Real‑world pharmacovigilance of rituximab‑associated adverse reactions in children and adolescents: Disproportionality signal detection and time‑to‑onset pattern characterization using the FAERS database

ABSTRACT

Rituximab is widely used for B‑cell malignancies and autoimmune diseases, but its pediatric safety profile requires systematic real‑world evaluation. This research mined safety signals of adverse events (AEs) linked to rituximab in pediatric patients using U.S. Food and Drug Administration Adverse Event Reporting System (FAERS) data from 2004 to 2025. A total of 2,231 pediatric AE reports involving rituximab were identified and analyzed using four disproportionality methods (ROR, PRR, BCPNN, and MGPS) for signal detection and Weibull distribution analysis for time-to-onset patterns. Results showed a balanced sex distribution in AE reports (47.7% male, 48.5% female). Most reports involved adolescents aged 12–17, with an overall increasing trend. A total of 149 positive safety signals were detected via disproportionality analysis. Several of these signals pertained to immune system – and infection‑related adverse events. Notably, the elevated signal magnitudes for certain signals were substantially confounded by indication bias, pre‑existing comorbidities, and reports associated with therapeutic failure. Stratified analyses revealed differences in the signal distribution of adverse‑event reports across age and sex subgroups. Time‑to‑onset analysis of adverse‑event reports indicated that most events documented in these reports occurred within 30 d of therapy initiation, with a median of 7 d. This study characterizes the reported safety profile of rituximab in pediatric patients based on FAERS spontaneous reports. Our findings represent exploratory signal hypotheses only and can serve as a reference for subsequent relevant investigations.

Introduction

Since its approval, rituximab, a chimeric mouse-human monoclonal antibody that targets the CD20 antigen, has significantly transformed the therapeutic landscape of B-cell non-Hodgkin lymphoma and various autoimmune diseases.¹ By inducing B‑cell lysis, complement‑dependent cytotoxicity, and antibody‑dependent cell‑mediated cytotoxicity, rituximab eliminates pathogenic B‑cell clones and has demonstrated substantial efficacy in conditions such as rheumatoid arthritis, systemic lupus erythematosus, vasculitis, and organ transplant rejection.² As its clinical indications have expanded, rituximab is increasingly used in pediatric populations and has become a key therapeutic option for refractory nephrotic syndrome in children, juvenile idiopathic arthritis, and related hematologic malignancies.³

Despite its growing use in pediatric clinical practice, epidemiologic evidence on the safety of rituximab in children remains limited and fragmented. Current knowledge is derived primarily from registration clinical trials and sporadic case reports. However, clinical trials are often constrained by small sample sizes, short follow‑up durations, and the exclusion of critically ill children or those with comorbidities, making it difficult to detect rare adverse events, delayed toxicities, or the potential consequences of long‑term immunosuppression. Whether adverse reactions that are well characterized in adults, including infusion reactions, hepatitis B virus reactivation, hypogammaglobulinemia, and neutropenia, manifest similarly in children undergoing rapid growth and development remains a subject of debate.⁴ A recent systematic review of rituximab in pediatric pemphigus vulgaris (PV) indicated that while the agent is generally tolerable, the risk of infections is significantly elevated, and the adverse event profile varies across different age subgroups.⁵ Children are not simply smaller adults; their physiology, immune system maturity, pharmacokinetic profile, and apparent volume of distribution differ fundamentally from those of adults, resulting in distinct patterns of drug safety responses in the pediatric population. Therefore, directly extrapolating adult safety data to minors carries potential clinical risks. The disease spectrum unique to children may expose them to risks not observed in adult studies. Moreover, the long‑term remodeling effects of rituximab on the developing immune system and its potential associations with rare pediatric complications cannot be adequately assessed within a framework dominated by adult data.⁶ Drug‑related adverse events may not only interrupt treatment and reduce adherence but also precipitate severe infections, hematologic toxicity, or irreversible immune injury, leading to prolonged hospitalization, permanent disability, or even death.⁷ These consequences impose a heavy burden on affected families and strain substantial healthcare resources. The current situation underscores the urgent need for research focused on pediatric‑specific safety signals.

From a regulatory perspective, rituximab represents a typical example of a biologic therapy in which pediatric clinical use has expanded beyond the scope of the original approval. While rituximab is approved for several adult B-cell malignancies and autoimmune diseases, its use in many pediatric immune-mediated disorders, such as refractory nephrotic syndrome and systemic autoimmune diseases, is frequently based on accumulating clinical experience rather than extensive pediatric randomized controlled trials.⁸ Therefore, regulatory evaluation of its pediatric safety profile requires continuous integration of post-marketing surveillance data.

The FDA Adverse Event Reporting System (FAERS), one of the largest spontaneous reporting systems globally, encompasses over two decades of drug safety surveillance data and can capture rare, serious, and long‑term adverse events that are difficult to detect in clinical trials. Compared with traditional observational cohort studies, FAERS offers advantages including a large sample size, broad population coverage, and diverse reporting sources, enabling more sensitive detection of weak drug–event association signals.⁹

To characterize the real-world safety profile of rituximab in pediatric patients and provide additional evidence for pediatric drug safety monitoring and regulatory assessment, this study employed a retrospective pharmacovigilance approach using FAERS data. Through data cleaning and standardization, we constructed a high‑quality adverse event dataset to provide data‑driven support for clinical medication risk management and regimen optimization.

Materials and methods

Data sources and processing

Rituximab-related records were identified from the FAERS DRUG file. Both the generic name “Rituximab” and brand names “Rituxan,” “Mabthera,” and “GP2013” were used as search terms to retrieve drug exposure records. Only records in which rituximab was coded as the primary suspect drug (ROLE_COD = “PS”) were retained to maximize the specificity of drug–adverse event associations. Patients aged 0–17 y were included in the analysis. The data extraction and cleaning procedures were as follows: the fields PRIMARYID, CASEID, and FDA_DT were extracted from the DEMO table; for reports sharing the same CASEID, the record with the most recent FDA_DT was retained; if both CASEID and FDA_DT were identical, the record with the largest PRIMARYID was kept.¹⁰ Adverse events were classified using preferred terms (PTs) and system organ classes (SOCs) according to the Medical Dictionary for Regulatory Activities (MedDRA) version 27.0.¹¹ Figure 1A illustrates the process of screening for rituximab‑related adverse events in the pediatric population from the FAERS database.

Figure 1.

Disproportionality analysis and AEs distribution across SOC in rituximab. (A) Flowchart of this study; (B) Global distribution of AE reporting rates for rituximab; (C) Trend of AE reports for rituximab (Q1 2004-Q3 2025); (D) Distribution of AEs categorized by SOC; (E) Volcano plot illustrating the signal strength of SOC based on ROR; (F) Venn diagram of positive AE signals identified by four algorithms. DEMO, demographic and administrative information; DRUG, drug information; REAC, preferred terminology for adverse events; AE, adverse event; ADR, adverse drug reaction; SOC, system organ class; PT, preferred term; ROR, reporting odds ratio; PRR, proportional reporting ratio; BCPNN, Bayesian confidence propagation neural network; MGPS, multi-item gamma poisson shrinker.

Data analysis

Four established algorithms were employed for disproportionality analysis: the reporting odds ratio (ROR), the proportional reporting ratio (PRR), the Bayesian confidence propagation neural network (BCPNN), and the multiple gamma Poisson shrinker (MGPS). ROR and PRR are frequentist methods characterized by high detection sensitivity. ROR can effectively adjust for reporting bias when AE counts are low, while PRR is less affected by underreporting of AEs. The Bayesian algorithms, represented by BCPNN and MGPS, offer notable advantages in signal specificity, consistency, and control of false‑positive rates. BCPNN efficiently integrates multi‑source data and supports cross‑validation, whereas MGPS is particularly effective for detecting rare AE signals. To ensure the accuracy of signal detection, all four algorithms were applied simultaneously; an AE signal was defined as positive only when it met the preset thresholds across all four methods.¹⁰ A positive AE signal detected at the PT level was defined as a positive PT; otherwise, it was considered negative. All four algorithms utilized 2 × 2 contingency tables (Supplementary Table 1), with the corresponding formulas and signal criteria provided in Supplementary Table 2.¹²

During signal interpretation, MedDRA PTs were reviewed considering their clinical relevance and reporting context. Certain PTs primarily describing medication-use-related information (such as off-label use, drug effective for unapproved indication, and intentional product use issue) were interpreted cautiously, as these terms may reflect patterns of drug utilization rather than biological adverse drug reactions.

Time to onset (TTO) was defined as the interval from the drug start date (START_DT) to the AE occurrence date (EVENT_DT) and was described using the median, minimum, maximum, and interquartile range (IQR). All analyses were performed following the exclusion of reports containing missing or anomalous data. The Kaplan–Meier estimator was adopted to describe the distribution of TTO for eligible reports with complete onset‑time data, and the log‑rank test was utilized for between‑group comparisons of TTO distributions. The Weibull distribution model was fitted to the temporal trend of TTO data. Specifically, the Weibull analysis used the calculated TTO values (in days) as the input variable. Only reports with available and valid START_DT and EVENT_DT information were included. The time interval was calculated as the difference between the event occurrence date and the treatment initiation date. Separate Weibull models were fitted for selected SOCs and PTs with sufficient numbers of reports and complete TTO information, and the shape parameter (β) with its 95% confidence interval (CI) was used for classification as follows: (1) if β < 1 and the upper limit of the 95% CI is <1, the risk decreases over time (early failure type); (2) if β ≈ 1 and the 95% CI contains 1, the risk remains constant over time (random failure type); (3) if β > 1 and the lower limit of the 95% CI is >1, the risk increases over time (wear‑out failure type).¹⁰

All statistical tests were two‑sided, and a P‑value < 0.05 was considered statistically significant. Data organization and statistical analyses were performed using Microsoft Excel 2019 and R software (version 4.4.3).

Results

Adverse events and population characteristics

A total of 940,359 adverse event reports for pediatric patients aged 0–17 y were extracted from the DEMO dataset between the first quarter of 2004 and the third quarter of 2025. After removing duplicate records, 2,231 reports suspected to be associated with rituximab therapy were included in the analysis (Table 1). The sex distribution was relatively balanced: females accounted for 48.5% (1,082 cases) and males for 47.7% (1,065 cases). The majority of patients were adolescents aged 12–17 y (45.4%, 1,013 cases), followed by children aged 0–5 y (27.4%, 611 cases) and 6–11 y (27.2%, 607 cases). Most reports were submitted by healthcare professionals, including other health professionals (41.2%, 920 cases) and physicians (37.7%, 842 cases). Regarding adverse event outcomes, hospitalization was reported in 17.8% (397 cases), death in 9.6% (215 cases), life‑threatening events in 5.6% (126 cases), and disability in 0.6% (14 cases). Geographically, the United States contributed the largest proportion of reports (33.7%, 752 cases), followed by Canada (9.4%, 210 cases), China (9.4%, 209 cases), Japan (9.0%, 201 cases), and the United Kingdom (7.4%, 165 cases); other countries collectively accounted for 31.1% (694 cases) (Figure 1B). Between 2004 and the third quarter of 2025, the overall number of rituximab‑related adverse event reports in children showed an increasing trend (Figure 1C). Few reports were recorded in the early years; however, a marked increase occurred after 2017, peaking at 498 cases in 2024. As of the third quarter of 2025, 175 cases had been reported.

Table 1.

Epidemiological characteristics of adverse event reports.

Parameters	Number of reports (%)
Sex
Female	1082 (48.5%)
Male	1065 (47.7%)
Missing	84 (3.8%)
Age (year)
0–5	611 (27.4%)
6–11	607 (27.2%)
12–17	1013 (45.4%)
Weight (Kg)
0–24.9	190 (8.5%)
25–49.9	180 (8.1%)
≥50	174 (7.8%)
Missing	1687 (75.6%)
Occupation reporter
Health Professional	920 (41.2%)
Physician	842 (37.7%)
Consumer	158 (7.1%)
Pharmacist	89 (4.0%)
Missing	222 (10.0%)
Outcome
Death	215 (9.6%)
Disability	14 (0.6%)
Hospitalization	397 (17.8%)
Life-threatening events	126 (5.6%)
Other	1479 (66.4%)
Reporting country
United States	752 (33.7%)
Canada	210 (9.4%)
China	209 (9.4%)
Japan	201 (9.0%)
United Kingdom	165 (7.4%)
Other countries	694 (31.1%)

Safety signal detection results

Among SOCs, injury, poisoning and procedural complications ranked first in terms of the number of reports (1,176 cases), followed by infections and infestations (1076 cases), general disorders and administration site conditions (932 cases) (Figure 1D, Supplementary Table 3). Regarding signal strength, immune system disorders had the highest ROR (4.31), followed by infections and infestations (ROR = 3.11) and blood and lymphatic system disorders (ROR = 2.68) (Figure 1E, Supplementary Table 3).

A total of 149 positive PTs were identified using a combination of four disproportionality algorithms (Figure 1F, Supplementary Table 4). The most frequently reported PTs included off‑label use (868 cases, ROR = 5.9, 95% CI: 5.49–6.34), hypogammaglobulinaemia (122 cases, ROR = 71.04, 95% CI: 58.24–86.66), drug effective for unapproved indication (100 cases, ROR = 57.92, 95% CI: 46.7–71.85), intentional product use issue (86 cases, ROR = 15.89, 95% CI: 12.77–19.76), and infusion related reaction (62 cases, ROR = 6.92, 95% CI: 5.38–8.91) (Table 2). Some frequently reported PTs, such as off-label use, drug effective for unapproved indication, and intentional product use issue, represent medication-use characteristics rather than biological adverse reactions. Therefore, these terms were considered separately from clinical adverse reaction signals during interpretation. In general, a higher ROR indicates a stronger association with rituximab. ROR is a statistical metric for disproportionality analysis. A higher ROR value indicates a stronger reported-level association of the AE within rituximab-related reports. Nevertheless, this metric is only intended for initial safety signal screening and cannot establish a causal drug–event relationship. In the present study, the top 30 PTs ranked by ROR were primarily analyzed (Figure 2). Recurrence of primary mediastinal large B‑cell lymphoma occurred in 3 cases and had the highest ROR (888.92, 95% CI: 92.45–8547.09). Other notable PTs included B‑lymphocyte abnormality (5 cases, ROR = 211.71, 95% CI: 67.18–667.23), paraneoplastic pemphigus (3 cases, ROR = 177.78, 95% CI: 42.48–744.08), CD19 lymphocyte decrease (4 cases, ROR = 169.34, 95% CI: 49.56–578.63), and B‑lymphocyte count decrease (22 cases, ROR = 118.86, 95% CI: 72.45–194.99). Notably, using the spontaneous reporting database, this study identified several potential safety signals not listed in the rituximab prescribing information, including polyomavirus-associated nephropathy (11 cases; ROR = 9.09, 95% CI: 4.99–16.57) and smooth muscle cell neoplasm (6 cases; ROR = 12.02, 95% CI: 5.31–27.19). Given the small absolute number of reported cases, these findings should be considered exploratory hypotheses only and require validation in prospective clinical cohort studies.

Table 2.

The top 30 adverse event signals for rituximab by case report count.

PT	N	ROR (95%CI)	PRR (χ2)	EBGM (EBGM05)	IC (IC025)
Off label use	868	5.9 (5.49–6.34)	5.26 (3018.3)	5.19 (4.83)	2.37 (2.26)
Hypogammaglobulinaemia	122	71.04 (58.24–86.66)	69.75 (6693.96)	56.65 (46.44)	5.82 (5)
Drug effective for unapproved indication	100	57.92 (46.7–71.85)	57.07 (4619.92)	48.01 (38.7)	5.59 (4.72)
Intentional product use issue	86	15.89 (12.77–19.76)	15.69 (1124.57)	14.95 (12.02)	3.9 (3.37)
Infusion related reaction	62	6.92 (5.38–8.91)	6.87 (304.17)	6.73 (5.23)	2.75 (2.26)
Drug ineffective for unapproved indication	56	5.89 (4.51–7.68)	5.84 (220.83)	5.75 (4.41)	2.52 (2.02)
Agranulocytosis	52	23.58 (17.76–31.31)	23.41 (1034.01)	21.76 (16.39)	4.44 (3.55)
Neutropenia	52	2.98 (2.26–3.92)	2.96 (67.05)	2.94 (2.24)	1.56 (1.1)
Anaphylactic reaction	44	4.08 (3.03–5.5)	4.06 (100.21)	4.02 (2.98)	2.01 (1.48)
Mucosal inflammation	42	5.38 (3.96–7.3)	5.35 (146.1)	5.27 (3.88)	2.4 (1.82)
Infection	42	3.73 (2.75–5.06)	3.71 (82.21)	3.68 (2.71)	1.88 (1.35)
Cytomegalovirus infection	40	8.21 (5.99–11.25)	8.16 (244.92)	7.97 (5.82)	3 (2.31)
Covid-19	40	4.76 (3.48–6.51)	4.74 (116.21)	4.68 (3.42)	2.23 (1.65)
Serum sickness	39	54.8 (38.91–77.17)	54.48 (1729.68)	46.17 (32.79)	5.53 (3.94)
Sepsis	39	3.26 (2.37–4.47)	3.24 (59.92)	3.22 (2.34)	1.69 (1.15)
Respiratory failure	38	2.79 (2.03–3.85)	2.78 (43.12)	2.77 (2.01)	1.47 (0.94)
Multiple organ dysfunction syndrome	34	3.59 (2.56–5.04)	3.58 (62.47)	3.55 (2.53)	1.83 (1.23)
Disease progression	31	3.69 (2.59–5.26)	3.68 (59.75)	3.64 (2.56)	1.87 (1.24)
Pneumocystis jirovecii pneumonia	29	21.84 (14.96–31.86)	21.74 (534.78)	20.33 (13.93)	4.35 (3.08)
Epstein-barr virus infection	29	8.3 (5.73–12.01)	8.27 (180.31)	8.07 (5.57)	3.01 (2.17)
Nephrotic syndrome	28	14.81 (10.13–21.67)	14.76 (342.13)	14.1 (9.64)	3.82 (2.73)
Treatment failure	28	3.77 (2.6–5.48)	3.76 (56.06)	3.72 (2.56)	1.9 (1.23)
Post transplant lymphoproliferative disorder	26	11.78 (7.95–17.45)	11.74 (245.76)	11.33 (7.65)	3.5 (2.47)
Septic shock	26	4.33 (2.94–6.38)	4.32 (65.36)	4.27 (2.9)	2.09 (1.37)
Chills	23	3.21 (2.13–4.85)	3.21 (34.6)	3.18 (2.11)	1.67 (0.95)
B-lymphocyte count decreased	22	118.86 (72.45–194.99)	118.47 (1830.43)	84.91 (51.76)	6.41 (3.5)
Myelosuppression	20	4.13 (2.66–6.43)	4.12 (46.73)	4.08 (2.62)	2.03 (1.2)
Staphylococcal infection	19	3.93 (2.5–6.18)	3.92 (40.84)	3.88 (2.47)	1.96 (1.11)
Blood immunoglobulin G decreased	18	44.92 (27.35–73.78)	44.8 (669.56)	39.04 (23.77)	5.29 (2.99)
Lymphopenia	17	7.7 (4.76–12.48)	7.69 (96.42)	7.52 (4.64)	2.91 (1.77)

PT, preferred terms; ROR, reporting odds ratio; PRR, proportional reporting ratio; BCPNN, Bayesian confidence propagation neural network; MGPS, multi-item gamma poisson shrinker; EBGM, empirical Bayesian geometric mean; CI, confidence interval; χ2, chi-square; IC, information component; IC025, the lower limit of the 95% one-sided confidence interval for IC; EBGM05, the lower limit of the 95% CI for EBGM.

Figure 2.

Disproportionality estimates for the top 30 preferred terms based on ROR. PT, preferred terms; SOC, system organ class; ROR, reporting odds ratio; CI, confidence interval.

Stratified analysis

In all age-and-sex subgroups of children treated with rituximab, safety signals were identified, including off-label use, infusion-related reactions, serum sickness, hypogammaglobulinaemia, nephrotic syndrome, cytomegalovirus infection, agranulocytosis, post-transplant lymphoproliferative disorder, decreased B-lymphocyte count, and decreased immunoglobulin G (Figure 3A–D).

Figure 3.

Stratified analysis of rituximab-associated adverse event signals: (A) Gender subgroup; (B) 0–5 y subgroup; (C) 6–11 y subgroup; (D) 12–17 y subgroup; (E) Autoimmune disease subgroup; (F) Malignant neoplasm subgroup.

Sex-stratified analysis showed that in male pediatric patients, the strongest signal was immunodeficiency common variable (ROR = 237.08, 95% CI: 53.04–1059.74), followed by B-lymphocyte abnormalities (ROR = 210.80, 95% CI: 59.46–747.39) (Figure 3A, Supplementary Table 5). In female pediatric patients, the AE with the highest signal strength was primary mediastinal large B-cell lymphoma recurrent (ROR = 833.56, 95% CI: 86.68–8015.86), followed by granulomatosis with polyangiitis (ROR = 108.18, 95% CI: 45.15–259.18) (Figure 3A, Supplementary Table 6).

The results of age-stratified analyses are shown in Figure 3B–D and Supplementary Table 7–9. In children aged 0–5 y, decreased CD19 lymphocytes showed the strongest signal (ROR = 1363.60, 95% CI: 152.33–12,206.55), with decreased blood immunoglobulin A ranking second (ROR = 210.26, 95% CI: 87.04–507.92). In children aged 6–11 y, decreased B-lymphocyte count was the leading high-risk signal (ROR = 104.81, 95% CI: 48.43–226.81), followed by X-linked lymphoproliferative syndrome (ROR = 89.09, 95% CI: 32.34–245.37). In the 12–17-year-old subgroup, decreased B-lymphocyte count exhibited the strongest signal (ROR = 144.45, 95% CI: 36.11–577.86), followed by granulomatosis with polyangiitis (ROR = 130.26, 95% CI: 59.26–286.30).

We further performed stratified analyses by indication (malignant tumors and autoimmune diseases) among pediatric reports of rituximab use. The results showed that in children with autoimmune diseases, the AE with the highest signal strength was Castleman’s disease (ROR = 161.05, 95% CI: 18.00–1441.22), followed by B-lymphocyte abnormalities (ROR = 100.68, 95% CI: 19.53–519.07) (Figure 3E, Supplementary Table 10). In children with malignant tumors, the AE with the highest signal strength was Castleman’s disease (ROR = 391.38, 95% CI: 45.69–3352.46), followed by drug effective for unapproved indication (ROR = 45.97, 95% CI: 23.72–89.06) (Figure 3F, Supplementary Table 11).

Time‑to‑onset analysis

Precise TTO information was available for 466 patients, with a median TTO of 7 d (95% CI: 4–10 d). Most adverse events (324/466, 69.5%) occurred within the first 30 d after rituximab initiation. Lower proportions were observed during the intermediate periods, with 5.2% (24/466) occurring at 31–60 d and 3.9% (18/466) at 61–90 d, whereas 7.5% (35/466) of events were reported after 1 y (Figure 4A–C). Stratified Kaplan–Meier analysis with log‑rank testing showed no statistically significant differences in incubation period distributions among any of the subgroups analyzed, including AE severity, sex, and age (Figure 4D–F).

Figure 4.

(A) TTO distribution of rituximab-associated AEs; (B) Log-rank test comparing TTO distributions stratified by sex, age, and severity; (C) Overall Kaplan–Meier curve depicting the time-to-onset distribution of reported AEs among cases with evaluable dates; (D–F) Kaplan–Meier curves stratified by sex (D), age (E), and severity (F), depicting the time-to-onset distribution of reported AEs among cases with complete temporal information; (G-H) Weibull distribution fitting for the top 3 preferred terms (PTs) (G) and top 3 system organ classes (SOCs) (H). TTO, time-to-onset; PT, preferred term; SOC, system organ class; IQR, interquartile range; CI, confidence interval.

In the temporal pattern analysis, we first screened SOCs and PTs with complete time‑to‑onset (TTO) information from Supplementary Table 2 and 3, and then selected the top three SOCs and the top three PTs, respectively. The TTO data of these three SOCs and three PTs were finally fitted to a Weibull distribution model to characterize the temporal occurrence pattern of rituximab‑related AE reports.

At the PT level, hypogammaglobulinaemia (β = 0.94, 95% CI: 0.62–1.26), agranulocytosis (β = 1.07, 95% CI: 0.74–1.40), and neutropenia (β = 0.88, 95% CI: 0.34–1.42) demonstrated random failure patterns (Figure 4G). This observation indicates that the reporting risk of these adverse events remained relatively stable throughout the observation period. At the SOC level, infections and infestations, general disorders and administration site conditions, and respiratory, thoracic and mediastinal disorders exhibited early failure patterns (Figure 4H). These events were therefore more likely to be reported early during follow‑up. Nevertheless, these findings should be interpreted strictly as the observed temporal pattern of reported onset times derived from spontaneous reporting data, rather than a direct reflection of the true underlying temporal risk profile for these adverse events.

Discussion

Rituximab exerts effective immunomodulatory and antitumor effects through targeted depletion of CD20-positive B cells and has significantly improved therapeutic outcomes in various pediatric immune-mediated diseases and hematological malignancies. However, with the expanding clinical application of rituximab in pediatric populations, the gradual extension of its indications, and improvements in pharmacovigilance surveillance, the number of pediatric adverse event reports associated with rituximab in the FAERS database has shown an increasing trend. It should be noted that an increase in the number of reported events does not necessarily indicate an elevation in the actual risk associated with the drug; rather, it may reflect multiple contributing factors, including increased drug exposure, heightened clinical awareness, and changes in reporting practices. From a pharmacological perspective, depletion of CD20-positive B cells, while representing the fundamental mechanism underlying the therapeutic efficacy of rituximab, may also alter immune regulation and consequently increase the risk of infections, hypogammaglobulinaemia, and other immune-related adverse events.¹³ Therefore, a systematic evaluation of the safety profile of rituximab in pediatric real-world settings is of considerable clinical importance. In this study, we conducted a comprehensive pharmacovigilance analysis of rituximab-associated adverse events in pediatric patients using the FAERS database to characterize the real-world adverse event spectrum and identify potential safety signals.

The high‑frequency signals identified in this study (hypogammaglobulinaemia, neutropenia) have plausible biological explanations and are related to the immune‑response cascade secondary to rituximab‑mediated CD20-positive B‑cell depletion. By contrast, high‑ROR signals such as recurrent primary mediastinal large B‑cell lymphoma and paraneoplastic pemphigus suggest more complex drug‑disease interaction mechanisms. Importantly, high ROR signals derived from disproportionality analysis of the FAERS database do not equate to direct drug‑induced adverse reactions; statistical artifacts caused by indication‑related confounding and disease relapse must be cautioned against. High‑signal PTs including recurrent primary mediastinal large B‑cell lymphoma and Castleman’s disease in the present study should not be directly interpreted as rituximab‑related toxic endpoints. Recurrent primary mediastinal large B-cell lymphoma (PMBCL) is more likely to represent treatment failure or relapse of the underlying malignancy, reflecting disease progression rather than a direct drug-induced adverse effect. Castleman’s disease, in contrast, is an established therapeutic indication for rituximab, particularly in pediatric populations; therefore, reports involving this condition may largely reflect the underlying disease being treated, resulting in substantial indication bias and potentially inflated ROR estimates. Paraneoplastic pemphigus represents a more complex scenario, as it may occur as a tumor-associated autoimmune manifestation. Meanwhile, emerging evidence suggests that anti-CD20 monoclonal antibody therapy may alter B-cell-mediated immune regulation and, in rare cases, contribute to paradoxical activation or exacerbation of autoimmune bullous disorders.¹⁴ Therefore, this signal should be interpreted cautiously as a potential drug–disease interaction rather than being considered evidence of direct drug toxicity.

Rituximab specifically binds to the CD20 antigen on B‑cell surfaces, activating complement‑dependent cytotoxicity and antibody‑dependent cell‑mediated cytotoxicity, which leads to rapid B‑cell depletion.¹⁵ The duration of B-cell depletion may extend beyond the period required for immune recovery in some pediatric patients, which may contribute to the development of hypogammaglobulinemia following rituximab treatment.¹⁶ Notably, previous studies have indicated that children with pre-existing hypogammaglobulinemia before rituximab treatment may have an increased risk of persistent or new-onset hypogammaglobulinemia. Therefore, baseline immunoglobulin assessment and longitudinal monitoring during therapy are of clinical importance.¹⁷ Furthermore, a nationwide survey of children with nephrotic syndrome suggested that agranulocytosis may represent an independent risk factor for severe infection among patients with hypogammaglobulinemia. The coexistence of hematologic toxicity and impaired humoral immunity may further increase susceptibility to serious infections.¹⁸ Notably, the development of anti‑drug antibodies can markedly accelerate rituximab clearance and inhibit B‑cell depletion, potentially leading to poor treatment response or disease relapse. This phenomenon is particularly evident in pediatric patients with systemic lupus erythematosus and offers a plausible pharmacokinetic rationale for some of the tumor relapse signals observed in our study.¹⁹ A review of the literature indicates that while rituximab can induce remission in pediatric pemphigus vulgaris, as an anti‑CD20 monoclonal antibody it may paradoxically trigger or exacerbate certain autoimmune bullous diseases by disrupting regulatory B‑cell function or inducing a cytokine storm.²⁰ This suggests that drug intervention may modulate the presentation of autoantigens or disrupt the body’s pre-existing immune tolerance.²¹

The boxed warning for rituximab already lists progressive multifocal leukoencephalopathy (PML) as a serious AE. PML results from JC virus reactivation, driven by B‑cell depletion‑induced cellular immune deficiency that permits latent virus reactivation.^22,23 Within this immunosuppressive context, our analysis identified two previously unrecognized safety signals that are not currently listed in the rituximab prescribing information. The first was polyomavirus-associated nephropathy, a finding that is biologically plausible given that B-cell depletion may impair immune surveillance and potentially contribute to viral reactivation or reduced viral control.²⁴ Second, smooth muscle cell tumors, a rare Epstein–Barr virus (EBV)-associated malignancy observed in immunocompromised patients, were also identified as a potential safety signal. Rituximab-induced B-cell depletion may theoretically impair EBV-specific immune control, thereby facilitating EBV reactivation and contributing to tumor development. EBV-driven oncogenic signaling pathways, including MAPK and PI3K/Akt pathways, have been implicated in the pathogenesis of EBV-associated tumors.²⁵ Although rare, these two complications indicate that beyond known PML, other diseases related to latent virus reactivation warrant high clinical vigilance in the setting of immune homeostasis disruption caused by B‑cell‑targeted therapy.²⁶ Furthermore, it should be emphasized that rituximab is frequently administered in combination with various immunosuppressive regimens.² Concomitant immunosuppressive agents represent a critical confounding factor, as they themselves can impair host immune surveillance, trigger reactivation of latent viruses and subsequent virus‑associated neoplastic complications. Since spontaneous reporting data cannot disentangle the individual effects of different drugs, polyomavirus‑associated nephropathy and EBV‑related smooth‑muscle cell neoplasm cannot be fully attributed to CD20‑positive B‑cell depletion mediated by rituximab.

In this study, strong ROR signals unique to male pediatric patients, such as common variable immunodeficiency, may be associated with lower functional reserves of X‑linked immunoregulatory genes (e.g., BTK or CD40L), thereby hindering the restoration of normal immunoglobulin levels after B‑cell depletion.^27,28 In contrast, recurrence of primary mediastinal large B-cell lymphoma (PMBCL) showed a stronger reporting signal among female pediatric patients in our analysis. This finding should be interpreted cautiously because FAERS cannot determine true relapse incidence or establish causal relationships. Nevertheless, PMBCL has a recognized female predominance, and recent studies suggest that sex-related molecular differences, including variations in gene expression and hormonal regulation pathways, may influence lymphoma biology and treatment response.^29–31 In addition, CD20 expression is a key determinant of rituximab efficacy, and alterations in CD20 expression or B-cell phenotype during disease evolution may affect sensitivity to anti-CD20 therapy.³² Therefore, the observed sex-specific reporting pattern may reflect the combined effects of disease biology, molecular heterogeneity, and treatment-related factors, which require further validation. Consistent with adult data, females exhibit a higher baseline prevalence of autoimmune diseases such as pemphigus and systemic lupus erythematosus, and they also show better treatment responses to rituximab, which may be a major contributor to the observed sex differences.^33–35 Moreover, in the context of malignancies, sex‑related differences in the tumor microenvironment may also contribute to divergent relapse patterns.³⁶ Other studies have shown that B‑cell depletion therapy can delay the progression of organ damage in female‑predominant diseases such as systemic lupus erythematosus, but the impact on long‑term immune reconstitution varies significantly between sexes.³⁷

This study also identified age‑specific AE patterns in pediatric patients receiving rituximab: the 12–17‑year‑old adolescent group had the highest number of reports, while the 0–5-year-old group exhibited distinct hematologic AE reporting patterns. These differences profoundly reflect the impact of developmental stage‑specific immune remodeling. Previous pediatric and young adult cohort studies have shown that rituximab‑induced B‑cell depletion can last for more than one year, and the recovery of memory B cells (CD27+) lags significantly behind that of naive B cells, resulting in long‑term humoral immune deficiency.⁴ The unique CD19-positive lymphocyte reduction signal observed in children aged 0–5 y may reflect immature B-cell development and limited immune reconstitution capacity. Consistently, studies in pediatric nephrotic syndrome have shown that younger children receiving rituximab are at higher risk of hypogammaglobulinemia, infection, and neutropenia, indicating that age is an important factor influencing treatment safety. Therefore, closer immune monitoring is particularly warranted in younger patients.³⁸ The higher frequency of AE reports in the adolescent group may be related not only to the higher prevalence of autoimmune diseases and lymphomas in this age range, but also to differences in clinical exposure patterns and reporting behaviors. The age‑specific differences in AEs of rituximab in the pediatric population essentially result from the interplay between the drug and the structure, function, and compensatory capacity of the immune system at different developmental stages. Although this study identified sex‑ and age‑specific characteristics of AEs, the inherent limitations of FAERS data mean that these findings require further validation in cohort studies with larger sample sizes and longer follow‑up periods.

Stratified analysis by indication suggested that the safety profile of rituximab may vary according to underlying disease conditions. In children with autoimmune diseases, signals related to immunoglobulin reduction and hypogammaglobulinemia were more prominent, which may reflect the vulnerability of humoral immune recovery in patients with pre-existing immune dysregulation. Previous studies have suggested that baseline immunoglobulin levels, age, and cumulative rituximab exposure may influence the risk of persistent hypogammaglobulinemia after treatment in pediatric autoimmune diseases.³⁹ In contrast, children with malignancies showed more signals associated with opportunistic infections and virus-related complications, including CMV infection and EBV-related lymphoproliferative disorders. These findings may be partially explained by the combined effects of underlying malignancy, intensive immunosuppressive therapy, and impaired immune surveillance after B-cell depletion.⁴⁰ Notably, signals such as Castleman disease and lymphoproliferative disorders should be interpreted cautiously, as they may largely reflect treatment indications rather than direct drug-related toxicity. Overall, these findings highlight the importance of considering disease background and concurrent therapies when evaluating rituximab-associated safety risks in pediatric populations.

Our study demonstrated that most reported AEs occurred within 30 d after drug administration, with a median time‑to‑onset of 7 d, characterized by an “early failure” distribution pattern. This distribution may be related to the pathophysiology of infusion‑related reactions and acute cytokine release syndrome. As rituximab is a chimeric monoclonal antibody, its xenogeneic protein components can trigger type I hypersensitivity reactions upon first infusion, or induce cytokine release via Fc cross‑linking, leading to acute inflammatory reactions.⁴¹ This pattern is consistent with the occurrence of allergic reactions and infusion-related reactions early after drug initiation. Previous reports have described cases of rituximab-associated anaphylactic shock in pediatric patients, with some cases accompanied by elevated serum tryptase levels, suggesting mast cell activation and the possible involvement of IgE-mediated or non-IgE-mediated immediate hypersensitivity mechanisms.⁴² Furthermore, real-world studies of biosimilars have shown that infusion-related reactions, including bronchospasm and injection site reactions, tend to occur more frequently during the initial treatment phase, with reported incidence generally decreasing over subsequent infusions. This pattern may reflect immune adaptation, treatment-related factors such as premedication, or reduced susceptibility after repeated exposure.⁴³ Unlike early acute reactions, infectious AEs exhibited a delayed or prolonged temporal pattern, which may be associated with impaired humoral immunity following B-cell depletion and delayed recovery of B-cell subsets. Notably, severe infections have been reported to occur months after treatment initiation, highlighting the potential for persistent immune alterations after rituximab therapy.^44,45 Overall, the temporal heterogeneity of AEs may reflect different dominant mechanisms ranging from acute immune activation to chronic immunosuppression, suggesting that clinical monitoring strategies may need to be adjusted according to the time window. Of particular note is the selection bias introduced by substantial missing timestamp data in this temporal analysis. Of the 2231 eligible reports, only 466 (20.9%) provided complete and valid dates for both drug initiation and adverse event onset, resulting in the exclusion of over 79% of cases from the TTO analysis. This substantial exclusion is non-random; while acute events such as infusion-related reactions are frequently documented with precise onset dates, delayed immune-related adverse events (e.g., late-onset neutropenia and hypogammaglobulinaemia occurring months post-infusion) are disproportionately subject to incomplete date reporting. Consequently, the estimated Weibull shape (β) and scale (α) parameters are susceptible to systematic bias and may not accurately reflect the true latency characteristics of rituximab-associated adverse events in real-world practice. Therefore, our findings are strictly descriptive of the subpopulation with complete timestamp data and cannot be generalized to the broader spectrum of rituximab events.

This study has several limitations. First, the data originate from a spontaneous reporting system, which is subject to inherent reporting biases. For example, serious AEs (e.g., death, hospitalization) are more likely to be reported, potentially leading to an overestimation of signal strength for certain AEs and an underestimation of mild or non‑specific events. Moreover, more than half of the reports lacked time‑to‑onset data, limiting the statistical power and generalizability of our temporal pattern analyses. Second, although we used a multi‑algorithm approach to enhance specificity, the stringent criterion requiring positivity across all four algorithms, while effective in controlling false positives, may have compromised sensitivity. Consequently, some true associations with small numbers of reports but potential clinical significance may have been missed, particularly in subgroup analyses with relatively limited sample sizes. Third, FAERS does not provide sufficient information for comprehensive adjustment of concomitant medications. Rituximab is frequently administered as part of combination regimens in pediatric oncology and immune-mediated diseases, including high-dose corticosteroids, cyclophosphamide, and other immunosuppressive agents. These background therapies may independently contribute to serious adverse outcomes, particularly infections, hematological abnormalities, and immune-related complications. Therefore, although the identified signals may reflect potential safety concerns associated with rituximab, residual confounding from concomitant medications cannot be completely excluded.

Conclusion

Using rituximab-associated AE reports involving pediatric patients from the FAERS database, this study evaluated the drug’s safety profile in this population. The results demonstrate that rituximab-associated adverse event reports involved multiple organ systems, showed sex- and age-specific differences in reporting patterns, and also identified several new potential positive signals. Given the limitations of this study, clinical use of rituximab should be guided by individualized monitoring based on patient characteristics, and its comprehensive safety profile requires further validation through further investigation.

Biographies

Wenjun Fang, Deputy Chief Pharmacist, currently serves as the Director of the Outpatient Pharmacy at Yancheng Maternal and Child Health Hospital. He is also Vice Chairperson of the Yancheng Committee on Centralized Intravenous Medication Preparation, a member of the Expert Database for Non-Expanded Program on Immunization (Non-EPI) Vaccines in Yancheng, and a member of the Expert Database of Jiangsu Yancheng Technician College. He previously completed advanced training at Yancheng First People’s Hospital and the First Affiliated Hospital of Soochow University. His main research interests include rational drug use and intravenous medication preparation.

Ben Liu, an attending physician, holds a Master’s degree from Nanjing University Medical School and currently works in the Pediatric Intensive Care Unit at Yancheng First People’s Hospital. He serves on the committees of the Critical Care Group and the Neurology Group within the Pediatric Branch of the Yancheng Medical Association and is a committee member of the Yancheng Nutrition Society. His main research interests include pediatric critical illnesses and neurophysiology, with expertise in bibliometric, database, Mendelian randomization, and bioinformatics analyses.

Li Cai, Associate Chief Pharmacist, earned a Master’s degree from Nantong University. She currently serves as the Director of the Pharmacy Department at the Children’s Hospital Campus of Yancheng First People’s Hospital. With extensive experience in clinical pharmacy management and rational drug use, she has deep expertise in the standardized use of antineoplastic drugs, clinical application of biological agents, and medication safety management. She possesses a strong theoretical foundation in pharmacy and extensive practical experience in clinical pharmacy administration. She is committed to advancing precision medicine and improving the quality of pharmaceutical care.

Funding Statement

This work was supported by the Research Project of Yancheng Municipal Health Commission in 2025 [YK2025004; YK2025044].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data presented in this study are available on request from the corresponding author.

Ethics statement

This study involves a secondary analysis of publicly accessible summary statistics and does not necessitate explicit ethical approval.

Supplemental material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21645515.2026.2731825.