Efficacy of early rituximab treatment in primary Sjögren’s syndrome: a systematic review and meta-analysis

Abstract

Objective

This systematic review and meta-analysis aimed to assess Rituximab (RTX)’s efficacy and safety in primary Sjögren’s syndrome (pSS), particularly how treatment timing influences outcomes.

Methods

The study included randomized controlled trials (RCTs) and quasi-experimental studies evaluating RTX in pSS patients, focusing on disease activity (European League Against Rheumatism Sjögren’s Syndrome Disease Activity Index [ESSDAI] score) and adverse events (AEs). Searches were conducted in MEDLINE, Embase, SCOPUS, and Cochrane Library databases up to July 2024. Risk of bias was assessed using Cochrane Risk of Bias 2.0 (RoB 2) and Joanna Briggs Institute (JBI) checklists. Meta-analysis was performed in Stata 17 with a random-effects model, reporting mean differences in ESSDAI and I² for heterogeneity.

Results

From 555 articles, 15 studies were included (4 RCTs and 11 quasi-experimental studies). RCT meta-analysis showed a mean difference of 0.09 (95% confidence interval [CI] –0.43, 0.61), indicating no significant RTX efficacy. In contrast, the pooled quasi-experimental analysis revealed a mean difference of –4.36 (95% CI –5.83, –2.89), suggesting a significant reduction in disease activity. Meta-regression indicated no significant correlation between RTX efficacy and mean disease duration. Subgroup analysis of disease duration (under vs. over 60 months) showed no significant difference. Safety assessment indicated no significant differences in AEs between RTX and placebo in RCTs. In quasi-experimental studies, infusion reactions and infections were the most common AEs, with serious infections being the most severe.

Conclusion

RTX did not show significant improvement in RCTs. However, RTX significantly reduced pSS activity at week 24 or month 6 following treatment, based on quasi-experimental studies. We found no significant correlation between RTX efficacy and disease duration.

INTRODUCTION

Primary Sjögren’s syndrome (pSS) is an autoimmune systemic disease primarily characterized by lymphocytic infiltration of exocrine glands. It is best known for affecting the salivary and lacrimal glands, causing dryness of the eyes and mouth (sicca syndrome) [1-3]. pSS also presents with systemic features such as fatigue and arthralgia and may involve pulmonary, renal, and neurological manifestations [4]. It predominantly affects female, with the most common age range being 40~60 years [5,6]. The symptoms of pSS are often non-specific, leading to delays in diagnosis and challenges in management [7,8].

Several markers are used for diagnosis, including anti-Ro, anti-La, and rheumatoid factor. Additionally, criteria such as the American-European Consensus Criteria for Sjögren’s Disease are employed for diagnostic purposes [9]. The pathogenesis of pSS involves B-cell activity, though the precise mechanisms remain unclear [10,11].

While some immunosuppressive treatments are available, many therapeutic approaches are limited to symptomatic relief. However, due to the crucial role of B-cells in pathogenesis, rituximab (RTX) has been explored as a potential treatment option [12].

RTX is a monoclonal anti-CD20 antibody that has shown effectiveness in various autoimmune disorders [13]. The use of RTX in pSS has been investigated in various observational studies and randomized controlled trials (RCTs), but the findings are conflicting, with some research indicating RTX is effective [14] for pSS and other research suggesting it is not [15]. It is still uncertain whether RTX is a viable therapeutic option, given conflicting evidence and reports suggesting only temporary efficacy in some studies [14-17].

A persistent concern with RTX treatment is its safety in pSS, including both infusion reactions and long-term adverse effects [18].

An essential factor influencing drug efficacy is the time between disease onset and treatment initiation. This is particularly relevant in pSS, where the irreversible nature of glandular infiltration makes early intervention critical; a topic that remains underexplored in the literature.

This systematic review and meta-analysis aim to provide distinct and meaningful contributions to the understanding of RTX in pSS. Specifically, it evaluates the influence of disease duration on RTX’s efficacy, an underexplored factor that could impact treatment outcomes, and the safety of RTX. Additionally, the study separates the analysis of RCTs and quasi-experimental studies, offering insights into RTX’s effectiveness in both settings.

MATERIALS AND METHODS

This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19].

Study selection

Studies focusing on patients diagnosed with pSS, with RTX used as the therapeutic intervention have been included. The comparator groups included placebo (PBO), standard care without RTX, or RTX administered at different stages of disease progression. Both RCTs and quasi-experimental before-and-after observational studies that reported disease activity levels both prior to and following the use of RTX were eligible for inclusion. Any non-human studies and studies which were not published in English were excluded

The primary outcome was disease activity, measured by the European League Against Rheumatism (EULAR) Sjögren’s Syndrome Disease Activity Index (ESSDAI), along with the occurrence of adverse events (AEs), including both infusion-related and long-term complications. Studies which did not use ESSDAI as the disease activity measure were also excluded.

Search strategy

We performed the search using Medical Subject Headings (MeSH) and relevant keywords (Supplementary Table 1). The terms used for pSS included: Sjogren’s Syndrome, Sjögren’s Syndrome, Syndrome, Sjögren’s, Sjögren Syndrome, Sicca Syndrome, Xerostomia, Hyposialia, Asialia, Mouth Dryness, and Hyposalivation.

For RTX, the following terms were used: CD20 Antibody, Rituximab, Rituximab CD20 Antibody, Rituxan, IDEC-C2B8 Antibody, Mabthera, GP2013, Genentech, Hoffmann-La Roche, and Roche brands of rituximab.

We applied a study design filter based on the Cochrane handbook for systematic reviews. Searches were conducted in MEDLINE, Embase, SCOPUS, and Cochrane Library databases, covering publications up to July 2024.

Study selection

Titles and abstracts were screened for eligibility, and the full texts of potentially relevant studies were reviewed against the inclusion criteria. Any uncertainties or disagreements during the selection process were resolved through discussion and consultation with another author. The selection process was documented using a PRISMA flow diagram.

Data extraction

The following study characteristics were extracted: Author(s), Year of Publication, Journal, Country, and Study Design.

For participant information: Sample Size, Age, Gender Distribution, Inclusion, and Exclusion Criteria.

For intervention details: RTX dosage, control group medication, study duration, and disease duration at baseline.

For outcomes: ESSDAI scores for disease activity and reported AEs.

Risk of bias assessment

The Cochrane Risk of Bias 2.0 (RoB 2) tool was employed for RCTs [20], while the Joanna Briggs Institute (JBI) checklist was used for quasi-experimental studies [21]. Each study was categorized as having high, moderate, or low risk of bias. Studies classified as high risk were excluded from the meta-analysis.

Data synthesis and analysis

Statistical formulas were applied to impute missing means or standard deviations. If data were available only in charts, WebPlotDigitizer (Ankit Rohatgi, Austin, TX, USA) was used for extraction. Meta-analysis was performed using Stata 17 (StataCorp LLC, College Station, TX, USA), focusing on data collected at week 24 or month 6 of follow-up.

In non-randomized studies, only the data regarding pre- and post-RTX treatment was extracted.

All results were reported with a 95% confidence interval (CI), and a p-value<0.05 was deemed statistically significant. A random-effects model was adopted to address heterogeneity across studies. The mean difference, standardized mean change, and standard deviation change were calculated for each study. Heterogeneity was evaluated using the I² statistic, classified as: Low, I²<25%; Moderate, 25%~75%; High, I²>75% [22].

To assess the correlation of RTX efficacy and disease duration meta-regression and subgroup analysis are conducted.

RESULTS

A total of 524 studies were found in the initial search, with an additional 31 added using references from other systematic reviews in the field. Of these studies, 189 were duplicates. After screening 366 study titles and abstracts, 49 were selected for full-text screening, of which 15 met the inclusion criteria for final analysis. Four studies were RCTs and 11 were quasi-experimental (Supplementary Figure 1). Out of the four RCTs, two shared the same data; thus, only one was included in meta-analysis.

These studies were published from 2012 to 2022. The study populations ranged from 25 to 67 cases and 13 to 66 controls in RCTs and from 9 to 78 in quasi-experimental studies. The majority of patients were female, consistent with the epidemiology of pSS.

All RCTs included an RTX group and a control group, which used either standard treatment or a PBO. All quasi-experimental studies observed pSS patients before and after RTX treatment courses. The follow-up ranged from 24 to 68 weeks in RCTs and from 35 weeks up to 3 years in quasi-experimental studies. All studies reported ESSDAI as a measure of disease activity. Results at week 24 or month 6 were used for the meta-analysis.

Using the JBI tool, an answer of “yes” scored 1 point, “no” scored 0 points, and “unclear” scored 0.5; Not applicable questions were not counted. The sum of scores was classified as low risk at 80% or higher, moderate risk for 50%~79%, and high risk for below 50%. This approach resulted in four low-risk and seven moderate-risk bias assessments (Supplementary Table 2).

The ROB2 tool identified one study as low risk, and two studies had some concerns (Supplementary Table 3).

The Mariette et al. [23] study included four arms: PBO, subcutaneous belimumab, intravenous RTX, or sequential belimumab+RTX. Only the RTX and PBO arms were included since belimumab is not yet considered standard treatment, and comparisons involving the belimumab and sequential belimumab+RTX arms do not meet our inclusion criteria. They reported a reduction in ESSDAI for RTX compared with the PBO group, but the effect size CI included zero, ranging from –1.17 to 0.19. Bowman et al. [16] reported a lower ESSDAI in PBO with an effect size of 0.51 and a CI of 0.16 to 0.86. Devauchelle-Pensec et al. [15]’s result was also non-significant for efficacy (Tables 1 and 2) [14-18, 23-31].

Table 1.

Data extraction table of RCTs

Study characteristic						Participant information
Author	Year of publication	Journal	Country	Study design		RTX group population	Control group population	Age mean (SD)	Sex distribution (female percentage)
Mariette et al. [23]	2022	JCI Insight	Multi-national (France, United Kingdom, Italy, Netherlands, Norway, USA)	RCT		25	13	RTX: 55.2 (15.07) PBO: 52.7 (12.67)	92%~100%
Devauchelle-Pensec et al. [15]	2014	Annals of Internal Medicine	France	RCT		63	57	RTX: 52.9 (13.3) PBO: 55.6 (13.6)	97 (80.8%)
Bowman et al. [16]	2017	Arthritis and Rheumatology	UK	RCT		67	66	RTX: 54.3 (11.5) PBO: 54.4 (11.6)	124 (93.2%)
Participant information					Intervention detail
Inclusion criteria		Exclusion criteria			Intervention (RTX details: dosage, frequency)		Control/comparison (PBO, other treatments)		Duration of intervention (wk)
Adults (≥18 years) with a confirmed diagnosis of pSS, an ESSDAI score of ≥5, symptomatic oral dryness (NRS≥5/10), and evidence of glandular reserve function with minimal salivary flow at baseline		Secondary Sjögren’s syndrome, life-threatening/organ-threatening complications, severe immunocompromise, severe cardiac/liver disease (except Gilbert’s syndrome, asymptomatic gallstones), major organ transplant, malignancy (past 5 years), serious infections, hypersensitivity to belimumab/rituximab/other mAbs, PML, suicide risk. Recent use of immunosuppressives, B-cell depleting agents, biologics, IVIG/plasmapheresis, high-dose steroids, or live vaccines (within specified timeframes). Laboratory exclusions: IgA deficiency, critically low WBC, neutrophils, or IgG levels, elevated liver function markers. Study withdrawal for severe infections, critical IgG/neutrophil reductions, PML, severe skin reactions, liver dysfunction, suicide risk, or pregnancy.			RTX IV (1,000 mg in 250 mL saline) in 2 courses at weeks 0, 2, 24, and 26		PBO		68
Patients with scores of 50 mm or greater on at least 2 of 4 visual analogue scales (global disease, pain, fatigue, and dryness) and recent-onset (10 years) biologically active or systemic pSS.		The exclusion criteria included secondary Sjögren’s syndrome, cytotoxic drug therapy within the past four months, severe renal or hematologic failure, a history of cancer, hepatitis B or C, HIV infection, tuberculosis, severe diabetes, or any other chronic illness. Additional exclusions were the presence of infection, a history of severe allergic or anaphylactic reactions to humanized or murine monoclonal antibodies, and an inability to comprehend the study protocol.			RTX 1 g at weeks 0 and 2		PBO		24
Aged 18~80 years, positive for Anti-Ro autoantibodies, had some (greater than zero) unstimulated salivary flow, symptomatic fatigue and oral dryness worse than 5/10 on a patient-completed Likert scale, on a stable dose of corticosteroids, NSAIDS, DMARDS, pilocarpine and antidepressants for 4 weeks prior and throughout the study and provided written informed consent to participate.		Secondary SS, Hepatitis B or C, tuberculosis, HIV or other immunodeficiency, prior RTX or monoclonal antibody usage, malignancies within 5 years prior, recent organ transplant, major surgery planned or 3 months prior, pregnancy/lactation and unwillingness to use contraception throughout the study.			RTX IV (1,000 mg in 250 mL) at weeks 0, 2, 24, and 26, with pre-and post-infusion medication including corticosteroids.		PBO IV		48
Result									AEs
Mean disease duration year-RTX group	Mean disease duration year-control group	ESSDAI week 24 RTX group mean (SD)	ESSDAI baseline RTX group mean (SD)	ESSDAI week 24 control group mean (SD)	ESSDAI baseline control group mean (SD)	SMD	SE of SMD		AEs reported
6.1	8.8	5.9 (4.65)	11.2 (5.2)	15.1 (4.065)	12.2 (5.23)	–0.49	0.35		100% AEs in PBO arm none was considered serious and there was no death reported. In the RTX arm, 96% AEs, 16% of which was considered serious, and no death in the RTX group either.
7.4	8.4	8.8 (7.57)	10 (6.9)	8.5 (7.79)	10.2 (6.8)	0.068	0.18		87.3% AEs, 20.6% serious in the RTX group and 93%, 14% serious in the PBO group.
5.3	6.2	4.1 (9)	5.3 (4.7)	4.4 (8.93)	6 (4.3)	0.51	0.18		10 patients AEs in both PBO and RTX groups.

RCTs: randomized controlled trials, RTX: rituximab, SD: standard deviation, JCI: Journal Citation Index, PBO: placebo, pSS: primary Sjögren’s syndrome, NRS: Numeric Rating Scale, PML: progressive multifocal leukoencephalopathy, IVIG: intravenous immunoglobulin, NSAIDS: nonsteroidal anti-inflammatory drugs, DMARDS: disease-modifying antirheumatic drugs, HIV: human immunodeficiency virus, ESSDAI: European League Against Rheumatism Sjögren’s Syndrome Disease Activity Index, SMD: standardized mean difference, SE: standard error, AEs: adverse events, Ig: immunoglobulins, WBC: white blood cells.

Table 2.

Data extraction table of quasi-experimental studies

Study characteristic						Participant information
Author	Year of publication	Journal	Country	Study design		Sample size	Mean age (yr)	Sex distribution (female percentage)	Inclusion criteria	Exclusion criteria
Mirouse et al. [26]	2019	Autoimmunity Reviews	France	Before-after		19	54	95%	Patients (≥18 years) with a diagnosis of pSS (AECG criteria) and at least one clinical and/or ultrasound synovitis during the follow-up	Patients with other diagnosis of inflammatory (p.e. ACR-EULAR criteria for rheumatoid arthritis), infectious or microcrystalline arthritis (gout, chondrocalcinosis) were excluded.
Pavlych et al. [27]	2020	Frontiers in Medicine	Italy	Case control		9	51.8	89%	pSS diagnosis Disease duration <5 years Systemic moderate-high activity (ESSDAI ≥5)	Secondary SS, severe cardiac, pulmonary, renal, or hematologic failure, history of cancer in the last 5 years Hepatitis B or C infection, HIV infection, tuberculosis, severe diabetes, or any other chronic disease, evidence of infection Inability to understand and adhere to the treatment
Carubbi et al. [28]	2013	Arthritis Research & Therapy	Italy	Case-control		19	40	94.70%	pSS diagnosis, recent onset of the disease (maximum 2 years), active disease (ESSDAI ≥6), values >50 mm for two out four visual analogical scales.	Secondary SS, severe cardiac, pulmonary, renal or hematologic failure, history of cancer in the last 5 years, hepatitis B or C infection, HIV infection, tuberculosis, severe diabetes, any other chronic disease, or evidence of infection, and inability to adhere to the protocol.
Mekinian et al. [29]	2012	Annals of the Rheumatic Diseases	France	Cohort		17	62	82.30%	Patients with pSS and PNS involvement, included in the French AIR registry, fulfilling the AECG criteria for pSS.	Not specified
Gottenberg et al. [30]	2013	Annals of the Rheumatic Diseases	France	Cohort		78	59.8	86%	Patients with pSS treated with RTX.	Not specified
Berardicurti et al. [18]	2022	The Journal of Rheumatology	Italy	Historical cohort study		35	54	91%	Patients ≥ 18 years, fulfilled the American-European Consensus Criteria for pSS, received at least 1 RTX infusion, they had moderate (5≤ESSDAI≤13) or high activity (ESSDAI ≥14), or if they had enlarged parotid glands or arthritis for at least 3 months and were unresponsive to glucocorticoids and immunosuppressives	Patients with missing data were excluded from the study.
Arends et al. [25]	2017	Clinical and Experimental Rheumatology	Nederland	Before-after		28	43	96%	All patients fulfilled the revised AECG criteria for pSS and were over 18 years of age.	Not specified
Pepple et al. [14]	2022	ACR Open Rheumatology	UK	Retrospective observational cohort study		40	54	95%	≥18 years old, fulfilling the revised 2002 AECG classification criteria for pSS, and having at least a 6-month follow-up post RTX.	Exclusion criteria were having secondary SS and concurrent anti-cyclic citrullinated peptide antibody positivity.
Meiners et al. [24]	2012	Annals of the Rheumatic Diseases	Netherlands	Before-after		28	43	96%	Aged 18 years and fulfilled the revised European-US criteria for pSS.	Not specified
Mekinian et al. [17]	2012	Clinical and Experimental Rheumatology	France	Before-after		11	55	91%	pSS AECG criteria and pSS-related CNS involvement defined by the presence of central nervous clinical impairment associated with diffuse white matter T2-weigted hyper signal after exclusion of other causes of CNS involvement.	Not specified
Chen et al. [31]	2016	Clinical and Experimental Rheumatology	Taiwan	Before-after		10	51.1	100%	All patients fulfilled the 2002 American-European Revised Classification Criteria for pSS and ILD diagnosed according to the British Thoracic Society Interstitial Lung Disease guidelines	Patients with other connective tissue diseases that may induce secondary SS such as systemic lupus erythematosus, rheumatoid arthritis, PM/DM, scleroderma, primary biliary cirrhosis, and mixed connective tissue disease, other medical illnesses that could cause ILD, such as drugs, infections, etc., were excluded
Intervention detail								Result		AEs
Intervention (RTX details: dosage, frequency)				Median duration of intervention (wk)		Mean disease duration baseline (mo)		ESSDAI after RTX mean (SD)	ESSDAI before RTX mean (SD)	AEs reported
RTX dose is not mentioned				173		9		4.13 (2.59)	6.11 (4.44)	Only 2 AEs after third line treatment of RTX
RTX (MabThera): 1,000 mg at day 1 and day 15, repeated after 24 weeks				48		21		6.4 (4.5)	12.6 (6.6)	Only 2 AEs in 12 month follow-up
RTX 1,000 mg on day 1 and day 15, repeated every 24 weeks for six courses				120		13		9.8 (8.71)	20.3 (12.64)	No AEs reported
RTX 1 g repeated 15 days later in 8 patients (47%). Corticosteroids were associated with RTX in 15 cases (88%) at a median daily dose of 10 mg (5~80 mg). Other immunosuppressive agents were associated in 5 cases (29%).				143		Unavailable		14 (6)	22.5 (8.5)	Six (35%) AEs, 2 were considered severe
RTX, 67 patients received two infusions of 1 g, 11 patients received four infusions of 375 mg				151		142.8		7.5 (6.5)	11 (7.25)	Four infusion reactions and 1 delayed serum sickness-like disease which lead to RTX discontinuation and also 3 serious infections (1.3/100 patient-years) and two cancer related deaths.
Patients received an infusion of 1,000 mg RTX on day 1 and day 15 and repeated every 24 weeks				155		60.72		4.22 (3.25)	7.76 (5.03)	Thirteen (37%) discontinuation of RTX among those Infusion-related reactions in 4 patients and hypogammaglobulinemia in 7 patients were reported.
RTX 1,000 mg intravenously at days 1 and 15				60		64.8		2.96 (2.77)	8 (4.5)	AEs were not explored
A first cycle of therapy consisting of 100 mg of methylprednisolone and 1,000 mg of RTX given intravenously on days 1 and 14				156		63.72		3.65 (3.92)	11.39 (6.45)	There were 12 SIEs in 10 patients (pneumonia=6; cellulitis and osteomyelitis=4; COVID-19 pneumonitis with subsequent recovery=2) and only one patient developed CD5-negative lymphoproliferative disorder 5 years after RTX infusion.
RTX (1,000 mg) infusions on days 1 and 15				60		80		3 (3)	8 (5)	AEs were not explored
Previous treatments Corticosteroids (45%) dose (mg/day) 10 [5–15], immunosuppressant agents (64%), RTX regimen 1 gx2 (82%), 375 mg/m2×4 (18%), RTX associated treatments Corticosteroids (64%) immunosuppressant agents (45%)				56		109		16.5 (5.5)	16 (5)	One mild infusion reaction, 1 severe infection and one patient died in the context of cognitive degradation and weight loss 11 months after RTX administration.
All patients underwent two doses of RTX 1,000 mg, intravenous infusion, 14 days apart, and repeated the same protocol every half a year depending on individual responsiveness.				26		35.16		2.4 (1.1)	4.6 (1.3)	No infusion reactions reported and only one patient experienced severe infection.

pSS: primary Sjögren’s syndrome, AECG: American-European Consensus Group, ACR: American College of Rheumatology, EULAR: European Alliance of Associations for Rheumatology, ESSDAI: European League Against Rheumatism Sjögren’s Syndrome Disease Activity Index, HIV: human immunodeficiency virus, PNS: peripheral nervous system, RTX: rituximab, CNS: central nervous system, ILD: interstitial lung disease, PM/DM: polymyositis/dermatomyositis, SD: standard deviation, AEs: adverse events. SIEs: severe infection episodes.

While most quasi-experimental studies showed a significant ESSDAI difference between pre-RTX and week 24, Mekinian et al. [17] found a non-significant efficacy result. The pooled analysis of quasi-experimental studies using a random-effects model showed a mean difference of 4.36 in score reduction after RTX, with a 95% CI of –5.83 to –2.89 (Figure 1).

Figure 1.

Forest plot of quasi experimental studies. N: number, SD: standard deviation, CI: confidence interval, REML: restricted maximum likelihood estimation.

With I² of 73.87%, the heterogeneity is considered moderate for quasi-experimental studies.

Nine of the studies showed significant mean difference after RTX use and the other two did not show any statistically meaningful efficacy (Figure 1). The weight of each study along with its mean difference CI is also mentioned in Figure 1.

In contrast to the above results, the meta-analysis of RCTs using a random-effects model yielded a mean difference of 0.09, with a 95% CI of –0.43 to 0.61 (Figure 2).

Figure 2.

Forest plot of RCTs. RCTs: randomized controlled trials, CI: confidence interval, REML: restricted maximum likelihood estimation.

With I² of 76.39%, the heterogeneity is considered high for RCTs.

To combine the result of RCTs and quasi-experimental studies we have calculated the standardized mean difference and conducted the meta-analysis using random-effect model showing statistically significant efficacy of RTX resulting in 1.12 with 95% CI of –1.54 to –0.69 ESSDAI score reduction (Figure 3).

Figure 3.

Forest plot of combining both RCTs and quasi-experimental studies. RCTs: randomized controlled trials, CI: confidence interval, REML: restricted maximum likelihood estimation.

With a high heterogeneity I² of 90.16%.

The results of meta-analysis in each methodology and the combination of them is summarized in Supplementary Table 4.

To examine the impact of disease duration on RTX efficacy, meta-regression was conducted for both RCTs and quasi-experimental studies. Mekinian et al. [28] in their 2012’s study of 17 patients did not report disease duration, so this study was excluded from further analysis. The result of meta-regression for 10 quasi-experimental studies showed a coefficient of 0.0099 with a 95% CI of –0.03 to 0.05 and a p-value of 0.632 for mean disease duration. The constant was –4.7677 with a 95% CI of –7.79 to –1.73 and a p-value of 0.002, indicating that the significant efficacy of RTX is not correlated with mean disease duration (Supplementary Table 5).

The result of meta-regression for RCTs was a coefficient of –0.1568 with a 95% CI of –0.98 to 0.66 and a p-value of 0.70 for mean disease duration. The constant was 1.0481 with a 95% CI of –4.15 to 6.25 and a p-value of 0.69, showing no significant efficacy of RTX or correlation with mean disease duration (Supplementary Table 6).

The meta-regression was also done on the combined results showing the coefficient of 0.0078 with 95% CI of –0.0039 to 0.0195 and the non-significant p-value of 0.195 for the disease duration and the coefficient of –1.6077 with 95% CI of –2.5165 to –0.6988 and the significant p-value of 0.001 for the constant (Supplementary Table 7).

As only three RCTs were included, subgroup analysis was not conducted on RCTs. The quasi-experimental studies were divided into two groups based on disease duration. The first group included studies with a mean disease duration under 60 months, and the second group included studies with a mean disease duration over 60 months. The result of the subgroup analysis showed between-group differences of 0.09 with a p-value of 0.77, supporting the previous meta-regression results (Figure 4). As mentioned before one of 11 quasi-experimental studies was not involved in this analysis because of not reporting disease duration. Also both subgroups showed significant reduction in ESSDAI after RTX use with –3.90 reduction in first group and –4.42 in the second group (Figure 4).

Figure 4.

Subgroup meta-analysis of quasi-experimental studies biased on disease duration. N: number, SD: standard deviation, CI: confidence interval, REML: restricted maximum likelihood estimation.

Safety

In the included RCTs, Mariette et al. [23] reported 100% AEs in the PBO arm; none were considered serious, and no deaths were reported. In the RTX arm, they reported 96% AEs, 16% of which were serious, with no deaths in the RTX group either. This was not considered significantly different between the two groups [23]. Devauchelle-Pensec et al. [15] reported 87.3% AEs, 20.6% serious in the RTX group, and 93%, 14% serious in the PBO group. Bowman et al. [16] also found a similar number of 10 AEs in both the PBO and RTX groups. These findings suggest no additional hazard associated with RTX use compared with PBO.

Of the 11 quasi-experimental studies included, only nine reported AEs. Meiners et al. [24] and Arends et al. [25] did not assess the AEs of RTX. Mirouse et al. [26] did not explain the AEs in detail and reported two AEs after the third-line RTX treatment. Pavlych et al. [27] reported only two AEs in a 12-month follow-up and pretreated patients with prednisolone, paracetamol, and chlorpheniramine to minimize AEs. Carubbi et al. [28] found no AEs. Mekinian et al. [29]’s study on pSS with peripheral nervous system involvement reported six (35%) AEs, with two considered severe. Gottenberg et al. [30]’s study reported four infusion reactions and one delayed serum sickness-like disease that led to RTX discontinuation, as well as three serious infections (1.3/100 patient-years) and two cancer-related deaths.

Berardicurti et al. [18] reported 13 (37%) RTX discontinuations; infusion-related reactions in four patients and hypogammaglobulinemia in seven patients were the most common causes. Two hypogammaglobulinemia events were followed by severe infections [18]. Pepple et al. [14] had 12 severe infections in 10 patients, and only one patient developed CD5-negative lymphoproliferative disorder 5 years after RTX infusion. Mekinian et al. [17] in his study on patients with central nervous system manifestations reported one mild infusion reaction, one severe infection, and one death due to cognitive degradation and weight loss 11 months post-RTX administration. Finally, in Chen et al. [31]’s study, no infusion reactions were reported, and only one patient experienced a severe infection.

Overall, infusion reactions and infections were the most common AEs. Although infusion reactions were generally not severe, infections were more likely to be severe.

DISCUSSION

The number of studies assessing the question of RTX efficacy in pSS has been growing lately.

Regarding the results of RCTs, many immunosuppressants have shown limited or no response in treating pSS [32]. Many studies have demonstrated substantial B-cell activity in pSS [11], and since RTX specifically targets these cells, it is logical to investigate its efficacy in pSS. Therefore, our study addressed this critical subject. Most studies in this field have not used similar or reliable criteria to evaluate disease activity and treatment efficacy, likely due to the complexity of pSS diagnosis and its nature [9]. Consequently, we selected ESSDAI as a standard, reliable tool for measuring disease activity.

While RTX significantly affects B-cells, which play a crucial role in the physiopathology of pSS [33], its efficacy on disease activity remains controversial, especially in RCTs. In this study, through a meta-analysis of both before-after studies and RCTs, we found that RTX significantly reduces pSS activity based on multiple observations from before-after studies. However, the results from RCTs did not align. This inconsistency is likely due to the small number of high-quality RCTs on this topic, and the higher heterogeneity. Besides the mean disease duration in RCTs were 93.6 months while for quasi-experimental studies was 59.9 months which means there were difference in baseline of two study designs. Another cause could be the differences between intervention regimens. Additionally, the open-label nature of quasi-experimental studies can influence patient and clinician perceptions, potentially leading to more favorable reporting of outcomes, and publication bias should also always be considered because positive results are more likely to be published in observational studies, contributing to an apparent better response and this could case both heterogeneity and inconsistency between RCT and quasi-experimental studies (Supplementary Figure 2). We believe that additional RCTs will help reconcile these differences and align the findings of RCTs with those of before-after studies.

Our study distinguishes itself from previous meta-analyses in several ways. First, we included a broader range of databases, and incorporated more recent studies published up to July 2024. This enabled a more comprehensive and updated synthesis of available evidence. Second, we utilized ESSDAI as a measure of disease activity, which is considered a reliable scale for pSS assessment. Finally, our separate analysis of RCTs and quasi-experimental studies provided a combined understanding of RTX efficacy. Additionally, some older studies were not included because they did not report outcomes using ESSDAI due to this scale’s development date. This decision was based on the need for consistency and reliability in evaluating disease activity using a validated tool. The work of Chen et al. [12] which highlights these older studies, underscores their importance in establishing foundational knowledge. However, the heterogeneity of outcome measures in earlier trials made direct comparisons and meta-analyses challenging, further justifying our approach.

In our study, both RCTs and quasi-experimental studies were analyzed. We understand the potential bias due to the methodological differences. However, the inclusion of quasi-experimental studies was necessitated by the limited number of high-quality RCTs available in this field. While quasi-experimental studies inherently carry a higher risk of bias, they provide valuable real-world insights that complement the controlled evidence from RCTs. By combining these approaches, we aimed to enhance the breadth and reliability of our conclusions. To minimize biases, we conducted separate analyses for each methodology, allowing for a clearer comparison and a deeper understanding of the results.

There is still no consensus on the best time to assess RTX efficacy. However, we used 24 weeks or 6 months as our time point, as this was the most commonly referenced period in other studies. Some studies reported a decrease in RTX efficacy over the long term, possibly due to discontinuation of the treatment [34-37]. RTX imposes considerable financial burdens, limiting its use in many studies, which could result in transient effects. Another possible reason for these findings is that RTX may have only a temporary effect in pSS. To address this, further studies are suggested to involve long-term RTX administration and follow-up.

Lacking long-term use of RTX in high quality studies doesn’t allow a long-term efficacy precise conclusion so the long-term use and efficacy of RTX should be explored more in future studies regarding its short term efficacy.

The heterogeneity of the studies was high in RCTs and moderate in before-after studies. This may be due to variations in patient characteristics, RTX dosing regimens, and disease severity across studies. There were differences in baseline ESSDAI between studies and also wide differences in sample sizes in studies. Another cause of this heterogeneity could be the fact that some studies combined their RTX intervention with corticosteroids as a safer therapeutic regimen. The differences in designs both between RCTs and quasi-experimental studies and also between studies in each group are increasing the heterogeneity too. It is worth mentioning that the risk of bias of few studies were low (Supplementary Tables 2 and 3) also there is a wide geographical distribution in the studies (Tables 1 and 2) which results in clinical settings differences adding disease durations wide variety the heterogeneity is explained. Additionally, the publication bias addressed in Supplementary Figure 2 is also another factor to increase the heterogeneity.

Another key question is whether disease duration affects RTX efficacy. Given the permanent nature of infiltrative diseases, it is essential to begin treatment before irreversible stages are reached. Although pSS stages are not well-defined in terms of reversibility, it is reasonable to assume parallels with other diseases that share similar characteristics. In this study, we explored this question through both meta-regression and subgroup analysis, focusing on disease duration at the start of RTX treatment. Our findings did not reveal a significant correlation between these factors, but two points are worth considering. First, the wide variation in disease duration among the studies likely increased heterogeneity. Second, none of the studies administered RTX within 9 months of diagnosis, which may represent a significant delay in starting treatment. Further studies should explore RTX as an urgent therapeutic option. However, this could pose challenges given the complex and often delayed diagnosis of pSS following the onset of symptoms.

Even though limited data on safety assessment makes it difficult to make a definite conclusion, RTX appears to be safe for use in pSS, with limited critical adverse effects reported in most studies. There were also minimal differences between the intervention and control groups in RCTs. One observation from multiple studies was that RTX infusion reactions were less commonly reported when patients had received prior corticosteroids or immunosuppressive. But also, most of the patients with infection AEs had also received an immunosuppressive. This observation requires further investigation through dedicated studies. Although immunosuppressive use alongside RTX could confound the assessment of RTX efficacy, if future studies confirm a reduction in adverse reactions with combined use, it would suggest that the therapeutic effect comes primarily from RTX or its co-administration. These findings would also allow for safer experimentation. We have to keep in mind that missing or inconsistent reporting of adverse effects in some studies limited the safety conclusions.

For future studies, we recommend exploring RTX as an immediate treatment option. Additionally, further research is needed to develop safer infusion protocols for RTX. More high-quality RCTs are also essential to address the remaining discrepancies between RCTs and quasi-experimental studies. Which could also focus on reducing disease duration before beginning of the treatment. Also, the use of corticosteroids alongside RCTs could be experimented for both combined efficacy and higher safety. Another point worth mentioning is that using a unanimous criterion for pSS is important to get comparable results. ESSDAI could be a proper choice for this purpose in future studies.

CONCLUSION

RTX can significantly reduce pSS activity at week 24 or month 6 following treatment, based on quasi-experimental studies and the combination of both quasi-experimental and randomized studies. However, the findings from RCTs remain inconsistent. There is a limited number of high-quality studies, especially RCTs, evaluating the efficacy of RTX in pSS. Although we found no significant correlation between RTX efficacy and disease duration, multiple confounding factors may have influenced these results.

SUPPLEMENTARY DATA

Supplementary data can be found with this article online at https://doi.org/10.4078/jrd.2024.0149