Background
Late-onset neutropenia (LON) is now recognized as one of the significant side effects of rituximab therapy that may lead to serious infections. LON is defined as an absolute blood neutrophil count <1.5 × 10
9/L occurring 4 weeks after the last rituximab infusion. The frequency of LON in hematological malignancies and autoimmune diseases is comparable. LON has been reported in 5–27% of patients with lymphoma treated with rituximab [
1,
2] and in 1.5–23% of patients with rheumatic diseases [
3‐
6]. Because of the morbidity associated with LON, it is important to understand the causes and mechanisms of this phenomenon.
The binding of rituximab to Fcγ receptors (FcγRs) of macrophages and natural killer (NK) cells is supposed to influence therapeutic efficacy of rituximab [
7]. A certain single-nucleotide polymorphism (SNP) in the
FCGR gene, the
FCGR3A 158V/V genotype (also called the
176V/V genotype), enhances ligation of rituximab to this receptor [
8,
9], and it has been associated with improved clinical response in non-Hodgkin lymphoma and rheumatoid arthritis (RA) [
10‐
12], as well as with B-lymphocyte depletion in patients with systemic lupus erythematosus (SLE) [
13]. The clinical relevance of other
FCGR genotypes is not known.
After depletion, B lymphocytes return to the peripheral blood after a mean of 8 months [
14]; this return is preceded by increases in blood levels of B-cell-activating factor (BAFF, also called
B-lymphocyte stimulator or
BLyS). BAFF, a cytokine expressed mainly by neutrophils and monocytes, plays a central role in the stimulation of B-lymphocyte proliferation, differentiation, immunoglobulin production, and survival [
15]. Elevated serum BAFF levels have been correlated to a poorer clinical outcome in patients with rheumatic diseases [
16,
17]. Thus, regulation of BAFF expression may affect the efficacy of rituximab. Enhanced generation of BAFF has been associated with the presence of a certain SNP of the
BAFF gene promoter (the −871C/T genotype) [
18‐
20]. Recently, the
BAFF −871C/T promoter polymorphism has been reported to influence the response to rituximab in patients with RA [
20,
21].
Predisposing factors for LON and the potential clinical consequences of LON are poorly defined. We therefore conducted a nested case-control analysis of a cohort of rituximab-treated patients with rheumatic diseases. We hypothesized that three established
FCGR polymorphisms—
FCGR2A 131H/R,
FCGR2B 232I/T, and
FCGR3A 158V/F—and the
BAFF −871C/T promoter polymorphism are related to occurrence of LON through modified therapeutic activity of rituximab. We thus investigated the possible predictive role of the polymorphisms for LON. Second, in this study with an extended initial control population [
3], we examined the relationship between LON and serum levels of immunoglobulin M (IgM) and also BAFF levels. Finally, we asked whether the efficacy of rituximab differs between patients with LON and those without LON.
Discussion
In this prospective case-control analysis nested within a large cohort of rituximab-treated patients with rheumatic diseases, we examined possible genetic bases for the occurrence of LON and the consequences of presentation with that complication in the largest dataset of reported LON cases reported to date. We showed, for the first time to our knowledge, the relationship between the occurrence of LON and possession of the high-affinity FCGR3A V allele. We also confirmed a previous observation of lower serum IgM levels in patients with LON than in those who did not develop LON. Further, an association between the FCGR3A V allele and lower serum IgM levels was demonstrated. Finally, we observed a favorable association between both LON and the FCGR3A V allele and a longer duration of flare-free survival following rituximab therapy. Taken together, these findings suggest that the variation in quality of clinical response to rituximab may be explained by genotypic polymorphism in a certain FcγR, and that a more pronounced B-cell depletion could likely account for both occurrence of LON and longer therapeutic activity of rituximab. Although patients with the BAFF −871T/T allotype had a longer flare-free survival than patients with the BAFF −871C/T or C/C allotype, no significance of serum BAFF levels related to LON occurrence was found.
LON represents a potentially severe clinical phenomenon occurring after rituximab treatment in patients with lymphoma and patients with rheumatic disease. At the time of LON occurrence, up to a 90% infection rate is reported in patients with rheumatic disease diagnoses, and in many patients granulocyte colony-stimulating factor (G-CSF) is required [
3,
11]. It is therefore of clinical importance to define predisposing factors for LON. To some extent, this has been investigated in patients with lymphoma. Thus, high cumulative doses of myelotoxic agents, including autologous stem cell transplant, have been proposed as a predictive factor for LON [
27‐
29].
The biological rationale to investigate
FCGR polymorphisms is based on the hypothesis that LON is related to higher binding of the Fc part of rituximab to the FcγR on macrophages and NK cells, conferring an increased efficacy of the antibody-dependent cellular cytotoxicity (ADCC) process that leads to depletion of B lymphocytes. The FcγRIIIa and FcγRIIa receptors, which mainly mediate proinflammatory activity, as well as the FcγRIIb receptor with anti-inflammatory activity, may be of significance for the ADCC process [
30].
FCGR3A V allotype is the most thoroughly investigated and is known to display higher affinity for IgG1 antibody (such as rituximab) than the
FCGR3A F allotype. Patients carrying the V/V genotype, compared with the V/F and F/F genotypes, demonstrate higher biological and clinical responses to rituximab in the treatment of B-lymphoproliferative malignancies, RA and SLE [
9,
13,
31].
In addition to SNPs in the
FCGR gene, the role of copy number variation has been proposed to be related to the efficacy of antibody-based drugs [
32] and differential susceptibility to autoimmune diseases [
33].
FCGR gene copy number variation was not assessed in our sample. Investigation of whether the
FCGR gene copy number genotypic profile is associated with occurrence of LON requires larger study populations.
To the best of our knowledge, this report is the first regarding the relationship between LON occurrence and possession of the high-affinity
FCGR3A V allele in patients with rheumatic disease. This finding is in line with a report on patients with lymphoma in which the
FCGR3A 158V/F polymorphism was suggested as a novel risk factor for LON [
34,
35]. We also extended our analysis to several FcγR types. In this context, the role of possession of the anti-inflammatory FcγRIIb was of particular interest, not only because of its largely opposing effects compared with FcγRIIIa in mediating actions of antibodies but also because it is expressed mainly on B lymphocytes [
30,
36]. The significance of polymorphisms in the
FCGR2A and
FCGR2B genes for LON occurrence was not documented here, however. We acknowledge that our analysis had limited statistical power to estimate the contribution of the
FCGR2B 232I/I polymorphism; taking into account its prevalence in the LON and non-LON groups, a sample size of 96 patients in each group would have been required to detect a difference with 80% statistical power (based on the normal approximation to the binomial distribution). Considering the low numbers of patients in our study, the negative findings do not exclude a role of other polymorphisms and FcγRs for developing LON. Of note, we observed differences in the distribution of
FCGR polymorphisms in the LON and non-LON groups, with more frequent
FCGR2A 131H/H and
FCGR2B 232I/I polymorphisms among patients who developed LON than among control subjects.
Previously, we reported that occurrence of LON is associated with a more pronounced and sustained B-lymphocyte depletion and lower levels of serum IgM following rituximab than in matched control subjects with a similar history of previous antirheumatic therapy [
3]. In the present study with an extended control group of patients who did not develop LON following rituximab therapy, the presence of lower serum IgM levels in the LON group than in non-LON control subjects was confirmed. In the present study, we further found that the homozygosis for the high-affinity V allele of
FCGR3A was correlated to lower IgM levels over time compared with the low-affinity F allele.
These findings suggest that the high-affinity V allele of FCGR3A, leading to an enhanced binding of rituximab to FcγRs, might confer deeper and more prolonged B-lymphocyte depletion that may result in both LON and more pronounced and prolonged decrease of serum IgM. Whether monitoring of serum IgM levels in patients with a certain genotypic polymorphism may be useful for prediction of LON development is to be addressed in larger prospective studies.
In this study, we found no significant association between LON and serum BAFF levels or
BAFF promoter polymorphisms (which may influence serum BAFF levels), but the number of available blood samples was limited, and only a few samples were drawn during the LON period. Patients with the
BAFF −871T/T allotype, however, had a better clinical response with a longer flare-free survival than patients with the
BAFF −871C/T or C/C allotype. Though caution is needed in interpretation of the findings of our analysis in patients with different rheumatic diseases, our findings are in line with a reported improved response to rituximab in patients with seropositive RA carrying the
BAFF −871T/T allotype [
20]. To avoid findings by chance, we did not proceed with a subanalysis in patients possessing autoantibodies. Further study of BAFF regulation in the context of severe acute neutropenia is of particular interest because BAFF is produced largely by neutrophils and monocytes and because BAFF expression is augmented by G-CSF [
37], which is often used for patients who present with infectious complications due to LON.
The present analysis demonstrated that the occurrence of LON and the
FCGR3A 158V/V genotype both were related to a longer time to flare of the rheumatic diseases. Transient neutropenia is recognized in idiosyncratic drug-induced neutropenia and agranulocytosis [
38]. Our findings of lower serum IgM levels in patients who developed LON and lower serum IgM levels in association with the
FCGR3A V allele, together with a longer time to flare, indicate that the phenomenon of LON is of a different nature than drug-induced agranulocytosis and that LON is likely explained by biological mechanisms of rituximab action modified through
FCGR genotypes (i.e., a more pronounced B-cell depletion).
Although we report a potentially important finding, LON is a rare clinical phenotype, and there were only 11 cases of LON associated with rituximab in our study. Our results should be verified in larger cohorts performed in disease-specific patient populations with a larger number of cases presenting with LON. Statistical power to evaluate the effect of genetic variation on a rare event such as LON could be improved by performing meta-analyses of combined data from several cohorts. The small number of cases in the present study probably limited the chance of detection of effects of some genotypic polymorphisms (type II error), but the main results appear to be in line with those in previous reports. The strength of the study lies in its advantage of a large cohort of 214 consecutive patients treated with rituximab who were followed prospectively with a variety of important clinical and laboratory data. The possible confounding effects of exposure to other significant drugs was minimized (as could not formally be overcome in the observational study) by individual matching of the LON cases and control cases by the cumulative number of previously used DMARDs; use of cytotoxic and biologic therapies; and concurrent treatment with DMARDs, cyclophosphamide, and corticosteroids. The patients represented a typical, clinic-based, “real-life” patient population treated with rituximab; thus, the selection bias was minimal because our center provides care for patients with the whole spectrum of rheumatic diseases.
Acknowledgements
We thank research nurse Margareta Wörnert for her assistance with blood sampling. We thank all the participants and staff of the rheumatology department at Karolinska University Hospital Huddinge for their contributions.