Note from the National Guideline Clearinghouse (NGC): The National Institute for Health and Clinical Excellence (NICE) commissioned an independent academic centre to perform a systematic literature review on the technology considered in this appraisal and prepare an Evidence Review Group (ERG) report. The ERG report for this technology appraisal was prepared by the Liverpool Reviews and Implementation Group (see the "Availability of Companion Documents" field).
Clinical Effectiveness
Clinical Evidence
The two relevant randomized controlled trials (RCTs) were included in the manufacturer's systematic review (SR) are the European Organisation for Research and Treatment of Cancer (EORTC) and German Low grade Lymphoma Study Group-Fludarabine, Cyclophosphamide and Mitoxantrone (GSLG-FCM) trials. Both trials are phase III, multi-centre, randomized, open-label trials designed to evaluate the efficacy and safety of: (i) rituximab in combination with chemotherapy in inducing a remission in follicular non-Hodgkin's lymphoma (FL) and (ii) rituximab as maintenance therapy for FL patients.
Description and Critique of Manufacturers Approach to Validity Assessment
The manufacturer's submission (MS) provides completed validity assessments for the two included trials; no details on who conducted the validity assessment or how it was conducted are detailed. Both trials are considered to be high quality trials although it is noted both were open-label trials and it is unclear from the published papers whether assessors were aware of treatment allocation. The MS also highlights that the GLSG-FCM trial did not use an intention to treat analysis as stated, but rather 19 patients were excluded due to inadequate documentation or were withdrawn between randomisation and delivery of study treatment. Also, analysis of the clinical evidence from the GLSG-FCM maintenance phase is limited as it is unclear how many patients completed the maintenance treatment. The table of validity assessment provided in the MS is included in Appendix 7.2 of the ERG report.
Description and Critique of the Statistical Approach Used in the MS
The MS includes a thorough description of the statistical approaches used in the EORTC trial and the GLSG-FCM trial.
EORTC
The EORTC trial used a chi-square test for trend to examine response rates, with a significance threshold of P<0.001, with early stopping allowed if this threshold was crossed. For the secondary endpoints of event-free survival (EFS), overall survival (OS) and the exploratory endpoint progression-free survival (PFS), a log-rank test using a two-sided alpha level of 5% was used. Kaplan-Meier (K-M) curves were produced to graphically display the unadjusted difference between the treatment arms. Results were presented as risk ratios with 95% confidence intervals (CIs) reported. In the maintenance phase, the primary endpoint of PFS was based on a log-rank test stratified according to induction treatment; secondary and exploratory endpoints were unstratified. For OS in the maintenance phase, an unstratified log-rank test using a two-sided alpha level of 5% was used for the primary analysis and secondary analyses were done by the Cox regression analysis and the results presented as risk ratios including 95% CIs.
The statistical approaches used in trial EORTC are generally appropriate. However, it is unclear as to why a significance threshold of P<0.001 was used instead of P<0.05 as is usually used when Haybittle-Peto's rule is applied. Furthermore, the trial protocol outlines that secondary analyses in the maintenance phase of the trial will use Cox regression analysis with adjustment for stratification factors and other potential prognostic factors and that the secondary endpoint of OS will be analysed with a log-rank test, stratified for the same factors as PFS. In the MS all analyses of secondary outcomes were unstratified; the reason for this deviation from the protocol is unclear. Finally, EFS results are not reported.
Whilst the EORTC trial allows comparison of the four alternative treatment strategies contained within the trial, as depicted by the 4-arm economic model, the trial was not powered or designed for this specific purpose.
GLSG-FCM
The GLSG-FCM trial used a 1-sided triangular sequential test with a significance level of 0.05, for both the induction and maintenance phases. Exploratory analyses were performed for histological subgroups, the PFS from the start of therapy and OS. The Fisher test was used for analyses of binary responses and the log-rank test and univariate Cox regression for time-censored analyses.
The statistical analyses performed in the GLSG-FCM trial appear to be appropriate. However, only limited results are available for FL patients.
Summary Statement
The SR was adequately conducted by the manufacturer. The two trials included in the SR were of good quality and the primary outcome measures reported in the MS were considered to be appropriate. As specific clinical results for the FL patients in the GLSG-FCM trial were not fully reported in the published papers, the value of the trial results is therefore limited, particularly for patients in the maintenance phase, as they are not focussed on the patient population of interest in this single technology appraisal.
Cost Effectiveness
Critique of Data Extraction and Quality Assessment
The manufacturer presented summary details of the nine studies included in the review in a table which included the following categories: study, aims, methods, results and relevance to decision-making in England and Wales. It is unknown whether or not a second reviewer conducted independent data abstraction or how any discrepancies were discussed. The manufacturer did not state whether quality assessment of the included studies had been undertaken.
Summary of Cost-Effectiveness Evidence Identified
The manufacturer presented summary details of the nine included studies. The studies were published during the period 1999-2006. Seven of the studies were cost analyses. Only one cost-effectiveness study included rituximab maintenance as a comparator (rituximab maintenance versus autologous stem cell transplant). In a cost-minimisation analysis, the authors assumed that there was no significant difference between the treatments in terms of response rates and disease duration yet went on to describe differences in the incidence and severity of drug-related adverse events. Only two of the nine included studies were conducted in the United Kingdom.
Conclusions
The systematic literature review of the economic evidence conducted by the manufacturer was poor. The ERG concludes that direct or meaningful comparison of the included studies was not possible due to the fact that the economic analyses were very different. In particular, the studies were heterogeneous in terms of the comparators, approaches to costing and country of origin.