肿瘤干细胞Ottevanger-2017-Ovarian cancer stem cells moreVIP

Contents lists available at ScienceDirect Seminars in Cancer Biology journal homepage: Review Ovarian cancer stem cells more questions than answers Petronella Beatrix Ottevanger Department of Medical Oncology (452), Radboud University Medical Center, PO 9101, 6500 HB Nijmegen, The Netherlands A R T I C L E I N F O Keywords: Ovarian cancer Stem cell Epithelial mesenchymal transformation A B S T R A C T Epithelial ovarian cancer is a highly lethal disease, which is usually diagnosed at a late stage with extensive metastases in the abdominal cavity. Ovarian cancer either develops from the ovarian surface epithelium (OSE) or from serous intra-epithelial carcinoma (STIC). Primary therapy consists of debulking surgery and platinum based chemotherapy. The success of debulking surgery depends on surgical skills but also on the gene signature of the tumour. Debulking surgery combined with fi rst line platinum based chemotherapy, frequently leads to complete remission. However, most ovarian cancers relapse. Once the disease has relapsed, the interval between subsequent therapies decreases steadily due to rapid progression and therapy resistance. Research on therapy resistance of ovarian cancer is frequently devoted to genetic alterations in cancer cells, leading to drug inactivation, enhanced DNA repair mechanisms and intracellular pathway derangements. However the knowledge of ovarian cancer stem cells (OCSC) and the role they play in the development of cancer and therapy resistance is sparse. In this review current knowledge on the characteristics of OCSCs and the micro environmental mechanisms leading to the development or activation of OCSCs resulting in ovarian cancer is reviewed. Moreover the role of OCSC in both surgical and systemic therapy resistance and the relation with epithelial mesenchymal transformation (EMT) is discussed, as are micro-environmental signals leading to OCSC or EMT activation. 1. Introduction Ovarian cancer is the 7th most common cancer in women world- wide and the third most common female cancer after breast and cervix cancer, with an incidence rate of 1015 per 100,000. It is also the 5th most lethal cancer in women worldwide 1. Mortality rates are falling in Western countries, but this is more likely due to falling incidence rates resulting from life style changes, for example the use of contraceptives and the declined use of postmeno- pausal hormone replacement therapies than to better treatment options 2. On the contrary, current overall age standardised 5-year survival rates for ovarian cancer in Western European countries range from 30.3 to 44.1%, and these have not increased signifi cantly in the last 20 years 3,4. The poor survival rate is related to the fact that the disease is usually diagnosed at a late stage because the seeding of cancer cells in the peritoneal cavity goes for a long time without signs or symptoms. The other reason is that although many patients achieve a clinical complete remission after primary therapy, the disease recurs in most patients. With more lines of therapy multi-drug resistance develops in all patients, leading to increasingly shorter periods of remission or stable disease, fi nally resulting in death. Many explanations for this therapy resistance are available: genetic and epigenetic mutations leading to expelling or inactivation of cytotoxic drugs, impaired apoptosis en- hanced repair mechanisms and a micro environment leading to inhibi- tion of the immune system all contribute to the poor prognosis of this disease. However the role of cancer stem cells in ovarian cancer is being little explored, especially in comparison to other tumour types such as colon and prostate cancer. In this review the hypotheses of the development of ovarian cancer stem cells (OCSC) and the current knowledge of their role in treatment failure as well factors that infl uence their activity will be discussed. 2. Aetiology The aetiology of ovarian cancer is not completely elucidated. Epidemiology studies indicate that oral contraceptives play a protective role, while besides proven genetic predispositions such as mutations in the tumour suppressor genes BRCA 1 and 2, nulliparity and ovulation inducing agents increase the chance of developing ovarian cancer. Frequent ovulation, related to hormonal infl uences, might play a crucial role, although discussions remain whether this is due to an infertile state or to the use of the ovulation stimulating agents 57. Two theories have been hypothesised: the fi rst is related to the ovarian surface epithelium (OSE) (Fig. 1). This is the pelvic mesothe- /10.1016/j.semcancer.2017.04.009 Received 23 February 2017; Received in revised form 19 April 2017; Accepted 20 April 2017 E-mail address: nelleke.ottevangerradboudumc.nl. Seminars in Cancer Biology 44 (2017) 6771 Available online 24 April 2017 1044-579X/ 2017 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY license (/licenses/BY/4.0/). MARK lium covering the ovaries, which consists of cuboidal cells (OSEC). The repeated rupture of this epithelium during ovulation, the accompanying exposure to infl ammatory cytokines and the subsequent repair resulting in epithelial mesenchymal transformation (EMT), might give rise to genetic alterations leading to cancer. In the OSE also cortical inclusion cysts (CIC) develop with increasing age. It is believed that the OSE in these cysts might become metaplastic or even neoplastic due to accumulated concentrations of infl ammatory cytokines in the cysts or increased stimulation of stromal factors 8,9. It remains unclear, however in this theory, why diff erent and distinct phenotypes of ovarian cancer develop, such as high grade serous ovarian carcinoma (HSOC), endometrioid, clear cell, low grade serous carcinoma and borderline endometrioid ovarian carcinoma, each with its own genetic profi le and outcome. Neither has it been clarifi ed how extra-ovarian peritoneal cancer can develop, without malignant cells in the ovaries. The second theory is that ovarian cancer develops in the fallopian tube. Piek et al. discovered dysplastic changes in a series of prophy- lactically removed macroscopically normal fallopian tubes of women with a predisposition for ovarian cancer. These changes were accom- panied by changes in proteins related to the development of cancer, such as P53, resulting in continuous cell divisions and inhibition of apoptosis. In women without a predisposition, no such changes were found 10. Later these precursor lesions were called serous tubal intraepithelial carcinoma (STIC) lesions. These lesions were not only associated with tubal carcinoma but also with high grade serous ovarian cancer (HGSOC) 11. Moreover, endometrioid and clear cell carcinoma would, according to this theory, result from shedding of more proximal tubal tissue or from endometriosis (Fig. 1) (Table 1). Both theories have been merged into one: while the low grade ovarian cancers, such as borderline tumours and low grade serous and endometrioid carcinoma originate from CIC lesions, the high grade serous carcinoma are believed to develop from the secretory cells of the fallopian tube and especially the fi mbriae, probably induced by infl ammatory stimuli caused by ovulation. Since culturing of the secretory cells from the fallopian tube (FTSECs) is now possible more theories have evolved. In general, it is widely accepted that repeated cell divisions lead to mutations that can give rise to cancer cells. Whether these changes need to occur in somatic stem cells or in normal more diff erentiated epithelial cells is not clear. Coetzee et al. investigated the genome of the non-protein coding fraction of the human genome harbouring the cell regulatory DNA in FTSECs, OSECs and endometriosis epithelial cells (EECs). When focusing on 17 genomic regions associated with the risk of HGSOC, large similarities in the regulatory DNA of these cell types were found, such as a signifi cant enrichment for regulatory biofeatures coinciding with risk-associated single nucleotide polymorphisms (SNPs) in FTSECs and OSECs but not in EECs. The close similarity between these two cell types based on their regulatory architecture at risk loci, implies that both could be cells of origin for HGSOC 12. Interestingly enough, Flesken-Nikitin et al. put the previously mentioned theory in a completely new light. In their search for stem cells of the OSE and the location of a stem cell niche, knowing this epithelium needs continuous regeneration, they used the immunohis- tochemical stem cell marker aldehyde dehydrogenase1 (ALDH1). Previous experiments had already shown that about 7% of the OSE contained ALDH1+ cells. Using fl uorescence-activated cell sorting they separated ALDH1Highfrom ALDH1Lowcells and were able to grow spheroids from the ALDH1highpopulation. The spheroids were formed from a single-cell suspension in at least fi ve consecutive rounds of spheroid dissociation and regeneration, suggesting the selection of a stem cell like population. They then looked for the stem cell niche in the ovaries of mice of diff erent ages by immunohistochemical staining of ALDH1 in cells in the corpus luteum region, the tubal epithelium and the distal antral and hilum region of the ovary. Cells with high ALDH1 staining were predominantly located at the hilum of the ovary, where nerves and vessels enter the ovary (Fig. 1). This area contains epithelium showing the transition of OSE, tubal epithelium and mesothelium. Moreover, they showed migration of these cells to areas of damage after ovulation and an increased malignant potential of these cells after P53 mutation 13. These interesting data in mice however have not yet been confi rmed in humans. Mutant p53 is regarded as an important and early trait of develop- ing HGSOC and might be induced by release of follicular fl uid at ovulation. Follicular fl uid also induces intracellular reactive oxygen species (ROS), resulting in up-regulation of ROS-induced miR-182. In normal FTSECs, miR-182 overexpression triggers cellular senescence by p53-mediated up-regulation of p21. In FTSECs with mutant P53 (mP53), however, this miRNA acts as an oncogene leading to further DNA damage and cancer 14. Recently KDM3A, a lysine demethylase, was shown in cell line experiments and xenograft models to demethy- late P53, leading to a cancer stem cell phenotype 15. Moreover it has been postulated that mP53 also induces early cell migration leading to intra-abdominal metastases 16. 3. Prognosis and therapy For epithelial ovarian cancer FIGO stage I to IV the 5 year survival rates fall from almost 90% to approximately 20% and results have not improved signifi cantly in recent years (Table 2) 1719. Treatment of ovarian cancer depends on the stage of the disease at the moment of diagnosis and usually consists of surgery and che- motherapy. Fig. 1. Putative ovarian cancer origin. 1. Ovary. 2. Fallopian tube 3. Endometrium/ endometriosis. 4. Hilum of the ovary (stem cell niche). Table 1 Histological types of epithelial ovarian cancer and their origin. LocationCells of originOvarian cancer histological type OvaryOvarian surface epithelium transforming to CICLow grade serous carcinoma High grade serous carcinoma Distal fallopian tubeSecretory epithelial cells transforming to STICLow grade serous carcinoma High grade serous carcinoma Proximal fallopian tube and endometrium/endometriosisSecretory epithelial cells and basal cellsEndometrioid carcinoma and clear cell carcinoma Hilum of the ovaryOvarian cancer stem cellsLow grade serous carcinoma High grade serous carcinoma CIC: cortical inclusion cysts, STIC: serous tubal intraepithelial carcinoma. P.B. OttevangerSeminars in Cancer Biology 44 (2017) 6771 68 3.1. Surgery For early stage disease (FIGO stage IIIA) it is important to remove all visible tumour, but also to exclude higher stages of disease by sampling of lymph nodes and tissue from fi xed locations throughout the abdominal cavity. For patients with stage IIB to IV it is of utmost importance to accomplish a complete debulking, i.e. removing all visible tumour. In a meta-analysis of trials, Dubois et al. showed a stage dependent improvement in median overall survival (OS) if all visible tumour could be removed: for stage IIB-IIIB median OS increased from 48.3 to 100.6 months, for IIIC from 34.2 to 81.1 months and for stage IV from 24.6 to 54.6 months 20. If complete debulking is not to be expected due to extensive disease, neo-adjuvant chemother- apy followed by surgery is an option 21,22. Complete debulking should always be the goal. Surgical skills are, of course, very important, but it is questionable whether this is all that matters. Evolving evidence points to the fact that tumour characteristics may be crucial for successful debulking surgery. Two studies have now shown that the genetic make-up of the tumour infl uences the chance of a successful debulking: Tucker et al. used 2 datasets, from The Cancer Genome Atlas (TCGA) and Tothill cohorts, selecting patient cohorts with chemotherapy nave tumour tissue after primary debulking. By using a reverse transcription polymerase chain reaction (RT-PCR), they discovered that high fatty acid binding protein 4 (FABP4) and alcohol dehydrogenase 1B (ADH1B) expression were signifi cantly related to residual disease after debulking. These fi ndings were validated in 2 other separate datasets and results remained signifi cant 23. The role of ADH1B in ovarian cancer is unclear, FABP4 is related to angiogenesis and omental metastasis by stimulating C-KIT and stem cell factor 24,25. Riester et al. even used 13 publicly available datasets containing 1525 patients in total. They trained prediction models using a meta-analysis variation on the compound co-variable method and tested predictive models by a “leave-one- dataset-out” procedure. They validated the results in 2 other separate cohorts by quantitative RT-PCR(qRT-PCR) and immunohistochemistry. POSTN, CXCL14, FAP, NUAK1, PTCH1, and TGFBR2 were validated by qRT-PCR (P .05) and POSTN, CXCL14, and phosphorylated Smad2/ 3 were also validated by immunohistochemistry (P .001) as inde- pendent predictors of debulking status. With these last 3 genes 92.8% of the debulking surgeries could be predicted adequately in complete or incomplete debulking 26. The selected genes were positively related to adhesion, invasion, angiogenesis and EMT. 3.2. Systemic therapy For almost all tumour stages, except for well or intermediately diff erentiated FIGO stage Ia ovarian carcinoma, additional chemother- apy is indicated. Chemotherapy should consist of carboplatin or cisplatin and paclitaxel as agreed during the 4th Ovarian Cancer Consensus meeting in Vancouver 27. Schedules with intra-peritoneal administration of cisplatin and paclitaxel or dose-dense schedules are also allowed, since both have shown increased OS compared to the standard 3-weekly carboplatin or cisplatin combinations with paclitaxel 28,29. Although these chemotherapy schedules lead to clinical complete remission in a large proportion of patients, most tumours recur within 2 years. The addition of various angiogenesis inhibitors in fi rst line therapy has resulted in only a small improvement in progression free survival (PFS) but OS benefi t has not been accomplished 3033. Once the disease has recurred, the interval between subsequent therapies decreases because progression and chemotherapy resistance develop fast. No second or further lines of therapy, consisting of either chemotherapy or targeted therapy or a combination thereof have shown survival benefi t, yet. For platinum sensitive disease (recurrence more than 6 months after last platinum administration) impressive PFS gains have been reported recently, by the use of maintenance therapy with poly ADP ribose polymerase (PARP) inhibitors 3436. Hope exists that with longer follow up these drugs will improve OS for at least a part of the ovarian cancer population. 3.3. Resistance to platinum Focusing on HGSOC, various molecular mechanisms play a role: reduced infl ux of platinum due to down regulation of trans membrane transporters, increased effl ux of platinum by multidrug resistant proteins and increased intracellular inactivation by gluthation S transferase. Additionally increased DNA repair mechanisms such as PARP, are able to counteract cell death caused by DNA damage. Finally, defects in signalling transduction pathways may prohibit apoptosis. The most important protein in this last failing mechanism is mutant P53, but defects in the MAPkinase pathway might also impair apoptosis 37. Whether these changes occur primarily in the OCSCs and are trans- ferred to the progenitor cell populations, or occur de novo in the more diff erentiated cancer cells as well, is unknown. 3.4. Ovarian cancer stem cells Only more recently, focus on ovarian cancer stem cells as the origin of therapy resistance has increased. Bapat et al. were the fi rst who cultured so-called cancer stem cells or induced progenitor stem cells from the ascites of an ovarian cancer patient. They isolated a single tumourigenic clone among a mixed population of cells from the ascites. From this clone another clone evolved. Both clones showed diff erentia- tion features but were also able to continuously give rise to new tumours if transplanted in the mouse peritoneal cavity. They suggested that stem cell transformation could be the cause of ovarian cancer and that random events of stem and progenitor cell transformation ulti- Table 2 5-Year overall survival rates of ovarian cancer according to FIGO stage. FIGO stageTumour extension5-year survival rate IaTumour limited to one ovary or one fallopian tube, no tumour on the ovarian or fallopian tube surface (capsule intact), no malignant cells in ascites or peritoneal washings 94% IbT