- Email: [email protected]
Summary of the 2015 American Association for Cancer Research (AACR) Annual Meeting
Gynecologic Oncology 138 (2015) 7–10
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Meeting Report Summary of the 2015 American Association for Cancer Research (AACR) Annual Meeting Keywords: Personalized therapy Immunotherapy Resistance to cancer therapies Tumor heterogeneity Preclinical models Tumor microenvironment Dormancy Cancer genetics
The 2015 American Association for Cancer Research (AACR) Annual Meeting was held in Philadelphia, PA from April 18th to April 22nd, 2015. The theme of the meeting was “Bringing Discoveries to Patients”. This year there were few gynecologic cancer-specific sessions, yet many talks on basic cancer biology have significance to cancers specific to women. True to its name, there were more sessions than prior years with updates on clinical trials, reports of programs seeking to globally profile patient samples, and efforts to understand cancer genomics as they relate to treatment and outcome. This conference report summarizes key abstracts and presentations relevant to the understanding and therapy of gynecologic malignancies.
1. Tumor genomics, heterogeneity, evolution, and resistance Mike Stratton reported genomic insights gained from pan-cancer analysis of The Cancer Genome Atlas and International Cancer Genome Consortium and described several processes that could account for mutation patterns observed in various tumors. This was the result of analysis of 12,000 carcinoma samples from 40 different cancer types and about 8 million somatic substitutions from these samples. The intent is to identify patterns in types of mutations that can be used to categorize cause, prognosis, or basic biology of different cancers. They concluded that there were only thirty different mutational categories, a few of which were tumor type specific, but most were found across multiple tumor types. Certain environmental exposures produced signatures, such as a common mutational profile noted by exposure to UV radiation where C N T somatic substitutions predominate, or C N A mutations in lung cancer associated with smoking. In ovarian cancer, three major mutational signatures were noted. One of these was a signature of APOBEC's, a family of cytidine deaminases. A similar APOBEC mutational signature was also observed in cervical cancer associated with viral-induced carcinogenesis. Interestingly, viral DNA entry and retrotransposon remobilization may serve as mechanisms that switch on APOBEC enzymes. Mutational signatures associated with HR defects were also identified. Overall, he concluded that cancer genomes frequently contain “ketaegis” events, or clusters of hypermutated regions which are relatively random and not consistent from patient to patient. The underlying etiologies of these mutations patterns are not understood. A better
http://dx.doi.org/10.1016/j.ygyno.2015.05.032
understanding of underlying mutational processes associated with these signatures may lead to a better understanding of the initiating oncogenic switch, or treatment options specific to the signature, instead of tumor type. Investigating the hierarchical organization of breast and ovarian cancer cells, John Stingl from the Cancer Research UK Cambridge Institute, discussed his findings in high grade serous ovarian carcinomas. He hypothesized that subpopulations of these tumors cell have stemlike qualities which render them resistant to chemotherapy. Analyzing 86 freshly isolated tissues, he found three distinct subpopulations of epithelial cells, one which expressed high levels of epithelial cell protein (EpCAM) and the other two which were both EpCAM negative, expressed podoplanin and varying levels of CD43, but had different types of daughter cells. A xenotransplantation model treated with cisplatin revealed that EpCAM negative cells were less sensitive to treatment, however, ultimately it was concluded that both EpCAM positive and negative cell populations have stem-like properties. In a poster session, Paul Goodfellow presented a study that identified CTCF and ZFHX3 deletions and loss of function mutations in endometrial cancers. CTCF and ZFHX3 are tumor suppressor genes located near 16q21.2, and The Cancer Genome Atlas dataset on uterine corpus endometrioid carcinomas showed truncating nonsense and frameshift insertions and deletions in these genes. His group performed targeted sequencing of 541 endometrioid endometrial carcinomas and found single nucleotide variations and indels in 24% and 18% for CTCF and ZFHX3, respectively. His group found a significant co-occurrence of alterations in both genes, and these alterations are associated with high grade, lymphovascular space invasion, and shorter recurrencefree survival in endometrial cancers. Charles Swanton described clonal evolution of tumors over time using the “tree” model, whereby the trunk represents initial (and persistent) mutations, and branches represent subsequent mutations that may be disparate in different metastatic lesions. The difficulty lies in knowing whether subsequent “branch” mutations are drivers or passengers, in which case they would be irrelevant to patient-directed therapeutics. His analysis indicated that chemotherapy may be “trimming” the multiple branches, but leaving cells with the more crucial trunk mutations that can persist. Additionally, he provided evidence that patients with heterogeneous tumors consisting of multiple subclonal drivers showed poor overall survival. This would be the case for ovarian cancer, since the TP53 “trunk” mutation would repeatedly give rise to multiple branches. However, his dataset was not specific to ovarian cancer, being concentrated on colon cancers, and additional research is required to know if these findings are generalizable. Dr. Levi Garraway then expanded on the theme of evolving cancer resistance by describing molecular mechanisms associated with resistance to targeted therapy. Using functional screens, his group addressed whether candidate genes mediating resistance are necessary and sufficient for resistance to targeted therapies, whether these candidate genes reactivate signaling pathways downstream of the drug targets, and the relevance of such candidate genes in clinical resistance. He described that pathway reactivation is a common mechanism of resistance
8
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
to pathway-targeted cancer therapies. For example, C-RAF overexpression can overcome B-RAF inhibition, and so is downstream activation of Mek1/2. Other growth factor singling pathways that crosstalk with a targeted pathway may also result in resistance to pathway-targeted therapies. For example, resistance to RAS/Raf pathway inhibitors can be mediated by downstream transcription factors, upstream G-protein coupled receptor, or activators of Protein Kinase A. Pathway reactivation, pathway bypass, pathway indifference (alternative oncogenic transcriptional outcome) are generalized concepts involved in pathway-targeted cancer therapy resistance. Pathway inhibition will select for pathway-independent resistant mechanisms, and these mechanisms may converge on transcription factors that are actual effectors of the pathway. Dr. Garraway also described known challenges to clinical drug resistance studies, such as the multi-factorial nature of drug resistance, under-sampling, and intratumor heterogeneity. Therefore, it would be important to target points of convergence to reverse resistance. 2. Epigenetics Dr. Steve Baylin described epigenetic alterations in cancer, including focal regions of hypermethylation and wide regions of hypomethylation in cancer genomes. He also described IDH1 mutations and their association with CpG island methylator phenotype (CIMP). He then discussed the extent to which abnormal epigenetics programming can contribute to tumorigenesis and how abnormal epigenetics programming in established tumor cells may be targeted for therapeutic benefits. One exciting area that Dr. Baylin highlighted is the use of low dose 5azacidine (DNA methyltransferase inhibitor) for the purpose of immunostimulation and enhancing immunotherapy. He then described a potential role of endogenous retroviral remobilization in stimulating IFNγ and the therapeutic potential of epigenetics reprogramming by 5 azaC in immunoreactive molecular subtype of ovarian cancer. 3. Mouse models Dr. Tyler Jacks from MIT reviewed traditional and novel approaches to generate genetically engineered mouse models of cancer to characterize the role of candidate genes in tumor immunology and tumor biology. He described development of a transgenic mouse line with FoxP3driven diphtheria toxin receptor for immunotherapy. Upon administration of diphtheria toxin, FoxP3-expressing Treg cells were depleted, and T cell infiltration of tumor was observed in these mice, demonstrating the potent role of Tregs in the suppression of tumor immunity. Also described was the exciting development of the CRIPSR/Cas9 system, whereby genetic editing can be performed to specifically knock in or out genes, or even change one nucleotide to induce a mutation. Dr. Jacks and his colleagues generated lentiviral expression constructs that expressed both guide RNA (gRNA) and Cas9, and disrupted p53 and PTEN. These mice showed the onset of tumor development within 6 months of injection of lentiviral particles. Unlike conventional knockout models, where construction of targeting vectors, screening of ES cells, and generation of knockout mouse lines are quite laborious and time consuming, lentiviral delivered CRISPR/Cas9 system is rapid, reproducible, and relatively simple. 4. Immunology There were numerous sessions on advances in preclinical immunologic approaches to cancer, one of which was specific to ovarian cancer. George Coukos from the University Hospital of Lausanne presented “Opportunities for Immunotherapy in Ovarian Cancer”, which emphasized the importance of evaluating the presence or absence of tumorinfiltrating lymphocytes (TILs) in ovarian cancer. Of their cohort of patients with TILs present in tumors, greater than 60% had improved survival rates and 50% never relapsed. In comparing tumors with and
without TILs, their group found that high levels of FASL, mediated by VEGF and PGE2, encouraged T-cell death in the tumors. Blocking VEGF and PGE2 reduced FASL levels, allowing T-cell accumulation and restoration of an immunogenic tumor. In another important approach, they demonstrated that the absence of TILs predicted failure to anti-PD-1 therapy. However, this could be partially overcome by addition of a vaccine generated specifically to tumor antigens, increasing therapeutic response. Jeong Kim (Genentech) presented “Unleashing anti-tumor immunity through anti-OX40 monotherapy and in combination with anti-PDL1”. OX40 (aka Tumor necrosis factor receptor superfamily member 4 (TNFRSF4) or CD134) is a receptor expressed only on activated CD4 and CD8 lymphocytes. It is typically upregulated 48–72 h after T-cell activation, and promotes proliferation and clonal expansion of effector and memory populations. Anti-OX40 is an agonistic monoclonal antibody that is effective via a dual mechanism of 1) co-stimulating effector T-cells increasing their proliferation and the production of cytokines and 2) inhibiting regulatory T-cells. Mouse studies demonstrated that anti-OX40 treatment reduced tumor burden and established immune memory, making the mice resistant to tumor rechallenge. Additionally, combining anti-OX40 agonistic MAb with anti-PD-L1 treatment increased the therapeutic response compared to either treatment alone. Dr. Robert Schreiber from Washington University at St. Louis described the application of genomics to personalize cancer immunotherapy. He first described the concept of immunoediting involving the “3E's”: elimination, equilibrium, and escape that may account for the mechanisms of tumor cell evolution to immune evasion. For example, recombination-activating gene 2 (RAG2) is one of the 2 genes responsible for rearrangement and recombination of genes during V(D)J recombination as immunoglobulin and T-cell receptors undergo maturation. RAG2(−/−) mice develop tumors quicker than wild type, and tumors taken from Rag2(−/−) mice are rejected in Rag+/+. Tumors that developed in Rag2(+/+) mice showed reduced immunogenicity providing evidence that immunogenicity is “edited”. He went on to describe how genomics might be used to personalize cancer immunotherapy. NetMHC is a server at the Center for Biological Sequence Analysis at the Technical University of Denmark that can be used to predict which tumor-specific mutations form tumor-specific mutant rejection antigens. These could be used for rapid generation of immunotherapy approaches by tumor profiling. Finally, he presented evidence that checkpoint blocking therapy (such as anti-PD-1-based therapies) targets specific mutant neoantigens. Here, mutated proteins can serve as neoantigens that are recognized by immune cells, even if the mutant proteins were not necessarily drivers of disease progression. These neoantigens can be used to develop personalized vaccines, with data presented that such vaccines are effective in blocking immune checkpoints. Furthermore, combining neoantigen personalized vaccine and checkpoint blocking therapy was more effective than either alone. Specifically, while immune checkpoint blocking therapies are less effective when started late in the disease course, adding a personalized vaccine increased checkpoint therapy efficacy even in late stages of tumor progression. 5. Tumor dormancy and senescence Several talks were presented examining the mechanisms underlying tumor dormancy. Lewis Chodosh used an inducible HER2overexpression breast cancer model to demonstrate that the initial oncogenic pathways responsible for primary tumor cell development were not identical to those mediating recurrent disease. For example, when HER2-positive tumors were treated to no evidence of disease, recurrent tumors could be induced by re-expression of HER2, but recurrent tumors then overexpressed other genes or pathways that were minimally expressed in primary tumors, including single-stranded binding protein Ssb1, mediators of Notch signaling, c-Met, and autophagy. Recurrence could be prevented in some
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
cases by inhibition of these pathways with SsB1 knockout, gamma secretase inhibitors, c-Met inhibitors, or chloroquine. Interestingly, in their model the targeted therapies were more effective when given after primary therapy, instead of in combination with it. Although this was a HER2-inducible breast cancer model, these insights into tumor dormancy may have parallels in the dormancy noted in surviving ovarian cancer cells after primary chemotherapy. In a cellular senescence and aging symposium Scott Lowe described cell autonomous and non-cell autonomous programs of senescence. The cell autonomous pathway frequently engages p53 and Rb tumor suppressor pathways in a repressive chromatin state. The non-cell autonomous or secretory phenotype involves MMP3 and PAI-1, and can stimulate and transform pre-malignant cells. He showed that Brd4, a chromatin reader protein that regulates transcription of Myc, works with NFκB to induce senescence associated secreted phenotype (SASP). He showed that Brd4 is required for secretion, but not cell cycle arrest. Secretory factors from senescent cells are required for communication to other cells such as NK cells. Immune clearance of senescent cells occurs normally, but with impairment of Brd4, clearance of senescent cells was not observed. Instead these cells were able to escape immune surveillance. It is unclear if this action is pro-oncogenic or pro-senescent, but it likely depends on the neighboring cell. In the same session Judith Campisi discussed the induction of DNA damage by chemotherapy, which may also evoke senescence. She showed that SASP from both fibroblast and endothelial cells were involved in wound healing via the matrix protein CCN1. These data suggests that senescent cells play a role in migration and repair mechanisms. This begs the question: if chemotherapy potentially promotes senescence of tumor cells then how are these tumor cells contributing to chemoresistance and metastatic disease? Campisi hypothesized that chemotherapeutics might cause a senescent burden that actually may be detrimental. However if the clearance of senescent cells could be reactivated, such potential for a nidus of recurrent disease might be prevented. 6. Tumor microenvironment Several presentations at the meeting highlighted new findings by which tumor cells and stroma may be regulating one another in relation to the development and persistence of metastatic disease in ovarian cancer patients. Haria and co-workers determined that the homeobox gene DLX4 promotes inflammatory signaling and peritoneal metastasis in ovarian cancer patients. Specifically, DLX4 expressed in tumor cells stimulated CD44 on mesothelial cells, through activation of IL-1β and NF-κB. Inhibition of either IL-1β or NF-κB diminished the effect of DLX4 on tumor-stroma interactions. The authors suggested that inflammatory response pathways could be a potential therapeutic target for ovarian cancer. Work presented from Samuel Mok's group focused on understanding how omental obesity contributes to poor prognosis in high grade serous ovarian carcinoma. Transcriptome profiling of adipose tissue from normal patients and those with high grade serous ovarian cancer suggest that a novel adipokine, omentin (Intestinal Lactoferrin Receptor, IITLN1), was downregulated in ovarian cancer patient tissues compared to normal or benign patient's omentum. The authors propose that ITLN1 may sequester lactoferrin, which is highly expressed in ascites, preventing it from binding to low density lipoprotein receptor-related protein 1 (LRP-1) on the surface of ovarian cancer cells and activating downstream signaling pathways that regulate MMP1 expression. Furthermore, the authors showed that patients with low levels of ITLN1 prior to initial treatment had poor survival compared to those with higher levels. The authors concluded that cancer cells are capable of modifying visceral adipose tissue via down regulation of ITLN1. In the joint TME/Cancer Immunology group meeting Raghu Kalluri discussed the functional role of the stroma in tumor cell
9
immunity in pancreatic cancer. He noted that myofibroblasts in TGFBR2 flox/flox mice were able to maintain some type of control to keep tumors differentiated. Previously it has been reported that fibroblasts and the collagen-rich matrices frequently found in pancreatic cancer patients impair access of gemcitabine to cancer cells. Kalluri and co-workers determined that removing fibroblasts and then adding gemcitabine to tumor cells had no effect on cell death. However, treating these tumors with the immune checkpoint therapy anti-CTLA4 improved survival in these mice. Kalluri concluded that more myofibroblasts improve survival in pancreatic patients. These findings demonstrate the two sides of the microenvironment: 1) an instigator of tumor growth and 2) a suppressor of tumor growth. In the same session Padmanee Sharma showed that CTLA4 blockade enhances T-cell anti-tumor response, decreasing tumor burden. However, she noted that PD-L1 expression in tumors changes as a function of treatment. It is therefore important to establish methods by which immune targets can be evaluated in patients to predict response to certain treatments. 7. Emerging mechanisms of PARP inhibition Alan D'Andrea presented data on potential extension of PARP inhibitor therapy beyond BRCA-mutated cases. PARPi kills cancer cells through synthetic lethality by suppressing non-homologous endjoining DNA repair. In cells lacking BRCA1/2, this promotes alternative end-joining as the required DNA repair process, which is error-prone and ultimately lethal. A key enzyme in alternative end-joining is polymerase Q (POLQ), and data were presented that POLQ simultaneously functions to inhibit homologous repair by binding and inhibiting RAD51. Therefore POLQ inhibitors, likely in combination with PARPi, may increase the therapeutic response, or sensitize malignancies that are inherently resistant to PARP inhibitors. 8. Promising agents in clinical trials Parames Thavasu presented results on a novel mTORC1/2 inhibitor (AZD2014). In inhibiting both mTORC 1 and 2, the feedback-initiated increase in phospho-Akt that accompanies and limits effectiveness of mTORC1-selective inhibitors is avoided. The inhibitor was studied in a phase I trial in which ovarian cancer patients received a 3-day on, 4day off schedule in combination with weekly paclitaxel (80 mg). A DLT of 50 mg BID (or 75 mg BID for 2-day on, 5-day off) was identified. Regarding responses, 7 out of 10 patients demonstrated a partial response, and most of those were durable for more than 6 months. A randomized phase II trial is slated to begin later this year. 9. Cervical cancer Dr. Douglas Lowy, who initially developed the vaccine against the L1 surface protein of the Human Papilloma Virus, and has been named interim director of the National Cancer Institute, provided an overview and update of the HPV vaccination program. He highlighted that only half of HPV-associated cancers are cervical, the other half predominantly due to oropharyngeal and anal cancers, which have a high prevalence in men, arguing for screening programs in boys as well as girls. He presented supporting data that the vaccination program not only reduces cases through direct immunization in the recipient, but also through herd immunity — decreased cases due to decreased prevalence of the virus in the population. An Australian study was highlighted which showed a dramatic reduction in the incidence of genital warts, even in patients who were not vaccinated. Additionally, two new developments were presented. The first was evidence that in adolescents (9–14 yoa), 2 doses for 6 months apart are as effective as 3 doses, due to the higherthan-expected immune response in these young patients. Trials examining efficacy of just 1 dose are being designed. Secondly, the 9-valent vaccine (Gardasil-9, Merck) offers better protection against CIN2
10
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
disease than the quadrivalent vaccine, leading to FDA approval of this vaccine in December 2014. Acknowledgments The authors would like to thank Jill Madden and Megan Cooley at the University of Kansas Medical Center, and Robert Cornelison at the University of Virginia for their assistance in writing this review. Jeremy Chien Department of Cancer Biology, University of Kansas Medical Center, United States Translational Genomics, University of Kansas Cancer Center, United States
Charles N. Landen Department of Obstetrics and Gynecology, University of Virginia, United States Women's Oncology Program, University of Virginia Cancer Center, United States Corresponding author at: Department of Obstetrics and Gynecology, University of Virginia, United States. E-mail address: [email protected].
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Meeting Report Summary of the 2015 American Association for Cancer Research (AACR) Annual Meeting Keywords: Personalized therapy Immunotherapy Resistance to cancer therapies Tumor heterogeneity Preclinical models Tumor microenvironment Dormancy Cancer genetics
The 2015 American Association for Cancer Research (AACR) Annual Meeting was held in Philadelphia, PA from April 18th to April 22nd, 2015. The theme of the meeting was “Bringing Discoveries to Patients”. This year there were few gynecologic cancer-specific sessions, yet many talks on basic cancer biology have significance to cancers specific to women. True to its name, there were more sessions than prior years with updates on clinical trials, reports of programs seeking to globally profile patient samples, and efforts to understand cancer genomics as they relate to treatment and outcome. This conference report summarizes key abstracts and presentations relevant to the understanding and therapy of gynecologic malignancies.
1. Tumor genomics, heterogeneity, evolution, and resistance Mike Stratton reported genomic insights gained from pan-cancer analysis of The Cancer Genome Atlas and International Cancer Genome Consortium and described several processes that could account for mutation patterns observed in various tumors. This was the result of analysis of 12,000 carcinoma samples from 40 different cancer types and about 8 million somatic substitutions from these samples. The intent is to identify patterns in types of mutations that can be used to categorize cause, prognosis, or basic biology of different cancers. They concluded that there were only thirty different mutational categories, a few of which were tumor type specific, but most were found across multiple tumor types. Certain environmental exposures produced signatures, such as a common mutational profile noted by exposure to UV radiation where C N T somatic substitutions predominate, or C N A mutations in lung cancer associated with smoking. In ovarian cancer, three major mutational signatures were noted. One of these was a signature of APOBEC's, a family of cytidine deaminases. A similar APOBEC mutational signature was also observed in cervical cancer associated with viral-induced carcinogenesis. Interestingly, viral DNA entry and retrotransposon remobilization may serve as mechanisms that switch on APOBEC enzymes. Mutational signatures associated with HR defects were also identified. Overall, he concluded that cancer genomes frequently contain “ketaegis” events, or clusters of hypermutated regions which are relatively random and not consistent from patient to patient. The underlying etiologies of these mutations patterns are not understood. A better
http://dx.doi.org/10.1016/j.ygyno.2015.05.032
understanding of underlying mutational processes associated with these signatures may lead to a better understanding of the initiating oncogenic switch, or treatment options specific to the signature, instead of tumor type. Investigating the hierarchical organization of breast and ovarian cancer cells, John Stingl from the Cancer Research UK Cambridge Institute, discussed his findings in high grade serous ovarian carcinomas. He hypothesized that subpopulations of these tumors cell have stemlike qualities which render them resistant to chemotherapy. Analyzing 86 freshly isolated tissues, he found three distinct subpopulations of epithelial cells, one which expressed high levels of epithelial cell protein (EpCAM) and the other two which were both EpCAM negative, expressed podoplanin and varying levels of CD43, but had different types of daughter cells. A xenotransplantation model treated with cisplatin revealed that EpCAM negative cells were less sensitive to treatment, however, ultimately it was concluded that both EpCAM positive and negative cell populations have stem-like properties. In a poster session, Paul Goodfellow presented a study that identified CTCF and ZFHX3 deletions and loss of function mutations in endometrial cancers. CTCF and ZFHX3 are tumor suppressor genes located near 16q21.2, and The Cancer Genome Atlas dataset on uterine corpus endometrioid carcinomas showed truncating nonsense and frameshift insertions and deletions in these genes. His group performed targeted sequencing of 541 endometrioid endometrial carcinomas and found single nucleotide variations and indels in 24% and 18% for CTCF and ZFHX3, respectively. His group found a significant co-occurrence of alterations in both genes, and these alterations are associated with high grade, lymphovascular space invasion, and shorter recurrencefree survival in endometrial cancers. Charles Swanton described clonal evolution of tumors over time using the “tree” model, whereby the trunk represents initial (and persistent) mutations, and branches represent subsequent mutations that may be disparate in different metastatic lesions. The difficulty lies in knowing whether subsequent “branch” mutations are drivers or passengers, in which case they would be irrelevant to patient-directed therapeutics. His analysis indicated that chemotherapy may be “trimming” the multiple branches, but leaving cells with the more crucial trunk mutations that can persist. Additionally, he provided evidence that patients with heterogeneous tumors consisting of multiple subclonal drivers showed poor overall survival. This would be the case for ovarian cancer, since the TP53 “trunk” mutation would repeatedly give rise to multiple branches. However, his dataset was not specific to ovarian cancer, being concentrated on colon cancers, and additional research is required to know if these findings are generalizable. Dr. Levi Garraway then expanded on the theme of evolving cancer resistance by describing molecular mechanisms associated with resistance to targeted therapy. Using functional screens, his group addressed whether candidate genes mediating resistance are necessary and sufficient for resistance to targeted therapies, whether these candidate genes reactivate signaling pathways downstream of the drug targets, and the relevance of such candidate genes in clinical resistance. He described that pathway reactivation is a common mechanism of resistance
8
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
to pathway-targeted cancer therapies. For example, C-RAF overexpression can overcome B-RAF inhibition, and so is downstream activation of Mek1/2. Other growth factor singling pathways that crosstalk with a targeted pathway may also result in resistance to pathway-targeted therapies. For example, resistance to RAS/Raf pathway inhibitors can be mediated by downstream transcription factors, upstream G-protein coupled receptor, or activators of Protein Kinase A. Pathway reactivation, pathway bypass, pathway indifference (alternative oncogenic transcriptional outcome) are generalized concepts involved in pathway-targeted cancer therapy resistance. Pathway inhibition will select for pathway-independent resistant mechanisms, and these mechanisms may converge on transcription factors that are actual effectors of the pathway. Dr. Garraway also described known challenges to clinical drug resistance studies, such as the multi-factorial nature of drug resistance, under-sampling, and intratumor heterogeneity. Therefore, it would be important to target points of convergence to reverse resistance. 2. Epigenetics Dr. Steve Baylin described epigenetic alterations in cancer, including focal regions of hypermethylation and wide regions of hypomethylation in cancer genomes. He also described IDH1 mutations and their association with CpG island methylator phenotype (CIMP). He then discussed the extent to which abnormal epigenetics programming can contribute to tumorigenesis and how abnormal epigenetics programming in established tumor cells may be targeted for therapeutic benefits. One exciting area that Dr. Baylin highlighted is the use of low dose 5azacidine (DNA methyltransferase inhibitor) for the purpose of immunostimulation and enhancing immunotherapy. He then described a potential role of endogenous retroviral remobilization in stimulating IFNγ and the therapeutic potential of epigenetics reprogramming by 5 azaC in immunoreactive molecular subtype of ovarian cancer. 3. Mouse models Dr. Tyler Jacks from MIT reviewed traditional and novel approaches to generate genetically engineered mouse models of cancer to characterize the role of candidate genes in tumor immunology and tumor biology. He described development of a transgenic mouse line with FoxP3driven diphtheria toxin receptor for immunotherapy. Upon administration of diphtheria toxin, FoxP3-expressing Treg cells were depleted, and T cell infiltration of tumor was observed in these mice, demonstrating the potent role of Tregs in the suppression of tumor immunity. Also described was the exciting development of the CRIPSR/Cas9 system, whereby genetic editing can be performed to specifically knock in or out genes, or even change one nucleotide to induce a mutation. Dr. Jacks and his colleagues generated lentiviral expression constructs that expressed both guide RNA (gRNA) and Cas9, and disrupted p53 and PTEN. These mice showed the onset of tumor development within 6 months of injection of lentiviral particles. Unlike conventional knockout models, where construction of targeting vectors, screening of ES cells, and generation of knockout mouse lines are quite laborious and time consuming, lentiviral delivered CRISPR/Cas9 system is rapid, reproducible, and relatively simple. 4. Immunology There were numerous sessions on advances in preclinical immunologic approaches to cancer, one of which was specific to ovarian cancer. George Coukos from the University Hospital of Lausanne presented “Opportunities for Immunotherapy in Ovarian Cancer”, which emphasized the importance of evaluating the presence or absence of tumorinfiltrating lymphocytes (TILs) in ovarian cancer. Of their cohort of patients with TILs present in tumors, greater than 60% had improved survival rates and 50% never relapsed. In comparing tumors with and
without TILs, their group found that high levels of FASL, mediated by VEGF and PGE2, encouraged T-cell death in the tumors. Blocking VEGF and PGE2 reduced FASL levels, allowing T-cell accumulation and restoration of an immunogenic tumor. In another important approach, they demonstrated that the absence of TILs predicted failure to anti-PD-1 therapy. However, this could be partially overcome by addition of a vaccine generated specifically to tumor antigens, increasing therapeutic response. Jeong Kim (Genentech) presented “Unleashing anti-tumor immunity through anti-OX40 monotherapy and in combination with anti-PDL1”. OX40 (aka Tumor necrosis factor receptor superfamily member 4 (TNFRSF4) or CD134) is a receptor expressed only on activated CD4 and CD8 lymphocytes. It is typically upregulated 48–72 h after T-cell activation, and promotes proliferation and clonal expansion of effector and memory populations. Anti-OX40 is an agonistic monoclonal antibody that is effective via a dual mechanism of 1) co-stimulating effector T-cells increasing their proliferation and the production of cytokines and 2) inhibiting regulatory T-cells. Mouse studies demonstrated that anti-OX40 treatment reduced tumor burden and established immune memory, making the mice resistant to tumor rechallenge. Additionally, combining anti-OX40 agonistic MAb with anti-PD-L1 treatment increased the therapeutic response compared to either treatment alone. Dr. Robert Schreiber from Washington University at St. Louis described the application of genomics to personalize cancer immunotherapy. He first described the concept of immunoediting involving the “3E's”: elimination, equilibrium, and escape that may account for the mechanisms of tumor cell evolution to immune evasion. For example, recombination-activating gene 2 (RAG2) is one of the 2 genes responsible for rearrangement and recombination of genes during V(D)J recombination as immunoglobulin and T-cell receptors undergo maturation. RAG2(−/−) mice develop tumors quicker than wild type, and tumors taken from Rag2(−/−) mice are rejected in Rag+/+. Tumors that developed in Rag2(+/+) mice showed reduced immunogenicity providing evidence that immunogenicity is “edited”. He went on to describe how genomics might be used to personalize cancer immunotherapy. NetMHC is a server at the Center for Biological Sequence Analysis at the Technical University of Denmark that can be used to predict which tumor-specific mutations form tumor-specific mutant rejection antigens. These could be used for rapid generation of immunotherapy approaches by tumor profiling. Finally, he presented evidence that checkpoint blocking therapy (such as anti-PD-1-based therapies) targets specific mutant neoantigens. Here, mutated proteins can serve as neoantigens that are recognized by immune cells, even if the mutant proteins were not necessarily drivers of disease progression. These neoantigens can be used to develop personalized vaccines, with data presented that such vaccines are effective in blocking immune checkpoints. Furthermore, combining neoantigen personalized vaccine and checkpoint blocking therapy was more effective than either alone. Specifically, while immune checkpoint blocking therapies are less effective when started late in the disease course, adding a personalized vaccine increased checkpoint therapy efficacy even in late stages of tumor progression. 5. Tumor dormancy and senescence Several talks were presented examining the mechanisms underlying tumor dormancy. Lewis Chodosh used an inducible HER2overexpression breast cancer model to demonstrate that the initial oncogenic pathways responsible for primary tumor cell development were not identical to those mediating recurrent disease. For example, when HER2-positive tumors were treated to no evidence of disease, recurrent tumors could be induced by re-expression of HER2, but recurrent tumors then overexpressed other genes or pathways that were minimally expressed in primary tumors, including single-stranded binding protein Ssb1, mediators of Notch signaling, c-Met, and autophagy. Recurrence could be prevented in some
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
cases by inhibition of these pathways with SsB1 knockout, gamma secretase inhibitors, c-Met inhibitors, or chloroquine. Interestingly, in their model the targeted therapies were more effective when given after primary therapy, instead of in combination with it. Although this was a HER2-inducible breast cancer model, these insights into tumor dormancy may have parallels in the dormancy noted in surviving ovarian cancer cells after primary chemotherapy. In a cellular senescence and aging symposium Scott Lowe described cell autonomous and non-cell autonomous programs of senescence. The cell autonomous pathway frequently engages p53 and Rb tumor suppressor pathways in a repressive chromatin state. The non-cell autonomous or secretory phenotype involves MMP3 and PAI-1, and can stimulate and transform pre-malignant cells. He showed that Brd4, a chromatin reader protein that regulates transcription of Myc, works with NFκB to induce senescence associated secreted phenotype (SASP). He showed that Brd4 is required for secretion, but not cell cycle arrest. Secretory factors from senescent cells are required for communication to other cells such as NK cells. Immune clearance of senescent cells occurs normally, but with impairment of Brd4, clearance of senescent cells was not observed. Instead these cells were able to escape immune surveillance. It is unclear if this action is pro-oncogenic or pro-senescent, but it likely depends on the neighboring cell. In the same session Judith Campisi discussed the induction of DNA damage by chemotherapy, which may also evoke senescence. She showed that SASP from both fibroblast and endothelial cells were involved in wound healing via the matrix protein CCN1. These data suggests that senescent cells play a role in migration and repair mechanisms. This begs the question: if chemotherapy potentially promotes senescence of tumor cells then how are these tumor cells contributing to chemoresistance and metastatic disease? Campisi hypothesized that chemotherapeutics might cause a senescent burden that actually may be detrimental. However if the clearance of senescent cells could be reactivated, such potential for a nidus of recurrent disease might be prevented. 6. Tumor microenvironment Several presentations at the meeting highlighted new findings by which tumor cells and stroma may be regulating one another in relation to the development and persistence of metastatic disease in ovarian cancer patients. Haria and co-workers determined that the homeobox gene DLX4 promotes inflammatory signaling and peritoneal metastasis in ovarian cancer patients. Specifically, DLX4 expressed in tumor cells stimulated CD44 on mesothelial cells, through activation of IL-1β and NF-κB. Inhibition of either IL-1β or NF-κB diminished the effect of DLX4 on tumor-stroma interactions. The authors suggested that inflammatory response pathways could be a potential therapeutic target for ovarian cancer. Work presented from Samuel Mok's group focused on understanding how omental obesity contributes to poor prognosis in high grade serous ovarian carcinoma. Transcriptome profiling of adipose tissue from normal patients and those with high grade serous ovarian cancer suggest that a novel adipokine, omentin (Intestinal Lactoferrin Receptor, IITLN1), was downregulated in ovarian cancer patient tissues compared to normal or benign patient's omentum. The authors propose that ITLN1 may sequester lactoferrin, which is highly expressed in ascites, preventing it from binding to low density lipoprotein receptor-related protein 1 (LRP-1) on the surface of ovarian cancer cells and activating downstream signaling pathways that regulate MMP1 expression. Furthermore, the authors showed that patients with low levels of ITLN1 prior to initial treatment had poor survival compared to those with higher levels. The authors concluded that cancer cells are capable of modifying visceral adipose tissue via down regulation of ITLN1. In the joint TME/Cancer Immunology group meeting Raghu Kalluri discussed the functional role of the stroma in tumor cell
9
immunity in pancreatic cancer. He noted that myofibroblasts in TGFBR2 flox/flox mice were able to maintain some type of control to keep tumors differentiated. Previously it has been reported that fibroblasts and the collagen-rich matrices frequently found in pancreatic cancer patients impair access of gemcitabine to cancer cells. Kalluri and co-workers determined that removing fibroblasts and then adding gemcitabine to tumor cells had no effect on cell death. However, treating these tumors with the immune checkpoint therapy anti-CTLA4 improved survival in these mice. Kalluri concluded that more myofibroblasts improve survival in pancreatic patients. These findings demonstrate the two sides of the microenvironment: 1) an instigator of tumor growth and 2) a suppressor of tumor growth. In the same session Padmanee Sharma showed that CTLA4 blockade enhances T-cell anti-tumor response, decreasing tumor burden. However, she noted that PD-L1 expression in tumors changes as a function of treatment. It is therefore important to establish methods by which immune targets can be evaluated in patients to predict response to certain treatments. 7. Emerging mechanisms of PARP inhibition Alan D'Andrea presented data on potential extension of PARP inhibitor therapy beyond BRCA-mutated cases. PARPi kills cancer cells through synthetic lethality by suppressing non-homologous endjoining DNA repair. In cells lacking BRCA1/2, this promotes alternative end-joining as the required DNA repair process, which is error-prone and ultimately lethal. A key enzyme in alternative end-joining is polymerase Q (POLQ), and data were presented that POLQ simultaneously functions to inhibit homologous repair by binding and inhibiting RAD51. Therefore POLQ inhibitors, likely in combination with PARPi, may increase the therapeutic response, or sensitize malignancies that are inherently resistant to PARP inhibitors. 8. Promising agents in clinical trials Parames Thavasu presented results on a novel mTORC1/2 inhibitor (AZD2014). In inhibiting both mTORC 1 and 2, the feedback-initiated increase in phospho-Akt that accompanies and limits effectiveness of mTORC1-selective inhibitors is avoided. The inhibitor was studied in a phase I trial in which ovarian cancer patients received a 3-day on, 4day off schedule in combination with weekly paclitaxel (80 mg). A DLT of 50 mg BID (or 75 mg BID for 2-day on, 5-day off) was identified. Regarding responses, 7 out of 10 patients demonstrated a partial response, and most of those were durable for more than 6 months. A randomized phase II trial is slated to begin later this year. 9. Cervical cancer Dr. Douglas Lowy, who initially developed the vaccine against the L1 surface protein of the Human Papilloma Virus, and has been named interim director of the National Cancer Institute, provided an overview and update of the HPV vaccination program. He highlighted that only half of HPV-associated cancers are cervical, the other half predominantly due to oropharyngeal and anal cancers, which have a high prevalence in men, arguing for screening programs in boys as well as girls. He presented supporting data that the vaccination program not only reduces cases through direct immunization in the recipient, but also through herd immunity — decreased cases due to decreased prevalence of the virus in the population. An Australian study was highlighted which showed a dramatic reduction in the incidence of genital warts, even in patients who were not vaccinated. Additionally, two new developments were presented. The first was evidence that in adolescents (9–14 yoa), 2 doses for 6 months apart are as effective as 3 doses, due to the higherthan-expected immune response in these young patients. Trials examining efficacy of just 1 dose are being designed. Secondly, the 9-valent vaccine (Gardasil-9, Merck) offers better protection against CIN2
10
J. Chien, C.N. Landen Jr. / Gynecologic Oncology 138 (2015) 7–10
disease than the quadrivalent vaccine, leading to FDA approval of this vaccine in December 2014. Acknowledgments The authors would like to thank Jill Madden and Megan Cooley at the University of Kansas Medical Center, and Robert Cornelison at the University of Virginia for their assistance in writing this review. Jeremy Chien Department of Cancer Biology, University of Kansas Medical Center, United States Translational Genomics, University of Kansas Cancer Center, United States
Charles N. Landen Department of Obstetrics and Gynecology, University of Virginia, United States Women's Oncology Program, University of Virginia Cancer Center, United States Corresponding author at: Department of Obstetrics and Gynecology, University of Virginia, United States. E-mail address: [email protected].