Can response duration after the first rituximab treatment be used in timing of rituximab retreatment?

Abstract

The optimal strategy concerning the timing of rituximab (RTX) retreatment in patients with rheumatoid arthritis (RA) is yet to be determined (1). Four different approaches can be conceived. The most prevalent approach is ondemand retreatment when the disease activity increases, based on judgement of the physician, after initial response. This, however, inevitably results in temporary deterioration of disease activity and inferior disease control and is possibly associated with worse outcome (2, 3). A second strategy includes a fixed retreatment schedule. However, because response duration varies considerably between patients, the choice of the optimal interval at a group level is difficult. These two strategies have been compared head to head, and no differences in disease activity and joint damage progression were found after a 1-year follow-up (4). A third approach is a treatment-to-target strategy, which is a more formalized adaptation of the first strategy. Emery et al compared a treatment-to-target strategy with a treatment-asneeded strategy, and revealed more patients with major clinical response in the treatment-to-target strategy group (5). A fourth alternative strategy could be to retreat a patient when there is loss of response (LoR), thereafter using a fixed interval based on the first response duration. However, this is only feasible when the intra-individual variation in response duration is low. In this retrospective study, RA patients treated with at least three courses of RTX, according to the on-demand retreatment strategy, were included. Data were collected from patient charts by two research physicians. The date of LoR was operationalized in two different ways: (i) the start date of the (first infusion of the) next cycle; (ii) the date of clinical LoR, which was based on patient chart review by the research physicians. This last (sensitivity) analysis was added because the date of retreatment could be biased by other factors, such as patient preferences or (non)availability of outpatient-care infusion capacity. Seventy RA patients were included; Table 1 shows their baseline characteristics. The dosage of the first RTX treatment cycle was 2 1000 mg in 69 patients, and in one patient the dosage was unknown. The dosage of the second RTX cycle was 2 1000 mg in 57 patients, 2 500 mg in six patients, and 1 1000 mg in seven patients. Concomitant disease-modifying anti-rheumatic drug (DMARD) treatment changes were infrequent and comparable between the first and second intervals. Forty patients were treated with systemic corticosteroids at baseline; changes in systemic steroid treatment were also comparable between the first and second intervals. The mean duration until retreatment was 301 (SD 95) days for the first and 341 (SD 123) days for the second RTX infusion interval. When the interval was measured between infusion until LoR, the mean duration was 252 (SD 93) days for the first and 307 (SD 126) days for the second interval. Thus, the mean difference between the first and second intervals was 40 days (SD 119, p 1⁄4 0.003) when measured as time between infusions, and 55 days (SD 127, p 1⁄4 0.0003) when operationalized as clinical LoR. The correlation between the first and second RTX intervals was only low to moderate with r 1⁄4 0.43 (p 1⁄4 0.0002), and when using the clinical LoR r 1⁄4 0.36 (p 1⁄4 0.002). Figure 1 presents the Bland–Altman analyses between the first and second infusion intervals, with limits of agreement of –190 and þ272 days. We hypothesized that the duration of response after the first RTX cycle might be used for individual timing of retreatment, thus reducing both overand undertreatment.