© The Author 2006. Published by Oxford University Press.
ARTICLE |
Survival Benefits With Diverse Chemotherapy Regimens for Ovarian Cancer: Meta-analysis of Multiple Treatments
Affiliations of authors: Department of Obstetrics and Gynaecology, Hammersmith Hospital, Queen Charlotte's and Chelsea Hospital, London, U.K. (MK); Department of Obstetrics and Gynaecology, University Hospital of Ioannina, Ioannina, Greece (MK, EP); Medical Research Council, Biostatistics Unit, Cambridge, U.K. (GS); Division of Medical Oncology, University of Ioannina School of Medicine, Ioannina, Greece (NP); Clinical Trials and Evidence-Based Medicine Unit, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece (JPAI); Biomedical Research Institute, Foundation for Research and Technology-Hellas, Ioannina, Greece (JPAI); Institute for Clinical Research and Health Policy Studies, Department of Medicine, Tufts University School of Medicine, Boston, MA (JPAI)
Correspondence to: John P. A. Ioannidis, MD, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, University Campus, Ioannina 45110, Greece (e-mail: jioannid{at}cc.uoi.gr).
 |
ABSTRACT |
|---|
 TopNotes Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background: Numerous randomized trials have compared different chemotherapy regimens in women with ovarian cancer. Although ovarian cancer survival has improved in recent years, the magnitude of these incremental benefits across diverse regimens is unclear. Methods: We used multiple-treatment meta-analysis methodology to combine information from direct and indirect comparisons of all chemotherapy regimens used in randomized trials of ovarian cancer in the last 40 years. Chemotherapy was categorized by the use or not of platinum and/or taxanes, combinations of agents, and intraperitoneal administration. Monte Carlo simulations were used to determine which regimen most improved survival. Analyses of trials that examined first- and second-line treatments were also performed separately. Results: We found 198 trials (N = 38 440 women) involving 120 different chemotherapy regimens published in 1971–2006. Eighty-two trials compared different types of chemotherapy, among which 60 had usable survival information (N = 15 609 women). Monte Carlo simulations showed a 92% probability that the regimen that best prolonged survival is a platinum and taxane combination with intraperitoneal administration; this regimen resulted in a 55% relative risk reduction (95% confidence interval [CI] = 39% to 67%) for mortality as compared with nonintraperitoneal monotherapy using neither platinum nor taxane. Against that same monotherapy comparator, platinum-based combinations with and without intraperitoneal administration achieved 40% (95% CI = 21% to 54%) and 30% (95% CI = 20% to 38%) relative risk reductions for mortality, respectively, and combinations involving platinum and taxane without intraperitoneal administration achieved a 42% (95% CI = 31% to 51%) relative risk reduction. Results were similar when analyses were limited to first-line treatment. Data on second-line treatment were consistent with the superiority of platinum and taxane combinations. Conclusions: Distinct incremental improvements in survival have been achieved for ovarian cancer chemotherapy over time, with the possibility to achieve a doubling or more of time to mortality with platinum and taxane combinations, especially when intraperitoneal administration is used.
|
INTRODUCTION |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Ovarian cancer is a major cause of death worldwide, and it is especially prevalent in developed countries (1–5). The overall 5-year survival rate is poor, at approximately 30% for women older than 65 years (6). Typically, primary treatment consists of surgery and some form of chemotherapy (7,8). Alkylating agents were used initially, with poor results, but platinum-based chemotherapy has improved outcomes in the last 25 years (9–12). In the last 10 years, evidence has also started to accumulate that taxanes and their combination with platinum agents may further improve outcomes (7,13–17). Data from recent studies also suggest that additional benefits are achieved by intraperitoneal chemotherapy (18).
Although there is wide consensus that chemotherapy prolongs survival in women with epithelial ovarian cancer, the exact magnitude of the survival benefits with various regimens is unclear. Approximately, 200 randomized trials comparing different chemotherapy regimens for ovarian cancer have been published. Because newer or more intensive regimens also have higher toxicity, it is important to quantify the incremental survival benefits, if any, conferred by each type of chemotherapy as compared with older regimens.
Here we performed an overview of all randomized trials comparing chemotherapy regimens in patients with ovarian cancer. We aimed to present the evolution of the randomized evidence over time as newer regimens were tested. We examined whether specific regimens are superior to others and evaluated the magnitude of the benefit. Given the wide spectrum of comparisons available, we used multiple-treatment meta-analysis (19–21). This method allowed us to integrate data from both direct and indirect comparisons of diverse regimens.
|
METHODS |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Search Strategy and Eligibility Criteria
We searched MEDLINE, EMBASE, and the Central Registry of Controlled Trials of the Cochrane Library (1965 until January 2006). The search strategy used "ovarian cancer or neoplasia" and "chemotherapy" with an array of terms suggestive of randomized trials (22). The full strategy is available on request. We also perused the references of retrieved articles and of previous meta-analyses. Additional cross-searches were performed in MEDLINE using the names of investigators who were the lead authors of at least one eligible trial. In addition, we hand-searched several years of the journals with the highest number of electronically identified trials (23).
We considered all randomized controlled trials published in English, German, French, or Italian that compared at least two arms of different chemotherapy regimens (different agents and/or schedules) in at least five patients with ovarian cancer. Trials were included regardless of stage and line of treatment. Trials with at least three arms were included if at least two arms addressed an eligible comparison; noneligible arms were excluded. We excluded meeting abstracts; nonrandomized trials and pseudorandomized trials with alternate allocation of subjects; data on other malignancies or nonepithelial ovarian cancer; trials comparing radiotherapy, immunotherapy, hormonal therapy, and gene therapy (unless the above regimens were the same in all compared arms and the difference between arms pertained to the chemotherapeutic regimens only); and comparisons of chemotherapy against no chemotherapy (best supportive care). In cases of overlap or duplicate reports, we retained only the survival data with the maximal follow-up.
Data Extraction
From each eligible trial, we recorded the first author, publication year, journal, country or countries of the investigators, sample size (randomized and considered eligible for survival analyses [total and per arm]), regimens compared, the line of chemotherapy, previous chemotherapy, stage(s) of the disease, and percentage of patients with performance status 2 or worse (Karnofsky score 
70) per arm.
The type of chemotherapy regimen was categorized according to whether it involved platinum, taxane, both, or neither; used a combination of agents or monotherapy; and used intraperitoneal administration of any agents or not. Theoretically, this categorization gives rise to 14 possible types of chemotherapy, not all of which have necessarily been tested in trials.
For each trial, we recorded the median survival and the number of deaths per arm, if available, and whether there was a statistically significant difference in survival between the compared arms (two-tailed P<.05). When several analyses were reported, we preferred the log-rank test results over other statistics and unadjusted analyses over adjusted analyses. For trials including at least two arms, statistical significance was assessed for an analysis considering all arms.
For trials that compared at least two different types of chemotherapy regimens, we also extracted or estimated the hazard ratio (HR) and its variance. We used the reported hazard ratios and 95% confidence intervals (CIs) from Cox regression models. Unadjusted hazard ratios were preferred over multivariable ones. For trials that did not provide this information on hazard ratios and their uncertainty, we imputed these data by using the number of events (E1, E2) and patients (T1, T2) in each arm and the presented log-rank P value. We estimated the variance by the formula (T1 + T2)2/[(E1 + E2)T1T2] and then estimated the natural logarithm of the hazard ratio such that it would have the P value denoted by the log-rank test (24). If the log-rank P value was also not available, we estimated the hazard ratio as the inverse of the ratio of the median survival times from the presented Kaplan–Meier curves assuming exponential survival curves and proportional hazards.
Data were extracted independently by two investigators (MK and JPAI). Discrepancies were discussed to reach consensus.
Statistical Analyses
We generated descriptive statistics for trial and study population characteristics across all eligible trials. We described the types of comparisons and how these had evolved over time.
We conducted a series of meta-analyses summarizing the log hazard ratios for each comparison of two different types of treatment applying random effects (25). Between-study heterogeneity was estimated using the I2 statistic (26).
Multiple-treatment meta-analysis is a method of synthesizing information from a network of trials (27,28). Trials comparing treatments A and B provide direct evidence on the relative effect size 
(here the log hazard ratio). However, it is possible that indirect information may also be available through a common comparator, e.g., treatment C. Studies comparing A versus B, B versus C, and C versus A form a simple network. Indirect evidence about the effectiveness of A versus B is given by 
The combination of the direct and indirect estimates and 
into a single effect size can increase precision while randomization is respected (27,28). The combination of direct and indirect evidence for any given treatment comparison can be extended when ranking more than three types of treatments according to their effectiveness: every study contributes evidence about a subset of these treatments.
Multiple-treatment meta-analysis should be used with caution, and the underlying assumptions of the exchangeability of studies across the entire network should be investigated carefully (19,28,29). Exchangeability means that characteristics of the trials would not greatly modify the magnitude of the treatment effects. Exchangeability was tested by examining heterogeneity and incoherence. Joint analysis of treatments can be misleading if the network is substantially incoherent, i.e., if there is disagreement between indirect and direct estimates and 
We performed multiple-treatment meta-analysis with a linear mixed-effects model. Incoherence was taken into account by applying a model that suggests two levels in random effects—one at the level of general comparison being made (e.g., A versus B) and the other at the level of each specific study (30). The comparison-specific effects follow a distribution with zero mean and standard deviation that reflects the incoherence of the network.
One set of multiple-treatment meta-analysis addressed all eligible trials comparing different types of chemotherapy. However, line of treatment may be an important confounder. Therefore, we also considered two separate analyses: one that included only trials that addressed first-line treatment and another that included trials that addressed second-line therapy.
Analyses were conducted in R, version 2.2.1 (R Institute, Vienna, Austria) using lme functions. We also estimated the probability for each type of treatment to be the best at prolonging survival, given the results of the multiple-treatment meta-analysis, using Monte Carlo simulations in WinBUGS, version 1.4 (MRC Biostatistics Unit, Cambridge, U.K., http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml). A major advantage of working within a Bayesian framework is that probabilistic statements about the effectiveness of the different treatments can be made. The prior assumption is that all treatments are equally likely to be the best. For example, the probability of a treatment Ti to be the best out of N treatments can be calculated as the posterior probability P(Rank(Ti) = N). In WinBUGS code, P.most.effective<-equals(rank(mean[],i),N).
|
RESULTS |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Eligible Trials
The electronic searches yielded 4653 items from MEDLINE, 3276 from EMBASE, and 1916 from Cochrane Central. Of these, 727 potentially eligible articles were scrutinized. We excluded 529 reports that did not meet eligibility criteria (Fig. 1), leaving 198 trials that qualified for the overview.
|
The 198 trials (Table 1; Supplementary Data, available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22) had been published from 1971 to January 2006. The median sample size was 103 patients (interquartile range = 53–234 patients). Only four trials had more than 1000 subjects; 18 had more than 500 subjects. A total of 38 440 subjects were randomly assigned across 421 arms, and 36 029 subjects were included in survival analyses. In most reports, no previous chemotherapy had been given. There was considerable variability in the percentage of patients with poor performance status, but the percentage of such enrolled patients decreased over time (–6% per decade, P = .003). In only two trials did the majority of patients have stage I disease. Most trials were done in the United States and in Europe (Table 1). Only 5% of the studies included one or more intraperitoneally administered agents.
|
Eighty-two of the 198 trials compared different types of regimens according to our a priori categorization. Overall, this set of 82 trials had a similar profile as the total dataset of 198 trials (Table 1).
Compared Chemotherapy Regimens
A total of 120 different regimens were tested in the 198 trials. Only six monotherapies and 15 combinations of different agents had been tested in at least four trial arms each (Supplementary Table 1, available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). Moreover, doses and schedules of the same regimen often differed across trials. The most commonly used regimen—cisplatin, cyclophosphamide, and doxorubicin—had been used in 34 arms, followed by cisplatin and cyclophosphamide (n = 33 arms), paclitaxel (n = 24), cisplatin (n = 23), carboplatin (n = 21), and carboplatin and paclitaxel (n = 19). Nonplatinum, nontaxane regimens dominated trials published in the 1970s, platinum-based combinations prevailed in the 1980s and 1990s, and combinations including platinum and taxane are the most common regimens used recently (Table 2).
|
Of 116 trials that compared only regimens of the same type, only six found statistically significant differences in survival (Supplementary Table 2, available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22). The percentage (5%) is not beyond what is expected by chance. In two trials with statistically significant differences, the median survival difference between the compared arms was less than 2 months. Three trials showed differences among different regimens involving platinum, but there were 59 trials comparing different platinum-based regimens. One small trial found differences of borderline statistical significance between two regimens involving neither platinum nor taxanes among 29 trials that compared such regimens. Therefore, although minor differences between regimens using the same classes of drugs cannot be totally excluded, these differences would probably be very small overall, if not chance findings.
Comparisons of Different Types of Regimens
Of the 82 trials comparing different types of regimens, 60 had survival data (Table 3; Supplementary Table 3 and Supplementary Fig. 1, available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22), and these were included in quantitative syntheses (16 478 randomly assigned subjects, 15 609 in survival analyses). The other 22 trials (1380 randomly assigned subjects) were typically phase 1/2 or phase 2 trials with no usable survival data.
|
We first made pairwise comparisons of regimens from the 60 trials based on direct evidence only (Fig. 2). Heterogeneity (I2 > 35%) was found only for the comparison of platinum-based combinations versus platinum and taxane–based combinations (I2 = 70%, four studies), among which three trials using a cisplatin background found a statistically significant survival benefit for the addition of taxane, whereas the trial with a carboplatin background found no such benefit. Direct comparisons showed that platinum monotherapy was statistically significantly better than monotherapy with a nonplatinum, nontaxane agent, and a platinum-based combination was better than monotherapy or combinations involving neither platinum nor taxanes (relative risk reduction of mortality [RRR] = 37%, 22%, and 23%, respectively). Moreover, combinations involving neither platinum nor taxane were better than monotherapy with such agents (RRR = 17%), platinum and taxane–based combinations were better than platinum-based combinations (RRR = 21%), and intraperitoneal administration improved survival with platinum and taxane–based combinations (RRR = 21%).
|
Multiple-Treatment Meta-analysis
We next performed multiple-treatment meta-analysis of all 60 trials with survival data (Table 4). The combination of nonplatinum and nontaxane agents (not given intraperitoneally) was associated with a small decrease in mortality risk (RRR = 13%) compared with monotherapy with a nonplatinum, nontaxane agent. Platinum monotherapy (RRR = 32%) or platinum-based combination therapy with (RRR = 40%, 95% CI = 21% to 54%) or without (RRR = 30%, 95% CI = 20% to 38%) intraperitoneal administration was associated with more prominent statistically significant survival benefits than nonplatinum agents. Data on taxane-based regimens were sparse, and no formally statistically significant survival benefits were found. Combinations involving both platinum and taxane achieved the most impressive benefits (when intraperitoneal administration was involved, RRR = 55%, 95% CI = 39% to 67%, otherwise RRR = 42%, 95% CI = 31% to 51%).
|
Monte Carlo simulations showed a 92% probablilty that combinations of platinum and taxane with intraperitoneal administration were the most effective regimens, whereas there was a 6% probability that the best regimen was platinum-based combination chemotherapy with intraperitoneal administration. Therefore, it was 98% likely that the most effective regimen included intraperitoneal administration. The probability that any other type of regimen was the best for prolonging survival was less than 2%. Similar results were seen also in a more simplified multiple-treatment meta-analysis that merged the nonplatinum–nontaxane monotherapy and combination regimens in the same treatment category (Supplementary Table 4, available at: http://jncicancerspectrum.oxfordjournals.org/jnci/content/vol98/issue22).
According to the two-level random-effects model, no important incoherence between comparisons was detected (incoherence = .001). Only one particular comparison loop showed incoherence. Specifically, in the direct comparison, platinum monotherapy was marginally better than taxane monotherapy (HR = 0.87, 95% CI = 0.71 to 1.06), but it was much better based on indirect evidence from the comparison of these two regimens against monotherapy with nonplatinum, nontaxane agents (HR = 0.53, 95% CI = 0.36 to 0.80). The reason for this incoherence was that this comparison loop includes both trials with exclusively first-line treatment and trials with patients on second-line treatment. The nonplatinum, nontaxane regimens for second-line treatment included topotecan, which was not included in first-line therapy. Furthermore, hazard ratios estimated from direct analyses limited to first-line treatments were not materially different from those estimated from analyses of both first- and second-line therapies (Table 4).
Given the small number of trials assessing second-line treatment, the multiple-treatment meta-analysis method is unstable and thus potentially unreliable for the analysis of second-line treatments. There was a suggestion that platinum and taxane–containing combinations could result in a borderline statistically significant relative risk reduction of mortality against platinum monotherapy (HR = 0.53, 95% CI = 0.29 to 0.96) and taxane monotherapy (HR = 0.47, 95% CI = 0.22 to 0.99). On direct comparison, the hazard ratio for mortality of platinum and taxane–containing combinations versus platinum monotherapy was 0.82 (95% CI = 0.69 to 0.97), while there was no direct comparison of the combination versus taxane monotherapy.
|
DISCUSSION |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We examined 198 randomized trials that compared various chemotherapy regimens in women with ovarian cancer during the last 40 years. We show conclusively the major progress that has been achieved in prolonging the survival of these patients and we have quantified the magnitude of the survival benefit. Compared with the early days, when neither platinum nor taxanes were available and chemotherapy was clearly no better than supportive care for mortality outcomes (10), median survival can now be more than doubled using currently available regimens. A platinum and taxane–based combination with intraperitoneal administration (18) can reduce the risk of death by 55%. More than half (30%) of this benefit may be achieved with a standard platinum-based combination. Taxanes without platinum do not appear to confer any clear survival benefit. Although the data pertain predominantly to first-line treatment, the sparse data on second-line treatment are consistent with a superiority of combined platinum and taxane regimens.
The absolute magnitude of the increase in survival time is impressive. For an expected median survival of 2.5 years without effective treatment, approximately 3 years are gained with the most effective current treatments. However, these benefits should be tailored to the individual patient and balanced against tolerability. Multidrug combinations including both platinum and taxanes can cause cumulative toxicities (14,31,32). Intraperitoneal administration in particular has not been used in many centers around the world, and its implementation may cause some challenges (33) and extra toxicities (18). Platinum and taxane–based combinations with intraperitoneal administration are the best regimens for survival in our multiple-treatment meta-analysis with very high probability; we estimate that some kind of intraperitoneal regimen is 98% likely to be the best. However, if intraperitoneal administration proves difficult to implement, survival can still be almost doubled with a combination of platinum and taxanes without intraperitoneal administration.
The long-term evolution of cancer chemotherapy therapeutics creates difficulty in estimating the relative merits of regimens that are introduced over a long period of successive steps. Newer regimens are unavoidably compared with previous regimens, rather than to very old agents or no treatment. The multiple-treatment meta-analysis approach is attractive in this setting because it can accommodate such successive steps through loops of indirect comparisons. Multiple-treatment meta-analysis can prove a particularly useful approach to synthesize evidence on medical domains among which data on many treatments accumulate, a scenario that is becoming relatively common in clinical therapeutics (20). However, caution should be used for trials that are distant in time because of the potential danger of confounding with other factors that vary over time, such as supportive or adjunct treatment practices (34).
We should also caution that a full evaluation of the possibility of incoherence for nine treatments would require up to 112 loops to be evaluated in the multiple-treatment meta-analysis framework. In our network, there were many treatment comparisons for which no direct evidence was available and the evaluation of incoherence was impossible. Therefore, some conservative interpretation is warranted, especially for treatments that have been involved in few trials. In this regard, more evidence on regimens that include intraperitoneally administered agents, as well as longer term follow-up of the existing trials, would be useful.
Another possible limitation is that the current analysis is based on published group data, rather than individual-level information. Therefore, we have included women across a wide range of prognoses, from stage I to advanced disease. However, the vast majority of trials did not include any meaningful proportion of data for very early stage disease, a reflection of the fact that most women with ovarian cancer are still diagnosed at relatively advanced stages. Individual-level information might allow a more detailed appraisal of outcomes at different levels of risk (35). Individual-level data would also be useful to address whether there is any effect modification by the type of surgical intervention, stage, and whether residual tumor is left after surgery. In addition to depending on chemotherapy, optimization of outcomes may depend also on the surgical approach. However, the power to detect effect modifications might still be limited, even with individual-level information, in such a complex network of multiple treatments.
It is also notable that, even though almost 40 000 women have participated in published randomized trials of chemotherapy regimens in ovarian cancer, the large majority of investigations pertain to comparisons of similar types of treatment with minor modifications in the dosing or schedule or minor differences in the compared drugs that belong to the same class of agents. Although such trials may still help to improve our understanding of minor variants of different regimens, especially regarding toxicity and tolerability, they are unlikely to result in major progress regarding improvement of survival. The few exceptions of statistically significant survival differences that we have seen in such trials may be real but may also simply reflect chance findings. With many investigators working in this field, it would be useful to design future trials that will fill in important gaps where major loops of evidence have minimal or no information. An efficient strategic plan for clinical research in the field would require infrastructure support. Moreover, there is a need for clinical trials cooperative groups to develop complementary trials, to decrease duplication, and when appropriate, to also perform joint trials. The multiple-treatment meta-analysis can offer an additional impetus to such an effort, and it may also be useful as new regimens, including molecularly targeted treatments (36,37), are developed.
|
NOTES |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The authors have full responsibility for the study design, data collection and analysis, interpretation of the results, and the preparation of and decision to submit the manuscript.
|
REFERENCES |
|---|
|
TopNotesAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
(1) Harries M, Gore M. Part I: chemotherapy for epithelial ovarian cancer-treatment at first diagnosis. Lancet Oncol 2002;3:529–36.[CrossRef][Web of Science][Medline]
(2) Greenlee RT, Hill-Harmon MB, Murray T, Thun M. Cancer statistics, 2001. CA Cancer J Clin 2001;51:15–36.
(3) American Cancer Society. Cancer facts and figures: 2000. New York (NY): American Cancer Society; 2000.
(4) Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, et al. Cancer statistics, 2005 [erratum in CA Cancer J Clin 2005;55:259]. CA Cancer J Clin 2005;55:10–30.
(5) Quinn M, Babb P, Brock A, Kirby L, Jones J. Cancer trends in England and Wales 1950–1999. National Statistics SMPS No. 66. Available at: http://www.statistics.gov.uk/downloads/theme_health/cancertrends_5099.pdf. [Last accessed: October 5, 2006.]
(6) Cannistra SA. Cancer of the ovary. N Engl J Med 2004;351:2519–29.
(7) McGuire WP, Hoskins WJ, Brady MF, Kucera PR, Partridge EE, Look KY, et al. Cyclophosphamide and cisplatin compared with paclitaxel and cisplatin in patients with stage III and stage IV ovarian cancer. N Engl J Med 1996;334:1–6.
(8) Bristow RE, Tomacruz RS, Armstrong DK, Trimble EL, Montz FJ. Survival effect of maximal cytoreductive surgery for advanced ovarian carcinoma during the platinum era: a meta-analysis. J Clin Oncol 2002;20:1248–59.
(9) Advanced Ovarian Cancer Trialists' Group. Chemotherapy in advanced ovarian cancer: four systematic meta-analyses of individual patient data from 37 randomized trials. Br J Cancer 1998;78:1479–87.[Web of Science][Medline]
(10) Advanced Ovarian Cancer Trialists' Group. Chemotherapy in advanced ovarian cancer: an overview of randomised clinical trials. BMJ 1991;303:884–93.
(11) Ovarian Cancer Meta-analysis Project. Cyclophosphamide plus cisplatin versus cyclophosphamide, doxorubicin and cisplatin chemotherapy of ovarian carcinoma: a meta-analysis. J Clin Oncol 1991;9:1668–74.[Abstract]
(12) A'Hern RP, Gore ME. Impact of doxorubicin on survival in advanced ovarian cancer. J Clin Oncol 1995;13:726–32.
(13) Piccart MJ, Bertelsen K, James K, Cassidy J, Mangioni C, Simonsen E, et al. Randomized intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epithelial ovarian cancer: three-year result. J Natl Cancer Inst 2000;92:699–708.
(14) Muggia FM, Braly PS, Brady MF, Sutton G, Niemann TH, Lentz SL, et al. Phase III randomized study of cisplatin versus paclitaxel versus cisplatin and paclitaxel in patients with suboptimal stage III or IV ovarian cancer: a Gynecologic Oncology Group Study. J Clin Oncol 2000;18:106–15.
(15) Ozols RF, Bundy BN, Greer BE, Fowler JM, Clarke-Pearson D, Burger RA, et al. Phase III trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected stage III ovarian cancer. J Clin Oncol 2003;21:3194–200.
(16) du Bois A, Luck HJ, Meier W, Adams HP, Mobus V, Costa S, et al. A randomized clinical trial of cisplatin/paclitaxel vs carboplatin/paclitaxel as first-line treatment of ovarian cancer. J Natl Cancer Inst 2003;95:1320–9.
(17) Neijt JP, Engelholm SA, Tuxen MK, Sorensen PG, Hansen M, Sessa C, et al. Exploratory phase III study of paclitaxel and cisplatin versus paclitaxel and carboplatin in advanced ovarian cancer. J Clin Oncol 2000;18:3084–92.
(18) Armstrong DK, Bundy B, Wenzel L, Huang HQ, Baergen R, Lele S, et al. Intraperitoneal cisplatin and paclitaxel in ovarian cancer. N Engl J Med 2006;354:34–43.
(19) Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ 2005;331:897–900.
(20) Lumley T. Network meta-analysis for indirect treatment comparisons. Stat Med 2002;21:2313–24.[CrossRef][Web of Science][Medline]
(21) Psaty BM, Lumley T, Furberg CD, Schellenbaum G, Pahor M, Alderman MH, et al. Health outcomes associated with various antihypertensive therapies used as first-line agents: a network meta-analysis. JAMA 2003;289:2534–44.
(22) Cochrane Collaboration. Authors' handbook. Available at: http://www.cochrane.org. [Last accessed: October 5, 2006.]
(23) Hopewell S, Clarke M, Lusher A, Lefebvre C, Westby M. A comparison of handsearching versus MEDLINE searching to identify reports of randomized controlled trials. Stat Med 2002;21:1625–34.[CrossRef][Web of Science][Medline]
(24) Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815–34.[CrossRef][Web of Science][Medline]
(25) Lau J, Ioannidis, Schmid CH. Quantitative synthesis in systematic reviews. Ann Intern Med 1997;127:820–6.
(26) Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58.[CrossRef][Web of Science][Medline]
(27) Glenny AM, Altman DG, Song F, Sakarovitch C, Deeks JJ, D'Amico R, et al. Indirect comparisons of competing interventions. Health Technol Assess 2005;9:1-iv.[Medline]
(28) Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med 2004;23:3105–24.[CrossRef][Web of Science][Medline]
(29) Baker SG, Kramer BS. The transitive fallacy for randomized trials: if A bests B and B bests C in separate trials, is A better than C? BMC Med Res Methodol 2002;2:13.[CrossRef][Medline]
(30) Lumley T, Keech A. Meta-meta-analysis with confidence. Lancet 1995;346:576–7.[Web of Science][Medline]
(31) Herzog TJ. Recurrent ovarian cancer: how important is it to treat to disease progression? Clin Cancer Res 2004;10:7439–49.
(32) ICON2: randomised trial of single-agent carboplatin against three-drug combination of CAP (cyclophosphamide, doxorubicin, and cisplatin) in women with ovarian cancer. ICON Collaborators. International Collaborative Ovarian Neoplasm Study. Lancet 1998;352:1571–6.[CrossRef][Web of Science][Medline]
(33) Fujiwara K, Suzuki S, Ishikawa H, Oda T, Aotani E, Kohno I. Preliminary toxicity analysis of intraperitoneal carboplatin in combination with intravenous paclitaxel chemotherapy for patients with carcinoma of the ovary, peritoneum, or fallopian tube. Int J Gynecol Cancer 2005;15:426–31.[CrossRef][Web of Science][Medline]
(34) Trikalinos TA, Ioannidis JP. Predictive modeling and heterogeneity of baseline risk in meta-analysis of individual patient data. J Clin Epidemiol 2001;54:245–52.[CrossRef][Web of Science][Medline]
(35) Ioannidis JPA. Indirect comparisons: the mesh and mess of clinical trials. Lancet 2006;368:1470–2.[CrossRef][Web of Science][Medline]
(36) Muggia F, Kosloff R. Investigational agents for epithelial ovarian cancer. Expert Rev Anticancer Ther 2005;5:855–68.[CrossRef][Web of Science][Medline]
(37) Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, et al. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature 2006;439:353–7.[CrossRef][Medline]
Manuscript received June 7, 2006; revised August 29, 2006; accepted September 21, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. A. Bookman, M. F. Brady, W. P. McGuire, P. G. Harper, D. S. Alberts, M. Friedlander, N. Colombo, J. M. Fowler, P. A. Argenta, K. De Geest, et al. Evaluation of New Platinum-Based Treatment Regimens in Advanced-Stage Ovarian Cancer: A Phase III Trial of the Gynecologic Cancer InterGroup J. Clin. Oncol., March 20, 2009; 27(9): 1419 - 1425. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Mozzetti, R. Iantomasi, I. De Maria, S. Prislei, M. Mariani, A. Camperchioli, S. Bartollino, D. Gallo, G. Scambia, and C. Ferlini Molecular Mechanisms of Patupilone Resistance Cancer Res., December 15, 2008; 68(24): 10197 - 10204. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Salanti, J. P. Higgins, A. Ades, and J. P. Ioannidis Evaluation of networks of randomized trials Statistical Methods in Medical Research, June 1, 2008; 17(3): 279 - 301. [Abstract] [PDF]
|
|
|
|
 |
|
 A. M. C. Swart, S. Burdett, J. Ledermann, P. Mook, and M. K. B. Parmar Why i.p. therapy cannot yet be considered as a standard of care for the first-line treatment of ovarian cancer: a systematic review Ann. Onc., April 1, 2008; 19(4): 688 - 695. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Salanti, F. K. Kavvoura, and J. P.A. Ioannidis Exploring the Geometry of Treatment Networks Ann Intern Med, April 1, 2008; 148(7): 544 - 553. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. L. Trimble, S. Thompson, M. C. Christian, and L. Minasian Intraperitoneal Chemotherapy for Women with Epithelial Ovarian Cancer Oncologist, April 1, 2008; 13(4): 403 - 409. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Rao, M. Crispens, and M. L. Rothenberg Intraperitoneal Chemotherapy for Ovarian Cancer: Overview and Perspective J. Clin. Oncol., July 10, 2007; 25(20): 2867 - 2872. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P Zola and A Ferrero Is carboplatin-paclitaxel combination the standard treatment of elderly ovarian cancer patients? Ann. Onc., February 1, 2007; 18(2): 213 - 214. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||