Cytoreductive surgery for recurrent ovarian cancer: A meta-analysis

Available online at www.sciencedirect.com

Gynecologic Oncology 112 (2009) 265–274www.elsevier.com/locate/ygyno

Review

Cytoreductive surgery for recurrent ovarian cancer: A meta-analysis

Robert E. Bristow a,⁎, Isha Puri a, Dennis S. Chi b

a The Kelly Gynecologic Oncology Service, Departments of Gynecology and Obstetrics and Oncology, The Sidney Kimmel Comprehensive Cancer Center, The JohnsHopkins Medical Institutions, 600 North Wolfe Street, Phipps #281, Baltimore, Maryland 21287, USA

b Gynecology Service, Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, New York, USA

Received 29 July 2008Available online 19 October 2008

Abstract

Objective. To determine the relative effect of multiple prognostic variables on overall post-recurrence survival time among cohorts of patientswith recurrent ovarian cancer undergoing cytoreductive surgery.

Methods. Forty cohorts of patients with recurrent ovarian cancer (2019 patients) meeting study inclusion criteria were identified from theMEDLINE database (1983–2007). Simple and multiple linear regression analyses, with weighted correlation calculations, were used to assess theeffect on median post-recurrence survival time of the following variables: year of publication, age, disease-free interval, localized disease, tumorgrade and histology, the proportion of patients undergoing complete cytoreductive surgery, requirement for bowel resection, and the sequence ofcytoreductive surgery and salvage chemotherapy.

Results. The mean weighted median disease-free interval prior to cytoreductive surgery was 20.2 months, and the mean weighted medianoverall post-recurrence survival time was 30.3 months. The weighted mean proportion of patients in each cohort undergoing completecytoreductive surgery was 52.2%. Median survival improved with increasing year of publication (p=0.009); however, the only statisticallysignificant clinical variable independently associated with post-recurrence survival time was the proportion of patients undergoing completecytoreductive surgery (p=0.019). After controlling for all other factors, each 10% increase in the proportion of patients undergoing completecytoreductive surgery was associated with a 3.0 month increase in median cohort survival time.

Conclusions. Among patients undergoing operative intervention for recurrent ovarian cancer, the proportion of patients undergoing completecytoreductive surgery is independently associated with overall post-recurrence survival time. For this select group of patients, the surgicalobjective should be resection of all macroscopic disease.© 2008 Elsevier Inc. All rights reserved.

Keywords: Ovarian cancer; Recurrence; Cytoreductive surgery

Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266

Study selection and data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266Statistical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267Study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267Simple linear regression analyses of predictor variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268Multiple linear regression analysis of predictor variables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270

⁎ Corresponding author. Fax: +1 410 614 8718.E-mail address: [email protected] (R.E. Bristow).

0090-8258/$ - see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.ygyno.2008.08.033

266 R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272

Introduction

The American Cancer Society has estimated that 21,650women in the United States will be diagnosed with ovariancancer annually, and 15,520 women will die of this disease [1].The concept of cytoreductive surgery for ovarian cancer hasevolved since Meigs, in 1934, first proposed that as much tumoras possible should be removed to enhance the effects of post-operative irradiation [2]. Forty years after Meigs' initialproposition, Griffiths published the landmark study that firstclearly delineated the inverse relationship between post-operative residual tumor size and patient survival [3]. Morecontemporary studies published by Hoskins et al., writing forthe Gynecologic Oncology Group, demonstrated two importantprinciples with respect to residual disease after primary surgeryfor advanced-stage ovarian cancer. First, there is a thresholdeffect, or a maximal diameter of residual disease above whicheven extensive efforts at cytoreduction will not impact survival[4]. Secondly, below this threshold there is also a continuumeffect, such that the smaller the residuum, the better the survivaloutcome, with patients left with no gross residual disease havingthe most favorable prognosis [5].

Although the basic treatment paradigm of a maximumcytoreductive surgical effort prior to initiating platinum andtaxane-based chemotherapy is well established, the majority ofpatients with advanced-stage epithelial ovarian cancer willultimately experience tumor recurrence [6,7]. For this reason, thetherapeutic value of repeating the initial surgical treatment plan(cytoreduction) has been widely debated. Since the publicationby Berek et al. in 1983, which first introduced the term“secondary cytoreduction”, the clinical scenarios, indicationsfor, and anticipated outcomes of repeat tumor reductiveoperations for recurrent ovarian cancer have beenmore preciselydefined [8,9]. By most accounts, repeat or secondary cyto-reductive surgery for recurrent ovarian cancer is defined as anoperative procedure performed at some time remote (disease-free interval of more than 6 to 12months) from the completion ofprimary therapy with the intended purpose of tumor reduction.Even within this narrowly defined clinical scenario, the potentialutility of surgical cytoreduction remains controversial. Specifi-cally, the survival impact of successful tumor reduction has beendifficult to quantify in relation to other relevant clinical andbiological prognostic characteristics. The objective of thecurrent study was to aggregate the published literature oncytoreductive surgery for recurrent ovarian cancer to determinethe relative effect of multiple prognostic variables on overallpost-recurrence survival time among cohorts of surgical patients.The main research hypothesis concerning residual disease wasthat median overall post-recurrence survival time would bepositively correlated with the proportion of patients in eachcohort undergoing successful secondary cytoreductive surgery.

Methods

Study selection and data extraction

Using the headings and keywords “recurrent ovariancarcinoma,” “recurrent ovarian cancer,” “secondary cytoreduc-tive surgery,” and “secondary surgical cytoreduction,” aMEDLINE search for English-language articles publishedbetween January 1, 1983 and July 31, 2007 was conducted.Research published only in abstract format was not included.Publications were selected for initial review if the studycontained at least one cohort of patients with recurrent epithelialovarian cancer undergoing attempted secondary cytoreductivesurgery. The bibliography of each selected article was reviewedfor other potentially relevant citations. Published reports ofpatients thought to be clinically free of disease and undergoingcytoreduction at the time of second-look re-assessment surgerywere not included. Similarly, studies describing patients under-going interval cytoreductive surgery during the prescribedprimary chemotherapy treatment program after an initialsuboptimal or deferred attempt at primary surgery were alsonot included.

The formal published versions of all eligible studies werereviewed for content and screened according to the followingminimal inclusion requirements: internal study eligibilitycriteria specifying a disease-free interval of at least 6 monthsprior to attempted cytoreductive surgery for recurrent ovariancancer, stated median cohort post-recurrence survival time, thesurgical objective or criteria for optimal residual disease, andthe proportion of patients in each cohort undergoing an optimalcytoreductive surgery for disease recurrence. The followingadditional variables were extracted from the published reportswhen available but were not specifically required for inclusionin the final dataset: accrual interval (in years), median cohortpatient age, median disease-free interval preceding cytoreduc-tive surgery, proportion of patients in each cohort undergoingcomplete cytoreductive surgery (no gross residual disease),proportion of patients with localized recurrence (according tothe author's definition), proportion of patients with ascites at thetime of surgery, proportion of patients with serous histology,proportion of patients with poorly differentiated or grade 3tumors, proportion of patients submitted to bowel resectionduring cytoreductive surgery, proportion of patients undergoingcytoreductive surgery prior to salvage chemotherapy, medianestimated blood loss, median operative time, proportion ofpatients experiencing significant post-operative morbidity, andpost-operative mortality rate. When possible, additional datapoints not reported in the formal manuscript were obtainedthrough personal contact with the study author. Post-cytor-eductive surgery adjuvant treatment was also recorded ifavailable.

267R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

Statistical methods

Simple linear regression models were generated to examinethe effect of the following variables on median cohort post-recurrence survival time: year of publication of the study,accrual interval (in years), median age, median disease-freeinterval, proportion of patients undergoing optimal cytoreduc-tive surgery, proportion of patients undergoing completecytoreductive surgery, proportion of patients with localizedrecurrent disease, proportion of patients with ascites, proportionof patients with serous histology, proportion of patients withgrade 3 (or poorly differentiated) tumors, proportion of patientssubmitted to bowel resection, and the proportion of patientsundergoing cytoreductive surgery prior to receiving salvagechemotherapy. The regressions were weighted by the number ofpatients in each cohort. This tended to give more influence tothe larger studies, because they necessarily contained moreprecise estimates of the key measures. Examination of the datarevealed no nonlinear effects for the variables under study whenexpressed in this fashion.

To overcome the limitations of incomplete data, a multipleimputation procedure was employed to create values formissing observations among the predictor variables using theMarkov Chain Monte Carlo (MCMC) method [10,11]. Not allpredictor variables had missing values; however, all predictorvariables were used in the imputation process to yield improvedestimates for those with missing values. Following the multipleimputation procedure, multiple linear regression analysis wasused to derive the independent effects of the aforementionedvariables on median post-recurrence survival time, simulta-neously controlling for all other measured variables. Multi-collinearity did not emerge as a serious problem based on thevariance inflation factor (VIFb4). For predictor variables with astrong positive or negative correlation at the bivariate level, thevariable more strongly associated with median overall post-recurrence survival was chosen for incorporation into themultiple linear regression model.

All results are quoted as two-sided p values and 95%confidence intervals (CI); pb0.05 was considered statisticallysignificant. Analyses were carried out using the SAS statisticalsoftware package (Version 9.1.3, Research Triangle Institute,Research Triangle Park, Cary, NC).

Results

Study characteristics

The initial MEDLINE search yielded 876 articles. The full-length published reports of 101 of these studies were formallyreviewed. Ultimately, 40 studies, encompassing 2019 patients,were identified as meeting the minimum study inclusion criteria(Table 1) [8,12–50]. Thirteen studies were published between1983 and 2000, while 27 studies (67.5%) were publishedbetween 2001 and 2007. Twelve of the 40 reports utilizedprospective non-randomized data collection methodology; therewere 27 retrospective analyses and 1 retrospective case-controlstudy. The number of subjects in each study cohort ranged from

12 to 267, with a median of 34 subjects and an average of 50.5subjects in each cohort. The definition of an optimal secondarycytoreductive surgical resection varied across studies asfollows: b2.5 cm in 1 study (24 patients or 1.2%), b2 cm in7 studies (265 patients or 13.1%), b1.5 cm in 1 study (32patients or 1.6%), ≤1 cm in 17 studies (972 patients or 48.1%),≤0.5 cm in 7 studies (381 patients or 18.9%), ≤0.25 cm in 1study (30 patients or 1.5%), and no gross residual disease(complete cytoreduction) in 6 studies (315 patients or 15.6%).The mean weighted proportion of patients in each cohortundergoing an optimal surgical resection was 70.3%, with arange from 22.2% to 100%. The weighted mean proportion ofpatients in each cohort undergoing complete cytoreduction was52.2%, with a range from 9.4% to 100%.

Analysis of operative parameters at the time of cytoreductivesurgery for recurrent ovarian cancer revealed that the medianestimated blood loss ranged from 300 cm3 to 1000 cm3, with aweighted mean of 587 cm3. The median operative time rangedfrom 130 min to 588 min, with a weighted mean of 233 min.The weighted mean proportion of patients in each cohortundergoing bowel resection at the time of cytoreductive surgerywas 40.5%, with a range of 0 to 80.0%. Thirty-six studies (1613patients) reported the rate of significant peri-operative morbid-ity, which had a weighted mean of 19.2% (range 0 to 88.8%).The peri-operative mortality rate was reported in 37 studies,which ranged from 0 to 5.5% and had a weighted mean of 1.2%.For all study cohorts, the mean weighted median survival timefollowing cytoreductive surgery for ovarian cancer recurrencewas 30.3 months, with a range of 10.0 months to 62.0 months.

Thirty studies (1599 patients) offered at least limited data onpost-operative therapy. There were 1221 patients that weredocumented to have received systemic or intraperitonealchemotherapy. Platinum, either alone or in combination wasadministered intravenously or intraperitoneally to 838 patients.Among these patients, platinum agents were combined withpaclitaxel, docetaxel, gemcitabine, doxorubicin, liposomaldoxorubicin, cyclophosphamide, or etoposide. One hundredand forty-four patients were specifically documented as havingreceived platinum and paclitaxel combination therapy post-operatively. Post-operative single agent non-platinum therapyincluded the following drugs: melphalan, cyclophosphamide,paclitaxel, docetaxel, hexamethylmelamine, 5-fluorouracil,mitomycin C, epidoxorubicin, gemcitabine, ifosfamide, topo-tecan, doxorubicin, liposomal doxorubicin, vindesine, vincris-tine, and etoposide. For 165 patients, there was documentationthat some post-operative chemotherapy was administered, butno detail was provided regarding the specific agents or whetherplatinum was included in the treatment program. Seven patientsunderwent high-dose chemotherapy or bone marrow transplantwith unspecified agents. Sixty-two patients underwenthyperthermic intraperitoneal chemotherapy at the time ofsecondary cytoreductive surgery (cisplatin, n=45; paclitaxel,n=14; mitomycin C, n=3). Twenty-four patients receivedintra-operative radiation therapy during secondary surgery, and49 patients received some form of radiation therapy post-operatively. Six patients were treated with hormonal therapypost-operatively. The large variety and inconsistent reporting of

Table 1Study characteristics

Author Reference Yearpublication

Accrualinterval(months)

N(total)

Age(median)

Median overallsurvival(months)

Disease-freeinterval(months)

Optimalcriteria(cm)

Optimalcytoreduction(%)

Completecytoreduction(%)

Berek et al. a [8] 1983 60 32 54.5 10 6 b1.5 37.5 NAMorris et al. a [12] 1989 96 30 50 16.3 36 b2.0 56.7 30.0Janicke et al. a [13] 1992 NA 30 53 18 16 b2.0 86.7 46.7Segna et al. a [14] 1993 144 100 55 16.6 NA b2.0 61.0 NAEisenkop et al. b [15] 1995 54 36 60.6 43 22 ≤1.0 91.7 83.3Vaccarello et al. a [16] 1995 120 57 57 18 20 b0.5 36.8 17.5Lichtenegger et al. a [17] 1998 73 81 58 23 NA No gross 22.2 22.2Landoni et al. a [18] 1998 156 38 51 c (mean) 29 22 No gross 100.0 100.0Cormio et al. a [19] 1999 84 21 58 29 25 b2.0 90.5 71.4Gadducci et al. a [20] 2000 112 30 58.5 21 17.5 b2.0 83.3 56.7Zang et al. a [21] 2000 144 60 50 11 12 ≤1.0 38.3 NAChen et al. a [22] 2000 142 22 56.5 41 26 b1.0 86.4 63.6Eisenkop et al. b [23] 2000 108 106 60.5 35.9 16.8 No gross 82.1 82.1Tay et al. a [24] 2002 99 46 50.3 22.5 26 ≤1.0 71.7 41.3Bristow et al. a [25] 2002 96 21 46 56.2 15.7 ≤1.0 71.4 61.9⁎⁎

Yoon et al. b [26] 2003 163 24 47.5 62 36.5 ≤1.0 100.0 87.5Zang et al. a [27] 2003 144 60 52 17 NA ≤1.0 38.3 NAMeredith et al. a [28] 2003 288 26 62 26.3 23.4 ≤1.0 80.8 69.2Look et al. a [29] 2003 161 24 54 45.8 NA b2.5 87.5 20.8Loizzi et al. d [30] 2003 105 31 57 38 34 b2.0 90.3 NALeitao et al. a [31] 2004 156 26 55.5 33.4 13.4 ≤0.5 73.1 53.8Zang et al. b [32] 2004 48 117 53 22 15.4 ≤1.0 61.5 9.4Zanon et al. b [33] 2004 69 30 60 28.1 NA ≤0.25 76.7 NAUzan et al. a [34] 2004 221 12 51 50 21 No gross 100.0 100.0Gronlund et al. b [35] 2005 73 38 59 c 27.4 e 16.3 No gross 42.1 42.1Gungor et al. a [36] 2005 120 44 54.3 16 27.1 b1.0 77.3 NAYap et al. a [37] 2005 106 22 57.4 26 48.2 b0.5 100.0 NAOnda et al. b [38] 2005 165 44 52 32 18.5 b1.0 84.1 59.1Ayhan et al. a [39] 2006 144 64 50.6 18.6 15.5 ≤1.0 82.8 43.8Matsumoto et al. a [40] 2006 192 23 55.7 41.7 22.5 b2.0 43.4 30.4Manci et al. a [41] 2006 120 24 54 56 26 ≤0.5 100.0 66.7Chi et al. a [42] 2006 180 153 56.5 41.7 NA ≤0.5 51.6 40.5Harter et al. b [43] 2006 48 267 60 29.2 NA ≤1.0 75.7 49.8Rufian et al. b [44] 2006 96 14 55 57 NA ≤1.0 85.0 52.0Helm et al. a [45] 2007 32 18 64 31 24.6 ≤0.5 94.4 61.1Salani et al. a [46] 2007 91 55 57.7 48 26 ≤1.0 89.1 74.5Santillan et al. a [47] 2007 92 25 59 37 16 ≤1.0 100.0 96.0Benedetti Panici et al. b [48] 2007 126 40 51 60 c 14 No gross 72.5 72.5Benedetti Panici et al. b [49] 2007 60 47 52 49 15 ≤1.0 87.2 78.7Cotte et al. b [50] 2007 180 81 54.3 28.4 NA ≤0.5 80.2 55.6

NA = data not available.a Retrospective review.b Prospective, non-randomized.c Median not yet reached.d Retrospective case-control.e Personal communication.

268 R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

post-operative treatment programs administered during thestudy period, coupled with a lack of correlative survivaloutcomes, made inclusion of post-operative therapy as apredictor variable in the statistical analysis impractical.

Simple linear regression analyses of predictor variables

For the predictor variables of year of publication, mediancohort age, and proportion of patients undergoing optimalcytoreductive surgery, there were no missing values. For theremaining predictor variables, the proportion of missing valuesranged from 3.7% to 40.1% of subjects. The results of simplelinear regression analyses are shown in Table 2.

The mean weighted median disease-free interval prior tocytoreductive surgery for recurrent ovarian cancer was20.2 months. Cohorts with a median disease-free intervalb12 months prior to cytoreductive surgery had a meanweighted median survival time of 28.5 months, compared to31.7 months for cohorts with a median disease-free intervalbetween 12 and 24 months and 34.1 months for cohorts witha disease-free interval N24 months. However, these differ-ences were not statistically significant (p=0.31). There was astatistically significant association between year of publica-tion and median cohort survival time. Each one yearincrement in the year of publication was associated with a1.12 month increase in median survival time (p=0.002, 95%

Localizeddisease (%)

Carcinomatosis(%)

Ascites(%)

Seroushistology(%)

Grade 3tumors (%)

Bowelresection(%)

Surgery prior tochemotherapy (%)

Median estimatedblood loss (ml)

Medianoperative time(min)

Morbidity(%)

Mortality(%)

NA 12.5 50.0 78.1 50.0 53.1 NA 835 NA 34.4 0.0NA NA NA 46.7 NA 70.0 NA 892 261 36.7 0.0NA NA NA NA 36.7 59.4 NA 270 23.3 0.0NA NA NA NA NA 38.0 NA 500 13.0 1.013.9 33.3 88.9 86.1 47.2 50.0 735 186 30.1 2.8NA NA NA 63.2 59.6 26.3 66.7 NA NA 23.7 0.0NA NA NA NA NA 48.0 NA NA NA 25.9 6.063.0 76.3 65.7 21.0 0.0 NA NA 23.7 5.071.4 28.6 0.0 71.4 47.6 38.1 100.0 NA NA 42.9 0.036.7 NA NA 70.0 53.3 36.7 100.0 NA 240 20.0 3.3NA NA 25.0 48.3 58.3 26.7 81.7 450 NA 5.0 0.036.4 NA 18.2 63.6 90.9 50.0 NA 838 222 28.6 0.017.9 NA 17.0 NA 90.6 46.2 60.4 680 192 32.1 1.954.3 45.7 NA 52.2 58.7 34.8 100.0 748 133 8.7 4.3NA NA 34.6 100.0 0.0 34.6 100.0 400 NA 23.8 0.025.0 54.2 45.8 20.8 NA 750 20.8 0.0NA NA NA 52.7 NA 36.7 100.0 470 NA 8.3 1.723.1 61.5 92.3 42.3 NA 1000 246 23.1 0.0NA 100.0 NA 71.4 NA NA NA 1109 528 10.7 0.0NA NA NA 64.5 61.3 NA NA NA NA NA NA42.3 NA NA 46.2 57.7 61.5 0.0 NA NA 23.1 0.028.2 71.8 19.7 52.1 38.5 88.0 60.7 300 150 7.7 0.0NA NA NA NA NA NA NA NA 410 16.7 3.375.0 16.7 0.0 66.7 NA 0.0 100.0 NA NA 25.0 0.031.6 NA NA 39.5 NA 10.5 71.1 NA NA 21.1 0.063.6 0.0 NA NA 43.2 100.0 NA NA 36.4 0.0100.0 0.0 0.0 45.5 63.6 18.2 NA 490 390 22.7 0.036.4 NA 6.8 79.5 NA 25.0 52.3 NA NA NA 0.020.2 57.8 32.8 62.5 54.7 26.6 71.9 NA NA NA NA43.4 NA NA 65.2 NA 30.4 NA NA NA 4.3 0.037.5 45.8 58.3 70.8 29.2 NA 610 155 12.5 0.026.8 28.8 19.0 45.8 59.5 52.3 NA NA NA 3.9 0.026.6 50.0 13.5 NA NA NA 68.9 NA NA NA NANA NA 57.0 NA 57.1 NA NA 300 14.3 0.00.0 NA NA 83.3 NA 38.9 NA 1260 588 88.8 5.576.4 NA 0.0 81.8 56.4 16.4⁎⁎ 80⁎⁎ 200 NA 25.5 1.876.0 4.0 4.0 76.0 100.0 0.0 84.0 100 NA 0.0 0.035.0 NA 0.0 72.5 70.0 12.5 100.0 420 130 32.9 0.059.6 NA 0.0 68.1 68.0 21.3 80.9 500 210 19.1 0.014.8 66.7 NA NA 60.5 48.6 NA NA 258 13.6 2.5

269R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

CI=0.43 months to 1.82 months) (Fig. 1). The only clinicalparameters significantly associated with survival were themeasures of residual disease. Each 10% increase in theproportion of patients left with optimal residual disease wasassociated with a 2.69 month increase in median survivaltime (95% CI 0.90 months to 4.49 months, p=0.004).Similarly, each 10% increase in the proportion of patientsundergoing complete surgical resection was associated withan increase in median cohort survival time of 2.84 months(95% CI 1.29 months to 4.38 months, p=0.0008). Theproportion of patients in each cohort undergoing completecytoreductive surgery plotted versus median cohort survivaltime is shown in Fig. 2. The regression line was computed

weighted by the number of observations in each study, andthe effects of other variables were ignored. Simple linearregression analysis revealed no statistically significantassociation between year of publication and the proportionof patients in each cohort undergoing complete cytoreductivesurgery for ovarian cancer recurrence (p=0.62).

Multiple linear regression analysis of predictor variables

At the bivariate level, there was a strong positive correlationbetween the proportion of patients in each cohort undergoingoptimal cytoreductive surgery and the proportion of patientsundergoing complete cytoreductive surgery (r=0.81). Based on

Table 2Simple linear regression analyses of predictor variables versus median cohort survival time

Predictor variable Weighted mean Range Increment Change inmedian survival

95% confidenceinterval

p-value Missing data a

(%)

Year of publication NA 1983 to 2007 (+) 1 year 1.12 0.43 to 1.82 0.002 0.0Accrual interval 111.7 months 32 to 228 months (+) 1 month 0.03 (−)0.05 to 0.10 0.51 3.7Median age 55.6 years 46 to 64 years (+) 1 year 0.19 (−)0.85 to 1.24 0.71 0.0Median disease-free interval 20.2 months 6 to 48.2 months (+) 1 year 0.35 (−)0.33 to 1.03 0.31 40.1Localized disease 35.1% 0 to 100% (+) 10% 0.57 (−)1.63 to 2.78 0.609 31.5Ascites 15.6% 0 to 50% (+) 10% (−)4.33 (−)9.13 to 0.47 0.089 34.9Serous histology 58.2% 39.5 to 100% (+) 10% 3.02 (−)0.52 to 6.58 0.09 38.1Grade 3 tumor 61.3% 0 to 100% (+) 10% 1.54 (−)1.51 to 4.59 0.32 36.7Surgery before chemotherapy 71.5% 0 to 100% (+) 10% (−)0.38 (−)23.67 to 22.91 0.974 37.9Bowel resection 40.5% 0 to 80% (+) 10% (−)1.77 (−)2.90 to 0.51 0.129 17.4Optimal cytoreduction 70.3% 22.2 to 100% (+) 10% 2.69 0.90 to 4.49 0.004 0.0Complete cytoreduction 52.2% 9.4 to 100% (+) 10% 2.84 1.29 to 4.38 0.0008 15.8

a Proportion of total subjects.

270 R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

a greater degree of reproducibility associated with the surgicaloutcome of no gross residual disease (i.e. not variably defined),the proportion of patients undergoing complete cytoreductionwas the only surgical outcome measure incorporated into themultiple linear regression model.

Multiple linear regression analysis was performed on theimputed data set to derive the independent effect on mediancohort survival time of the predictor variables. After controllingfor all other factors, the only two statistically significantpredictor variables were the proportion of patients undergoingcomplete cytoreduction and year of publication (Table 3). Each10% increase in the proportion of patients undergoing completecytoreductive surgery was associated with a 3.00 monthincrease in median cohort survival time (95% CI 0.50 monthsto 5.53 months, p=0.019). According to the regression modelestimate, the median cohort survival time would range from18.0 months to 48.0 months as the proportion of patientsundergoing complete cytoreductive surgery increased from 0 to100%.

Fig. 1. Simple linear regression analysis: median cohort survival time plottedagainst year of publication for patient cohorts undergoing cytoreductive surgeryfor recurrent ovarian cancer. Circle size is proportional to the number of subjectsin each study.

Discussion

For women with advanced-stage ovarian cancer, thetreatment paradigm of maximal primary cytoreductive surgeryfollowed by platinum and paclitaxel-based chemotherapy iswell established, and a complete clinical response can beexpected in over 50% of patients [51,52]. For patients withprogressive disease on front-line therapy or recurring shortlyafter completing initial chemotherapy, treatment options arelimited and the prognosis is poor. On the other hand, ovariancancer recurrence after an extended treatment-free interval (atleast 6 to 12 months), or so-called platinum-sensitive disease,can often be treated successfully with an expectation ofprolonged survival, although cure is uncommon. For this selectgroup of patients, repeating the initial surgical treatment plan(tumor cytoreduction) has been advocated as a means ofaugmenting durability of response to subsequent adjuvantchemotherapy. Establishing the proper role of cytoreductivesurgery in the management of patients with recurrent ovarian

Fig. 2. Simple linear regression analysis: median cohort survival time plottedagainst the proportion of patients in each cohort undergoing completecytoreductive surgery for recurrent ovarian cancer. Circle size is proportionalto the number of subjects in each study.

Table 3Multiple linear regression analysis of selected predictor variables versus mediancohort survival time using imputed data set

Predictor variable(incremental increase)

Change inmedian survivaltime (months)

95% confidenceinterval

p-value

Proportion completecytoreduction (+10%)

(+) 3.00 0.50 to 5.53 0.02

Year of publication(+1 year)

(+) 1.00 0.26 to 1.77 0.01

Study accrual interval(+1 month)

(+) 0.0007 (−) 0.09 to 0.087 0.98

Median cohort age(+1 year)

(−) 0.55 (−) 1.84 to 0.725 0.39

Proportion surgery beforechemotherapy (+10%)

(+) 0.43 (−) 1.64 to 2.51 0.66

Proportion seroushistology (+10%)

(+) 1.09 (−) 4.29 to 6.47 0.68

Proportion grade 3tumor (+10%)

(+) 0.43 (−) 2.92 to 3.79 0.79

Proportion localizeddisease (+10%)

(−) 1.16 (−) 3.33 to 1.003 0.28

Proportion bowelresection (+10%)

(+) 0.83 (−) 1.97 to 3.64 0.56

271R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

cancer has been hampered by the fact that most studiesaddressing this topic are retrospective, span a large period oftime, and are not randomized [53]. In an attempt to overcomesome of these limitations, the technique of meta-analysis maybe used to extract clinically useful observations from thecollective literature. The primary objective of this study,therefore, was to determine the relative effect of multipleprognostic variables on overall post-recurrence survival timeamong cohorts of patients with recurrent ovarian cancerundergoing cytoreductive surgery.

Of all of the predictor variables analyzed in the currentdataset, the only clinical parameter that was independently andstatistically significantly associated with post-recurrence survi-val time was the proportion of patients undergoing completecytoreductive surgery. Bolstering the validity of this observa-tion, a similar positive association between survival outcomeand an increasing proportion of patients undergoing maximalsurgical resection has been demonstrated in the settings ofprimary and interval cytoreductive surgery for advanced-stageovarian cancer [7,54]. The current study is the first toquantitatively define the expected survival benefit fromsuccessful cytoreductive surgery for recurrent ovarian canceramong a large collection of surgical patient cohorts. On simplelinear regression analysis, the magnitude of the incrementalimprovement in survival with each 10% increase in theproportion of patients left with optimal residual disease orundergoing complete cytoreduction was 2.69 months and2.84 months, respectively. Since these two variables wereassessing two different points along the same clinical outcomecontinuum, complete cytoreduction was selected for inclusionin the final multiple linear regression model based on theinherently greater degree of specificity in its definition of thesurgical result. After controlling for the effects of all otherstudied variables, each 10% increase in complete secondary

cytoreduction was associated with an incremental increase inmedian overall survival of 3.00 months. The only otherpredictor variable independently associated with survivaltime was year of publication. This effect has also beenobserved in other meta-analyses of this nature and pre-sumably reflects improvements in the overall quality ofcancer care over time, although the possible contributions ofunmeasured variables, such as post-operative treatmentregimens, cannot be discounted [7,54–56].

In addition to characterizing variables predictive of survivaloutcome, a collective database of this nature also lends itselfto an examination of several other important aspects ofsurgery for recurrent ovarian cancer. The mean operative timeof approximately 4 h and estimated blood loss of just under600 cm3 are comparable to that of primary debulkingoperations [5,57,58]. Given the positive association betweensurgical outcome and median post-recurrence survival time,the 19.2% incidence of significant post-operative morbidityand 1.2% post-operative mortality rate are also within therange of acceptable risk. Additionally, an important negativefinding was that we were unable to identify a clinical profilethat could be utilized to select appropriate candidates forcytoreductive surgery, at least from the predictor variablesutilized in the current study. Previous investigators haveidentified disease-free interval, the extent of disease or numberof lesions, pre-cytoreduction size of largest tumor, and theperformance of secondary surgery prior to salvage chemother-apy as being associated with post-recurrence survival time[23,24,42]. It is possible that such characteristics are indeeduseful as surgical selection criteria but that the current analysiswas either unable to incorporate these variables, due to limitedor inconsistent data, or incapable of teasing out theirsignificance because of the coarseness of the measuresselected for methodological purposes.

Admittedly, a critical review of the data presented requiresan appreciation for the methodological limitations of a meta-analysis of this nature. First, although we made everyreasonable effort to include as many studies as possible, thecurrent analysis is inherently limited by the potential forselection bias, both for patient inclusion in the originalpublished reports as well as inclusion of the studies for meta-analysis. Secondly, the coarseness of data and inconsistentreporting of predictor variables over the 25-year datacollection interval, during which time concepts of cytoreduc-tive surgery, residual disease objectives, and availableadjuvant therapies greatly evolved, may have detracted fromour ability to fully evaluate the prognostic impact of certainvariables on survival. Along these lines, a third limitation ofthe current study is that the large variety of differentchemotherapeutic agents and administrations schedules usedduring the study period precluded an analysis of the impact ofthe type of adjuvant therapy on survival. Although thisomission may be a potential source of confounding, similarmeta-analyses examining the primary management ofadvanced ovarian cancer have not found these factors toaffect survival significantly [59,60]. A fourth limitation is thatwe did not examine additional prognostic factors, such as

272 R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

tumor size, number of lesions, and performance status thatmight have influenced either survival or the proportion ofpatients undergoing successful cytoreductive surgery. At leastone meta-analysis has found that performance status, inaddition to optimal surgery and the use of platinumchemotherapy, had a measurable effect on survival in theprimary treatment setting [61].

Despite the above limitations, several conclusions relevant toclinical practice can be drawn from the current study. Usingweighted calculations, these data indicate that, among patientsundergoing surgery for recurrent ovarian cancer, optimalcytoreduction, although variably defined, was achievable inapproximately 70% of cases, with just over 50% of patients ableto be surgically rendered free of macroscopic residual tumor(complete cytoreduction). The average weighted mediandisease-free interval of 20.2 months undoubtedly reflects aselection bias on the part of clinicians toward operating onpatients with chemo-sensitive disease. As such, a directcomparison of survival outcomes for patients with recurrentovarian cancer treated with surgery and chemotherapy versuschemotherapy alone would be statistically invalid. Never-theless, from an observational standpoint, the median post-recurrence survival time of 30.3 months following cytoreduc-tive surgery matches up favorably to the bulk of phase II andphase III studies of patients with recurrent platinum-sensitiveepithelial ovarian cancer treated with chemotherapy alone. Infact, only two phase III studies of chemotherapy treatment alonefor recurrent platinum-sensitive disease have reported mediansurvival times approaching 30 months [62,63]. In both instancesthe study populations were dominated by patients with extendeddisease-free intervals following primary therapy, with one studyreporting a median disease-free interval prior to re-treatmentwith platinum-based chemotherapy of 38.8 months [63].Notably, the majority of such chemotherapy-alone trialsdescribe median post-recurrence survival times ranging from15 months to 18.0 months [64–66]. Although clearly not arigorous scientific evaluation, from the more practical stand-point of clinical management these observations at least suggestsome additive survival benefit from successful cytoreductivesurgery for ovarian cancer recurrence when combined withadjuvant chemotherapy. Finally, in view of the consistentcontinuum effect between residual disease and survival out-come (i.e. “less is more”) and the intrinsic specificity in thedefinition of no gross residual disease, it seems reasonable toadopt complete cytoreduction as the most appropriate surgicalobjective for women undergoing attempted cytoreductivesurgery for recurrent ovarian cancer.

Conflict of interest statementDSC has served on the Speakers' Bureau for Genzyme Inc. REB and IP have noconflicts of interest to declare.

Acknowledgments

This work supported by a grant from the EntertainmentIndustry Foundation and the Callaway Foundation Women'sCancer Initiative. The authors gratefully acknowledge the

assistance of Dr. Bo Gronlund for providing additionalunpublished data for inclusion in this study.

References

[1] Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, et al. Cancer statistics2008. CA Cancer J Clin 2008;58:71–96.

[2] Meigs JV. Tumors of the female pelvic organs. New York: MacMillan;1934.

[3] Griffiths CT. Surgical resection of tumor bulk in the primary treatment ofovarian carcinoma. Natl Cancer Inst Monogr 1975;42:101–4.

[4] Hoskins WJ, Bundy BN, Thigpen JT, Omura GA. The influence ofcytoreductive surgery on recurrence-free interval and survival in small-volume stage III epithelial ovarian cancer: a Gynecologic Oncology Groupstudy. Gynecol Oncol 1992;47:159–66.

[5] Hoskins WJ, McGuire WP, Brady MF, Homesley HD, Creasman WT,Berman M, et al. The effect of diameter of largest residual disease onsurvival after primary cytoreductive surgery in patients with suboptimalresidual epithelial ovarian carcinoma. Am J Obstet Gynecol 1994;170:974–9.

[6] Chi DS, Liao JB, Leon LF, Venkatraman ES, Hensley ML, Bhaskaran D,et al. Identification of prognostic factors in advanced epithelial ovariancarcinoma. Gynecol Oncol 2001;82:532–7.

[7] Bristow RE, Tomacruz RS, Armstrong DK, Trimble EL, Montz FJ.Survival effect of maximal cytoreductive surgery for advanced ovariancarcinoma during the platinum era: a meta-analysis. J Clin Oncol2002;20:1248–59.

[8] Berek JS, Hacker NF, Lagasse LD, Nieberg RK, Elashoff RM. Survival ofpatients following secondary cytoreductive surgery in ovarian cancer.Obstet Gynecol 1983;61:189–93.

[9] Bristow RE, Lagasse LD, Karlan BY. Secondary surgical cytoreduction foradvanced epithelial ovarian cancer: patient selection and review of theliterature. Cancer 1996;78:2049–62.

[10] Rubin DB. Inference and missing data. Biometrika 1976;63:581–92.[11] Schafer JL. Analysis of incomplete multivariate data. London: Chapman

and Hall/CRC; 1997.[12] Morris M, Gershenson DM, Wharton JT, Copeland LJ, Edwards CL,

Stringer CA. Secondary cytoreductive surgery for recurrent epithelialovarian cancer. Gynecol Oncol 1989;34:334–8.

[13] Jänicke F, Hölscher M, Kuhn W, von Hugo R, Pache L, Siewart JR, et al.Radical surgical procedure improves survival time in patients withrecurrent ovarian cancer. Cancer 1992;70:2129–36.

[14] Segna RA, Dottino PR, Mandell JP, Konsker K, Cohen CJ. Secondarycytoreduction for ovarian cancer following cisplatin therapy. J Clin Oncol1993;11:434–9.

[15] Eisenkop SM, Friedman RL, Wang HJ. Secondary cytoreductive surgeryfor recurrent ovarian cancer. A prospective study. Cancer 1995;76:1606–14.

[16] Vaccarello L, Rubin SC, Vlamis V, Wong G, Jones WB, Lewis JL, et al.Cytoreductive surgery in ovarian carcinoma patients with a documentedpreviously complete surgical response. Gynecol Oncol 1995;57:61–5.

[17] Lichtenegger W, Sehouli J, Buchmann E, Karajanev C, Weidmann H.Operative results after primary and secondary debulking-operations inadvanced ovarian cancer (AOC). J Obstet Gynaecol Res 1998;24:447–51.

[18] Landoni F, Pellgrino A, Cormio G, Milani R, Maggioni A, Mangioni C.Platin-based chemotherapy and salvage surgery in recurrent ovarian cancerfollowing negative second-look laparotomy. Acta Obstet Gynecol Scand1998;77:233–7.

[19] CormioG, di Vagno G, Cazzolla A, Bettocchi S, di Gesu G, Loverro G, et al.Surgical treatment of recurrent ovarian cancer: report of 21 cases and reviewof the literature. Eur J Obstet Gynecol Reprod Biol 1999;86:185–8.

[20] Gadducci A, Iacconi P, Cosio S, Fanucchi A, Cristofani R, Genazzani R.Complete salvage surgical cytoreduction improves further survival ofpatients with late recurrent ovarian cancer. Gynecol Oncol 2000;79:344–9.

[21] Zang RY, Zhang ZY, Li ZT, Chen J, Tang MQ, Liu Q, et al. Effect ofcytoreductive surgery on survival of patients with recurrent epithelialovarian cancer. J Surg Oncol 2000;75:24–30.

273R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

[22] Chen LM, Leuchter RS, Lagasse LD, Karlan BY. Splenectomy andsurgical cytoreduction for ovarian cancer. Gynecol Oncol 2000;77:362–8.

[23] Eisenkop SM, Friedman RL, Spirtos NM. The role of secondarycytoreductive surgery in the treatment of patients with recurrent epithelialovarian cancer. Cancer 2000;88:144–53.

[24] Tay EH, Grant PT, Gebski V, Hacker NF. Secondary cytoreductive surgeryfor recurrent epithelial ovarian cancer. Obstet Gyenecol 2002;99:1008–13.

[25] Bristow RE, Gossett DR, Shook DR, Zahurak ML, Tomacruz RS,Armstrong DK, et al. Recurrent micropapillary serous carcinoma. The roleof secondary cytoreductive surgery. Cancer 2002;95:791–800.

[26] Yoon SS, Jarnagin WR, Fong Y, DeMatteo RP, Barakat RR, Blumgart LH,et al. Resection of recurrent ovarian or fallopian tube carcinoma involvingthe liver. Gynecol Oncol 2003:383–8.

[27] Zang RY, Li ZT, Zhang ZY, Cai SM. Surgery and salvage chemotherapyfor Chinese women with recurrent advanced epithelial ovarian carcinoma:a retrospective case-control study. Int J Gynecol Cancer 2003;13:419–27.

[28] Merideth MA, Cliby WA, Keeney GL, Lesnick TG, Nagorney DM,Podratz KC. Hepatic resection for metachronous metastases from ovariancarcinoma. Gynecol Oncol 2003;89:16–21.

[29] Look M, Chang D, Sugarbaker PH. Long-term results of cytoreductivesurgery for advanced and recurrent epithelial ovarian cancers and papillaryserous carcinoma of the peritoneum. Int J Gynecol Cancer 2003;13:764–70.

[30] Loizzi V, Chan JK, Osann K, Cappuccini F, DiSaia PJ, Berman ML.Survival outcomes in patients with recurrent ovarian cancer who weretreated with chemoresistance assay-guided chemotherapy. Am J ObstetGynecol 2003;189:1301–7.

[31] Leitao MM, Kardos S, Barakat RR, Chi DS. Tertiary cytoreduction inpatients with recurrent ovarian carcinoma. Gynecol Oncol 2004;95:181–8.

[32] Zang RY, Li ZT, Tang J, Cheng X, Cai SM, Zhang ZY, et al. Secondarycytoreductive surgery for patients with relapsed ovarian carcinoma: whobenefits? Cancer 2004;100:1152–61.

[33] Zanon C, Clara R, Chiappino I, Bortolini M, Cornaglia S, Simone P, et al.Cytoreductive surgery and intraperitoneal chemohyperthermia forrecurrent peritoneal carcinomatosis from ovarian cancer. World J Surg2004;28:1040–5.

[34] Uzan C, Morice P, Rey A, Pautier P, Camatte S, Lhommé C, et al.Outcomes after combined therapy including surgical resection in patientswith epithelial ovarian cancer recurrence(s) exclusively in lymph nodes.Ann Surg Oncol 2004;11:658–64.

[35] Gronlund B, Lundvall L, Christensen IJ, Knudsen JB, Hogdall C. Surgicalcytoreduction in recurrent ovarian carcinoma in patients with completeresponse to paclitaxel-platinum. Eur J Surg Oncol 2005;31:67–73.

[36] Güngör M, Ortac F, Arvas M, Kösebay D, Sönmezer M, Köse K. The roleof secondary cytoreductive surgery for recurrent ovarian cancer. GynecolOncol 2005;97:74–9.

[37] Yap OWS, Kapp DS, Teng NNH, Husain A. Intraoperative radiationtherapy in recurrent ovarian cancer. Int J Radiation Oncology Biol Phys2005;63:1114–21.

[38] Onda T, Yoshikawa H, Yasugi T, Yamada M, Matsumoto K, Taketani Y.Secondary cytoreductive surgery for recurrent epithelial ovariancarcinoma: proposal for patient selection. Br J Cancer 2005;92:1026–32.

[39] Ayhan A, Gultekin M, Taskiran C, Aksan G, Celik NY, Dursun P, et al. Therole of secondary cytoreduction in the treatment of ovarian cancer:Hacettepe University experience. Am J Obstet Gynecol 2006;194:49–56.

[40] Matsumoto A, Higuchi T, Yura S, Mandai M, Kariya M, Takakura K, et al.Role of salvage cytoreductive surgery in the treatment of patients withrecurrent ovarian cancer after platinum-based chemotherapy. J ObstetGynecol Res 2006;37:580–7.

[41] Manci N, Bellati F, Muzii L, Calcagno M, Alon SA, Pernice M, et al.Splenectomy during secondary cytoreduction for ovarian cancer diseaserecurrence: surgical and survival data. Ann Surg Oncol 2006;13:1717–23.

[42] Chi DS, McCaughty K, Diaz JP, Huh J, Schwabenbauer S, Hummer AJ,et al. Guidelines and selection criteria for secondary cytoreductivesurgery in patients with recurrent, platinum-sensitive epithelial ovariancarcinoma. Cancer 2006;106:1933–9.

[43] Harter P, du Bois A, Hahmann M, et al. Surgery in recurrent ovariancancer: the Arbeitsgemeinschaft Gynaekologische Onkologie (AGO)DESKTOP OVAR trial. Ann Surg Oncol 2006;13:1702–10.

[44] Rufian S, Munoz-Casares F, Briceno J, Diaz CJ, Rubio MJ, Ortega R, et al.Radical surgery-peritonectomy and intraoperative chemotherapy fortreatment of peritoneal carcinomatosis in recurrent or primary ovariancancer. J Surg Oncol 2006;94:316–24.

[45] Helm CW, Randall-Whitis L, Martin II RS, Metzinger DS, Gordinier ME,Parker LP, et al. Hyperthermic intraperitoneal chemotherapy inconjunction with surgery for the treatment of recurrent ovarian carcinoma.Gynecol Oncol 2007;105:90–6.

[46] Salani R, Santillan A, Zahurak ML, Giuntoli II RL, Gardner GJ,Armstrong DK, et al. Secondary cytoreductive surgery for localized,recurrent epithelial ovarian cancer: analysis of prognostic factors andsurvival outcome. Cancer 2007;109:685–91.

[47] Santillan A, Karam AK, Li AJ, Giuntoli R, Gardner GJ, Cass I, et al.Secondary cytoreductive surgery for isolated nodal recurrence in patientswith epithelial ovarian cancer. Gynecol Oncol 2007;104:686–90.

[48] Benedetti Panici P, DeVicoA, Bellati F,Manci N, PerniolaG, Basile S, et al.Secondary cytoreductive surgery in patients with platinum-sensitiverecurrent ovarian cancer. Ann Surg Oncol 2007;14:1136–42.

[49] Benedetti Panici P, Perniola P, Angioli R, ZulloMA,Manci N, Palaia I, et al.Bulky lymph node resection in patients with recurrent epithelial ovariancancer: impact of surgery. Int J Gynecol Cancer 2007;17:1245–51.

[50] Cotte E, Glehen O, Mohamed F, Lamy F, Falandry C, Gofier F, et al.Cytoreductive surgery and intraperitoneal chemohyperthermia forchemoresistant and recurrent advanced epithelial ovarian cancer:prospective study of 81 patients. World J Surg 2007;31:1813–20.

[51] Ozols RF, Bundy BN, Greer BE, Fowler JM, Clarke-Pearson D, BurgerRA, et al, Gynecologic Oncology Group. Phase III trial of carboplatinand paclitaxel compared with cisplatin and paclitaxel in patients withoptimally resected stage III ovarian cancer: a Gynecologic OncologyGroup study. J Clin Oncol 2003;21:3194–200.

[52] Ozols RF, Rubin SC, Thomas GM, Robboy SJ. Epithelial ovarian cancer.In: Hoskins WJ, Perez CA, Young RC, Barakat R, Markman M, RandallM, editors. Principles and practice of gynecologic oncology. 4th ed.Philadelphia, PA: Lippincott-Raven Publishers; 2005. p. 895–987.

[53] Munkarah AR, Coleman RL. Critical evaluation of secondary cytoreduc-tion in recurrent ovarian cancer. Gynecol Oncol 2004;95:273–80.

[54] Bristow RE, Chi DS. Platinum-based neoadjuvant chemotherapy andinterval surgical cytoreduction for advanced ovarian cancer: a meta-analysis. Gynecol Oncol 2006;103:1070–6.

[55] Balvert-Locht HR, Coeberg JW, Hop WC, Brölmann HA, Crommelin M,van Wijck DJ, et al. Improved prognosis of ovarian cancer in theNetherlands during the period 1975–1985: a registry-based study. GynecolOncol 1991;42:3–8.

[56] Bjorge T, Engeland A, Sundfor K, Tropé CG. Prognosis of 2,800patients with epithelial ovarian cancer diagnosed during 1975–1994 andtreated at the Norwegian Radium Hospital. Acta Obstet Gynecol Scand1998;77:777–81.

[57] Eisenkop SM, Friedman RL, Wang HJ. Complete cytoreductive surgery isfeasible and maximizes survival in patients with advanced epithelialovarian cancer: a prospective study. Gynecol Oncol 1998;69:103–8.

[58] Aletti GD, Dowdy SC, Gostout BS, Jones MB, Stanhope CR, Wilson TO,et al. Aggressive surgical effort and improved survival in advanced-stageovarian cancer. Obstet Gynecol 2006;107:77–85.

[59] Hunter RW, Alexander NDE, Soutter WP. Meta-analysis of surgery inadvanced ovarian carcinoma: is maximum cytoreductive surgery anindependent determinant of prognosis? Am J Obstet Gynecol 1992;166:504–11.

[60] Ben-David Y, Rosen B, Franssen E, Einarson T, Szyfer I. Meta-analysiscomparing cisplatin total dose intensity and survival. Gynecol Oncol1995;59:93–101.

[61] Voest EE, van Houwelingen JC, Neijt JP. A meta-analysis of prognosticfactors in advanced ovarian cancer with median survival and overallsurvival (measured with the log [relative risk]) as main objective. Eur JCancer Clin Oncol 1989;25:711–20.

[62] The ICON and AGO Collaborators. Paclitaxel plus platinum-basedchemotherapy versus conventional platinum-based chemotherapy inwomen with relapsed ovarian cancer: the ICON4/AGO-OVAR-2.2 trial.Lancet 2003;361:2099–106.

274 R.E. Bristow et al. / Gynecologic Oncology 112 (2009) 265–274

[63] Cantù MG, Buda A, Parma G, Rossi R, FLoriani I, Bonazzi C, et al.Randomized controlled trial of single-agent paclitaxel versus cyclopho-sphamide, doxorubicin, and cisplatin in patients with recurrent ovariancancer who responded to first-line platinum-based regimens. J Clin Oncol2002;20:1232–7.

[64] Bolis G, Scarfone G, Giardina G, Villa A, Mangili G, Melpignano M,et al. Carboplatin alone vs carboplatin plus epidoxorubicin as second-line therapy for cisplatin- or carboplatin-sensitive ovarian cancer, 81;2001. p. 3–9.

[65] Pfisterer J, Plant M, Vergote I, du Bois A, Hirte H, Lacave AJ, et al.Gemcitabine plus carboplatin compared with carboplatin in patients withplatinum-sensitive recurrent ovarian cancer: an intergroup trial of theAGO-OVAR, the NCIC CTG, and the EORTC GCG. J Clin Oncol2006;24:4699–707.

[66] Gordon AN, TondaM, Sun S, Rackoff W, Doxil Study 30–49 Investigators.Long-term survival advantage for women treated with pegylated doxo-rubicin compared with topotecan in a phase 3 randomized study of recurrentand refractory epithelial ovarian cancer. Gynecol Oncol 2004;95:1–8.