Carboplatin plus paclitaxel once a week versus every 3 weeks in patients with advanced ovarian cancer (MITO-7): a randomised, multicentre, open-label, phase 3 trial

Articles

Carboplatin plus paclitaxel once a week versus every 3 weeks in patients with advanced ovarian cancer (MITO-7): a randomised, multicentre, open-label, phase 3 trial Sandro Pignata, Giovanni Scambia, Dionyssios Katsaros, Ciro Gallo, Eric Pujade-Lauraine, Sabino De Placido, Alessandra Bologna, Beatrice Weber, Francesco Raspagliesi, Pierluigi Benedetti Panici, Gennaro Cormio, Roberto Sorio, Maria Giovanna Cavazzini, Gabriella Ferrandina, Enrico Breda, Viviana Murgia, Cosimo Sacco, Saverio Cinieri, Vanda Salutari, Caterina Ricci, Carmela Pisano, Stefano Greggi, Rossella Lauria, Domenica Lorusso, Claudia Marchetti, Luigi Selvaggi, Simona Signoriello, Maria Carmela Piccirillo, Massimo Di Maio, Francesco Perrone, on behalf of the Multicentre Italian Trials in Ovarian cancer (MITO-7), Groupe d’Investigateurs Nationaux pour l’Etude des Cancers Ovariens et du sein (GINECO), Mario Negri Gynecologic Oncology (MaNGO), European Network of Gynaecological Oncological Trial Groups (ENGOT-OV-10), and Gynecologic Cancer InterGroup (GCIG) Investigators*

Summary Lancet Oncol 2014; 15: 396–405 Published Online February 28, 2014 http://dx.doi.org/10.1016/ S1470-2045(14)70049-X See Comment page 363 *See appendix pp 7–8 for a full list of investigators Dipartimento di Oncologia Uroginecologica (S Pignata MD, C Pisano MD, S Greggi MD) and Unità Sperimentazioni Cliniche (M C Piccirillo MD, M Di Maio MD, F Perrone MD), Istituto Nazionale per lo Studio e la Cura dei Tumori, Fondazione G Pascale, IRCCS, Napoli, Italy; Dipartimento per la Tutela della Salute della Donna, della Vita Nascente, del Bambino e dell’ Adolescente, Università Cattolica del Sacro Cuore, Roma, Italy (Prof G Scambia MD, V Salutari MD, C Ricci MD); Ginecologia Oncologica, Azienda Ospedaliera Città della Salute e della Scienza, Presidio S Anna e Università, Torino, Italy (Prof D Katsaros MD); Statistica Medica, Seconda Università di Napoli, Napoli, Italy (Prof C Gallo MD, S Signoriello PhD); Hôpital Hôtel-Dieu, Paris, France (Prof E Pujade-Lauraine MD); Dipartimento di Medicina Clinica e Chirurgia, Università di Napoli Federico II, Napoli, Italy (Prof S De Placido MD, R Lauria MD); Oncologia Medica, Azienda Ospedaliera ASMN, IRCCS, Reggio Emilia, Italy (A Bologna MD); Centre Alexis Vautrin, Vandoeuvre-Les-Nancy, France (B Weber MD); Unità di Ginecologia Oncologica, Istituto Nazionale per lo Studio e la Cura dei Tumori, IRCCS, Milano, Italy (F Raspagliesi MD, D Lorusso MD); Dipartimento di

396

Background Carboplatin plus paclitaxel administered every 3 weeks is standard first-line chemotherapy for patients with advanced ovarian cancer. A weekly paclitaxel schedule combined with carboplatin every 3 weeks prolonged progression-free survival and overall survival in a Japanese phase 3 trial. The aim of our study was to assess whether a weekly schedule of carboplatin plus paclitaxel is more effective than the same drugs given every 3 weeks. Methods We did a multicentre, randomised, phase 3 study at 67 institutions in Italy and France. Women with FIGO stage IC–IV ovarian cancer, an ECOG performance status of 2 or lower, and who had never received chemotherapy were randomly allocated in a 1:1 ratio to receive either carboplatin (AUC 6 mg/mL per min) plus paclitaxel (175 mg/m²) every 3 weeks for six cycles or carboplatin (AUC 2 mg/mL per min) plus paclitaxel (60 mg/m²) every week for 18 weeks. Randomisation was done by computer-based minimisation, stratified by centre, residual disease after surgery, and ECOG performance status. The study was not blinded. Coprimary endpoints were progression-free survival and quality of life (assessed by the Functional Assessment of Cancer Therapy Ovarian Trial Outcome Index [FACT-O/TOI] score), and analysis was by modified intention to treat. This report presents the final analysis. The study is registered with ClinicalTrials.gov, number NCT00660842. Findings 822 patients were enrolled into the study between Nov 20, 2008, and March 1, 2012; 12 withdrew their consent immediately after randomisation and were excluded, and 810 were eligible for analysis. 404 women were allocated treatment every 3 weeks and 406 were assigned to the weekly schedule. After median follow-up of 22·3 months (IQR 16·2–30·9), 449 progression-free survival events were recorded. Median progression-free survival was 17·3 months (95% CI 15·2–20·2) in patients assigned to treatment every 3 weeks, versus 18·3 months (16·8–20·9) in women allocated to the weekly schedule (hazard ratio 0·96, 95% CI 0·80–1·16; p=0·66). FACT-O/TOI scores differed significantly between the two schedules (treatment-by-time interaction p<0·0001); with treatment every 3 weeks, FACT-O/TOI scores worsened at every cycle (weeks 1, 4, and 7), whereas for the weekly schedule, after transient worsening at week 1, FACT-O/TOI scores remained stable. Fewer patients assigned to the weekly group than those allocated treatment every 3 weeks had grade 3–4 neutropenia (167 [42%] of 399 patients vs 200 [50%] of 400 patients), febrile neutropenia (two [0·5%] vs 11 [3%]), grade 3–4 thrombocytopenia (four [1%] vs 27 [7%]), and grade 2 or worse neuropathy (24 [6%] vs 68 [17%]). Three deaths during the study were attributed to chemotherapy; two women died who were allocated treatment every 3 weeks and one death was recorded in the group assigned the weekly regimen. Interpretation A weekly regimen of carboplatin and paclitaxel might be a reasonable option for first-line treatment of women with advanced ovarian cancer. Funding None.

Introduction Ovarian cancer is a frequent cause of cancer-related death in women worldwide. Surgery followed by chemotherapy is the typical treatment approach. The combination of carboplatin plus paclitaxel is considered standard chemotherapy.1 Nevertheless, this regimen is toxic, causing alopecia, neurotoxicity, and fatigue that can affect patients’ quality of life negatively.

Weekly scheduling of paclitaxel has been proposed to reduce toxic effects in patients with malignant disease, by lowering the peak concentration of the drug in plasma, and to increase efficacy, by reducing tumour regrowth between cycles and maximising the antiangiogenic drug effect.2 In patients with ovarian cancer who had previously received chemotherapy, single-agent paclitaxel (80–100 mg/m²) given once a week caused fewer www.thelancet.com/oncology Vol 15 April 2014

Articles

haematological and neurological toxic effects and was active, with responses noted in up to 20% of patients.3–6 Paclitaxel administered once a week has been combined with carboplatin given either every 3 weeks or once a week. In both cases, findings of phase 2 studies show that the weekly regimen is well-tolerated and active.7–11 Based on the above considerations, in 2008, we launched the Multicentre Italian Trials in Ovarian Cancer (MITO-7), European Network of Gynaecological Oncological Trial Groups (ENGOT-OV-10), Gynecologic Cancer InterGroup (GCIG) trial (hereafter called MITO-7), to investigate whether a weekly schedule of carboplatin plus paclitaxel was more effective than the standard regimen administered every 3 weeks as first-line treatment for women with advanced ovarian cancer. The experimental schedule and doses of each drug were selected to compare regimens in an unbiased manner, with roughly the same planned total dose of both drugs in each group.

Methods Study population MITO-7 is an open-label, randomised, phase 3 study based at the National Cancer Institute in Napoli, Italy. We recruited patients from 67 institutions located in Italy and France (appendix pp 5–6). Women were eligible for the study if they: were older than 18 years; had a cytological or histological diagnosis of epithelial ovarian, fallopian tube, or peritoneal cancer; had stage IC–IV disease, according to the International Federation of Gynecology and Obstetrics (FIGO) staging system; had an Eastern Cooperative Oncology Group (ECOG) performance status of 2 or lower; and had a life expectancy of at least 3 months. We excluded patients who: had a history of clinically relevant heart disease or other concurrent disorders that were a contraindication to treatment drugs; had previous or concomitant other malignant disease (except nonmelanoma skin cancer or in-situ carcinoma of the uterine cervix); or had received previous chemotherapy. Further inclusion criteria were adequate bone marrow, kidney, and liver function, assessed with routine haematological and biochemical tests. The ethics committee of every participating institution approved the study, and we obtained written informed consent from every participant before enrolment and randomisation.

Randomisation and masking We used a computer-based minimisation procedure to randomly allocate participants in a 1:1 ratio to either the standard regimen of carboplatin plus paclitaxel every 3 weeks or the experimental schedule of these same drugs once a week. Randomisation was done centrally at the Clinical Trials Unit of the National Cancer Institute in Napoli, Italy. We stratified the random allocation by centre, residual disease after surgery (absent, ≤1 cm, >1 cm, or no primary surgery), and ECOG performance status (0–1 or 2). We enrolled patients either via the Clinical Trials Unit website or by telephone; if clinicians www.thelancet.com/oncology Vol 15 April 2014

contacted the Clinical Trials Unit by telephone, data managers used the same computer-based minimisation procedure that was available through the website. We obtained data through the Clinical Trials Unit website using dedicated electronic case report forms. Patients and clinicians were not masked to the random allocation.

Procedures Randomisation was done after initial surgery and staging of the disease. Women assigned to the treatment regimen administered every 3 weeks received carboplatin (area under the curve [AUC] 6 mg/mL per min, according to the Calvert formula, with creatinine clearance estimated according to the Cockcroft-Gault formula) and paclitaxel (175 mg/m²) intravenously every 21 days for six cycles (appendix p 2). Individuals allocated to the weekly schedule received carboplatin (AUC 2 mg/mL per min) and paclitaxel (60 mg/m²) intravenously every week for 18 weeks consecutively. We diluted carboplatin in 250 mL of 5% glucose and infused it over a 30 min period. We diluted paclitaxel in 250 mL of physiological saline and infused it either over a 3 h period (for patients assigned to treatment every 3 weeks) or over 1 h (for those allocated to the weekly schedule). In both treatment groups, we stopped treatment early in case of progressive disease or unacceptable toxic effects. Because of the different risks of acute toxic effects in each study treatment group, we required more restrictive haematological measurements at every administration of chemotherapy in patients allocated to the regimen every 3 weeks (leucocytes >3000 cells per μL, neutrophils >1500 cells per μL, platelets ≥100 000 cells per μL) than in those assigned to the weekly schedule (leucocytes >3000 cells per μL, neutrophils >1000 cells per μL, platelets ≥75 000 cells per μL); for retreatment in both study treatment groups, we required an absence of organ toxic effects (excluding alopecia) of grade 2 or higher. We allowed treatment discontinuation because of prolonged toxic effects in patients needing a treatment delay of 2 weeks or longer. In both study groups, the doses of all drugs were reduced by 20% if neutrophils fell to fewer than 500 cells per μL or platelets to less than 50 000 cells per μL for 7 days or longer, and we reduced drug doses by 25% if grade 2 neuropathy arose. For baseline staging of patients, we did a clinical examination and imaging studies (chest radiography, abdominopelvic ultrasound, and either abdominopelvic CT or nuclear magnetic resonance [NMR]) and we measured serum CA-125 concentrations. For women whose cancer was not resected completely during the initial surgical intervention, we repeated abdominal staging with CT or NMR after surgery. After three and six cycles of chemotherapy, we repeated the CT or NMR scan and measured response with Response Evaluation Criteria in Solid Tumors (RECIST) version 1.0. In accordance with current clinical practice, French centres did not do radiological restaging after cycle three if clinical suspicion of progression was absent.

Scienze GinecologicoOstetriche e Scienze Urologiche, Università La Sapienza, Roma, Italy (Prof P Benedetti Panici MD, C Marchetti MD); Ginecologia ed Ostetricia, Policlinico, Bari, Italy (G Cormio MD, Prof L Selvaggi MD); Oncologia Medica C, Centro di Riferimento Oncologico, Aviano-PN, Italy (R Sorio MD); Oncologia ed Ematologia, AO Carlo Poma, Mantova, Italy (M G Cavazzini MD); Ginecologia Oncologica, Centro di Ricerca e Formazione ad Alta Tecnologia nelle Scienze Biomediche, Università Cattolica del Sacro Cuore, Campobasso, Italy (G Ferrandina MD); Oncologia Medica Ospedale S Giovanni Calibita Fatebenefratelli, Roma, Italy (E Breda MD); Oncologia Medica Ospedale S Chiara, Trento, Italy (V Murgia MD); Dipartimento di Oncologia AO S Maria della Misericordia, Udine, Italy (C Sacco MD); and Oncologia Medica, Ospedale Antonio Perrino, Brindisi, and Istituto Europeo di Oncologia, Milano, Italy (S Cinieri MD) Correspondence to: Dr Sandro Pignata, Istituto Nazionale per lo Studio e la Cura dei Tumori, Fondazione G Pascale IRCCS, 80131 Napoli, Italy [email protected] See Online for appendix

For the Clinical Trials Unit enrolment site see http://www. usc-intnapoli.net

397

Articles

For patients undergoing interval debulking surgery after three cycles, we could only assess radiological response after three cycles and before debulking. During follow-up, we defined disease progression as the first occurrence of one of the following events: an increase of greater than 20% in the sum of largest diameters of known lesions; appearance of a new lesion; or a rise in the amount of CA-125 in serum of more than 50% compared with the previous value, confirmed in two consecutive assays. We assessed quality of life with Functional Assessment of Cancer Therapy for Ovarian Cancer (FACT-O) version 4 and the neurotoxicity module of this same survey (FACT/GOG-Ntx).12,13 The FACT-O questionnaire has 39 items and six subscales that assess physical (seven items), social (seven items), emotional (six items), and functional (seven items) wellbeing and symptoms specific to ovarian cancer (12 items). We calculated the FACT-O trial outcome index (FACT-O/TOI) score by adding together the scores from the physical and functional wellbeing subscales and the ovarian cancer-specific subscale. The FACT/GOG-Ntx questionnaire has 11 questions for assessment of the presence and severity of chemotherapy-induced peripheral neuropathy, with five response options for every question. Patients completed the FACT-O and FACT/GOG-Ntx questionnaires at baseline before randomisation then every week (same day of the week) for the first 9 weeks of the study, starting 1 week after the first cycle of chemotherapy. We graded

822 underwent random allocation

409 assigned treatment every 3 weeks 5 withdrew consent immediately after randomisation 2 had missing information on treatment 402 started assigned treatment

413 assigned treatment once a week 7 withdrew consent immediately after randomisation 1 had missing information on treatment 1 received treatment every 3 weeks (violation) 3 never started treatment (2 worsening of condition; 1 refused to start chemotherapy) 401 started assigned treatment

223 events for primary endpoint 218 progressed 5 died without documented progression 181 alive and without progression 38 followed up ≤12 months 78 followed up 12–24 months 65 followed up >24 months

226 events for primary endpoint 214 progressed 12 died without documented progression 180 alive and without progression 27 followed up ≤12 months 85 followed up 12–24 months 68 followed up >24 months

404 analysed for progression-free survival (modified intention to treat) 404 analysed for overall survival (modified intention to treat) 400 available for toxic-effect analysis (including 1 patient assigned to weekly treatment but receiving treatment every 3 weeks) 301 available for quality-of-life analysis (baseline questionnaire available)

406 analysed for progression-free survival (modified intention to treat) 406 analysed for overall survival (modified intention to treat) 399 available for toxic-effect analysis 308 available for quality-of-life analysis (baseline questionnaire available)

Figure 1: Trial profile

398

toxic effects according to Common Terminology Criteria for Adverse Events (CTCAE) version 3.0.

Outcomes Initially, quality of life was the only primary endpoint of our study, measured by FACT-O/TOI score. However, in 2010, we added progression-free survival as coprimary endpoint after publication of findings from the Japanese Gynaecological Oncology Group (JGOG) NOVEL trial,14 in which a significant 11-month prolongation of progression-free survival was recorded with paclitaxel given once a week plus carboplatin every 3 weeks, compared with the standard regimen of both drugs every 3 weeks. We defined progression-free survival as the time between randomisation and either progression or death (whichever occurred first) or last follow-up for patients alive at the end of the study without progression. Secondary endpoints included overall survival (defined as the time between randomisation and death or date of last follow-up for patients alive at the end of the study), toxic effects, and the proportion of patients who achieved an objective response (ie, a complete or partial response as assessed by RECIST version 1.0; patients who died or stopped treatment because of toxic effects or who refused before restaging were classified as non-responders).

Statistical analysis We based our study-size calculation initially on the comparison of FACT-O/TOI changes between baseline and week 9 assessments. Assuming an effect size of 0·30 (ie, a difference between arms equal to 30% SD), we needed to enrol 350 patients, with a two-tailed α of 0·05 and power of 80% (EAST version 3.1; Cytel, Cambridge, MA, USA). We planned to enrol 400 participants to offset potentially 15% missing baseline quality-of-life questionnaires. When we added progression-free survival as coprimary endpoint, we readjusted the study size to 810 patients. Furthermore, we needed 382 events to have 80% power of detecting a 0·75 hazard ratio of progression (median progression-free survival from 18 months to 24 months), with a two-tailed α of 0·05 (EAST version 5.4; Cytel) and an interim analysis of efficacy, according to O’Brien-Fleming approach, after half the events had occurred. With the sample size planned on the basis of progression-free survival, statistical power to test the initial quality-of-life hypothesis would be higher than 95%. We did the interim analysis on Feb 6, 2012, when 197 events had been recorded. Stopping rules for the trial were not met (log-rank p=0·424, whereas significance for stopping was p=0·00356). We did efficacy analyses on the basis of modified intention to treat—ie, patients who withdrew consent immediately after randomisation (for whom no follow-up data were available after baseline) were excluded. We calculated median (IQR) follow-up according to the reverse Kaplan-Meier technique.15 We used the Kaplan-Meier product-limit method to estimate www.thelancet.com/oncology Vol 15 April 2014

Articles

progression-free and overall survival, and we compared curves with the log-rank test. For progression-free survival, we used the Cox proportional-hazards model to assess treatment effect, adjusted by FIGO stage, ECOG performance status, residual disease, patient’s age, and size of the study centre. We used the χ² test to measure significance of the difference in the proportion of patients who had an objective response between study treatment groups. We analysed quality of life according to the FACT manual for FACT-O/TOI, FACT-O, the FACT/GOG-Ntx module, and the neurotoxicity subscale. We assessed the effect of each treatment schedule on every quality-of-life scale by fitting a linear mixed-effects model,16 using no within-patient correlation structure among repeated quality-of-life measures. For every patient, as many items were entered in the database as were observed quality-oflife values (up to ten). We included time as a categorical covariate using the week of the observation, from 0 (baseline) to 9 (last assessment); if an intermediate value was missing, the corresponding time was skipped. We entered the following covariates: treatment, FIGO stage, ECOG performance status, residual disease, patient’s age, and size of the study centre. For every quality-of-life scale, the interaction of treatment by time of quality-of-life assessment was first tested; if we judged it significant (p=0·05)—ie, enough evidence was available to show that quality of life was modified in a different way by the two schedules—we did a separate model for each study treatment group to test the difference in mean scores between quality-of-life values at different times and the baseline value. We included in the analysis of toxic effects all patients who received chemotherapy at least once. We calculated the worst toxicity grade for every patient and compared grades for every type of toxic effect with the exact WilcoxonMann-Whitney test. We compared rates of severe events (grade 2 vs grade 0–1 for hair loss; grade ≥2 vs grade 0–1 for neuropathy; grade ≥3 vs grade 0–2 for other adverse events) with the χ² or Fisher’s exact test, as appropriate. We used S-Plus version 6.1 and R software, version 3.0.0 for statistical analyses; exact tests were done with StatXact version 7 (Cytel). This study is registered with ClinicalTrials.gov, number NCT00660842. The protocol is publicly available at the website of the Clinical Trials Unit of the National Cancer Institute of Napoli, Italy.

Role of the funding source There was no funding source for this study. SP, CG, SS, MCP, MDM, and FP had access to raw data, and the corresponding author (SP) had final responsibility for the decision to submit for publication.

excluded from analysis because they withdrew consent immediately after randomisation. Three patients were discovered to be ineligible for the study after randomisation—one woman allocated to treatment every 3 weeks had been pretreated, and two patients assigned to the weekly schedule had the wrong histological diagnosis (rectal cancer and mesothelioma)—but were included in the modified intention-to-treat analysis. 404 women were assigned to treatment every 3 weeks and 406 were allocated to the weekly regimen; baseline characteristics were balanced between groups (table 1). Interval debulking surgery was done in 79 (20%) of 404 patients in the group allocated treatment every 3 weeks and in 71 (17%) of 406 women assigned to the weekly schedule. Information on treatment actually received was incomplete for nine patients. Of 400 women with complete Treatment every 3 weeks (n=404)

Treatment once a week (n=406)

59 (29–83)

60 (23–87)

IC

25 (6%)

31 (8%)

II

33 (8%)

31 (8%)

III

255 (63%)

234 (58%)

IV

91 (23%)

110 (27%)

Age (years) FIGO stage

Primary tumour site Peritoneal Ovarian Fallopian tube

21 (5%)

22 (5%)

366 (91%)

369 (91%)

17 (4%)

15 (4%)

Tumour histology Serous

290 (72%)

274 (67%)

Endometrioid

32 (8%)

49 (12%)

Clear cell

25 (6%)

19 (5%)

Mucinous

7 (2%)

8 (2%)

Undifferentiated

24 (6%)

24 (6%)

Mixed or other

18 (4%)

19 (5%)

8 (2%)

13 (3%)

Missing information Tumour grading G1

14 (3%)

15 (4%)

G2

33 (8%)

45 (11%)

G3

287 (71%)

266 (66%)

70 (17%)

80 (20%)

0

302 (75%)

301 (74%)

1

90 (22%)

93 (23%)

2

12 (3%)

12 (3%)

None

166 (41%)

167 (41%)

≤1 cm

48 (12%)

48 (12%)

>1 cm

91 (23%)

92 (23%)

Not operated

99 (25%)

99 (24%)

Missing information ECOG performance status

Residual disease

Results

Data are median (range) or n (%). FIGO=International Federation of Gynecology and Obstetrics. ECOG=Eastern Cooperative Oncology Group.

Between Nov 20, 2008, and March 1, 2012, 822 patients underwent random allocation (figure 1). 12 patients were

Table 1: Baseline characteristics

www.thelancet.com/oncology Vol 15 April 2014

For the protocol see https://uscintnapoli.net/v2-uosc-servizi/ studidocs/MITO-7-EN/pubdocs/ Mito-7%20protocol_version%20 2%20english.pdf

399

Articles

information who were allocated treatment every 3 weeks, 360 (90%) received all treatment cycles and 17 (4%) stopped treatment because of toxic effects or refusal to continue. At least one delay (attributable to toxic effects or logistic reasons including debulking surgery) was reported in 249 (62%) patients and at least one dose reduction occurred in 142 (36%) individuals. Of 401 patients with complete information in the group assigned to weekly treatment, 334 (83%) received all treatment cycles, 34 (8%) stopped chemotherapy because of toxic effects or they refused to continue, and four (1%) never started treatment. At least one delay was reported in 315 (79%) women in this group, and at least one dose reduction occurred in 76 (19%) patients who received at least one treatment dose. Overall, median relative dose intensity was 82·7% for carboplatin and 86·8% for paclitaxel in women assigned to treatment every 3 weeks, and 76·6% and 79·1% for the two drugs, respectively, in patients allocated treatment on a weekly schedule (appendix pp 2–4). Data for time-to-event analyses were updated to Oct 31, 2013. After median follow-up of 22·3 months (IQR 16·2–30·9), 449 progression-free survival events were recorded, 223 events in 404 women allocated to the standard treatment regimen every 3 weeks (median follow-up A Every 3 weeks Weekly

Proportion without progression (%)

100 80 60 40 20

21·6 months [IQR 15·9–30·4]) and 226 events in 406 patients assigned to the weekly schedule (22·6 months [16·5–31·1]). Estimated median progression-free survival was 17·3 months (95% CI 15·2–20·2) in patients assigned chemotherapy every 3 weeks and 18·3 months (16·8–20·9) in those allocated to treatment once a week (hazard ratio 0·96, 95% CI 0·80–1·16; p=0·66; figure 2). On multivariable analysis, adjusted by FIGO stage, ECOG performance status, residual disease, patient’s age, and size of the institution, the difference between treatments remained unchanged (hazard ratio 0·94, 0·78–1·14; p=0·53); residual disease and FIGO stage were independent predictors of progression-free survival (table 2). Exploratory analysis by subgroups according to FIGO stage, ECOG performance status, patient’s age category, residual disease after surgery, and size of institution showed no heterogeneity of treatment effect (figure 3). 180 deaths were recorded, 82 in the group assigned to treatment every 3 weeks and 98 in the group allocated to chemotherapy once a week. At 24 months the estimated probability of survival was 78·9% (95% CI 74·3–83·8) in patients allocated treatment every 3 weeks and 77·3% (72·8–82·1) in those assigned to chemotherapy once a week (hazard ratio 1·20, 95% CI 0·90–1·61; p=0·22; figure 2). On multivariable analysis, the difference between treatments remained unchanged (1·21, 0·90–1·62; p=0·21). 339 (42%) of 810 participants were eligible for RECIST response analysis: 170 (42%) of 404 patients allocated to treatment every 3 weeks and 169 (42%) of 406 women assigned to the weekly schedule. The proportion of patients who achieved an objective response did not differ between groups (p=0·63): 100 (58·8%, 95% CI 51·0–66·3) women allocated to chemotherapy every 3 weeks had an

p=0·66 0

Number at risk Every 3 weeks 404 Weekly 406

357

240

142

82

39

20

4

1

352

255

151

80

43

20

9

3

B

Proportion surviving (%)

80

0·94 (0·78–1·14)

0·53

FIGO stage (III–IV vs IC–II)

2·71 (1·73–4·24)

<0·0001

40 20 p=0·22 0

6

Number at risk Every 3 weeks 404 383 Weekly 406 377

12 18 24 30 36 Time since randomisation (months)

42

328 323

13 12

231 231

142 140

Figure 2: Kaplan-Meier survival analyses Progression-free survival (A) and overall survival (B).

80 80

43 38

48

2 4

0·068

1·29 (1·03–1·61)

··

2 vs 0

1·38 (0·81–2·34)

··

··

<0·0001

≤1 cm vs absent

2·10 (1·52–2·92)

>1 cm vs absent

2·54 (1·94–3·33)

··

Not operated vs absent

2·84 (2·15–3·75)

··

Age (≥70 years vs <70 years)

0·93 (0·73–1·18)

0·56

Size of the institution

0

··

1 vs 0 Residual after surgery 60

p

Treatment (once a week vs every 3 weeks) ECOG performance status

100

400

Hazard ratio (95% CI)

··

··

0·073

Intermediate (20–89 patients) vs large (≥90 patients)

1·11 (0·88–1·40)

··

Small (<20 patients) vs large (≥90 patients)

0·85 (0·67–1·09)

··

FIGO=International Federation of Gynecology and Obstetrics. ECOG=Eastern Cooperative Oncology Group.

Table 2: Cox proportional-hazards model for progression-free survival

www.thelancet.com/oncology Vol 15 April 2014

Articles

objective response (35 complete responses and 65 partial responses), as did 95 (56·2%, 48·4–63·8) patients assigned to treatment every week (23 complete responses and 72 partial responses). 609 (75%) of 810 patients completed a valid baseline quality-of-life questionnaire, 301 allocated to treatment every 3 weeks and 308 assigned to chemotherapy once a week. Figure 4 presents quality-of-life analyses. Baseline FACT-O/TOI scores were similar between treatment assignments. The median score at baseline in patients allocated to treatment every 3 weeks was 63·0 (range 20·0–93·1), and in those assigned to the weekly schedule it was 64·0 (20·3–100). The treatment-by-time interaction was significant (p<0·0001), implying that FACT-O/TOI was modified in a different way by the two schedules. With treatment every 3 weeks, quality-of-life scores worsened after every chemotherapy cycle (weeks 1, 4, 7), whereas with the weekly schedule, after transient worsening at week 1, scores remained stable. Similar findings were recorded for FACT-O, FACT/GOG-Ntx, and the neurotoxicity subscale. In all analyses, the treatment-by-time interaction favoured chemotherapy every week (p<0·0001). Table 3 shows the number of patients who had a toxic effect, according to the worst grade recorded. Three

deaths were attributed, by the investigators, to chemotherapy: two deaths in women allocated to treatment every 3 weeks were caused by bowel perforation (n=1) and febrile neutropenia (n=1); one death in the group assigned to weekly treatment was due to bowel perforation. Five further deaths—not deemed to be treatment-related—happened while on treatment without sure signs of progressive disease. One death was noted in a patient allocated to treatment every 3 weeks, during the fifth treatment cycle, and the woman had abdominal pain, vomiting, dehydration, and renal failure. Four women died in the group allocated to weekly chemotherapy: one died from an unknown cause during the first cycle; one woman who had pre-existing deep venous thrombosis died from pulmonary embolism during cycle four; one patient who had grade 1 arrhythmia during the first treatment cycle died for unknown reasons during cycle two; and one woman who was enrolled with grade 2 anaemia died from acute myocardial infarction during the first cycle. Compared with patients allocated to chemotherapy every 3 weeks, fewer women assigned to treatment once a week had grade 3–4 neutropenia (200 [50%] of 400 women vs 167 [42%] of 399 women), febrile neutropenia (11 [3%] vs two [0·5%]), grade 3–4 thrombocytopenia (27 [7%] vs

Number of events/patients

Unadjusted hazard ratio (95% CI)

Treatment once a week

Treatment every 3 weeks

Overall (n=810)

226/406

223/404

0·96 (0·80–1·16)

ECOG performance status 0 (n=603) 1 (n=183) 2 (n=24)

151/301 67/93 8/12

156/302 60/90 7/12

0·93 (0·74–1·16) 0·97 (0·68–1·38) 1·07 (0·39–2·97)

Age <70 years (n=659) ≥70 years (n=151)

183/327 43/79

179/332 44/72

1·03 (0·84–1·27) 0·70 (0·46–1·07)

FIGO stage IC–II (n=120) III–IV (n=690)

12/62 214/344

11/58 212/346

0·96 (0·42–2·17) 0·97 (0·80–1·17)

Residual disease Absent (n=333) ≤1 cm (n=96) >1 cm (n=183) No surgery (n=198)

55/167 29/48 65/92 77/99

53/166 30/48 71/91 69/99

1·02 (0·70–1·49) 1·25 (0·74–2·10) 0·72 (0·51–1·01) 0·97 (0·70–1·35)

Size of institution* Large (n=216) Intermediate (n=329) Small (n=265)

66/109 88/162 72/135

66/107 91/167 66/130

0·97 (0·69–1·36) 0·98 (0·73–1·32) 0·93 (0·66–1·29)

0·5

1·0

Favours treatment once a week

1·5

2·0

2·5

Favours treatment every 3 weeks Hazard ratio of progression

Figure 3: Forest plot showing treatment effect on progression-free survival within specific subgroups ECOG=Eastern Cooperative Oncology Group. FIGO=International Federation of Gynecology and Obstetrics. Vertical dotted line represents the hazard ratio (treatment once a week vs treatment every 3 weeks) in the overall study population. *Large institutions had 90 patients or more, intermediate centres had 20–89 patients, and small institutions had fewer than 20 patients.

www.thelancet.com/oncology Vol 15 April 2014

401

Articles

A

B

90

120

Weekly Every 3 weeks

110 Mean FACT-O score

Mean FACT-O/TOI score

80

70 *

60

*

†

†

† 50

100

*

90

†

80 Treatment vs time interaction p<0·0001

Treatment vs time interaction p<0·0001 40

70 0

Number of patients Weekly 308 Every 3 weeks 301

1

2

3

4

5

6

7

8

9

0

1

2

3

4

5

6

7

8

9

266 229

254 208

237 250

239 209

238 195

218 221

212 193

223 177

177 169

307 301

266 226

254 207

236 250

239 209

238 195

218 221

211 193

222 177

177 169

*

*

†

†

†

†

†

†

†

†

†

†

†

†

†

7

8

9

C

D

140 40 130

†

30 120 †

*

*

110

† †

†

*

*

* †

†

†

†

Mean Ntx score

Mean FACT/GOG Ntx score

†

†

20

10

100 Treatment vs time interaction p<0·0001

Treatment vs time interaction p<0·0001

0

90 0

1

2

3

4

5

6

7

8

9

0

1

Time since randomisation (weeks)

2

3

4

5

6

Time since randomisation (weeks)

Number of patients Weekly 298

263

252

232

240

237

217

208

220

175

299

264

252

233

240

238

217

209

222

175

Every 3 weeks 291

221

202

242

208

192

219

190

174

169

291

225

204

243

208

192

222

192

175

169

Figure 4: Quality-of-life analyses for the first 9 weeks (A) Coprimary endpoint analysis of FACT-O/TOI scores. (B–D) Secondary quality-of-life analyses. FACT-O/TOI=functional assessment of cancer therapy for ovarian cancer, trial outcome index. FACT-O=functional assessment of cancer therapy for ovarian cancer. FACT/GOG Ntx=functional assessment of cancer therapy/ Gynecologic Oncology Group, neurotoxicity module. Ntx=neurotoxicity subscale. *p<0·05 vs baseline. †p<0·001 vs baseline.

four [1%]), grade 2 or higher neuropathy (68 [17%] vs 24 [6%]), and grade 2 hair loss (235 [59%] vs 116 [29%]). Considering all grades of toxic effect, the patients allocated to treatment once a week showed a much more favourable pattern for neutropenia, thrombocytopenia, vomiting, hair loss, and neuropathy, but a worse pattern for pulmonary toxic effects.

Discussion Our findings show that administration of carboplatin and paclitaxel once a week, compared with the standard regimen every 3 weeks, augments quality of life and has a better toxicity profile, but does not prolong progressionfree survival (panel). The positive effect on quality of life recorded with administration once a week of carboplatin and paclitaxel might be accounted for simply by the frequency and severity of side-effects. Indeed, toxic effects differed between schedules, with more frequent and 402

severe haematological toxic effects, vomiting, neuropathy, and hair loss noted in women allocated to treatment every 3 weeks. Nevertheless, adherence to planned treatment was slightly better for women assigned to chemotherapy every 3 weeks (appendix p 2), suggesting that despite the weekly schedule being less toxic and better tolerated than the standard scheme, it remains a challenging regimen. Quality-of-life data from the NOVEL trial have been reported,28 and no relevant differences in overall quality of life were noted between the dose-dense scheme (paclitaxel 80 mg/m² once a week plus carboplatin AUC 6 mg/mL per min every 3 weeks) and the standard regimen administered every 3 weeks. However, worse quality-of-life scores were recorded in patients assigned to the dose-dense treatment on the questionnaire dedicated to neurotoxicity.28 It is noteworthy that the quality-of-life assessment in the Japanese study was done at baseline, after three and six cycles, and at 12 months; www.thelancet.com/oncology Vol 15 April 2014

Articles

Treatment every 3 weeks (n=400*), CTCAE grade 1 Anaemia

2

3

4

2

3

0

130 (33%)

102 (26%)

5

0

0

54 (14%)

104 (26%)

68 (17%)

10 (3%)

0

70 (18%)

117 (29%)

50 (13%)

5 (1%)

0

0·48

0·03

Neutropenia

29 (7%)

49 (12%)

104 (26%)

96 (24%)

0

46 (12%)

71 (18%)

141 (35%)

26 (7%)

0

0·003

0·021

Febrile neutropenia

NA

NA

0

1 (<1%)

NA

1 (<1%)

1 (<1%)

0

0·02

0·012

Neutropenic infection

NA

6 (2%)

1 (<1%)

1 (<1%)

0

NA

5 (1%)

3 (1%)

0

0

0·92

0·69

Non-neutropenic infection

NA

11 (3%)

2 (<1%)

0

0

NA

19 (5%)

2 (<1%)

2 (<1%)

0

0·074

0·45

Thrombocytopenia

56 (14%)

41 (10%)

21 (5%)

3 (1%)

1 (<1%)

0

<0·0001

0

0

0

0·051

7 (2%)

0

20 (5%) 0

7 (2%)

0

49 (12%)

0

0

15 (4%)

1 (<1%)

0·34

<0·0001 NA

Allergy

15 (4%)

6 (2%)

14 (4%)

2 (<1%)

0

16 (4%)

21 (5%)

2 (<1%)

0

0·058

Kidney

8 (2%)

2 (<1%)

3 (1%)

3 (1%)

0

12 (3%)

9 (2%)

1 (<1%)

0

0

0·37

Heart, rhythm

4 (1%)

0

0

0

0

7 (2%)

0

0

0

0

0·38

Heart, general

10 (3%)

4 (1%)

4 (1%)

0

0

15 (4%)

7 (2%)

0

0

0

0·59

0·12

7 (2%)

7 (2%)

4 (1%)

0

0

19 (5%)

11 (3%)

3 (1%)

1 (<1%)

0

0·028

1

Fatigue

97 (24%)

75 (19%)

18 (5%)

15 (4%)

0

NA

0·12

0·49

Fever

10 (3%)

9 (2%)

0

0

0·33

0·50

Hair loss

26 (7%)

235 (59%)

NA

NA

<0·0001

<0·0001

Skin

22 (6%)

10 (3%)

1 (<1%)

Anorexia

28 (7%)

7 (2%)

1 (<1%)

Constipation

61 (15%)

27 (7%)

6 (2%)

Pulmonary

Diarrhoea

15 (4%)

0·19

p‡

Leucopenia

NA

25 (6%)

p†

4

85 (21%)

10 (3%)

2 (<1%)

1

116 (29%)

Bleeding

30 (8%)

Treatment once a week (n=399), CTCAE grade 5

0·85 0·12 NA

1 (<1%)

NA

111 (28%)

90 (23%)

0

0

0

19 (5%)

6 (2%)

NA

NA

NA

85 (21%)

116 (29%)

0

0

30 (8%)

16 (4%)

1 (<1%)

1 (<1%)

0

0·071

0·62

0

0

19 (5%)

5 (1%)

2 (<1%)

1 (<1%)

0

0·28

0·37

0

0

87 (22%)

24 (6%)

5 (1%)

0

0

0·12

0·76

1 (<1%) NA

44 (11%)

19 (5%)

4 (1%)

1 (<1%)

0

51 (13%)

19 (5%)

2 (<1%)

0

0

0·76

0·45

101 (25%)

62 (16%)

8 (2%)

0

0

120 (30%)

51 (13%)

6 (2%)

0

0

0·99

0·59

Vomiting

68 (17%)

32 (8%)

5 (1%)

0

0

32 (8%)

23 (6%)

5 (1%)

0

0

0·0002

Mucositis

20 (5%)

6 (2%)

0

0

0

32 (8%)

6 (2%)

0

0

0

0·14

Liver

23 (6%)

1 (<1%)

1 (<1%)

0

0

27 (7%)

9 (2%)

2 (<1%)

0

0

0·066

Nausea

Neurology

16 (4%)

4 (1%)

2 (<1%)

0

0

18 (5%)

8 (2%)

0

0

0

0·53

Neuropathy

95 (24%)

58 (15%)

8 (2%)

2 (<1%)

0

98 (25%)

24 (6%)

0

0

0

0·0002

Toxic death§

NA

NA

2 (<1%)

NA

NA

NA

1 (<1%) NA

NA

NA

NA

1 NA 0·62 0·50 <0·0001 0·62

Data are number of patients (%). CTCAE=Common Terminology Criteria for Adverse Events, version 3.0. NA=not applicable. *Including one patient assigned to treatment once a week who received treatment every 3 weeks. †Any grade (test for trend). ‡Comparison of severe (grade 2 hair loss, grade ≥2 neuropathy, grade ≥3 any other adverse events) versus non-severe toxic effects (χ² or Fisher test, as appropriate). §Treatment-related deaths were bowel perforation (one in each group) and febrile neutropenia (treatment every 3 weeks).

Table 3: Worst grade of toxic effect, according to treatment assignment

therefore, the NOVEL researchers could not catch the differences we noted in the first 9 weeks because we administered questionnaires every week. Although direct comparison between our study findings and those of the NOVEL study is not feasible, our data suggest that the weekly schedule (incorporating a lower dose of paclitaxel) could be preferable in terms of quality of life. We recognise that adoption of a weekly regimen of carboplatin and paclitaxel into clinical practice cannot be based exclusively on quality-of-life findings. However, we believe that patients should be informed of the different pattern of toxic effects seen when chemotherapy is given every 3 weeks versus once a week, and the increased quality-of-life outcomes with weekly scheduling. We acknowledge that a change from administration of www.thelancet.com/oncology Vol 15 April 2014

chemotherapy every 3 weeks to once a week might affect the organisation of gynaecological cancer units negatively, because of the escalation in number of admissions for chemotherapy infusions; an analysis of the economic effect might be needed. However, the possibility to reduce toxic effects of chemotherapy has relevance for antiangiogenic treatments that not only augment efficacy but also increase the burden of side-effects.25,26 Findings of prospective trials should add clarity. The absence of a progression-free and overall survival advantage with the weekly schedule does not accord with findings of the NOVEL study.12,27 Differences between our study and the Japanese trial could account partly for outcome inconsistencies with respect to progressionfree and overall survival. First, even in view of the 403

Articles

Panel: Research in context Systematic review We searched PubMed between 1966 and December, 2013, with the term “ovarian cancer”. We restricted our search to clinical trials only, and to articles published in English. According to current guidelines, the combination of paclitaxel 175 mg/m² and carboplatin AUC 5–6 mg/mL per min, administered intravenously every 3 weeks, represents standard first-line chemotherapy for patients with ovarian cancer.17 This recommendation is based on findings of randomised trials showing that this combination is as effective as cisplatin plus paclitaxel, with a more favourable toxic effect profile.18–20 Substitution of paclitaxel with docetaxel or pegylated liposomal doxorubicin produced similar efficacy in two randomised trials,21,22 whereas addition of a third drug did not show a benefit for progression-free or overall survival.23 Reports indicate that intraperitoneal chemotherapy has better efficacy, but this approach has not been accepted as a common standard of care, in view of greater toxic effects and difficult adherence.24 In two large randomised clinical trials (GOG-218 and ICON-7) of ovarian cancer, addition of bevacizumab to the standard combination of paclitaxel and carboplatin every 3 weeks was assessed as first-line treatment;25,26 in both trials, progression-free survival was prolonged with bevacizumab. Addition of bevacizumab during chemotherapy and as maintenance treatment is currently recommended for patients with advanced ovarian cancer with poor prognostic features, such as FIGO stage IV disease or suboptimum debulking, as defined in the ICON-7 trial. Dose-dense scheduling was tested in the NOVEL study,12,27 in which the standard regimen of paclitaxel plus carboplatin every 3 weeks was compared with paclitaxel 80 mg/m² every week plus carboplatin AUC 6 mg/mL per min every 3 weeks, with a relevant benefit for the dose-dense schedule in terms of progression-free and overall survival. These results have been judged potentially practice-changing, but they need to be confirmed in populations from other ethnic backgrounds.17 Apart from our study, two other trials testing a weekly schedule are in progress—GOG-262 (NCT01167712) and ICON-8 (NCT01654146). Interpretation In our study, a weekly regimen of carboplatin plus paclitaxel (compared with the standard 3-week schedule) did not prolong progression-free survival, but was associated with better quality of life and a lower toxicity profile. On the basis of the data we present, while waiting for the results of GOG-262 and ICON-8, a weekly schedule might be considered among the possible first-line treatment options for advanced ovarian cancer in clinical practice.

limitations of indirect study comparisons, the frequency and severity of toxic effects were strikingly different between the two studies. In patients allocated to chemotherapy every 3 weeks (standard treatment) in the Japanese trial, grade 3–4 neutropenia was noted in 88%, thrombocytopenia in 38%, and anaemia in 44% of patients, versus 50%, 7%, and 8%, respectively, in our study. Such large differences indicate possible genetic causes for drug sensitivity. Polymorphisms in genes that play a part in DNA repair or drug metabolism can lead to differences in the toxic effects of a drug. Variations in paclitaxel metabolism between European and Asian patients have been attributed to genetic polymorphisms.29 Findings of the OCTAVIA trial,30 in which carboplatin and paclitaxel given once a week (plus bevacizumab) were tolerated much better in European patients than in those from Asian populations, lend support to this hypothesis. Whether genetic polymorphisms affect the efficacy of carboplatin and paclitaxel in ovarian cancer is 404

not known, and why and whether an effect might vary according to the dosing schedule is not clear. However, progression-free survival for patients allocated to chemotherapy every 3 weeks was very similar in our study to that reported in the NOVEL trial, whereas it differed clearly between studies for patients assigned to the weekly schedule. Second, the experimental schedules in our study and in the NOVEL trial varied according to both the dose of paclitaxel given and the weekly administration of carboplatin. When planning our study, we chose doses of drugs for the weekly regimen that would maintain the same total dose of drugs administered every 3 weeks in the standard scheme, to produce an unbiased estimation of the scheduling effect. Therefore, the dose of paclitaxel for patients assigned to the weekly schedule was 60 mg/m² in our study, which was a rounding up of a third of the 175 mg/m² given every 3 weeks in the standard regimen. This dose, however, is 25% lower than the 80 mg/m² paclitaxel dose used in the Japanese study. Weekly scheduling might possibly not be enough to prolong progression-free survival, and an increase of individual paclitaxel doses (dose-dense treatment) could be needed to improve efficacy. However, we remain unsure about this hypothesis, because findings of a previous study,31 in which two different doses of paclitaxel (175 mg/m² vs 225 mg/m²) were compared, did not show any dose–effect relation, but more toxic effects were noted with the higher dose. Two ongoing trials—GOG-262 (NCT01167712) and ICON-8 (NCT01654146)—aim to address the hypothesis that dosedense weekly paclitaxel is effective, but the design of both studies does not include an unconfounded arm with weekly—rather than dose-dense—paclitaxel. Preliminary data of GOG-262 showed progression-free survival did not differ between dose-dense weekly treatment and the standard regimen given every 3 weeks (most patients also received bevacizumab).32 With respect to carboplatin given once a week, this schedule might contribute substantially to a reduction in toxic effects and to quality of life, an idea supported by indirect comparison of our data with those of the quality-of-life neurotoxicity scale in the NOVEL trial.28 However, we cannot rule out that carboplatin given every 3 weeks might be more effective for progression-free survival than carboplatin administered every week, and we cannot exclude that weekly carboplatin might antagonise the effect of paclitaxel, as suggested by findings of an in-vitro study.33 The ICON-8 trial aims to compare the standard carboplatin and paclitaxel regimen given every 3 weeks with both a full weekly schedule (similar to our study) and a mixed regimen (similar to the NOVEL study), and these findings will add useful information. In conclusion, carboplatin and paclitaxel given once a week compared with the standard regimen administered every 3 weeks did not significantly prolong progressionfree survival in women with ovarian cancer, but it was associated with better quality of life and fewer toxic effects. On the basis of these data, we believe that a www.thelancet.com/oncology Vol 15 April 2014

Articles

weekly schedule might be a reasonable alternative for first-line treatment of advanced ovarian cancer in clinical practice. Contributors SP, CG, and FP designed the study. SP was the principal investigator. SP, GS, DK, EP-L, SDP, AB, BW, FR, PBP, GC, RS, MGC, GF, EB, VM, CS, SC, VS, CR, CP, SG, RL, DL, CM, and LS enrolled and treated patients and gathered data. MCP, MDM, and FP gathered data. CG, SS, MDM, and FP did statistical analyses. All authors interpreted data, wrote the report, and approved the final version. Declaration of interests All authors declare that they have no competing interests. Acknowledgments MITO-7 is an academic, non-profit trial promoted by the National Cancer Institute of Napoli (NCI-NAP), Italy. The institutions directed by SP and FP at NCI-NAP are supported by the non-profit charity Associazione Italiana per la Ricerca sul Cancro (AIRC). References 1 Thigpen T, duBois A, McAlpine J, et al, and the Gynecologic Cancer InterGroup. First-line therapy in ovarian cancer trials. Int J Gynecol Cancer 2011; 21: 756–62. 2 Saito K, Kikuchi Y, Fujii K, Kita T, Furuya K. Effect of paclitaxel on vascular endothelial growth factor (VEGF) and interleukin (IL)-8 in serum of patients with recurrent ovarian cancer: a comparison of weekly vs triweekly regimens. Target Oncol 2006; 1: 86–89. 3 Fennelly D, Aghajanian C, Shapiro F, et al. Phase I and pharmacologic study of paclitaxel administered weekly in patients with relapsed ovarian cancer. J Clin Oncol 1997; 15: 187–92. 4 Markman M, Blessing J, Rubin SC, Connor J, Hanjani P, Waggoner S, and the Gynecologic Oncology Group. Phase II trial of weekly paclitaxel (80 mg/m²) in platinum and paclitaxel-resistant ovarian and primary peritoneal cancers: a Gynecologic Oncology Group study. Gynecol Oncol 2006; 101: 436–40. 5 Klaassen U, Wilke H, Strumberg D, Eberhardt W, Korn M, Seeber S. Phase I study with a weekly 1 h infusion of paclitaxel in heavily pretreated patients with metastatic breast and ovarian cancer. Eur J Cancer 1996; 32A: 547–49. 6 Markman M, Hall J, Spitz D, et al. Phase II trial of weekly single-agent paclitaxel in platinum/paclitaxel-refractory ovarian cancer. J Clin Oncol 2002; 20: 2365–69. 7 Kikuchi A, Sakamoto H, Yamamoto T. Weekly carboplatin and paclitaxel is safe, active, and well tolerated in recurrent ovarian cancer cases of Japanese women previously treated with cisplatin-containing multidrug chemotherapy. Int J Gynecol Cancer 2005; 15: 45–49. 8 Rose PG, Smrekar M, Fusco N. A phase II trial of weekly paclitaxel and every 3 weeks of carboplatin in potentially platinum-sensitive ovarian and peritoneal carcinoma. Gynecol Oncol 2005; 96: 296–300. 9 Sehouli J, Stengel D, Mustea A, et al, and the Ovarian Cancer Study Group of the Nord-Ostdeutsche Gesellschaft für Gynäkologische Onkologie. Weekly paclitaxel and carboplatin (PC-W) for patients with primary advanced ovarian cancer: results of a multicenter phase-II study of the NOGGO. Cancer Chemother Pharmacol 2008; 61: 243–50. 10 van der Burg MEL, Boere IA, Berns PMJJ. Dose-dense therapy is of benefit in primary treatment of ovarian cancer: contra. Ann Oncol 2011; 22 (suppl 8): viii33–39. 11 Watanabe Y, Nakai H, Ueda H, Hoshiai H. Evaluation of weekly lowdose paclitaxel and carboplatin treatment for patients with platinumsensitive relapsed ovarian cancer. Gynecol Oncol 2005; 96: 323–29. 12 Basen-Engquist K, Bodurka-Bevers D, Fitzgerald MA, et al. Reliability and validity of the functional assessment of cancer therapy-ovarian. J Clin Oncol 2001; 19: 1809–17. 13 Huang HQ, Brady MF, Cella D, Fleming G. Validation and reduction of FACT/GOG-Ntx subscale for platinum/ paclitaxel-induced neurologic symptoms: a gynecologic oncology group study. Int J Gynecol Cancer 2007; 17: 387–93. 14 Katsumata N, Yasuda M, Takahashi F, et al, for the Japanese Gynecologic Oncology Group. Dose-dense paclitaxel once a week in combination with carboplatin every 3 weeks for advanced ovarian cancer: a phase 3, open-label, randomised controlled trial. Lancet 2009; 374: 1331–38.

www.thelancet.com/oncology Vol 15 April 2014

15 16 17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials 1996; 17: 343–46. Pinheiro JC, Bates DM. Mixed-effects models in S and S-Plus. New York: Springer, 2000. Colombo N, Peiretti M, Garbi A, Carinelli S, Marini C, Sessa C, and the ESMO Guidelines Working Group. Non-epithelial ovarian cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol 2012; 23 (suppl 7): vii20–26. du Bois A, Lück HJ, Meier W, et al, and the Arbeitsgemeinschaft Gynäkologische Onkologie Ovarian Cancer Study Group. A randomized clinical trial of cisplatin/paclitaxel versus carboplatin/paclitaxel as first-line treatment of ovarian cancer. J Natl Cancer Inst 2003; 95: 1320–29. Neijt JP, Engelholm SA, Tuxen MK, 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. Ozols RF, Bundy BN, Greer BE, et al, and the Gynecologic Oncology Group. Phase III trial of carboplatin and paclitaxel compared with cisplatin and paclitaxel in patients with optimally resected stage III ovarian cancer: a Gynecologic Oncology Group study. J Clin Oncol 2003; 21: 3194–200. Pignata S, Scambia G, Ferrandina G, et al. Carboplatin plus paclitaxel versus carboplatin plus pegylated liposomal doxorubicin as first-line treatment for patients with ovarian cancer: the MITO-2 randomized phase III trial. J Clin Oncol 2011; 29: 3628–35. Vasey PA, Jayson GC, Gordon A, et al, and the Scottish Gynaecological Cancer Trials Group. Phase III randomized trial of docetaxel-carboplatin versus paclitaxel-carboplatin as first-line chemotherapy for ovarian carcinoma. J Natl Cancer Inst 2004; 96: 1682–91. Bookman MA, Brady MF, McGuire WP, 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 2009; 27: 1419–25. Armstrong DK, Bundy B, Wenzel L, et al, and the Gynecologic Oncology Group. Intraperitoneal cisplatin and paclitaxel in ovarian cancer. N Engl J Med 2006; 354: 34–43. Burger RA, Brady MF, Bookman MA, et al, and the Gynecologic Oncology Group. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011; 365: 2473–83. Perren TJ, Swart AM, Pfisterer J, et al, and the ICON7 Investigators. A phase 3 trial of bevacizumab in ovarian cancer. N Engl J Med 2011; 365: 2484–96. Katsumata N, Yasuda M, Isonishi S, et al, and the Japanese Gynecologic Oncology Group. Long-term results of dose-dense paclitaxel and carboplatin versus conventional paclitaxel and carboplatin for treatment of advanced epithelial ovarian, fallopian tube, or primary peritoneal cancer (JGOG 3016): a randomised, controlled, open-label trial. Lancet Oncol 2013; 14: 1020–26. Harano K, Terauchi F, Katsumata N, et al. Quality-of-life outcomes from a randomized phase III trial of dose-dense weekly paclitaxel and carboplatin compared with conventional paclitaxel and carboplatin as a first-line treatment for stage II–IV ovarian cancer: Japanese Gynecologic Oncology Group Trial (JGOG3016). Ann Oncol 2014; 25: 251–57. Ma BBY, Hui EP, Mok TSK. Population-based differences in treatment outcome following anticancer drug therapies. Lancet Oncol 2010; 11: 75–84. Gonzalez-Martin A, Gladieff L, Tholander B, et al, and the OCTAVIA Investigators. Efficacy and safety results from OCTAVIA, a single-arm phase II study evaluating front-line bevacizumab, carboplatin and weekly paclitaxel for ovarian cancer. Eur J Cancer 2013; 49: 3831–38. Bolis G, Scarfone G, Polverino G, et al. Paclitaxel 175 or 225 mg per meters squared with carboplatin in advanced ovarian cancer: a randomized trial. J Clin Oncol 2004; 22: 686–90. Chan J, Brady M, Penson R, et al. Phase III trial of every-3-weeks paclitaxel vs. dose dense weekly paclitaxel with carboplatin +/bevacizumab in epithelial ovarian, peritoneal, fallopian tube cancer: GOG 262 (NCT01167712). Int J Gynecol Cancer 2013; 23 (suppl 1): 9. Xiong X, Sui M, Fan W, Kraft AS. Cell cycle dependent antagonistic interactions between paclitaxel and carboplatin in combination therapy. Cancer Biol Ther 2007; 6: 1067–73.

405