Late-onset neutropenia following rituximab treatment in systemic lupus erythematosus-a role of the BAFF/APRIL pathway

Background: Rituximab-mediated late-onset neutropenia (LON) has been studied in various diseases, but data from systemic lupus erythematosus (SLE) are limited. Objectives: To study the prevalence and contributing factors for LON following treatment with rituximab in patients with SLE, including B cell related cytokines and growth factors of the myeloid lineage. Methods: Patients from the Karolinska SLE cohort treated with rituximab (n=107) were enrolled in this observational study. Rituximab was given according to the lymphoma course (weekly for four weeks), the arthritis course (at week 0 and 2), or as a single infusion, with or without concomitant pulses of cyclophosphamide. LON was defined as an absolute neutrophil count <1,500 cells/μL, occurring four weeks to two years after initiation of rituximab treatment, provided that other apparent causes were excluded. Neutropenia occurring later than two years after treatment initiation but during sustained B cell depletion were also considered LON. B lymphocyte stimulator (BLyS/BAFF), a proliferation-inducing ligand (APRIL), interleukin 6 (IL-6), granulocyte colony stimulating factor (G-CSF) and granulocyte-macrophage colony stimulating factor (GM-CSF) were measured by ELISA prior to treatment (n=70) and either at the incidence of LON in patients who developed LON or after approximately the same median time following rituximab treatment in patients who did not develop LON (n=52). Results: Thirty-four of 107 patients developed LON after a median time of 222 days (IQR: 105-355 days). BLyS levels increased from baseline (median: 0.62 ng/mL; IQR: 0.42-1.07 ng/mL) through the post-treatment measurement, both in patients who developed LON (median: 1.73 ng/mL; IQR: 1.03-2.13 ng/mL; P=0.005) and patients who did not (median: 1.03; IQR: 0.67-1.56 ng/mL; P<0.001), but the increase was greater in patients who developed LON, resulting in significantly higher post-treatment BLyS levels (P=0.029). BLyS levels did not differ between the two groups at baseline (P=0.745). We observed a numerical increase in APRIL levels from baseline (median: 1.29 ng/mL; IQR: 0.85-2.3 ng/mL) through the post-treatment measurement in patients who developed LON (median: 2.39; IQR: 1.08-5.16 ng/mL; P=0.074) and a numerical decrease in patients who did not (median: 1.11 ng/mL; IQR: 0.77-1.64 ng/mL; P=0.064), resulting in significantly higher post-treatment APRIL levels in the LON group (P=0.032), from being similar at baseline (P=0.125). We found no difference in levels of G-CSF, GM-CSF or IL-6 between patients who developed LON and patients who did not, either at baseline or at the post-treatment measurement. Higher prednisone dose administered concomitantly to rituximab (P=0.003) and younger age (P=0.001) were found to be associated with the development of LON, whereas neither the use nor the doses of cyclophosphamide were found to have any impact. Conclusions: The prevalence of rituximab-mediated LON within the SLE patients of the current study (31.8%) was higher compared to previous reports on patients with lymphoma (3-27%), ANCA-associated vasculitis (11.9%) and rheumatoid arthritis (3%). Our results imply a role of the BAFF/APRIL pathway in the immunologic mechanisms underlying this phenomenon and demonstrate that LON following rituximab treatment is a common complication in SLE patients.