Late-onset neutropenia following rituximab results from a hematopoietic lineage competition due to an excessive BAFF-induced B-cell recovery

Rituximab is used in the treatment of lymphoma and autoimmunediseases, for which late-onset neutropenia (LON) were reported.LON-related mechanisms remain unclear. To obtain insights intothe mechanisms, we assessed serum, peripheral blood and bonemarrow (BM) samples of a patient with LON. Factors classicallyassociated with neutropenia such as anti-neutrophil antibodies,T-LGL, soluble Fas Ligand were not detectable. We then evaluatedthe kinetics of various cytokines involved in B-cell and granulocytehomeostasis. We found that LON is related to a lack of granulopoiesisin the BM that coincides with a very high level of BAFF, a strongstimulator of B-cell recovery, and hypothesized a hematopoieticlineage competition due to an excessive B-cell recovery in theBM by promotion of B-cell lymphopoiesis over granulopoiesiswithin common developmental niches. Assessment of serum BAFFlevels following rituximab could detect patients at risk ofdeveloping LON.

Key words: Neutropenia, rituximab, BAFF, niches, granulopoiesis, lymphopoiesis, lineage competition.

Introduction

TOP

Introduction

Rituximab, an anti-CD20 monoclonal antibody, is increasinglyused in the treatment of lymphoma and autoimmune diseases. CD20is expressed on malignant B-cells, normal differentiated B-cellsand pre-B-cells, but not on stem cells and granulocyte precursors.Late-onset neutropenia following rituximab (LON) have been reported,^1–7occurring 1 to 6 months after last rituximab infusion, as beingsevere (<0.5x10⁹/L), spontaneously reversible and withoutlife-threatening infections.

Several studies suggest that LON could be related to an excessof T-Large Granular Lymphocyte (LGL) in the bone marrow (BM)and peripheral blood (PB) which express and secrete large amountsof Fas and Fas Ligand (FasL) leading to apoptosis of matureneutrophils,⁴ or to a production of autoantibodies binding tothe neutrophil surface during recovery of a new immune repertoire.^2,6On the other hand, a recent study suggests that LON is not relatedto circulating factors but to perturbations of Stromal-derivedFactor 1 (SDF-1) and granulopoiesis homeostasis during B-cellrecovery.⁷ This is reinforced by previous studies showing thehypocellularity of the marrow at the time of LON^1–3,6and the absence of anti-neutrophil antibodies in the serum orT-LGL in the PB.^,2,6,7

We report here that LON is related to a lack of granulopoiesisin the BM that coincides with the time of maximum B-cell depletionin PB, and proposed the hypothesis that LON is due to a hematopoieticlineage competition in the BM by promotion of B-cell lymphopoiesisover granulopoiesis within common developmental niches.

Study design

TOP

Study design

Case report
In August 2005, a 55-year-old woman was referred for severeneutropenia of fortuitous discovery. Her past medical historywas remarkable for Waldenström macroglobulinemia (WM) associatedwith retroperitoneal and renal infiltration diagnosed in March2005. The patient was treated with rituximab (375 mg/m² day1), fludarabine (30 mg/m² day 1 to 3) and cyclophosphamide (300mg/m² day 1 to 3) from March to June 2005 (last treatment beingdefined as day 0), for 4 courses monthly, leading to completeremission of WM. After the end of treatment, blood tests performedon day 37 were unremarkable except for a mild decrease in neutrophilcount (1.6x10⁹/L). At admission on day 84, blood tests revealeda severe neutropenia (0.2x10⁹/L), with normal hemoglobin leveland platelet count, but the exact time from last treatment tothe development of neutropenia could not be precisely determinedwith the available data. The serum monoclonal component wasstill not detectable. There was no recent or concomitant medication.Bone marrow aspiration showed granulocytic hypoplasia (22%)without maturation blockade, and no excess of T-LGL or hemophagocytosis.In the PB, smear analysis was normal without T-LGL, lymphocyteimmunophenotyping showed the absence of B cells (CD19/CD20:0%) and T-LGL, and TCR rearrangement analysis did not revealany circulating T-cell clone. Viral PCR in PB and BM were negativefor parvovirus B19, CMV, EBV, and HHV6. Circulating anti-neutrophilantibodies were not detected by immunofluorescence. Thus, weput forward the diagnosis of rituximab-induced neutropenia sinceregular causes of neutropenia were excluded. Neutropenia wasspontaneously reversible within 10 days (5400/mm³), withoutthe use of G-CSF. A second episode of neutropenia (340/mm³)occurred few days later, spontaneously reversible within 6 days.After 15 months, the patient did not relapse of either neutropeniaor WM. Peripheral blood stem cells were collected after stimulationby G-CSF 6 months after the recovery from neutropenia, suggestingthat the patient’s bone marrow has returned to normal.

Serum samples and cytokines levels
Patient serum samples were either available or have been collectedspecifically, at -198 (before rituximab), 37, 84 (first episodeof LON), 138 and 222 (after resolution of LON)-day time points;day 0 being defined as the last rituximab infusion. Cytokineslevels at the five selected time points have been assessed byEnzyme Linked ImmunoSorbent Assay (ELISA) for tumor necrosisfactor (TNF)-alpha, interleukin (IL)-6 (BioSource Europe, Belgium),Thymic Stromal Lymphopoietin (TSLP), SDF-1 and B-cell activatingfactor (BAFF) (R&D systems, MN, USA); by immunoblot forsoluble FasL.

Bone marrow samples, CFU-GM culture and analysis of the serum inhibitory effect
Patient’s BM sample was collected during the LON episode,at 84-day time point, but not after the neutrophils or B-cellsrecovery. On this sample, the Colony-forming units granulocyte-macrophage(CFU-GM) growth have been assessed by the quantification ofthe number of CFU-GM after 10 days of culture. The inhibitoryeffect of the patient’s sera collected at the 5 time-pointson CFU-GM growth have been assessed by incubation of CFU-GMfrom an healthy subject BM with sera and quantification of thenumber of CFU-GM after 10 days of culture.

Results and discussion

TOP

Results and discussion

To obtain insights into the LON-related mechanisms, we assessedserum, PB and BM samples of a patient with LON following rituximab.

BM assessment revealed granulocytic hypoplasia and the absenceof T-LGL on BM smear. No anti-neutrophil antibodies in the serumor T-LGL in the PB were detected and at the selected time points,serum assessment revealed the absence of soluble FasL (Figure 1A).The patient’s sera did not inhibit the growth of healthydonor CFU-GM, and showed a possible stimulation at the 84-daytime point (Figure 1B). Finally, the patient’s CFU-GMculture from BM at day 84 showed a disturbance of granulocytedifferentiation, with a decrease number of CFU-GM after 10 daysof culture (12 vs. 60 to 300 in healthy subjects) (Figure 1C).Our findings suggest: first, a lack of early granulocyte precursorsat this time point, since this patient did not have any argumentfor an intrinsic defect of granulocyte differentiation untilthe episode of LON; second, the commitment of progenitors tothe lymphoid lineage since CFU-GM growth quantifies the progenitorsalready committed to the myeloid lineage.

View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
Figure 1. LON is not related to circulating factors but rather to a lack of granulopoiesis in BM. (A) Detection of soluble FasL in the patient’s sera by immunoblot. (B) Analysis of the inhibitory effect of the patient’s sera at the 5 time-points on the growth of healthy donor CFU-GM. (C) Assessment of patient CFU-GM growth by quantification of the number of CFU-GM after 10 days of culture (normal values are within the dotted lines) at time of LON. Day 0 is defined as the end of treatment (last rituximab infusion).

We then assessed variations in the serum levels of cytokinesinvolved in B-cell and granulocyte homeostasis: SDF-1 involvedin human B-cell homeostasis and granulopoiesis,⁸ TSLP involvedin murine B-cell homeostasis and granulopoiesis,⁹ TNF-alphawhich inhibits granulopoiesis¹⁰ and IL-6 which exhibits stimulatoryeffect on granulocyte production.^11,12 No fluctuations of TNF-alpha,TSLP and SDF-1 in serum were detected (Figure 2A, 2B, 2C, 2D).Regarding SDF-1, perturbations of this cytokine were recentlyunderlined in the pathophysiology of LON, because of its centralrole in both the regulation of neutrophil egress from the BMand early B-cell lymphopoiesis, raising the notion that B-cellrecovery following rituximab involves early stages of B-celllymphopoiesis.⁷ We did not find significant fluctuations incirculating levels of SDF-1, but SDF-1 is not directly accessibleto quantitative analysis. IL-6 paralleled the evolution of thenumber of CFU-GM after incubation with patient’s sera,with an increase following rituximab, a maximum at the timeof LON, and a decrease with the neutrophils recovery. IL-6 variationwas not linked to an inflammatory process since C-reactive proteinlevel and temperature curve were normal at each time point.The evolution of IL-6 serum level could be explained by itssynergistic action with G-CSF on myeloid progenitor cells, asa positive feedback to normalize granulopoiesis.

View larger version (14K):
[in this window]
[in a new window]
[Download PPT slide]
Figure 2. Assessment of the level of (A) TNF-alpha, (B) IL-6, (C) TSLP and (D) SDF-1 in the patient’s sera at the selected time points (the upper limits for TNF-alpha and IL-6 are represented by the dotted lines). Day 0 is defined as the end of treatment (last rituximab infusion)

LON occurred during the time of peripheral B-cell depletion(Figure 3A, 3B, 3C). Interestingly, the neutropenia was alsoconcomitant to the maximum of BAFF serum level, a cytokine involvedin human B-cell survival, expansion and development.^13,14 BAFF,almost undetectable prior to therapy, increased following rituximabuntil a maximum coinciding with the episode of LON, and thendecreased with the neutrophils and B-cells recovery in the PB(Figure 3D). The increase in BAFF levels following rituximabhas been already reported and is in part, linked to the physicalloss of BAFF-binding B-cells,^16,17 but also the result of atranscriptional up-regulation of BAFF mRNA after B-cell depletionby a yet unknown mechanism. Thereby, BAFF serum level is increasedduring the time of peripheral B-cell depletion and the declinein BAFF serum levels is associated with the reemergence of Bcells in the PB. Our results are consistent with the fact thatLON could occur just before B-cell recovery as suggested byDunleavy et al.,⁷ and are concordant with the usual timing ofoccurrence of LON after rituximab therapy. In our patient, BAFFserum level after rituximab was much higher (22 ng/ml) thanin patients treated with rituximab for autoimmune systemic disordersbut who did not develop LON (n=5, mean 5.0±2.3 ng/mLwith the same assay).¹⁷ Such high BAFF serum level is very unusualand represents a very strong activating factor of B-cell recovery.Thus, we hypothesize that post-rituximab LON could be the consequenceof a promotion of B-cell lymphopoiesis over granulopoiesis,determined by B-cell depletion-induced variations of growthand retention factors, including SDF-1 and BAFF, to supportthe reemergence of peripheral B-cells. Ueda et al. demonstratedthat the reciprocal dynamics of the BM lymphocyte and neutrophilpopulations were consistent with a cellular competition withinthe developmental niche defined by a common location and sharedgrowth resources, and that there was a balance between granulopoiesisand lymphopoiesis in the BM in normal condition, balance whichcould be disturbed following rituximab.¹⁸ BAFF does not seemto be directly involved in LON-related mechanism but could representa surrogate biomarker of the risk of developing LON followingrituximab. The high increase of BAFF following rituximab couldbe linked to the profundity of tissular B-cell depletion orto other unknown genetic factors. Another hypothesis could bethat a cytokine, which was not measured in our study, exhibitan inhibitory effect on granulopoiesis. However, this hypothesisdoes not seem relevant since our patient’s sera did notinhibit the growth of healthy donor CFU-GM. Studies on a largercohort of patients are needed to confirm our hypothesis.

View larger version (9K):
[in this window]
[in a new window]
[Download PPT slide]
Figure 3. Kinetics of (A) LON, (B) serum immunoglobulins level and (C) B-cell count in PB. (D) Assessment of BAFF level in the patient’s sera at the selected time points (the mean value of BAFF serum level after rituximab of 5 patients treated with rituximab but who did not develop LON is represented by the dotted line). Day 0 is defined as the end of treatment (last rituximab infusion).

In conclusion, LON following rituximab would occur in patientswith an excessive B-cell depletion and recovery, in relationto a hematopoietic lineage competition by promotion of B-celllymphopoiesis over granulopoiesis within common developmentalniches in the BM. Assessment of serum BAFF levels followingrituximab could detect patients at risk of developing LON.

References

TOP