- Email: [email protected]
The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions
The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions Ben Davidson MD PhD, Arild Holth BSc, Ellen Hellesylt MSc, Tuan Zea Tan PhD, Ruby Yun-Ju Huang MD PhD, Claes Trop´e MD PhD, Jahn M. Nesland MD PhD, Jean Paul Thiery PhD PII: DOI: Reference:
S0046-8177(14)00410-9 doi: 10.1016/j.humpath.2014.10.004 YHUPA 3431
To appear in:
Human Pathology
Received date: Revised date: Accepted date:
21 July 2014 29 September 2014 1 October 2014
Please cite this article as: Davidson Ben, Holth Arild, Hellesylt Ellen, Tan Tuan Zea, Huang Ruby Yun-Ju, Trop´e Claes, Nesland Jahn M., Thiery Jean Paul, The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions, Human Pathology (2014), doi: 10.1016/j.humpath.2014.10.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions
T
Running head: EMT markers in ovarian serous carcinoma
RI P
Ben Davidson, MD PhD1,2, Arild Holth, BSc1, Ellen Hellesylt, MSc1, Tuan Zea Tan, PhD3, Ruby Yun-Ju Huang, MD PhD3,4, Claes Tropé, MD PhD2,5, Jahn M. Nesland, MD PhD1,2,
SC
Jean Paul Thiery, PhD3,6,7
MA NU
Departments of Pathology1 and Gynecologic Oncology5, Oslo University Hospital, Norwegian Radium Hospital, N-0424 Oslo, Norway; 2University of Oslo, Faculty of Medicine, Institute of Clinical Medicine, N-0424 Oslo, Norway; Cancer Science Institute (CSI)3, Department of Obstetrics and Gynecology4 and Department of Biochemistry6, Yong
ED
Loo Lin School of Medicine, National University of Singapore, MD6 Centre for Translational Medicine, 14 Medical Drive, Singapore, 117599; Institute of Molecular and Cell Biology
CE
PT
(IMCB) A*STAR, 61 Biopolis Drive, Proteos, Singapore, 1386737
Financial acknowledgment: This work was supported by Inger and John Fredriksen
AC
Foundation for Ovarian Cancer Research.
Corresponding author Ben Davidson, MD PhD Department of Pathology Norwegian Radium Hospital Oslo University Hospital Montebello N-0310 Oslo Norway Tel: (47) 22934871 Fax: (47) 22508554 Email: [email protected]
ACCEPTED MANUSCRIPT 2
Summary We recently identified gene signatures which allow classification of ovarian carcinoma into 5
T
distinct clinically-relevant groups. In the present study, we investigated the clinical role of 10
RI P
protein products of the discriminating genes, with focus on epithelial-mesenchymal transition and stem cell markers. Expression of E-cadherin, N-cadherin, P-cadherin, Zeb1, HMGA2,
SC
Rab25, CD24, NCAM (CD56), Sox11 and vimentin was assessed in 100 advanced-stage (FIGO III-IV) serous ovarian carcinoma effusions using immunohistochemistry. Results were
MA NU
analyzed for association with clinicopathologic parameters, including chemotherapy response, and survival. All 10 proteins were frequently expressed in carcinoma cells. HMGA2 expression was related to older age (p=0.015). HMGA2 and NCAM expression was related to
ED
stage III disease (p=0.011 and p=0.023, respectively) and NCAM was overexpressed in peritoneal compared to pleural effusions (p=0.001). Vimentin and Zeb1 expression was
PT
significantly related to poor chemotherapy response at diagnosis (p=0.005 and p=0.017, respectively). The associations between NCAM and peritoneal localization and of vimentin
CE
and poor chemoresponse were retained after Bonferroni correction. NCAM expression was
AC
associated with a trend for shorter overall survival in univariate survival analysis (p=0.187), but emerged as independent prognosticator in Cox multivariate analysis (p=0.042). This study identifies Vimentin and Zeb1 as markers of poor chemoresponse in metastatic serous ovarian carcinoma effusions and suggests NCAM as potential prognostic marker in metastatic disease. The generally limited prognostic role of the studied markers emphasizes the difficulty in applying data obtained in studies of primary ovarian carcinomas to analyses of ovarian carcinoma effusions, reflecting the unique biology of the latter.
Keywords: ovarian carcinoma; epithelial-mesenchymal transition; effusion; chemoresistance; survival
ACCEPTED MANUSCRIPT 3
Introduction Ovarian cancer, consisting predominantly of ovarian carcinoma (OC), is the most fatal
T
gynecological malignancy. The combination of aggressive cytoreductive surgery and
RI P
neoadjuvant or adjuvant platinum/paclitaxel-based chemotherapy has prolonged survival in OC. However, chemotherapy resistance remains a major determinant of treatment failure and
SC
unfavorable clinical outcome, and the majority of patients consequently still die of their disease [1,2]. The development of malignant peritoneal, and less commonly, pleural effusions
MA NU
constitutes an almost universal finding in advanced-stage OC [3], and OC cells in effusions have been growingly perceived to represent a chemoresistant population with cancer stem cell characteristics [4,5], highlighting the need to better understand their genotypic and phenotypic
ED
characteristics.
PT
Epithelial-mesenchymal transition (EMT) is a process by which epithelial cells assume mesenchymal characteristics, facilitating migration through the extracellular matrix (ECM)
CE
and settlement in areas of new organ formation during embryogenesis. Wound-healing
AC
represents another form of physiological EMT, whereas pathological EMT occurs in tissue fibrosis and cancer [6-8]. EMT is induced by multiple signals, including growth factors, the Wnt signaling pathway, integrins, Notch transcription factors, prostaglandin E2 (PGE2), cyclooxygenase-2 (COX-2) and hormones [8]. During EMT, carcinoma cells lose their epithelial characteristics and acquire mesenchymal properties that promote ECM invasion and distant metastasis. This occurs through downregulation of E-cadherin, cytokeratins, ZO-1, claudins, occludin, laminin-1, entactin, MUC-1, and the microRNA 200 (miR-200) family, and acquisition of the transcription factors Snail1, Snail2, Twist, Zeb1 and Zeb2/SIP1, E47, KLF8, E2.2, Goosecoid, LEF-1 and FoxC2, as well as N-cadherin, vimentin, fibronectin, miR10b and miR21 [8-10].
ACCEPTED MANUSCRIPT 4
In OC, particularly serous carcinoma, the balance between the epithelial and mesenchymal phenotypes is complex, in part due to the inherent nature of serous carcinoma cells to express
T
mesenchymal markers, such as vimentin and N-cadherin. During OC progression, both EMT
RI P
and the reverse process, mesenchymal-epithelial transition (MET) occur, and OC cells in effusions upregulate E-cadherin and repress Snail1 and Snail2 expression compared to
SC
primary carcinomas [11-13].
MA NU
We recently generated a classification system for OC based on the gene expression patterns of 1,538 OC, including both public databases and our tumor material. OC could be classified into 5 sub-groups – Epi-A, Epi-B, Mes, Stem-A and Stem-B, of which Epi-B and Stem-A signatures were independent prognosticators of good and poor survival, respectively [14]. In
ED
the current study, we evaluated the expression of 10 protein products of genes identified by
PT
Significance Analysis of Microarray (SAM) [15] as differentiators between the above groups. Among these 10 genes, 4 were significantly down-regulated and 6 were significantly up-
CE
regulated in either Mes or Stem-A tumors (SAM q<10), the groups which had the worst
AC
outcome. Proteins were chosen based on antibody availability and performance from an initial group of 20 markers. In order to minimize the effect of clinical parameters on outcome, we analyzed a homogenous series of 100 pre-chemotherapy effusions tapped at diagnosis from 100 patients with FIGO stage III-IV serous OC. All patients received platinum-based chemotherapy. Immunohistochemistry (IHC) results were analyzed for association with clinicopathologic parameters, chemoresponse and survival.
ACCEPTED MANUSCRIPT 5
Materials and methods Patients and material
T
Specimens and clinical data were obtained from the Department of Gynecologic Oncology,
RI P
Norwegian Radium Hospital during the years 1998-2006. Informed consent was obtained
Committee for Medical Research Ethics in Norway.
SC
according to institutional and national guidelines. Study approval was given by the Regional
Fresh non-fixed serous carcinoma effusions (n=100; 72 peritoneal, 28 pleural; 80 OC, 16
MA NU
peritoneal carcinomas, 4 tubal carcinomas) were obtained pre-chemotherapy at diagnosis from 100 patients. Due to their closely-linked histogenesis and phenotype, all tumors are referred to as OC henceforth. Fifty-two patients had primary surgery, 36 had secondary debulking and 12
combined with paclitaxel.
ED
patients only received chemotherapy. All patients received platinum-based therapy, 82
PT
Effusions were submitted for routine diagnostic purposes and processed immediately after tapping. Cell blocks were prepared using the Thrombin clot method. Diagnoses were
CE
established using morphology and IHC. Clinicopathologic data are detailed in Table 1.
IHC
AC
Grading was according to the International Federation of Gynecology and Obstetrics (FIGO).
Formalin-fixed paraffin-embedded sections were analyzed for protein expression of Ecadherin, N-cadherin, P-cadherin, Zeb1, CD24, NCAM (CD56), Sox11 and vimentin using the Dako FLEX+ protocol (Dako, Glostrup, Denmark), with the exception of CD24, stained using the Dako Envision protocol. HMGA2 and Rab25 immunostaining data were retrieved from previous studies in which larger series, obtained both at diagnosis and disease recurrence, were analyzed [16,17]. Antibody and IHC data are provided in Table 2. Visualization was achieved using 3`3-diaminobenzidine tetrahydrochloride substrate (DAB)
ACCEPTED MANUSCRIPT 6
and hematoxylin counterstaining. Positive controls, chosen based on manufacturer datasheets, were satisfactory in all reactions.
T
Negative controls consisted of sections that underwent similar staining procedure with normal
nonrelevant rabbit IgG1 (Sigma Aldrich, St Louis MO).
RI P
IgG of the corresponding isotype as the primary antibody (for mouse antibodies) or with a
SC
Staining extent was scored as 0-4, corresponding to 0%, 1-5%, 6-25%, 26-75% and 76-100% stained carcinoma cells (minimum 100 cells required).
MA NU
Scoring was performed by two surgical pathologists (BD and JMN), both blinded to clinical data. Inter- and intra-observer concordance was >80%, with discrepant cases generally of one scoring level. These were discussed until agreement was reached.
ED
Statistical analysis
PT
Statistical analysis was performed applying the SPSS-PC package (Version 18.0, Chicago IL). Probability of <0.05 was considered significant. Analysis of the association between different
CE
biomarker immunostains were performed using the Chi-square test. Analysis of the
AC
association between biomarker expression and clinicopathologic parameters were undertaken using the Kruskal-Wallis H test (for grade, 3-tier analysis) or the Mann-Whitney U test (all other analyses). Clinicopathologic parameters were grouped as follows: Age: ≤60 vs. >60 years; grade: 1 vs. 2 vs. 3; effusion site: peritoneal vs. pleural; FIGO stage: III vs. IV; chemotherapy status: pre- vs. post-chemotherapy specimens; residual disease volume (RD): ≤1cm vs. >1cm; chemotherapy response: complete vs. partial response/stable disease/progression; primary (intrinsic) chemoresistance: progression-free survival (PFS) at ≤6 months vs. >6 months. Chemotherapy response was defined according to RECIST criteria. PFS and overall survival (OS) were calculated from the date of the last chemotherapy treatment/diagnosis to the date of recurrence/death or last follow-up, respectively. Univariate
ACCEPTED MANUSCRIPT 7
survival analyses of PFS and OS were executed using the Kaplan-Meier method and log-rank test. Expression categories in survival analyses were clustered as high vs. low based on IHC
T
score of 0-2 vs. 3-4, so as to allow for a sufficient number of cases to be included in each
RI P
category. Multivariate analysis was performed using the Cox proportional hazard model
AC
CE
PT
ED
MA NU
SC
(Enter function).
ACCEPTED MANUSCRIPT 8
Results EMT and stem cell-related markers are frequently expressed in OC effusions
T
IHC analysis demonstrated expression of all studied proteins except Sox11 in the majority of
RI P
OC effusions (Figure 1). E-cadherin, N-cadherin and P-cadherin expression was predominantly seen in >25% of cells, whereas the remaining markers had generally more
SC
limited expression (Table 3). The high number of cells expressing all 3 cadherins precluded any statistical analysis with respect to the association between these proteins. However, we
MA NU
studied the relationship between cadherin expression and staining for HMGA2, Zeb1 and vimentin. N-cadherin expression was significantly associated with that of vimentin (p=0.014) and a trend was observed with respect to Zeb1 (p=0.068). The remaining analyses were
ED
negative.
PT
Vimentin expression is related to primary (intrinsic) chemoresistance and poor chemotherapy response at diagnosis
CE
Expression of the 10 studied proteins was analyzed for association with clinicopathologic
AC
parameters and chemotherapy response at diagnosis. HMGA2 expression was significantly related to older age (p=0.015). HMGA2 and NCAM were significantly overexpressed in FIGO stage III compared to stage IV tumors (p=0.011 and p=0.023, respectively), and NCAM was overexpressed in peritoneal compared to pleural effusions (p=0.001). Expression of P-cadherin, NCAM and CD24 was higher in grade 1 compared to grade 2 and 3 tumors (p=0.005, p=0.044, p=0.041, respectively). However, grade 1 group consisted of only 5 tumors. Vimentin and Zeb1 expression was significantly related to poor (other than complete) response to chemotherapy at diagnosis (p=0.005 and p=0.017, respectively; Table 4-A). The observation for vimentin remained significant when complete or partial response was
ACCEPTED MANUSCRIPT 9
analyzed against stable disease or progression (p=0.008) and became still stronger when patients with complete response were analyzed against those with progression (p=0.001; data
T
not shown). Vimentin expression was further significantly associated to primary (intrinsic)
RI P
chemoresistance (p=0.038; Table 4-A).
SC
NCAM expression is significantly associated with poor OS
Survival data were available for all patients. The follow-up period ranged from 6-110 months
MA NU
(median=28 months). PFS ranged from 0-103 months (median=7 months), with 16 patients never achieving a progression-free period. At last follow-up, 3 patients were alive without disease, 8 were alive with disease and 89 were dead of disease. In univariate survival analysis, low E-cadherin expression was significantly related to poor OS (p=0.043), with trends for
ED
poor OS observed for NCAM (p=0.187) and vimentin (p=0.111; Figure 2). Notably, low
PT
expression of E-cadherin was observed in only 4 patients. Clinical parameters associated with OS were RD (p=0.037) and, marginally, FIGO stage (p=0.05).
CE
The parameters entered into the Cox multivariate analysis were E-cadherin, NCAM and
AC
vimentin protein expression, RD volume and FIGO stage. In this analysis, RD after primary surgery or secondary debulking (p=0.035), FIGO stage (p=0.046) and NCAM expression (p=0.042) were independent predictors of OS. Data for IHC and clinicopathologic parameters with p<0.2 in univariate analysis are summarized in Table 4-B.
ACCEPTED MANUSCRIPT 10
Discussion The present study validated the role of 10 proteins related to EMT and cancer stem cell
T
phenotype in sub-classifying OC with respect to chemoresponse and survival. There are
RI P
several notable differences between the present cohort and the reference publication [14], as well as most papers dealing with this issue. The present study focused exclusively on
SC
metastatic effusions rather than on primary carcinomas, analyzed only chemo-naïve specimens obtained at diagnosis, and included only patients diagnosed with FIGO stage III-IV
MA NU
serous OC and treated with platinum-based chemotherapy, the majority with combination platinum-paclitaxel. Such inclusion criteria render it difficult to obtain significant differences with respect to both established clinicopatologic parameters and biological markers. The
ED
relative homogeneity of this series is further reflected in the almost universally poor patient outcome, which renders prognostic stratification based on biological endpoints difficult.
PT
There are nevertheless several noteworthy findings in our data.
CE
The clinical role of cadherins in OC is still undecided, based on studies which have largely
AC
focused on primary OC [reviewed in 13]. An obvious difficulty in assessing this relationship in OC effusions lies in the ubiquitous expression of cadherins, mainly E-cadherin and Ncadherin, at this anatomic site. Our current data suggest that cadherin expression in serous OC effusion specimens is not a strong predictor of chemoresponse or prognosticator. It is nevertheless noteworthy that the 4 patients who had tumors with low E-cadherin expression had significantly shorter OS, suggesting that the inability to undergo MET at this anatomic site may be a characteristic of aggressive cancers. Further evaluation of this hypothesis requires analysis of a larger series.
ACCEPTED MANUSCRIPT 11
These results suggest that the majority of OC cells in effusions have reacquired epithelial-like features such as E-cadherin through partial MET. However, OC effusions exhibit a similar
T
distribution of molecular histotypes as primary OC [14]. Primary tumors exhibiting a
RI P
mesenchymal or stem-A molecular histotype have EMT scores which favor a mesenchymallike phenotype [manuscript in preparation]. Thus, it would be of interest to establish the score
SC
of effusions to determine whether E-cadherin re-expression affects the EMT score rather than the molecular subtype. Notably, co-expression of E-cadherin with N-cadherin has been
MA NU
hypothesized to affect E-cadherin function through a trans-dominant effect of N-cadherin promoting invasion and metastasis, albeit through mechanisms which remain partially uncovered [18]. The co-expression of the two cadherins could then also impact invasion of the mesothelial layer upon re-attachment and induce chemoresistance.
ED
A study carried out on OC cell lines has shown that those with the intermediate mesenchymal
PT
phenotype characterized by N-cadherin, vimentin and cytoketatin expression are more clonogenic and resistant to anoikis than some intermediate epithelial cell lines. However,
CE
intermediate epithelial cell lines were more migratory and invasive than epithelial cell lines.
AC
OC Effusions can thus be considered a mixed phenotype between intermediate epithelial and intermediated mesenchymal as defined by Huang et al. [19]. It is intriguing to consider that OC effusions retain very aggressive behavior albeit expressing E-cadherin. A similar situation has been encountered in luminal breast cancers which are much insensitive to anthracycline and platinum-based therapy than the more mesenchymal-like triple-negative breast cancers [20,21]. Using a similar collection of OC cell lines, a recent study showed that OC epithelial-like cell lines are more resistant to platinum-salt than their mesenchymal-like counterparts through the inhibition of apoptosis mediated by NF-κB [22].
ACCEPTED MANUSCRIPT 12
In contrast to cadherins, higher vimentin expression in this cohort was strong predictor of chemoresponse. Higher vimentin expression has been reported in OC cells resistant to both
T
platinum and paclitaxel compared to their chemosensitive counterparts [23-25]. However, to
RI P
the best of our knowledge, the present study is the first to demonstrate this association in clinical material. Of note, vimentin expression was both related to chemoresponse as
SC
measured after treatment at diagnosis and to intrinsic resistance, evidenced as progression within 6 months following completion of primary treatment. Significant association with poor
MA NU
chemoresponse at diagnosis was additionally observed for Zeb1, mediator of EMT via Ecadherin repression, though less strongly than for vimentin. Our present data with respect to vimentin differ from the above-discussed finding that OC lines with epithelial phenotype are more platinum-resistant [22], reflecting inherent differences between in vitro models and
PT
ED
clinical specimens studied unaltered by culturing conditions.
NCAM is a well-established marker of sex cord-stromal tumors and neuroendocrine OC [26].
CE
However, its expression in OC without neuroendocrine differentiation has been reported.
AC
NCAM was expressed in 33/95 (34.7%) of serous carcinomas compared to 5/40 (12.5%) serous borderline tumors and 1/23 (4.3%) of serous cystadenomas and its expression in carcinomas was significantly associated with advanced FIGO stage and peritoneal metastasis in one study [27]. Analysis of 252 primary OC and 206 solid metastases of different histotype showed overexpression of NCAM compared to normal ovaries and cystadenomas, with expression seen in 23.8% and 34.5% of primary carcinomas and metastases, respectively. NCAM expression was higher in tumors of higher histological grade. NCAM further increased tumor migration and invasion in vitro and metastasis in mouse model via interaction with FGFR [28]. In our cohort, NCAM was an independent prognostic marker of poor OS,
ACCEPTED MANUSCRIPT 13
this constituting the first report of such role in OC, further supporting its association with
T
aggressive disease in this cancer.
RI P
Sox11 expression was previously reported to be marker of longer survival in 2 studies of 76 and 154 primary OC of different histotype by one group [29,30]. Our data argue against such
SC
role in serous OC effusions. Similarly, the present study supports our previous report [17] regarding the absence of predictive or prognostic role for Rab25 at this anatomic site.
MA NU
CD24 was found to be present in exosomes isolated from OC ascites [31]. Its expression in this tumor was reported to be related to stem cell phenotype and to an invasive mesenchymal phenotype [32,33]. Cytoplasmic expression of CD24 in solid OC specimens is associated with poor survival [34]. In contrast, absence of CD24 (CD44+/CD24- phenotype) rather than
ED
expression was related to poor chemoresponse and shorter PFS in another study [35]. Our
PT
data do not support any of these roles for CD24 in serous OC effusions.
CE
In conclusion, analysis of 100 serous OC effusions for the expression of 10 EMT and stem
AC
cell markers revealed a potential relationship to more aggressive clinical course for NCAM, Zeb1 and vimentin. While several of the associations observed with parameters such as patient age, histological grade, effusion site and FIGO stage are lost following Bonferroni correction, the ones between NCAM and peritoneal localization, between P-cadherin and grade 1 disease, and between vimentin expression and poor chemoresponse is retained, of which the latter merits further attention. Similarly, further analysis of the role of NCAM in metastatic OC may identify a role as therapeutic target in this cancer. As previously shown by our group and evidenced in the present study, EMT along OC progression may not follow the same sequence as in other epithelial cancers.
ACCEPTED MANUSCRIPT 14
References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62:10-
T
29.
RI P
2. Hennessy BT, Coleman RL, Markman M. Ovarian cancer. Lancet 2009;374;1371-82. 3. Davidson B. Ovarian and Primary Peritoneal Carcinoma. In: Davidson, Firat, Michael
SC
(Editors). Serous Effusions- Etiology, Diagnosis, Prognosis and Therapy. London: Springer; 2012.
MA NU
4. Latifi A, Luwor RB, Bilandzic M, et al. Isolation and characterization of tumor cells from the ascites of ovarian cancer patients: molecular phenotype of chemoresistant ovarian tumors. PLoS One 2012;7:e46858.
ED
5. Vathipadiekal V, Saxena D, Mok SC, Hauschka PV, Ozbun L, Birrer MJ. Identification of a potential ovarian cancer stem cell gene expression profile from advanced stage papillary
PT
serous ovarian cancer. PLoS One 2012;7:e29079. 6. Acloque H, Adams MS, Fishwick K, Bronner-Fraser M, Nieto MA. Epithelial-
CE
mesenchymal transitions: the importance of changing cell state in development and disease. J
AC
Clin Invest 2009;119:1438-49. 7. Thiery JP. Epithelial-mesenchymal transitions in tumour progression. Nat Rev Cancer 2002;2:442-54.
8. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in development and disease. Cell 2009;139:871-90. 9. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Invest 2009;119:1420-8. 10. Zeisberg M, Neilson EG. Biomarkers for epithelial-mesenchymal transitions. J Clin Invest 2009;119:1429-37.
ACCEPTED MANUSCRIPT 15
11. Davidson B, Berner A, Nesland JM, et al. E-cadherin and alpha-, beta-, and gammacatenin protein expression is up-regulated in ovarian carcinoma cells in serous effusions. J
T
Pathol 2000;192:460-9.
RI P
12. Elloul S, Silins I, Tropé CG, Benshushan A, Davidson B, Reich R. Expression of Ecadherin transcriptional regulators in ovarian carcinoma. Virchows Arch 2006;449:520-8.
SC
13. Davidson B, Trope' CG, Reich R. Epithelial-mesenchymal transition in ovarian carcinoma. Front Oncol 2012;2:33.
MA NU
14. Tan TZ, Miow QH, Huang RY, et al. Functional genomics identifies five distinct molecular subtypes with clinical relevance and pathways for growth control in epithelial ovarian cancer. EMBO Mol Med 2013;5:983-98.
15. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the
ED
ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116-21.
PT
16. Hetland TE, Holth A, Kærn J, Flørenes VA, Tropé CG, Davidson B. HMGA2 protein expression in ovarian serous carcinoma effusions, primary tumors and solid metastases.
CE
Virchows Arch 2012;460:505-13.
AC
17. Brusegard K, Stavnes HT, Nymoen DA, Flatmark K, Trope' CG, Davidson B. Rab25 is overexpressed in Müllerian serous carcinoma compared to malignant mesothelioma. Virchows Arch 2012;460:193-202. 18. Gheldof A, Berx G. Cadherins and epithelial-to-mesenchymal transition. Prog Mol Biol Trans Sci 2013;116:317-36. 19. Huang RY, Wong MK, Tan TZ, et al. An EMT spectrum defines an anoikis-resistant and spheroidogenic intermediate mesenchymal state that is sensitive to e-cadherin restoration by a src-kinase inhibitor, saracatinib (AZD0530). Cell Death Dis 2013;4:e915. 20. Carey LA, Dees EC, Sawyer L, et al. The triple negative paradox: primary tumor chemosensitivity of breast cancer subtypes. Clin Cancer Res 2007;13:2329-34.
ACCEPTED MANUSCRIPT 16
21. Silver DP, Richardson AL, Eklund AC, et al. Efficacy of neoadjuvant Cisplatin in triplenegative breast cancer. J Clin Oncol 2010;28:1145-53.
T
22. Miow QH, Tan TZ, Ye J, et al. Epithelial-mesenchymal status renders differential
RI P
responses to cisplatin in ovarian cancer. Oncogene 2014 May 26. doi: 10.1038/onc.2014.136. [Epub ahead of print]
SC
23. Kajiyama H, Shibata K, Terauchi M, et al. Chemoresistance to paclitaxel induces epithelial-mesenchymal transition and enhances metastatic potential for epithelial ovarian
MA NU
carcinoma cells. Int J Oncol 2007;31:277-83.
24. Haslehurst AM, Koti M, Dharsee M, et al. EMT transcription factors snail and slug directly contribute to cisplatin resistance in ovarian cancer. BMC Cancer 2012;12:91. 25. Rosanò L, Cianfrocca R, Spinella F, et al. Acquisition of chemoresistance and EMT
ED
phenotype is linked with activation of the endothelin A receptor pathway in ovarian
PT
carcinoma cells. Clin Cancer Res 2011;17:2350-60. 26. McCluggage WG, McKenna M, McBride HA. CD56 is a sensitive and diagnostically
AC
2007;26:322-7.
CE
useful immunohistochemical marker of ovarian sex cord-stromal tumors. Int J Gynecol Pathol
27. Cho EY, Choi Y, Chae SW, Sohn JH, Ahn GH. Immunohistochemical study of the expression of adhesion molecules in ovarian serous neoplasms. Pathol Int 2006;56:62-70. 28. Zecchini S, Bombardelli L, Decio A, et al. The adhesion molecule NCAM promotes ovarian cancer progression via FGFR signalling. EMBO Mol Med 2011;3:480-94. 29. Brennan DJ, Ek S, Doyle E, et al. The transcription factor Sox11 is a prognostic factor for improved recurrence-free survival in epithelial ovarian cancer. Eur J Cancer 2009;45:1510-7. 30. Sernbo S, Gustavsson E, Brennan DJ, et al. The tumour suppressor SOX11 is associated with improved survival among high grade epithelial ovarian cancers and is regulated by reversible promoter methylation. BMC Cancer 2011;11:405.
ACCEPTED MANUSCRIPT 17
31. Runz S, Keller S, Rupp C, et al. Malignant ascites-derived exosomes of ovarian carcinoma patients contain CD24 and EpCAM. Gynecol Oncol 2007;107:563-71.
T
32. Gao MQ, Choi YP, Kang S, Youn JH, Cho NH. CD24(+) cells from hierarchically
RI P
organized ovarian cancer are enriched in cancer stem cells. Oncogene 2010;29:2672-80. 33. Kang KS, Choi YP, Gao MQ, et al. CD24+ ovary cancer cells exhibit an invasive
SC
mesenchymal phenotype. Biochem Biophys Res Commun 2013;432:333-8. 34. Kristiansen G, Denkert C, Schlüns K, Dahl E, Pilarsky C, Hauptmann S. CD24 is
Am J Pathol 2002;161:1215-21.
MA NU
expressed in ovarian cancer and is a new independent prognostic marker of patient survival.
35. Meng E, Long B, Sullivan P, et al. CD44+/CD24- ovarian cancer cells demonstrate cancer
AC
CE
PT
ED
stem cell properties and correlate to survival. Clin Exp Metastasis 2012;29:939-48.
ACCEPTED MANUSCRIPT 18
Table 1: Clinicopathologic data of the study cohort (100 patients) Age (Mean; Range)
63; 39-86
52
IV
48
SC
Grade a
RI P
III
T
FIGO stage
5
II
MA NU
I
26
III
51
NA b
18
≤1 cm
PT
>1 cm
ED
Residual disease
CE
NA
48 38 14
Chemotherapy response at diagnosis 50
Partial response
20
Stable disease
7
Progressive disease
17
NA
6
AC
Complete
a
Corresponding to 5 low-grade and 77 high-grade serous carcinomas based on the 2014
WHO classification b
NA= non available
ACCEPTED MANUSCRIPT 19
Table 2: Antibodies and staining conditions Antibody
Clone
Source
Manufacturer
Dilution
Antigen retrieval
Mouse
Zymed Laboratories (San Francisco, CA)
1:3000
T
HECD-1
RI P
E-cadherin
Tris-EDTA
3B9
Mouse
Invitrogen (Carlsbad, CA)
1:300
Tris-EDTA
P-cadherin
56
Mouse
BD Transduction Laboratories
1:500
Tris-EDTA
1:200
Citrate
1:50
Tris-EDTA
1:1600
Tris-EDTA
1:500
Tris-EDTA
1:500
Tris-EDTA
1:500
Tris-EDTA
1:1000
Citrate
SC
N-cadherin
(San Jose, CA) CD24
SN3b
Mouse
Thermo Fisher Scientific
NCAM
1B6
Mouse
(CD56) Vimentin
MA NU
(Waltham, MA)
Novocastra (Newcastle-uponTyne, UK)
V9
Mouse
Dako Cytomation (Glostrup, Denmark
3F12F3
Zeb1
CE
Rabbit
AC
SOX11
Rabbit
PT
HMGA2
Mouse
Novus Biologicals (Littleton,
ED
RAB25
Rabbit
CO)
GenWay Biotech (San Diego CA)
Atlas Antibodies (Stockholm, Sweden) / Sigma-Aldrich (St. Louis, MO) Atlas Antibodies / SigmaAldrich
ACCEPTED MANUSCRIPT 20
Table 3: Protein expression results in the studied effusions Protein
Staining
Staining extent (percentage of stained cells) 0%
1-5%
6-25%
26-75%
E-cadherin
Membrane
1
1
2
18
N-cadherin
Membrane
3
6
12
P-cadherin
Membrane
0
4
4
CD24
Membrane
5
39
NCAM
Membrane
35
21
Vimentin
Cytoplasm
13
RAB25
Cytoplasm
7
HMGA2
Nuclear
4
SOX11
Nucleus
Zeb1
Nuclear
76-100% 78
100
23
56
100
21
71
100
19
28
8
99 a
16
22
6
100
18
12
22
35
100
3
11
15
43
79 b
12
12
21
51
100
63
11
16
5
5
100
15
14
20
30
21
100
SC
MA NU
PT
ED
RI P
T
pattern
One effusion with missing value.
b
Twenty-one effusions not analyzed due to discontinued production of antibody used in
AC
CE
a
previous study [20].
Total
ACCEPTED MANUSCRIPT 21
Table 4. Significant associations between protein expression and chemotherapy response
Vimentin
1-5%
6-25%
Complete (n=50)
9
8
14
Other (n=50)
6
6
6
16
16
Complete (n=50)
10
12
6
11
11
Other (n=50)
3
6
6
11
24
Primary
6
5
3
12
21
7
13
9
10
14
chemoresistance (n=47) No primary
CE
PT
By Mann-Whitney U test.
AC
a
ED
chemoresistance (n=53)
26-75%
T
0%
14
RI P
Vimentin
Staining extent (percentage of stained cells)
SC
Zeb1
Response
MA NU
Protein
p-value a
76-100% 5
0.017
0.005
0.038
ACCEPTED MANUSCRIPT 22
Table 5. Univariate and multivariate survival analysis for 100 patients with prechemotherapy serous carcinoma effusions (n=100) a Univariate
95%
Multivariate
cases
p-value
Confidence
p-value
Interval
4
High
96
p=0.187
Low
72
High
28
43
High
57
4 RD c
p=0.05
33.109-
p=0.084
45.038
p=0.046
45.038
52
AC
3
CE
FIGO stage
33.109-
PT
Low
p=0.042
45.038
p=0.111
Vimentin
33.109-
ED
CD56 (NCAM)
p=0.623
45.038
MA NU
Low b
33.109-
SC
p=0.043
E-cadherin
T
No. of
RI P
Parameter
48 p=0.037
≤1 cm
48
>1 cm
38
35.768-
p=0.035
49.007
a
Only parameters with p<0.2 in univariate analysis shown
b
Low = staining of 0-25% tumor cells; High = staining in >25 of cells
c
Unavailable for 14 patients
ACCEPTED MANUSCRIPT 23
Figure legends Figure 1: EMT and stem cell marker expression in ovarian carcinoma (OC) effusions by
T
immunohistochemistry
RI P
Examples of immunostaining for the 10 studied proteins (X200 magnification all).
SC
Figure 2: The association between overall survival (OS) and E-cadherin, NCAM and vimentin protein expression in 100 serous carcinoma effusions
MA NU
A: Kaplan-Meier survival curve showing the association between E-cadherin expression and OS. Patients with effusions showing E-cadherin expression in ≤25% of tumor cells (n=4; solid line) had mean OS of 19 months compared to 40 months for patients with tumors displaying
ED
high (>25% of cells) expression (n=96, dashed line; p=0.043). B: Kaplan-Meier survival curve showing the association between NCAM expression and OS.
PT
Patients with effusions showing NCAM expression in >25% of tumor cells (n=28; solid line) had mean OS of 33 months compared to 41 months for patients with tumors displaying low
CE
(≤25% of cells) expression (n=72, dashed line; p=0.187).
AC
C: Kaplan-Meier survival curve showing the association between vimentin expression and OS. Patients with effusions showing vimentin expression in >25% of tumor cells (n=57; solid line) had mean OS of 35 months compared to 45 months for patients with tumors displaying low (≤25% of cells) expression (n=43, dashed line; p=0.111).
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
26
S0046-8177(14)00410-9 doi: 10.1016/j.humpath.2014.10.004 YHUPA 3431
To appear in:
Human Pathology
Received date: Revised date: Accepted date:
21 July 2014 29 September 2014 1 October 2014
Please cite this article as: Davidson Ben, Holth Arild, Hellesylt Ellen, Tan Tuan Zea, Huang Ruby Yun-Ju, Trop´e Claes, Nesland Jahn M., Thiery Jean Paul, The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions, Human Pathology (2014), doi: 10.1016/j.humpath.2014.10.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT 1
The clinical role of epithelial-mesenchymal transition and stem cell markers in advanced-stage ovarian serous carcinoma effusions
T
Running head: EMT markers in ovarian serous carcinoma
RI P
Ben Davidson, MD PhD1,2, Arild Holth, BSc1, Ellen Hellesylt, MSc1, Tuan Zea Tan, PhD3, Ruby Yun-Ju Huang, MD PhD3,4, Claes Tropé, MD PhD2,5, Jahn M. Nesland, MD PhD1,2,
SC
Jean Paul Thiery, PhD3,6,7
MA NU
Departments of Pathology1 and Gynecologic Oncology5, Oslo University Hospital, Norwegian Radium Hospital, N-0424 Oslo, Norway; 2University of Oslo, Faculty of Medicine, Institute of Clinical Medicine, N-0424 Oslo, Norway; Cancer Science Institute (CSI)3, Department of Obstetrics and Gynecology4 and Department of Biochemistry6, Yong
ED
Loo Lin School of Medicine, National University of Singapore, MD6 Centre for Translational Medicine, 14 Medical Drive, Singapore, 117599; Institute of Molecular and Cell Biology
CE
PT
(IMCB) A*STAR, 61 Biopolis Drive, Proteos, Singapore, 1386737
Financial acknowledgment: This work was supported by Inger and John Fredriksen
AC
Foundation for Ovarian Cancer Research.
Corresponding author Ben Davidson, MD PhD Department of Pathology Norwegian Radium Hospital Oslo University Hospital Montebello N-0310 Oslo Norway Tel: (47) 22934871 Fax: (47) 22508554 Email: [email protected]
ACCEPTED MANUSCRIPT 2
Summary We recently identified gene signatures which allow classification of ovarian carcinoma into 5
T
distinct clinically-relevant groups. In the present study, we investigated the clinical role of 10
RI P
protein products of the discriminating genes, with focus on epithelial-mesenchymal transition and stem cell markers. Expression of E-cadherin, N-cadherin, P-cadherin, Zeb1, HMGA2,
SC
Rab25, CD24, NCAM (CD56), Sox11 and vimentin was assessed in 100 advanced-stage (FIGO III-IV) serous ovarian carcinoma effusions using immunohistochemistry. Results were
MA NU
analyzed for association with clinicopathologic parameters, including chemotherapy response, and survival. All 10 proteins were frequently expressed in carcinoma cells. HMGA2 expression was related to older age (p=0.015). HMGA2 and NCAM expression was related to
ED
stage III disease (p=0.011 and p=0.023, respectively) and NCAM was overexpressed in peritoneal compared to pleural effusions (p=0.001). Vimentin and Zeb1 expression was
PT
significantly related to poor chemotherapy response at diagnosis (p=0.005 and p=0.017, respectively). The associations between NCAM and peritoneal localization and of vimentin
CE
and poor chemoresponse were retained after Bonferroni correction. NCAM expression was
AC
associated with a trend for shorter overall survival in univariate survival analysis (p=0.187), but emerged as independent prognosticator in Cox multivariate analysis (p=0.042). This study identifies Vimentin and Zeb1 as markers of poor chemoresponse in metastatic serous ovarian carcinoma effusions and suggests NCAM as potential prognostic marker in metastatic disease. The generally limited prognostic role of the studied markers emphasizes the difficulty in applying data obtained in studies of primary ovarian carcinomas to analyses of ovarian carcinoma effusions, reflecting the unique biology of the latter.
Keywords: ovarian carcinoma; epithelial-mesenchymal transition; effusion; chemoresistance; survival
ACCEPTED MANUSCRIPT 3
Introduction Ovarian cancer, consisting predominantly of ovarian carcinoma (OC), is the most fatal
T
gynecological malignancy. The combination of aggressive cytoreductive surgery and
RI P
neoadjuvant or adjuvant platinum/paclitaxel-based chemotherapy has prolonged survival in OC. However, chemotherapy resistance remains a major determinant of treatment failure and
SC
unfavorable clinical outcome, and the majority of patients consequently still die of their disease [1,2]. The development of malignant peritoneal, and less commonly, pleural effusions
MA NU
constitutes an almost universal finding in advanced-stage OC [3], and OC cells in effusions have been growingly perceived to represent a chemoresistant population with cancer stem cell characteristics [4,5], highlighting the need to better understand their genotypic and phenotypic
ED
characteristics.
PT
Epithelial-mesenchymal transition (EMT) is a process by which epithelial cells assume mesenchymal characteristics, facilitating migration through the extracellular matrix (ECM)
CE
and settlement in areas of new organ formation during embryogenesis. Wound-healing
AC
represents another form of physiological EMT, whereas pathological EMT occurs in tissue fibrosis and cancer [6-8]. EMT is induced by multiple signals, including growth factors, the Wnt signaling pathway, integrins, Notch transcription factors, prostaglandin E2 (PGE2), cyclooxygenase-2 (COX-2) and hormones [8]. During EMT, carcinoma cells lose their epithelial characteristics and acquire mesenchymal properties that promote ECM invasion and distant metastasis. This occurs through downregulation of E-cadherin, cytokeratins, ZO-1, claudins, occludin, laminin-1, entactin, MUC-1, and the microRNA 200 (miR-200) family, and acquisition of the transcription factors Snail1, Snail2, Twist, Zeb1 and Zeb2/SIP1, E47, KLF8, E2.2, Goosecoid, LEF-1 and FoxC2, as well as N-cadherin, vimentin, fibronectin, miR10b and miR21 [8-10].
ACCEPTED MANUSCRIPT 4
In OC, particularly serous carcinoma, the balance between the epithelial and mesenchymal phenotypes is complex, in part due to the inherent nature of serous carcinoma cells to express
T
mesenchymal markers, such as vimentin and N-cadherin. During OC progression, both EMT
RI P
and the reverse process, mesenchymal-epithelial transition (MET) occur, and OC cells in effusions upregulate E-cadherin and repress Snail1 and Snail2 expression compared to
SC
primary carcinomas [11-13].
MA NU
We recently generated a classification system for OC based on the gene expression patterns of 1,538 OC, including both public databases and our tumor material. OC could be classified into 5 sub-groups – Epi-A, Epi-B, Mes, Stem-A and Stem-B, of which Epi-B and Stem-A signatures were independent prognosticators of good and poor survival, respectively [14]. In
ED
the current study, we evaluated the expression of 10 protein products of genes identified by
PT
Significance Analysis of Microarray (SAM) [15] as differentiators between the above groups. Among these 10 genes, 4 were significantly down-regulated and 6 were significantly up-
CE
regulated in either Mes or Stem-A tumors (SAM q<10), the groups which had the worst
AC
outcome. Proteins were chosen based on antibody availability and performance from an initial group of 20 markers. In order to minimize the effect of clinical parameters on outcome, we analyzed a homogenous series of 100 pre-chemotherapy effusions tapped at diagnosis from 100 patients with FIGO stage III-IV serous OC. All patients received platinum-based chemotherapy. Immunohistochemistry (IHC) results were analyzed for association with clinicopathologic parameters, chemoresponse and survival.
ACCEPTED MANUSCRIPT 5
Materials and methods Patients and material
T
Specimens and clinical data were obtained from the Department of Gynecologic Oncology,
RI P
Norwegian Radium Hospital during the years 1998-2006. Informed consent was obtained
Committee for Medical Research Ethics in Norway.
SC
according to institutional and national guidelines. Study approval was given by the Regional
Fresh non-fixed serous carcinoma effusions (n=100; 72 peritoneal, 28 pleural; 80 OC, 16
MA NU
peritoneal carcinomas, 4 tubal carcinomas) were obtained pre-chemotherapy at diagnosis from 100 patients. Due to their closely-linked histogenesis and phenotype, all tumors are referred to as OC henceforth. Fifty-two patients had primary surgery, 36 had secondary debulking and 12
combined with paclitaxel.
ED
patients only received chemotherapy. All patients received platinum-based therapy, 82
PT
Effusions were submitted for routine diagnostic purposes and processed immediately after tapping. Cell blocks were prepared using the Thrombin clot method. Diagnoses were
CE
established using morphology and IHC. Clinicopathologic data are detailed in Table 1.
IHC
AC
Grading was according to the International Federation of Gynecology and Obstetrics (FIGO).
Formalin-fixed paraffin-embedded sections were analyzed for protein expression of Ecadherin, N-cadherin, P-cadherin, Zeb1, CD24, NCAM (CD56), Sox11 and vimentin using the Dako FLEX+ protocol (Dako, Glostrup, Denmark), with the exception of CD24, stained using the Dako Envision protocol. HMGA2 and Rab25 immunostaining data were retrieved from previous studies in which larger series, obtained both at diagnosis and disease recurrence, were analyzed [16,17]. Antibody and IHC data are provided in Table 2. Visualization was achieved using 3`3-diaminobenzidine tetrahydrochloride substrate (DAB)
ACCEPTED MANUSCRIPT 6
and hematoxylin counterstaining. Positive controls, chosen based on manufacturer datasheets, were satisfactory in all reactions.
T
Negative controls consisted of sections that underwent similar staining procedure with normal
nonrelevant rabbit IgG1 (Sigma Aldrich, St Louis MO).
RI P
IgG of the corresponding isotype as the primary antibody (for mouse antibodies) or with a
SC
Staining extent was scored as 0-4, corresponding to 0%, 1-5%, 6-25%, 26-75% and 76-100% stained carcinoma cells (minimum 100 cells required).
MA NU
Scoring was performed by two surgical pathologists (BD and JMN), both blinded to clinical data. Inter- and intra-observer concordance was >80%, with discrepant cases generally of one scoring level. These were discussed until agreement was reached.
ED
Statistical analysis
PT
Statistical analysis was performed applying the SPSS-PC package (Version 18.0, Chicago IL). Probability of <0.05 was considered significant. Analysis of the association between different
CE
biomarker immunostains were performed using the Chi-square test. Analysis of the
AC
association between biomarker expression and clinicopathologic parameters were undertaken using the Kruskal-Wallis H test (for grade, 3-tier analysis) or the Mann-Whitney U test (all other analyses). Clinicopathologic parameters were grouped as follows: Age: ≤60 vs. >60 years; grade: 1 vs. 2 vs. 3; effusion site: peritoneal vs. pleural; FIGO stage: III vs. IV; chemotherapy status: pre- vs. post-chemotherapy specimens; residual disease volume (RD): ≤1cm vs. >1cm; chemotherapy response: complete vs. partial response/stable disease/progression; primary (intrinsic) chemoresistance: progression-free survival (PFS) at ≤6 months vs. >6 months. Chemotherapy response was defined according to RECIST criteria. PFS and overall survival (OS) were calculated from the date of the last chemotherapy treatment/diagnosis to the date of recurrence/death or last follow-up, respectively. Univariate
ACCEPTED MANUSCRIPT 7
survival analyses of PFS and OS were executed using the Kaplan-Meier method and log-rank test. Expression categories in survival analyses were clustered as high vs. low based on IHC
T
score of 0-2 vs. 3-4, so as to allow for a sufficient number of cases to be included in each
RI P
category. Multivariate analysis was performed using the Cox proportional hazard model
AC
CE
PT
ED
MA NU
SC
(Enter function).
ACCEPTED MANUSCRIPT 8
Results EMT and stem cell-related markers are frequently expressed in OC effusions
T
IHC analysis demonstrated expression of all studied proteins except Sox11 in the majority of
RI P
OC effusions (Figure 1). E-cadherin, N-cadherin and P-cadherin expression was predominantly seen in >25% of cells, whereas the remaining markers had generally more
SC
limited expression (Table 3). The high number of cells expressing all 3 cadherins precluded any statistical analysis with respect to the association between these proteins. However, we
MA NU
studied the relationship between cadherin expression and staining for HMGA2, Zeb1 and vimentin. N-cadherin expression was significantly associated with that of vimentin (p=0.014) and a trend was observed with respect to Zeb1 (p=0.068). The remaining analyses were
ED
negative.
PT
Vimentin expression is related to primary (intrinsic) chemoresistance and poor chemotherapy response at diagnosis
CE
Expression of the 10 studied proteins was analyzed for association with clinicopathologic
AC
parameters and chemotherapy response at diagnosis. HMGA2 expression was significantly related to older age (p=0.015). HMGA2 and NCAM were significantly overexpressed in FIGO stage III compared to stage IV tumors (p=0.011 and p=0.023, respectively), and NCAM was overexpressed in peritoneal compared to pleural effusions (p=0.001). Expression of P-cadherin, NCAM and CD24 was higher in grade 1 compared to grade 2 and 3 tumors (p=0.005, p=0.044, p=0.041, respectively). However, grade 1 group consisted of only 5 tumors. Vimentin and Zeb1 expression was significantly related to poor (other than complete) response to chemotherapy at diagnosis (p=0.005 and p=0.017, respectively; Table 4-A). The observation for vimentin remained significant when complete or partial response was
ACCEPTED MANUSCRIPT 9
analyzed against stable disease or progression (p=0.008) and became still stronger when patients with complete response were analyzed against those with progression (p=0.001; data
T
not shown). Vimentin expression was further significantly associated to primary (intrinsic)
RI P
chemoresistance (p=0.038; Table 4-A).
SC
NCAM expression is significantly associated with poor OS
Survival data were available for all patients. The follow-up period ranged from 6-110 months
MA NU
(median=28 months). PFS ranged from 0-103 months (median=7 months), with 16 patients never achieving a progression-free period. At last follow-up, 3 patients were alive without disease, 8 were alive with disease and 89 were dead of disease. In univariate survival analysis, low E-cadherin expression was significantly related to poor OS (p=0.043), with trends for
ED
poor OS observed for NCAM (p=0.187) and vimentin (p=0.111; Figure 2). Notably, low
PT
expression of E-cadherin was observed in only 4 patients. Clinical parameters associated with OS were RD (p=0.037) and, marginally, FIGO stage (p=0.05).
CE
The parameters entered into the Cox multivariate analysis were E-cadherin, NCAM and
AC
vimentin protein expression, RD volume and FIGO stage. In this analysis, RD after primary surgery or secondary debulking (p=0.035), FIGO stage (p=0.046) and NCAM expression (p=0.042) were independent predictors of OS. Data for IHC and clinicopathologic parameters with p<0.2 in univariate analysis are summarized in Table 4-B.
ACCEPTED MANUSCRIPT 10
Discussion The present study validated the role of 10 proteins related to EMT and cancer stem cell
T
phenotype in sub-classifying OC with respect to chemoresponse and survival. There are
RI P
several notable differences between the present cohort and the reference publication [14], as well as most papers dealing with this issue. The present study focused exclusively on
SC
metastatic effusions rather than on primary carcinomas, analyzed only chemo-naïve specimens obtained at diagnosis, and included only patients diagnosed with FIGO stage III-IV
MA NU
serous OC and treated with platinum-based chemotherapy, the majority with combination platinum-paclitaxel. Such inclusion criteria render it difficult to obtain significant differences with respect to both established clinicopatologic parameters and biological markers. The
ED
relative homogeneity of this series is further reflected in the almost universally poor patient outcome, which renders prognostic stratification based on biological endpoints difficult.
PT
There are nevertheless several noteworthy findings in our data.
CE
The clinical role of cadherins in OC is still undecided, based on studies which have largely
AC
focused on primary OC [reviewed in 13]. An obvious difficulty in assessing this relationship in OC effusions lies in the ubiquitous expression of cadherins, mainly E-cadherin and Ncadherin, at this anatomic site. Our current data suggest that cadherin expression in serous OC effusion specimens is not a strong predictor of chemoresponse or prognosticator. It is nevertheless noteworthy that the 4 patients who had tumors with low E-cadherin expression had significantly shorter OS, suggesting that the inability to undergo MET at this anatomic site may be a characteristic of aggressive cancers. Further evaluation of this hypothesis requires analysis of a larger series.
ACCEPTED MANUSCRIPT 11
These results suggest that the majority of OC cells in effusions have reacquired epithelial-like features such as E-cadherin through partial MET. However, OC effusions exhibit a similar
T
distribution of molecular histotypes as primary OC [14]. Primary tumors exhibiting a
RI P
mesenchymal or stem-A molecular histotype have EMT scores which favor a mesenchymallike phenotype [manuscript in preparation]. Thus, it would be of interest to establish the score
SC
of effusions to determine whether E-cadherin re-expression affects the EMT score rather than the molecular subtype. Notably, co-expression of E-cadherin with N-cadherin has been
MA NU
hypothesized to affect E-cadherin function through a trans-dominant effect of N-cadherin promoting invasion and metastasis, albeit through mechanisms which remain partially uncovered [18]. The co-expression of the two cadherins could then also impact invasion of the mesothelial layer upon re-attachment and induce chemoresistance.
ED
A study carried out on OC cell lines has shown that those with the intermediate mesenchymal
PT
phenotype characterized by N-cadherin, vimentin and cytoketatin expression are more clonogenic and resistant to anoikis than some intermediate epithelial cell lines. However,
CE
intermediate epithelial cell lines were more migratory and invasive than epithelial cell lines.
AC
OC Effusions can thus be considered a mixed phenotype between intermediate epithelial and intermediated mesenchymal as defined by Huang et al. [19]. It is intriguing to consider that OC effusions retain very aggressive behavior albeit expressing E-cadherin. A similar situation has been encountered in luminal breast cancers which are much insensitive to anthracycline and platinum-based therapy than the more mesenchymal-like triple-negative breast cancers [20,21]. Using a similar collection of OC cell lines, a recent study showed that OC epithelial-like cell lines are more resistant to platinum-salt than their mesenchymal-like counterparts through the inhibition of apoptosis mediated by NF-κB [22].
ACCEPTED MANUSCRIPT 12
In contrast to cadherins, higher vimentin expression in this cohort was strong predictor of chemoresponse. Higher vimentin expression has been reported in OC cells resistant to both
T
platinum and paclitaxel compared to their chemosensitive counterparts [23-25]. However, to
RI P
the best of our knowledge, the present study is the first to demonstrate this association in clinical material. Of note, vimentin expression was both related to chemoresponse as
SC
measured after treatment at diagnosis and to intrinsic resistance, evidenced as progression within 6 months following completion of primary treatment. Significant association with poor
MA NU
chemoresponse at diagnosis was additionally observed for Zeb1, mediator of EMT via Ecadherin repression, though less strongly than for vimentin. Our present data with respect to vimentin differ from the above-discussed finding that OC lines with epithelial phenotype are more platinum-resistant [22], reflecting inherent differences between in vitro models and
PT
ED
clinical specimens studied unaltered by culturing conditions.
NCAM is a well-established marker of sex cord-stromal tumors and neuroendocrine OC [26].
CE
However, its expression in OC without neuroendocrine differentiation has been reported.
AC
NCAM was expressed in 33/95 (34.7%) of serous carcinomas compared to 5/40 (12.5%) serous borderline tumors and 1/23 (4.3%) of serous cystadenomas and its expression in carcinomas was significantly associated with advanced FIGO stage and peritoneal metastasis in one study [27]. Analysis of 252 primary OC and 206 solid metastases of different histotype showed overexpression of NCAM compared to normal ovaries and cystadenomas, with expression seen in 23.8% and 34.5% of primary carcinomas and metastases, respectively. NCAM expression was higher in tumors of higher histological grade. NCAM further increased tumor migration and invasion in vitro and metastasis in mouse model via interaction with FGFR [28]. In our cohort, NCAM was an independent prognostic marker of poor OS,
ACCEPTED MANUSCRIPT 13
this constituting the first report of such role in OC, further supporting its association with
T
aggressive disease in this cancer.
RI P
Sox11 expression was previously reported to be marker of longer survival in 2 studies of 76 and 154 primary OC of different histotype by one group [29,30]. Our data argue against such
SC
role in serous OC effusions. Similarly, the present study supports our previous report [17] regarding the absence of predictive or prognostic role for Rab25 at this anatomic site.
MA NU
CD24 was found to be present in exosomes isolated from OC ascites [31]. Its expression in this tumor was reported to be related to stem cell phenotype and to an invasive mesenchymal phenotype [32,33]. Cytoplasmic expression of CD24 in solid OC specimens is associated with poor survival [34]. In contrast, absence of CD24 (CD44+/CD24- phenotype) rather than
ED
expression was related to poor chemoresponse and shorter PFS in another study [35]. Our
PT
data do not support any of these roles for CD24 in serous OC effusions.
CE
In conclusion, analysis of 100 serous OC effusions for the expression of 10 EMT and stem
AC
cell markers revealed a potential relationship to more aggressive clinical course for NCAM, Zeb1 and vimentin. While several of the associations observed with parameters such as patient age, histological grade, effusion site and FIGO stage are lost following Bonferroni correction, the ones between NCAM and peritoneal localization, between P-cadherin and grade 1 disease, and between vimentin expression and poor chemoresponse is retained, of which the latter merits further attention. Similarly, further analysis of the role of NCAM in metastatic OC may identify a role as therapeutic target in this cancer. As previously shown by our group and evidenced in the present study, EMT along OC progression may not follow the same sequence as in other epithelial cancers.
ACCEPTED MANUSCRIPT 14
References 1. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012;62:10-
T
29.
RI P
2. Hennessy BT, Coleman RL, Markman M. Ovarian cancer. Lancet 2009;374;1371-82. 3. Davidson B. Ovarian and Primary Peritoneal Carcinoma. In: Davidson, Firat, Michael
SC
(Editors). Serous Effusions- Etiology, Diagnosis, Prognosis and Therapy. London: Springer; 2012.
MA NU
4. Latifi A, Luwor RB, Bilandzic M, et al. Isolation and characterization of tumor cells from the ascites of ovarian cancer patients: molecular phenotype of chemoresistant ovarian tumors. PLoS One 2012;7:e46858.
ED
5. Vathipadiekal V, Saxena D, Mok SC, Hauschka PV, Ozbun L, Birrer MJ. Identification of a potential ovarian cancer stem cell gene expression profile from advanced stage papillary
PT
serous ovarian cancer. PLoS One 2012;7:e29079. 6. Acloque H, Adams MS, Fishwick K, Bronner-Fraser M, Nieto MA. Epithelial-
CE
mesenchymal transitions: the importance of changing cell state in development and disease. J
AC
Clin Invest 2009;119:1438-49. 7. Thiery JP. Epithelial-mesenchymal transitions in tumour progression. Nat Rev Cancer 2002;2:442-54.
8. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in development and disease. Cell 2009;139:871-90. 9. Kalluri R, Weinberg RA. The basics of epithelial-mesenchymal transition. J Clin Invest 2009;119:1420-8. 10. Zeisberg M, Neilson EG. Biomarkers for epithelial-mesenchymal transitions. J Clin Invest 2009;119:1429-37.
ACCEPTED MANUSCRIPT 15
11. Davidson B, Berner A, Nesland JM, et al. E-cadherin and alpha-, beta-, and gammacatenin protein expression is up-regulated in ovarian carcinoma cells in serous effusions. J
T
Pathol 2000;192:460-9.
RI P
12. Elloul S, Silins I, Tropé CG, Benshushan A, Davidson B, Reich R. Expression of Ecadherin transcriptional regulators in ovarian carcinoma. Virchows Arch 2006;449:520-8.
SC
13. Davidson B, Trope' CG, Reich R. Epithelial-mesenchymal transition in ovarian carcinoma. Front Oncol 2012;2:33.
MA NU
14. Tan TZ, Miow QH, Huang RY, et al. Functional genomics identifies five distinct molecular subtypes with clinical relevance and pathways for growth control in epithelial ovarian cancer. EMBO Mol Med 2013;5:983-98.
15. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the
ED
ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116-21.
PT
16. Hetland TE, Holth A, Kærn J, Flørenes VA, Tropé CG, Davidson B. HMGA2 protein expression in ovarian serous carcinoma effusions, primary tumors and solid metastases.
CE
Virchows Arch 2012;460:505-13.
AC
17. Brusegard K, Stavnes HT, Nymoen DA, Flatmark K, Trope' CG, Davidson B. Rab25 is overexpressed in Müllerian serous carcinoma compared to malignant mesothelioma. Virchows Arch 2012;460:193-202. 18. Gheldof A, Berx G. Cadherins and epithelial-to-mesenchymal transition. Prog Mol Biol Trans Sci 2013;116:317-36. 19. Huang RY, Wong MK, Tan TZ, et al. An EMT spectrum defines an anoikis-resistant and spheroidogenic intermediate mesenchymal state that is sensitive to e-cadherin restoration by a src-kinase inhibitor, saracatinib (AZD0530). Cell Death Dis 2013;4:e915. 20. Carey LA, Dees EC, Sawyer L, et al. The triple negative paradox: primary tumor chemosensitivity of breast cancer subtypes. Clin Cancer Res 2007;13:2329-34.
ACCEPTED MANUSCRIPT 16
21. Silver DP, Richardson AL, Eklund AC, et al. Efficacy of neoadjuvant Cisplatin in triplenegative breast cancer. J Clin Oncol 2010;28:1145-53.
T
22. Miow QH, Tan TZ, Ye J, et al. Epithelial-mesenchymal status renders differential
RI P
responses to cisplatin in ovarian cancer. Oncogene 2014 May 26. doi: 10.1038/onc.2014.136. [Epub ahead of print]
SC
23. Kajiyama H, Shibata K, Terauchi M, et al. Chemoresistance to paclitaxel induces epithelial-mesenchymal transition and enhances metastatic potential for epithelial ovarian
MA NU
carcinoma cells. Int J Oncol 2007;31:277-83.
24. Haslehurst AM, Koti M, Dharsee M, et al. EMT transcription factors snail and slug directly contribute to cisplatin resistance in ovarian cancer. BMC Cancer 2012;12:91. 25. Rosanò L, Cianfrocca R, Spinella F, et al. Acquisition of chemoresistance and EMT
ED
phenotype is linked with activation of the endothelin A receptor pathway in ovarian
PT
carcinoma cells. Clin Cancer Res 2011;17:2350-60. 26. McCluggage WG, McKenna M, McBride HA. CD56 is a sensitive and diagnostically
AC
2007;26:322-7.
CE
useful immunohistochemical marker of ovarian sex cord-stromal tumors. Int J Gynecol Pathol
27. Cho EY, Choi Y, Chae SW, Sohn JH, Ahn GH. Immunohistochemical study of the expression of adhesion molecules in ovarian serous neoplasms. Pathol Int 2006;56:62-70. 28. Zecchini S, Bombardelli L, Decio A, et al. The adhesion molecule NCAM promotes ovarian cancer progression via FGFR signalling. EMBO Mol Med 2011;3:480-94. 29. Brennan DJ, Ek S, Doyle E, et al. The transcription factor Sox11 is a prognostic factor for improved recurrence-free survival in epithelial ovarian cancer. Eur J Cancer 2009;45:1510-7. 30. Sernbo S, Gustavsson E, Brennan DJ, et al. The tumour suppressor SOX11 is associated with improved survival among high grade epithelial ovarian cancers and is regulated by reversible promoter methylation. BMC Cancer 2011;11:405.
ACCEPTED MANUSCRIPT 17
31. Runz S, Keller S, Rupp C, et al. Malignant ascites-derived exosomes of ovarian carcinoma patients contain CD24 and EpCAM. Gynecol Oncol 2007;107:563-71.
T
32. Gao MQ, Choi YP, Kang S, Youn JH, Cho NH. CD24(+) cells from hierarchically
RI P
organized ovarian cancer are enriched in cancer stem cells. Oncogene 2010;29:2672-80. 33. Kang KS, Choi YP, Gao MQ, et al. CD24+ ovary cancer cells exhibit an invasive
SC
mesenchymal phenotype. Biochem Biophys Res Commun 2013;432:333-8. 34. Kristiansen G, Denkert C, Schlüns K, Dahl E, Pilarsky C, Hauptmann S. CD24 is
Am J Pathol 2002;161:1215-21.
MA NU
expressed in ovarian cancer and is a new independent prognostic marker of patient survival.
35. Meng E, Long B, Sullivan P, et al. CD44+/CD24- ovarian cancer cells demonstrate cancer
AC
CE
PT
ED
stem cell properties and correlate to survival. Clin Exp Metastasis 2012;29:939-48.
ACCEPTED MANUSCRIPT 18
Table 1: Clinicopathologic data of the study cohort (100 patients) Age (Mean; Range)
63; 39-86
52
IV
48
SC
Grade a
RI P
III
T
FIGO stage
5
II
MA NU
I
26
III
51
NA b
18
≤1 cm
PT
>1 cm
ED
Residual disease
CE
NA
48 38 14
Chemotherapy response at diagnosis 50
Partial response
20
Stable disease
7
Progressive disease
17
NA
6
AC
Complete
a
Corresponding to 5 low-grade and 77 high-grade serous carcinomas based on the 2014
WHO classification b
NA= non available
ACCEPTED MANUSCRIPT 19
Table 2: Antibodies and staining conditions Antibody
Clone
Source
Manufacturer
Dilution
Antigen retrieval
Mouse
Zymed Laboratories (San Francisco, CA)
1:3000
T
HECD-1
RI P
E-cadherin
Tris-EDTA
3B9
Mouse
Invitrogen (Carlsbad, CA)
1:300
Tris-EDTA
P-cadherin
56
Mouse
BD Transduction Laboratories
1:500
Tris-EDTA
1:200
Citrate
1:50
Tris-EDTA
1:1600
Tris-EDTA
1:500
Tris-EDTA
1:500
Tris-EDTA
1:500
Tris-EDTA
1:1000
Citrate
SC
N-cadherin
(San Jose, CA) CD24
SN3b
Mouse
Thermo Fisher Scientific
NCAM
1B6
Mouse
(CD56) Vimentin
MA NU
(Waltham, MA)
Novocastra (Newcastle-uponTyne, UK)
V9
Mouse
Dako Cytomation (Glostrup, Denmark
3F12F3
Zeb1
CE
Rabbit
AC
SOX11
Rabbit
PT
HMGA2
Mouse
Novus Biologicals (Littleton,
ED
RAB25
Rabbit
CO)
GenWay Biotech (San Diego CA)
Atlas Antibodies (Stockholm, Sweden) / Sigma-Aldrich (St. Louis, MO) Atlas Antibodies / SigmaAldrich
ACCEPTED MANUSCRIPT 20
Table 3: Protein expression results in the studied effusions Protein
Staining
Staining extent (percentage of stained cells) 0%
1-5%
6-25%
26-75%
E-cadherin
Membrane
1
1
2
18
N-cadherin
Membrane
3
6
12
P-cadherin
Membrane
0
4
4
CD24
Membrane
5
39
NCAM
Membrane
35
21
Vimentin
Cytoplasm
13
RAB25
Cytoplasm
7
HMGA2
Nuclear
4
SOX11
Nucleus
Zeb1
Nuclear
76-100% 78
100
23
56
100
21
71
100
19
28
8
99 a
16
22
6
100
18
12
22
35
100
3
11
15
43
79 b
12
12
21
51
100
63
11
16
5
5
100
15
14
20
30
21
100
SC
MA NU
PT
ED
RI P
T
pattern
One effusion with missing value.
b
Twenty-one effusions not analyzed due to discontinued production of antibody used in
AC
CE
a
previous study [20].
Total
ACCEPTED MANUSCRIPT 21
Table 4. Significant associations between protein expression and chemotherapy response
Vimentin
1-5%
6-25%
Complete (n=50)
9
8
14
Other (n=50)
6
6
6
16
16
Complete (n=50)
10
12
6
11
11
Other (n=50)
3
6
6
11
24
Primary
6
5
3
12
21
7
13
9
10
14
chemoresistance (n=47) No primary
CE
PT
By Mann-Whitney U test.
AC
a
ED
chemoresistance (n=53)
26-75%
T
0%
14
RI P
Vimentin
Staining extent (percentage of stained cells)
SC
Zeb1
Response
MA NU
Protein
p-value a
76-100% 5
0.017
0.005
0.038
ACCEPTED MANUSCRIPT 22
Table 5. Univariate and multivariate survival analysis for 100 patients with prechemotherapy serous carcinoma effusions (n=100) a Univariate
95%
Multivariate
cases
p-value
Confidence
p-value
Interval
4
High
96
p=0.187
Low
72
High
28
43
High
57
4 RD c
p=0.05
33.109-
p=0.084
45.038
p=0.046
45.038
52
AC
3
CE
FIGO stage
33.109-
PT
Low
p=0.042
45.038
p=0.111
Vimentin
33.109-
ED
CD56 (NCAM)
p=0.623
45.038
MA NU
Low b
33.109-
SC
p=0.043
E-cadherin
T
No. of
RI P
Parameter
48 p=0.037
≤1 cm
48
>1 cm
38
35.768-
p=0.035
49.007
a
Only parameters with p<0.2 in univariate analysis shown
b
Low = staining of 0-25% tumor cells; High = staining in >25 of cells
c
Unavailable for 14 patients
ACCEPTED MANUSCRIPT 23
Figure legends Figure 1: EMT and stem cell marker expression in ovarian carcinoma (OC) effusions by
T
immunohistochemistry
RI P
Examples of immunostaining for the 10 studied proteins (X200 magnification all).
SC
Figure 2: The association between overall survival (OS) and E-cadherin, NCAM and vimentin protein expression in 100 serous carcinoma effusions
MA NU
A: Kaplan-Meier survival curve showing the association between E-cadherin expression and OS. Patients with effusions showing E-cadherin expression in ≤25% of tumor cells (n=4; solid line) had mean OS of 19 months compared to 40 months for patients with tumors displaying
ED
high (>25% of cells) expression (n=96, dashed line; p=0.043). B: Kaplan-Meier survival curve showing the association between NCAM expression and OS.
PT
Patients with effusions showing NCAM expression in >25% of tumor cells (n=28; solid line) had mean OS of 33 months compared to 41 months for patients with tumors displaying low
CE
(≤25% of cells) expression (n=72, dashed line; p=0.187).
AC
C: Kaplan-Meier survival curve showing the association between vimentin expression and OS. Patients with effusions showing vimentin expression in >25% of tumor cells (n=57; solid line) had mean OS of 35 months compared to 45 months for patients with tumors displaying low (≤25% of cells) expression (n=43, dashed line; p=0.111).
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
24
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
25
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
26