Interpreting long-term response to rituximab in immune thrombocytopenia: Cohort construction, treatment attribution, and dynamic prognostication

Abstract

We read with great interest the study by Tsykunova et al.¹ published in HemaSphere. Drawing on a large Nordic cohort, the authors comprehensively characterized initial and sustained long-term responses to rituximab in patients with immune thrombocytopenia (ITP), providing valuable real-world evidence. We wish to add several considerations regarding cohort construction, concomitant corticosteroid use, and long-term outcome assessment that may further inform the interpretation and clinical implications of these findings.

Of the 759 patients, 444 received rituximab before fulfilling the registry criteria for chronic ITP, although cohort eligibility required at least two ITP diagnoses more than 6 months apart. The upfront group, comprising 123 previously untreated patients, may therefore represent early-treated patients with subsequent disease documentation rather than all patients receiving rituximab early. Durable early responders may have fewer subsequent ITP encounters, whereas patients with persistent disease remain more likely to be documented and followed. Such selection related to subsequent disease trajectories may affect the representativeness of this group, although its direction and magnitude remain uncertain.²

The authors appropriately acknowledge that concomitant corticosteroids may partly explain the longer response duration in the upfront group. Corticosteroid use differed markedly between the upfront and pretreated groups (74.0% vs. 15.6%), suggesting that the observed durability may reflect both treatment timing and combination therapy. Randomized trials have shown that rituximab plus dexamethasone improves sustained response and delays relapse compared with dexamethasone alone.^{3, 4} Moreover, excluding platelet-elevating therapy only during the 8 weeks preceding response assessment may not fully separate the contributions of corticosteroids given at rituximab initiation. If available, outcomes stratified by concomitant corticosteroid use and regimen could further clarify the relationship between early rituximab use and durable response.

A further consideration concerns the assessment of long-term response. Because both rituximab treatment and registry follow-up ended in 2018, patients had unequal opportunities to complete 2- or 5-year follow-up. Although SR2 and SR5 were calculated among initial responders, the numbers with sufficient potential follow-up at each landmark remain unclear. These estimates may therefore reflect both response durability and observable follow-up time. Reporting landmark-eligible responders and numbers at risk, or providing time-to-event estimates, could help contextualize the 31.4% SR5 rate. The STROBE statement likewise emphasizes transparent reporting of follow-up duration, numbers analyzed, and loss to follow-up.⁵

Complete response (CR) at 3–6 months was closely associated with SR2 and SR5; however, it is inherently a dynamic post-treatment measure rather than a pretreatment predictor. Its clinical value may therefore lie in identifying patients likely to maintain long-term remission and informing subsequent risk stratification and follow-up, rather than guiding pretreatment selection for rituximab. Consistent with this interpretation, Patel et al.⁶ similarly found that initial CR was associated with sustained response at 5 years.

More broadly, the findings highlight two distinct questions that are often combined in observational studies of rituximab: whether earlier initiation causally improves treatment durability and which treated patients are likely to maintain response. The first requires comparison of clearly defined treatment strategies from a common time zero, whereas the second concerns prognostic updating after treatment. Future registry studies could emulate a target trial by aligning eligibility assessment, treatment-strategy initiation, and the start of follow-up at a common time zero, while distinguishing rituximab monotherapy, rituximab–corticosteroid combination therapy, and relevant alternative strategies.⁷ Among rituximab-treated patients, prognosis could then be updated dynamically using early CR, platelet trajectories, and rescue-treatment requirements. Such a two-stage causal–prognostic framework may help separate treatment-effect estimation from post-treatment risk stratification and better translate real-world evidence into individualized ITP management.

Overall, this study provides important real-world evidence on the long-term effectiveness of rituximab. Accounting for cohort construction, concomitant treatment, and observable follow-up time may clarify the relationship between early rituximab use and durable response. Separating causal treatment-effect estimation from dynamic post-treatment prognostication may further strengthen the translation of registry evidence into individualized ITP management.

Xin Zhou: Writing—original draft. Ningning Shan: Writing—review and editing.

The authors declare no conflicts of interest.

This work was supported by grants from the Natural Science Foundation of Shandong Province (Grant No. ZR2026MH266) and the Shandong University Horizontal Research Projects (Grant Nos. 6010226027 and 6010225019).

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.