3 The manufacturer's submission
3.1 The manufacturer's submission approached the decision problem by comparing gemcitabine plus paclitaxel with licensed taxane-based regimens: paclitaxel, docetaxel monotherapy, and docetaxel plus capecitabine. The population consisted of people who had relapsed and developed metastatic breast cancer following anthracycline-based adjuvant or neo-adjuvant chemotherapy, or non-anthracycline-based chemotherapy where anthracyclines were contraindicated. The manufacturer stated that gemcitabine plus paclitaxel would be considered for people who are younger and fitter than the general population of patients with metastatic breast cancer, and who are considered suitable for taxane-based therapy. The patients for whom gemcitabine plus paclitaxel would be considered also require 'a higher level of efficacy than would be achieved with a monotherapy regimen, without the toxicity usually associated with a combination regimen', for example because of visceral metastasis. The primary outcome measure considered was overall survival. Secondary outcome measures included time to documented progression of disease, progression-free survival, overall response rates, pain and analgesia, quality of life and incidence of adverse events.
3.2 The manufacturer's submission presented evidence on the clinical effectiveness of gemcitabine plus paclitaxel based on a single randomised controlled trial (RCT), the JHQG trial, which compared gemcitabine plus paclitaxel with paclitaxel monotherapy. Final analyses of the JHQG trial showed that, compared with the 263 people in the paclitaxel arm of the trial, the 266 people in the gemcitabine plus paclitaxel arm of the trial had greater median overall survival (18.6 months versus 15.8 months, p = 0.0489; hazard ratio: 0.82, 95% confidence interval 0.67 to 1.00, p = 0.0495) and time to documented progression of disease (5.4 months versus 3.5 months, p = 0.0013; hazard ratio: 0.73, 95% confidence interval 0.61 to 0.89, p = 0.0015). The final results of the JHQG trial have not yet been published in a peer-reviewed journal.
3.3 Evidence on cost effectiveness presented in the manufacturer's submission was based on a Markov state-transition model with a 3-year horizon, equivalent to the typical life expectancy of people diagnosed with metastatic breast cancer. A series of pairwise economic analyses comparing gemcitabine plus paclitaxel with docetaxel monotherapy, paclitaxel monotherapy and docetaxel plus capecitabine was presented by the manufacturer. All these analyses were based on an indirect comparison in which weighted absolute treatment outcomes (including survival data) were pooled from single arms of different trials in published literature. In order to compare gemcitabine plus paclitaxel with paclitaxel monotherapy, the median overall survival estimate for gemcitabine plus paclitaxel was taken from the RCT comparing gemcitabine plus paclitaxel with paclitaxel monotherapy. However, for paclitaxel monotherapy, the manufacturer did not use overall survival estimates from this comparative study, but instead used the average of the pooled, weighted absolute survival data from single arms of different studies.
3.4 The base-case analysis compared gemcitabine plus paclitaxel with docetaxel monotherapy and resulted in an incremental cost-effectiveness ratio (ICER) of £17,200 per quality-adjusted life year (QALY). A comparison of gemcitabine plus paclitaxel with paclitaxel monotherapy resulted in an ICER of £30,100 per QALY. A comparison of gemcitabine plus paclitaxel with docetaxel plus capecitabine resulted in an ICER of £23,200 per QALY. The manufacturer presented a scenario analysis for gemcitabine plus paclitaxel against docetaxel monotherapy where the price of non-proprietary paclitaxel is assumed to be 55% less than that of proprietary paclitaxel: the ICER in this case fell from £17,200 per QALY to £4700 per QALY.
3.5 The Evidence Review Group (ERG) reviewed the evidence submitted for clinical and cost effectiveness. The ERG judged that when only the results of the JHQG trial were considered, the manufacturer's submission contained a reasonable estimate of the clinical effectiveness of gemcitabine plus paclitaxel when compared with paclitaxel monotherapy. It was noted that the overall survival benefits of gemcitabine plus paclitaxel may have been diluted by a number of patients in the paclitaxel arm of the trial receiving second-line treatments that included gemcitabine, docetaxel, vinorelbine and capecitabine. The use of second-line treatments was similar in both arms of the trial except for a four‑fold greater use of gemcitabine in the paclitaxel arm.
3.6 The ERG reviewed the economic model and judged its structure to be reasonable and based on previous economic studies. The main drivers of cost effectiveness are the estimates of overall survival, the cost of paclitaxel, and the utilities assigned to the health states in the model. The ERG's main source of concern was the indirect comparison method used by the manufacturer to generate the survival estimates for the economic model, which involved pooling treatment outcome data from single arms of different trials. The ERG commented that the method used by the manufacturer ignored the fact that RCTs are designed to measure relative treatment effects. The indirect comparison method used does not preserve the benefits of randomisation and it is at best equivalent to observational studies.
3.7 The ERG raised concerns about the comparability of the trials from which the data were pooled. In particular, the ERG highlighted underlying differences in the patient characteristics in the trials, notably the lines of prior therapies received. Finally, the ERG noted that the manufacturer's indirect comparison estimated median overall survival with paclitaxel monotherapy to be longer than with docetaxel monotherapy. This contradicts the results of a head-to-head trial in which patients randomised to docetaxel monotherapy had greater median overall survival than those randomised to paclitaxel monotherapy.
3.8 By using the treatment efficacy data from both arms of the RCT comparing gemcitabine plus paclitaxel with paclitaxel monotherapy instead of the pooled estimates from the manufacturer's indirect comparisons, the ERG estimated the ICER for a comparison between gemcitabine plus paclitaxel and paclitaxel monotherapy to be £42,800 per QALY. In an illustrative analysis, the ERG found that using relative treatment effects to estimate overall survival for docetaxel monotherapy resulted in an ICER of £45,800 per QALY for a comparison of gemcitabine plus paclitaxel against docetaxel monotherapy.
3.9 Full details of the evidence is in the manufacturer's submission and the ERG report.