Second‐line chemotherapy for non‐small cell lung cancer

Abstract

Background

The role of second‐line chemotherapy for the treatment of patients with non‐small cell lung cancer (NSCLC) who have relapsed or failed to respond to first‐line treatment was unclear.

Objectives

To determine the effectiveness of any second‐line chemotherapy in patients with NSCLC.

Search methods

Medline (1966‐July 2001), Embase (1974‐July 2001), Cancerlit (1993‐July) and tthe Cochrane Central Register of Controlled Trials (CENTRAL, issue 2 2001) were searched. In addition a handsearch was performed and experts in the field contacted to identify any further studies that had not been found by the electronic searches.

Selection criteria

Randomised controlled clinical trials in which any second‐line chemotherapy was compared with placebo or best supportive care in patients with NSCLC who had failed to respond to any previous chemotherapy regimen.

Data collection and analysis

Data were extracted by 2 independent authors and revised by a third author.

Main results

Only one study was included. This study included a total of 204 patients who were randomised to receive either doxetaxel or best supportive care. Following an unacceptably high toxic death rate the dose of doxetaxel was reduced from 100 mg/m² to 75 mg/m². Docetaxel gave an extra 2.4 months survival ‐ an average of 7.0 months vs 4.6 months on best supportive care. At 1 year after diagnosis 29% of doxetaxel treated patients were alive compared with 19% of the best supportive care group.

Authors' conclusions

Definitive recommendations cannot be made since evidence is only available from one randomised controlled trial which, though of reasonable quality had a number of limitations. There is currently no evidence to support second‐line treatment of patients with poor performance status. Larger, well‐designed controlled trials are needed to further evaluate whether the benefits of second‐line chemotherapy to patients with non‐small cell lung cancer outweigh its risks and costs.

Plain language summary

Not enough evidence to give a second round of chemotherapy to patients with lung cancer in a poor state

Patients with lung cancer and a good physical condition who have not been cured by a first round of chemotherapy often receive a second round of chemotherapy (second‐line). A second round of chemotherapy may not increase the survival chances of these patients and may make them feel worse because of bad side effects. This review has found only one study that compared the effects of a second round of chemotherapy with treatment showing no benefits for the patients, apart from keeping them comfortable. This study does not provide enough evidence to judge whether such treatment causes more benefits than harms and further larger studies are needed before firm conclusions can be drawn.

Background

Chemotherapy has been widely used for the treatment of non‐small cell lung cancer (NSCLC). It is administered as a single agent or in combination with surgery or radiotherapy, either as radical or palliative treatment (Ginsberg 1997). Although some survival benefits have been reported (NSCLCCG 1995; NSCLCCG 2001), the risk‐benefit ratio for chemotherapy in lung cancer is rather narrow.

It is therefore not surprising that there are uncertainties about the effectiveness of second‐line chemotherapy administered either for clinical failure (i.e. recurrence or progression) or lack of response to previous treatment. Patients with unresponsive or progressive disease are usually in an advanced stage of a highly lethal disease. Furthermore, they may be suffering from serious toxicities as a consequence of treatment received, or have cross‐resistance to the drugs already administered. However, despite these potential contraindications, second‐line chemotherapy for lung cancer is often used in oncology, in some cases with reported apparent benefits (Rigas 1998). The need for a systematic review to analyse the effectiveness of second‐line chemotherapy administered for lung cancer was therefore identified. This review focuses exclusively on NSCLC because this is a clearly different biological and clinical entity from small cell lung cancer.

Objectives

To determine the effectiveness of any form of second‐line chemotherapy in patients with NSCLC.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials (RCTs) or controlled clinical trials (CCTs) in patients with NSCLC, in which second‐line chemotherapy is compared with placebo or conventional supportive care.

Types of participants

Patients of any age and stage with histologically proven NSCLC, who have received any course of previous chemotherapy for the same disease.

Types of interventions

Any second‐line chemotherapy (any single agent or combination of different drug treatments, at any dosage or number of cycles), regardless of whether previous chemotherapy had been given as adjuvant, neoadjuvant or primary treatment for limited or advanced disease.

All studies where second‐line chemotherapy was administered either for local or regional recurrence, metastasis or lack of response, were included. Any chemotherapy administered as "third‐line", "fourth‐line", etc… were also considered for inclusion. Although the need to conduct a separate subgroup analysis by the interval between chemotherapy treatments or response to previous courses of different chemotherapies was planned, in view of the lack of evidence this was not possible.

Studies where second‐line chemotherapy was exclusively justified for reasons of excessive toxicity due to an initial treatment and those where second‐line chemotherapy was administered in combination with a second radiotherapy or surgical treatment, were excluded.

Types of outcome measures

The primary outcome was lung cancer mortality or mortality due to any cause, or survival.

Other outcomes considered included:
 a) Disease‐free survival, measured in weeks
 b) Quality of life (QOL) or functional status measures
 c) Response rate (complete and/or partial)
 d) Treatment‐related toxic events

The therapeutic response criteria usually followed in oncology settings are listed below:

Complete response: total disappearance of all known disease, over a period of at least 4 weeks. 
 Partial response: in the case of bi‐dimensionally measurable disease, reduction of at least 50% in the sum of the products of the greatest perpendicular diameters of all measurable lesions, determined through two observations not more than 4 weeks apart. In the case of uni‐dimensionally measurable disease, reduction of at least 50% in the sum of the greatest diameters of all lesions, determined through two observations not more than 4 weeks apart.

The WHO scale was used to measure toxicity whenever possible, taking into account any symptom or sign attributable to the treatment reported in the corresponding study.

Search methods for identification of studies

Medline (1966‐July 2001), Embase (1974‐July 2001), Cancerlit (1993‐July) and the Cochrane Central Register of Controlled Trials (CENTRAL, issue 2 2001) were searched to identify all published randomised clinical trials and controlled clinical trials that have assessed chemotherapy for NSCLC. In addition any unpublished randomised clinical trials that were identified were considered for inclusion.

See: Collaborative Review Group Search Strategy.

1. randomized controlled trial.pt. 
 2. controlled clinical trial.pt. 
 3. randomized controlled trials/
 4. random allocation/
 5. double blind method/
 6. single blind method/
 7. or/1‐6 
 8. clinical trial.pt. 
 9. exp clinical trials/ 
 10. (clin$ adj25 trial$).ti,ab. 
 11. ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab. 
 12. placebos/ 
 13. placebo$.ti,ab. 
 14. Random$.ti,ab. 
 15. research design/ 
 16. or/8‐15 
 17. comparative study/ 
 18. exp evaluation studies/ 
 19. follow‐up studies/ 
 20. prospective studies/ 
 21. (control$ or prospectiv$).ti,ab. or volunteer$.ti,ab.
 22. or/17‐21
 23. 7 or 16 or 22 
 24. limit 23 to human
 25. exp lung neoplasms/ 
 26. ((lung$ or pulmon$) adj25 (neoplas$ or cancer or adenocarcinom$ or carcinom$ or sarcom$ or lymphom$ or carcino$ or tumor$ or tumour$ or blastom$. or squamous or oat cell or small cell)).ti,ab.
 27. exp carcinoma, non‐small cell lung/ or exp carcinoma, small cell/ 
 28. exp pneumonectomy 
 29. pneumonectom$.ti,ab,sh. 
 30. or/25‐30 
 31. 24 and 30

This search strategy in each of the above mentioned databases was complemented with other terms in order to focus the results on second‐line chemotherapy for non‐small cell lung cancer trials:

second‐line chemotherapy
 or 
 rescue chemotherapy
 failed chemotherapy
 relapsed non‐small cell lung cancer
 or recurrent non‐small cell lung cancer
 or progressive non‐small cell lung cancer
 or resistant non‐small cell lung cancer

2. Identification of further studies from references of identified trials or reviews

3. Colleagues, collaborators and other experts in the field were asked to identify missing and unreported trials

4. Other: The abstracts of the 9th World Conference on Lung Cancer (Tokyo, 2000) and 7th Central European Lung Cancer Conference (Prague, 2001) were also reviewed.

Data collection and analysis

All RCTs and CCTS were assessed for inclusion independently by 2 authors. Data were extracted from the included studies (those that fulfilled the inclusion criteria) using a specific methodology and previously tested form. Methodological quality was assessed according to Cochrane Collaboration handbook guidelines and using the CONSORT Statement as a basis (Begg 1996). No formal scoring system was adopted. The trial that met the selection criteria was registered using ProCite. The data were extracted by 2 authors independently to ensure validity and discrepancies were resolved by a third author.

Results

Description of studies

A total of 151 potentially eligible trials were identified by the literature searches. All except one (Shepherd 2000) were found to be ineligible. The great majority of these studies were phase II studies without a control group and others did not include a comparison of a second‐line chemotherapy with the best supportive care (BSC) or a chemotherapeutic regimen administered as a real second option. The exclusion details for each study are presented in the Characteristics of excluded studies. This review therefore focuses on the trial by Shepherd et al (Table 1).

1. DESCRIPTION OF RESULTS.

Study ID	Interventions	Outcomes	Results	Comments
SHEPHERD 2000	Docetaxel 100 mg/m2 every 21 days (48 patients) compared with Best Supportive Care (100 patients); but the protocol was amended to reduce the dose of docetaxel from 100 mg/m2 to 75 mg/m2 every 21 days (55 patients) for the second half of the trial because a high toxic death rate was identified in the chemotherapy arm	1) Median Survival (months) 2) 1‐year survival (%) 3) Time to progression (weeks) 4) QOL	1) 7.0 months (95% CI, 5.5 to 9.0) for chemotherapy group (5.9 months for 100 mg/m2 subgroup and 7.5 for 75 mg/m2 subgroup,CI for subgroups not available) vs 4.6 months (95% CI, 3.7 to 6.0) for BSC group 2) 29% for chemotherapy group (19% for the 100 mg/m2 subgroup and 37% for the 75 mg/m2 subgroup) vs 19% for BSC groups 3) 10.6 weeks for docetaxel 100 mg/m2 vs 6.7 weeks for docetaxel 75 mg/m2	Details on the quality of life analysis were not available for this review.

The Shepherd trial (Shepherd 2000) which included a total of 204 eligible patients, was published in 2000, although the inclusion period lasted 4 years (November 1994 to December 1998). Thirty‐five centres from different countries contributed to the study: USA , Canada, Finland, United Kingdom, Poland, Hungary, Puerto Rico and Sweden.

The inclusion criteria of the trial were: prior treatment with one or more platinum‐containing chemotherapy regimen; unresectable locally advanced or metastatic NSCLC; lesions that were both measurable and evaluable; performance status of 2 or lower on the ECOG (Eastern Cooperation Oncology Group) scale; adequate haematological, serum creatinine, bilirubin and hepatic enzyme (except alkaline phosphatase) parameters; patients may have received prior radiotherapy, but should not have been previously treated with taxanes; appropriate time interval between ending the last therapy and entering the study. Patients were excluded if they had received prior paclitaxel or if they had symptomatic brain metastases.

Docetaxel 100 mg/m² every 21 days was the planned intervention compared with BSC, but only 47% of the eligible patients received that dose, because a high toxic death rate was identified in the chemotherapy arm. The protocol was amended to reduce the dose of docetaxel from 100 mg/m² to 75 mg/m² for the second half of the trial and dose adjustments were made as appropriate to reduce toxicity. Treatment was continued until disease progression or unacceptable toxicity. BSC was defined as any appropriate therapy (including radiotherapy) that did not include chemotherapy.

The primary end point of the study was survival and secondary outcomes included objective tumour response, duration of response and QOL measures. Analysis was performed on an intention‐to‐treat basis. Survival time was censured for loss of contact or initiation of any further anti‐tumour therapy.

Risk of bias in included studies

Randomisation
 The study by Shepherd et al. (Shepherd 2000) was a randomised controlled trial, which has been reported as an oral communication (Shepherd 2000), original article (Shepherd 2000) and as a part of a review (Shepherd 2001). No information is provided in the article about the randomisation or concealment methods that were followed, but the baseline characteristics of patients in the chemotherapy and the BSC arms were very similar.

Study design and sample size
 The sample size was calculated to detect a 3‐month difference in median survival (the primary outcome) between the treatment and control groups. Although eligible patients were stratified on the basis of their performance status (0‐1 vs 2) measured by the ECOG scale and on their best response to previous platinum‐based treatment, the sample size was too small to allow analysis by these subgroups.

Although it is reported that the time interval from last chemotherapy treatment among responders varied between 1 and 21 months there is no information on whether or how this interval varied among the rest of the included patients.

Quality of life measurement
 All patients were asked to complete quality of life (QOL) questionnaires but although p values are given for differences in favour of docetaxel‐treated patients with respect to pain and fatigue, detailed results, planned for subsequent publication, were not available for this review.

Authors have been contacted to clarify details on the QOL aspects and these will be incorporated in the next update of this review.

Effects of interventions

Results of this review are based only on the 204 patients included in the Shepherd trial. All patients except one (subsequently excluded) had confirmed NSCLC. The remaining 203 were randomised into 2 groups: 103 into the docetaxel arm (48 receiving 100 mg/m² every 21 days and 55 receiving 75 mg/m² also every 21 days) and 100 into the BSC group.

The median age of patients was 61 years for both groups and in both groups a third of patients were women. Eighty‐one per cent (81%) in the BSC group had stage IV disease, compared with 76% in the chemotherapy group. About 74% in both groups had 0‐1 scores on the ECOG scale. About 75% of patients in both groups had received only one prior regimen and 37% of the BSC and 33.7% of the CT groups had had a partial or a complete response to previous CT treatments.

Survival
 The median survival was 7 months (CI: 5.5 to 9 months) in the chemotherapy group compared with 4.6 months (CI: 3.7 to 6 months) in the BSC group. When the results of the chemotherapy group were analysed separately for the different doses (100 mg/m² and 75 mg/m² ), the median survivals were 5.9 (CI:not available) and 7.5 months (CI:not available), respectively. The percentage of survivors at one year was 29% for the chemotherapy group (19% for the 100 mg/m² subgroup and 37% for the 75 mg/m² subgroup) and 19% for the BSC group. Nine patients received chemotherapy after the initial second‐line chemotherapy (3 in the chemotherapy arm and 6 in the BSC arm) and were included in their respective groups for analysis.

The differences found in median survival were not statistically significant, since the 95 % CI for both groups overlapped slightly. The number needed to treat (NNT) was 9.9 (calculated as part of this review), that is ‐10 patients need to be treated with second‐line chemotherapy in order for one patient to survive for one year. The CI for the NNT ranges from 5 patients being harmed to 59 benefitting).

Response
 Almost six per cent (5.8%) of patients on chemotherapy achieved a partial response and none (0%) a complete response. The overall duration of that response (for 84 patients with a measurable lesion) was 26.1 weeks.

Toxicity
 Six patients died within 30 days of receiving chemotherapy (5 in the high‐dose and 1 in the low docetaxel dose group) from causes other than progressive disease: fatal neutropenia and/or pneumonia. For the 49 patients treated at the 100 mg/m² dose, the median number of cycles was two (range 1 to 17) and only 68% of cycles could be delivered at the initial planned dose. The patients treated with docetaxel 75 mg/m² received a median of 4 cycles. Haematological toxicities included a grade 3 or 4 neutropenia, suffered by 76% patients in the doxetaxel group overall (86% and 67% for the 100 mg/m² and 75 mg/m² doses respectively), and febrile neutropenia suffered by 22.4% (n=11) of those receiving the higher dose, compared with only 1.8% (n=1) of those on the lower dose. Grade 3 or 4 anaemia also occurred more frequently in the higher dose group than in the lower dose group (16.3% vs 5.5%, 10.6% overall) and thrombocytopenia occurred in less than 1% of all chemotherapy cycles. Non‐haematological toxicities were also reported for the chemotherapy patients and compared with BSC patients. For these toxicities the most striking differences between the two chemotherapy subgroups (100 and 75 mg/m² , respectively) were: asthenia (61.2 and 54.5%), diarrhoea (30.6 and 36.4%), fever (36.7 and 61.8%), infection (36.7 and 30.9%), nausea (34.7 and 36.4%), neuromotor change (16.3 and 14.5%), neurosensory change (26.5 and 20%), pulmonary toxicity (53.1 and 38.2%), stomatitis (26.5 and 25.5%) and vomiting (26.5 and 23.6%). When only grade 3/4 toxicities were taken into account, the percentages were much lower and closer (although always higher) to that recorded in the BSC group.

Quality of life
 A full report on the quality of life (QOL) assessments from the Shepherd trial are to be published in another article.The QOL data reported in this article (Shepherd 2000) have therefore not been considered for this review as their exhaustiveness and validity have not been clarified.

Discussion

Virtually all patients with NSCLC who respond to initial treatment will eventually relapse. Some of these patients have had a relatively long disease‐free interval and have a good performance status at the time of relapse. Patients who relapse at an earlier stage or who do not respond to induction treatment are considered refractory to first‐line chemotherapy and therefore have a high probability of also being resistant to second‐line treatment.

In spite of having a very poor prognosis, many patients are unwilling to accept BSC only, and although they may have already suffered severe toxicity from treatment, they or their relatives often demand further active intervention. Medical professionals are also sometimes equally reluctant to opt for a merely supportive strategy and prefer to offer their patients at least some therapeutic option. As a result many patients are given second‐line chemotherapy even though its clinical effectiveness is unclear.

Non‐small cell lung cancer is nevertheless a lethal condition with an extremely poor outlook once first‐line treatment has failed. Health care providers need to make resource allocation decisions based on an evaluation of the relative benefits and costs of treating patients in the advanced stages of this disease. A recent report for the UK National Institute of Clinical Excellence (NICE 2001) found that the use of docetaxel as second‐line treatment gave an incremental cost per life year gained of 14,098 pounds compared with BSC. The costs per life saved in treating patients with relapsed NSCLC need to be set not only against those of treating patients with other cancers but also against the range of competing priorities faced by a health care system with limited resources. Since these decisions may have serious implications in terms of both the length and quality of people's lives, it is imperative that they are informed by data of the best possible quality.

There are more than one million new cases of lung cancer each year and 5‐year survival for NSCLC has been estimated as being only 12%. Furthermore a systematic review (NSCLCCG 1995) found that the benefits of first‐line chemotherapy are at best modest . Given that the incidence of progression in this common disease is high, that second‐line treatment is frequently undertaken for clinical reasons but with scant supporting evidence, and that there is increasing competition for scarce health care resources, there is an urgent need for quality data on the effectiveness of second‐line treatment options for this disease.

We undertook an extensive search of electronic databases, and by way of handsearches and contacts with experts in the field identifed unpublished trials. Although 151 potentially eligible trials were identified, only one trial (Shepherd 2000) fulfilled the inclusion criteria. This paucity of trials was surprising: only two phase III trials were identified, of which one (Fosella 2000), was excluded from this review because it compared active agents (docetaxel vs vinorelbine or ifosfamide) and did not include a control group which only received BSC.

The Shepherd study thus appears to be the only existing RCT that examines the effectiveness of second‐line chemotherapy compared with BSC only. Although a range chemotherapeutic drugs have been assessed in phase II studies, docetaxel is the only agent that has been studied in phase III trials of second‐line treatment of NSCLC. The Shepherd trial is an international, randomised controlled trial with a low recruitment rate (50 patients per year over the 4 year period), although 35 centres in different countries participated. The trial has an acceptable level of quality, although some minor limitations were detected in the article: although the treatment and best supportive groups were well balanced, no details were provided about the method of randomisation or of allocation concealment; the analysis was done on an intention‐to‐treat basis, but insufficient information was provided about losses; the sample size (about 200 patients) was relatively small and prevented any subgroup analysis; the results of the quality of life questionnaires are still not available; there are some minor discrepancies between the figures appearing in the two references where the trial is reported (Shepherd 2000; Shepherd 2001 ) which need further clarification; and although there is no declaration of conflict of interests given in the original article, one of the authors is from a pharmaceutical company (Rhône‐Poulenc Rover). We have contacted the authors of the study to clarify these points and will update this review accordingly.

The sample size in this trial was calculated to detect a 3‐month difference in median survival between patients in the treatment and control groups ‐ the primary outcome, but not to compare results within groups defined by performance status and previous response to therapy. The difference in median survival found in the study was only 2.4 months and did not achieve statistical significance. With respect to 1‐year survival, the percentage of survivors in the group receiving docetaxel (29 %) was greater than in the control group (19 %). Although these differences did not reach statistical significance, they may be important from a clinical point of view. Ten patients need to be treated (NNT) in order for one to survive for one year, although all of them are exposed to a wide range of chemotherapy toxicities and some will be severe.

A judgement of the effectiveness of docetaxel from the Shepherd study must be made by considering all the patients together, irrespective of whether the dose received was that originally planned (100mg/m² ), or the dose amended in the light of high toxicity levels (75 mg/m² ). Although it is possible to estimate the different effects of each dose, definitive conclusions are precluded because of the consequent reduction in sample size. Given these constraints, the benefits of docetaxel therapy at 75 mg/m² appear to outweigh the risks. Compared with BSC, docetaxel (75 mg/m² ) gave an extra 3 months survival ‐ a median of 7.5 months vs 4.6 months. At 1 year after diagnosis, 37 % of the docetaxel treated patients were alive compared with 19% of the BSC group. When used at 100 mg/m² , docetaxel obtained poorer results (median survival of 5.9 months and 1‐year survival of 19%), probably because of high toxicity.

A balanced assessment of the impact of docetaxel on patients' quality of life must await publication of the full results of the quality of life assessment. Docetaxel was also found to reduce the need for opiate analgesia (32% of docetaxel patients vs 49% BSC) and palliative radiotherapy (26% vs 37%).

Authors' conclusions

Implications for practice.

It is difficult to draw definitive conclusions or make recommendations for treatment on the basis of findings from only one trial. Nevertheless NSCLC patients at advanced stages of disease who have failed to respond to previous platinum‐based chemotherapy or who have relapsed after a disease‐free period, and who have good performance status, could be offered the possibility of receiving second‐line chemotherapy treatment with docetaxel at 75 mg/m² every 21 days. This should be on the understanding that the potential survival gain would be very modest and that toxicity, though only severe in a minority of patients, occurs frequently. No complete information is available on the effect of such a treatment option on the patient's quality of life.

It is crucial that affected patients and involved doctors understand the potential benefits and risks of docetaxel as a second‐line treatment in order to make well‐informed decisions and they should be aware that current evidence only applies to patients with good performance status. There is currently no evidence to support the use of second‐line chemotherapy in patients with a poor performance status or symptomatic brain metastases.

Implications for research.

In order to inform pressing and difficult clinical and resource management decisions, there is an urgent need for high quality, randomised trials which include patients clearly categorised by baseline characteristics, pretreatment variables, response to first‐line therapy, type of first‐line therapy and time intervals between the last chemotherapy course, disease recurrence and commencement of second‐line therapy. In this way second‐line treatment options can be tested for the optimal dose and administration pattern required to achieve the best possible outcomes with the least possible toxicity.

Determination of the factors predictive of good response to second‐line chemotherapy would allow a more balanced trade‐off between the short and medium‐term benefits and risks of docetaxel. Additional trials comparing docetaxel (75 mg/m²) with BSC are necessary in order to confirm the promising results of Shepherd's trial and to analyse the cost‐effectiveness of these alternatives in more detail than is currently available.

Finally, if other drugs or combinations are to be compared in RCTs, a docetaxel group receving 75 mg/m² every 21 days should be the referent control group.

What's new

Date	Event	Description
21 November 2016	Amended	This review will soon be updated. Methods have changed since 2001. The authors will publish a new protocol in the coming weeks

History

Protocol first published: Issue 4, 2000
 Review first published: Issue 4, 2001

Date	Event	Description
13 March 2012	Amended	Additional table linked to text
18 September 2008	Amended	Converted to new review format.
31 July 2001	New citation required and conclusions have changed	Substantive amendment

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Shepherd 2000.

Methods	RCT Multicentre study in 35 centres from different countries (USA, Canada, Finland, UK, Poland, Hungary and Puerto Rico). Intention‐to‐treat analysis.
Participants	204 patients with histologically or cytologically proven unresectable locally advanced or metastatic NSCLC. Inclusion criteria: previous treatment with more than one chemotherapy regime, except taxanes; performance status =< 2 on ECOG scale and adequate haematological parameters. Excluded those with symptomatic or uncontrolled brain metastases or peripheral neuropathy grade >2 (National Cancer Institute, USA).
Interventions	1) Treatment arm: Docetaxel 100 mg/m2, 1‐hour intravenous infusion every 21 days (first part of the study) / 75 mg/m2, intravenously over 1 hour / 3 weeks (second part of the study). Premedication 24 hours before with dexamethasone (dose adapted to docetaxel regimen). 2) Control group: BSC with any therapy given by the treating physician (analgesics, antibiotics, transfusions and/or palliative treatment)
Outcomes	1) Survival: from date of randomisation until date of death; 2) Time to disease progression: from date of randomisation to date of disease progression; 3) Response duration: from date of randomisation to date of documentation of disease progression; Patients were evaluated every three weeks: complete medical history, physiscal examination, weight and ECOG status, vital signs, toxicities, QOL questionnaires, chest x‐ray and scans (every 6 weeks).
Notes	The docetaxel regimen was changed during the study because of high toxicity.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Unclear risk	B ‐ Unclear

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Adonizio 2001	Phase II study, no control group
Agelaki 2001	Phase II study, no control group
Alexopoulos 1999	Phase II study, no control group
Anderson 2000	CT plus BSC vs BSC, but CT not a second‐line treatment
Androulakis 1999‐DG	Phase II study, no control group
Androulakis 1999‐PG	Phase II study, no control group
Baas 1999	Phase II study, no control group
Barr 1999	Phase II study, no control group
Bervar 1994	Phase II study, no control group
Biesma 1999	Phase II study, no control group
Bonomi 1989	Phase II study, no control group
Buccheri 1989	CT vs BSC, but CT not second‐line treatment
Burris 1993	Phase II study, no control group
Camps 2000	Phase II study, no control group
Carreca 2000	Phase II study, no control group
Cartei 1993	CT plus BSC vs BSC, but the CT is not a second‐line treatment
Cassano 1998	Phase II study, no control group
Cellerino 1991	CT vs BSC, but the CT is not a second‐line treatment
Chang 1996	Phase II study, no control group
Cole 1999	Phase II study, no control group
Cormier 1982	CT vs placebo and supportive care when required
Crawford 2001	Phase II study, no control group
Crinó 1999	Phase II study, no control group
Dansin 1992	Phase II study, no control group
De Pas 2001	Phase II study, no control group
DiNunno 2000	Phase II study, no control group
Dongiovanni 2001	Phase II study, no control group
Dowell 2001	Phase II study, no control group
Estapé 1992	Phase II study, no control group
Fidias 2001	Phase II study, no control group
Fosella 2000	Comparing diferents CT regimens, not CT vs placebo or BSC
Fossella 1995	Phase II study, no control group
Fuks 1983	Not a randomized controlled trial. Compared different second line CT regimens, but with CT or RT or both in the first line treatment
Furnas 1982	Phase II study, no control group
Gandara 2000	Phase II study, no control group
Ganz 1989	CT plus BSC vs BSC, but CT not a second‐line treatment
García‐López 2000	Phase II study, no control group
Garfield 1998	Phase II study, no control group
Georgoulias 1997	Phase II study, no control group
Gervais 2000	Comparison between sequential chemotherapy and another sequential regimen, in chemonaïve patients.
Giaccone 1990	Phase II study, no control group
Gian 1999	Phase II study, no control group
Gomez 2001	Phase II study, no control group
Gralla 1979	Phase II study, no control group
Gridelli 1992	Phase II study, no control group
Gridelli 1998	CT plus BSC vs BSC, but the CT is not a second line treatment
Guerra 1998	Phase II study, no control group
Haas 1996	Phase II study, no control group
Hainsworth 1995	Phase II study, no control group
Hainsworth 2000	Phase II study, no control group
Helsing 1998	CT plus BSC vs BSC, but CT not a second‐line treatment
Herbst 1999	Phase II study, no control group
Herrero 2000	Phase II study, no control group
Hotton 1999	Phase II study, no control group
Iaffaioli 2000	Phase II study, no control group
Kaasa 1991	CT vs BSC, but the CT is not a second‐line treatment
Kakolyris 2000‐IC	Phase II study, no control group
Kakolyris 2001‐DG	Phase II study, no control group
Kakolyris 2001‐PCA	Phase II study, no control group
Koletsky 1999	Phase II study, no control group
Kosmas 2001‐GV	Phase II study, no control group
Kosmas 2001‐PIC	Phase II study, no control group
Kris 1985	Phase II study, no control group
Kunitoh 1998	Phase II study, no control group
Leon 2000	Phase II study, no control group
Lin 2001	Phase II study, no control group
Lopez‐Vivanco 2001	Phase II study, no control group
Mattson 1998	Phase II study, no control group
Morales 2001	Phase II study, no control group
Murphy 1994	Phase II study, no control group
Nakai 1991	Phase II study, no control group
Nakamura 1999	Phase II study, no control group
Nakanishi 1999	Phase II study, no control group
Nauman 1997	Phase II study, no control group
Negoro 1991	Phase II study, no control group
Niitani 1994	Phase II study, no control group
Papadakis 1998	Phase II study, no control group
Papadimitrakopoulou	Phase II study, no control group
Pronzato 1994	Phase II study, no control group
Quoix 1991	CT vs BSC, but CT not a second‐line treatment
Ranson 2000	CT plus BSC vs BSC, but CT not a second‐line treatment
Rapp 1988	Ct vs CT vs BSC, but CTs not second‐line treatments
Reddy 1999	Phase II study, no control group
Rigas 1994	Phase II study, no control group
Rinaldi 1994	Phase II study, no control group
Roa 1998	Phase II study, no control group
Robinet 1997	Phase II study, no control group
Rosati 2000	Phase II study, no control group
Rossi 1999	Phase II study, no control group
Rosvold 1998	Phase II study, no control group
Roszkowski 2000	Ct plus BSC vs BSC, but the CT is not a second‐line treatment
Ruckdeschel 1994	Phase II study, no control group
Santoro 1994	Phase II study, no control group
Sculier 1994	Phase II study, no control group
Sculier 2000	Phase II study, no control group
Serke 2001	Phase II study, no control group
Sherman 2000	Phase II study, no control group
Socinski 1999	Phase II study, no control group
Spiridonidis 2001	Phase II study, no control group
Stathopoulos 1999	CT second line vs BSC, but it is not a randomized controlled trial
Stathopoulos 1999‐PC	Phase II study, no control group
Stewart 1996	Phase II study, no control group
Takenaka 2001	Phase II study with control group, comparing different CT regimens
Tan 1995	Phase II study, no control group
Treat 2001	Phase II study, no control group
Trongprasert 1999	CT plus BSC vs BSC, but CT not a second‐line treatment
Tsavaris 2001	Phase II study, no control group
Van Kooten 1999	Phase II study, no control group
Van Putten 2001	Phase II study, no control group
Woods 1990	CT vs BSC, but CT not a second‐line treatment

Contributions of authors

MS, XB, FL and CS participated in the identificaction and selection of studies. Data extraction was carried by XB, MS, MN, FL and JM. XB and CS edited the text of the review and the rest of authors made additional comments.

Sources of support

Internal sources

• Hospital de la Santa Creu i Sant Pau. Barcelona, Spain.

• Iberoamerican Cochrane Centre, Spain.

External sources

• Instituto de Salud Carlos III (contract no. 10035), Ministry of Health, Spain.

Declarations of interest

None known

Edited (no change to conclusions)