3 The manufacturer's submission
3.1 The manufacturer's submission approached the decision problem by comparing fludarabine monotherapy and fludarabine plus cyclophosphamide with chlorambucil. The population under consideration was defined as people with CLL, who were chemotherapy naïve, had sufficient bone marrow reserves and had advanced symptomatic Binet stage B or C disease or evidence of progressive disease in Binet stage A. The primary outcome measures considered were progression-free survival and health-related quality of life. Secondary outcome measures included treatment response rates, incidence of adverse events and overall survival.
3.2 The manufacturer's submission included evidence on the clinical effectiveness of fludarabine monotherapy and fludarabine plus cyclophosphamide compared with chlorambucil. Only clinical evidence relating to fludarabine monotherapy, within its licensed indication, is presented in this section. Seven randomised controlled trials (RCTs) were identified to be relevant to the decision problem, of which two were published and five were available in abstract form. However, the CLL4 trial was considered to be the most relevant by the manufacturer because it is the largest study and the only one directly comparing fludarabine monotherapy, fludarabine plus cyclophosphamide and chlorambucil. The CLL4 trial enrolled 777 patients, of whom 194 were randomised to fludarabine monotherapy, 196 to fludarabine plus cyclophosphamide and 387 to chlorambucil. Follow-up is ongoing. Early results of the CLL4 trial published in abstract form and presented in the manufacturer's submission showed that overall treatment response rates were 77% for fludarabine monotherapy and 69% for chlorambucil (no p values or confidence intervals were reported). Early results from the CLL4 trial reported 3-year progression-free survival of 31% for fludarabine monotherapy and 23% for chlorambucil. Analysis of overall survival showed no difference between the treatment regimens.
3.3 The manufacturer's submission contained an economic analysis comparing fludarabine monotherapy, fludarabine plus cyclophosphamide and chlorambucil. Only the economic evidence for fludarabine monotherapy compared with chlorambucil is presented in this section. The economic analyses were based on a Markov state transition model with a 20-year time horizon. The economic model used patient-level data from the CLL4 trial to inform first-line treatment, with data for second-line and salvage treatments taken from a variety of published sources. The manufacturer submitted revised base-case economic analyses following clarifications requested by the ERG. These showed an incremental cost-effectiveness ratio (ICER) of £26,105 per quality-adjusted life year (QALY) for fludarabine monotherapy compared with chlorambucil.
3.4 The ERG raised a number of issues and uncertainties relating to the clinical and cost-effectiveness evidence presented in the manufacturer's submission. The ERG stated that the clinical effectiveness evidence had to be interpreted with caution because the follow-up period for the CLL4 trial was not complete at the time of submission. The ERG assessed the manufacturer's economic model and noted that the main drivers of the ICERs presented were time horizon and rates of response to retreatment with the same chemotherapeutic agent as that used in first-line treatment.
3.5 The ICER for fludarabine monotherapy compared with chlorambucil for a 15-year time horizon was £28,178 per QALY. For 10-year and 5-year time horizons the ICERs were £42,516 per QALY and £310,663 per QALY, respectively. The ERG stated that the extrapolation of model data is likely to be central to the validity of the ICERs presented. The ERG also noted that an assumption of constant transition probabilities over time was used within the manufacturer's model. Because patient-level data from the CLL4 trial were available, the ERG stated that this assumption should have been validated using formal survival analysis. It therefore performed a survival analysis using patient-level data from the CLL4 trial, the results of which showed that the assumption of constant transition probabilities is not supported. However, incorporating the results of the ERG's survival analysis into the economic model would have required a substantial restructuring of the model. Correcting this assumption was expected to increase the ICER for fludarabine monotherapy compared with chlorambucil.
3.6 The ERG report noted the way in which retreatment response rates for fludarabine monotherapy and chlorambucil were modelled. For fludarabine monotherapy, the first-line treatment response rate (77%) was taken from the CLL4 study and the retreatment response rate (74%) was taken from the existing literature as presented in the manufacturer's submission. For chlorambucil, the first-line treatment response rate (69%) was taken from the CLL4 study and the retreatment response rate (35%) was taken from the existing literature as presented in the manufacturer's submission. This led to a base-case ICER of £26,105 per QALY for fludarabine monotherapy compared with chlorambucil. Because no retreatment response rates were available from the CLL4 study, and because of the limited evidence available in existing literature, the manufacturer's submission presented a one-way sensitivity analysis in which retreatment response rates were assumed to be the same as first-line treatment response rates for all the treatment arms in the economic model. This resulted in an ICER of £86,770 per QALY for fludarabine monotherapy compared with chlorambucil.
3.7 The ERG noted that the manufacturer's submission stated that improved progression-free survival with fludarabine monotherapy was linked to improvements in quality of life in the CLL4 trial. However, the impact of the adverse effects of fludarabine, specifically the potential additional costs related to increased hospitalisations due to infections, was not explored in the manufacturer's economic model. Although the manufacturer's model included sensitivity analysis to assess potential decreases in utilities and quality of life as a result of adverse events, the ERG considered that the omission of the treatment costs of adverse events was likely to have resulted in an underestimation of the ICERs for fludarabine monotherapy compared with chlorambucil.
3.8 Full details of all the evidence are in the manufacturer's submission and ERG report (see appendix B).