Search strategy

Twenty-three electronic databases and other databases of ongoing research and Internet resources were searched from inception to May 2002. The bibliographies of retrieved articles and submissions received from the drug company were also examined.

Inclusion/exclusion criteria

Two reviewers independently screened all titles and/or abstracts including economic evaluations. The full manuscript of any study judged to be relevant by either reviewer was obtained and assessed for inclusion or exclusion. Disagreements were resolved through discussion. For the assessment of capecitabine monotherapy, uncontrolled Phase II and observational studies were included which only recruited patients reported to have received previous treatment with an anthracycline and/or a taxane. For the assessment of capecitabine in combination with docetaxel, randomised controlled trials (RCTs) and uncontrolled studies were included which investigated only patients who had received previous treatment with an anthracycline. The outcomes were: survival, response, symptom relief, QoL, adverse events and costs.

Data extraction strategy

Data were extracted into an Access database by one reviewer and checked for accuracy by a second. Any disagreements were resolved by discussion.

Quality assessment strategy

The quality of each clinical study was assessed by one reviewer and checked by a second. Any disagreements were resolved by discussion. The same quality checklist was used regardless of study design to give a continuous measure of quality. The quality of the cost-effectiveness studies was assessed using a checklist updated from that developed by Drummond and colleagues in 1997. This checklist reflected the criteria for economic evaluation detailed in the methodological guidance developed by NICE. This information is presented in table form.

Methods of analysis

Owing to the small number of studies included in the review and the heterogeneity between the studies, statistical pooling was inappropriate, so statistical chi-squared tests of heterogeneity were not performed. Studies were grouped according to whether capecitabine was used alone or in combination with docetaxel. For the time to event data, where reported, hazard ratios (HR) with 95% confidence intervals (CI) were presented. For the remaining outcomes (tumour response, QoL and adverse events) relative risk (RR) estimates were calculated where appropriate, with 95% CI. RR data were also presented in the form of Forest plots without pooled estimates.

Details of each published economic evaluation, with an assessment of study quality, were presented in structured tables and as a narrative summary. Economic data were presented in the form of a summary and critique of the evidence. Additional analysis was undertaken to explore cost-effectiveness more fully. This included a careful assessment of assumptions underlying the submitted economic analyses using relevant experts, the estimation of differential mean survival duration, the use of Monte Carlo simulation to generate cost-effectiveness acceptability curves and the impact of differences in health-related QoL on cost-effectiveness.

Included studies

In total, 347 titles and abstracts were screened for inclusion in the review. Full paper manuscripts of 233 citations were assessed for inclusion.

Quality of the clinical effectiveness data

Quality of economic evaluations

Assessment of clinical effectiveness

Assessment of cost-effectiveness

Capecitabine monotherapy

The evidence base for the assessment of the effectiveness of capecitabine monotherapy was particularly poor. All of the studies identified for inclusion in the review lacked a control group, leaving them vulnerable to biases and confounding factors.

The evidence from these uncontrolled studies appears to indicate that capecitabine has antitumour activity when used as monotherapy in patients who have received previous treatment with anthracycline-containing regimens and taxanes. The toxicity profile appeared to indicate an increased risk of patients particularly experiencing hand-foot syndrome and diarrhoea. QoL was inadequately assessed; only one study included an assessment as part of the evaluation of capecitabine monotherapy.

In terms of cost-effectiveness, based on the available data, treatment with capecitabine, compared indirectly with treatment with vinorelbine, appears to be cost-effective. No comparative trials of these treatments were reported. Given the diverse patient population, in terms of disease and treatment history, it is likely that an RCT, comparing survival from point of randomisation for both treatments in a comparative trial, could provide different information on relative survival times.

In conclusion, good quality RCTs are urgently needed to compare the effectiveness of capecitabine monotherapy with the alternative third- and subsequent-line therapies currently available, as well as with best supportive care. These data should be collected in a form that facilitates cost-effectiveness analysis. The quality of the economic assessment reflects the poor level of clinical evidence. On the available evidence, capecitabine monotherapy is cost-effective, but there remain serious doubts about whether the quality of the clinical trials invalidates this conclusion. For a more complete picture, systematic reviews of vinorelbine, best supportive care and other relevant comparators in this setting need to be undertaken.

Capecitabine in combination with docetaxel

This review suggests that there is limited evidence in the form of RCTs on which to base an assessment of the effectiveness of capecitabine in combination with docetaxel in comparison to existing and new chemotherapy agents for the second-line treatment of advanced breast cancer. Only one RCT was identified for inclusion in the review comparing capecitabine in combination with docetaxel to treatment with single-agent docetaxel.

From the evidence available from the single trial, capecitabine in combination with docetaxel appears to be more effective than single-agent docetaxel in terms of overall survival, time to disease progression, time to treatment failure and overall tumour response (complete response plus partial response). There was no statistically significant difference between the two groups in any of the QoL domains. Statistically significant differences between combination and single-agent therapy were identified in terms of reported grade 3/4 treatment-related side-effects. Treatment with capecitabine-docetaxel was associated with higher incidences of hand-foot syndrome, nausea, diarrhoea and stomatitis.

In terms of costs, combination therapy seems to be cost-effective; however, the cost-effectiveness analysis did not directly consider the impact on QoL associated with the combination and monotherapy treatments themselves.

In conclusion, further RCTs investigating capecitabine in combination with docetaxel compared to alternative second-line therapies are required. From the limited evidence it would appear that capecitabine in combination with docetaxel is more effective in terms of median survival time, time to disease progression, time to treatment failure and overall response than single-agent docetaxel. The economic analysis indicates that combination therapy is very likely to be cost-effective. However, the method of calculation of QALYs ignores the potential for differences in adverse events between treatments to alter QoL estimates.