Should complete B cell depletion be maintained in patients treated long-term with rituximab for rheumatoid arthritis?

Abstract

Objective

To determine whether persistent complete B cell (BC) depletion was associated with a better clinical response in rheumatoid arthritis (RA) patients long-term treated with rituximab (RTX).

Methods

We conducted a retrospective study including RA patients admitted for a new infusion between 2019 and 2021. The primary endpoint was the comparison of the mean disease activity score based on 28 joints (DAS28)-CRP at each of the four last infusion visits between patients with persistent complete BC depletion (mean CD19 counts <18/µl at each of the last four visits) or without persistent complete BC depletion (mean CD19 counts of the last four visits ≥18/µl). Secondary endpoints included DAS28, pain/fatigue visual analogue scale, CRP, gammaglobulins and the frequency of self-reported RA flares.

Results

Of the 126 patients in maintenance therapy with RTX [exposure period: 76 (5) months, 14 (7) infusions received], 43 (34%) had persistent complete BC depletion at each of the four last infusions. The mean DAS28-CRP calculated at each of the four last infusion visits did not significantly differ according to persistence or not of complete BC depletion. This result remained unchanged after adjusting for antibody status, number of previous therapies, number of RTX infusion and cumulative RTX dose. All secondary outcomes were also not significantly different between the two groups.

Conclusion

Maintaining complete BC depletion does not appear to be a therapeutic target to achieve in RA patients in long-term maintenance therapy with RTX. There is a limited benefit of monitoring CD19 in RA patients long-term treated with RTX and having achieved low disease activity/remission.

rituximab, arthritis, rheumatoid, lymphocyte depletion, biologic therapy

Rheumatology key messages

• Persistence of complete B cell depletion is not associated with better long-term efficacy of RTX in RA.

• Maintaining persistent complete B cell seems not to be a target to achieve in RTX long-term treated patients.

• Systematic monitoring of CD19 may be of less interest in long-term RTX treated RA patients.

Introduction

Rituximab (RTX) in combination with methotrexate is recommended for the treatment of rheumatoid arthritis (RA) with an inadequate response or contraindication to other targeted biologic therapies [1, 2]. The depletion of peripheral B cells (BC) following RTX is usually rapid, deep and persists for at least 6–12 months [3]. Complete peripheral BC depletion has been considered as a relevant indicator of short-term treatment response. A first study indicated that a lack of complete depletion of BC after a single RTX infusion was associated with a poorer outcome at 9 and 12 months [4]. A second study has shown that the response rate at 6 months of RA to a first cycle of two RTX infusions was determined by the level of peripheral BC depletion rather than the dose of the drug [5]. In addition, BC repopulation has been shown to precede clinical RA relapse, suggesting that BC counts could serve as a tool to predict disease flares [6]. Similar results were also observed in ANCA-associated vasculitis renal injury, in which achievement of complete remission was associated with complete peripheral BC depletion and relapse was preceded by BC repopulation [7].

Since all of these data were gathered during the first months of RTX application, one important question is whether long-term sustained complete BC depletion is required to maintain a good clinical response to RTX. Moreover, this should be balanced with the potential risk for the development of hypogammaglobulinaemia and related infectious events [8]. Thus, the objective of our study was to determine whether the persistence of complete BC depletion was associated with a better clinical response in RA patients long-term treated with RTX.

Methods

Study design

We conducted a retrospective routine care study in the Rheumatology department of Cochin hospital in Paris, France.

Study population

We included adult patients (age >18 years) fulfilling the 2010 ACR/EULAR classification criteria for RA [9] who were hospitalized between January 2019 and December 2021 for a new RTX infusion (most recent visit). All patients had to be on maintenance long-term therapy with RTX, meaning that they had to receive at least three prior RTX infusions.

The protocol and the informed consent document received Institutional Review Board/Independent Ethics Committee (IRB/IEC) approval before initiation of the study (‘Comité de Protection des Personnes’ Paris Ile de France I). The study was declared to the Commission Nationale de l’Informatique et des Libertés (reference 2221678). All patients in our institution (AP-HP) are informed that their clinical data can be used for research and give their consent for the use of their data unless they provide opposition to it. All patients agreed to participate in this study after written informed consent, which was recorded in the medical source file.

Setting

All patients initiated RTX with a course of two 500 mg or 1000 mg intravenous infusions separated by 2 weeks. Maintenance RTX therapy following this initiation consisted of the administration of a single infusion of RTX performed at least every 6 months at a dose of 500 mg or 1000 mg. All patients included in this study were on maintenance therapy with RTX. RTX management decisions (modification of dose or interval between maintenance infusions) were at the discretion of the treating physicians. Patients were routinely evaluated in our centre at each infusion visit for treatment efficacy and tolerance.

Clinical and laboratory data

Data related to the last four infusion visits (last visit between 2019 and 2021 and the three previous infusion visits) were collected from the electronic medical report that included: demographic parameters the day of the new RTX infusion visit (age, gender, underlying disease, disease duration, BMI), current and past medication use (including concomitant treatment with methotrexate), duration of treatment with RTX, number of infusions and cumulative dose.

Joint disease activity was assessed using the disease activity score based on evaluation of 28 joints (DAS28 and DAS28-CRP using C-reactive protein) [10]. We collected the values of pain and fatigue visual analog scale (VAS), and the presence of patient self-reported RA flares between two RTX courses (defined as worsening of RA accompanied by at least one swollen and tender joint, as perceived by the patient) [10]. We also collected reports of end-of-dose effect, defined by the resurgence of RA symptoms perceived by the patient near the next rituximab infusion.

Laboratory tests were performed at each infusion visit and included: complete blood cell counts, erythrocyte sedimentation rate (ESR), CRP, gammaglobulin levels, and T, B and NK cell immunophenotyping (Aquios, Beckman Coulter, Brea, CA, USA). The limit of detection of B cells was defined by CD19 <18/µl.

Exposure of interest

Persistent complete BC depletion was defined as CD19 value <18/µl at each of last four infusion visits. Thus, the mean CD19 count of the last four visits was <18/µl. Since all patients maintained CD19 counts <18/µl at all four time points, the grouping was constant and no group changes occurred from visit to visit.

Patients with no persistent complete BC depletion had a mean CD19 value of the last four visits ≥18/µl. This subgroup included patients with incomplete B cell depletion (mean CD19 of last four visits between 18 and 65/µl) and no B cell reduction (mean CD19 of last four visits ≥65/µl).

Endpoints

Primary endpoint

Our hypothesis was that a better treatment response in patients with persistence of complete BC depletion would be observed. To that end, the primary endpoint was the difference of treatment response at each of the last four infusion visits, assessed by the DAS28-CRP, between patients with or without persistent complete B cell depletion.

Secondary endpoints

We considered as secondary endpoints the DAS28, the frequency of end-of-dose effect and patient self-reported RA flares, pain/fatigue VAS, CRP and gammaglobulin levels between patients with or without persistent complete B cell depletion at each of the four infusion visits.

We also performed a second analysis of the DAS28-CRP considering the three following groups (Supplementary Table S1, available at Rheumatology online): persistent complete depletion, incomplete B cell depletion (mean CD19 of last four visits between 18 and 65/µl) and no B cell reduction (mean CD19 of last four visits ≥65/µl).

Statistical analysis

All data are presented as mean values (s.d.) or number and percentage for continuous and categorical variables, respectively. Statistical analysis was performed using GraphPad Prism (v9.1.2; GraphPad Software, Boston, MA, USA) and XLSTAT (Lumivero, Denver, Colorado) 2023.1.6.1410. We used Student’s unpaired t-test for two-group comparisons (continuous variables) and the χ² test for differences in frequency (binary variables). Two- and three-group continuous variable multiple comparisons at each time points were performed by an ANOVA followed by Šidák’s and Tukey’s multiple comparison tests, respectively. The analysis of the primary endpoint DAS28-CRP patients with or without persistent BC depletion was performed before and after adjustment for potential confounders following linear regression (antibody status, number of previous therapies, number of RTX infusion and cumulative RTX dose). An analysis of the power of our sample was performed to determine its ability to show a difference of DAS28-CRP between patients with or without complete B cell depletion. A P-value <0.05 was considered statistically significant.

Results

Patient and disease characteristics

We included 126 patients meeting our inclusion criteria: 43 with persistent complete and 83 with absence of persistent complete BC depletion. This sample had sufficient power for detecting a moderate difference of DAS28-CRP between patients with or without complete BC depletion. The power of our sample was 0.89 for an effect size of 0.6 and 0.75 for an effect size of 0.5. Among the 126 included patients, 105 (83%) were women with a mean age of 64 (12) years and mean disease duration of 22 (5) years the day of the last RTX infusion. Positive rheumatoid factor (RF) and positive anti-citrullinated protein antibody (ACPA) antibodies were detected in 110 (87%) and 101 (80%) patients, respectively; and 98 patients (78%) had erosive disease. RTX was predominantly used after the failure of at least one targeted therapy (n = 88, 70%).

Included patients have received a mean total number of RTX infusions of 14 (7) over a mean period of 76 (2) months, corresponding to a mean cumulative dose of 9 (6) g. The mean retreatment interval over the last four RTX infusions was 7.5 (2) months. Detailed characteristics at inclusion (collected the day of the last of the four infusion visits) are presented in Table 1.

Table 1.

Patient characteristics at inclusion (day of the last of the four infusions) according to persistence or not of complete B cell depletion

	Total (n = 126)	Persistent complete BC depletion (n = 43)	No persistent complete BC depletion (n = 83)	P-value
Age, mean (s.d.), years	64 (12)	63 (14)	64 (11)	0.87
Females, n (%)	105 (83)	36 (84)	69 (83)	0.89
Disease duration, mean (s.d.), years	22 (9)	21 (9)	22 (11)	0.67
Rheumatoid factor, n (%)	110 (87)	34 (77)	76 (92)	0.018
ACPA, n (%)	101 (80)	31 (72)	70 (84)	0.11
Radiographic damage, n (%)	98 (78)	34 (79)	64 (77)	0.80
DAS28, mean (s.d.)	2.9 (1.1)	3.1 (1.1)	2.8 (1.1)	0.42
DAS28-CRP, mean (s.d.)	2.6 (0.9)	2.8 (0.8)	2.5 (0.9)	0.10
Concomitant treatment, n (%)
Methotrexate	79 (63)	25 (58)	54 (65)	0.44
Corticosteroids	56 (44)	25 (58)	31 (37)	0.20
Previous lines of targeted therapies prior to RTX, n (%)
0	38 (30)	9 (21)	29 (35)	0.11
1	33 (26)	8 (19)	25 (30)	0.19
2	37 (29)	16 (37)	21 (25)	0.16
>2	18 (14)	10 (23)	8 (10)	0.049
Duration of RTX exposurea, mean (s.d.), months	76 (52)	69 (47)	99 (57)	0.003
Number of infusionsa, mean (s.d.)	14 (7)	12 (6)	14 (7)	0.037
Cumulative RTX dosea, mean (s.d.), g	9 (6)	8 (5)	10 (6)	0.010
Retreatment intervalb, mean (s.d.), months	7.5 (2)	6 (1)	8 (3)	0.001

a

Since RTX onset.

b

Over the last four RTX infusions. BC: B cells; RTX: rituximab.

Patients’ characteristics according to BC depletion

Forty-three patients (34%) presented persistent complete BC depletion at each of the last four infusions (Supplementary Fig. S1A and B, available at Rheumatology online). Patients with no persistent complete BC depletion (n = 83) did not differ from those who maintained complete BC depletion in terms of age [64 (11) years vs 63 (14) years, P = 0.87], gender (83% vs 84% women, P = 0.89), disease duration [21 (9) years vs 22 (11) years, P = 0.67], radiographic damage (77% vs 79% bone erosions, P = 0.80) and concomitant treatment by methotrexate or corticosteroids (65% vs 58% on MTX, P = 0.44; 37% vs 58% on corticosteroids, P = 0.20) (Table 1).

Patients with no persistent complete BC depletion had a higher frequency of RF (92% vs 77%, P = 0.018) and ACPA (84% vs 72%, P = 0.11); these patients had received less previous targeted therapies (54/83, 65% vs 17/43, 39.5% with ≤1 targeted therapy, P = 0.006), they have been treated with RTX for a longer period [99 (57) months vs 69 (47) months, P = 0.003], with a significantly higher number of infusions [14 (7) vs 12 (6) infusions, P = 0.037] and increased cumulative dose [10 (6) vs 8 (5) g, P = 0.10] compared with patients with persistent complete BC depletion. On the other hand, their interval between two infusions was significantly longer [8 (3) months vs 6 (1) months, P = 0.001].

Endpoints

Primary endpoint

The course of the DAS28-CRP over the last four infusion visits was comparable between the two groups (Fig. 1A). The mean DAS28-CRP calculated at each of the last four infusion visits did not significantly differ between patients with no persistent complete BC depletion and those who maintained complete BC depletion (Table 2 and Fig. 1B). This result was not changed after adjusting the DAS28-CRP for antibody status, number of previous therapies, number of RTX infusion and cumulative RTX dose (Table 2). The analysis of the DAS28-CRP considering the three groups of persistent complete depletion, incomplete B cell depletion and no B cell reduction (Supplementary Table S1 and Fig. S2A, available at Rheumatology online) did not reveal any significant difference between groups at each of the last four infusion visits (Supplementary Fig. S2B, available at Rheumatology online).

Figure 1.

Evaluation of the DAS28-CRP as the primary outcome. (A) Changes of mean (s.d.) of the DAS28-CRP according to the persistence or not of complete B cell depletion. (B) Comparison of the DAS28-CRP at each of the last four infusion visits between patients with or without persistent complete B cell depletion. NS: not significant by ANOVA followed by Šidák’s multiple comparison test. DAS28: disease activity score based on 28 joints

Table 2.

Comparison of the DAS28-CRP at each of the last four infusion visits according to the persistence of complete B cell depletion

	Infusion n − 3	Infusion n − 2	Infusion n − 1	Infusion n
DAS28-CRP, mean (s.d.)
No persistent complete B cell depletion	2.56 (1.09)	2.55 (1.23)	2.59 (1.08)	2.40 (0.97)
Persistent complete B cell depletion	2.71 (0.95)	2.99 (1.31)	2.76 (0.95)	2.72 (0.81)
Mean difference (s.d.)	0.15 (0.20)	0.43 (0.20)	0.16 (0.20)	0.31 (0.20)
P-value	0.91	0.12	0.88	0.39
Adjusted DAS28-CRPa, mean (s.d.)
No persistent complete B cell depletion	2.58 (0.24)	2.66 (0.36)	2.62 (0.26)	2.48 (0.20)
Persistent complete B cell depletion	2.67 (0.20)	2.79 (0.30)	2.71 (0.22)	2.53 (0.16)
Mean difference (s.d.)	0.09 (0.04)	0.13 (0.06)	0.09 (0.04)	0.05 (0.03)
P-value	0.12	0.14	0.13	0.53

a

Adjusted for antibody status, number of previous therapies, number of RTX infusion and cumulative RTX dose. DAS28: disease activity score based on 28 joints

Secondary endpoints

No difference was observed regarding the course of the DAS28 over the last four infusion visits, and the mean DAS28 calculated at each of the last four infusion visits was not statistically significant between patients with or without persistent complete BC depletion (Fig. 2A and B). The frequency of an end-of-dose effect and self-reported flares assessed the day of the last four infusion visits did not significantly differ between patients with or without persistent complete BC depletion (Table 3). The course of pain VAS, fatigue VAS, CRP and gammaglobulin levels was also similar between the two groups (Fig. 3A–D), and the mean value of each parameter, calculated at each of the last four infusion visits, was not significantly different between patients with or without persistent complete B cell depletion.

Figure 2.

Evaluation of the DAS28. (A) Changes of mean (s.d.) of the DAS28 according to the persistence or not of complete B cell depletion. (B) Comparison of the DAS28 at each of the last four infusion visits between patients with or without persistent complete B cell depletion. NS: not significant by ANOVA followed by Šidák’s multiple comparison test. DAS28: disease activity score based on 28 joints

Figure 3.

Evaluation secondary outcomes. (A–D) Changes of mean (s.d.) of pain (A) and fatigue VAS (B), CRP (C) and gammaglobulin levels (D) according to the persistence or not of complete B cell depletion. (E–H) Comparison of pain VAS (E), fatigue VAS (F), CRP (G) and gammaglobulin levels (H) at each of the last four infusion visits between patients with or without persistent complete B cell depletion. PCBCD: persistent complete B cell depletion; NPCBCD: no persistent complete B cell depletion. NS: not significant by ANOVA followed by Šidák’s multiple comparison test. VAS: visual analogue scale

Table 3.

Frequency of end-of-dose effect and self-reported RA flares

	Infusion n − 3	Infusion n − 2	Infusion n − 1	Infusion n
End-of-dose effect, n (%)
No persistent complete B cell depletion	23/77 (30)	29/81 (36)	25/78 (32)	24/81 (30)
Persistent complete B cell depletion	16/40 (40)	19/41 (46)	15/41 (37)	16/40 (40)
P-value	0.28	0.054	0.58	0.097
Self-reported flares, n (%)
No persistent complete B cell depletion	10/79 (13)	13/82 (16)	15/82 (18)	12/82 (15)
Persistent complete B Cell depletion	4/42 (9.5)	10/43 (23)	3/43 (7)	4/42 (9.5)
P-value	0.57	0.34	0.095	0.39

The frequency of an end-of-dose effect and the occurrence of self-reported RA flares was assessed at each of the last four RTX infusion visits according to persistence or not of complete B cell depletion.

Discussion

While associated with a better short-term response rate [4, 5, 7], complete BC depletion was not associated with a better clinical response in our population of RA patients long-term treated with RTX.

We identified two distinct populations according to their level of BC depletion. The first displayed persistent complete BC depletion over four consecutive sets of infusions. This group was characterized by 6-month intervals between each set of infusions. This subset had lower RF and ACPA positivity, shorter RTX exposure, and higher number of previous targeted therapies. Interestingly, low disease activity or remission was maintained in these patients over the follow-up period of four infusions, which could lead to the discussion of potential infusion spacing.

The second population displayed incomplete BC depletion or no BC reduction probably related to a longer spacing between infusions (8 months). These patients were more likely to be RF and/or ACPA positive, have received fewer previous targeted therapies in the past and have been exposed to RTX for a longer period. This population may correspond to a less refractory and better responsive population to RTX than patients with persistent complete BC depletion, explaining the effective spacing of the infusions. Indeed, ACPA and RF positivity have been identified as predictive markers of response to RTX [11]. Higher autoantibody positivity in this subgroup may also be related the persistence of antibody secreting cells favoured by the persistence of circulating and tissular B cells. Higher frequency of disease flares and end-of-dose effect may have been expected in this subgroup related to the spacing of the infusions but was not observed. Thus, considering the good disease control of patients with incomplete BC depletion or no BC reduction during the last four infusions, the systematic monitoring of CD19 may be of less interest since it may not influence treatment management.

The gradual spacing of RTX infusions may increase humoral response in patients secondary to the progressive BC cell repopulation. This may improve the response to vaccination, which may be particularly important during the COVID pandemic because RTX is associated with a reduced humoral response to the anti-SARS-CoV-2 mRNA vaccine [12–15].

Although RTX does not directly deplete immunoglobulin secreting CD20 negative plasma cells, it may reduce their repopulation by depleting CD20 pre-plasma cells leading to decreased production of immunoglobulins in some cases. Hypogammaglobulinaemia associated with RTX treatment depends on many factors, including immunoglobulin levels at baseline and concomitant glucocorticoid treatment, and is not directly associated with peripheral cell levels [8, 16–19]. In our study, immunoglobulin levels were similar between the two groups, which is consistent with the observations in previous studies.

Another important point to consider is the use of lower doses of maintenance RTX in RA patients who achieve a good disease control [20–22]. Our patients were all initiated on 2 × 500 mg or 2 × 1000 mg of RTX with retreatment with 500 mg or 1000 mg when a good clinical response was achieved. None of our patients were treated with an ultra-low dose of 200 mg. While the non-inferiority of the ultra-low dose (200 mg or 500 mg) could only be demonstrated in the intention-to-treat analysis of the REDO trial [23–24], a pre-planned secondary analysis of the REDO trial showed no statistically significant differences in BC counts nor in maintained clinical response between ultra-low dose and low-dose (1000 mg) groups at different time points at baseline [25]. This could raise the possibility of using lower doses of RTX in maintenance of RA patients.

The strength of our study is the long-term clinical and biological follow-up of RTX-treated RA patients in real life routine practice. Our population was composed of patients under treatment maintenance with RTX with a good disease control.

Limitations to our study were its retrospective nature and the exclusion of early non-responder patients to RTX (patients who received <4 RTX infusions and had DAS28-CRP showing persistent moderate to severe disease activity). These patients had RTX discontinuation and were not included in our study. Thus, our results may not be extrapolated to all RTX-treated RA patients. The design of the study did not allow the identifications of predictive factors leading to transient or persistent BC depletion. Moreover, it is important to mention that in our department, patients were not significantly delayed between their RTX infusions during the COVID pandemic, and this situation may have been different in other countries.

Conclusion

While it is associated with a better response to RTX at the start of treatment, the persistence of complete BC depletion is not associated with better efficacy of RTX in RA patients receiving this treatment over the long term. There is a limited benefit of monitoring CD19 in RA patients long-term treated with RTX and having achieved low disease activity/remission. Incomplete BC depletion and no BC reduction were frequent in our patients and were probably explained by the lengthening of the intervals between infusions linked to good control of the disease. Maintaining complete BC depletion does not seem to be a therapeutic goal to achieve in RA patients receiving long-term RTX. During the COVID-19 pandemic, these data are all the more important as the humoral anti-SARS-CoV-2 vaccine response is preferentially obtained in patients with no complete BC depletion.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data that support the findings of this study are available on request from the corresponding author (J.A.).

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declare no conflicts of interest.

Ethics: The protocol and the informed consent document have received Institutional Review Board/Independent Ethics Committee (IRB/IEC) approval before initiation of the study (‘Comité de Protection des Personnes’ Paris Ile de France I). The study was declared to the Commission Nationale de l’Informatique et des Libertés (reference 2221678).

Author notes

R.G. and O.A.T. contributed equally.