously-untreated epithelial ovarian, primary peritoneal or fallopian tube cancer who underwent adequate surgical staging and cytoreduction with FIGO stage III disease with any gross or palpable residual disease or FIGO stage IV disease (Table 3). A series of TR studies are leveraging specimens and resources including those designed to validate published genomic profiles, angiogenic markers, single nucleotide polymorphisms (SNPs), cell-free DNA, EGFR family members/ligands, markers of VEGF-targeted therapy as well as new markers and profiles including those identified by The Genome Atlas Project in ovarian cancer. New discovery and validation efforts are also built into GOG 218 including a number of genomic profiling, SNP and genome-wide association studies.

GOG 136 Protocol

GOG 136, is a CTEP-sponsored, specimen banking protocol by Cibull and colleagues for women undergoing a surgery for a gynecologic malignancy or a prophylactic oophorectomy which has supported a number of note worthy studies including those reported by Zorn et al. [42,43] and Beck et al. [44–46].

GOG 143/144 Protocols

The CTEP-sponsored GOG 143 protocol demonstrated that in an unselected, clinic-based series of ovarian cancer cases, 12 patients exhibited protein truncation mutations in BRCA1 and another 12 displayed BRCA1 mutations of unknown significance [47]. GOG 144 was a CTEP-sponsored protocol for women with familial ovarian cancer that screened 26 women for mutations in BRCA1/BRCA2 and detected 12 deleterious alterations; 8 in BRCA1 and 4 in BRCA2 [48]. Mutations in BRCA1 or BRCA2 were present in about 50% of ovarian cancer patients with at least one first-degree relative with disease, and in 70% of patients with two or more relatives with ovarian/breast cancer (p=0.0002) [48].

GOG 199 Protocol

The GOG 199 protocol is a prospective, international, two-cohort, non-randomized study by Greene and colleagues, in the Clinical Genetics Branch an NCI Intramural Research Program, the GOG, and the Cancer Genetics Network, in women at genetic risk of ovarian cancer, who undergo risk-reducing salpingo-oophorectomy (RRSO) or screening [49]. A series of TR studies leverage specimens and resources from the GOG 199 protocol including GOG 246^CIMBA-4 which demonstrated that the minor allele of the rs3817198 SNP in LSP1 was associated with increased breast cancer risk only for BRCA2 mutation carriers (HR=1.16, 95%CI=1.07–1.25, p-trend=2.8×10⁻⁴) whereas the rs13387042 SNP at 2q35, but not the rs13281615 SNP at 8q24, was associated with breast cancer risk for BRCA1 mutation carriers (HR=1.14, 95% CI=1.04–1.25, p=0.005) and BRCA2 mutation carriers (HR=1.18, 95%CI=1.04–1.33, p=.0079) [50]. In addition, GOG 8008^CIMBA-5 will evaluate the association between SNPs in rs16942 in BRCA1, rs2237060 in RAD50, SNP3 and rs2241193 in IGFBP5 and breast cancer risk in BRCA1/BRCA2 mutation carriers. GOG 8009 and 8010 represent genome-wide association studies examining modifiers of breast cancer risk in BRAC1 and BRCA2 mutation carriers, respectfully. Additional studies are also under development including CIMBA-6 and CIMBA-7.