Abstract
There are limited data on the relative efficacy of initial therapies for follicular NHL, in part due to the large sample size and prolonged follow-up required for comparative trials of induction regimens, and now complicated by the use of maintenance rituximab. To gain perspective in the comparison of different induction regimens (separate from rituximab maintenance), we performed a meta-analysis of trials in frontline indolent NHL, focusing on the clinical outcome measure of complete response rate (CRR) and its relation to disease progression. A literature search of clinical trials in untreated follicular NHL published from 2001–2006 was performed on Medline, Embase, Current Contents, and Biosis. Baseline characteristics, ORRs, CRRs, and time-to-event rates (TTP, PFS, EFS) were extracted. Studies were included only if they reported follicular-specific data. Therapies were induction and/or consolidation therapy, excluding trials with rituximab maintenance. Treatment categories were chemotherapy combinations without fludarabine (CHEMO); rituximab as a single-agent or in combination with chemotherapy (R±CHEMO); fludarabine as a single-agent or in combination (FLU); and ibritumomab tiuxetan or tositumomab as single agents or in combination (RIT). The Kruskal-Wallis test was used to compare baseline characteristics. CRRs were analyzed using the fixed effect model (Mantel-Haenszel) and the random effect model (DerSimonain and Laird). Reported time-to-event rates required a conversion to a standard monthly event rate (HR), assuming an exponential distribution. The Pearson correlation (rho) between the CRR and HR was estimated. 32 treatments from 25 publications with 2421 patients met the inclusion criteria—8 CHEMO, 13 R±CHEMO, 5 FLU and 6 RIT. Only median age and disease stage were consistently reported. Median age and proportion of stage 3–4 disease did not differ between treatment groups. Heterogeneity among the treatments was analyzed using an analog of ANOVA. There was a significant difference in CRRs between CHEMO, R±CHEMO, FLU and RIT (P=0.03). Random effect estimation showed a CRR of 37% with CHEMO (95% CI: 18%–57%), 53% with R±CHEMO (34%–71%), 68% with FLU (49%–87%), and 79% with RIT (73%–85%) (fig 1). There was a highly significant linear correlation between the CRR and HR for disease progression with all treatments (rho= −0.79; 95% CI: −0.57 to −0.91; P<0.001). Despite dissimilarities in therapy and patient populations, and heterogeneity of ORRs and CRRs within treatment groupings, significant differences in CRRs persist. The analysis suggests that a higher CRR is correlated with a lower hazard of disease progression. These data support the selection of regimens that achieve high CRRs for future trials of initial therapy and could provide an additional basis for the design of long-term therapeutic strategies.
Complete Response Rate by region strategy
Disclosures: Regimens for the treatment of follicular NHL were included in the analysis regardless of their licensure.; M. Wayne Saville, Mohamed Darif, and Craig Park are employees of Biogen Idec.; M. Wayne Saville, Mohamed Darif, and Craig Park have stock options in Biogen Idec.; Stephen J. Schuster from Biogen Idec.; Mitchell Smith on Speakers bureau for Biogen Idec, Genentech, Berlex.
