Chancellors Distinguished Chair in Biomedical Sciences and Endowed Professor
Professor and Director of Molecular Oncology, Pathology and Laboratory Medicine
Deputy Director, The University of Kansas Cancer Center
Director, Biospecimen Shared Resource
Kansas Bioscience Authority Eminent Scholar
Lab Phone: 913-945-6373
Godwin and members of his group have been involved in both basic and translational research for many years, filling several important niches within the translational research community as a whole (e.g., early detection, predictive biomarkers and clinical trials, molecular pathology, biosample ascertainment). The central theme of his work centers on the idea of obtaining a molecular definition of a tumor to define its treatment-sensitive elements. As such, he has a long-standing interest in the field of cancer genetics and molecular targeted therapies. His studies primarily involve three different types of cancer: breast, ovarian, and gastrointestinal stromal tumor (GIST). His group uses these models to investigate how to improve patient care.
The incidence of epithelial ovarian cancer has remained stable over the last 20 years, with approximately 20,200 cases estimated to occur annually, and a lifetime risk in the general population of 1.4%. Aggressive debulking surgery to remove the primary tumor and its metastatic implants has long been the accepted initial step in the management of ovarian cancer. The achievement of optimum cytoreduction at the end of the initial surgery has been shown as a favorable prognostic factor for the disease course. It has been recently demonstrated that clinical response may not correlate with survival due to the preferential survival of chemotherapy resistant tumor-initiating cells (also referred to as cancer stem cells, cancer progenitor cells, and cancer initiating cells) that eventually grow and repopulate the tumor. Our group studies the origins of ovarian cancer, the role of the extracellular environment (stromal/epithelial/ECM interactions) in regulating tumor cell growth and the use of molecular targeted therapies for the treatment of this disease. Below are examples of ongoing studies in these areas.
The majority of advanced epithelial ovarian, primary peritoneal and fallopian tube cancers are invasive serous adenocarcinomas; recently classified as Type II epithelial ovarian cancers. Identification of morphologic or molecular precursors for Type II ovarian cancer would be paramount to the development of effective prevention strategies. Preliminary studies in hereditary breast-ovarian cancer (HBOC) kindred suggest that Type II ovarian cancer may arise from dysplastic tubal epithelium (tubal intraepithelial carcinoma, TIC) and are related to early mutational loss of TP53. Our preliminary data show strongly stained p53 in both areas of TIC and the concurrent ovarian carcinoma. We hypothesize that TIC represents a precursor of invasive disease for Type II tumors in HBOC (and perhaps sporadic cases). In order to test this hypothesis, we are using molecular genetic approaches to evaluate concurrent TIC and ovarian adenocarcinoma to determine additional molecular alterations that lead to disease development and to determine whether sporadic cases of ovarian cancer share these molecular features. If our data support the model of TIC as an origin of Type II serous tumors, this would justify development of in situ imaging technologies specific to morphologic or molecular changes associated with TIC in the prevention of Type II serous adenocarcinomas.
The tumor microenvironment is complex and contains non-cancerous stromal cells, such as fibroblasts, immune cells, and endothelial cells and products secreted from these cells that intermingle with the tumor cells. Since the trigger for tumor growth and spread may actually come from the stroma, these studies explore how to best use therapies that target the stromal components. We are utilizing 3 dimensional (3D) models to study how the tumor microenvironment supports ovarian tumor cell growth. We have reported that the basement membrane (a thin layer of extracellular proteins that anchors epithelial tissue to underlying connective tissue and provides mechanical structure, separates different cell types, and signals for cell differentiation, migration, and survival) is disturbed very early in the initiation of ovarian cancer. Using this 3D model we have shown that different ovarian cancer cell lines migrate and invade in different ways and these cancer cells change their phenotype by acquiring tumor stem cell characteristics depending on signals provided by their immediate surroundings. Although it is believed that a normal microenvironment has a negative effect on tumor growth, the tumor-associated microenvironment actively contributes to tumor development and progression. Our research is focusing on ways to alter how ovarian tumor cells interact with their microenvironment as a means to interfere with tumor development.
We have utilized recent technological advancements in the field of RNAi and high-throughput screening to perform several types of RNAi screens to identify genes essential for ovarian tumor growth and to identify genes that help sensitize tumor cells to molecular targeted therapies. As a first step we employed a druggable library of 24,088 siRNAs targeting ~6,022 genes (pool of 4 siRNAs/gene) and screened across A1847 cells. From this screen we identified 300 unique human genes that were considered potential targets for drug development and represent essential genes for tumor growth and survival. A custom siRNA library was derived and this library has been screened across a panel of ovarian cancer and non-tumorigenic human ovarian surface epithelium cell lines. For each tumor and non-tumor cell line a gene signature was established. A total of 50 common genes across the 7 ovarian cancer cell lines were discovered and found to induce varying degrees of cytotoxic effects as assessed by a cell viability assay. Of these 50, seven genes were found to have increased cytotoxicity for the tumorigenic cell lines but not for the non-tumorigenic cell lines. These genes were validated by deconvolution of the initial pool of 4 siRNAs targeting each gene and evaluating the effectiveness of individual siRNAs in all of the cell lines. Quantitative RT-PCR was performed to further assess the specificity of the individual siRNAs in all of the cell lines. We will also evaluate the functional significance of these genes in primary cultures obtained by processing ascites from ovarian cancer patients as a basic step towards establishing personalized medicine. As siRNA therapy holds a promising future, we believe that these studies lay the foundation to develop novel molecular approaches and targets that may ultimately improve the treatment of patients with ovarian cancer.
At the time of presentation most epithelial ovarian cancers are no longer dependent on single genetic determinants for growth and/or survival. The overall response to oncogene-targeted therapies used as single agents is not highly significant in the general population, and therefore, treatment of ovarian cancer poses particularly formidable challenges. We hypothesize that targeted agents will become potent when used in combinations that simultaneously block multiple oncogenic pathways, where such simultaneous blockade will reduce the ability of cancer cells to select resistance mutations. Therefore, we have used second-site siRNA lethality screens to help identify critical pathways to target in combination with novel biologics. Our studies have used bioinformatics to devise a tryrosine kinase-centered network, and probed a corresponding siRNA library targeting 638 genes in synthetic lethality screens. Using this library, we have identified a number of genes that synergize with a multi-kinase inhibitor targeting the SRC-family of kinases among other kinases, all of which are implicated in the formation and maintenance of cancer. Our goal is to identify critical pathways which sensitize ovarian cancer cells to this novel biologic that was recently evaluated in a Phase II clinical trial (GOG-0170M) to treat patients with recurrent ovarian cancer. We have completed the siRNA screening and its subsequent validation and are now refining this list of sensitizing genes by performing additional screens using alternate biologics and other tumor cell types to identify drug-specific and ovarian-specific sensitizers. We have analyzed DNA microarray data sets to explore the expression patterns of transcripts for genes identified through our screens in patient samples to assess their clinical relevance. Furthermore, we have access to clinical samples from GOG-0170M (pre- and post-treatment serum samples and tumor tissue) for evaluation of the prognostic value of these dasatinib-sensitizing genes. We have correlated the sensitivity of each cell line to dasatinib with the expression of each of the dasatinib-sensitizing genes using quantitative RT-PCR. We have performed in vitro drug-drug combination experiments using the Chou-Talalay combination index method to evaluate drug combinations and have found strong synergy in several two-drug combinations at various molar ratios of the two drugs across a panel of ovarian cancer cell lines. For highly promising drug combinations, we have initiated in vivo orthotopic xenograft studies to evaluate the efficacy of combination therapy in preventing ovarian tumor growth (see section on “Evaluation of therapeutic agents in treating human ovarian cancer”). The sensitizing genes identified from these studies represent potential targets for combination therapies with biologics in clinical trials. We view this work as ultimately enabling the next-generation approach for cancer therapies: combining targeted drugs in coherently designed multi-agent combinations for use in the clinic.
Our group uses various in vitro and in vivo models to assess the activity of novel anti-tumor drugs alone or in combination for the treatment of ovarian cancer, including 2D- and 3D-cell culture models, subcutaneous/intraperitoneal/intrabursal tumor xenograft models, and bioluminescence-based animal imaging. We are examining many biologics including cetuximab, panitubimab, gefitinib (EGFR inhibitors), dasatinib (SRC family kinases inhibitor), and AMG706 (VEGFR inhibitor) in vitro and in vivo. In addition to drug evaluations, our studies are also focused on understanding the mechanisms of action at the cellular level. The goals of these therapeutic studies are to identify important drug targets, promising agents, and treatment combinations that can ultimately help improve the clinical outcome of human ovarian cancer treatment.
BREAST CANCER SUSCEPTIBILITY:
Women who carry mutations in the BRCA1 and BRCA2 (BRCA1/2) genes have a substantially increased risk of developing breast and/or ovarian cancer as compared to the general population. However, risk estimates for breast cancer range from 20-80% suggesting the presence of genetic and/or environmental effect modifiers. We have recently reported that loss of allele specific expression (also referred to as allelic imbalance or AI) of BRCA1 is associated with increased breast cancer risk and demonstrated that these expression patterns can be transmitted by Mendelian inheritance, suggesting that alternate mechanisms, other than deleterious coding mutations, may contribute to breast cancer. We are also participating in the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA), an international consortium that is evaluating large cohorts (>20,000) of BRCA1 and BRCA2 mutation carriers for genetic modifiers of breast cancer risk. Our group continues to explore additional high-penetrant breast cancer susceptibility genes. We have shown that BRCC36 is aberrantly expressed in breast cancer, is mutated in cancer prone kindreds, and plays an important role in the regulation of the ubiquitin E3 ligase activities of the BRCA1 containing complex (BRCC). We have also recently initiated studies to identify proteomic markers of benign breast disease (BBD) that can be used to better predict a woman’s risk of developing invasive breast cancer. Below are examples of ongoing studies in these areas.
We have previously reported a novel multiprotein complex, termed BRCC, containing seven polypeptides including BRCA1, BRCA2, BARD1, and RAD51 (Dong et al., Mol. Cell 12:1087, 2003). BRCC is an ubiquitin E3 ligase complex exhibiting an E2-dependent ubiquitination of the tumor suppressor p53. In this multiprotein complex, one of these proteins, referred to as BRCC36, was found to be associated with BRCA1 and BRCA2 and was shown to play an important role in the regulation the ubiquitin E3 ligase activity of BRCC. We have shown that cancer-associated mutations in BRCA1 abrogated the association of BRCC36 with BRCC and BRCA1. Furthermore, reconstitution of a recombinant four-subunit BRCC complex containing BRCA1/BARD1/BRCC45/BRCC36 revealed an enhanced E3 ligase activity compared to that of BRCA1/BARD1 heterodimer. In addition, we have reported over-expression of BRCC36 in the majority of breast cancer cell lines and invasive ductal carcinomas. The mechanism and consequences of abnormal BRCC36 expression in breast cancer are presently unknown. To further elucidate the functional consequence of abnormal BRCC36 expression in breast cancer, we performed in vivo silencing studies using small interfering RNAs (siRNA) targeting BRCC36 in the MCF-7 breast cancer cell line. Knock-down of BRCC36 alone did not affect cell growth, but when combined with ionizing radiation (IR) exposure, led to an increase in apoptotic cells when compared to the siRNA control group. Immunoblot analysis showed that inhibition of BRCC36 had no effect on activation of ATM, expression of p21 and p53, or BRCA1- BARD1 interaction following IR exposure (Figure, left panel). Importantly, BRCC36 depletion disrupted IR-induced phosphorylation of BRCA1. Immunofluorescent staining of BRCA1 and g-H2AX indicated that BRCC36 depletion prevented the formation of BRCA1 nuclear foci in response to DNA damage (Figure, right panel). These results imply that down-regulation of BRCC36 expression impairs the DNA repair pathway activated in response to IR via abolishing BRCA1 activation and thereby to sensitizes breast cancer cells to IR-induced apoptosis.
Although breast cancer rates have been declining in recent years, the number of women receiving benign breast diagnoses is increasing. Some benign lesions have a higher relative risk of subsequent breast cancer development than others; however, it is unknown which lesions are most likely to progress to invasive breast cancer. As a proof of principle study, laser capture microdissection was employed to isolate histologically identified regions of BBD (i.e, ductal hyperplasia; DH) and normal breast epithelium from mastectomy specimens. Using 2D gel electrophoresis and mass spectrometry, we identified differentially expressed proteins and demonstrated that a set of these proteins could identify a subpopulation of DH cells exhibiting columnar cell changes (CCC). This is clinically relevant given that cells with atypical CCC have characteristics similar to those observed in atypical DH (ADH) and well-differentiated in situ and invasive carcinomas; thus atypical CCC may represent nonobligate precursors to neoplasia. We have since expanded our studies using more sensitive proteomic technologies to compare ADH from women who remain disease free or subsequently develop invasive disease to those proteins from normal tissue specimens. We compared the proteins and generated a heat map to illustrate proteins that are upregulated in ADH tissues associated with cancer compared to those proteins in normal or ADH tissues not associated with cancer. Together, these markers may prove to be clinically relevant in identifying regions of BBD and in assigning prognostic significance in biopsy samples. We are collaborating with members of the Early Detection Research Network (EDRN) to access clinical specimens to test the value of our markers in predicting women diagnosed with ADH who are at increased risk of developing invasive breast cancer and who might benefit from risk reduction with the use of chemoprevention agents such as Tamoxifen.
The goals of these studies are to evaluate the utility of protein biomarkers in blood at identifying women at increased risk of developing breast or ovarian cancer and to develop a panel of biomarkers to be used to detect these cancers early in their clinical course. To date, we have developed approximately 10 custom bead-based suspension immunoassays (luminex assays) in which fluorescent polystyrene microspheres are coated with antibodies towards the diagnostic biomarkers identified through previous gene microarray studies. We also have in hand approximately 100 commercially available bead-based assays. In many instances, these assays can be multiplexed and provide a fast and effective means to analyze hundreds of biomarkers in precious patient serum or plasma samples. We are currently evaluating our panel of >100 biomarkers by screening blood serum and plasma samples from disease-free women who have a normal level of risk of developing breast and/or ovarian cancer and from women who have an elevated level of risk of developing these cancers due to either being carriers of BRCA1 and/or BRCA2 mutations or by having a family history of these diseases. These tests are being performed to ascertain if single or multiple markers can be used to detect elevated risk in an individual. We have also included blood samples from women who have sporadic breast and/or ovarian tumors to determine the capacity of our marker panel for disease diagnosis in addition to risk diagnosis. Preliminary data show some promising markers for disease diagnosis in ovarian cancer patients as observed by the statistically significant elevation of these markers in women with ovarian cancer relative to healthy women and in women with benign cases. Additional improvements in disease diagnosis were also found when our markers were combined with CA125, a traditional diagnostic marker. Our studies are likely to result in the identification of several single or multi-marker blood-based protein biomarkers which provide the means to assess breast and/or ovarian cancer risk and early detection in women.
TUMOR-DERIVED EXOSOMES AS POTENTIAL BIOMARKERS FOR CANCER:
Tumor-derived microvesicles or exosomes (TDE) are abundantly found in the plasma and malignant effusions derived from cancer patients. Growing evidence supports the view that tumors constitutively shed exosomes with pleiotropic immunosuppressive effects that are protective and supportive of the tumor to facilitate escape from lymphocyte immunosurveillance. These effects range from regulation of tumor growth, to invasion, angiogenesis and metastasis. Several tudies support a role for exosomes in adapting the host microenvironment to allow escape from immune surveillance via stimulation of angiogenesis and metastasis which suggests that tumors may use exosomes to keep the host immune system under control without a direct interaction with host immune cells. Moreover, TDE are used by tumor cells to remodel the tumor microenvironment into a tumor supportive milieu. The constitutive release and enrichment of certain proteins into TDE prompts consideration of the use of TDE as diagnostic tools in cancer. The use of proteomic analysis on malignant effusion-derived exosomes from various sources has increased our knowledge on exosome protein composition and likewise, our understanding of the role of exosomes in vivo. Exosome-derived proteome studies have been conducted in a number of cancer-derived exosomes including: mesothelioma, melanoma, gastric carcinoma, breast carcinoma, ovarian, prostate, malignant pleural effusions, brain, and colorectal. Since exosomes are known to represent an important mechanism of intercellular communication, numerous studies have been performed which reveal the presence of common subsets of proteins in exosome preparations from different biological fluids, some being unique to the tissue from which they are derived. All of these isolated exosomal proteins constitute a "cancer signature" which may help in improving the diagnosis and follow-up of cancers. We are isolating and characterizing exosomes from blood or ascites fluids as a potentially effective alternative for diagnosis, prognosis, and treatment of cancer.
PATHOGENESIS AND MOLECULARLY TARGETED THERAPY OF GASTROINTESTINAL STROMAL TUMORS (GISTs):
Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the digestive tract, with an estimated annual occurrence of 3,300-6,000 in the United States. GISTs are believed to arise from the Interstitial Cells of Cajal (ICCs), the pacemaker cells of the gut, or from interstitial mesenchymal precursor stem cells. GISTs express and are clinically diagnosed by immunohistochemical staining of CD117, the 145 kDa transmembrane glycoprotein KIT. The most common primary sites for these neoplasms are the stomach (60-70%), followed by the small intestine (25-35%). The majority (~80%) of GISTs possess gain-of-function mutations in KIT in either exons 9, 11, 13 or 17, causing constitutive activation of the kinase receptor, whereas smaller subsets of GISTs possess either gain-of-function mutations in PDGFRA (exons 12, 14, or 18) (~5-8%) or no mutations in either KIT or PDGFRA (~12-15%). Imatinib mesylate (IM), an oral drug that inhibits the KIT/PDGFRA is very effective at controlling metastatic disease and preventing recurrence after initial surgery. Unfortunately, in patients with advanced disease, IM stops controlling disease after approximately two years. Our group is studying what happens to GIST cells when they are treated with imatinib and what leads to clinical resistance. Our initial studies identified genetic markers that could predict the response of patients with metastatic/recurrent GIST to imatinib and current studies are focusing on genomic and proteomic changes associated with the pathogenesis of GIST and response to molecular targeted therapies. We have also shown an important role for IGF signaling in adult and pediatric GISTs that lack activating kinase mutations. Based on these studies clinical trials are being developed with the goal of ultimately eradicating this disease. Below are examples of ongoing studies in these areas.
A subset of GISTs lack gain-of-function mutations in the KIT and PDGFRA genes. These GISTs tend to be less responsive to imatinib-based therapies and have a worse prognosis than mutation positive GISTs. We have shown that the aberrant expression of IGF-1R, as opposed to oncogenic RTK-mutations, may be the driving event in these GISTs. This observation suggests an alternative and/or complementary therapeutic regimen in the clinical management of all GISTs, especially in the subset of tumors that respond less favorably to imatinib. Immunoblotting of GIST specimens demonstrated that IGF-1R was present in all tumors but was markedly overexpressed (10- to 30- fold) in GISTs lacking RTK-mutations as compared to mutant GISTs. IGF-1R overexpression in these tumors was confirmed by IHC analysis using an extended set of GIST specimens. In addition, qPCR analysis demonstrated that IGF-1R mRNA is overexpressed ~17- to 19-fold in GISTs lacking RTK-mutations versus mutant GISTs (P = 0.0013). Interestingly, IGF-1R is constitutively activated in all GISTs irrespective of KIT/PDGFRA mutational status. Using a genomic-based qPCR assay, we found that the IGF-1R gene copy number was increased in 70% of GISTs lacking mutation in KIT and PDGFRA (copy number range, 2.5 to 4 copies/tumor cell), as compared with 28% of mutant GISTs (P = 0.04). Fluorescent in situ hybridization analysis confirmed the IGF-1R amplification (3 to 10 copies/tumor cell). Inhibition of IGF-1R signaling with NVP-AEW541 (Novartis, Basel, Switzerland), a specific inhibitor of IGF-1R or down-regulation of expression of IGF-1R with siRNA led to cytotoxicity and induced apoptosis in GIST cells. Combination of NVP-AEW541 and imatinib in GIST cells induced a strong cytotoxicity response. These findings are particularly exciting given the number of agents targeting IGF-1R that are currently being tested in clinical trials. We are currently testing neutralizing monoclonal antibodies and small molecule inhibitors specific for IGF-1R alone and in combination with imatinib, with the hope of improving current therapeutic modalities in GIST patients. Clinical trials using IGF-1R-targeted therapies for imatinib-refractory GIST patients, initially focusing on adult and pediatric GIST patients lacking KIT or PDGFRA mutations are in development within our sarcoma program.
In these studies we have shown that not all GIST that lack mutations in KIT or PDGFRA, the so-called “wild-type” tumors, are alike. Using whole-genome SNP-arrays we have been able to obtain high-resolution maps of the GIST’s genome. We used high-density SNP arrays containing ~1.8 million markers genome-wide to analyze GISTs lacking KIT/PDGFRA mutations as well as mutant GISTs. As expected, the mutant GISTs exhibited multiple regions of chromosomal copy number (CN) loss (mean ~ 6.2 per sample) as well as CN gains (mean ~ 2.4 per sample). Importantly, our studies found that the majority of KIT/PDGRFA mutation-negative, both pediatric and adult GISTs which are clinically more resistant to imatinib-based therapies, had either no broad-based CN changes, or CN change across a single chromosome arm. We refer to these tumors as KIT/PDGFRA mutation negative/genome stable GISTs, while the remaining tumors (including both KIT/PDGFRA mutation positive and some mutation negative GISTs) are genome unstable. We also identified smaller (0.2-6 Mb) or focal regions of CN aberration. We have compiled a list of ~80 of these focal CN-variant regions, including 4 regions identified within the mutation-negative, genome stable GISTs. These specific focal changes may be important in the oncogenesis of genome-stable GISTs. Furthermore, as discussed above we have shown that these KIT/PDGFRA mutation negative/genome stable GISTs greatly overexpress IGF1R relative to other GISTs. Importantly, in these studies we have shown the connection between the mutation negative GISTs that have a stable genome and overexpressing of IGF1R. Clinical trials are being developed to explore IGF1R as a target based on these and other molecular studies. Finally, we have shown that a portion of previous classified “wild-type” GISTs that demonstrate genome instability have mutations in the BRAF gene. However, the genetic changes in GISTs without KIT/PDGFRA/BRAF mutations is not known, so we are currently using second-generation sequencing approaches to further explore the GIST genome in these patients to uncover additional clues to help define the origins of these tumors (see Figure “Not all GISTs are created equal”).
Despite initial efficacy of imatinib in most GIST patients, many experience primary and secondary drug resistance. Therefore, clinical management of GIST may benefit from further molecular characterization of tumors before and after IM treatment. As part of a recent Phase II Trial of neoadjuvant/adjuvant IM treatment for advanced primary and recurrent operable GISTs (RTOG-S0132), gene expression profiling using oligonucleotide microarrays was performed on tumor samples obtained before and after IM therapy. Patients were classified according to changes in tumor size after treatment based on CT scan measurements. Gene profiling data were evaluated with SAM analysis to identify differentially expressed genes (in pre-treatment GIST samples). Based on SAM analysis, thirty-eight genes were expressed at significantly lower levels in the pre-treatment samples of those tumors that significantly responded to 8 to 12 weeks of IM, i.e., >25% tumor reduction. Eighteen of these genes encoded KRAB domain containing zinc finger (KRAB-ZNF) transcriptional repressors. Interestingly, ten KRAB-ZNF genes mapped to a single chromosomal locus, and a subset of these predicted likely response to IM-based therapy in a naïve panel of GISTs. We designed a custom siRNA library targeting these genes and used a mid-throughput siRNA synthetic lethal screening approach to evaluate the ability of each to sensitize GIST cells to IM. We found that modifying expression of genes within this predictive signature can enhance the sensitivity of GIST cells to IM. These studies are very exciting given that a very high percentage of the siRNAs can sensitize GIST cells to IM, suggesting that many members of the gene signature may not only have predictive value but functional relevance to IM’s activity in vivo.
Our group is actively involved in evaluating clinical trial samples for predictive markers of drug response. As such, Dr. Godwin is the Translational Science Co-chair or the lead translational researcher for several Gynecologic Oncology Group (GOG), Eastern Cooperative Oncology Group (ECOG), and Southwestern Oncology Group (SWOG) clinical trials evaluating molecularly targeted agents, such as AMG 102, AMG 386, AMG 479, AMG 706, cetuximab, dasatinib, everolimus, enzastaurin, gefitinib, lapatinib, RAD001, sorafenib, temsirolimus, and VEGF-TRAP. Through the Clinical Molecular Oncology Laboratory (Director, Godwin) at KUMC we are evaluating clinical trial samples for predictive markers of response to therapy. For example, our group demonstrated in a phase II clinical trial that certain serologic markers were predictive of disease control when women with recurrent epithelial ovarian carcinoma were treated with cetuximab alone. We have also recently evaluated a series of tumor vascular proteins identified through gene profiling studies as serum prognostic and diagnostic markers and are assessing the predictive value in large cohorts of patient samples. In a collaborative phase II clinical trial of patients with colorectal cancer we demonstrated that tumors with high gene expression levels of epiregulin, amphiregulin, and/or wild-type K-RAS were more likely to have disease control with cetuximab treatment. This study led to numerous other confirmatory studies and a recent Provisional Clinical Opinion statement by the American Society of Clinical Oncology that all patients with metastatic colorectal cancer who are candidates for anti-EGFR therapy (i.e., cetuximab and panitumumab) have their tumors tested for K-RAS mutations. In general, our translational correlatives are aimed at extracting proteomic and genomic information from patients’ blood and/or tumor tissue samples prior to and during the course of treatment to identify potential biomarkers associated with clinical activity and/or predictive of outcome. The ultimate goal of these efforts is to better identify patients more likely to benefit from a particular therapy. A targeted approach to patient selection could potentially improve survival, spare needless toxicity, and reduce expenses associated with futile therapy. Our clinical research efforts will continue to explore and expand the use of molecular pathology in personalizing patient care.