4 Evidence

4.1 Clinical effectiveness

4.1.1 There are two randomised controlled trials (RCT) evaluating intravenous fludarabine as a second line agent in CLL. In one of these, comparing fludarabine with cladribine, an unknown number of patients had B-cell CLL. It is not possible therefore to derive from this study, results relevant to the indication under consideration.

4.1.2 The second RCT compared fludarabine with CAP. Only 96 of 196 patients studied had been previously treated with a first line agent. Response rates (RR) for these previously treated patients were 48% with fludarabine versus 27% with CAP (95% confidence interval on the difference of 21 percentage points was 2 to 40 percentage points). The fludarabine response rate was much higher than that seen in the case series (see below). Mostly, the responses were partial, with just 13% complete responses for fludarabine versus 6% for CAP (the difference is not statistically significant). Time to progression for previously treated patients (responders only) had a median of 324 days for fludarabine versus 179 days for CAP (the difference is not statistically significant). In addition there were 9 deaths in 48 patients with fludarabine compared with 3 out of 48 in the control arm (the difference is not statistically significant). No data on quality of life were collected as part of the trial. Adverse events, mostly haematological, were common in both arms of the trial.

4.1.3 For intravenous fludarabine, six case series with an aggregate population size of almost 1000 were found. Response rates averaged 32%. The response rate appears to decline in patients who have been treated more heavily prior to treatment with fludarabine, and is generally lowest for patients for whom fludarabine is the treatment of last resort. Where reported, mild or moderate side effects were common although 72% of patients experienced severe haematological adverse events (based on laboratory results). The degree and length of response in these studies is greater than those of comparable studies for combination therapy, although it is not known whether there is an impact on length of survival. The quality of all such studies is lower than that of randomised controlled trials, due to the possibility of unrepresentative or biased patient selection, heterogeneity between studies, and placebo effects of unknown size.

4.1.4 Quality of life data have not been formally collected as part of these studies. Some evidence, however, is available from the Lymphoma Association, whose survey showed that 74% of patients consider that they are in as good or better health during fludarabine treatment as they would have been with other chemotherapy.

4.1.5 Recent data from the MRC-sponsored CLL3 study show that, for patients who failed to respond to first-line treatment with chlorambucil or chlorambucil plus epirubicin at 6 months, 80% responded to fludarabine (17% complete, 63% partial). This evidence is in general supportive of the use of fludarabine second line. It is, however, subject to the biases of being an uncontrolled case series, the difficulty in defining non-responders and the variation in times to recruitment into the study following 'failure' of first line treatment.

4.1.6 Other evidence in the form of four randomised controlled trials for fludarabine alone against other chemotherapeutic agents for first line treatment (for which fludarabine is not currently licensed) shows that fludarabine gives higher rates of remission and longer remission than CAP, CHOP, or chlorambucil plus prednisolone, but no demonstrated survival advantage. For fludarabine against chlorambucil, one study favours fludarabine as above, but the other finds no difference between the two drugs. Assuming that a high positive correlation exists between the relative performance of fludarabine against its comparators in first and second line therapies, this evidence supports the case for the use of fludarabine in second-line therapy.

4.1.7 For oral fludarabine, an open study of 78 previously treated patients with CLL showed that 46% of patients responded (20% complete and 26% partial remission). Evidence on the bioavailability of the oral formulation and equivalence of blood levels of oral fludarabine with those of the intravenous preparation indicates that orally administered fludarabine is likely to have similar clinical efficacy to the intravenously administered preparation.

4.2 Cost effectiveness

4.2.1 No single piece of evidence is strong enough to establish whether fludarabine is a clinically effective agent for second line treatment of CLL. When the evidence is taken in aggregate, however, it is supportive of fludarabine as a clinically effective drug for this indication. The assessment of cost effectiveness, however, is also subject to considerable uncertainty, because reliable figures for the costs of dealing with adverse side effects are not available.

4.2.2 Three sources have been used to examine the costs of overcoming adverse events in using fludarabine and CHOP. The first (described as the low estimate), submitted by the manufacturer, is based on very small numbers of patients, and is therefore subject to large sampling error. The second (described as the high estimate) was submitted by a different manufacturer for the treatment of a different type of lymphoma, and was also based on a small sample. This source yielded cost estimates over six times as high as the first source, much of which appears to be attributable to the disease having advanced much further, on average, in patients from the second source. The third was from the MRC trial CLL3, which may have included costs associated with the disease per se, as well as of side effects of fludarabine, yielded costs comparable with those of the high estimate.

4.2.3 The cost effectiveness estimates have been examined separately for oral and intravenous fludarabine, and for both of these against the comparators of CHOP and of no treatment, for both high and low estimates of the cost of side effects.

4.2.4 Oral fludarabine is less costly to administer than intravenous fludarabine. The estimated combined cost of acquisition and administration, based on an average of 4.1 cycles administered, is £3,000 for oral fludarabine (comprising £2,700 for acquisition and £300 for administration) against £5,300 for intravenous fludarabine (£2,700 acquisition, £2,600 administration).

4.2.5 The mean cost per year of remission from CLL of oral fludarabine against no treatment is estimated to be £9,000 (low cost of treating side effects) or £21,000 (high cost of treating side effects); for intravenous fludarabine, the corresponding estimates are £14,000 and £28,000; and for CHOP are £22,000 and £67,000.

4.2.6 The mean incremental cost-effectiveness ratio for oral fludarabine against CHOP is estimated to be £2,700 per year of remission (low cost of treating side-effects) and £200 per year of remission (high cost of treating side-effects) (the latter figure is smaller than the former because the cost of treating side-effects for CHOP is also much higher in the high-cost scenario). The corresponding estimates for intravenous fludarabine against CHOP are £10,600 and £10,500. Even if there is no increase in overall survival using fludarabine compared with that of CHOP, it is probable that oral fludarabine is cost effective against CHOP.

4.2.7 In clinical practice it is likely that if fludarabine were not given, then patients who would otherwise have been prescribed it would be prescribed CHOP or an equivalent combination. In this situation oral fludarabine is both a more clinically effective and a more cost effective drug than its alternatives, as well as being more likely to be acceptable to patients.