Rituximab for the treatment of relapsed or refractory stage 3 or 4 follicular non-Hodgkin's lymphoma

The Committee considered the use of rituximab as recommended (only in the context of a prospective case series) in the original appraisal (TA37): as monotherapy for the treatment of patients with relapsed or refractory stage 3 or 4 follicular non-Hodgkin's lymphoma, when all alternative treatment options have been exhausted (that is, when there is resistance to or intolerance of chemotherapy). Resistance was defined as the absence of a response to the last course of chemotherapy following several courses of treatment. The guidance also applied to people who have or are likely to become intolerant of chemotherapy because of adverse effects that severely limit the safety of further treatment. The Committee was concerned that no new evidence for this use of rituximab was made available, particularly because it had been a condition of use that patients only received rituximab if their data were collected as part of a case series. It agreed with the ERG that there was a lack of clarity in the manufacturer's search for relevant new data. The clinical specialists pointed to the diminishing use of rituximab at last line because it has been licensed and recommended for earlier use as first-line therapy (TA110; replaced by TA243), and they felt that no new evidence would now be collected at later stages of the treatment pathway. They stated that rituximab was still needed as a last-line option for frail patients in whom the use of toxic chemotherapy may not be possible. The Committee agreed that the previous recommendation should stand, but recognised the limitations of treatment recommendations based on the need for future data collection. The Committee concluded that rituximab monotherapy should continue to be recommended as an option for the treatment of patients with relapsed or refractory stage 3 or 4 follicular non-Hodgkin's lymphoma, when all alternative treatment options have been exhausted.

The Committee considered the evidence for the clinical effectiveness of rituximab monotherapy as maintenance therapy and the use of rituximab as induction therapy in relapsed follicular non-Hodgkin's lymphoma. Both of these indications were evaluated in the EORTC trial, and in both indications rituximab-treated patients had higher response rates and longer remissions than control patients. The Committee heard from clinical specialists that rituximab was a valuable treatment and that the possibility of maintaining remission was particularly encouraging. The Committee also heard from the patient representatives of the importance attached by patients to the remission state and the value of maintenance treatment in prolonging it.

The Committee accepted that the EORTC trial demonstrated the effectiveness of rituximab for maintenance therapy and for the induction of remission in patients at first and second relapse. The Committee was mindful that the results from the EORTC trial were based on rituximab-naive patients. The number of rituximab-naïve patients is decreasing due to increased prescribing of rituximab as first-line therapy. The Committee considered that it was necessary to be cautious about the assumption that rituximab is as efficacious in patients who had already received it as in patients who are rituximab-naive. The clinical specialists stated that the evidence indicated that follicular non-Hodgkin's lymphoma could be re‑treated with rituximab with little or no loss of efficacy. Although it noted this as an area of uncertainty, the Committee accepted that this was biologically plausible given its mechanism of action. The Committee noted that rituximab maintenance therapy was associated with similar increases in progression-free survival in patients who had received rituximab at induction as in those who had only chemotherapy for induction. It also noted that the use of rituximab maintenance therapy following induction with chemotherapy alone resulted in a greater reduction in the risk of progression compared with rituximab maintenance therapy following induction with combined chemotherapy plus rituximab. However, the Committee was aware that the trial was not powered or designed to evaluate the relative efficacy of rituximab maintenance in patients who had or had not received rituximab at induction. The Committee was therefore unable to draw firm conclusions on the relative efficacy of rituximab maintenance therapy in patients who had and had not received rituximab for induction.

The Committee also queried the extent to which induction therapy with CHOP was representative of the comparator treatments used in the NHS. The clinical specialists stated that CHOP was the main chemotherapy used in follicular non-Hodgkin's lymphoma patients at this stage but that fludarabine-containing regimens are also used to some extent. The Committee was also aware that the GLSG trial, using a fludarabine-based regimen, showed a similar magnitude of benefit as the EORTC trial, which used CHOP. However, the regimen of fludarabine used in the GLSG trial differed from that commonly used in the NHS and the trial population included people with other indolent lymphomas. The Committee considered it appropriate to focus its considerations on CHOP as the most important comparator within the NHS.

The Committee considered the manufacturer's economic model and the critique of it by the ERG. In particular, it discussed the costs included in the model and the approach to survival modelling and extrapolation. The Committee considered the changes to the costs in the manufacturer's model suggested by the ERG. It thought that it was appropriate to calculate costs at progression by aggregating treatments into categories, and it agreed with the ERG's assumptions as to how these would vary across the treatment strategies. It heard from clinical and patient specialists that, although the duration of second and subsequent infusions can sometimes be reduced to as little as 2 hours, for the most part, approximately 4 hours are necessary. The Committee also understood that the practice of rapid administration of rituximab was increasingly followed because its safety was now accepted by clinicians. The Committee concluded that it would currently be more appropriate to cost administration of rituximab as a day-case procedure than as an outpatient visit. The Committee also concurred with the ERG's approach of adding a terminal-care cost to the model and that the amounts assumed were appropriate. The Committee was not satisfied that the survival modelling adopted by the manufacturer was optimal and regarded the estimates resulting from the manufacturer's initial model as unreliable and requiring further analysis (see section 3.14).

The Committee first considered the cost effectiveness of rituximab when used as maintenance therapy. When considering rituximab monotherapy as maintenance, it was mindful that the clinical and patient specialists had strongly supported this use based on its potential to prolong remission, and the lack of alternative therapies for doing so. The Committee considered the results of the exploratory analysis performed by the ERG, which did not use model-based extrapolation but limited the analysis to 1,500 days of the Kaplan–Meier estimates and made changes to costs as suggested by the ERG above (see section 4.7). The Committee agreed that this approach overcame some of the concerns regarding the initial survival modelling. Based on this approach, the use of rituximab as maintenance only when compared with no use at all was associated with an ICER of approximately £13,000 per QALY gained. The Committee noted that the limited time horizon used in this approach could result a higher ICER than if a longer time horizon was used. The Committee concluded that, despite its concern that the EORTC population (from whom these calculations were derived) was not fully representative of UK patients (see section 4.5), the use of rituximab for maintenance therapy following induction of remission with chemotherapy was likely to be a cost-effective use of NHS resources.

Having accepted that rituximab maintenance therapy following induction with chemotherapy was likely to be a cost-effective use of NHS resources, the Committee then considered the use of rituximab as induction with chemotherapy. Given that patient experts and clinical specialists identified maintenance therapy as the clinical priority and that this is cost effective when compared with current standard treatment, the cost effectiveness of rituximab at induction was considered as an additional strategy over its use as maintenance alone. It is expected that rituximab maintenance will become standard therapy following this appraisal, and the Committee agreed that the appropriate comparator for the dual-use strategy was the next best use of resources – single use of rituximab as maintenance therapy as opposed to no use of rituximab.

The Committee therefore considered the use of rituximab in combination with chemotherapy for induction in addition to use for maintenance. It noted that the use of rituximab in induction increased the proportion of patients who entered remission and became eligible for rituximab maintenance. It also noted that the ERG's exploratory analysis suggested an ICER of approximately £43,000 per QALY gained when compared with chemotherapy only for induction followed by rituximab maintenance therapy. However, the Committee was aware that this figure resulted from a curtailed time horizon and required further analysis to obtain more reliable estimates. The Committee was also aware that there were a number of concerns with the manufacturer's survival modelling raised by the ERG (see section 4.7) and that further analysis (see section 4.11) was required before obtaining reliable estimates of cost effectiveness.

The manufacturer's original approach to the modelling of survival did not take account of the initial zero-hazard period, although there was a protocol-driven event-free period in the data. The Committee did not accept this approach because excluding the event-free period when fitting distributions could change the goodness-of-fit of any distribution fitted to trial data and influence the outcome of the cost-effectiveness analysis. The Committee noted that the additional analyses by the manufacturer excluded the event-free period when fitting functions and this resulted in an ICER of £21,379 using the Weibull model, with the exponential model being the best fit and resulting in an ICER of £16,183 per QALY gained (see section 3.14). These analyses were calculated over a lifetime horizon, assuming a duration of treatment benefit of 5 years, which the Committee considered to be reasonable.

The Committee also discussed that, in initially fitting the Weibull model to the RCT data, the manufacturer had made the assumption of proportional hazards such that the only difference between the fitted distributions was as a result of a treatment effect. The Committee requested further analysis because this strong assumption had not been substantiated by RCT data and could have resulted in an overestimation of the clinical and cost effectiveness of rituximab. The Committee noted that relaxing the proportional hazards assumption and independently fitting Weibull functions caused the ICERs to decrease to £15,775 per QALY gained (see section 3.14). It also noted that, in the further analysis from the manufacturer, terminal-care costs were not included, but it was aware that including such a cost made little difference to the ICERs. The Committee considered the ICERs of approximately £16,000 per QALY using extrapolation from distributions that had been shown to be a good fit to clinical data to be the most appropriate of those presented in the manufacturer's reanalysis. It was, however, mindful that this figure could be an underestimate of the cost per QALY gained because in practice patients would not usually be rituximab-naive, whereas those in the evidence base were. On balance, the Committee concluded that the use of rituximab in combination with chemotherapy as induction therapy was likely to be a cost-effective use of resources.