Ovarian cancer: prevention, detection, and treatment of the disease and its recurrence. Molecular mechanisms and personalized medicine meeting report.

Cancer stem cells are self-renewing cells capable of initiating tumorigenesis, recurrence and metastasis. Patricia K. Donahoe, MD proposed that at diagnosis, ovarian cancers have both stem and non-stem cell populations which must be differentially treated in order to ensure both cell populations are effectively targeted. An ovarian cancer stem cell-enriched population marked by three markers conserved across primary cancers and normal Fallopian tube fimbria (CD44, CD24 and Epcam) and by negative selection for Ecadherin has been identified. ( 20 ) These cells comprise less than 1% of cancer cells, have increased colony formation and shorter tumor-free intervals in vivo . Moreover, they are resistant to but stimulated by standard chemotherapeutic agents. They are inhibited by Mullerian inhibiting substance (MIS). These data support the use of combination of markers to develop targeted “tumor stem cell therapies” individualized to the specific tumor-initiating population identified in a lesion. Further work is needed to identify and understand mechanistic differences between different putative stem cell populations that could serve as therapeutic targets. Understanding the mechanisms underlying stem cell self-renewal or differentiation can also shed light on how these cells contribute to chemoresistence and whether modulation of these mechanisms can impact patient outcome. Kunle Odunsi, MD, PhD discussed vaccine development. An effective immunotherapy will generate a robust, clonal expansion of T-cells that can differentiate into both effector cells with capacity to kill tumor targets and memory cells with capacity for recall response. Identifying targets for immune recognition is the first step in vaccine development. NY-ESO-1, a tumor-specific antigen, is one such target. Vaccination with NY-ESO-1 epitope induces integrated humoral CD4+ and CD8+ T cell response with the capacity to recognize tumor targets. ( 21 ) However, as the time from vaccination increases, functional immune response decreases. Thus, while the vaccine generates substantial effector T-cells, it does not generate a high frequency of memory T-cells. Understanding the mechanisms underlying this phenomenon as well as identifying agents that can influence the type of T-cells generated are critical in designing vaccines with durable protection. mTOR blockade may be one possibility. In In vitro and animal studies, mTOR blockade influences T-cell differentiation towards memory cells, suggesting that including an agent that blunts mTOR may improve vaccine efficacy. However, even when an effective long-lived functional T-cell response is generated, most subjects will relapse. Improved understanding of how tumors escape immune attack is needed. Mechanisms underlying immune system escape and exploiting those mechanisms to enhance vaccine efficacy was addressed by Pawel Kalinksi, MD, PhD . One way tumors escape the immune system is by creating a highly immunosuppressive environment through both the production of MDSCs and the suppression of type-1 immune effector cells. Local production of prostaglandin E 2 (PGE2) and COX2 appear to play a role. A COX2-PGE2 positive feedback loop controls CXCR4/CXCL12-guided accumulation of MDSCs as well as induction and stability of the immunosuppressive MDSC phenotype and function. PGE2 and COX2 also suppress induction and function of type-1 effector cells (Teff) and selectively inhibit production of Teff-attracting chemokines. Disrupting this feedback loop by suppressing COX2 restores local immunosurveillance in vitro . These findings have important implications for vaccine development. Cancer vaccine adjuvants can amplify PGE2-driven suppressive events when used alone. However, when combined with COX2 inhibitors, they induce Teff-attracting chemokines and also suppress MDSC and Treg attracting chemokines in tumor but not marginal tissue ( 22 ). This suggests that conditioning the tumor microenvironment by disrupting the PGE2-COX2 feedback loop prior to vaccination may enhance vaccine efficacy. Studies to test that hypothesis are needed. William Zamboni, PhD presented data on the translational development of nanoparticles for drug delivery. Relative to non-encapsulated drugs, nanoparticle encapsulated drugs can have prolonged circulation and may selectively accumulate in different organs and blood components. They may also have different cellular distributions. Moreover, the size and surface properties of the encapsulating nanomaterial (“carrier”) may lead to greater accumulation in tumors as a result of the enhanced permeability and retention (EPR) effect. Together these data suggest that nanoparticle encapsulation can alter the pharmacokinetics and distribution of a drug in ways that can improve drug efficacy while reducing toxicity. ( 23 ) However, recent studies suggest nanoparticle encapsulation results in greater pharmacokinetic variability – the individual rate of clearance of an encapsulated drug is highly variable between patients. Some agents can have 10 to 100 times the variability of their non-encapsulated counterparts. The mechanisms underlying this variability are poorly understood, although emerging data suggest that it may be due to the ability of cells to recognize the carrier and activate the drug, coupled with the effect of drug activation on those cells. A greater understanding of these mechanisms and their effects on toxicity and efficacy is needed. David G. Huntsman, MD discussed how current genomic work can lead to new opportunities for cancer control. Advances in cellular and molecular biology confirm that “ovarian cancer” is not a single disease but rather a group of molecularly- and etiologically-distinct diseases that share an anatomical location. Prevention and treatment efforts must therefore exploit subtype-specific precursor lesions and tumor biology. Conducting subtype-specific clinical trials is challenging because of the sparse number of cases for subtypes of an already rare cancer. One possibility is to group cancers not by their site of origin and clinical presentation but by their molecular characteristics. Thus, clear cell and endometrioid ovarian cancers could be grouped with a subset of uterine cancers with which they share similar molecular profiles. However, successfully targeting a tumor-specific mutation in one cancer does not necessarily mean the same approach will be effective in another cancer expressing the same mutation: BRAF inhibition is highly effective in treating melanoma but shows limited response in colon cancers harboring the same oncogenic lesion. ( 24 ) Understanding other factors that influence targeted therapeutics within a specific molecular context will be key to implementing this new paradigm in clinical trials. Michael V. Seiden, MD, PhD presented data on the success of several therapies targeted at molecular alterations in other cancers. In almost all cases, these therapies targeted an activated oncogene (eg, EGFR, HER-2). Yet when these molecularly-targeted therapies have been tested in ovarian cancer, they have failed to show significant efficacy in progression-free survival or in clinical response. Data from TCGA provide insight: within HGSOC there is a great deal of genomic instability/variability with no single oncogene or group of oncogenes to target. ( 25 ) However, most HGSOC have lost at least three tumor suppressor genes (p53, OPCML, BRCA1/2) early in carcinogenesis. Mutated tumor suppressor genes cannot be directly targeted for reversal of function and the large number of pathways each affects makes targeting all subsequently activated genes prohibitive. Together, these observations raise the question of whether personalized, targeted, molecular therapies can really be developed in ovarian cancer and if so, what resources will be needed to accomplish that goal. Therapies targeting the tumor microenvironment and stem cells may prove more effective for ovarian cancer.

Ovarian cancer is the sixth most common cancer worldwide among women in developed countries and the most lethal of all gynecologic malignancies.( 1 ) Currently, most women have advanced stage disease at the time of diagnosis. Despite aggressive surgery and chemotherapy, the prognosis for these women is poor, with a 5-year survival rate of less than 30%. This poor outcome is due in part to the lack of effective prevention and early detection strategies: when diagnosed at an early stage, the survival rate is approximately 85–90%. Thus, prevention and early detection are key to overcoming this disease. With the exception of oral contraceptives, there are no successful chemopreventive agents available. Bilateral oophorectomy has also been shown to reduce disease incidence, but the procedure has several drawbacks in terms of women’s health.( 2 ) Existing screening techniques (CA125, transvaginal ultrasound) have not been demonstrated to reduce morbidity or mortality. Thus, better prevention, detection and screening methods are urgently needed. As well, because of the virulent and usually fatal nature of the disease, most women with ovarian cancer live with fear of recurrence, which happens in about 85% of cases. Current treatments offer little hope and survival has remained virtually unchanged for almost three decades. New methods to prevent, detect and treat recurrence are urgently needed. With advances in molecular biology and the emergence of new technologies, scientists are gathering remarkable knowledge about the genetic and biologic basis of ovarian cancer carcinogenesis. Such knowledge opens the door to new strategies for prevention, early detection and treatment of the disease. Importantly, it allows for the development of “personalized medicine,” wherein prevention, detection and treatment modalities are aimed at the specific molecular mechanisms of an individual tumor and its microenvironment, as well as at the specific genetic and biologic profile of the host. Science stands on the precipice of a new era for making profound progress in ovarian cancer research. To facilitate this progress, we convened a scientific symposium on ovarian cancer. The meeting was held May 10–11, 2012 in Pittsburgh, Pennsylvania. The meeting brought together over 300 researchers, scientists, clinicians, policy makers and advocates for an intensive two-day discussion of molecular mechanisms and personalized medicine in the prevention, detection and treatment of ovarian cancer and its recurrence. This article summarizes the highlights of the main presentations. Also included are abstracts chosen by the program committee as among the top submissions, as well as abstracts of the presentations by members of the Department of Defense Ovarian Cancer Academy. Readers are referred to the conference website in order to view videos of the complete talks and interactive panel discussions that were part of each session ( www.upci.upmc.edu/ovarian ).

Dana Bovbjerg, PhD discussed biobehavioral models for symptom management. Managing symptoms of both the disease and treatment present significant challenges for women and clinicians. In addition to medical complications, behavioral comorbidities, such as depression, fatigue, disrupted sleep and cognitive dysfunction, often initiate with diagnosis and treatment, and may continue into the survivorship period. Post-operative pain is a significant symptom which often influences other comorbidities throughout treatment and beyond. However, there are great inter-individual differences in pain perception even in healthy adults and the determinants of pain remain poorly understood. Recent data suggest sleep disruption may be one such determinant. In women undergoing breast conserving surgery, lower sleep efficiency (i.e., frequently disrupted sleep the night before surgery) has been associated with greater pain severity and interference with daily activities in the week following surgery. ( 17 ) Similar results were found in women undergoing more invasive gynecologic surgeries. These data illustrate the intricate relationships among psychological, behavioral and physical factors. However, the biologic mechanisms underlying these relationships remain unclear. Recent work has focused on neuroendocrine-immune factors, which jointly influence central nervous system functions. Understanding these mechanisms will help identify at-risk women who can then be targeted for therapeutic interventions. Sandra Mitchell, PhD, CRNP discussed using patient-reported outcomes (PROs) to measure therapeutic response and toxicities in clinical trials. PROs provide important insight into treatment efficacy, but to date have been underutilized. Many issues are involved in selecting an appropriate PRO instrument, including understanding the purpose of the measurement, assessing the correspondence between the instrument domain and scientific questions, and having clear scoring guidelines. Practical aspects must be considered, too, such as respondent burden, which can negatively impact results. The ability of current PRO systems to detect cancer treatment effects and their sensitivity to organ-specific issues remain unknown. Many challenges remain in incorporating PRO into gynecologic cancer trials, including measuring the value of the PRO instrument in the clinical trial setting, understanding when and how to incorporate a PRO system into trials, how to correlate PRO with diverse clinical and biomedical endpoints, how to handle missing data, developing methodologies to define responders and analyze results over time, understanding the clinical significance of changes in PRO over time, and how to report results in conjunction with other scientific findings. ( 18 ) Heidi Donovan, PhD, RN presented data on multi-symptom management. Women with ovarian cancer report 10–14 concurrent symptoms during treatment. Identifying and prioritizing symptoms as well as providing clinicians and women with both medical and self-care strategies remain areas for further research. Effective symptom management poses many challenges because some symptoms, such as fatigue, have no efficacious medical treatment, while the treatment of some symptoms often causes or exacerbates others. Symptom management is an ongoing part of care for women with ovarian cancer and may continue after active treatment ends. Thus, symptom management requires effective patient-clinician communication as well as substantial self-management on the part of women. Many trials exist to investigate symptom management, from those investigating a single method to address a single symptom to those investigating multiple methods to address multiple symptoms. Translating the findings on symptom management from the research setting to the clinic is an area for future investigation. The substantial self-management component for symptom management raises questions about its consequences on women with advanced disease. Identifying ways to support women in symptom management as well as to overcome barriers to self-management represent fruitful research directions. Diane C. Bodurka, MD discussed cancer survivorship.( 19 ) The number of cancer survivors, especially long term survivors, continues to grow. Many factors influence the health care and other needs of this growing population, including disease site and treatment, age at and time since diagnosis, comorbidities, lifestyle and behavioral factors and social support. Health issues affecting survivors, both pre-diagnosis issues and those resulting from treatment, are also not well understood. Such factors include fatigue, sexual dysfunction, sleep disturbance, neurological issues, urinary complaints and bowel complaints. Different treatments and combination of treatments impart different risk for these health effects and are experienced at all points on the survivorship spectrum. Interventions to mitigate these treatment-related effects throughout the entire survivorship period are needed. As well, educating both patients and primary care clinicians on managing short- and long-term survivor care is an important but currently unmet need.

The meeting opened with a keynote address by Nelly Auersperg, MD, PhD who provided a look at the progress made in ovarian cancer biology during the last 40 years. The ovarian cancer cell of origin remains an ongoing debate. Early studies implicated the ovarian surface epithelium (OSE) as the origin of high-grade serous ovarian carcinomas (HGSOCs) based on the observation of early transformed cells within ovarian epithelial inclusion cysts ( 3 ) This view was held until quite recently when it was observed that lesions resembling HGSOC were found in the oviductal fimbriae of BRCA1 carriers suggesting that some HGSOCs arise in the fallopian tubes.( 4 ) Despite evidence for both theories, none of the evidence supports either theory 100%. Thus it is possible that a subset of HGSOCs arises from the OSE while another subset has a tubal origin. This raises the question: how do two epithelia from different organs with very different structure and function give rise to identical carcinomas? Emerging molecular evidence suggests that the OSE and distal fimbrial epithelium are a continuous, incompletely determined zone of epithelial transition from one epithelial type to another.( 5 ) Such transitional epithelia in other parts of the body, such as the squamo-columnar junction of the cervix, are known to be prone to neoplastic transformation. This theory provides a partial explanation for why after prophylactic salpingo-oopohorectomies the remaining portion of the fallopian tube does not represent a cancer risk and why salpingectomy alone may not provide adequate protection against ovarian cancer development. Future work will provide greater insight into this theory and how it can help explain ovarian carcinogenesis.

Setsuko K. Chambers, MD presented data showing that micro RNAs (miRNAs) interact with messenger RNAs (mRNAs) to influence ovarian cancer etiology. RNA binding proteins control translational regulation of mRNA through a binding element typically located in the 3’ untranslated region (3’UTR). mRNA binding proteins can have both oncogenic and tumor suppressor roles. Emerging data also show that miRNAs are important mRNA regulatory components through both translational control as well as modulation of mRNA decay. miRNAs are frequently dysregulated in cancers, have both oncogenic and tumor suppressive roles, and are involved in ovarian proliferation, invasion and metastasis. Interaction between RNA binding proteins and miRNAs is another mechanism whereby each influences disease. For example, CSF-1, which plays a key role in ovarian cancer etiology,( 15 ) is regulated by miR130a and miR301a. Both miRNAs are dependent on the RNA binding protein nucleolin for gene expression of CSF-1 and as well as for tumor cell motility. These data suggest that regulators of the 3’UTR can control gene expression and tumor behavior. Whether these mechanisms can be exploited for therapeutic intervention warrants further investigation. There is limited understanding of the role of nuclear receptors (NRs) in HGSOC. Steffi Oesterreich, PhD presented in silico analyses of the Cancer Genome Atlas (TCGA) Project data sets, which identified members of the NR4A family of orphan receptors as potential drivers of a subset of HGSOC. The relevance of this finding is unclear and further studies of the mechanics and function of these receptors are needed. The association between hormonal exposures and HGSOC suggests that hormones play a role in the etiology of the diseases and, thus, endocrine therapies may be fruitful at least for some subset of cancers. However, little is known about the role of steroid hormone receptors such as the estrogen receptor (ER) in HGSOC. Preliminary studies in ovarian cancer cell lines indicate that ER expression is not a reliable biomarker of estrogen response and better predictive markers are needed. Melanie Flint, PhD discussed the role of stress on cancer initiation and progression with a focus on the adaptive immune system. Stress triggers a complex response mechanism that affects various systems, including the immunes system. In vitro , release of stress hormones induces T cell activation and migration, while decreasing cell proliferation. ( 16 ) The mechanism underlying these changes may be through rearrangement of the actin cytoskeleton. In a transgenic mouse model of ovarian cancer, chronic stress decreases CD3+ T cell activation and results in earlier onset of tumors. However, the tumors appear to be more confined to the ovary compared cancers induced in unstressed mice. This suggests that stress hormones, in addition to affecting the immune system, may also directly interact with cancer cells thereby impacting proliferation. Elucidating the molecular mechanisms underlying these findings will be important in understanding the impact of chronic and acute stress on cancer initiation and progression.

Partnerships, Martha E. Gaines, JD. Dr. Gaines, an 18 year ovarian cancer survivor, founded the Center for Patient Partnerships to address the need to enhance the many partnerships that result from a diagnosis of ovarian cancer. Central to successful treatment for ovarian cancer is the physician-patient partnership. Choosing a healing path requires many elements, including understanding the roles and goals of both the physician and patient, clear and honest communication and a balance of power. Only through an effective patient-clinician partnership will a woman be truly empowered to face the fight of her life.

Robert C. Bast, Jr, MD discussed biomarkers for detecting and treating incident and recurrent ovarian cancer. Identifying sensitive (>75%) and specific (>99.6%) screening modalities are key to early detection. CA125 is currently the standard marker; however, it is only 99% specific. UKCTOCS showed improved detection when using ultrasound (US) after assessing CA125 levels relative to a woman’s baseline value rather than a “standard” clinical value: early stage detection was doubled (48%) and the number of operations per cancer case detected was remarkably low (4 versus 36 for US alone). ( 10 ) Despite these promising findings, 20% of ovarian cancers will not be detected with this two-tier approach because they fail to express CA125. Moreover, cancers originating in the fallopian tube cannot be visualized on ultrasound prior to metastasis. Additional biomarkers are needed. Promising directions include panels of serum-based tumor markers and autoantibodies. To date neither has demonstrated the requisite sensitivity and specificity to be clinically useful. Better imaging techniques able to detect cancers at even smaller volumes are also needed. Finally, work is needed to understand how to combine both serum marker and visualization approaches in a way that will produce a cost-effective screening tool that can easily be adapted to the clinic. Kristin Zorn, MD presented data on ovarian cancer development in “high-risk” women, defined as women who possess a germ-line mutation that confers an increased risk of ovarian cancer. In addition to garnering insight into how to best care for these women, studying the high-risk population helps understand the pathogenesis of sporadic disease. The discovery of incident fallopian tube cancers among women undergoing risk-reducing bilateral salpingo-oophorectomy (RRBSO) raises the intriguing notion that the fallopian tube and not the ovary is the origin for some HGSOCs. Further laboratory data support this hypothesis and even suggest that tubal intraepithelial carcinomas (TICs) may serve as the precursor lesion for HGSOC: both lesions exhibit cytological atypicia, high proliferative indices, and the presence of p53. ( 11 ) The presence of TICs in women diagnosed with HGSOC further supports this hypothesis. ( 12 ) However, the data are not consistent. For example, high rates of p53 foci are found in the tubal epithelium of normal-risk women. Thus, other molecular alterations must be necessary for the progression to malignant disease. Elucidating these alterations as well as developing techniques to more carefully profile specimens from high-risk women represent areas for future research. ( 13 ) Marian Mourits, MD, PhD talked about the impact of the new fallopian tube hypothesis on preventive strategies in high-risk woman. Screening is ineffective for detecting ovarian cancer at an early stage and has not been shown useful in managing high-risk women. RRBSO is currently the only preventive option available. However, the procedure has many side effects that may negatively impact a woman’s health and quality of life, including bone and cardiovascular health as well as sexual functioning.( 2 ) Thus, care must be taken to manage postmenopausal symptoms arising from a RRBSO. In light of the tubal origin of the disease, another option is for high-risk women to have a “risk-reducing” salpingectomy. However, no studies have been conducted to assess the effectiveness of this strategy. As well, while data support a tubal origin of the disease especially in murine models, ( 14 ) the evidence in humans is far from conclusive. Many questions remain, such as how cancerous or even pre-cancerous cells from the fallopian tubes migrate to the ovary and become HGSOC, what is the precursor to TICs, what is necessary for a TIC to convert to HGSOC, and through what mechanism does ovulation suppression affect the tubal epithelium?

Technological advances in the last decade have increased our knowledge of the molecular mechanisms involved in a host of biological activities related to normal ovarian function as well as to ovarian cancer development. The advantages of applying molecular approaches and supporting technologies to ovarian cancer prevention and early detection are many. The most effective way to prevent any disease is to understand its underlying cause and change the conditions that permit it to occur. Identifying the precise molecular and biologic steps that characterize pre-malignant change will provide the foundation for the search to find agents that reverse these changes or block the steps critical to the full development of cancer. Similar steps can also be used to detect and prevent disease recurrence. Moreover, a precise characterization and understanding of the mechanisms involved in cancer initiation and progression will help lead to the development of prevention and treatment modalities that can be personalized to each patient, thereby helping to overcome this highly-fatal malignancy.

C. Leigh Pearce, PhD presented work on endometriosis and ovarian cancer. Women with endometriosis are at an increased risk of ovarian cancer. Despite the prevalence of endometriosis (about 10% in the general population ( 6 )), only a small percentage of women with the condition develop ovarian cancer. Identifying these high-risk women remains elusive. Moreover, the association between endometriosis-associated disease and well-known ovarian cancer protective factors, such as OC use and parity, remains unclear. The relationship between ovarian cancer risk and endometriosis-related factors, such as anatomical location, type and timing of treatments, and symptom and treatment response, are not also known. Clarifying these factors will help establish risk estimates based on an individual’s profile and will help tailor prevention interventions. The Ovarian Cancer Association Consortium (OCAC) confirmed that the endometriosis-ovarian cancer link is limited to the invasive clear cell and endometrioid subtypes.( 7 ) No association was found for borderline tumors, suggesting that contrary to current theories, borderline clear cell and endometrioid tumors may not be the pre-cursors of their invasive counterparts. Identifying the pre-cursor lesion for endometriosis-associated ovarian cancer as well as factors associated with progression from pre-malignant to malignant disease will support developing targeted prevention therapies. Ellen L. Goode, PhD, MPH contended that ovarian cancer has a genetic component beyond the rare, high-risk, low prevalence (<1%) mutations in genes such as BRCA1/2 . Combinations of common genetic variants with minor allele frequencies (MAFs) greater than 5% and which confer modest risk likely account for the remaining heritability. Candidate gene studies have identified many potential susceptibility variants. However, data have been inconsistent due to small sample sizes and heterogeneity across study populations. OCAC provides a large sample size, pooling of data and examining of between-study heterogeneity in order to address these limitations. OCAC has confirmed some variants associated with ovarian cancer while refuting others. Using GWAS, 6 novel susceptibility loci (in 2q31, 3q25, 8q24, 9p22,17q21, 19p13) were identified.( 8 , 9 ) Each of these variants is common (frequency >8%) and confers only a modest change in risk (< 20%); however, their functions are mostly unknown. Further work in understanding how they impact ovarian cancer and how they are affected by host factors is needed. Further research is also needed to understand the relationship between these loci, specific disease subtypes and other disease phenotypes such as survival. This will help evaluate the clinical utility of the markers as well as uncover novel prevention and treatment targets. Kirsten Moysich, PhD examined the role of immunosuppression pathways in ovarian cancer and explored the hypothesis that a strong immunosuppressive genotypic and phenotypic profile is associated with ovarian carcinogenesis. Specifically, her work is investigating the roles of immunosuppressive regulatory T cells (Tregs) and myeloid derived suppressor cells (MDSCs) in ovarian cancer. Ovarian cancers appear to have much higher levels of Tregs and MDSCs compared to benign tumors. Work clarifying the meaning of immunosuppressive cells in ovarian cancer development and outcome is needed. In addition to the phenotypic associations, functional SNPs in genes involved in the Treg and MDSC pathways were also associated with ovarian cancer risk and poorer survival in initial studies. When these same candidate genes were examined among 17,421 cases and 25,878 controls in OCAC, only a handful of associations were found and the effect sizes were of questionable relevance (less than 5–10%). However, when examining associations by histologic subtypes, several associations of modest size were found. The associations differed among the histologic types in both magnitude and direction. This raises interesting questions about the functionality of SNPs, the subsequent phenotype and the relationship to specific histology.