4 Evidence and interpretation
The Appraisal Committee considered evidence from a number of sources (see Appendix A).
4.1.1 Two phase III randomised controlled trials (RCTs), recruiting 602 and 605 individuals, and one pooled analysis of these study data were reviewed. Both RCTs compared capecitabine with a bolus (Mayo) 5-FU/FA regimen and were identical in design. Both studies included individuals with untreated, locally advanced or metastatic colorectal cancer, most of whom had undergone previous surgery. Although neither RCT was undertaken under blinded conditions because of the different routes of administration (oral or intravenous), both studies used an independent committee to review outcomes. The primary outcome measure in both trials was tumour response rate. Both studies were adequately powered to demonstrate equivalence in overall response rates.
4.1.2 Differences in median overall survival were not statistically different at the 5% significance level in either RCT, with values of 12.5 and 13.3 months for capecitabine and 5-FU/FA respectively in one study, and 13.2 and 12.1 months respectively in the other study. The pooled study also did not report any statistically significant difference in overall survival.
4.1.3 In both studies, the overall response rate was statistically significantly higher in the capecitabine groups than in the 5-FU/FA groups when outcomes were assessed by study investigators (p = 0.005 and p = 0.013). However, when the independent review committee assessed outcomes, these response rates were statistically significantly higher in only one of the studies (p = 0.0001). When the data from both studies were pooled, response rates statistically favoured capecitabine irrespective of who carried out the assessment (p < 0.0002 for the investigator assessment and p < 0.0001 for the independent review committee assessment).
4.1.4 Neither study reported a statistically significant difference in mean duration of response between the capecitabine and 5-FU/FA groups, nor was a difference reported for the pooled data. Neither study, nor the pooled analysis, reported any statistically significant differences in time to disease progression, death or treatment failure between the capecitabine and 5-FU/FA groups.
4.1.5 Neither of the studies reported any statistically significant difference in global quality of life as measured using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30).
4.1.6 With regard to treatment-related adverse events, the pooled analysis of the two trials indicated that individuals in the capecitabine groups reported less diarrhoea (48% vs 58%, p < 0.001), stomatitis (24% vs 62%, p < 0.001), nausea (38% vs 47%, p < 0.001) and alopecia (6% vs 21%, p < 0.001) of all grades than those in the 5-FU/FA groups. Patients in the capecitabine groups also had less grade III/IV neutropenia (2% vs 21%, no p-value available) and grade III stomatitis (2% vs 15%, p < 0.0001), and less frequent hospitalisation for adverse events (12% vs 18%, p = 0.002), but reported more hand–foot syndrome (54% vs 6.0%, no p-value available) and grade III hyperbilirubinaemia (18% vs 3%, p < 0.0001). In the pooled analysis, treatment mortality was 1% for each group.
4.1.7 Two large, open, phase III RCTs (Studies 011 and 012) were reviewed. These trials recruited 816 and 380 individuals respectively. No independent review committee was used to compensate for the fact that the assessors were aware of treatment allocation. Study 011 compared UFT/FA with 5-FU/FA administered using the Mayo regimen, whereas Study 012 compared UFT/FA with 5-FU/FA administered using a modification of the Mayo regimen, where treatment was repeated every 35 days instead of the standard 28 days. This non-standard variation of the Mayo regimen is a less dose-intensive regimen and has not been tested for efficacy. In Study 011, individuals recruited in the USA received UFT plus FA 75 mg/day, while those in non-USA centres received UFT plus FA 90 mg/day. Study 011 used overall survival as the primary endpoint and was powered to demonstrate equivalence of the two treatments as non-inferiority of survival. Study 012 used time to disease progression as the primary endpoint, and was powered to detect a hazard ratio (HR) of 1.4 between the groups. A third study of crossover design was also identified, which assessed patient preference for UFT/FA compared with intravenous 5-FU/FA.
4.1.8 In Study 011, median survival time was 12.4 months in the UFT/FA group and 13.4 months in the 5-FU/FA group. The HR for 5-FU/FA over UFT/FA was 0.96 (95% confidence interval [CI]: 0.83 to 1.13). In Study 012, median survival time was 12.2 months in the UFT/FA group and 10.3 months in the 5-FU/FA group. The HR for 5-FU/FA over UFT/FA in this study was 1.14 (95% CI: 0.92 to 1.42). A secondary analysis showed that individuals from the USA sites in Study 011, who received lower-dose FA, had worse overall survival than the total study population.
4.1.9 In Study 011 the median time to disease progression was statistically significantly greater in the 5-FU/FA group than in the UFT/FA group (3.8 months vs 3.5 months, p = 0.01), although the actual difference was 10 days. No statistically significant difference in time to disease progression was reported in Study 012.
4.1.10 In both studies there were no statistically significant differences between treatment arms with regard to overall tumour response rates. The rates in the UFT/FA and 5-FU/FA groups were 11.7% and 14.5% respectively in Study 011, and 10.5% and 9.0% respectively in Study 012. No statistically significant differences in duration of response were reported (actual values were not reported).
4.1.11 In Study 011, compared with 5-FU/FA, UFT/FA was associated with statistically significantly less diarrhoea (67% vs 76%, p = 0.006), nausea/vomiting (67% vs 75%, p = 0.02), mucositis (24% vs 75%, p < 0.001), neutropenia (13% vs 77%, p < 0.001) and thrombocytopenia (21% vs 31%, p < 0.001) of all toxicity severity grades. UFT/FA was also associated with less grade III/IV mucositis (1% vs 20%, p < 0.001), neutropenia (1% vs 56%, p < 0.001), thrombocytopenia (0% vs 2%, p = 0.003) and anaemia (3% vs 7%, p = 0.03). Increased bilirubin, without other liver function abnormalities, was statistically significantly more common in individuals treated with UFT/FA than in those treated with 5-FU/FA (39% vs 22%, p < 0.001). In Study 012, UFT/FA treatment resulted in statistically significantly fewer episodes of stomatitis/mucositis (18% vs 55%, p < 0.001), neutropenia (11% vs 67%, p < 0.001), thrombocytopenia (18% vs 28%, p = 0.025) and anaemia (76% vs 89%, p = 0.002) of any grade than 5-FU/FA treatment. UFT/FA treatment resulted in statistically significantly less grade III/IV stomatitis/mucositis (2% vs 16%, p < 0.001) and neutropenia (3% vs 31%, p < 0.001). In all, 127 individuals were hospitalised during the study: 59 (31%) in the UFT/FA group and 68 (37%) in the 5-FU/FA group (p-values not reported).
4.1.12 Health-related quality of life was measured in both studies using either the Functional Living Index-Cancer or EORTC QLQ-C30; no statistically significant differences in outcomes were reported between the treatment groups in either study. An unpublished preliminary report of a phase II randomised study in 202 individuals indicated that scores for functional and symptom scales were either improved or unchanged in the UFT/FA group but worse in the 5-FU/FA group.
4.1.13 The only information available on preferences for treatment was a 37-patient crossover study in which individuals received either UFT (300 mg/m2/day) plus FA (90 mg/m2/day) for 28 days every 5 weeks, or intravenous FU (425 mg/m2/day) plus FA (20 mg/m2/day) for 5 days every 4 weeks. They were then crossed-over to the other treatment regimen for the second treatment cycle. Therapy preference questionnaires were completed before the first and after the second treatment cycle. Of the 31 individuals who completed the questionnaire, 84% preferred the UFT/FA regimen. The reasons for this preference included being able to take medication at home, experiencing less stomatitis and diarrhoea, and being able to use a tablet instead of having an injection.
4.2.1 Two economic evaluations of capecitabine compared with 5-FU/FA were identified, one conducted by the manufacturer and the other by the Assessment Group. Both evaluations assumed equivalent effectiveness, and thus only evaluated associated costs from an NHS perspective. Both models included costs associated with drug acquisition, chemotherapy administration (including inpatient stays) and adverse event management.
4.2.2 The manufacturer estimated the costs of capecitabine and 5-FU/FA (using the Mayo bolus regimen) to be approximately £2700 and £5000, respectively. The Assessment Group estimated these costs to be £2100 and £3600, respectively. The Assessment Group also estimated the cost of 5-FU/FA to be £6300 when the de Gramont infusional regimen was used and £3500 when the modified de Gramont infusional method, which does not generally require inpatient administration, was used. In all instances, capecitabine was the least costly treatment option.
4.2.3 Both the manufacturer and the Assessment Group conducted economic analyses that compared UFT/FA with 5-FU/FA; both assessed costs from an NHS perspective and included categories of costs such as drug acquisition, chemotherapy administration (including inpatient stays), and adverse event management. A cost-minimisation study was also identified, although it was of limited use because it was from a non-UK perspective and did not specify the comparator regimen (for example Mayo or de Gramont).
4.2.4 The manufacturer's cost-effectiveness analysis compared UFT/FA with 5-FU/FA based on a Mayo regimen. The analysis used adverse events as the health outcome of interest although the evaluation was conducted separately for the two RCTs, and the costs of UFT/FA and 5-FU/FA were found to be £3600 and £6100 respectively for Study 011, and £3200 and £4900 respectively for Study 012.
4.2.5 The Assessment Group's cost-minimisation analysis showed a cost of approximately £3500 both for UFT/FA and for 5-FU/FA, administered using either the Mayo or modified de Gramont outpatient-based regimen.
4.3.1 The Committee reviewed the evidence available on the clinical and cost effectiveness of capecitabine and UFT, having considered evidence on the nature of the condition and the value placed by users on the benefits of capecitabine and UFT/FA by people with colorectal cancer, those who represent them, and clinical experts. It was also mindful of the need to take account of the effective use of NHS resources.
4.3.2 In the absence of patient preference data from adequately designed studies, the Committee took particular note of the opinions of both the professional and patient representatives regarding the advantages of oral compared with intravenous administration of chemotherapy, and of the potential problems of concordance with oral treatments. The patient representatives particularly emphasised that the vast majority of individuals expressed a strong preference for oral drugs provided that effectiveness was not compromised, because they reduce the disruptive impact of chemotherapy on individuals' lives and give them greater control over the management of their disease.
4.3.3 The Committee was satisfied that the phase III RCT data demonstrated that both capecitabine and UFT/FA were likely to have clinical effectiveness similar to that of 5-FU/FA administered by the bolus Mayo regimen. The appropriateness of using the Mayo regimen as a comparator was questioned because of evidence that suggests that infusional regimens may be more effective and less toxic. Indirect comparison of the oral drugs with infusional regimens might therefore suggest that the oral drugs are less effective. However, this evidence has not been formally appraised, and both the professional experts and the Assessment Group questioned its robustness. The Committee did not therefore consider it sufficiently conclusive to be the basis of a recommendation against the use of the oral treatments. However, the Committee also firmly believed that an appropriately designed RCT was required to carry out a direct comparison of the effectiveness of the oral treatments versus the infusional regimens. In addition, the Committee considered there was insufficient evidence to enable a distinction to be made in terms of effectiveness between the two oral agents.
4.3.4 There are also differences in the contraindications and side-effect profiles of the individual oral and intravenous regimens, and the Committee appreciated that the choice of the most appropriate treatment regimen might depend on the individual's circumstances. The Committee therefore concluded that intravenous regimens may be preferable under certain circumstances, and that capecitabine and UFT/FA should thus be available as options for treatment rather than as the preferred choice.
4.3.5 The Committee considered that, given the lack of compelling evidence for a difference in effectiveness between the regimens, the correct approach to evaluation of cost effectiveness was cost minimisation. They took note of the variations in the estimates of the total costs obtained from the submitted models, and overall were convinced that the oral drugs were cost-effective compared with 5-FU/FA regimens, principally on the basis of the potential cost savings related to the method of administration. They were also aware that the reduced burden of preparation and administration on specialist staff might potentially allow reallocation of clinical resources.