Potential predictors of chemotherapy response in ovarian cancer—How do we define chemosensitivity?

Result . Quantifying VEGF proved to be a valuable independent prognostic indicator in progression-free survival (PFS) ( p < 0.05) and overall survival (OS) ( p < 0.0001). VEGF correlated with response to chemotherapy at the 6-month interval ( r = 0.446, p < 0.001) but failed to correlate at the 1-year interval. Increased staining with CD31 was associated with decreased PFS ( p < 0.01) and OS ( p < 0.01) in univariate but not multivariate analysis. MDR1 failed to act as a prognostic marker or as a predictor of response to chemotherapy. Conclusion . VEGF correlates with response to chemotherapy at the 6-month but not the 12-month interval. What should our criteria be for determining sensitivity to chemotherapy? CD31, VEGF and MDR1 do play a role in some ovarian malignancies but other factors are likely to be involved and perhaps molecular profiling will determine which factors will be important for determining the response to chemotherapy. Keywords VEGF CD31 MDR1 Ovarian cancer Prognostic indicator Chemotherapy response Introduction Ovarian cancer is one of the most common cancers in the female and the leading cause of death from gynaecological malignancy in the western world [1] . About 190,000 cases of ovarian cancer occur worldwide each year [2] . The majority of ovarian cancers present in advanced stages (III or IV) and are treated by surgery and systemic chemotherapy. Despite an initial 70–80% response rate, current therapy is frequently followed by recurrence, which is often resistant to chemotherapy, as demonstrated by the < 20% long-term survivors. Growth and metastatic dissemination of solid tumours requires vascular support for nutritive supply and access [ 3 , 4 ]. For a tumour to get larger than 1 mm 3 , a process of angiogenesis or neovascularisation must occur [ 5 , 6 ]. CD31 has been largely used for immunohistochemical analysis on formalin-fixed, paraffin-embedded tissue sections to assess tumour angiogenesis (microvessel density (MVD)) [7] . Various growth factors have been shown to stimulate angiogenesis in physiological and pathological conditions, including neoplastic disease. Among these, vascular endothelial growth factor (VEGF) has been shown to play a major role in the proliferation and migration of endothelial cells, providing nourishment to the growing tumours and allowing the tumour cells to establish continuity with the host vasculature [8] . The biological and clinical significance of angiogenic activity in ovarian cancer is controversial. Some studies have shown a prognostic role for VEGF in ovarian cancer but others disagree [ 9–12 ]. Conflicting evidence also surrounds the correlation of MVD to the clinical outcome of the patient [10,13] . The present study attempted to establish the significance of angiogenesis in ovarian cancer. There is very little published data on the relationship between the tumour level of VEGF or CD31 and the efficacy of response to chemotherapy. Any data published remains contradictory and clearly depends on the definition of response to chemotherapy. Should response be determined by WHO Standards [14] or should a longer follow-up period be used such as that used by the Gynecologic Oncology Group (GOG) [15] . The influence of VEGF expression and MVD in ovarian cancer still remains controversial [10,16–19] . P-glycoproteins (P-gps) are a homologous family of integral membrane proteins and function as an energy-dependent drug efflux pump that reduces intracellular drug accumulation, thereby conferring resistance to many different drugs. P-glycoprotein is encoded by a small family of closely related genes, and of the two genes in humans, only the multidrug resistant (MDR1) gene (also known as PGY1) causes the MDR phenotype [20] . Whereas increased levels of P-gp appear to play a role in the drug resistance phenotype of certain human cancers, evidence for the involvement of this protein in ovarian cancer is conflicting [21,22] . Methods Patients Ovarian specimens were obtained from patients undergoing oophorectomy for ovarian cancer, benign ovarian disease and prophylactic oophorectomy in St. James's Hospital Dublin. Ethical approval was received. Tumours were staged according to the International Federation of Gynecology and Obstetrics (FIGO). Details of the specimens are shown in Table 1 . The patients' charts were monitored retrospectively every 3–6 months, during and after chemotherapy treatment to determine the in vivo patient response . Patients received a platinum-based or a paclitaxel/platinum-based chemotherapy regimen. Response to chemotherapy was assessed after 6 months and again after 1 year. Sensitive was defined as having a good response to chemotherapy and remaining disease free. Resistant was defined as a recurrence of the disease or progressive disease. Immunohistochemistry Specimens were fixed in 10% neutral buffered formalin. Four-micrometer sections were cut, dewaxed using xylene and hydrated. The DAKO EnVision™ + System, Peroxidase kit (3,3′-diaminobenzidine (DAB), mouse) (Dako, Denmark) was used for immunohistochemical analysis. The monoclonal antibodies used were CD31, (1:50) (Clone JC/70A, Oncogene, Cambridge, MA), VEGF (Ab-3) (1:20) (Clone 14-124, Oncogene) and P170-Glycoprotein (1:20), multidrug resistance related, clone JSB-1 (Monosan, The Netherlands). Antigen retrieval was by pressure cooker. Positive control slides for CD31 were from an appendix section and from a gastric tumour for VEGF and MDR. Negative controls were incubated with the antibody diluent only. Evaluation of staining All sections were quantified by the same technique using unified analysis of a representative section. Sections were scored from 0 to 2 based on the degree of staining: 0- no staining, 1- low degree of staining (< 50% of section), 2- high degree of staining (> 50% of section stained). Various methods exist for measurement of MVD and we used a semiquantitative grading method in preference to the vascular hot spot technique [23,24] . The obvious advantage of MVD grading is its time efficiency but also we used the same technique for grading VEGF and MDR sections. Several studies have reported a positive correlation between quantitative and semiquantitative MVD scores [24,25] . CD31, VEGF and MDR were correlated with patient age, tumour stage, tumour grade, progression-free survival and overall survival. Statistical analysis SPSS for Windows, Rel. 11.0.1. 2001. Chicago: SPSS Inc. was used for statistical analysis. The criterion of statistical significance applied was p < 0.05. For statistical evaluation, Pearson correlation, one-way ANOVA and chi-square were used for univariate analysis. Kaplan Meier survival curves were used, log-rank was used for univariate analysis and a Cox proportional hazards regression model was used for multivariate evaluation of survival rates. Results Examples of immunostaining are shown in Fig. 1 (ovarian adenocarcinoma stained with CD31), Fig. 2 (ovarian adenocarcinoma stained with VEGF) and Fig. 3 (ovarian adenocarcinoma stained with MDR). CD31 ( p < 0.01) and MDR ( p < 0.05) staining were significantly increased in the malignant group in comparison to the borderline or benign/normal group. The high staining VEGF specimens consisted of just malignant samples but this relationship was not significant. The degree of CD31, VEGF and MDR staining did not correlate with tumour grade, stage or patient age. Progression-free survival (PFS) and overall survival (OS) were measured for the stained specimens. The average follow-up period was 21 months. Only malignant specimens were included in the survival analysis. Kaplan Meier curves were constructed for PFS and patients that were alive and free from disease were censored ( Figs. 4, 6 and 8 ). OS entries were censored for patients still alive at time of follow-up ( Figs. 5, 7 and 9 ). Eighty-two percent (18/22) of the high staining CD31 group had tumour progression as compared to 62% (16/26) in the low staining group and 48% (14/29) in the no staining group ( p < 0.01) ( Fig. 4 ). Patient samples that had a CD31 score of 2 had a median PFS of 6 months. Median PFS time in the group that displayed a low degree of staining was 14.5  months and 51 months in the no staining group ( p < 0.01). At the time of follow-up 69% (13/19) of the high CD31 staining group had died, 39% (9/23) of low staining and 29% (8/28) of the no staining group ( p < 0.01) ( Fig. 5 ). For VEGF the entire high staining group presented with tumour progression compared to 60% (12/20) of the low staining and 55% (26/47) of the no staining group ( p < 0.0001) ( Fig. 6 ). Patient samples that had a score of 0 had a median PFS of 27 months. Median PFS time in the group that displayed a low degree of staining was 9 months and in the group with a high degree of staining the median was 5 months ( p < 0.0001). Eighty-nine percent (8/9) of the high staining group had died at time of follow-up compared to 50% (9/18) in the low staining group and 30% (13/43) in the no staining group ( p < 0.0001) ( Fig. 7 ). Corrected median OS for the group of specimens with no staining was not calculable but the 75th percentile was 25 months. The group that displayed a low degree of staining had a median OS of 23 months and for the high staining group the median OS was 7 months. VEGF remained significant in multivariate analysis for PFS ( p < 0.05) and OS ( p < 0.001). A significant correlation was detected between CD31 and VEGF ( r = 0.508, p < 0.0001). The survival curves for MDR were not significant, although a trend was observed ( Figs. 8 and 9 ). Eighty-four percent (16/19) of the high staining group had disease progression compared to 57% (17/30) in the low staining and 56% (15/28) in the no staining group. Patient samples that had a score of 2 had a median PFS of 11 months. Those with a low degree of staining had a median PFS of 14 months and in the group with no staining the median was 15 months. Median OS for the specimens that displayed no staining was 23 months. The 75th percentiles were calculated for the low and no staining groups with values of 9 and 13 months, respectively, as compared to 8 months in the high staining group. Sixty-seven percent (10/15) of the high staining group had died of disease at time of follow-up, 39% (11/28) of the low staining group and 35% (9/27) of the no staining group. MDR was detected in previously untreated specimens. Sixty of the cases that were stained for CD31, VEGF and MDR were followed up in vivo for their response to chemotherapy. After a follow-up period of 12 months no significant correlations were observed between the staining and chemotherapy response. However, if the follow-up period was decreased to 6 months, there was a significant correlation between VEGF and chemotherapy response ( r = 0.446, p < 0.001). Conclusion The role of VEGF and MVD in ovarian cancer remains conflicting but our results suggest VEGF acts as a prognostic marker in ovarian cancer and its association with CD31 suggests it plays a role in angiogenesis. Significantly more MVD staining was observed in the malignant group than in the borderline or benign group. The malignant group in the present study consisted mainly of serous papillary adenocarcinomas so it was not possible to distinguish between malignant tumour types. VEGF and MVD did not correlate with tumour stage, grade or patient age which is in agreement with some studies [16,26–28] . Other studies have found a correlation with stage [9,29] and grade [19,30,31] and this is probably due to the larger cohort of early stage tumours. A statistically significant reduction in progression-free and overall survival was found with increasing degree of CD31 staining in the present study, which is in agreement with Darai et al. [32] . Also Hollingsworth et al. [26] concluded that increased intratumour microvessel density is associated with decreased overall and disease-free survival in ovarian cancer but only disease-free survival remained significant in multivariate analysis. The present results are similar to Obermair et al. [33] who found that microvessel density was a significant prognostic indicator in univariate analysis but this failed to attain statistical significance in multivariate analysis. Raspollini et al. [10] found that MVD remained significant for cause-specific survival and brief disease relapse in multivariate analysis. Other studies, however, have failed to repeat these observations [16,34] . In the present study, patients whose disease recurred or progressed or who died from disease displayed significantly higher VEGF staining. This result is in agreement with other studies reporting a correlation between induced neoangiogenesis and tumour progression and/or metastases [10,11,12,16] . Paley et al. [11] suggested that VEGF expression by neoplastic cells was of independent prognostic value in patients with early stage ovarian cancers and patients with VEGF-rich tumours had a shorter disease-free survival. Yamamoto et al. [12] found that ovarian cancer patients with strong VEGF immunoreactivity showed poorer survival rates than those with weak or no immunostaining but the prognostic significance was related to its correlation with FIGO stage and it was not an independent prognostic indicator. The present results demonstrate that VEGF is an independent prognostic indicator in disease-free survival and overall survival. A limited number of studies have investigated the relationship between MVD and VEGF but have shown conflicting results. Our results are in agreement with others [16,35] who found a significant association between MVD and VEGF expression in that the MVD of VEGF-rich tumours was significantly higher than that of VEGF-poor tumours suggesting VEGF has a crucial role in angiogenesis of ovarian cancer. Raspollini et al. [10] observed a trend between MVD and VEGF but other studies [17,19,27] observed no relationship between VEGF and MVD but the antibody used for detection of MVD was FVIII-related antigen, which has been shown to be less sensitive than CD31 [30,36–38] . Increased vascularity may suggest improved tumour oxygenation and drug delivery and thereby improved response to chemotherapy. However, this remains contradictory in many cases [10,39,40] . Studies analysing the relationship between angiogenesis and chemotherapy response in ovarian cancer are limited. In the present study, no correlation existed between angiogenesis and in vivo response after a 1-year follow-up; however, a large percentage of the resistant tumours were in the high staining groups. Gasparini et al. [39] examined the relationship between the degree of vascularisation and response to platinum-based combined induction chemotherapy in patients with FIGO stages III–IV ovarian disease. There was a significant association with the degree of staining and lack of response. Due to the small number of patients with a pathological complete response, they were unable to examine the association of the probability of complete response with degree of vascularisation. Raspollini et al . [10] used the WHO criteria for determining response to chemotherapy and found a significant relationship with MVD and VEGF. The association between degree of staining and chemotherapy response clearly depends on the follow-up time. What do we define as chemosensitive 6 months or 12 months progression-free survival? We have used the Gynecologic Oncology Group guidelines of 6 months after completing chemotherapy, approximately 12 months post surgery. Many of the previous studies looking at angiogenesis markers have used the WHO method of determining response which has a very short follow-up time. If a patient responds to chemotherapy, then a reasonable disease-free survival should be expected, 1 year minimum. However, in this study the markers failed to provide an adequate measure of chemotherapy response after a 1-year period but VEGF did correlate with response after 6 months. Clearly a longer follow-up period is needed when assessing the suitability of a marker as an indicator of chemotherapy response. A general consensus on the definition of chemosensitive and chemoresistant is needed when assessing markers of chemotherapy response. Previous studies suggested MDR1 was associated with acquired resistance [22] ; however, this was not the case in this study as MDR1 expression was detected in 24% of previously untreated ovarian tumours, a finding reported by other investigators [41–43] . MDR1 failed to act as a prognostic marker or as a predictor of response to chemotherapy that is in agreement with Van der Zee et al. [44] although a trend was observed. The mechanisms responsible for intrinsic and acquired drug resistance in ovarian cancer are not clearly understood. P-gp, which is encoded by the MDR gene, has been extensively studied in cell lines and human tumour samples [45] . Conflicting evidence on the frequency and prognostic significance of P-gp expression in ovarian carcinomas can be found in various studies [21,22,46,47] . Kavallaris et al. [47] found that patients with MDR1 positive tumours had a significantly shorter event-free survival time than patients with MDR1 negative tumours suggesting that even low-level expression of the MDR1 gene may offer a selective growth advantage to these tumours. A recent study demonstrated that P-glycoprotein was an independent predictor of both overall and progression-free survival [43] . Raspollini et al. [10] found MDR1 to be associated with chemotherapy response evaluated using the WHO criteria. Mechanisms responsible for drug resistance in ovarian cancer appear to be variable and further studies are required to elucidate the role MDR associated genes play in this carcinoma. It is likely that CD31, VEGF and MDR play a role in the resistance to chemotherapy and could act as a prognostic factor in some ovarian cancers but as yet it is not possible to determine which tumours. It is postulated that the role of these genes in resistance may be governed by hypoxia. The hypoxia inducible factor-1 (HIF-1) gene pathway regulates the tumour response to hypoxia and VEGF and MDR1 are known to be regulated by HIF-1 [48] . The role of hypoxia and the correlation with these genes will be further investigated and perhaps molecular profiling of tumours will identify other potential markers of chemotherapy response. A consensus on evaluation of chemotherapy response is needed for the accurate determination of biomarkers of chemotherapy response. Acknowledgments The authors wish to thank the Royal City of Dublin Hospital Trust Fund for their financial contribution to this project. We are grateful to Dr. Noreen Gleeson, the gynaecological oncology team and the theatre staff in St. James's Hospital Dublin. References [1] W.J. Hoskins Prospective on ovarian cancer J. Cell. Biochem. Suppl. 23 1995 189 199 [2] D.M. Parkin P. Pisani J. Ferlay Estimates of the worldwide incidence of twenty-five major cancers in 1990 Int. J. Cancer 80 1999 827 841 [3] J. Folkman The role of angiogenesis in tumor growth Semin. Cancer Biol. 3 2 1992 65 71 [4] J. Folkman K. Watson D. Ingber D. Hanahan Induction of angiogenesis during the transition from hyperplasia to neoplasia Nature 339 6219 1989 58 61 [5] N. Paweletz M. Knierim Tumor-related angiogenesis Crit. Rev. Oncol. Hematol. 9 3 1989 197 242 [6] D. Ribatti A. Vacca L. Roncali F. Dammacco Angiogenesis under normal and pathological conditions Haematologica 76 4 1991 311 320 [7] B.R. DeYoung M.R. Wick J.F. Fitzgibbon K.E. Sirgi P.E. Swanson CD31: an immunospecific marker for endothelial differentiation in human neoplasms Appl. Immunohistochem. 1 1993 97 100 [8] N. Ferrara The role of vascular endothelial growth factor in pathological angiogenesis Breast Cancer Res. Treat. 36 1995 127 137 [9] H. Brustmann P. Riss S. Naude The relevance of angiogenesis in benign and malignant epithelial tumors of the ovary: a quantitative histologic study Gynecol. Oncol. 67 1 1997 20 26 [10] M.R. Raspollini G. Amunni A. Villanucci G. Baroni V. Boddi G.L. Taddei Prognostic significance of microvessel density and vascular endothelial growth factor expression in advanced ovarian serous carcinoma Int. J. Gynecol. Cancer 14 5 2004 815 823 [11] P.J. Paley K.A. Staskus K. Gebhard Vascular endothelial growth factor expression in early stage ovarian carcinoma Cancer 80 1 1997 98 106 [12] S. Yamamoto I. Konishi H. Kuroda Expression of vascular endothelial growth factor (VEGF) in epithelial ovarian neoplasms: correlation with clinicopathology and patient survival, and analysis of serum VEGF levels Br. J. Cancer 76 9 1997 1221 1227 [13] N. Weidner Intratumor microvessel density as a prognostic factor in cancer Am. J. Pathol. 147 1 1995 9 19 [14] World Health Organization Offset Publication No. 48. WHO handbook for reporting results of cancer treatment . Geneva (Switzerland): WHO;1979. [15] J.T. Thigpen H. Ball S.J. Hummel R.J. Barrett Phase II trial of paclitaxel in patients with progressive ovarian carcinoma after platinum-based chemotherapy: a Gynecologic Oncology Group study J. Clin. Oncol. 12 9 1994 1748 1753 [16] Y. Nakanishi J. Kodama M. Yoshinouchi The expression of vascular endothelial growth factor and transforming growth factor-beta associates with angiogenesis in epithelial ovarian cancer Int. J. Gynecol. Pathol. 16 3 1997 256 262 [17] E.M. Hartenbach T.A. Olson J.J. Goswitz D. Mohanraj L.B. Twiggs L.F. Carson Vascular endothelial growth factor (VEGF) expression and survival in human epithelial ovarian carcinomas Cancer Lett. 121 2 1997 169 175 [18] M. Orre P.A. Rogers VEGF, VEGFR-1, VEGFR-2, microvessel density and endothelial cell proliferation in tumors of the ovary Int. J. Cancer 84 2 1999 101 108 [19] G.H. Shen M. Ghazizadeh O. Kawanami Prognostic significance of vascular endothelial growth factor expression in human ovarian carcinoma Br. J. Cancer 83 2 2000 196 203 [20] K. Ueda C. Cardarelli M.M. Gottesman I. Pastan Expression of a full-length cDNA for the human MDR1 gene confers resistance to colchicines, doxorubicin and vinblastine Proc. Natl. Acad. Sci. U. S. A. 84 1987 3004 3008 [21] D.R. Bell J.H. Gerlach N. Kartner R.N. Buick V. Ling Detection of P-glycoprotein in ovarian cancer: a molecular marker associated with multidrug resistance J. Clin. Oncol. 3 3 1985 311 315 [22] J. Bourhis L.J. Goldstein G. Riou I. Pastan M.M. Gottesman J. Benard Expression of a human multidrug resistance gene in ovarian carcinomas Cancer Res. 49 18 1989 5062 5065 [23] L.A. Liotta J. Kleinerman G.M. Saidel Quantitative relationships of intravascular tumor cells, tumor vessels, and pulmonary metastases following tumor implantation Cancer Res. 34 5 1974 997 1004 [24] N. Weidner J.P. Semple W.R. Welch J. Folkman Tumor angiogenesis and metastasis-correlation in invasive breast carcinoma N. Engl. J. Med. 324 1 1991 1 8 [25] N. Weidner J. Folkman F. Pozza Tumor angiogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma J. Natl. Cancer Inst. 84 24 1992 1875 1887 [26] H.C. Hollingsworth E.C. Kohn S.M. Steinberg M.L. Rothenberg M.J. Merino Tumor angiogenesis in advanced stage ovarian carcinoma Am. J. Pathol. 147 1995 33 41 [27] M. Sonmezer M. Gungor A. Ensari F. Ortac Prognostic significance of tumor angiogenesis in epithelial ovarian cancer: in association with transforming growth factor beta and vascular endothelial growth factor Int. J. Gynecol. Cancer 14 1 2004 82 88 [28] J. Fujimoto H. Sakaguchi R. Hirose S. Ichigo T. Tamaya Biologic implications of the expression of vascular endothelial growth factor subtypes in ovarian carcinoma Cancer 83 12 1998 2528 2533 [29] H. Brustmann S. Naude Vascular endothelial growth factor expression in serous ovarian carcinoma: relationship with high mitotic activity and high FIGO stage Gynecol. Oncol. 84 1 2002 47 52 [30] P.J. Stone M.J. Goodheart S.L. Rose B.J. Smith B.R. DeYoung R.E. Buller The influence of microvessel density on ovarian carcinogenesis Gynecol. Oncol. 90 3 2003 566 571 [31] G.G. Garzetti A. Ciavattini G. Lucarini Expression of vascular endothelial growth factor related to 72-kilodalton metalloproteinase immunostaining in patients with serous ovarian tumors Cancer 85 10 1999 2219 2225 [32] E. Darai A.F. Bringuier F. Walker-Combrouze CD31 expression in benign borderline, and malignant epithelial ovarian tumors: an immunohistochemical and serological analysis Gynecol. Oncol. 71 1998 122 127 [33] A. Obermair R. Wasicky A. Kaider Prognostic significance of tumor angiogenesis in epithelial ovarian cancer Cancer Lett. 138 1–2 1999 175 182 [34] A. Gadducci P. Viacava S. Cosio Intratumoral microvessel density, response to chemotherapy and clinical outcome of patients with advanced ovarian carcinoma Anticancer Res. 23 1B 2003 549 556 [35] R. Ali-Fehmi R.T. Morris S. Bandyopadhyay Expression of cyclooxygenase-2 in advanced stage ovarian serous carcinoma: correlation with tumor cell proliferation, apoptosis, angiogenesis, and survival Am. J. Obstet. Gynecol. 192 3 2005 819 825 [36] M. Orre M. Lofti-Miri P. Maners P.A.W. Rogers Increased microvessel density in mucinous compared with malignant serous and benign tumours of the ovary Br. J. Cancer 77 12 1998 2204 2209 [37] A. Giatromanoali M.I. Koukourakis D. Theodossiou Comparative evaluation of angiogenesis assessment with anti-factor-VIII and anti-CD31 immunostaining in non-small cell lung cancer Clin. Cancer Res. 3 1997 2485 2492 [38] E.R. Horak R. Leek N. Klenk Angiogenesis, assessed by platelet/endothelial cell adhesion molecule antibodies, as indicator of node metastases and survival in breast cancer Lancet 340 1992 1120 1124 [39] G. Gasparini E. Bonoldi G. Viale Prognostic and predictive value of tumour angiogenesis in ovarian carcinomas Int. J. Cancer 69 1996 205 211 [40] P.J. van Diest J.P. Zevering L.C. Zevering J.P. Baak Prognostic value of microvessel quantitation in cisplatin treated FIGO 3 and 4 ovarian cancer patients Pathol. Res. Pract. 191 1 1995 25 30 [41] T.A. Holzmayer S. Hilsenbeck D.D. Von Hoff I.B. Roninson Clinical correlates of MDR1 (P-glycoprotein) gene expression in ovarian and small-cell lung carcinomas J. Natl. Cancer Inst. 84 19 1992 1486 1491 [42] T. Schondorf A. Scharl C.M. Kurbacher Amplification of the mdr1-gene is uncommon in recurrent ovarian carcinomas Cancer Lett. 146 2 1999 195 199 [43] M.M. Baekelandt R. Holm J.M. Nesland C.G. Tropé G.B. Kristensn P-glycoprotein expression is a marker for chemotherapy resistance and prognosis in advanced ovarian cancer Anticancer Res. 20 2000 1061 1068 [44] A. Van der Zee H. Hollema A. Suurmeijer Value of P-glycoprotein, glutathione S-transferase pi, c-erbB-2, and p53 as prognostic factors in ovarian carcinomas J. Clin. Oncol. 13 1 1995 70 78 [45] I.B. Roninson Molecular and Cellular Biology of Multidrug Resistance in Tumour Cells 1991 Plenum New York [46] H.M. Coley Drug resistance studies using fresh human ovarian carcinoma and soft tissue sarcoma samples Keio J. Med. 46 3 1997 142 147 [47] M. Kavallaris J.A. Leary J.A. Barrett M.L. Friedlander MDR1 and multidrug resistance-associated protein (MRP) gene expression in epithelial ovarian tumors Cancer Lett. 102 1–2 1996 7 16 [48] G.L. Semenza Targeting HIF-1 for cancer therapy Nat. Rev., Cancer 3 10 2003 721 732