Combination versus sequential single agent chemotherapy for metastatic breast cancer

  • Review
  • Intervention

Authors


Abstract

Background

Combination chemotherapy can cause greater tumour cell kill if the drug dose is not compromised, while sequential single agent chemotherapy may allow for greater dose intensity and treatment time, potentially meaning greater benefit from each single agent. In addition, sequentially using single agents might cause less toxicity and impairment of quality of life, but it is not known whether this might compromise survival time.

Objectives

To assess the effect of combination chemotherapy compared to the same drugs given sequentially in women with metastatic breast cancer.

Search methods

We searched the Cochrane Breast Cancer Group Specialised Register, using the search terms "advanced breast cancer" and "chemotherapy", MEDLINE and EMBASE on 31 October 2013. The World Health Organization International Clinical Trials Registry Platform and ClinicalTrials.gov were also searched (22 March 2012).

Selection criteria

Randomised controlled trials of combination chemotherapy compared to the same drugs used sequentially in women with metastatic breast cancer in the first-, second- or third-line setting.

Data collection and analysis

Two authors independently extracted data from published trials. Hazard ratios (HR) were derived from time-to-event outcomes where possible, and a fixed-effect model was used for meta-analysis. Response rates were analysed as dichotomous variables (risk ratios (RR)), and toxicity and quality of life data were extracted where available.

Main results

Twelve trials reporting on nine treatment comparisons (2317 patients randomised) were identified. The majority of trials (10 trials) had an unclear or high risk of bias. Time-to-event data were collected for nine trials for overall survival and eight trials for progression-free survival. All 12 trials reported results for tumour response. In the 12 trials there were 1023 deaths in 2317 women randomised. There was no difference in overall survival, with an overall HR of 1.04 (95% confidence interval (CI) 0.93 to 1.16; P = 0.45), and no significant heterogeneity. This result was consistent in the four subgroups analysed (risk of bias, line of chemotherapy, type of schema of chemotherapy, and relative dose intensity). In particular, there was no difference in survival according to the type of schema of chemotherapy, that is whether chemotherapy was given on disease progression or after a set number of cycles. In the seven trials that reported progression-free survival (time to first progression in the sequential arm), 637 women progressed out of the 846 women randomised. There was weak evidence of a higher risk of progression in the combination arm (HR 1.11; 95% CI 0.99 to 1.25; P = 0.08) with no significant heterogeneity. This result was consistent in all subgroups. Overall tumour response rates were higher in the combination arm (RR 1.16; 95% CI 1.06 to 1.28; P = 0.001) but there was significant heterogeneity for this outcome across the trials. In the seven trials that reported treatment-related deaths, there was no significant difference between the two arms, although the CIs were very wide due to the small number of events (RR 1.53; 95% CI 0.71 to 3.29; P = 0.28). The risk of febrile neutropenia was higher in the combination arm (RR 1.32; 95% CI 1.06 to 1.65; P = 0.01). There was no statistically significant difference in the risk of neutropenia, nausea and vomiting, or treatment-related deaths. Overall quality of life showed no difference between the two groups, but only three trials reported this outcome.

Authors' conclusions

There is weak evidence that sequential single agent chemotherapy has a positive effect on progression-free survival, whereas combination chemotherapy has a higher response rate and a higher risk of febrile neutropenia in metastatic breast cancer. There is no difference in overall survival time between these treatment strategies, both overall and in the subgroups analysed. In particular, there was no difference in survival according to the schema of chemotherapy (giving chemotherapy on disease progression or after a set number of cycles) or according to the line of chemotherapy (first-line versus second- or third-line). Generally this review supports the recommendations by international guidelines to use sequential monotherapy unless there is rapid disease progression.

摘要

联合化疗与序贯单药化疗治疗转移性乳腺癌的比较

研究背景

联合化疗需要调整药物剂量,否则会导致肿瘤细胞死亡数过多;而采用单药序贯疗法则可加大剂量强度并延长治疗时间,也就意味着单剂药能够发挥更大效用。另外,连续使用单剂药可能降低化疗毒性并减少对患者生活质量的影响,但关于这种疗法是否会减少存活时间至今尚不可知。

研究目的

评估联合化疗与单药序贯疗法对女性转移性乳腺癌(metastatic breast cancer)的疗效。

检索策略

我们于2013年10月31日以“晚期乳腺癌(advanced breast cancer)”和“化疗(chemotherapy)”为检索词,检索了Cochrane乳腺癌组专业注册库(the Cochrane Breast Cancer Group Specialized Register)、美国国立医学图书馆数据库(MEDLINE)和荷兰医学文摘数据库(EMBASE),并于2012年3月22日检索了世界卫生组织国际临床注册平台与美国临床试验数据库(World Health Organization International Clinical Trials Registry Platform and ClinicalTrials.gov).

标准/纳入排除标准

与单药序贯化疗相比研究联合化疗对处于一线、二线、三线化疗期的转移性乳腺癌女性患者的随机对照试验。

数据收集与分析

两名作者独立对已发表的试验进行资料提取。两名作者从可获取的生存曲线结局中计算风险比(HR),使用固定效应模型(fixed-effect model)进行meta分析。我们将治疗应答率作为二分变量(危险比(RR))进行分析,并从可获取资料中提取了毒性和生活质量的数据。

主要结果

本研究纳入了12项试验,其中包括9组不同治疗方式的比较,共涉及2317名患者。大部分试验(10项)偏倚风险不明或偏倚风险较高。本研究从9项试验获取了关于总生存率的生存曲线数据,从8项试验中获取了无进展生存率的数据。共12项试验均报告了肿瘤反应结局。在这12项试验中,随机选择的2317名女性患者中有1013例死亡。其总生存期没有差异,总风险比为1.04 (95% CI: 0.93-1.16; P = 0.45)。无显著异质性。此结果与4项亚组分析(偏倚风险、化疗方案、化疗计划的类型和相对剂量浓度)的结果一致。特别地,不同的化疗计划在生存率方面没有差异,无论化疗在疾病进展过程中进行,还是经历一定的化疗疗程后。在报告无进展生存率(依据首次恶化时间)的7项试验中,共846位随机选择的女性患者中有637位出现恶化。有较少证据 (HR 1.11; 95% CI 0.99 to 1.25; P = 0.08)显示使用联合化疗的对照组呈现高恶化率,无显著异质性。此结果与亚组分析结果一致。联合化疗组的总体肿瘤反应率较高(RR 1.16; 95% CI 1.06 to 1.28; P = 0.001),但此结果在各项试验中有显著异质性。在报道治疗相关死亡事件的7项试验中两组没有显著差异,但因事件例数少,可信区间跨度很大。联合化疗组的发热性嗜中性白血球减少症的风险较高(RR 1.32; 95% CI 1.06 to 1.65; P = 0.01),但嗜中性白血球减少症、恶心、呕吐或治疗相关性死亡风险无显著统计学差异。两组总体生活质量方面没有差异,但只有三项试验有相关报告。

作者结论

有较少证据显示单药序贯疗法能较为有效地提高无进展生存率,而使用联合化疗则呈现较好的肿瘤治疗疗效,但并发发热性嗜中性白血球减少症的风险也较高。无论是总体上还是亚组分析中,不同治疗方案的总生存时间没有差异。特别地,在不同化疗计划(在疾病发展期进行化疗或在经历一定治疗后进行化疗)或化疗方案(一线化疗与二、三线化疗对比)方面,患者的生存率没有明显差异。国际肿瘤治疗指导中提出,除非疾病进程很快,推荐使用单剂序贯疗法。总体而言,本综述支持了这一观点。 总体而言,本篇综述文章支持国际标准的建议,除病情急剧恶化以外,应使用序贯单药化疗。.

翻译注解

译者:王海玉、刘欣然(北京中医药大学中医英语专业本科生)审校:王海玉、刘欣然、李迅(北京中医药大学循证医学中心)

Plain language summary

Combination (several drugs at the same time) versus sequential chemotherapy (same drugs given one after the other) for metastatic breast cancer

Metastatic breast cancer is not currently a curable disease but one that can be very effectively treated with chemotherapy, endocrine therapy and targeted therapies. Average survival is about two years but some women live for many years longer. It is important to investigate the best way to give chemotherapy to treat metastatic breast cancer in order to optimise survival and quality of life and to minimise the side effects from treatment.

This review investigated whether giving a combination of drugs at the same time was more effective than giving the same drugs one at a time (sequential treatment).

A literature search conducted in October 2013 resulted in 12 randomised controlled studies with 2317 patients that could be included in the analysis. The patients had metastatic breast cancer and either they had not been treated or had received one or two treatments after their diagnosis of metastatic breast cancer. The primary outcomes were overall survival and progression-free survival (time from randomisation to the time of disease progression). Secondarily, we compared the degree the tumour shrunk in response to chemotherapy (overall response rate), toxicity and quality of life.

There was no difference in overall survival between the two groups but we found that when drugs were given one at a time there maybe more time before the tumours grew back again (longer progression-free survival). However, combination chemotherapy caused tumours to shrink more, although this did not result in longer survival than when using sequential chemotherapy. Rates of febrile neutropenia (infection) were higher in the combination arm but there was no difference in the rates of neutropenia (low white blood cells). There was no difference in quality of life between the two groups but there were only three trials that reported this information. Quality of life should be included as an outcome in future trials addressing this question. Overall, the studies did not consistently report the way patients were randomised and this may be a source of bias in the results.

Generally this review supports the recommendations by international guidelines to use sequential monotherapy unless there is rapid disease progression.

Laienverständliche Zusammenfassung

Kombinations- (verschiedene Medikamente zu einem Zeitpunkt)- versus sequenzielle (die selben Medikamente nacheinander verabreicht) ein-Komponenten Chemotherapie bei metastasierendem Brustkrebs

Metastasierender Brustkrebs ist derzeit nicht heilbar, kann aber mit Chemotherapie, Hormontherapie und zielgerichteten Therapien sehr wirksam behandelt werden. Die durchschnittliche Überlebensdauer ist etwa zwei Jahre, aber einige Frauen leben viele Jahre länger. Es ist wichtig zu untersuchen, welches die beste Anwendungsmethode der Chemotherapie bei metastasierendem Brustkrebs ist, um das Überleben und die Lebensqualität zu optimieren und um die Nebenwirkungen der Behandlung zu minimieren.

Dieser Review untersucht, ob die Verabreichung einer Kombination von Medikamenten zur gleichen Zeit wirksamer war als die gleichen Medikamente nacheinander zu verabreichen (sequentielle Behandlung).

Eine Literaturrecherche von Oktober 2013 fand 12 randomisierte kontrollierte Studien mit 2317 Patienten, die in die Analyse miteinbezogen werden konnten. Die Patienten hatten metastasierenden Brustkrebs und sie waren entweder noch nicht behandelt worden, oder sie hatte bereits ein oder zwei Behandlungen nach ihrer Diagnose erhalten. Die primären Endpunkte waren Gesamtüberleben und das progressionsfreie Überleben (Zeit von der Randomisierung bis zum Zeitpunkt des Fortschreitens der Krankheit). Außerdem verglichen wir, wie sehr der Tumor in Reaktion auf Chemotherapie schrumpfte (Gesamtreaktionsrate), Toxizität und Lebensqualität.

Es gab keinen Unterschied im Gesamtüberleben zwischen den beiden Gruppen, aber wir fanden heraus, dass, wenn Medikamente nacheinander verabreicht wurden, möglicherweise mehr Zeit verging, bis die Tumore wieder wuchsen (längeres progressionsfreies Überleben). Jedoch verursachte die Kombinations-Chemotherapie, dass Tumore mehr schrumpften, auch wenn dies zu keiner längeren Überlebenszeit als bei Verwendung von sequentieller Chemotherapie führte. Raten der febrilen Neutropenie (Infektion) waren unter Kombinations-Chemotherapie höher, aber es gab keinen Unterschied in den Raten der Neutropenie (niedrige Anzahl weißer Blutkörperchen). Es gab keinen Unterschied in der Lebensqualität zwischen den beiden Gruppen, aber es gab nur drei Studien, die diese Information berichteten. In zukünftigen Studien sollte die Lebensqualität als Endpunkt zu dieser Frage enthalten sein. Insgesamt haben die Studien nicht konsistent berichtet, wie Patienten randomisiert wurden und dies könnte Bias in den Ergebnissen verursachen.

Im Allgemeinen unterstützt dieser Review die Empfehlungen internationaler Leitlinien sequenzielle Monotherapie zu verwenden, es sei denn es gibt ein schnelles Fortschreiten der Krankheit.

Anmerkungen zur Übersetzung

K. Kunzweiler und I. Töws, Koordination durch Cochrane Schweiz.

Laički sažetak

Kombinirana (nekoliko lijekova u isto vrijeme) naspram sekvencijske kemoterapije (isti lijekovi dani u slijedu jedan za drugim) za metastatski rak dojke

Metastatski rak dojke trenutno nije izlječiva bolest, ali se može vrlo učinkovito liječiti kemoterapijom, endokrinom terapijom i ciljanim terapijama. Prosječno preživljenje je oko dvije godine, ali neke žene žive mnogo godina dulje. Važno je istražiti najbolji način davanja kemoterapije za liječenje metastatskog karcinoma dojke, kako bi se optimizirali preživljenje i kvaliteta života i smanjile nuspojave od liječenja.

Ovaj pregled je istražio je li istovremeno davanje kombinacije lijekova bilo učinkovitije od davanja istih lijekova u slijedu jedan za drugim (sekvencijsko liječenje).

Pretraživanjem literature provedenim u listopadu 2013 pronađeno je 12 randomiziranih kontroliranih studija s 2317 bolesnika, a koje bi mogle biti uključene u analizu. Bolesnici su imali metastatski karcinom dojke i ili nisu bili liječeni ili su primili jedan ili dva tretmana nakon postavljanja dijagnoze metastatskog karcinoma dojke. Primarni ishodi bili su ukupno preživljenje i preživljenje bez progresije bolesti (vrijeme od randomizacije do trenutka progresije bolesti). Sekundarno, usporedili smo veličinu smanjenja tumora u odgovoru na kemoterapiju (stopa ukupnog odgovora), toksičnost i kvalitetu života.

Nije bilo razlike u ukupnom preživljenju između dvije skupine, ali smo otkrili da je kada se lijekovi primijenjuju u slijedu jedan za drugim možda produljeno vrijeme do ponovnog javljanja tumora (duže preživljenje bez progresije bolesti). Međutim, kombinirana kemoterapija dovela je do većeg smanjenja tumora, iako to nije rezultiralo dužim preživljenjem nego kada se koristi sekvencijska kemoterapija. Stope febrilne neutropenije (infekcije) bile su više u skupini s kombiniranom kemoterapijom, ali nije bilo razlike u stopama neutropenije (sniženog broja bijelih krvnih stanica). Nije bilo razlika u kvaliteti života između dviju skupina, ali su samo tri istraživanja koja su prikazala ove informacije. Kvaliteta života treba biti uključena kao ishod u budućim istraživanjima koja se bave ovim pitanjem. Sveukupno, studije nisu dosljedno pokazale način na koji su bolesnici randomizirani i to može biti izvor pristranosti u rezultatima.

Općenito ovaj pregled podržava preporuke međunarodnih smjernica za korištenje sekvencijske monoterapije, osim ako postoji brza progresija bolesti.

Bilješke prijevoda

Cochrane Hrvatska
Prevela: Katarina Vučić
Ovaj sažetak preveden je u okviru volonterskog projekta prevođenja Cochrane sažetaka. Uključite se u projekt i pomozite nam u prevođenju brojnih preostalih Cochrane sažetaka koji su još uvijek dostupni samo na engleskom jeziku. Kontakt: cochrane_croatia@mefst.hr

概要

联合化疗(同时使用予多种药物)与序贯化疗(同种药物接连依次使用)治疗转移性乳腺癌的比较

目前,转移性乳腺癌目前尚是不可治愈的,但是通过化疗、,内分泌治疗及靶向治疗可获得显著疗效。该病平均存活时间约为两年,但是有些女性存活时间可延长长达数年。重点是要研究出治疗转移性乳腺癌的最佳化疗方式,从而延长生存增加存活时间、,优化生活质量并将治疗副作用降至最低,这些尤为重要。

本综述文章探讨同时使用予多种药物联合治疗较每依次使用一种药物治疗(序贯治疗)是否更为有效。

作者于2013年10月进行文献检索,检索获得了12项个随机对照试验共涉及2317 位患者,这些数据结果可能被纳入分析。纳入对象为从未接受过治疗或确诊诊断后接受过一到两次治疗的转移性乳腺癌患者。主要结果为总生存存活时间及无进展生存恶化存活[微软用户1]时间(从随机分组时间至疾病进展恶化时间)。其次,我们对化疗后肿瘤缩小程度(总有效率)与,毒性及生活质量进行了对比。 [微软用户1]翻译此术语时咨询过肿瘤专业的研究生师兄

两组总生存存活时间无差异,但我们发现一次使用一种药物时肿瘤再次形成所需时间较长(无进展生存恶化存活时间较长)。但是,联合化疗可使肿瘤缩小更为显著明显,尽管即便这没有不能使其存活时间长于序贯治疗组。联合化疗组发热性中性粒细胞减少症(感染)发病率较高,但两组在嗜中性粒细胞白血球减少症(白细胞水平低)发病率(白细胞水平低)上无差异。两组患者的生活质量无差异,但是仅有3项试验报告了相关信息。日后进行试验应关注该问题,将生活质量纳入试验结果。总的来说,这些研究没有全部对患者随机分组的方式进行报告本研究的患者是随机的,且没有进行持续报道,这可能是造成结果的偏倚[微软用户1]的一个原因。 [微软用户1]原文中”this may be a source of results ”似乎没有肯定结果中已存在偏倚,不知舍弃 ”a source of “而译成”可能造成偏倚”是否更妥。

总体而言,本篇综述文章支持国际标准的建议,除病情急剧恶化以外,应使用序贯单药化疗。.

翻译注解

译者:王海玉、刘欣然(北京中医药大学中医英语专业本科生)审校:王海玉、刘欣然、李迅(北京中医药大学循证医学中心)

Background

Description of the condition

Breast cancer is the most common type of cancer and the second leading cause of cancer deaths in women. In 2002, there were over 1.1 million new cases and approximately 410,000 deaths from breast cancer, worldwide (Ferlay 2004).

If breast cancer becomes metastatic, it is treatable but not curable. The aims of treatment include improving survival and quality of life. There has never been a randomised trial comparing systemic therapy to best supportive care to demonstrate that systemic therapy can improve survival in women with metastatic breast cancer. Despite this, data from single institution series and population-based registries show that the survival of patients with metastatic breast cancer has improved over time, from approximately 16 months in the early 1990s to beyond 24 months in 2001 (Chia 2007; Giordano 2004), although some women may live for many years (Hayes 1995). This improvement coincides with the availability of new systemic agents (taxanes, aromatase inhibitors, trastuzumab and capecitabine) and a higher proportion of women with metastatic breast cancer receiving systemic therapy (Chia 2007). While longer survival cannot be attributed conclusively to the use of improved systemic therapies it is believed they have contributed. The optimal combination, sequence and timing of systemic agents for metastatic breast cancer remains unresolved. Selection of treatment is based on consideration of the site of recurrence, symptoms, anticipated response to treatment, expected toxicities, quality of life and patient preference. Ultimately most women with metastatic breast cancer will receive chemotherapy because they have hormone receptor negative disease, their disease has become refractory to endocrine therapy, they are symptomatic, or they have rapidly progressive visceral disease (Hortobagyi 1996).

Description of the intervention

Combination chemotherapy can achieve several important objectives that are less probable with single agent chemotherapy. Combination therapy firstly provides increased cell kill within the range of acceptable toxicity for each drug, as long as dosing is not compromised and the tumour is sensitive to each drug; secondly used with an heterogeneous tumour cell population it increases the probability of some cells being responsive compared with a single agent; and thirdly the combination may delay the development of drug resistance if there is rapid cell kill and a reduced cell mass.

The scheduling of multiple drug regimens may influence both the effectiveness and toxicity of the regimen. The Goldie and Coldman model predicts that drug resistant cells may be present even in small tumours (Goldie 1982). They propose alternating non cross-resistant drug combinations as early as possible as a means of eliminating resistant clones. Norton and Day predicted that the sequential use of drug combinations would outperform alternating cycles because no two combinations were likely to be strictly non-cross resistant or have equal cell killing capacity (Day 1986; Norton 1991). In the adjuvant setting, a randomised trial observed that four three-week cycles of doxorubicin followed by eight three-week cycles of cyclophosphamide methotrexate fluorouracil (CMF) in women with high-risk primary breast cancer (four or more positive lymph nodes) was more effective than an alternating schedule of doxorubicin and CMF in terms of disease-free and overall survival (Bonadonna 1995).

A potential limitation of the effectiveness of combination chemotherapy is that dose levels of the component drugs may need to be reduced to limit toxicity. Another limitation of combination treatment is that it may use non-cross resistant agents rapidly. In contrast, these agents given as sequential monotherapy may allow for more cycles of effective treatment. The use of sequential monotherapy may allow for greater dose intensity and allow for a treatment approach that gets maximum time and benefit from each agent.

Optimal dosing of chemotherapy is challenging because of the different effects on normal and tumour tissues. Most cancer chemotherapeutic agents show a steep dose-response curve in the linear phase of tumour growth. A reduction in drug dose when the tumour is in this linear phase of the dose-response curve almost always results in a reduction in the probability of cure before a reduction in antitumour activity is observed. This observation has led cancer clinicians and researchers to conclude that when cure is a reasonable treatment expectation, increasing the dose intensity of treatment will increase the rate and duration of tumour response and improve the survival rate (Hryniuk 1987). However, there are limitations to the effectiveness of dose intensity, as demonstrated by the use of high-dose chemotherapy. A Cochrane Review (Farquhar 2008) of six randomised trials comparing high-dose chemotherapy and autograft with conventional chemotherapy for women with metastatic breast cancer demonstrated no statistically significant evidence of benefit in overall survival. Increased dose intensity also comes at a cost of increased toxicity to normal tissues. If palliation is the primary goal of treatment, and anticipated survival is limited, then toxicity and quality of life become important factors when deciding on a treatment regimen.

Combination therapy can also refer to the combination of chemotherapy with a biological agent. This regimen is most commonly used in women with human epidermal growth factor receptor 2-positive (HER2-positive) metastatic breast cancer, for example a taxane and trastuzumab or capecitabine and lapatinib. This review focuses only on the combination of chemotherapeutic agents compared to the same agents given as sequential monotherapy.

How the intervention might work

The question of interest is whether patients receiving combination chemotherapy regimens have better health outcomes, including longer survival and less toxicity, than for sequential single agent chemotherapy. Sequential single agent use has evolved to maintain dose intensity while limiting toxicity, and without sacrificing survival. Sequential therapy may refer to the consecutive administration of single agents with each agent introduced after disease progression or it may refer to a planned sequence of single agents given without interruption between the regimens.

Why it is important to do this review

The comparison of combination chemotherapy with the same agents given sequentially as monotherapy is a clinically important question that addresses a common clinical scenario facing medical oncologists treating women with metastatic breast cancer. This is because the role of combination chemotherapy in metastatic breast cancer is still unresolved. Combination chemotherapy is often given to obtain a response in immediately life-threatening circumstances whereas sequential monotherapy is given for less urgent treatment needs. A Cochrane review showed that combination chemotherapy gave a statistically significant improvement in response rate, time to progression and overall survival but was more toxic compared to single agent chemotherapy (Carrick 2009). Few studies included in this review formally compared women's quality of life between combination and single agent therapy. While the review by Carrick et al tested the benefit of two agents given in combination compared to a single agent it did not address the benefit of two agents given in combination versus the same drugs given as sequential monotherapy. It is not known if it is more beneficial to give a combination of drugs at the same time or to give the same drugs as single agents administered sequentially. Furthermore, regarding sequential single agents, there is still debate about whether single agents should be given until disease progression (when the drugs are changed) or whether a planned sequence of single agents should be given that involves changing the drug before resistance develops and disease progression occurs (see Objectives).

Objectives

To assess the effect of combination chemotherapy compared to the same drugs given sequentially in women with metastatic breast cancer.

We compared the use of single chemotherapy agents given sequentially with regimens using a combination of the same agents for the management of women with metastatic breast cancer to determine their effect on survival. This involved examining two different chemotherapy schemata:

  • schema 1: the combination given until disease progression versus sequential single agents given on disease progression, e.g. cycles of combination AB until disease progression versus cycles of drug A until disease progression followed by cycles of drug B until disease progression;

  • schema 2: the combination given for a set number of cycles versus sequential single agents given as a set number of cycles, e.g. eight cycles of AB versus four cycles of A followed by four cycles of B.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled clinical trials that compared a combination of drugs with the same single chemotherapy agents given sequentially for women with metastatic breast cancer.

Types of participants

Women diagnosed with all types of advanced (metastatic) breast cancer. Metastatic breast cancer was defined as stage IV by the TNM classification of malignant tumours. Adults aged 18 years and above were included.

Types of interventions

The intervention was any chemotherapy regimen containing a combination of chemotherapeutic agents given as first-line or second-line and third-line treatment.

The comparison was sequential single agent use of the drugs in the combination chemotherapy regimen given as first-line or second-line and third-line treatment. The sequential single agents may have been administered according to either schema 1 or schema 2 as described under Objectives.

Trials were excluded that included a single agent chemotherapeutic agent alone (drug A alone).

Types of outcome measures

Primary outcomes
  1. Overall survival

  2. Progression-free survival (or time to progression)

Secondary outcomes
  1. Tumour response rate

  2. Quality of life measures

  3. Toxicity

Outcome definitions
  1. Overall survival (OS): time from date randomised to date of death (any cause).

  2. Progression-free survival (PFS): time from date randomised to date of progression or death from any cause. We used time to progression (TTP) (time from randomisation to date of progression) when PFS was not reported.

  3. Overall response rate (RR): the percentage of patients with a complete or partial response. Complete and partial responses were defined according to Response Evaluation Criteria in Solid Tumors (RECIST) criteria (Therasse 2000). That is, a complete response was the disappearance of all target lesions and a partial response was at least a 30% decrease in the sum of the longest diameter of target lesions. The composite response rate was used for sequential single agent chemotherapy. This was calculated as the percentage of patients responding at some point during any of the single agent regimens.

  4. Quality of life was defined as an expression of well-being and measured using a validated scale (e.g. Short Form-36 (SF-36), European Organisation for Research and Treatment of Cancer (EORTC), Functional Assessment of Cancer Therapy (FACT)).

  5. Toxicities: restricted in our analyses to serious side effects as classified by the World Health Organization (WHO) or National Cancer Institution Criteria (NCIC) grading 3 or more.

Search methods for identification of studies

See: Breast Cancer Group methods used in reviews.

Only trials published in English were included.

Electronic searches

We searched the following databases.

  1. The Cochrane Breast Cancer Group (CBCG) Specialised Register (searched on 31 October 2013). Details of search strategies used by the CBCG for the identification of studies and the procedures used to code references are outlined in the CBCG's module at www.mrw.interscience.wiley.com/cochrane/clabout/articles/BREASTCA/frame.html). Trials with the key words 'advanced breast cancer', 'metastatic breast cancer', 'combination chemotherapy', 'sequential single agent chemotherapy' and 'single agent chemotherapy' were extracted and considered for inclusion in the review.

  2. MEDLINE (via OvidSP) (searched on 31 October 2013), see Appendix 1.

  3. EMBASE (via Embase.com) (searched on 31 October 2013), see Appendix 2.

  4. The WHO International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/Default.aspx) for prospectively registered and ongoing trials (searched on 22 March 2012), see Appendix 3.

  5. ClinicalTrials.gov (http://clinicaltrials.gov/ct2/search/advanced) (searched on 22 March 2012), see Appendix 4.

Searching other resources

(a) Bibliography searching

We tried to identify further studies from reference lists of identified relevant trials or reviews. A copy of the full article for each reference reporting a potentially eligible trial was obtained, where possible. Where this was not possible, we attempted to make contact with the authors to provide additional information.

(b) Handsearching of journals

We handsearched the database of abstracts on the treatment of metastatic breast cancer from the annual meetings of the American Society of Clinical Oncology (1987 to 2000; 2003) and San Antonio Breast Cancer Conference (1981 to 2002). A list of journals currently being handsearched by The Cochrane Collaboration is available at the US Cochrane Center Handsearch master list page (http://us.cochrane.org/master-list).

Data collection and analysis

We performed the analysis in accordance with the guidelines published in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 (Higgins 2011).

Selection of studies

Two authors (RD, MT) independently screened the results of the search. We applied the selection criteria described above to each trial and recorded any reasons for exclusion. A full copy of trials likely to be included in the review were retrieved and read. For unpublished trials, we assessed the available information from conference proceedings. Any discrepancies were resolved by a third author. Only manuscripts published in English were included.

Data extraction and management

Data were extracted from trial publications by two authors (RD, MT). A standardised data extraction form was used to record details of study design, participants, setting, interventions, follow-up, quality components, efficacy outcomes, toxicity and quality of life. The form is available from the authors upon request. A third author (NW) was available to resolve any discrepancies regarding the extraction of quantitative data and a fourth author (AB) was available to resolve any discrepancies regarding the assessment of the quality of the randomised controlled trials. When a trial was presented in abstract form, we sought further information, as necessary, from the internet, contacting the authors, or checking for the next best available resource or publication.

In the case of published papers, we also contacted the authors for additional information if it was not reported in the paper. For studies with more than one publication, we extracted data from all the publications, however we considered the final or updated version of the trial as the primary reference.

Assessment of risk of bias in included studies

Quality assessment was based on the assessment of a list of procedures related to the design and reporting of trials that are known to minimise the risk of bias. Therefore, trials were assessed by the degree they addressed each of the following items.

Random sequence generation
  1. Low risk: if the allocation sequence was generated by a computer or random number table. Drawing of lots, tossing of a coin, shuffling of cards or throwing dice were considered as adequate if a person who was not otherwise involved in the recruitment of participants performed the procedure.

  2. Unclear risk: if the trial was described as randomised but the method used for the allocation sequence generation was not described.

  3. High risk: if the trial used a system involving dates, names, admittance numbers or alternating allocation.

Allocation concealment
  1. Low risk: if the allocation of patients involved a central independent unit; sequentially numbered identical drug containers; or serially numbered, sealed, opaque envelopes.

  2. Unclear risk: if the trial was described as randomised but the method used to conceal the allocation was not described.

  3. High risk: if the allocation sequence was known to the investigators who assigned participants; the envelopes did not have all three safeguards; or if a non-random sequence was used.

Blinding of participants, personnel and outcome assessors
  1. Low risk: if the trial was described as double-blind or single-blind i.e. double-blind (method described and technique used meant that neither the participant nor the care provider or assessor knew which treatment was given) and single-blind (participant or care provider or assessor was aware of the treatment given).

  2. Unclear risk: if the trial was described as double-blind or single-blind but the method of blinding was not described.

  3. High risk: if the trial was open label (all parties were aware of the treatment).

Incomplete outcome data addressed

The completeness of outcome data were described for each main outcome. The description included whether attrition and exclusions were reported, the numbers in each intervention group (compared with total randomised participants), reasons for attrition and exclusions and any re-inclusions in the analysis.

  1. Low risk: no missing data; or reasons for missing data were not related to outcome, were balanced across groups with similar reasons or numbers of participants missing, or plausible effect size was not enough to change observed effect.

  2. Unclear risk: attrition and exclusions reported but reasons were not included.

  3. High risk: incomplete outcome data were not addressed.

Free of selective outcome reporting

We stated how the possibility of selective outcome reporting was examined by the review authors and what was found.

  1. Low risk: all pre-specified outcomes were reported in a pre-specified way.

  2. Unclear risk: expected outcomes were reported but not pre-specified.

  3. High risk: outcomes were not reported as planned or expected; or the outcomes were reported in a way they could not be used in the review, e.g. only direction of effect and significance.

Other sources of bias

Comments were made regarding the following.

  • If the treatment groups were similar at baseline regarding the most important prognostic factors.

  • If the trial was stopped early using a data-dependent process (including formal stopping rule).

  • If analysis was by intention to treat (or enough data were provided to replicate such an analysis i.e. number of withdrawals, dropouts and losses to follow-up in each group).

We contacted authors of the primary trial reports, when necessary, to clarify data and provide missing information. We entered the characteristics and outcomes of the included trials and details of the excluded trials into our database.

Each selected study was assessed as low, unclear or high risk of bias based on assessments of the domains described above (random sequence generation, allocation concealment, blinding, incomplete data addressed, free of selective outcome reporting and other sources of bias). Since the selected studies had different levels of risk of bias (with most being of high or unclear risk of bias), we conducted a primary analysis including all trials and then performed a sensitivity analysis to show how the conclusions were affected if studies of unclear or high risk of bias were excluded.

Measures of treatment effect

We extracted the hazard ratio (HR) and associated variances for overall survival and progression-free survival (or time to progression) directly from the trial publications.

We obtained a pooled HR through the generic inverse variance approach, from the log HR and the standard error of the log HR, using fixed-effect and random-effects models along with assessments of heterogeneity (Deeks 2006). The pooled HR represented the overall risk of an event on combination chemotherapy regimens compared to single agent sequential chemotherapy.

We reported ratios of treatment effects for time-to-event outcomes so that HRs less than 1.0 favoured combination chemotherapy regimens and values greater than 1.0 favoured single agent sequential chemotherapy regimens. The forest plots for overall survival and progression-free survival presented the hazard ratio (HR) and 95% confidence interval (CI).

We analysed an indicator of response as a dichotomous variable (complete or partial response versus stable disease or no response) and derived a pooled relative risk (RR). We reported responses on an intention-to-treat basis. If the majority of trials only reported results based on assessable (not randomised) patients, we used assessable responses. We reported ratios of treatment effects for response so that risk ratios (RRs) more than 1.0 favoured combination chemotherapy regimens and values less than 1.0 favoured single agent sequential chemotherapy regimens.

Unit of analysis issues

Four trials were three-arm studies (Cresta 2004; Koroleva 2001; Sledge 2003; Soto 2006). Two out of three of the arms in each of these trials were combined. For Cresta 2004, there was one combination arm of doxorubicin and docetaxel, a single agent arm where these two drugs were given on an alternating schedule, and another single agent arm where the drugs were given sequentially. The two single agent arms were combined in this trial. In Koroleva 2001 there were two combination arms using doxorubicin and docetaxel but at different doses, and one sequential single agent arm. The two combination arms were combined in this trial. In Sledge 2003 there was one combination arm that included doxorubicin and paclitaxel, one single agent arm that started with doxorubicin, and another single agent arm that began treatment with paclitaxel. The two single agent arms were combined in this arm. In Soto 2006 there was one sequential single agent arm that included capecitabine followed by a taxane (either docetaxel or paclitaxel), a combination arm that included capecitabine and docetaxel, and another combination arm that included capecitabine and paclitaxel. The two combination arms were combined in this trial.

Dealing with missing data

If data were missing, we made multiple attempts to contact the original investigators. This was successful in the case of four studies (Alba 2004; Fountzilas 2001; Park 2010; Sledge 2003). When we were unable to obtain data, two authors extracted data from published Kaplan-Meier curves (Tierney 2007). If there was a discrepancy between authors, we took an average of the two values at each time point and used the averaged values to generate a new HR. To allow for premature follow-up, we adjusted the numbers at risk based upon estimated minimum and maximum follow-up times. If these were not reported we estimated minimum follow-up using the estimated time taken to complete treatment. We estimated maximum follow-up using the last event reported in the relevant time-to-event curve. We have presented these follow-up estimates in the Characteristics of included studies table under 'Notes'.

Assessment of heterogeneity

We included all outcomes available from the individual studies in the meta-analysis with heterogeneity reported by the Q (Chi2) and the I2 statistics. Statistical significance of the Q statistic was judged as P < 0.10 because of the low statistical power of the test. The I2 statistic indicates the per cent variability due to between-study (or inter-study) variability, as opposed to within-study (or intra-study) variability. An I2 statistic greater than 50% was considered large (Higgins 2002), however this was considered alongside the magnitude and direction of effects and the P value from the Chi2 test for heterogeneity.

Assessment of reporting biases

Reporting bias was assessed using the Cochrane Collaboration's assessment tool (Table 8.6a) in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Data synthesis

We used a fixed-effect model for meta-analysis. HRs and RRs were calculated with a generic inverse variance fixed-effect model analysis. Response rates were analysed as dichotomous variables and we derived a pooled relative risk. We used the Mantel-Haenszel method with zero-cell corrections for the analysis of treatment-related deaths because this was a rare event, however we used the inverse variance method to pool the RRs for the other toxicity outcomes (neutropenia, febrile neutropenia and nausea and vomiting) because these outcomes are not rare events. We also analysed all toxicity outcomes using conditional logistic regression.

Subgroup analysis and investigation of heterogeneity

As outlined in the protocol, we pre-specified three subgroups. We extracted data on these subgroups from all trial publications, where available.

  1. Schedule of single agent: schema 1 versus schema 2 (see Objectives).

  2. Dose of chemotherapy: trials where the dose used in the combination regimen was the same as the doses of the single agents compared to trials where the doses were not equivalent in the two arms of the study.

  3. Line of chemotherapy for metastatic breast cancer (first-line versus second-line or third-line chemotherapy).

Sensitivity analysis

A sensitivity analysis comparing the outcomes using studies at low risk of bias with outcomes of studies of unclear or high risk of bias was performed. Trials were assessed for quality by the degree to which they addressed the following seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete outcome data, selective outcome reporting, and if the analysis was by intention to treat. Each selected study was assessed as low, unclear or high risk of bias based on assessing each of these domains. If the majority of the seven categories (that is four or more out of seven) were considered at unclear or high risk of bias the trial was considered to be at risk of bias.

Results

Description of studies

See: Characteristics of included studies and Characteristics of excluded studies.

Results of the search

We searched the Cochrane Breast Cancer Group Specialised Register, EMBASE, MEDLINE, the WHO ICTRP and ClinicalTrials.gov. From these databases 2299 were coded as references to studies of advanced breast cancer and chemotherapy (Figure 1). Of these 160 were duplicate studies, leaving 2139 remaining studies. Of these 18 were trials that compared combination chemotherapy with drugs given as sequential single agents. Two were excluded because the drugs given in the sequential arm were different from those in the combination arm. Two studies (Campone 2013; Zhang 2013) were categorised as 'study awaiting classification' because the trials were completed but with insufficient data for inclusion at present. The results from these two small trials will be included in an updated version of this review. The remaining 14 references reported the results of 14 randomised trials. One report (Chlebowski 1989) combined the results from two separate but similar trials, both of which were reported earlier (Chlebowski 1979; Smalley 1976). Both of these publications were therefore excluded from the review. The 12 eligible trials reported on nine treatment comparisons, summarised in Table 1. The details of the chemotherapy protocols used are detailed in Table 2. The trials included in the forest plots are labelled by primary author and date of publication. Nine of the trials were published in full while three were only available in abstract form.

Figure 1.

Results of search strategy applied 31 October 2013 for combination versus sequential single agent chemotherapy for metastatic breast cancer.

Table 1. Summary of treatment comparisons
CombinationSequentialNumber of trials
doxorubicin + docetaxeldoxorubicin → docetaxel

Alba 2004

Cresta 2004(included alternating regimen)

Koroleva 2001 (included 2 combination arms with different doses)

5-fluorouracil + cyclophosphamide + vincristine5-fluorouracil → cyclophosphamide → vincristine Baker 1974
capecitabine + docetaxel or paclitaxelcapecitabine → docetaxel or paclitaxel

Beslija 2006

Soto 2006

5-fluorouracil + cyclophosphamide + prednisone + thyroxine or vincristine5-fluorouracil → cyclophosphamide → thyroxine or vincristine → prednisone Chlebowski 1989
epirubicin + paclitaxelepirubicin → paclitaxel Conte 2004
epirubicin + paclitaxeldose dense epirubicin→ paclitaxel Fountzilas 2001
gemcitabine + vinorelbinegemcitabine → vinorelbine Park 2010
doxorubicin + paclitaxeldoxorubicin → paclitaxel or paclitaxel → doxorubicin Sledge 2003
docetaxel + gemcitabinedocetaxel → gemcitabine Tomova 2010
Table 2. Chemotherapy details
Trial nameArm IArm II
Alba 2004

Arm I: AT=

Doxorubicin 50 mg/m2 and docetaxel 75 mg/m2 both on day 1.

Cycles repeated every 21 days for 6 cycles.

If prior anthracyclines:

given 3 cycles of AT at above doses followed by 3 cycles of docetaxel 100 mg/m2

Arm II: A→T=

Doxorubicin 75 mg/m2 intravenously day 1 for 3 cycles followed by docetaxel 100 mg/m2 intravenously day 1 for 3 cycles.

If prior anthracyclines given 2 cycles of doxorubicin 75 mg/m2 followed by 4 cycles of docetaxel 100 mg/m2.

Cycles repeated every 21 days

Baker 1974

Arm I: FCV=

5-fluorouracil 7.5 mg/kg intravenously days 1-5 plus cyclophosphamide 4 mg/kg intravenously days 1 to 5 plus vincristine 0.015 mg/kg intravenously days 1 and 8.

Cycles repeated every 28 days until disease progression

Arm II: F→C→V=

5-fluorouracil 15 mg/kg intravenously days 1 to 5 every 28 days until disease progression then

cyclophosphamide 8 mg/kg intravenously days 1 to 5 every 28 days until disease progression then

vincristine 0.02 mg/kg intravenously weekly until disease progression

Beslija

2006

Arm I: XT=

Capecitabine (Xeloda, X) 1250 mg/m2 twice daily orally from days 1 to 14 and docetaxel (Taxotere, T) 75 mg/m2 intravenously day 1.

Cycle repeated every 21 days until disease progression

Arm II: T→X=

Docetaxel 100 mg/m2 intravenously day 1 until disease progression then

capecitabine 1250 mg/m2 twice daily orally from days 1 to 14 until disease progression.

Cycles repeated every 21 days

Chlebowski 1989

WCSG Arm I: CMFTP=

Cyclophosphamide 2 mg/kg/day orally, plus 5-fluorouracil 15 mg/kg every 2 weeks intravenously from day 1, plus methotrexate 30 mg/m2 every 2 weeks intravenously beginning on day 8, plus prednisone 0.5 mg/kg/day orally, plus triiodothyronine 0.005 mg daily. Cycle repeated until disease progression.

SECSG Arm I:

Cyclophosphamide 400 mg/m2 intravenously day 1 every 28 days, plus 5-fluorouracil 400 mg/m2 intravenously day 1 and day 8 every 28 days, plus methotrexate 30 mg/m2 intravenously day 1 and day 8 every 28 days, plus vincristine 1 mg/m2 intravenously day 1 and day 8 every 28 days, plus prednisone 80mg orally daily from days 1 to 7 every 28 days. Cycle repeated until disease progression.

or

Cyclophosphamide 100 mg orally daily, plus 5-fluorouracil 400 mg/m2 intravenously weekly, plus methotrexate 20 mg/m2 orally weekly, plus vincristine 1 mg/m2 intravenously weekly, plus prednisone 45 mg orally daily for 14 days, then 30 mg daily for 14 days then 15 mg daily for 28 days. Cycle repeated until disease progression

WCSG Arm II: F→C→TP→M=

5-fluorouracil 15 mg/kg weekly intravenously from day 1 for a minimum of 4 weeks until disease progression then

cyclophosphamide 2 mg/kg/day orally for a minimum of 4 weeks until disease progression then

triiodothyronine 0.005 mg daily plus prednisone 0.5 mg/kg/day for a minimum of 6 weeks until disease progression then

methotrexate 30 mg/m2 intravenously weekly for a minimum of 4 weeks

SECSG Arm II: F→MC→V→P=

5-fluorouracil 600 mg/m2 intravenously weekly until disease progression then

methotrexate 20 mg/m2 orally biweekly until disease progression then

cyclophosphamide 100 mg/m2 orally daily until disease progression then

vincristine 1 mg/m2 intravenously weekly until disease progression then

prednisone 45 mg orally daily for 14 days then 30 mg daily for 14 days then 15 mg daily for 30 days

Conte

2004

Arm I: EP=

Epirubicin 90 mg/m2 plus paclitaxel 200 mg/m2 intravenously day 1.

Cycles repeated every 21 days for 8 cycles

Arm II: E→P=

Epirubicin 120 mg/m2 intravenously day 1 for 4 cycles then

paclitaxel 250 mg/m2 intravenously day 1 for 4 cycles.

Cycles repeated every 21 days

Cresta

2003

Arm I: AT=

Doxorubicin 60 mg/m2 plus docetaxel 60 mg/m2 intravenously day.

Cycles repeated every 21 days for 8 cycles

Arm II: A→T (sequential regimen)=

Doxorubicin 75 mg/m2 intravenously on day 1 for 4 cycles followed by docetaxel 75 mg/m2 intravenously on day 1 for 4 cycles. Cycles repeated every 21 days. Maximum 8 cycles.

Arm III: T then A (alternating regimen)=

Docetaxel 100 mg/m2 intravenously on day 1 for 4 cycles alternating with doxorubicin 75 mg/m2 intravenously on day 1 for 4 cycles. Cycles repeated every 21 days. Maximum 8 cycles

Fountzilas

2001

Arm I: P=

Epirubicin 80 mg/m2 plus paclitaxel 175 mg/m2 intravenously day 1.

Cycles repeated every 21 days for 6 cycles

Arm II: E→P

Epirubicin 110 mg/m2 intravenously day 1 for 4 cycles followed by

paclitaxel 225 mg/m2 intravenously day 1 for 4 cycles.

Cycles repeated every 14 days with G-CSF support (dose dense regimen)

Koroleva

2001

Arm I: AT=

Doxorubicin 50 mg/m2 plus docetaxel 75 mg/m2 intravenously day 1.

Cycles repeated every 21 days for 8 cycles

Arm II: T→A=

Docetaxel 100 mg/m2 intravenously day 1 for 4 cycles followed by

doxorubicin 75 mg/m2 intravenously day 1 for 4 cycles.

Cycles repeated every 21 days.

Arm III: AT=

Doxorubicin 60 mg/m2 plus docetaxel 60 mg/m2 intravenously day 1.

Cycles repeated every 21 days for 8 cycles

Park

2010

Arm I: GV=

Gemcitabine 1,000 mg/m2 plus vinorelbine 25 mg/m2 intravenously days 1 and 8.

Cycles repeated every 21 days until disease progression

Arm II: G→V=

Gemcitabine 1,200 mg/m2 intravenously on days 1 and 8 until disease progression then

vinorelbine 30 mg/m2 intravenously days 1 and 8 until disease progression.

Cycles repeated every 21 days

Sledge

2003

Arm I: AT=

Doxorubicin 50 mg/m2 plus paclitaxel 150 mg/m2 over 24 hours intravenously day 1.

Cycles repeated every 21 days until disease progression

Arm II: A (→P)=

Doxorubicin 60 mg/m2 intravenously day 1 for a maximum of 8 cycles. If disease progressed crossed over to paclitaxel 175 mg/m2 intravenously over 24 hours on day 1.

Cycles repeated every 21 days.

Arm III: P (→A)=

Paclitaxel 175 mg/m2 intravenously over 24 hours on day 1. If disease progressed crossed over to doxorubicin 60 mg/m2 intravenously day 1.

Cycles repeated every 21 days

Soto

2006

Arm I: XT=

Capecitabine 825 mg/m2 twice daily orally from days 1 to 14 plus docetaxel 75 mg/m2 intravenously on day 1 until disease progression.

Cycles repeated every 21 days

Arm II: X→T (docetaxel or paclitaxel)=

Capecitabine 1250 mg/m2 twice daily orally days 1 to 14 until disease progression then

docetaxel 100 mg/m2 or paclitaxel 175 mg/m2 intravenously on day 1 until disease progression.

Cycles repeated every 21 days.

Arm III: XP=

Capecitabine 825 mg/m2 twice daily orally from days 1 to 14 plus paclitaxel 175 mg/m2 intravenously on day 1 until disease progression.

Cycles repeated every 21 days

Tomova

2010

Arm I: TG=

Docetaxel 75 mg/m2 intravenously on day 1 plus gemcitabine 1,000 mg/m2 intravenously on days 1 and 8.

Cycles repeated every 21 days for 8 cycles

Arm II: T→G=

Docetaxel 100 mg/m2 intravenously on day 1 for 4 cycles followed by

gemcitabine 1,250 mg/m2 intravenously on days 1 and 1 for 4 cycles.

Cycles repeated every 21 days

Of the 12 eligible trials, only one allowed crossover from one arm to the other. This occurred in Sledge 2003 between the two single agent arms but did not include the combination arm. At the time of progression, patients in the doxorubicin arm crossed over to paclitaxel, and those in the paclitaxel arm crossed over to doxorubicin. These two single agent arms were therefore equivalent to two sequential single agent arms and were analysed as such in this review.

Risk of bias in included studies

See: 'Risk of bias' graph (Figure 2).

Figure 2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Sequence generation was clearly reported in three trials (Baker 1974; Conte 2004; Fountzilas 2001) and was unclear in the remaining nine trials (Alba 2004; Beslija 2006; Chlebowski 1989; Cresta 2004; Koroleva 2001; Park 2010; Sledge 2003; Soto 2006; Tomova 2010). Two trials had adequate allocation concealment (Chlebowski 1989; Conte 2004). In nine trials allocation concealment was unclear (Alba 2004; Beslija 2006; Cresta 2004; Fountzilas 2001; Koroleva 2001; Park 2010; Sledge 2003; Soto 2006; Tomova 2010). However, in all of these trials there were no large differences in baseline characteristics between the two treatment arms. In one trial (Baker 1974) there was no allocation concealment: "all patients treated with chemotherapy during this time period were entered onto the study, despite the fact that the physician might have known which branch the patient would be entered into prior to the initiation of therapy...this...probably accounted for the difference in size of the two divisions".

Blinding

In all but one trial it was unclear if there was blinding of participants or personnel. In the one trial where it was clear, there was not blinding of study personnel (Baker 1974). In two trials there was blinding of outcome assessors (Fountzilas 2001; Park 2010) but in the remaining 10 trials it was unclear if there was blinding of the outcome assessors.

Incomplete outcome data

Completeness of outcome data was adequate in 10 trials. The remaining two trials (Koroleva 2001; Soto 2006) reported attrition rates or exclusions but reasons were not provided so complete reporting of outcomes was judged as unclear.

Selective reporting

For eight trials, completeness of all outcome data was adequate for use in the review. Outcome data were either directly extracted from the paper (Beslija 2006; Park 2010), provided after contacting the authors (Alba 2004; Fountzilas 2001; Sledge 2003) or calculated from Kaplan-Meier curves in the case of HRs (Baker 1974; Conte 2004; Tomova 2010). In one trial the majority of outcomes had adequate completeness although there was one outcome which was unclear, progression-free survival (Chlebowski 1989). The remaining three trials had inadequate completeness of outcome data for both overall and progression-free survival (Cresta 2004; Koroleva 2001; Soto 2006). Only the direction of effect and P value were provided in these trials. All trials reported the response rate and toxicity. Only three trials reported quality of life (Conte 2004; Park 2010; Sledge 2003).

Three trials in the review were in the form of abstracts (Beslija 2006; Koroleva 2001; Soto 2006) so in the background section they hypothesised that combination chemotherapy prolongs overall survival, progression-free survival and response rate, which implied that these were pre-specified outcomes. Toxicity was not usually pre-specified. The abstracts did not describe how the outcomes were reported and this was not explained in the methods section. However, reporting of outcomes was adequate in Beslija 2006 but not in Koroleva 2001 or Soto 2006.

Other potential sources of bias

The main other potential source of bias was that in seven trials the analysis was not by intention to treat (Baker 1974; Conte 2004; Koroleva 2001; Park 2010, Sledge 2003; Soto 2006; Tomova 2010). To overcome this problem, when we extracted the data from these trials we analysed the results by intention to treat.

The review has been amended (21 November 2014) to highlight an important source of bias which was not sufficiently acknowledged in the original review. This source of bias is the comparison of progression-free survival in the combination arm to progression-free survival in the sequential single agent arm. This problem arises for the trials that use Schema 1 (where drugs changed on disease progression). The issue does not arise for drugs that use Schema 2 (where the drugs are changed after a set number of cycles before disease progression has occurred). There are four trials that use Schema 1 in the review: Baker 1974, Beslija 2006, Park 2010 and Sledge 2003. Among the schema 1 trials, Baker 1974 and Beslija 2006 reported time to first progression in the sequential single agent arm compared to time to first progression in the combination arm. However, the other two Schema 1 trials, Park 2010 and Sledge 2003, measured time to progression from the date of randomization to the date of progression or death after second progression for the single agents but after first progression for the combination arm. This biases progression-free survival in favour of sequential therapy. Our original progression-free survival results (now labelled as Analysis 2.2, Analysis 9.2, Analysis 16.2, Analysis 23.2, Analysis 30.2) are heavily weighted to the Sledge 2003 results because it is the largest trial. This means that in our original progression-free survival analysis it is likely that we have over-estimated the benefit of sequential single agent chemotherapy on progression-free survival. In this amendment, we have included an updated progression-free survival analysis that compares time to progression in the combination arm to time to first progression in the sequential arm. For this analysis Park 2010 was excluded (data were not available) and new data were included for Sledge 2003 after contacting the authors for this new information (Analysis 2.1, Analysis 9.1, Analysis 16.1, Analysis 23.1, Analysis 30.1). We have retained the original progression-free survival analysis for comparison.

Effects of interventions

The ratios of treatment effects for time-to-event outcomes were reported so that hazard ratios (HRs) less than 1.0 favoured the combination arm and HRs more than 1.0 favoured the sequential arm. We have reported ratios of treatment effects for response so that a risk ratio (RR) of more than 1.0 favoured the combination arm.

Of the 12 trials, the primary outcome in most trials (eight out of 12) was the objective response rate (Beslija 2006; Conte 2004; Cresta 2004; Fountzilas 2001; Koroleva 2001; Park 2010; Sledge 2003; Soto 2006). In two trials the primary outcome was overall survival (Baker 1974; Chlebowski 1989); in one trial the primary outcome was time to progression (Tomova 2010); and in one trial the primary outcome was toxicity (Alba 2004). Most trials had a superiority design, although one trial used a noninferiority design (Conte 2004). See: the Characteristics of included studies, Table 1 on treatment comparisons used in the included studies, and Table 2 detailing the chemotherapy regimens used in the included studies.

The results for the pooled data, from all trials that had extractable data, are detailed first followed by four subgroup analyses.

All trials

Overall survival - all trials

There are 12 trials in this review, which included nine treatment comparisons. A total of 2317 women were randomised. Time-to-event data were extractable for overall survival from nine trials, which included nine treatment comparisons. This represented 77% of all patients randomised (1786/2317). There were 1023 deaths in 2317 women randomised. The overall HR for overall survival was 1.04 (95% CI 0.93 to 1.16; P = 0.45) (Figure 3) and there was no statistically significant heterogeneity (I2 = 24%; P = 0.23).

Figure 3.

Forest plot of comparison: 1 Overall survival, outcome: 1.1 Overall survival (all trials).

Progression-free survival - all trials

In the amended analysis (comparing time to progression in the combination arm to time to first progression in the sequential single agent arm), progression-free survival data were available from seven trials, which included seven treatment comparisons. Park 2010 was excluded because these data were not available. There were 637 women who progressed out of 846 randomised. There was weak evidence that combination chemotherapy had a higher risk of progression that sequential single agent chemotherapy with an overall HR of 1.11 (95% CI 0.99 to 1.25; P = 0.08) and there was no statistically significant heterogeneity (I2 = 21%; P = 0.27) (Analysis 2.1; Figure 4).

Figure 4.

Forest plot of comparison: 2 Progression-free survival, outcome: 2.1 Progression-free survival (amended analysis).

In the original analysis, progression-free survival data were available from eight trials, which included eight treatment comparisons. This represented 68% of all patients randomised (1564/2317). There were 678 women who progressed out of 886 randomised. Combination chemotherapy had a higher risk of progression than sequential single agent chemotherapy with an overall HR of 1.16 (95% CI 1.03 to 1.31; P = 0.01; Analysis 2.2) and there was no statistically significant heterogeneity (I2 = 26%; P = 0.22).

Overall tumour response rate - all trials

Tumour response rates for assessable patients were available for all 12 trials in the review. This represented 92% of all patients randomised (2140/2317). Combination chemotherapy was associated with a higher overall tumour response rate compared to sequential single agent chemotherapy with an RR of 1.16 (95% CI 1.06 to 1.28; P = 0.001) (Figure 5). There was statistically significant heterogeneity for this outcome across the trials (I2 = 72%; P < 0.0001).

Figure 5.

Forest plot of comparison: 3 Overall response, outcome: 3.1 Overall response (all trials).

Toxicity - all trials

Treatment-related deaths were reported in seven trials. This represented 39% of all patients randomised (902/2317). There were 24 treatment-related deaths of which 16 occurred in the combination arm and eight occurred in the sequential arm. This was not statistically significant (RR 1.53; 95% CI 0.71 to 3.29; P = 0.28) (Analysis 4.1). There was no statistically significant heterogeneity (I2 = 29%; P = 0.22).

All 12 trials provided extractable data on grade III or IV toxicity. Of these, 12 reported on neutropenia or leukopenia (2351 patients), nine on febrile neutropenia (2328 patients), and eight on nausea and vomiting (1719 patients). Regarding neutropenia and leukopenia, data on grade III or IV leukopenia was included if neutropenia was not reported (Baker 1974; Chlebowski 1989; Tomova 2010). In trials where the toxicity was described as 'severe' we assumed they assessed toxicity similarly to the trials with toxicity grade III or IV using the WHO criteria (Baker 1974; Chlebowski 1989).

There was no difference in the risk of neutropenia or leukopenia (RR 0.94; 95% CI 0.87 to 1.02; P = 0.12) (Analysis 5.1) but there was significant heterogeneity between trials (I2 = 78%; P <0.00001). However, combination chemotherapy was associated with an increased likelihood of febrile neutropenia (RR 1.32; 95% CI 1.06 to 1.65; P = 0.01) (Analysis 6.1). There was no significant heterogeneity (I2 = 42%; P = 0.09). There was no difference in the risk of nausea and vomiting (RR 0.88; 95% CI 0.57 to 1.34; P = 0.55) (Analysis 7.1) and there was no significant heterogeneity (I2 = 0%; P = 0.58).

When the toxicity outcomes were analysed using conditional logistic regression the results were similar to those provided by the Mantel-Haenszel method for treatment-related deaths and the inverse variance method for the other toxicity measures (neutropenia, febrile neutropenia, and nausea and vomiting).

Quality of life - all trials

Quality of life data were reported in three trials. This represented 25% of all patients randomised (589/2317). Two of these trials used the European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ C-30) (Conte 2004; Park 2010) and one used FACT-B (Sledge 2003) to measure quality of life. In Conte 2004 emotional functioning was better in the combination arm (score differences 3.53; 95% CI -6.99 to -0.06). There was no significant difference for the seven symptom scales or for financial impact. In Park 2010 there was no difference in baseline subscale scores between the two arms, and no significant difference in symptom scale scores during treatment. However, the combination arm had statistically better scores in role functioning, social functioning, and global health compared to the sequential arm (P < 0.05). In Sledge 2003 there was no statistically significant difference between treatment arms on any quality of life subscales.

Subgroup analyses

Low versus high or unclear risk of bias

We conducted a pre-specified subgroup analysis to investigate the treatment effect in trials with a low risk of bias compared to the trials considered at high or unclear risk of bias. In total, there were two trials considered low risk of bias and 10 trials with a high or unclear risk of bias.

Overall survival

Data for overall survival were available for both trials with a low risk of bias. In the trials with a low risk of bias, there was no difference in overall survival between the combination and sequential arms, with a HR of 1.21 (95% CI 0.93 to 1.58; P = 0.16) (Analysis 8.1), and no significant heterogeneity (I2 = 0%; P = 0.93).

Data for overall survival was available for seven out of the 10 trials with a high or unclear risk of bias. In these trials there was no difference in overall survival, with a HR of 1.01 (95% CI 0.90 to 1.14; P = 0.85) (Analysis 8.1), and no significant heterogeneity (I2 = 34%; P = 0.17).

A test for interaction between the two groups for overall survival revealed no significant interaction (P = 0.23).

Progression-free survival

In the amended analysis, data for progression-free survival were available for both trials with a low risk of bias. In the trials with a low risk of bias there was no difference in progression-free survival, with a HR of 1.15 (95% CI 0.94 to 1.41; P = 0.17) (Analysis 9.1), and no significant heterogeneity (I2 = 0%; P = 0.54). Data for progression-free survival was available for five of the 10 trials with a high or unclear risk of bias. In these trials there was no difference in progression-free survival with a HR of 1.09 (95% CI 0.94 to 1.26, P = 0.26) (Analysis 9.1) and no significant heterogeneity (I2 = 43%; P = 0.13).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.66).

In the original analysis, the data for progression-free survival in trials with a low risk of bias were the same as the amended analysis (refer to Analysis 9.2). Data for progression-free survival was available for six of the 10 trials with a high or unclear risk of bias. In these trials there was a higher risk of progression in the combination arm compared to the sequential arm with a HR of 1.17 (95% CI 1.01 to 1.35, P = 0.03) (Analysis 9.2) and no significant heterogeneity (I2 = 45%; P = 0.11).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.93).

Overall response

Data for overall response rate were available for both trials with a low risk of bias. There was no difference in overall tumour response, with a HR of 0.91 (95% CI 0.76 to 1.10; P = 0.35) (Analysis 10.1), and no significant heterogeneity (I2 = 49%; P = 0.16).

Tumour response data were available for all 10 trials with a high or unclear risk of bias. There was a higher overall tumour response rate in these trials for the combination arm compared to the sequential arm with a HR of 1.26 (95% CI 1.13 to 1.39; P < 0.0001) (Analysis 10.1) but there was significant heterogeneity (I2 = 70%; P = 0.0005).

A test for interaction between the two group for overall tumour response revealed a significant interaction (P = 0.004).

Treatment-related deaths

Data were available for one out of the two trials with a low risk of bias. In this trial with a low risk of bias (Conte 2004) there were three deaths in the combination arm compared to one death in the sequential arm with no difference in treatment-related deaths (HR 2.60; 95% CI 0.28 to 24.60; P = 0.40) (Analysis 11.1).

Data were available for six out of the 10 trials with a high or unclear risk of bias. In these trials there were 13 deaths in the combination arm and seven deaths in the sequential arm. There was no statistically significant difference, with a HR of 1.41 (95% CI 0.62 to 3.20; P = 0.42) (Analysis 11.1), and no significant heterogeneity (I2 = 42%; P = 0.14).

A test for interaction between the two groups for treatment-related deaths revealed no significant interaction (P = 0.61).

Neutropenia

Data were available for both trials with a low risk of bias. In these two trials with a low risk of bias there was a decreased risk of neutropenia in the sequential arm compared to the combination arm (RR 0.62; 95% CI 0.50 to 0.76; P < 0.00001) (Analysis 12.1) however there was significant heterogeneity (I2 = 91%; P = 0.0006).

Data were available for all 10 trials with a high or unclear risk of bias. In these trials there was no difference in the rates of neutropenia (RR 1.00; 95% CI 0.92 to 1.09; P = 0.98) (Analysis 12.1) however there was significant heterogeneity (I2 = 59%; P = 0.009).

A test for interaction between the two groups for neutropenia revealed a significant interaction (P < 0.0001).

Febrile neutropenia

Data were available for both of the trials with a low risk of bias. In these two trials with a low risk of bias there was no significant difference in the rates of febrile neutropenia (RR 1.60; 95% CI 0.72 to 3.57; P = 0.25) (Analysis 13.1) and there was no significant heterogeneity (I2 = 0%; P = 0.55).

Data were available for seven out of the 10 trials with a high or unclear risk of bias. In these trials there was an increased risk of febrile neutropenia in the combination arm compared to the sequential arm (RR 1.30; 95% CI 1.03 to 1.64; P = 0.03) (Analysis 13.1) but with significant heterogeneity (I2 = 54%; P = 0.04).

A test for interaction between the two groups for febrile neutropenia revealed no significant interaction (P = 0.63).

Nausea and vomiting

Data were available for one out of the two trials with a low risk of bias. This trial showed no difference in the rates of nausea and vomiting (RR 1.03; 95% CI 0.21 to 4.99; P = 0.97) (Analysis 14.1).

Data were available for seven out of the 10 trials with a high or unclear risk of bias. In these trials there was no difference in the rates of nausea and vomiting (RR 0.87; 95% CI 0.56 to 1.35; P = 0.52) (Analysis 14.1) and no significant heterogeneity (I2 = 0%; P = 0.46).

A test for interaction between the two groups for nausea and vomiting revealed no significant interaction (P = 0.83).

First-line versus second-line or third-line chemotherapy

We conducted a pre-specified subgroup analysis to investigate the treatment effect in trials of first-line chemotherapy compared to trials of second-line or third-line chemotherapy. In total, there were nine trials with first-line chemotherapy and three trials for second-line or third-line chemotherapy.

Overall survival

Data were available for seven out of nine trials with first-line chemotherapy. In the trials of first-line chemotherapy there was no difference in overall survival between the combination and sequential arms, with a HR of 1.04 (95% CI 0.93 to 1.18; P = 0.47) (Analysis 15.1), and no significant heterogeneity (I2 = 34%; P = 0.17).

Data were available for two out of the three trials of second- or third-line chemotherapy. In these trials there was no difference in overall survival, with a HR of 1.03 (95% CI 0.76 to 1.40; P = 0.84) (Analysis 15.1), and no significant heterogeneity (I2 = 28%; P = 0.24).

A test for interaction between the two groups for overall survival revealed no significant interaction (P = 0.94).

Progression-free survival

In the amended analysis, data were available for six out of nine trials of first-line chemotherapy. In the trials of first-line chemotherapy there was weak evidence of a higher risk of progression in the combination arm compared to the sequential arm with a HR of 1.11 (95% CI 0.99 to 1.26; P = 0.08) (Analysis 16.1) and no significant heterogeneity (I2 = 33%; P = 0.19). Data were available for one out of the three trials of second- or third-line chemotherapy. In this trial, there was no difference in progression-free survival, with a HR of 0.85 (95% CI 0.24 to 2.99; P = 0.80) (Analysis 16.1).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.67).

In the original analysis, data were available for six out of nine trials of first-line chemotherapy. In the trials of first-line chemotherapy there was a higher risk of progression in the combination arm compared to the sequential arm with a HR of 1.16 (95% CI 1.02 to 1.31; P = 0.02) (Analysis 16.2) and no significant heterogeneity (I2 = 44%; P = 0.11). Data were available for two out of the three trials of second- or third-line chemotherapy. In these trials there was no difference in progression-free survival, with a HR of 1.25 (95% CI 0.80 to 1.95; P = 0.33) (Analysis 16.2), and no significant heterogeneity (I2 = 0%; P = 0.52).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.74).

Overall response

Data were available for all nine trials investigating first-line chemotherapy. In these trials of first-line chemotherapy there was evidence of a difference in overall tumour response rate in the combination arm compared to the sequential arm, with a HR 1.10 (95% CI 1.00 to 1.22; P = 0.05) (Analysis 17.1), but there was significant heterogeneity (I2 = 74%; P = 0.0001).

Data were available for all three trials of second- or third-line chemotherapy. In these trials there was a higher overall tumour response rate in the combination compared to the sequential arm with a HR of 1.54 (95% CI 1.22 to 1.93; P = 0.0002) (Analysis 17.1) and there was no significant heterogeneity (I2 = 0%; P = 0.40).

A test for interaction between the two groups for overall tumour response revealed a significant interaction (P = 0.009).

Treatment-related deaths

Data were available for five of the nine trials of first-line chemotherapy. In the trials of first-line chemotherapy there were 14 deaths in the combination arm and six deaths in the sequential arm but this was not statistically different (HR 1.72; 95% CI 0.70 to 4.18; P = 0.23) (Analysis 18.1) and there was no significant heterogeneity (I2 = 33%; P = 0.21).

Data were available for two out of the three trials of second- or third-line chemotherapy. In these trials there were two deaths in the combination arm and two deaths in the sequential arm and this was not statistically different (HR 1.06; 95% CI 0.23 to 4.94; P = 0.94) (Analysis 18.1) and there was no significant heterogeneity (I2 = 59%; P = 0.12).

A test for interaction between the two groups for treatment-related deaths revealed no significant interaction (P = 0.59).

Neutropenia

Data were available for all nine trials of first-line chemotherapy. In the trials of first-line chemotherapy there was no difference in rates of neutropenia (RR 0.96; 95% CI 0.87 to 1.05; P = 0.39) (Analysis 19.1) but there was significant heterogeneity between the trials (I2 = 81%; P < 0.00001).

Data were available for all three trials of second- or third-line chemotherapy. In these trials there were higher rates of neutropenia in the sequential arm compared to the combination arm (RR 0.64; 95% CI 0.48 to 0.85; P = 0.002) (Analysis 19.1) and there was no significant heterogeneity (I2 = 0%; P = 0.62).

A test for interaction between the two groups for neutropenia revealed a significant interaction (P = 0.007).

Febrile neutropenia

Data were available for seven out of nine trials of first-line chemotherapy. In these trials there were higher rates of febrile neutropenia in the combination arm compared to the sequential arm (RR 1.34; 95% CI 1.07 to 1.68; P = 0.01) (Analysis 20.1) and there was significant heterogeneity (I2 = 48%; P = 0.07).

Data were available for two out of the three trials of second- or third-line chemotherapy. In these trials there was no difference in the rates of febrile neutropenia (RR 0.86; 95% CI 0.24 to 3.15; P = 0.82) (Analysis 20.1) and no significant heterogeneity (I2 = 43%; P = 0.18).

A test for interaction between the two groups for febrile neutropenia revealed no significant interaction (P = 0.51).

Nausea and vomiting

Data were available for five out of nine trials of first-line chemotherapy. There was no difference in the rate of nausea and vomiting (RR 0.85; 95% CI 0.55 to 1.33; P = 0.48) (Analysis 21.1) and no significant heterogeneity (I2 = 0%; P = 0.53).

Data were available for all three trials of second- or third-line chemotherapy. There was no difference in the rate of nausea and vomiting (RR 1.25; 95% CI 0.27 to 5.74; P = 0.78) (Analysis 21.1) and no significant heterogeneity (I2 = 25%; P = 0.25).

A test for interaction between the two groups for nausea and vomiting revealed no significant interaction (P = 0.64).

Chemotherapy given on disease progression (schema 1) versus chemotherapy given for a set number of cycles (schema 2)

We conducted a pre-specified subgroup analysis to investigate the treatment effect in trials where the combination or sequential arms were given until disease progression (schema 1) compared to trials where the combination or sequential arms were given for a set number of cycles (schema 2). In total, there were six trials that used schema 1 and six trials that used schema 2.

Overall survival

Data were available for five out of the six trials that used schema 1. In trials where chemotherapy was given on disease progression there was no difference in overall survival, with a HR of 0.98 (95% CI 0.86 to 1.12; P = 0.77) (Analysis 22.1), and there was no significant heterogeneity (I2 = 47%; P = 0.11).

Data were available for four out of the six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no difference in overall survival, with a HR of 1.22 (95% CI 0.99 to 1.49; P = 0.06) (Analysis 22.1), and there was no significant heterogeneity (I2 = 0%; P = 1.00).

A test for interaction between the two groups for overall survival revealed no significant interaction (P = 0.08).

Progression-free survival

In the amended analysis, data were available for three out of the six trials that used schema 1. In trials where chemotherapy was given on disease progression there was no difference in progression-free survival in the combination arm compared to the sequential arm with a HR of 1.11 (95% CI 0.94 to 1.31; P = 0.23) (Analysis 23.1) but there was significant heterogeneity (I2 = 70%; P = 0.03). Data were available for four out of the six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles, there was no difference in progression-free survival, with a HR of 1.12 (95% CI 0.94 to 1.33; P = 0.22) (Analysis 23.1), and there was no significant heterogeneity (I2 = 0%; P = 0.84).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.95).

In the original analysis, data were available for four out of the six trials that used schema 1. In trials where chemotherapy was given on disease progression there was a higher risk of progression in the combination arm compared to the sequential arm with a HR of 1.20 (95% CI 1.03 to 1.40; P = 0.02) (Analysis 23.2) but there was significant heterogeneity (I2 = 63%; P = 0.04). Data were available for four out of the six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no difference in progression-free survival, with a HR of 1.12 (95% CI 0.94 to 1.33; P = 0.22) (Analysis 23.2), and there was no significant heterogeneity (I2 = 0%; P = 0.84).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.54).

Overall response

Data were available for all six trials that used schema 1. In trials where chemotherapy was given on disease progression there was a higher overall tumour response in the combination arm compared to the sequential arm with a HR of 1.46 (95% CI 1.28 to 1.65; P < 0.00001) (Figure 6) but there was significant heterogeneity (I2 = 58%; P = 0.04).

Figure 6.

Forest plot of comparison: 24 Overall response - schema 1 versus schema 2, outcome: 24.1 Overall response - schema 1 versus schema 2.

Data were available for all six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no significant difference in overall tumour response in the combination compared to the sequential arm, with a HR of 0.92 (95% CI 0.80 to 1.04; P = 0.19) (Figure 6), and there was no significant heterogeneity (I2 = 0%; P = 0.65).

A test for interaction between the two groups for overall tumour response revealed a significant interaction (P < 0.00001).

Treatment-related deaths

Data were available for three out of the six trials that used schema 1. In trials where chemotherapy was given on disease progression there were nine deaths in the combination arm and seven deaths in the sequential arm, with no significant difference in treatment-related deaths (HR of 0.94; 95% CI 0.36 to 2.44; P = 0.90) (Analysis 25.1) and no significant heterogeneity (I2 = 58%; P = 0.09).

Data were available for four out of the six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there were seven deaths in the combination arm and one death in the sequential arm, with no significant difference (HR 4.00; 95% CI 0.87 to 18.35; P = 0.07) (Analysis 25.1) and no significant heterogeneity (I2 = 0%; P = 0.89).

A test for interaction between the two groups for treatment-related deaths revealed no significant interaction (P = 0.11).

Neutropenia

Data were available for all six trials that used schema 1. In trials where chemotherapy was given on disease progression there was no difference in the rates of neutropenia (RR 0.99; 95% CI 0.87 to 1.14; P = 0.92) (Analysis 26.1) and there was significant heterogeneity (I2 = 47%; P = 0.09).

Data were available for all six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no difference in the rates of neutropenia (RR 0.92; 95% CI 0.84 to 1.01; P = 0.07) (Analysis 26.1) but there was significant heterogeneity (I2 = 88%; P < 0.00001).

A test for interaction between the two groups for neutropenia revealed no significant interaction (P = 0.35).

Febrile neutropenia

Data were available for three out of six trials that used schema 1. In trials where chemotherapy was given on disease progression there was an increased risk of febrile neutropenia in the combination arm compared to the sequential arm (RR 1.76; 95% CI 1.14 to 2.73; P = 0.01) (Analysis 27.1) and no significant heterogeneity (I2 = 19%; P = 0.29).

Data were available for all six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no difference in the risk of febrile neutropenia (RR 1.19; 95% CI 0.92 to 1.55; P =0.18) (Analysis 27.1) and no significant heterogeneity (I2 = 44%; P = 0.11).

A test for interaction between the two groups for febrile neutropenia revealed no significant interaction (P = 0.13).

Nausea and vomiting

Data were available for three out of six trials that used schema 1. In trials where chemotherapy was given on disease progression there was no difference in the risk of nausea and vomiting (RR 0.75; 95% CI 0.39 to 1.46; P = 0.40) (Analysis 28.1) and no significant heterogeneity (I2 = 0%; P = 0.97).

Data were available for five out of six trials that used schema 2. In trials where chemotherapy was given for a set number of cycles there was no difference in the risk of nausea and vomiting (RR 0.98; 95% CI 0.56 to 1.71; P = 0.94) (Analysis 28.1) and no significant heterogeneity (I2 = 8%; P = 0.36).

A test for interaction between the two groups for nausea and vomiting revealed no significant interaction (P = 0.56).

Relative dose intensity

We conducted a pre-specified subgroup analysis to investigate the treatment effect in trials that contained arms with a similar relative dose intensity (RDI) compared to trials where there was a different RDI. In total, there were three trials with a similar RDI and three trials with a different RDI.

Overall survival

Data were available for two out of the three trials with a similar RDI between the combination and sequential arms. In the trials with a similar RDI between arms there was no difference in overall survival, with a HR of 1.23 (95% CI 0.93 to 1.62; P = 0.15) (Analysis 29.1), and no significant heterogeneity (I2 = 0%; P = 0.96).

Data were available for two out of the three trials with a different RDI. There was no difference in overall survival, with a HR of 1.03 (95% CI 0.76 to 1.40; P = 0.84) (Analysis 29.1), and no significant heterogeneity (I2 = 28%; P = 0.24).

A test for interaction between the two groups for overall survival revealed no significant interaction (P = 0.42).

Progression-free survival

In the amended analysis, data were available for two out of the three trials with a similar RDI. There was no difference in progression, with a HR of 1.15 (95% CI 0.91 to 1.45; P = 0.26) (Analysis 30.1), and no significant heterogeneity (I2 = 0%; P = 0.44). Data were available for one of the three trials with a different RDI. There was no difference in progression, with a HR of 0.85 (95% CI 0.24 to 2.99; P = 0.80) (Analysis 30.1).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.65).

In the original analysis, data were available for two out of the three trials with a similar RDI and the results were the same as the amended analysis (refer to Analysis 30.2). Data were available for two out of the three trials with a different RDI. There was no difference in progression, with a HR of 1.25 (95% CI 0.80 to 1.95; P = 0.33) (Analysis 30.2), and no significant heterogeneity (I2 = 0%; P = 0.52).

A test for interaction between the two groups for progression-free survival revealed no significant interaction (P = 0.74).

Overall response

Data were available for all three trials with a similar RDI between the combination and sequential arms. There was no difference in overall tumour response, with a HR of 0.88 (95% CI 0.72 to 1.07; P = 0.20) (Analysis 31.1), and no significant heterogeneity (I2 = 0%; P = 0.43).

Data were available for all three trials with a different RDI between the combination and sequential arms. There was no difference in overall tumour response, with a HR of 0.93 (95% CI 0.72 to 1.20; P = 0.55) Analysis 31.1), and no significant heterogeneity (I2 = 46%; P = 0.16).

A test for interaction between the two groups for overall tumour response revealed no significant interaction (P = 0.76).

Treatment-related deaths

Data were available for one out of three trials with a similar RDI. In this trial with a similar RDI in the combination and sequential arms there were no treatment-related deaths in either arm.

Data were available for all three trials with a different RDI between the combination and sequential arms. There were four deaths in the combination arm and two deaths in the sequential arm but this was not statistically significant, with a HR of 1.66 (95% CI 0.45 to 6.10; P = 0.44) (Analysis 32.1), and was without significant heterogeneity (I2 = 37%; P = 0.20).

A test for interaction was not applicable.

Neutropenia

Data were available for all three trials with a similar RDI. There was no difference in the rates of neutropenia (RR 1.08; 95% CI 0.97 to 1.20; P = 0.15) (Analysis 33.1) and no significant heterogeneity (I2 = 60%; P = 0.08).

Data were available for all three trials with a different RDI. In trials with differing RDIs there were higher rates of neutropenia in the sequential arm compared to the combination arm (RR 0.67; 95% CI 0.51 to 0.87; P = 0.003) (Analysis 33.1) and no significant heterogeneity (I2 = 12%; P = 0.32).

A test for interaction revealed a significant difference for neutropenia between the two groups (P = 0.001).

Febrile neutropenia

Data were available for all three trials with a similar RDI. There was a higher rate of febrile neutropenia in the combination arm compared to the sequential arm (RR 2.36; 95% CI 1.21 to 4.62; P = 0.01) (Analysis 34.1) and there was no significant heterogeneity (I2 = 36%; P = 0.21).

Data were available for all three trials with a different RDI. In trials with differing RDIs there was no difference in the rates of febrile neutropenia (RR 1.07; 95% CI 0.80 to 1.42; P = 0.66) (Analysis 34.1) and no significant heterogeneity (I2 = 0%; P = 0.39).

A test for interaction revealed a significant difference for febrile neutropenia between the two groups (P = 0.03).

Nausea and vomiting

Data were available for all three trials with a similar RDI. There was no difference in the rates of nausea and vomiting (RR 1.14; 95% CI 0.50 to 2.60; P = 0.75) (Analysis 35.1) and no significant heterogeneity (I2 = 19%; P = 0.29).

Data were available for all three trials with a different RDI. There was no difference in the rates of nausea and vomiting (RR 0.86; 95% CI 0.40 to 1.83; P = 0.69) (Analysis 35.1) and no significant heterogeneity (I2 = 39%; P = 0.20).

A test for interaction between the two groups for nausea and vomiting revealed no significant interaction (P = 0.62).

Discussion

Summary of main results

There was no statistically significant benefit in overall survival for combination compared to sequential single agent chemotherapy in the first-, second- or third-line treatment of metastatic breast cancer. This result was consistent in the four subgroups analysed (that is trials judged as high versus low risk of bias, first-line versus second- or third-line, schema 1 versus schema 2, and similar versus different dose intensity). In particular, there was no difference in outcome according to the type of sequential schema of chemotherapy.

In this amended version of the review, where progression-free survival in the combination arm was compared to time to first progression in the sequential arm, there was only weak evidence that combination therapy had a higher risk of progression than sequential therapy (HR 1.11; 95% CI 0.99 to 1.25; P = 0.08), with no significant heterogeneity. This result was consistent in all subgroups. In the original progression-free survival analysis, time to progression in the combination arm was compared to time to second progression in the sequential arm and showed that combination chemotherapy had a statistically significant higher risk of progression than sequential therapy (HR1.16; 95% CI 1.03 to 1.31; P = 0.01). However it is likely that this analysis is biased in favour of the sequential arm, particularly given that the Sledge 2003 is the largest trial, and is given the greatest weight.

Overall tumour response rates were higher for combination therapy compared to sequential therapy (RR 1.16; 95% CI 1.06 to 1.28; P = 0.001) but there was significant heterogeneity for this outcome across the trials. In trials with a high or unclear risk of bias there was a higher rate of tumour response (RR 1.26; 95% CI 1.13 to 1.39; P < 0.0001) than in the two trials which had a low risk of bias (RR 0.91; 95% CI 0.76 to 1.10; P = 0.35), however there was significant heterogeneity in the trials of high or unclear risk of bias. A test for interaction revealed a difference between these two groups (P = 0.004). There was evidence of an effect of combination chemotherapy on tumour response rate for trials of first-line chemotherapy (RR 1.10; 95% 1.00 to 1.22; P = 0.05) but there was significant heterogeneity. The effect was greater in trials of second- or third-line chemotherapy (RR 1.54; 95% CI 1.22 to 1.93; P = 0.0002) and the test for interaction revealed a difference in these subgroups (P = 0.009). The effect on tumour response was greater in trials where chemotherapy was changed on disease progression (schema 1, RR 1.46; 95% CI 1.28 to 1.65; P < 0.00001) but with significant heterogeneity compared to trials where chemotherapy was given for a set number of cycles (schema 2, RR 0.92; 95% CI 0.80 to 1.04; P = 0.19). The test for interaction confirmed a difference between schema 1 and schema 2 (P < 0.00001).

As aforementioned, there was no difference in overall survival or progression-free survival according to the schema of chemotherapy drugs given. Given this result, it seems preferable to use combination chemotherapy, and change treatment on disease progression (schema 1), in the setting of rapidly progressive disease when a greater response rate is desired.

In the seven trials that reported treatment-related deaths, there were 16 deaths for combination therapy compared to eight deaths for sequential therapy but there was no statistically significant difference (RR 1.53; 95% CI 0.71 to 3.29; P = 0.28). The confidence interval was wide due to the small number of events.

The risk of febrile neutropenia was higher for combination therapy (RR 1.32; 95% CI 1.06 to 1.65; P = 0.01). This outcome was consistent amongst all subgroups, although there was a higher risk of febrile neutropenia in trials with a similar RDI (RR 2.36; 95% CI 1.21 to 4.62; P = 0.01) compared to trials where there was a different dose intensity (RR 1.07; 95% CI 0.80 to 1.42; P = 0.66), and the test of interaction between these groups was significant (P = 0.03).

Overall, there was no significant difference in the risk of neutropenia (RR 0.94; 95% CI 0.87 to 1.02; P = 0.12) but there was significant heterogeneity between these trials (P < 0.00001). The risk of neutropenia was lower for combination therapy than for sequential therapy in trials which had a low risk of bias (RR 0.62; 95% CI 0.50 to 0.76; P < 0.00001) compared to those with a high risk of bias (RR 1.00; 95% CI 0.92 to 1.09; P = 0.98) and the test for interaction between these groups was significant (P < 0.0001). There was significant heterogeneity between the trials in both of these subgroups. Neutropenia rates were also lower for the combination therapy compared to the sequential therapy in trials of second- or third-line chemotherapy (RR 0.64; 95% CI 0.48 to 0.85; P = 0.002) compared to trials of first-line chemotherapy (RR 0.96; 95% CI 0.87 to 1.05; P = 0.39). The test of interaction between these groups was significant (P = 0.007). Neutropenia rates were lower in the combination therapy compared to the sequential therapy in trials using a different RDI (RR 0.67; 95% CI 0.51 to 0.87; P = 0.003) compared to trials that used a similar RDI (RR 1.08; 95% CI 0.97 to 1.20; P = 0.15). The test of interaction between these groups was significant (P = 0.001).

There was no significant difference in the risk of nausea and vomiting between the combination and sequential single agent therapies (RR 0.88; 95% CI 0.57 to 1.37; P = 0.55) and this was consistent across all subgroups analysed.

Quality of life was only measured in three out of the 12 trials in this review. In Sledge 2003 there was no difference in quality of life between the two arms, however in Conte 2004 and Park 2010 there was a suggestion of a better quality of life in some domains in the combination arm compared to the sequential arm. In Conte 2004 the trend towards inferior quality of life in the sequential arm was thought to be due to the higher doses of the single agents, which caused more haematological and neurological toxicity. In Park 2010 increased functional scores (role function, social function) and global health in the combination arm were postulated to be due to the better response rate in the combination arm. This is consistent with studies of quality of life in the palliative setting where quality of life was related to how responsive the tumour was to chemotherapy (Modi 2002; Shin 2008).

Overall completeness and applicability of evidence

While data for overall and progression-free survival could be extracted for the majority of trials (nine of 12 trials for overall survival, and seven of 12 trials for progression-free survival in the amended analysis), the evidence would be more complete if these data were available for all trials. In the trials where the data were missing, two were published in abstract form (Koroleva 2001; Soto 2006). In Cresta 2004, HRs were not published for overall survival or progression-free survival, and survival curves were not included in the paper. Chlebowski 1989 did not publish a HR for progression-free survival, or include a survival curve. Despite some missing data, the evidence is still applicable to clinical practice because survival data were available for the majority of trials and the results in the subgroup analyses were consistent with the overall HRs. All trials provided data on tumour response and neutropenia. There was a paucity of quality of life data, which means it is not possible to generalise about the effect of combination versus sequential single agent chemotherapy on quality of life.

Quality of the evidence

The majority of trials (10 out of 12) were classified as having an unclear or high risk of bias. For most of these trials this classification was made because it was not clearly explained in the trial methods if factors such as randomisation and blinding were done properly. Randomisation and blinding make up four out of a total of seven factors on which the risk of bias is assessed (random sequence generation, allocation concealment, blinding of participants and personnel, and blinding of outcome assessment). As a result, there was a tendency for most trials to score poorly for risk of bias. While unclear risk is categorised with a high risk of bias it may be that this is an unfair assumption and randomisation may in fact have been done properly. Also, blinding is rarely feasible in chemotherapy trials. The risk of bias in the randomised controlled trials included in this review may therefore be overestimated and the quality of the trials may be better than this scoring system implies.

Potential biases in the review process

There were no potential biases in the review process.

Agreements and disagreements with other studies or reviews

The higher overall tumour response rate for combination therapy is consistent with other reviews of combination chemotherapy in metastatic breast cancer (Butters 2011; Carrick 2009). Butters 2011 aimed to assess the effect of adding a chemotherapy drug, or drugs, to an established chemotherapy regimen in metastatic breast cancer and demonstrated an advantage for tumour response for adding one or more drugs, but there was no difference in overall survival or time to progression. There was more toxicity with the addition of one or more drugs to a regimen. Carrick 2009 aimed to assess the effect of combination chemotherapy compared to single agent chemotherapy in metastatic breast cancer and found that combination chemotherapy had superior overall survival, time to progression and response rates, but more toxicity. Both of these Cochrane reviews found a high rate of toxicity for neutropenia, alopecia and nausea and vomiting. Febrile neutropenia was not analysed separately. Our review found higher rates of febrile neutropenia for combination therapy but no overall difference in the rates of neutropenia or nausea and vomiting. We did not analyse rates of alopecia. The higher risk of detection of neutropenia for sequential therapy compared to combination therapy in some subgroups is likely due to the higher doses of the single agents used.

Authors' conclusions

Implications for practice

Combination chemotherapy does not provide a survival benefit compared to sequential single agent chemotherapy. It is notable that there was no difference in outcome according to the treatment strategy of sequential single agents; the single agents can either be changed on disease progression, or planned sequences of the single agents can be administered. There was weak evidence that sequential single agent chemotherapy has a positive effect on progression-free survival while combination chemotherapy produces a higher tumour response rate. Combination chemotherapy also has a higher risk of febrile neutropenia, with no overall difference in rates of neutropenia or nausea and vomiting. This review supports the recommendations by international guidelines to use sequential monotherapy in most clinical scenarios unless there is rapid clinical progression, life-threatening visceral metastases, or need for rapid symptom or disease control, where combination chemotherapy is preferred (Cardoso 2009; Cardoso 2012).

Implications for research

Future trials need to provide good quality evidence so that the best treatment decisions can be made for women with metastatic breast cancer. More research is needed to determine the impact of combination therapy versus sequential single agent chemotherapy on quality of life, and also potentially include the outcome Q-TWIST (quality-adjusted time without symptoms of disease and toxicity), which is clearly an important goal in the treatment of metastatic breast cancer.

Acknowledgements

We would like to thank Fergus Tai for his work identifying studies through the Cochrane Breast Cancer Group Specialised Register. We would also like to thank Melina Willson for her advice during the review process.

Data and analyses

Download statistical data

Comparison 1. Overall survival
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival (all trials)91786Hazard Ratio (Fixed, 95% CI)1.04 [0.93, 1.16]
Analysis 1.1.

Comparison 1 Overall survival, Outcome 1 Overall survival (all trials).

Comparison 2. Progression-free survival
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Progression-free survival (amended analysis)71483Hazard Ratio (Fixed, 95% CI)1.11 [0.99, 1.25]
2 Progression-free survival (all trials)8 Hazard Ratio (Fixed, 95% CI)1.16 [1.03, 1.31]
Analysis 2.1.

Comparison 2 Progression-free survival, Outcome 1 Progression-free survival (amended analysis).

Analysis 2.2.

Comparison 2 Progression-free survival, Outcome 2 Progression-free survival (all trials).

Comparison 3. Overall response
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall response (all trials)122140Risk Ratio (IV, Fixed, 95% CI)1.16 [1.06, 1.28]
Analysis 3.1.

Comparison 3 Overall response, Outcome 1 Overall response (all trials).

Comparison 4. Treatment-related deaths
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment-related deaths (all trials)7902Risk Ratio (M-H, Fixed, 95% CI)1.53 [0.71, 3.29]
Analysis 4.1.

Comparison 4 Treatment-related deaths, Outcome 1 Treatment-related deaths (all trials).

Comparison 5. Neutropenia
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Neutropaenia12 Risk Ratio (Fixed, 95% CI)0.94 [0.87, 1.02]
Analysis 5.1.

Comparison 5 Neutropenia, Outcome 1 Neutropaenia.

Comparison 6. Febrile neutropenia
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Febrile neutropenia9 Risk Ratio (Fixed, 95% CI)1.32 [1.06, 1.65]
Analysis 6.1.

Comparison 6 Febrile neutropenia, Outcome 1 Febrile neutropenia.

Comparison 7. Nausea and vomiting
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Nausea and vomiting8 Risk Ratio (Fixed, 95% CI)0.88 [0.57, 1.34]
Analysis 7.1.

Comparison 7 Nausea and vomiting, Outcome 1 Nausea and vomiting.

Comparison 8. Overall survival - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival - risk of bias9 Hazard Ratio (Fixed, 95% CI)1.04 [0.93, 1.16]
1.1 Low risk of bias2 Hazard Ratio (Fixed, 95% CI)1.21 [0.93, 1.58]
1.2 High/unclear risk of bias7 Hazard Ratio (Fixed, 95% CI)1.01 [0.90, 1.14]
Analysis 8.1.

Comparison 8 Overall survival - risk of bias, Outcome 1 Overall survival - risk of bias.

Comparison 9. Progression-free survival - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Progression-free survival - risk of bias (amended analysis)7 Hazard Ratio (Fixed, 95% CI)1.11 [0.99, 1.25]
1.1 Low risk of bias2 Hazard Ratio (Fixed, 95% CI)1.15 [0.94, 1.41]
1.2 High/unclear risk of bias5 Hazard Ratio (Fixed, 95% CI)1.09 [0.94, 1.26]
2 Progression-free survival - risk of bias8 Hazard Ratio (Fixed, 95% CI)1.16 [1.03, 1.31]
2.1 Low risk of bias2 Hazard Ratio (Fixed, 95% CI)1.15 [0.94, 1.41]
2.2 High/unclear risk of bias6 Hazard Ratio (Fixed, 95% CI)1.17 [1.01, 1.35]
Analysis 9.1.

Comparison 9 Progression-free survival - risk of bias, Outcome 1 Progression-free survival - risk of bias (amended analysis).

Analysis 9.2.

Comparison 9 Progression-free survival - risk of bias, Outcome 2 Progression-free survival - risk of bias.

Comparison 10. Overall response - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall response - risk of bias122140Risk Ratio (IV, Fixed, 95% CI)1.16 [1.06, 1.28]
1.1 Low risk of bias2381Risk Ratio (IV, Fixed, 95% CI)0.91 [0.76, 1.10]
1.2 High/unclear risk of bias101759Risk Ratio (IV, Fixed, 95% CI)1.26 [1.13, 1.39]
Analysis 10.1.

Comparison 10 Overall response - risk of bias, Outcome 1 Overall response - risk of bias.

Comparison 11. Treatment-related deaths - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment-related deaths - risk of bias7902Risk Ratio (M-H, Fixed, 95% CI)1.53 [0.71, 3.29]
1.1 Low risk of bias1198Risk Ratio (M-H, Fixed, 95% CI)2.60 [0.28, 24.60]
1.2 High/unclear risk of bias6704Risk Ratio (M-H, Fixed, 95% CI)1.41 [0.62, 3.20]
Analysis 11.1.

Comparison 11 Treatment-related deaths - risk of bias, Outcome 1 Treatment-related deaths - risk of bias.

Comparison 12. Neutropenia - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Neutropaenia - risk of bias12 Risk Ratio (Fixed, 95% CI)0.94 [0.87, 1.02]
1.1 Low risk of bias2 Risk Ratio (Fixed, 95% CI)0.62 [0.50, 0.76]
1.2 High/unclear risk of bias10 Risk Ratio (Fixed, 95% CI)1.00 [0.92, 1.09]
Analysis 12.1.

Comparison 12 Neutropenia - risk of bias, Outcome 1 Neutropaenia - risk of bias.

Comparison 13. Febrile neutropenia - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Febrile neutropenia - risk of bias9 Risk Ratio (Fixed, 95% CI)1.32 [1.06, 1.65]
1.1 Low risk of bias2 Risk Ratio (Fixed, 95% CI)1.60 [0.72, 3.57]
1.2 High/unclear risk of bias7 Risk Ratio (Fixed, 95% CI)1.30 [1.03, 1.64]
Analysis 13.1.

Comparison 13 Febrile neutropenia - risk of bias, Outcome 1 Febrile neutropenia - risk of bias.

Comparison 14. Nausea and vomiting - risk of bias
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Nausea and vomiting - risk of bias8 Risk Ratio (Fixed, 95% CI)0.88 [0.57, 1.34]
1.1 Low risk of bias1 Risk Ratio (Fixed, 95% CI)1.03 [0.21, 4.99]
1.2 High/unclear risk of bias7 Risk Ratio (Fixed, 95% CI)0.87 [0.56, 1.35]
Analysis 14.1.

Comparison 14 Nausea and vomiting - risk of bias, Outcome 1 Nausea and vomiting - risk of bias.

Comparison 15. Overall survival - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival - line of chemotherapy9 Hazard Ratio (Fixed, 95% CI)1.04 [0.93, 1.16]
1.1 First line chemotherapy7 Hazard Ratio (Fixed, 95% CI)1.04 [0.93, 1.18]
1.2 Second/third line chemotherapy2 Hazard Ratio (Fixed, 95% CI)1.03 [0.76, 1.40]
Analysis 15.1.

Comparison 15 Overall survival - line of chemotherapy, Outcome 1 Overall survival - line of chemotherapy.

Comparison 16. Progression-free survival - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Progression free survival - line of chemotherapy (amended analysis)7 Hazard Ratio (Fixed, 95% CI)1.11 [0.99, 1.25]
1.1 First line chemotherapy6 Hazard Ratio (Fixed, 95% CI)1.11 [0.99, 1.26]
1.2 Second/third line chemotherapy1 Hazard Ratio (Fixed, 95% CI)0.85 [0.24, 2.99]
2 Progression free survival - line of chemotherapy8 Hazard Ratio (Fixed, 95% CI)1.16 [1.03, 1.31]
2.1 First line chemotherapy6 Hazard Ratio (Fixed, 95% CI)1.16 [1.02, 1.31]
2.2 Second/third line chemotherapy2 Hazard Ratio (Fixed, 95% CI)1.25 [0.80, 1.95]
Analysis 16.1.

Comparison 16 Progression-free survival - line of chemotherapy, Outcome 1 Progression free survival - line of chemotherapy (amended analysis).

Analysis 16.2.

Comparison 16 Progression-free survival - line of chemotherapy, Outcome 2 Progression free survival - line of chemotherapy.

Comparison 17. Overall response - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall response - subgroup analysis, line of chemotherapy122140Risk Ratio (IV, Fixed, 95% CI)1.16 [1.06, 1.28]
1.1 First line chemotherapy91778Risk Ratio (IV, Fixed, 95% CI)1.10 [1.00, 1.22]
1.2 Second/third line chemotherapy3362Risk Ratio (IV, Fixed, 95% CI)1.54 [1.22, 1.93]
Analysis 17.1.

Comparison 17 Overall response - line of chemotherapy, Outcome 1 Overall response - subgroup analysis, line of chemotherapy.

Comparison 18. Treatment-related deaths - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment-related deaths - line of chemotherapy7902Risk Ratio (M-H, Fixed, 95% CI)1.53 [0.71, 3.29]
1.1 First line chemotherapy5722Risk Ratio (M-H, Fixed, 95% CI)1.72 [0.70, 4.18]
1.2 Second/third line chemotherapy2180Risk Ratio (M-H, Fixed, 95% CI)1.06 [0.23, 4.94]
Analysis 18.1.

Comparison 18 Treatment-related deaths - line of chemotherapy, Outcome 1 Treatment-related deaths - line of chemotherapy.

Comparison 19. Neutropenia - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Neutropaenia - line of chemotherapy12 Risk Ratio (Fixed, 95% CI)0.92 [0.84, 1.01]
1.1 First-line chemotherapy9 Risk Ratio (Fixed, 95% CI)0.96 [0.87, 1.05]
1.2 Second/third-line chemotherapy3 Risk Ratio (Fixed, 95% CI)0.64 [0.48, 0.85]
Analysis 19.1.

Comparison 19 Neutropenia - line of chemotherapy, Outcome 1 Neutropaenia - line of chemotherapy.

Comparison 20. Febrile neutropenia - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Febrile neutropenia - line of chemotherapy9 Risk Ratio (Fixed, 95% CI)1.32 [1.06, 1.65]
1.1 First-line chemotherapy7 Risk Ratio (Fixed, 95% CI)1.34 [1.07, 1.68]
1.2 Second/third-line chemotherapy2 Risk Ratio (Fixed, 95% CI)0.86 [0.24, 3.15]
Analysis 20.1.

Comparison 20 Febrile neutropenia - line of chemotherapy, Outcome 1 Febrile neutropenia - line of chemotherapy.

Comparison 21. Nausea and vomiting - line of chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Nausea and vomiting - line of chemotherapy8 Risk Ratio (Fixed, 95% CI)0.88 [0.57, 1.34]
1.1 First-line chemotherapy5 Risk Ratio (Fixed, 95% CI)0.85 [0.55, 1.33]
1.2 Second/third-line chemotherapy3 Risk Ratio (Fixed, 95% CI)1.25 [0.27, 5.74]
Analysis 21.1.

Comparison 21 Nausea and vomiting - line of chemotherapy, Outcome 1 Nausea and vomiting - line of chemotherapy.

Comparison 22. Overall survival - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival - Schema 1 versus Schema 29 Hazard Ratio (Fixed, 95% CI)1.04 [0.93, 1.16]
1.1 Schema 15 Hazard Ratio (Fixed, 95% CI)0.98 [0.86, 1.12]
1.2 Schema 24 Hazard Ratio (Fixed, 95% CI)1.22 [0.99, 1.49]
Analysis 22.1.

Comparison 22 Overall survival - schema 1 versus schema 2, Outcome 1 Overall survival - Schema 1 versus Schema 2.

Comparison 23. Progression-free survival - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Progression-free survival - Schema 1 versus Schema 2 (amended analysis)7 Hazard Ratio (Fixed, 95% CI)1.11 [0.99, 1.25]
1.1 Schema 13 Hazard Ratio (Fixed, 95% CI)1.11 [0.94, 1.31]
1.2 Schema 24 Hazard Ratio (Fixed, 95% CI)1.12 [0.94, 1.33]
2 Progression-free survival - Schema 1 versus Schema 28 Hazard Ratio (Fixed, 95% CI)1.16 [1.03, 1.31]
2.1 Schema 14 Hazard Ratio (Fixed, 95% CI)1.20 [1.03, 1.40]
2.2 Schema 24 Hazard Ratio (Fixed, 95% CI)1.12 [0.94, 1.33]
Analysis 23.1.

Comparison 23 Progression-free survival - schema 1 versus schema 2, Outcome 1 Progression-free survival - Schema 1 versus Schema 2 (amended analysis).

Analysis 23.2.

Comparison 23 Progression-free survival - schema 1 versus schema 2, Outcome 2 Progression-free survival - Schema 1 versus Schema 2.

Comparison 24. Overall response - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall response - Schema 1 versus Schema 2122140Risk Ratio (IV, Fixed, 95% CI)1.16 [1.06, 1.28]
1.1 Schema 161344Risk Ratio (IV, Fixed, 95% CI)1.46 [1.28, 1.65]
1.2 Schema 26796Risk Ratio (IV, Fixed, 95% CI)0.92 [0.80, 1.04]
Analysis 24.1.

Comparison 24 Overall response - schema 1 versus schema 2, Outcome 1 Overall response - Schema 1 versus Schema 2.

Comparison 25. Treatment-related deaths - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment-related deaths - Schema 1 versus Schema 27902Risk Ratio (M-H, Fixed, 95% CI)1.53 [0.71, 3.29]
1.1 Schema 13379Risk Ratio (M-H, Fixed, 95% CI)0.94 [0.36, 2.44]
1.2 Schema 24523Risk Ratio (M-H, Fixed, 95% CI)4.00 [0.87, 18.35]
Analysis 25.1.

Comparison 25 Treatment-related deaths - schema 1 versus schema 2, Outcome 1 Treatment-related deaths - Schema 1 versus Schema 2.

Comparison 26. Neutropenia - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Neutropaenia - subgroup analysis, Schema 1 versus Schema 212 Risk Ratio (Fixed, 95% CI)0.94 [0.87, 1.02]
1.1 Schema 16 Risk Ratio (Fixed, 95% CI)0.99 [0.87, 1.14]
1.2 Schema 26 Risk Ratio (Fixed, 95% CI)0.92 [0.84, 1.01]
Analysis 26.1.

Comparison 26 Neutropenia - schema 1 versus schema 2, Outcome 1 Neutropaenia - subgroup analysis, Schema 1 versus Schema 2.

Comparison 27. Febrile neutropenia - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Febrile neutropenia - Schema 1 versus Schema 29 Risk Ratio (Fixed, 95% CI)1.32 [1.06, 1.65]
1.1 Schema 13 Risk Ratio (Fixed, 95% CI)1.76 [1.14, 2.73]
1.2 Schema 26 Risk Ratio (Fixed, 95% CI)1.19 [0.92, 1.55]
Analysis 27.1.

Comparison 27 Febrile neutropenia - schema 1 versus schema 2, Outcome 1 Febrile neutropenia - Schema 1 versus Schema 2.

Comparison 28. Nausea and vomiting - schema 1 versus schema 2
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Nausea and vomiting - Schema 1 versus Schema 28 Risk Ratio (Fixed, 95% CI)0.88 [0.57, 1.34]
1.1 Schema 13 Risk Ratio (Fixed, 95% CI)0.75 [0.39, 1.46]
1.2 Schema 25 Risk Ratio (Fixed, 95% CI)0.98 [0.56, 1.71]
Analysis 28.1.

Comparison 28 Nausea and vomiting - schema 1 versus schema 2, Outcome 1 Nausea and vomiting - Schema 1 versus Schema 2.

Comparison 29. Overall survival - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival - relative dose intensity4 Hazard Ratio (Fixed, 95% CI)1.14 [0.92, 1.39]
1.1 Similar dose intensity2 Hazard Ratio (Fixed, 95% CI)1.23 [0.93, 1.62]
1.2 Different dose intensity2 Hazard Ratio (Fixed, 95% CI)1.03 [0.76, 1.40]
Analysis 29.1.

Comparison 29 Overall survival - relative dose intensity, Outcome 1 Overall survival - relative dose intensity.

Comparison 30. Progression-free survival - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Progression-free survival - relative dose intensity (amended analysis)3 Hazard Ratio (Fixed, 95% CI)1.13 [0.90, 1.43]
1.1 Similar dose intensity2 Hazard Ratio (Fixed, 95% CI)1.15 [0.91, 1.45]
1.2 Different dose intensity1 Hazard Ratio (Fixed, 95% CI)0.85 [0.24, 2.99]
2 Progression-free survival - relative dose intensity4 Hazard Ratio (Fixed, 95% CI)1.17 [0.95, 1.44]
2.1 Similar dose intensity2 Hazard Ratio (Fixed, 95% CI)1.15 [0.91, 1.45]
2.2 Different dose intensity2 Hazard Ratio (Fixed, 95% CI)1.25 [0.80, 1.95]
Analysis 30.1.

Comparison 30 Progression-free survival - relative dose intensity, Outcome 1 Progression-free survival - relative dose intensity (amended analysis).

Analysis 30.2.

Comparison 30 Progression-free survival - relative dose intensity, Outcome 2 Progression-free survival - relative dose intensity.

Comparison 31. Overall response - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall response - relative dose intensity6679Risk Ratio (IV, Fixed, 95% CI)0.90 [0.77, 1.05]
1.1 Similar dose intensity3355Risk Ratio (IV, Fixed, 95% CI)0.88 [0.72, 1.07]
1.2 Different dose intensity3324Risk Ratio (IV, Fixed, 95% CI)0.93 [0.72, 1.20]
Analysis 31.1.

Comparison 31 Overall response - relative dose intensity, Outcome 1 Overall response - relative dose intensity.

Comparison 32. Treatment-related deaths - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Treatment-related deaths - relative dose intensity4406Risk Ratio (M-H, Fixed, 95% CI)1.66 [0.45, 6.10]
1.1 Similar dose intensity182Risk Ratio (M-H, Fixed, 95% CI)0.0 [0.0, 0.0]
1.2 Different dose intensity3324Risk Ratio (M-H, Fixed, 95% CI)1.66 [0.45, 6.10]
Analysis 32.1.

Comparison 32 Treatment-related deaths - relative dose intensity, Outcome 1 Treatment-related deaths - relative dose intensity.

Comparison 33. Neutropenia - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Neutropaenia - relative dose intensity6 Risk Ratio (Fixed, 95% CI)1.01 [0.92, 1.12]
1.1 Similar dose intensity3 Risk Ratio (Fixed, 95% CI)1.08 [0.97, 1.20]
1.2 Different dose intensity3 Risk Ratio (Fixed, 95% CI)0.67 [0.51, 0.87]
Analysis 33.1.

Comparison 33 Neutropenia - relative dose intensity, Outcome 1 Neutropaenia - relative dose intensity.

Comparison 34. Febrile neutropenia - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Febrile neutropenia - relative dose intensity6 Risk Ratio (Fixed, 95% CI)1.21 [0.93, 1.57]
1.1 Similar dose intensity3 Risk Ratio (Fixed, 95% CI)2.36 [1.21, 4.62]
1.2 Different dose intensity3 Risk Ratio (Fixed, 95% CI)1.07 [0.80, 1.42]
Analysis 34.1.

Comparison 34 Febrile neutropenia - relative dose intensity, Outcome 1 Febrile neutropenia - relative dose intensity.

Comparison 35. Nausea and vomiting - relative dose intensity
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Nausea and vomiting - relative dose intensity6 Risk Ratio (Fixed, 95% CI)0.98 [0.56, 1.71]
1.1 Similar dose intensity3 Risk Ratio (Fixed, 95% CI)1.14 [0.50, 2.60]
1.2 Different dose intensity3 Risk Ratio (Fixed, 95% CI)0.86 [0.40, 1.83]
Analysis 35.1.

Comparison 35 Nausea and vomiting - relative dose intensity, Outcome 1 Nausea and vomiting - relative dose intensity.

Appendices

Appendix 1. MEDLINE via OvidSP

1randomised controlled trial.pt.
2randomized controlled trial.pt.
3controlled clinical trial.pt.
4randomized.ab.
5randomised.ab.
6placebo.ab.
7randomly.ab.
8trial.ab.
9groups.ab.
101 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9
11exp Breast Neoplasms/
12metastatic breast cancer$.ti,ab,tw.
13metastatic breast neoplas$.ti,ab,tw.
14metastatic breast tumour$.ti,ab,tw.
15metastatic breast tumor$.ti,ab,tw.
16metastatic breast carcinoma$.ti,ab,tw.
17advance$ breast cancer$.ti,ab,tw.
18advance$ breast neoplas$.ti,ab,tw.
19advance$ breast tumour$.ti,ab,tw.
20advance$ breast tumor$.ti,ab,tw.
21advance$ breast carcinoma$.ti,ab,tw.
2212 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21
2311 and 22
24exp Drug Therapy, Combination/
25combination chemotherapy.mp.
26sequential single agent chemotherapy.mp.
27(sequential adj6 single adj6 agent adj6 chemotherap$).mp.
2824 or 25 or 26 or 27
2910 and 23 and 28
30Animals/
31Humans/
3230 not 31
3329 not 32

Appendix 2. EMBASE via Embase.com

  1. random* OR factorial* OR crossover* OR cross NEXT/1 over* OR placebo* OR (doubl* AND blind*) OR (singl* AND blind*) OR assign* OR allocat* OR volunteer*OR 'crossover procedure'/exp OR 'double blind procedure'/exp OR 'randomized controlled trial'/exp OR 'single blind procedure'/exp

  2. 'breast cancer'/exp

  3. advance* NEAR/6 'breast neoplasm'

  4. advance* NEAR/6 'breast carcinoma'

  5. advance* NEAR/6 'breast cancer'

  6. advance* NEAR/6 'breast tumour'

  7. advance* NEAR/6 'breast tumor'

  8. metastatic NEAR/6 'breast cancer'

  9. metastatic NEAR/6 'breast neoplasm'

  10. metastatic NEAR/6 'breast carcinoma'

  11. metastatic NEAR/6 'breast tumour'

  12. metastatic NEAR/6 'breast tumor'

  13. #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12

  14. #2 AND #13

  15. 'chemotherapy'/exp OR chemotherapy

  16. chemotherap*

  17. combin* NEAR/6 chemotherap*

  18. sequen* NEAR/6 'single agent'

  19. #16 AND #18

  20. #15 OR #16 OR #17 OR #19

  21. #1 AND #14 AND #20

  22. #21 AND [humans]/lim AND [embase]/lim

Appendix 3. WHO ICTRP

Basic search:

1.     advanced  breast cancer AND combination chemotherap*

2.     advanced breast cancer AND sequential single agent chemotherapy

3.     advanced breast cancer AND sequential single chemotherapy

4.     advanced breast cancer AND sequential chemotherapy

5.     advanced breast cancer AND single agent chemotherapy

6.     metastatic  breast cancer AND combination chemotherap*

7.     metastatic breast cancer AND sequential single agent chemotherapy

8.     metastatic breast cancer AND sequential single chemotherapy

9.     metastatic breast cancer AND sequential chemotherapy

10.  metastatic breast cancer AND single agent chemotherapy

Advanced search:

1.       Title: Combination versus sequential single agent chemotherapy for metastatic breast cancer

Recruitment Status: ALL

2.       Condition: advanced breast cancer OR metastatic breast cancer

Intervention: combination chemotherap*

Recruitment Status: ALL

3.       Condition: advanced breast cancer OR metastatic breast cancer

Intervention: sequential single agent chemotherapy

Recruitment Status: ALL

4.       Condition: advanced breast cancer OR metastatic breast cancer

Intervention: combination versus single sequential chemotherapy

Recruitment Status: ALL

Appendix 4. ClinicalTrials.gov

Advanced search:

1.       Conditions: metastatic breast cancer

Interventions: combination chemotherapy

2.       Conditions: metastatic breast cancer

Interventions: sequential single agent chemotherapy

3.       Conditions: metastatic breast cancer

Interventions: combination versus sequential single agent chemotherapy

4.       Conditions: advanced breast cancer

Interventions: combination chemotherapy

5.       Conditions: advanced breast cancer

Interventions: sequential single agent chemotherapy

6.       Conditions: advanced breast cancer

Interventions: combination versus sequential single agent chemotherapy

What's new

DateEventDescription
21 November 2014AmendedAdded new data for progression-free survival analysis for Schema 1 (where drugs changed on disease progression). The analysis now includes a comparison of time to first progression in the combination arm versus time to first progression (not after second progression) in the sequential arm (refer to analyses 2.1, 9.1, 16.1, 23.1 and 30.1). This review version also includes new data obtained for Sledge 2003.

History

Protocol first published: Issue 10, 2010
Review first published: Issue 12, 2013

DateEventDescription
21 January 2014AmendedCorrected labels for Analysis 10.1, 17.1, 24.1 (Figure 6) and 31.1, and also the number of participants in the Beslija trial for the outcome overall response

Contributions of authors

Rachel Dear (RD), Martin Tattersall (MT) and Nicholas Wilcken (NW) conceived the study objective. RD wrote the review, which was reviewed and revised by MT, Alexandra Barratt (AB), NW, Kevin McGeechan (KMcG), and Marisa Jenkins (MJ).

Declarations of interest

None

Sources of support

Internal sources

  • Nil, Other.

External sources

  • Nil, Other.

Differences between protocol and review

In the protocol we proposed only including first and second-line chemotherapy trials. Our review includes a trial of third-line chemotherapy (Park 2010, prior treatment with anthracyclines or taxanes) because the paper met all our other eligibility criteria.

A pre-specified secondary outcome in the protocol was stable disease. We have not reported this outcome because overall tumour response rate (which includes complete response and partial response) is a more commonly reported outcome that can then be compared to other similar reviews on this topic.

Another pre-specified outcome in the protocol was "QTWIST" (quality-adjusted time without symptoms of disease and toxicity). We have not reported this in our review because this outcome was not reported in any of the trials included in this review, and toxicity data were not consistently reported so that time with side effects from chemotherapy could not accurately be extracted.

In the review, the 'Human' limit in the MEDLINE search strategy and the 'Randomised controlled trial' limit in the EMBASE search strategy were revised.

Notes

Two studies (Campone 2013; Zhang 2013) were categorised under the section 'Characteristics of studies awaiting classification' because the trials were completed but with insufficient data for inclusion at present. The results from these two small trials will be included in an updated version of this review.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Alba 2004

Methods

Accrual dates: December 1999 to December 2001

Sample size: 144

Number of centres: multicentre - 23 sites in Spain

Randomisation method: randomisation done centrally but not reported how randomisation was achieved

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 31-78 years (median 58 in sequential arm, 61 in combination arm)

MBC

First-line chemotherapy

Interventions

AT versus A→T

Arm I: (AT) doxorubicin + docetaxel

Arm II: (A→T) doxorubicin followed by docetaxel

Outcomes

Toxicity (primary outcome)

Overall response rate (CR + PR)

Duration of response (for responding patients, calculated from the date of first objective response until the time of progression)

TTP

OS

Notes

Schema 2

Follow-up: minimum 0.36 months, maximum 35.59 months

Patients stratified prospectively based on previous administration of anthracyclines and study centre

Combination arm: 67% completed 6 cycles. Withdrawals mainly due to adverse events. Less than 95% relative dose intensity

Sequential arm: 81% completed 6 cycles. Greater than 95% relative dose intensity

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- unclear
Allocation concealment (selection bias)Unclear risk2- unclear
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- unclear
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- unclear
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes are reported
Other biasLow risk

Treatment groups were similar at baseline

The trial was not stopped early

Analysis was by intention to treat

Baker 1974

Methods

Accrual dates: October 1969 to March 1971

Sample size: 89

Number of centres: 1 site in the United States of America

Randomisation method: randomisation "based upon the last digit of their admission number to the service"

Baseline comparability: different sample size of 2 arms (more in combination than sequential arm) but no difference in Karnofsky Performance Scale (KPS), age, menopausal status or disease-free interval

Participants

Female

Age range: 26-77 years (53 in combination arm, median 54 in sequential arm)

MBC

First-line chemotherapy

Interventions

FCV versus F→C→ V

Arm I: (FCV) 5-FU + cyclophosphamide + vincristine

Arm II: (F→C→ V) 5-FU followed by cyclophosphamide followed by vincristine

Outcomes

Response

Duration of response

Overall survival

Notes

Schema 1

Follow-up: minimum 18 months, maximum 36 months

6/52 (12%) patients in combination arm, and 7/37 (19%) in sequential arm died prior to 30 days from the onset of the study

The median number of cycles received by the combination arm was not given

In sequential arm, 30/37 (81%) received second cycle of 5FU, 25/30 (83%) received cyclophosphamide, and 12/25 (40%) received vincristine. The median number of cycles of each drug was not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Women were randomly allocated to one of two branches of the study based upon the last digit of their admission number to the service"
Allocation concealment (selection bias)High risk3- No. "All patients treated with chemotherapy during this time period were entered onto the study, despite the fact that the physician might have known which branch the patient would be entered into prior to the initiation of therapy."
Blinding of participants and personnel (performance bias)
All outcomes
High risk3- No. All parties aware of treatment
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- unclear
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes are reported
Other biasHigh risk

Treatment groups had different numbers of patients, but were well balanced for baseline characteristics

The trial was not stopped early

Analysis was not by intention to treat

Beslija 2006

Methods

Accrual dates: not reported in abstract

Sample size: 100

Number of centres: not reported in abstract

Randomisation method: not reported in abstract

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 29-64 years (48 in combination arm, median 52 in sequential arm)

MBC

First-line chemotherapy

Interventions

XT versus X→T

Arm I: (XT) capecitabine + docetaxel

Arm II: (T→X) docetaxel followed by capecitabine

Outcomes

Toxicity

Overall response rate (CR + PR)

TTP

OS

Notes

Schema 1

Follow-up: not reported, not calculated because abstract only

100% had prior adjuvant anthracyclines

All patients had a good performance status - 100% had median KPS 100

X monotherapy data were not considered in the RR or TTP analyses but were included for OS and safety (number not reported)

In the T→X arm, 74% crossed over to X on progression

Median number of cycles given in the combination and sequential arms not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- unclear
Allocation concealment (selection bias)Unclear risk2- unclear
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- unclear
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- unclear
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- yes: all pre-specified outcomes are reported in a pre-specified way
Other biasLow risk

In the abstract report available, pre-specified outcomes were reported

The treatment groups were similar at baseline

It is not reported if the trial was stopped early

The analysis was by intention to treat

Chlebowski 1989

Methods

Accrual dates: 1971 and 1973

Sample size: 237 (126 from WCSG and 111 from SECSG)

Number of centres: unknown but multicentre (USA) - Western Cancer Study Group (WCSG) and Southeastern Cancer Study Group (SECSG)

Randomisation method: randomisation done centrally, but exact method not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: median age 55 years

MBC

First-line chemotherapy

Interventions

CMFTP or CMFVP versus F→C→TP→M or F→M→C→V→P

Arm I: (CMFTP or CMFVP) cyclophosphamide + 5-fluorouracil + methotrexate + triiodothyronine OR vincristine + prednisone

Arm II: (F→C→TP→M) 5-fluorouracil followed by cyclophosphamide followed by triiodothyronine plus prednisone followed by methotrexate OR (F→M→C→V→P) 5-fluorouracil followed by methotrexate followed by cyclophosphamide followed by vincristine followed by prednisone

Outcomes

Overall survival

Response rate

Duration of response

Toxicity

Notes

Schema 1

Follow-up: minimum 143 months, maximum 143 months

This report is a combined analysis from the WCSG trial (reported separately by Chlebowski in 1979) and the SECSG trial (reported by Smalley 1976). These trials were initiated at the same time and had similar patient eligibility, chemotherapy protocols (except WCSG used triiodothyronine and SECSG used vincristine) and response criteria

121/126 on WCSG were available for toxicity and response assessment

101/111 on SECSG were available for toxicity and response assessment

Median number of cycles for each arm not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- unclear
Allocation concealment (selection bias)Low risk1- adequate
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- unclear
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- unclear
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Unclear risk3- progression-free survival was not reported in a way that it could be used in the review
Other biasLow risk

Treatment groups were similar at baseline

The trial was not stopped early

The analysis was by intention to treat

Conte 2004

Methods

Accrual dates: December 1996 and June 2001

Sample size: 202

Number of centres: multicentre, 18 centres (15 Italian, 3 Spanish)

Randomisation method: central randomisation by phone call "a separate randomisation list for each centre was prepared with permuted blocks of varying size in random sequence"

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 30-73, median 58

MBC

First-line chemotherapy

Interventions

EP versus E→P

Arm I: (EP) epirubicin + paclitaxel

Arm II: (E→P) epirubicin followed by paclitaxel

Outcomes

Objective response rate (i.e. ORR = CR + PR). CR disappearance of all clinical evidence of tumour, determined by 2 observations not less than 4 weeks apart; PR- ≥50% decrease in the sum of the products of the largest dimension of the target lesions; SD- response less than PR or progression less than PD for ≥4 weeks; PD ≥25% increase in the product of measured lesions and/or new lesions

Overall survival (date of randomisation to date of death or last follow-up)

Progression-free survival (date of randomisation to date of death or PD)

Toxicity

Quality of life (QLQ-C30)

Pharmacoeconomic profile

Notes

Schema 2

Randomised phase 3 trial

Follow-up: minimum 25 months, maximum 79 months

Noninferiority design where the combination (Arm I) was considered the standard (control) arm

2 patients in combination arm and 2 patients in sequential arm withdrew and were lost to follow up immediately after randomisation

Patients who had disease progression during combination chemotherapy or when receiving paclitaxel in the sequential arm were released from the study. If disease progression occurred while on epirubicin, patients were given paclitaxel 250 mg/m2

Based on cost of chemo drugs- sequential arm 37% less expensive than combination arm- full analysis to be reported separately

60/106 (57%) of combination arm and 60/92 (65%) completed planned treatment; 15/106 (14%) in combination arm versus 31/92 (34%) in sequential arm completed all cycles without protocol modifications (P = 0.001)

Median of 8 courses (range 0 to 8) was given in both treatment arms

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk1- adequate
Allocation concealment (selection bias)Low risk1- adequate
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes reported
Other biasHigh risk

Treatment groups similar at baseline for important prognostic factors

Trial was stopped early before the planned sample size of 266 was reached because the sequential arm was associated with not a decreased but possibly an increased toxicity (neutropenia and neuropathy)

The analysis was not by intention to treat

Cresta 2004

Methods

Accrual dates: November 1996 to January 2000

Sample size: 123

Number of centres: multicentre, 9 centres

Randomisation method: not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 24-69. Median age 53 years

MBC

First-line chemotherapy

Interventions

AT versus alternating T then A versus sequential A→T

Arm I: (AT) doxorubicin + docetaxel

Arm II: (sequential A→T) doxorubicin followed by docetaxel

Arm III: (alternating T then A) docetaxel followed by doxorubicin

Outcomes

Rate of complete response

Overall response

Time to progression

Survival

Safety

Notes

Schema 2

Randomised phase 2 study

Follow-up: not reported

123 enrolled, 121 received study treatment. Of 123 enrolled patients, 13 didn't meet eligibility criteria (5 from Arm I, 2 from Arm II, and 6 from Arm III)

Median relative dose intensity for each arm ≥ 0.9 for each drug

Median total dose of doxorubicin: Arm I 460 mg/m2, Arm II 297 mg/m2, Arm III 294 mg/m2

Discontinuation due to adverse events: Arm I 7 patients, Arm II 5 patients, Arm III 2 patients

Median number of cycles received by each arm not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)High risk3- outcomes were not reported in a way that they could be used in the review (hazard ratio not reported, nor was a Kaplan-Meier curve reported)
Other biasLow risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was by intention to treat

Fountzilas 2001

Methods

Accrual dates: October 1997 to May 1999

Sample size: 183

Number of centres: unknown (Greece)

Randomisation method:done at central office and based on a random number list, stratification

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 26-77 years (median 57.5 years)

MBC

First-line chemotherapy

Interventions

Arm I: EP versus E→P

Arm I (EP) epirubicin + paclitaxel

Arm II (E→P) epirubicin followed by paclitaxel (dose dense)

Outcomes

Overall response rate

Complete response rate (complete disappearance of all clinical symptoms and signs of disease for at least 4 weeks)

Time to progression

Overall survival (from start of chemotherapy to date of death or last follow-up)

Toxicity

Notes

Schema 2

Randomised phase 2

Follow-up: minimum 6 months, maximum 84.5 months

Median relative dose intensity in both groups was similar (96-97)

80% completed planned combination therapy, 85% completed planned sequential therapy

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk1- adequate
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Low risk1- all imaging reviewed by 3 independent radiologists at the end of the study
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes are reported
Other biasLow risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was by intention to treat

Koroleva 2001

Methods

Accrual dates: August 1997 to August 2000

Sample size: 193

Number of centres: unknown (Europe)

Randomisation method: not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 21-70 (median 51 years)

MBC

First-line chemotherapy

Interventions

AT versus AT (different drug doses) versus T→A

Arm I: (AT) doxorubicin + docetaxel

Arm II: docetaxel then doxorubicin

Arm III: (AT, different drug doses) doxorubicin + docetaxel

Outcomes

Response rate

Toxicity

Time to progression

Overall survival

Notes

Schema 2

Randomised phase 2 study

Follow-up: not reported, not able to calculate because abstract only

Abstract only

Median relative dose intensity in both groups similar (97.9)

Median number of cycles received by each arm not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk2- attrition data reported (193 patients randomised, but data available for only 177 patients, and only 131 patients evaluated for response) but reasons not included
Selective reporting (reporting bias)High risk3- outcomes were not reported in a way that they could be used in the review (hazard ratio not reported, nor was a Kaplan-Meier curve reported)
Other biasHigh risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was not by intention to treat

Park 2010

Methods

Accrual dates: April 2004 to December 2006

Sample size: 82

Number of centres: not reported (Korea)

Randomisation method: not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: 27-77 (median 53 years)

MBC

Third-line chemotherapy and beyond (prior treatment with anthracyclines and taxanes)

Interventions

GV versus G→V

Arm I: (GV) gemcitabine + vinorelbine

Arm II: (G→V) gemcitabine → vinorelbine

Outcomes

Response rate (proportion with a complete or partial tumour response)

Time to treatment failure (from date of entry to date of progression, death, withdrawal etc)

Overall survival (survival from date of randomisation to date of death or last follow-up)

Quality of life (EORTC QLQ C-30, validated Korean version)

Notes

Schema 1

Randomised phase 2 study

Follow-up: minimum 1 month, maximum 55.4 months

81 patients evaluable for treatment efficacy and safety

The mean delivered doses higher in sequential arm, but higher relative dose intensity in combination arm

In the sequential arm, 33/40 (83%) crossed over to vinorelbine

In the combination arm the median number of cycles of GV was 4. In the sequential arm, the median number of cycles of gemcitabine was 2.5, and the median number of cycles of vinorelbine was 2

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Low risk1- all imaging was reviewed by an independent radiologist
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes reported
Other biasHigh risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was not by intention to treat

Sledge 2003

Methods

Accrual dates: February 1993 and September 1995

Sample size: 739

Number of centres: not reported (USA)

Randomisation method:

Baseline comparability:no significant imbalance reported

Participants

Female

Age range: 25-79 years (median 56 years combination, 58 in doxorubicin arm, 56 in paclitaxel arm)

MBC

First-line chemotherapy

Interventions

AP versus A versus P

Arm I: (AP) doxorubicin + paclitaxel

Arm II: (A) doxorubicin (→ paclitaxel)

Arm III: (P) paclitaxel (→ doxorubicin)

Outcomes

Response rate (complete response and partial response)

Time to treatment failure (date of study entry to date of disease progression pr date)

Overall survival (date of study entry to death or date of last follow-up)

Quality of life (FACT-B, at baseline and week 16)

Notes

Schema 1

Randomised phase 3 study

Follow-up: minimum 0.13 months, maximum 75.17 months

8 patients not eligible for inclusion (duplicate registration n = 3, became ineligible n = 3, patient needed radiotherapy n = 1, doctor preference n = 1), therefore 731 randomised, then 33 excluded

In the sequential arm, the overall rate of crossover was 57% (58% from doxorubicin to paclitaxel, and 56% from paclitaxel to doxorubicin)

The median number of cycles given in each arm was not reported

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes reported
Other biasHigh risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was not by intention to treat

Soto 2006

Methods

Accrual dates:

Sample size: 368 enrolled, 277 evaluable

Number of centres: unknown (Mexico)

Randomisation method: not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Age range: not reported in abstract

MBC

Second-line chemotherapy (prior anthracyclines)

Interventions

XT versus X→T

Arm I: (XT) capecitabine + docetaxel

Arm II: (X→T) capecitabine → docetaxel or paclitaxel

Arm III: (X→P) capecitabine + paclitaxel

Outcomes

Response rate

Progression-free survival

Overall survival

Toxicity

Median doses

Notes

Schema 1

Randomised phase 3 study

Median follow-up 15.5 months (minimum and maximum follow-up not reported)

Abstract only

In sequential arm 58/91 (64%) received the sequential taxane (the remainder did not either because still on X, or had a CR, or had a rapid PD)

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Unclear risk2- exclusions reported ("91 patients still on therapy or not evaluable") but reason incomplete (e.g. why the patients were not evaluable)
Selective reporting (reporting bias)High risk3- outcomes were not reported in a way that they could be used in the review (hazard ratio not reported, nor was a Kaplan-Meier curve reported)
Other biasHigh risk

1- treatment groups similar at baseline regarding important prognostic factors

The trial was not stopped early

The analysis was not by intention to treat

Tomova 2010

  1. a

    CR: complete response
    KPS: Karnofsky Performance Scale
    MBC: metastatic breast cancer
    OS: overall survival
    PD: progressive disease
    PR: partial response
    RR: response rate
    TTP: time to progression

Methods

Accrual dates: September 2002 to March 2006

Sample size: 100

Number of centres: 14 centres in 7 countries

Randomisation method: not reported

Baseline comparability: no significant imbalance reported

Participants

Female

Median age 53 years in combination arm, 55 years in sequential arm

MBC

Second-line chemotherapy (prior anthracyclines)

Interventions

TG versus T→ G

Arm I: (TG) docetaxel + gemcitabine

Arm II: (T→G) docetaxel → gemcitabine

Outcomes

Response rate

Response duration

Time to progression

Overall survival

Toxicity

Notes

Schema 2

Randomised phase 2 study

Follow-up: minimum 0 months, maximum 40 months

Sample size calculated to be 430, but due to poor recruitment the trial was closed after 100 patients were randomised

26% completed the combination treatment TG and 28% completed the sequential treatment T→G

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear risk2- not reported
Allocation concealment (selection bias)Unclear risk2- not reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear risk2- not reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear risk2- not reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk1- no missing data
Selective reporting (reporting bias)Low risk1- all pre-specified outcomes reported
Other biasHigh risk

3- the trial was stopped early due to poor recruitment

The analysis was not by intention to treat

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Chlebowski 1979The results of this study were updated and included in Chlebowski 1989 (which also includes the updated results from Smalley 1976)
Smalley 1976The results of this study were updated and included in Chlebowski 1989 (which also includes the updated results from Chlebowski 1979)

Characteristics of studies awaiting assessment [ordered by study ID]

Campone 2013

MethodsAccrual dates: June 2005 to April 2008
Sample size: 139 patients
Randomised multicentre study
ParticipantsMBC
Previously treated with anthracyclines
Interventions

Arm I: vinorelbine + capecitabine
Arm II: vinorelbine ↔ capecitabine

Arm II: docetaxel + capecitabine

Outcomes

Disease control rate
Safety profile
Response rate
Duration of response
Duration of disease control
Duration of stable disease
Progression-free survival
Time to treatment failure

Overall survival

NotesStudy results published in May-June 2013. Awaiting additional trial data from the trialists

Zhang 2013

Methods

Accrual dates: August 2007 to December 2010

Sample size: 60

Randomised multicentre phase II study

Participants

MBC

First-line therapy

Interventions

Arm I: vinorelbine + capecitabine

Arm II: vinorelbine → capecitabine

OutcomesProgression-free survival
Time to treatment failure
Quality of life
Time to progression
Overall survival
Notes

Study results published 2 October 2013. Awaiting additional trial data from the trialists

See: http://clinicaltrials.gov/show/NCT00629148