Original article. Introduction

Transcription

1 Original article Annals of Oncology 16: 81 89, 2005 doi: /annonc/mdi013 Sequential two-line strategy for stage IV non-small-cell lung cancer: docetaxel cisplatin versus vinorelbine cisplatin followed by cross-over to single-agent docetaxel or vinorelbine at progression: final results of a randomised phase II study J.-Y. Douillard 1 *, R. Gervais 2, G. Dabouis 3, A. Le Groumellec 4, M. D Arlhac 5, D. Spaeth 6, B. Coudert 7, D. Caillaud 8, A. Monnier 9, C. Clary 10, B. Maury 11, M. Mornet 12, A. Rivière 2, P. Clouet 13 & C. Couteau 1 1 Centre René Gauducheau, Saint Herblain; 2 Centre François Baclesse, Caen; 3 CHU Hotel Dieu, Nantes; 4 CHG Chubert, Vannes; 5 CHG Georges Renon, Niort; 6 Centre Alexis Vautrin, Vandoeuvre-les-Nancy; 7 Centre GF Leclerc, Dijon; 8 CHU Gabriel Montpied, Clermont-Ferrand; 9 CHG André Boulloche, Montbeliard; 10 CHU Pasteur, Nice; 11 CHG, Cholet; 12 CH J Cœur, Bourges; 13 Laboratoire Aventis, Paris, France Received 8 April 2004; revised 14 July 2004; accepted 9 August 2004 Background: This phase II trial compared docetaxel cisplatin (DC) with vinorelbine cisplatin (VC), both as first-line therapy followed by cross-over at progression to single-agent vinorelbine or docetaxel in advanced non-small-cell lung cancer (NSCLC). Methods: Overall, 115 patients received DC (docetaxel 75 mg/m 2 and cisplatin 100 mg/m 2 both on day 1, every 3 weeks, arm A1) and 118 VC (vinorelbine 30 mg/m 2 /week on days 1 and 8 and cisplatin 100 mg/m 2 on day 1, every 3 weeks, arm B1) for six cycles, and subsequently maintained by monotherapy with docetaxel (A1) or vinorelbine (B1) with cross-over on disease progression to vinorelbine 30 mg/m 2 days 1 and 8 (A2), or docetaxel 100 mg/m 2, day 1, both every 3 weeks (B2). The primary end point was overall response rate (ORR). Results: Patient characteristics were balanced; median follow-up was 8.8 months. First-line response rate was 33.9% with DC and 26.3% with VC (P = 0.20). In arms A1 and B1, respectively: duration of response was similar (8.2 versus 8.4 months); median time to progression was 5 months in both; median survival was 8 versus 9 months (P = 0.38); 1-, 2- and 3-year survival was 36% versus 35%, 17% versus 10% and 13% versus 6% (P not significant). However, with a low number of long-term survivors, statistical significance was not reached. Overall, almost half of the patients crossed over to second-line therapy; there were no response with vinorelbine and 6 (11.2%) partial responses with docetaxel. Considering the safety profile, the occurrence of febrile neutropenia was 9.6% with DC and 26.3% with VC. Treatment-related mortality was 2.5% with DC and 8.5% with VC. Conclusions: The trend in favour of the DC arm in ORR, even though statistical significance was not reached, is consistent with previous reports. This study suggests an activity of first-line DC in advanced NSCLC, and that second-line vinorelbine does not provide additional clinical benefit. As already shown in other studies, the use of DC in first-line should provide a better percentage of long-term survivors, despite the absence of efficacy of the second-line in our study. Key words: chemotherapy, docetaxel, first-line, non-small-cell lung cancer Introduction Lung cancer is a major worldwide health problem. Despite the development of new anticancer drugs, the overall benefit for patients with advanced, recurrent or metastatic non-small-cell *Correspondence to: Professor J.-Y. Douillard, Centre René Gauducheau, Boulevard J Monod, Saint Herblain, France. Tel: ; Fax: ; jy-douillard@nantes.fnclcc.fr lung cancer (NSCLC) remains modest [1]. A meta-analysis [2] of 52 randomised trials showed that cisplatin-based regimens provide survival benefits compared with best supportive care (BSC), a finding confirmed by additional studies [3]; therefore, the use of platinum-based doublets has become the first-line standard of care in advanced NSCLC. The impact of secondline chemotherapy is also beneficial compared with BSC [4], but its role in sequential regimens has not yet been established. Docetaxel and vinorelbine are approved in Europe q 2005 European Society for Medical Oncology

2 82 and the USA for use in first-line treatment of NSCLC. Docetaxel showed a demonstrable survival advantage in secondline therapy compared with BSC [4], vinorelbine and ifosfamide [5], leading to its establishment as the reference standard for second-line treatment of patients with advanced NSCLC. This activity of docetaxel led to numerous phase II studies of the drug as first-line treatment. A pooled analysis of the four main trials with docetaxel (100 mg/m 2 ) showed a 31% response rate, a median survival of 9.2 months and a 1-year survival rate of 37% [6]. In avanced NSCLC, standard firstline chemotherapy includes platinum combination. The recent report by Fossella et al. [7] has shown that the combinations of docetaxel cisplatin (DC) and vinorelbine cisplatin (VC) are active in the first-line setting. The study provided the results of a comparative phase III study involving 1200 patients, which showed that DC provided a greater benefit in terms of response, overall survival and quality of life compared with VC in the first-line setting [7]. This new setting in first-line chemotherapy for NSCLC will be responsible for more acute needs for the second-line setting therefore accessible to an increased population. However, in this phase III study, according to the protocol, second-line therapy was at the investigator s discretion. Our randomised, phase II study was designed to evaluate the activity of DC versus VC as first-line therapy in advanced NSCLC, and is the first to look at the effect of cross-over to single-agent docetaxel or vinorelbine at disease progression in NSCLC. Patients and methods Patient selection All patients enrolled in this study had histologically or cytologically confirmed stage IV NSCLC (squamous cell, large cell, adenocarcinoma or undifferentiated NSCLC). Patients were required to have at least one measurable or assessable lesion outside irradiated fields, i.e. cutaneous or lymph node >_ mm assessed by clinical measurement; limited pulmonary nodule >_ mm detected by standard chest X-ray or >_ mm using computed tomography (CT) scan; others lesions >_ mm at CT scan. Other inclusion criteria included: age years; World Health Organisation (WHO) performance status <_ 2; and adequate bone marrow (neutrophil count >_ /l, platelet count >_ /l), renal and hepatic functions [creatinine <_ 140 mmol/l, total bilirubin <_ 1.5 upper limit of normal (ULN), transaminases <_ 2.5 ULN, alkaline phosphatases <_ 5 ULN except for isolated bone metastases]. Previous radiotherapy was allowed if it involved <25% of bone marrow and was completed 4 weeks before study entry. Previously irradiated or clinically asymptomatic brain metastases and any weight loss during the last 6 months were admitted. Exclusion criteria were: stages IIIB (including wet T4); National Cancer Institute Common Toxicity Criteria (NCI CTC) peripheral neuropathy grade >1; prior chemotherapy or biological therapy for metastases; lymphangitis carcinomatosa, ascites or pleural effusion as the only target. This trial was approved by a local ethics committee and all patients provided written informed consent. Treatment administration Patients were randomised to one of two treatment arms: A (A1 + A2) and B (B1 + B2). Patients in arm A received as first-line (A1): docetaxel 75 mg/m 2 as a 1-h intravenous (i.v.) infusion followed by cisplatin 100 mg/m 2 as a 1-h infusion on day 1. Patients randomised to arm B received as first-line (B1): cisplatin 100 mg/m 2 as a 1-h infusion on day 1 followed by vinorelbine 30 mg/m 2 as a 15-min infusion on days 1 and 8. For both arms, cycles were administered every 3 weeks for up to six cycles. In the event of a complete response (CR), treatment was stopped until progressive disease (PD) occurred. When a partial response (PR) or stable disease (SD) was reported, cisplatin was stopped after six cycles and chemotherapy was continued with docetaxel in arm A1 and vinorelbine in arm B1 at the same dose until PD, unacceptable toxicity or patient s request to halt treatment. When PD had been documented, patients crossed over to the alternative monotherapy as second-line treatment: (A2) vinorelbine 30 mg/m 2 on days 1 and 8, or (B2) docetaxel 100 mg/m 2 every 3 weeks until PD (Figure 1). A docetaxel dose of 100 mg/m 2 in second-line was chosen at the time of the trial conception, when data from Shepherd et al. [4] and Fossella et al. [5] were not available. Before docetaxel administration, patients received steroid prophylaxis with 8 mg of oral dexamethasone, or equivalent, every 12 h for five consecutive doses starting 1 day before each docetaxel infusion. Cisplatincontaining regimens were administered on an inpatient basis. Antiemetics were given at the investigator s discretion. Granulocyte colony-stimulating factor support was not given prophylactically, except after febrile neutropenia. Figure 1. Trial profile.

3 83 Dose adjustment During treatment, patients underwent weekly complete blood cell count, white blood cell differential and platelet counts. Before treatment, patients were required to have an absolute neutrophil count of >_ /l and a platelet count >_ /l. Dose-adjustment criteria were based mainly on haematological measurements or clinical assessments, and were maintained for all subsequent cycles of treatment. The minimal dose reductions were 60 mg/m 2 (docetaxel), 80 mg/m 2 (cisplatin) and 25 mg/m 2 (vinorelbine). Docetaxel doses were reduced by 20% in cases of liver transaminase levels between 2.5 and 5 ULN values, and were interrupted in cases of transaminases >5 ULN and abnormal bilirubin. If patients experienced several toxicities, the most conservative dose adjustment was adopted. If a cisplatin-related toxicity occurred, either docetaxel or vinorelbine was continued as monotherapy until PD was assessed. In the case of NCI CTC (version 1) grade 3 toxicities (except alopecia, anaemia and asthenia), study treatment could be delayed for up to 2 weeks to allow recovery until grade <_ 1 levels were monitored. Study assessments Pretreatment assessments included a complete medical history and physical examination; complete blood cell count with differential and platelets, standard biochemical profile, aspartate aminotransaminase, alanine aminotransferase, total bilirubin, creatinine, albumin and a 1 -acid glycoprotein; electrocardiogram; chest X-ray; CT scan of the chest, abdomen and brain; and radionuclide total body scan. Tumour assessments were performed every three cycles. All lesions were followed and all measurable disease recorded. However, in cases of multiple lesions, up to four measurable target lesions representative of all organs involved were selected, giving the priority to bidimensionally measurable lesions. Response had to be confirmed at a minimum of 4-weekly interval. Radiological documentation of first-line treatment was reviewed by an external review committee to assess the best response and date of possible progression. Criteria of response were assessed according to WHO response criteria. Toxicity was evaluated according to the NCI CTC (version 1) when possible. Otherwise, the severity was assessed and graded as mild, moderate, severe or life-threatening. Statistical analysis The primary end point was the overall response rate (ORR) after first-line treatment as evaluated by the external review committee. Secondary objectives for both first- and second-line treatment were duration of response, time to progression (TTP), survival, toxicity and quality of life. The sample size estimation was based on the assumption that the response rate should be >_ 35% in each arm of the treated population after first-line treatment. With a precision of ±10% of the 95% confidence interval (CI), 88 patients per arm were required. For the analysis of secondary end points, 100 patients per arm were included. Statistical tests were only allowed for exploratory purposes. Analyses were performed using SAS software (version 6.12; SAS Institute, Cary, NC, USA) on all patients who had received at least one infusion (evaluable population). Median, minimum and maximum values were used to summarise continuous data. Time-dependent parameters were analysed using the Kaplan Meier method. All patients who received at least three cycles of treatment were assessable for response. Patients who received less than three cycles were assessable for toxicity and for response if clinical progression occurred >8 days after the first treatment infusion. Patients who had PD with combination therapy were considered as refractory to platinum. Results Patients characteristics From May 1998 to March 2000, 239 patients were randomised in 23 centres in France: 115 in arm A1 and 118 in arm B1 (Figure 1). Six patients were not treated and were excluded from the analysis. Fourteen patients were ineligible for analysis (5.8%) for the following reasons: radiotherapy stopped <2 weeks before entry (n = 4), liver dysfunction tests (n = 5), symptomatic brain metastases (n = 1), neuropathy grade >1 (n = 1), stage IIIA disease (n = 1), previous metastatic prostatic carcinoma (n = 1) and participation in another trial (n = 1). Patients characteristics are summarised in Table 1, and were well balanced between the two arms. The majority of patients were male and the most frequent histological subtype was adenocarcinoma. The most frequent metastatic site was the lung, and over a half of the patients had lymph node involvement. Local relapse was low and similar between the two groups. Table 1. Patients characteristics Characteristics Docetaxel cisplatin (n = 115) [n (%)] Vinorelbine cisplatin (n = 118) [n (%)] Age, years [median (range)] 58 (27 75) 57 (27 77) Sex Male 96 (83.5) 95 (80.5) Female 19 (16.5) 23 (19.5) Performance status 0 34 (29.6) 39 (33.1) 1 63 (54.8) 62 (52.5) 2 18 (15.6) 17 (14.4) Weight loss >5 kg 35 (30.4) 30 (25.4) within the last 3 months Histological subtype Adenocarcinoma 47 (40.9) 55 (46.6) Squamous cell carcinoma 38 (33.0) 38 (32.2) Large cell 12 (10.5) 16 (13.6) Undifferentiated lung 15 (13.0) 8 (6.8) cancer NSC Mixed histological subtype 3 (2.6) 1 (0.8) Organ involvement Lung 105 (91.3) 108 (91.5) Lymph node 57 (49.6) 61 (51.7) Bone 41 (35.7) 47 (39.8) Liver 24 (20.9) 28 (23.7) Brain 18 (15.7) 25 (21.2) Adrenal glands 20 (17.4) 19 (16.1) Prior treatment Thoracic surgery 19 (16.5) 16 (13.6) Radiotherapy 10 (8.7) 11 (9.3) NSC, non-small cancer.

4 84 Figure 1 shows that 103 patients (44%) received cross-over second-line therapy: 51 with vinorelbine in arm A2 and 52 with docetaxel in arm B2; three received second-line therapy but were not eligible due to the absence of PD after first-line. Over half of all patients in arms A1 and B1 did not receive second-line treatment, mainly because of deaths [18 (28.1%) and 24 (36.6%), respectively], change in general status and treatment outside the study, including radiotherapy after firstline therapy [28 patients (43.7%) and 34 (51.5%), respectively]. In arm A1, at the date of analysis, six patients had no PD after first-line therapy. Efficacy Response rate was 33.9% with DC (A1) and 26.3% with VC (B1) (Table 2) (P = 0.203). In A1, there was one (0.9%) CR and 38 (33.0%) PRs. In B1, there was no CR but 31 (26.3%) PRs. SD was similar in both arms: 34 (29.6%) in A1 and 41 (34.7%) in B1, and PD occurred in over one-quarter of patients in each arm, 29 (25.2%) in A1 and 26 (22%) in B1. Overall response rates were similar between the two arms. In descriptive histological subtypes, among patients with squamous cell carcinoma (38 each arm), 12 in A1 and seven in B1 had a PR. Among patients with large cell histological subtype (12 in A1 and 16 in B1), PR occurred in 6 and 1 patient, respectively. There was no difference between treatments in terms of median duration of response: 8.2 months (95% CI ) in A1 and 8.4 months (95% CI ) in B1. The median TTP was 5 months in both arms. In second-line therapy, no response was observed in patients treated with vinorelbine (A2); nine (17.6%) patients Table 2. Response rate by treatment group in first- and second-line Tumour response First-line treatment [n (%)] A1: Docetaxel cisplatin (n = 115) B1: Vinorelbine cisplatin (n = 118) Complete response 1 (0.9) 0 (0) Partial response 38 (33.0) 31 (26.3) Overall response rate 39* (33.9) 31* (26.3) 95% CI Stable disease 34 (29.6) 41 (34.7) Progressive disease 29 (25.2) 26 (22.0) Non-assessable 13 (11.3) 20 (16.9) Second-line treatment [n (%)] A2: Vinorelbine (n = 51) B2: Docetaxel (n = 52) Partial response 0 (0) 6 (11.5) Overall response rate 0 (0) 6 (11.5) 95% CI Stable disease 9 (17.6) 8 (15.4) Progressive disease 31 (60.8) 35 (67.3) Non-assessable 11 (21.6) 3 (5.8) *P = had SD and 31 (60.8%) PD. With docetaxel (B2), six (11.5%) patients had PR, eight (15.4%) had SD and PD occurred in 35 (67.3%) patients. Statistical difference was reached between the ORR for arm A2 and B2 (P = 0.038). Among patients on second-line therapy, a total of 27 (26.2%) were considered refractory to cisplatin: 12 (23.5%) in A2 and 15 (28.8%) in B2. Only one response in arm B2 was observed among these 27 patients. Median TTP was again similar in both arms at 2 months. Median survival in the intentionto-treat population was 8.3 and 9 months in arms A and B, respectively, at a median follow-up of 8.8 months (range ) (Figure 2). For arms A (first-line DC) and B (first-line VC), the 1-year survival rate was 37% and 36%, respectively. At 2 years, survival rate was 17% versus 10%, and at 3 years, 13% versus 6%, respectively. However, with a low number of long-term survivors, statistical significance was not reached. The median survival for all patients who received first- and second-line treatment was 9.9 (A1 + A2) and 11.4 (B1 + B2) months, respectively, considering that only half of the firstline population received the cross-over second-line therapy. Treatment exposure A total of 544 and 519 combination cycles were given as first-line in arms A1 and B1, respectively (Table 3), with a median of 6 (range 1 to 6) cycles of combination therapy in both arms. In A1, the median dose intensity of DC was >_ 98% of planned dose for both drugs, whereas for VC, it was 88% (vinorelbine) to 90% (cisplatin) of planned dose. The majority of combination therapy cycles were administered on time in A1 (93%) and B1 (73.4%). In A1, 38 cycles (7%) were delayed because of haematological (14 cycles), non-haematological toxicities (14 cycles) and for other reasons unrelated to treatment (10 cycles). In B1, 138 (26.6%) cycles were delayed for haematological (92 cycles, 66.7%) and non-haematological toxicities (27 cycles, 19.6%), and one cycle for both types of toxicity and 18 (13%) for other reasons. During combination therapy, the dose of docetaxel was reduced in 13 cycles (2.4%) and in 17 (3.1%) cycles for cisplatin. In B1, vinorelbine was reduced in 34 cycles (6.6%) and cisplatin was reduced in 53 (10.2%) cycles. Among the 233 patients who received first-line combination therapy, 55 continued with monotherapy: 34 in A1 (143 cycles) and 21 in B1 (90 cycles), with a similar median number of cycles in both arms (Table 3). Docetaxel was administered at the planned dose in 90.2% of cycles, compared with 73.3% for vinorelbine. For docetaxel, 4.9% of cycles were delayed compared with 12.2% of cycles in which vinorelbine was reduced. During second-line treatment, the median number of cycles administered was similar between both arms (three; range one to 18). However, more cycles were given in docetaxel compared with vinorelbine (Table 3), with only 12% of patients receiving at least 6 cycles of vinorelbine in A2 compared with 27% who received docetaxel in B2. As seen with first-line treatment, more cycles of vinorelbine were delayed for patients in arm A2 [24 (15.8%)] than with docetaxel in B2 [12 (5.6%)].

5 85 Figure 2. Overall survival: Kaplan Meier estimate. ITT, intention-to-treat. Table 3. Treatment exposure during first- and second-line therapy Cycles of treatment Arm A1 Arm B1 Docetaxel cisplatin (n = 115) Docetaxel (n = 34) Vinorelbine cisplatin (n = 118) Vinorelbine (n = 21) Total cycles administered (n) Cycles per patient [median (range)] 6 (1 6) 3 (1 19) 6 (1 6) 3 (1 15) No. full combination cycles [n (%)] 535 (98.3) 491 (94.6) Median relative dose intensity (per patient) a 1.0, , Range , , Median dose intensity (mg/m 2 /week) a 24.5, , Median cumulative dose (mg/m 2 ) a 427.4, , Cycles delayed (%) Arm A2: Vinorelbine (n = 51) Arm B2: Docetaxel (n = 52) Total cycles administered (n) Cycles per patient [median (range)] 3 (1 14) 3 (1 18) Cycles delayed (% of cycles) Cycles involving reduced doses [n (% of cycles)] 14 (9.2) 15 (7.0) Median duration of treatment [weeks (range)] 9 (3 43) 9 (3 54) a Docetaxel, cisplatin; or vinorelbine, cisplatin. In both arms, cycles were delayed because of haematological toxicity (A2, 10 cycles; B2, 2 cycles), non-haematological toxicity (A2, 2 cycles; B2, 4 cycles) and other reasons (A2, 11 cycles; B2, 6 cycles). Dose reductions occurred in <10% of cycles in both arms (Table 3). However, the main reason for dose reductions was non-haematological toxicity [six cycles (42.9%) in A2 and 6 (40%) in B2]. By the cut-off date, 101 patients (88%) had died in arm A and 111 in arm B (94%). In arm A (A1 + A2), 86 patients (85.1%) died due to PD and 10 patients (9.9%) due to other reasons. Five deaths (4.9%) were treatment-related: 3 patients died during first-line treatment (A1) and 2 in second-line (A2). Similarly in arm B, 86 deaths (77.5%) occurred due to PD and 15 (13.5%) from other reasons. Ten deaths (9%) were treatment-related, all occurring in the first-line phase (B1). Toxicity Toxicity was assessed at every cycle in all patients who received at least 1 cycle of treatment (Table 4). During first-line bitherapy, a similar percentage of patients

6 86 Table 4. Grade 3 and 4 haematological and non-haematological toxicities in first-line treatment Related adverse event Arm A1 Arm B1 Docetaxel cisplatin (n = 115) Grade 3 (%) Grade 4 (%) Vinorelbine cisplatin (n = 118) Grade 3 (%) Grade 4 (%) Haematological Anaemia Neutropenia Febrile neutropenia Thrombocytopenia Total (%) Grade 3 4 (%) Total (%) Grade 3 4 (%) Non-haematological Asthenia Constipation Diarrhoea Nausea Vomiting Peripheral neuritis Alopecia Nail disorders Renal failure Infection experienced at least one treatment-related grade 3 4 adverse event in arms A1 (42.8%) and B1 (49.2%). However, twice as many patients in B1 had serious treatment-related adverse events (50%) than did patients in A1 (24.3%). Grade 3 4 anaemia was seen in 17 patients (14.8% in A1) (grade 3 only) and 41 patients (35.1% in B1) (Table 4). Grade 3 4 neutropenia was observed in 74 patients (64.3%) in arm A1 versus 98 (83.7%) in B1. Among them, febrile neutropenia occurred in <10% of patients (n = 11) receiving DC compared with 26% (n = 31) receiving VC. Use of granulocyte colony-stimulating factor was rare, occurring only in 2.4% of 1063 cycles. However, the incidence of grade 3 4 thrombocytopenia was similar for both arms: four (3.4%) and six (5.1%) patients in arm A1 and B1, respectively. Only one patient (0.9%) experienced grade 3 4 infection in the DC arm (A1) compared with four (3.4%) in the VC arm (B1). As expected, alopecia occurred in more patients receiving DC (86 patients, 74.8%) than with VC (44 patients, 37.3%) (Table 4). Nail disorders occurred in 13 patients (11.3%) in DC versus one patient (<1%) in VC. Digestive toxicities were in accordance with the known toxicity profile of each agent. Grade 3 and 4 nausea and vomiting occurred in A1 [21 (18.3%) and 18 (15.7%) patients, respectively] and in B1 [eight (6.7%) and 12 (10.1%) patients, respectively]. Grade 3 asthenia was experienced by the same percentage of patients in both arms. Grade 3 peripheral neurotoxicity was experienced only by two (1.7%) patients in B1. Thirty-nine patients (17 in A1, 22 in B1) discontinued combination therapy and 18 patients (10 in A1, eight in B1) discontinued first-line monotherapy due to adverse events. The main reasons for patient withdrawal from combination therapy were: neurotoxicities (n = 4), asthenia (n = 3) and digestive toxicities (n = 3) in A1; and haematological toxicity (n = 6, of whom three had febrile neutropenia), renal failure (n = 4) and neurotoxicities (n = 3) in B1. The main grade 3 4 toxicities in second-line therapy were haematological: anaemia [12 (24%) in A2 and 10 (19.2%) in B2] and neutropenia, which was more frequent in B2 [35 (67.3%)] than in A2 [26 (52%)]. Febrile neutropenia occurred in five (9.8%) and three patients (5.8%) in A2 and B2, respectively. Twice as many patients withdrew from the study due to toxicity with vinorelbine [10 (19.6%)] compared with docetaxel [five (9.6%)]. In the vinorelbine arm (A2), the main toxicities arising in patients were: neurological (n = 7), haematological with concomitant vascular accident (n = 1), intestinal obstruction (n = 1) and pain after infusion (n = 1). In the docetaxel arm (B2), neutropenia and nail disorders each occurred in one patient. Neurological toxicity occurred in two patients: in one patient this was combined with haematological toxicity and in the other it was combined with disorders accompanied by worsening of the general condition of this patient. During second-line monotherapy, the major reason for patient discontinuation was neurological toxicity in both arms (seven patients in A2, six in B2). In first-line therapy, there were three treatment-related deaths (2.6%) in DC (one due to respiratory distress, one septic shock, one unknown cause) and 10 in VC (8.5%) (seven due to sepsis with febrile neutropenia, one pneumonia, one infection and one unknown cause despite autopsy). In secondline treatment, two patients died due to sepsis with febrile neutropenia in the vinorelbine arm (A2). No deaths due to toxicity were observed in the docetaxel arm. Discussion Treatment of stage IV NSCLC has been a controversial issue during the last decade, particularly when patients with poor performance status (WHO) and/or brain metastases are involved. Patients with favourable prognostic factors are more frequently treated first-line with platinum-based chemotherapy associated with vinorelbine, gemcitabine, paclitaxel or docetaxel. Conversely, patients with poor prognostic factors are preferably treated with single agents. An advantage for cisplatin combinations, regardless of prognostic factors, were shown by meta-analyses [2, 8], and confirmed by phase III trials comparing a single-agent with its combination with platinum and trials comparing platinum monotherapy with combination therapy [9 11]. This study is one of the first European randomised trials comparing docetaxel in combination with cisplatin versus one of the European standards of chemotherapy in NSCLC: vinorelbine-cisplatin. The other pivotal study was that recently published by Fossella et al. [7] describing the results of a comparative phase III study

7 87 involving 1200 patients and comparing the first-line treatment of docetaxel platinum combinations versus VC. A previous phase III study [12] demonstrated the superiority of the combination vinorelbine cisplatin compared with vindesine cisplatin or vinorelbine alone. VC has shown higher median survival, particularly among patients with good performance status (PS) [13]. Recently, an additional Japanese trial compared DC with vindesine cisplatin [14]. The aim of the present French study was to assess efficacy and toxicity of DC compared with VC. Then, based on previous phase II results for this new combination in NSCLC [15 18], we aimed to determine the effect of cross-over to second-line monotherapy on progression. Based on previous experience with weekly vinorelbine in VC, the treatment schedule was modified, with vinorelbine 30 mg/m 2 /day given on days 1 and 8 only. The two arms had the same cisplatin dosage (100 mg/m 2 ) and 3-weekly schedule in first-line. We report here the results of a strategy involving two treatment sequences: first-line cisplatin-based combination therapy with DC (arm A1) or VC (arm B1), followed by maintenance monotherapy with either docetaxel or vinorelbine until evidence of PD. Thereafter, patients crossed over to receive second-line monotherapy with either vinorelbine (arm A2) or docetaxel (arm B2). When this study began, the inefficacy of maintenance chemotherapy had not been published. Results of efficacy gave an ORR of 33.9% in the firstline DC arm that was consistent with other phase II and III trials using the same dosage of cisplatin [10, 11, 16] and with the pivotal study by Fossella et al. [7] in which patients received both cisplatin and docetaxel at 75 mg/m 2 every 3 weeks or 25 mg/m 2 weekly vinorelbine and 100 mg/m 2 of cisplatin every 4 weeks. Similarly, Kubota et al. [14] reported a response rate of 37% using only 60 mg/m 2 of docetaxel and 80 mg/m 2 of cisplatin. The ORRs from four phase II studies ranged from 33% to 45%, with a median survival of 8 11 months and 1-year survival of nearly 35% [15, 18]. Only one Greek study reported a 1-year survival of 48%, probably due to the higher proportion of stage IIIB patients at inclusion [15]. In contrast to most other phase III studies comparing the combination VC with alternative combinations, we accrued unselected patients, including poor-prognosis patients with low PS (15% of patients in each arm had PS 2), metastatic disease (all patients had stage IV disease at inclusion) and brain metastases. In other published studies, at least 10% of the patients had stage IIIB disease, good PS of 0 or 1 and no brain metastases [7, 14, 19 22]. In the current study, more favourable responses to first-line DC versus VC are suggested, respectively, in patients aged >_60 years (20 of 40 patients versus 12 of 52) and those with squamous cell carcinoma (12 of 38 patients versus seven of 38 patients), and in large cell histological subtypes (six of 12 patients versus 1 of 16 patients) and those with PS 2 (six of 18 patients versus three of 17 patients). The ORR for with first-line DC is consistent with published results according to disease stage and PS. Unlike the Greek study [23], adenocarcinoma histological subtype was not a prognostic factor for response. Amongst other efficacy parameters, TTP, median overall survival and 1-year survival were similar in both arms, consistent with other phase III trials [11, 13]. There was no statistical difference in first-line response rate, survival or treatment discontinuations due to toxicity. In this study, 2-year survival was higher in DC followed by vinorelbine than in VC followed by docetaxel (17% versus 10%), which is in agreement with the survival obtained with the other phase III trial using DC [7, 14]. Moreover, 3-year survival was also twice as high (13% versus 6%) with firstline DC followed by second-line vinorelbine compared with VC followed by docetaxel, although because of the low number of long-term survivors, statistical significance was not reached. This result suggests a long-term survival benefit with a sequential regimen involving DC as first-line followed by a second-line therapy in patients with NSCLC, including those with metastatic disease. Clinical trials will be developed including second-line treatment with further drugs to improve the results of the sequence. Despite its better activity in second-line treatment compared with vinorelbine, docetaxel was unable to compensate for the lower efficacy of VC that had been used as first-line treatment in arm B. Docetaxel is better tolerated than vinorelbine as both a single agent and in combination with cisplatin. For example, the percentage of discontinuation due to adverse events was lower with DC (14.3% versus 18.3%), and with docetaxel monotherapy in first-line (29.4% in A1 versus 38.1% in B1) or second-line (9.6% in B2 versus 19.6% in A2) therapy. Similarly, a greater percentage of patients experienced at least one serious treatment-related adverse event with VC (50.8%) than with DC (24.3%). Haematological toxicity, such as neutropenia, anaemia and thrombocytopenia, occurred more frequently in arm B1. In particular, more patients experienced febrile neutropenia with VC (26.3%) than with DC (9.6%). This incidence was higher than the results of a phase III study in which, for each arm, only 4% of patients experienced febrile neutropenia [7]. A possible explanation for the difference in the incidence of febrile neutropenia may be the different doses of vinorelbine and cisplatin used in the Fossella et al. [7] trial (25 versus 30 mg/m 2 and 75 versus 100 mg/m 2, respectively), or that the patient population showed more advanced disease and had worse PS in our study. Treatment-related deaths were twice as high overall with VC followed by docetaxel (9% of patients) as with DC followed by vinorelbine (4.9%), and this was mostly due to febrile neutropenia. This also includes second-line therapy, in which no treatment-related deaths occurred during treatment with docetaxel and two occurred with vinorelbine. In our study, a low percentage of patients experienced neurotoxicity leading to combination treatment discontinuation in both arms: four (3.4%) patients in arm A1 and three (2.6%) in arm B1. These results are similar to those reported by the TAX 326 study (3.6% of patients) [7]. Therefore, in terms of neurotoxicity, DC is equivalent to VC. Moreover, compliance with treatment was similar in both arms for the median number of cycles administered. Nevertheless, despite the modified vinorelbine schedule used in this study, the number of complete combination cycles administered was greater in arm

8 88 A1 (98.3%) than in B1 (94.6%). Similarly, more cycles were delayed with vinorelbine administered as combination therapy or monotherapy than with docetaxel. The efficacy and safety of DC in our study is consistent with that reported by the TAX 326 study [7]. Despite similar response rates, the benefit risk ratio seems to favour DC in first-line followed by second-line vinorelbine, regarding 2- and 3-year survivals as well as a manageable toxicity profile. There were more long-term survivors when DC was used in first-line. In our study, the second-line therapy did not seem to impact on overall survival, and alternative and more effective drugs are sought for second-line treatment. Despite a higher response rate in second-line, our data suggest that docetaxel gives a better response rate in both lines, but is not able to compensate for the reduced impact on survival of VC as first-line therapy. The benefit for 2- and 3-year survivals seems to be attributable to DC as first-line. In conclusion, the trend in favour of the DC arm in terms of ORR, even though statistical significance was not reached, is consistent with previous reports. This study suggests an activity of first-line DC in advanced NSCLC, and that second-line vinorelbine does not provide additional clinical benefit. As already shown in other studies, the use of DC in first-line should provide a better percentage of long-term survivors, despite the absence of efficacy of the second-line in our study. Acknowledgements We would like to thank all of our colleagues who contributed to the success of this study: C. Elkouri, H. Lacroix and S. Bordenave-Caffre (CHU Hôtel-Dieu, Nantes), J. Bennouna (Centre R. Gauducheau, Nantes), J. Bodin (CH Niort), T. Conroy (Centre A Vautrin, Vandoeuvre les Nancy), H. Janicot (CHU G Montpied, Clermont-Ferrand), F. Mayer, M. Darut-Jouve and P. Fargeot (Centre GF Leclerc, Dijon), X. Sun (CH A Boulloche, Montbéliard), P. Masson (CH Cholet), G. Adam (CH J Cœur, Bourges), PJ. Couderc and S. Friard (Hôpital Foch, Suresnes), JP. Duhamel (Clinique du Petit Colmoulins, Harfleur), G. Zalcman and A. Livartowski (Institut Curie, Paris), P. Foucher (CHU, Dijon), T. Pigeanne (CH Sables d Olonne), J. Lambert (Centre Médical Lalande, Lutterbach), B. Leduc (CHG, Brive), M. Peureux and CP. Hubscher (CH Montivilliers), X. Deguiral (CHG, Saint- Nazaire), D. Roux (Centre Jean Perrin, Clermont-Ferrand) and J. Guevenoux (CH les Oudairies, La Roche sur Yon). This study was supported by Laboratoire Aventis, Paris, France. References 1. Jemal A, Thomas A, Murray T et al. Cancer statistics CA Cancer J Clin 2002; 52: Non-small Cell Lung Cancer Collaborative Group. Chemotherapy in NSCLC: a meta-analysis using updated data on individual patients from 52 randomised clinical trials. BMJ 1995; 311: Rapp E, Pater JL, William A et al. Chemotherapy can prolong survival with advanced non-small cell lung cancer. Report of a Canadian multicenter randomized trial. J Clin Oncol 1988; 6: Shepherd F, Ramlau R, Mattson K et al. Prospective randomized trial of docetaxel versus best supportive care in patients with non smallcell lung cancer previously treated with cisplatin based chemotherapy. J Clin Oncol 2000; 18: Fossella FV, De Vore R, Kerr R et al. Randomized phase III trial of docetaxel vs vinorelbine or ifosfamide in patients with advanced nonsmall-cell lung cancer previously treated with platinum-containing chemotherapy regimens. J Clin Oncol 2000; 18: Rigas JR. Single agent docetaxel in previously untreated non smallcell lung cancer. Oncology 1997; 11 (Suppl 7): Fossella F, Pereira JR, Von Pawel J et al. Randomized, multinational, phase III study of docetaxel plus platinum combinations vs vinorelbine plus cisplatin for advanced NSCLC: the TAX 326 study group. J Clin Oncol 2003; 16: Souquet PF, Cahuvin F, Boissel JP et al. Polychemotherapy in advanced NSCLC: a metaanalysis. Lancet 1993; 342: Sandler A, Nemunaitis J, Denham C et al. Phase III study of gemcitabine plus cisplatin versus cisplatin alone in patients with locally advanced or metastatic non-small-cell lung cancer. J Clin Oncol 2000; 18: Scagliotti GV, De Marinis F, Rinaldi M et al. Phase III randomized trial comparing three platinum-based doublets in advanced non-smallcell lung cancer. J Clin Oncol 2002; 20: Wozniak AJ, Crowley JJ, Balcerzak SP et al. Randomized trial comparing cisplatin with cisplatin plus vinorelbine in the treatment of advanced non-small-cell lung cancer: a Southwest Oncology Group study. J Clin Oncol 1998; 16: Le Chevalier T, Brisgand D, Douillard JY et al. Randomized study of vinorelbine and cisplatin versus vindesine and cisplatin vs vinorelbine alone in advanced non-small cell lung cancer. Results of a European multicenter trial including 612 patients. J Clin Oncol 1994; 12: Le Chevalier T, Brisgand D, Soria JC et al. Prognostics analysis of survival in the European randomised trial comparing navelbine vs navelbine cisplatin vs vindesine. Lung Cancer 2000; 29 (Suppl 2): 82a. 14. Kubota K, Watanabe K, Kunitoh H et al. Phase III randomized trial of docetaxel plus cisplatin versus vindesine plus cisplatin in patients with stage IV non-small-cell lung cancer: the Japonese Taxotere Lung Cancer Study Group. J Clin Oncol 2004; 22: Georgoulias V, Androulakis N, Dimopoulos AM et al. First line treatment of advanced non-small-cell lung cancer with docetaxel and cisplatin: a multicenter phase II study. Ann Oncol 1998; 9: Le Chevalier T, Monnier A, Douillard JY et al. Docetaxel (Taxotere) plus cisplatin: an active and well tolerated combination in patients with advanced non-small cell cancer. Eur J Cancer 1998; 34: Zalcberg J, Millward M, Bishop J et al. Phase II study of docetaxel and cisplatin in advanced non-small-cell lung cancer. J Clin Oncol 1998; 16: Belani CP, Bonomi P, Dobbs T et al. Multicenter phase II study of docetaxel and cisplatin combination in patients with non-small cell lung cancer. Proc Am Soc Clin Oncol 1997; 16: 221a (Abstr 37). 19. Comella P, Frasci G, Panza N et al. Randomized trial comparing cisplatin, gemcitabine, and vinorelbine with either cisplatin and gemcitabine or cisplatin and vinorelbine, in advanced non-smallcell lung cancer: interim analysis of a phase III trial of the Southern Italy Cooperative Oncology Group. J Clin Oncol 2000; 18: Kelly K, Crowley J, Bunn PA et al. Randomized phase III trial of paclitaxel plus carboplatin versus vinorelbine plus cisplatin in the treatment of patients with advanced non-small-cell lung cancer:

9 89 a Southwest Oncology Group trial. J Clin Oncol 2001; 19: Schiller JH, Harrington D, Belani C et al. Comparison of four chemotherapy regimens for advanced non-small-cell lung cancer. N Engl J Med 2002; 346: Lilenbaum RC, Herndorn J, List C et al. Single agent versus combination chemotherapy in advanced non-small cell lung cancer: a CALGB randomised trial of efficacy, quality of life and cost-effectiveness. Proc Am Soc Clin Oncol 2002; 21: (Abstr 2). 23. Georgoulias V, Papadakis E, Alexopoulos A et al. Platinum-based and non-platinum-based chemotherapy in advanced non-small-cell lung cancer: a randomised multicenter trial. Lancet 2001; 357: