Introduction
Rituximab, a chimeric anti-CD20 monoclonal antibody, has become an established therapeutic option for children with difficult steroid-dependent and frequently relapsing nephrotic syndrome (SDNS/FRNS). Its efficacy in prolonging remission and reducing cumulative steroid exposure has led to endorsement in recent international guidelines, including those from Kidney Disease: Improving Global Outcome (KDIGO) and International Pediatric Nephrology Association (IPNA).1 2 With increasing use, attention has shifted towards its long-term safety and immunological consequences.
Hypogammaglobulinaemia is a recognised adverse effect of rituximab, resulting from B cell depletion and impaired immunoglobulin production.3 While immunoglobulin G (IgG) reduction is often transient and expected to recover following B cell reconstitution, a subset of patients develops persistent hypogammaglobulinaemia (PH), characterised by sustained low IgG levels beyond the anticipated recovery period. In other paediatric and adult autoimmune conditions, PH has been associated with increased infectious morbidity. However, in children with nephrotic syndrome (NS), reported prevalence varies widely and the clinical significance remains uncertain.
Most available data originate from high-income countries, where structured immunological monitoring and access to immunoglobulin replacement therapy are readily available. In low- and middle-income countries, where infectious disease burden is higher and laboratory surveillance may be limited, the implications of persistent IgG suppression may be more consequential. Furthermore, younger children—whose B cell compartment is still maturing—may be particularly vulnerable to sustained immunoglobulin depletion following rituximab exposure.
Data from resource-constrained environments evaluating long-term IgG recovery and its clinical impact are scarce.4 We aimed to examine the prevalence of hypogammaglobulinaemia detected ≥12 months after rituximab exposure and to identify associated clinical factors in children with NS treated at two tertiary care paediatric nephrology centres in India. We also explored the association between hypogammaglobulinaemia and infection-related hospitalisation. The results may help to refine treatment strategies, mitigate risks and aid shared decision-making with families.5
Method
This was a multicentre cross-sectional observational study evaluating a single serum total IgG measured at least 12 months post-rituximab in children with FRNS/SDNS. The study was not designed to assess longitudinal IgG recovery or persistence over time. All children visiting the Department of Paediatrics between May 2023 and January 2025 and fulfilling inclusion criteria were recruited.
Inclusion criteria were children <18 years with difficult to treat FRNS/SDNS who had previously received one or more doses of rituximab infusion at an interval of ≥12 months between the last rituximab dose and the study assessment. Excluded were children with history of steroid resistant NS as defined by KDIGO.1 Difficult to treat steroid sensitive NS were defined as per Indian Society of Pediatric Nephrology (ISPN) 2021 guideline.6
Primary objective of the study was to determine the prevalence of PH. Hypogammaglobulinaemia was defined as serum IgG level of <2 SD below the mean for age: 0–2 years: IgG<350 mg/dL; 2–6 years: IgG<400 mg/dL; >6 years: IgG<500 mg/dL. Severe hypogammaglobulinaemia was analysed using two thresholds: IgG<250 mg/dL7 or less than 100 mg/dL.4 PH was defined as hypogammaglobulinaemia detected ≥12 months after last dose of rituximab.
At both the institutions, the usual protocol for each course of intravenous rituximab was 375 mg/m2 for two doses 1 week apart followed up with a steroid-sparing immunosuppressant. The standard post-rituximab immunosuppressant has been mycophenolate mofetil (MMF), although it is modified based on physician preference or patient response. Pre-rituximab assessment of IgG levels was started routinely in the institutes from 2022, while previously, it was restricted to those receiving sequential rituximab or having recurrent significant infection post-rituximab.
The study was initiated after approval from institutional ethical committees (reference numbers: Institute of Child Health, Kolkata–ICH/IECBMHR/17/2023 and All India Institute of Medical Sciences, Kalyani: IEC/AIIMS/Kalyani/Certificate/2026/145) on submission of predefined study protocol (online supplemental file 1). Parental signed assent/consent was obtained allowing blood test for IgG measurement and data collection. For ethics-related queries, the Institutional Ethics Committee, Institute of Child Health, Kolkata, can be contacted at instecich@gmail.com.
Patients recruited were confirmed to be in NS remission by fulfilling all three criteria by (a) reviewing nephrotic diary (urinary dipstick negative/trace for albumin for the last 3 days), (b) confirming with a negative urine dipstick in the clinic and (c) checking serum albumin level. Those with urine albumin dipstick ≥1+ and/or serum albumin <3.5 g/dL in clinic were excluded, managed as per usual clinical care and recalled for taking part in the study once NS remission was confirmed and serum albumin had normalised. Only thereafter blood for total IgG levels were taken.
IgG levels were measured by nephelometry, a method using the light scattering properties of antigen-antibody complexes.
Clinical data collected included patient demographics, age of onset of NS, age of initiation of rituximab, number and timing of repeat courses of rituximab, number of relapses post-rituximab, use of additional immunosuppressant post-rituximab and infection-related hospitalisations were retrospectively ascertained from available hospital records. Events were considered infection-related hospitalisations when infection severity required hospital admission and intravenous antibiotics and/or paediatric intensive care unit (PICU) admission. No predefined prospective infection-surveillance protocol was used, as infection surveillance was beyond the scope of the present study.
Data of recruited patients were filled in predefined case record form (online supplemental file 2).
Sample size calculation
A previous study from a high resource country estimated prevalence of PH to be around 22%.7 To conduct a comparable study with 95% CI and 10% precision, we estimated that a minimum cohort size of 66 patients was required.8
Statistical analysis
The data were entered in MS Excel sheet, and statistical analysis was carried out in SPSS software V.31. The normality of the data was assessed by Shapiro-Wilk test. Continuous data were presented as median (IQR), while categorical data as percentages.
Primary outcome was a proportion of children with PH following rituximab therapy. The categorical variables between the PH and normal immunoglobulin group were compared using χ2 or Fisher exact test, while continuous variables were compared using Mann-Whitney U test.
A binary multivariate logistic regression model was conducted to identify independent predictors of PH along with their ORs and 95% CIs. Overfitting and testing for predictors that could be related derivatives were tested using parsimonious logistic regression modelling. Collinearity was assessed by examining variance inflation factors, tolerance values, condition indices and Eigenvalues in SPSS V.31.
We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline9 to draft this manuscript, and the STROBE reporting checklist10 when editing, included in online supplemental file 3.
Patient and public involvement
Patients and the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Results
Baseline characteristics
During the study period, 123 children met the inclusion criteria. Fourteen children were excluded (55% female, median age: 8.9 IQR 7.5 to 10.2 years, median follow-up post last dose of rituximab 18 IQR 14 to 28 months) as family refused consent for total IgG sampling. Therefore, 109 children were included in the study (44% female, median age: 10.2 years with an IQR: 7–12.3 years).
At the time of recruitment, all were confirmed to be in remission with median time since last relapse of 32 weeks (IQR: 18–50) and normal serum albumin level (median 4.2 g/dL; IQR: 3.9–4.5 g/dL). Participants last received rituximab at median 24 months previously (IQR: 14–30), and CD19 counts of all had normalised previously at a median of 7 (IQR:6-8) months after the last dose of rituximab.
Prevalence of persistent hypogammaglobulinaemia
PH was observed in 27/109 (25%) children, but severe PH (<100 mg/dL) was observed in only one child (0.9%) while <250 mg/dL in 3 (2.7%) children. There were only two children who were known to have low IgG levels prior to the study and for the rest PH was discovered only via the study. Table 1 compares different demographic and clinical parameters between the PH (n=27) and the normal-IgG (n=82) group.
In univariate analysis, compared with children with normal IgG, those with PH were younger both at first dose of rituximab (median (IQR): 4.5 (3.1–7) vs 8.3 (5.9–10) years; p<0.001) and at last dose of rituximab (5.5 (4–8.6) vs 8.5 (6–10.6) years; p=0.005). They were more likely to have received multiple (≥2) rituximab courses (13 (48%) vs 14 (17%); p=0.001); had PH despite longer time having elapsed since last rituximab dose (2.1 (1.5–3.2) vs 1.6 (1.2–2.1) years; p=0.021) and had lower serum albumin at time of measuring IgG (although all within normal range) (3.9 (3.5–4.2) vs 4.2 (4–4.5) g/dL; p=0.001). Among our sequential rituximab cohort, only two children received rituximab even when they were having hypogammaglobulinaemia.
Both of them received a dose of intravenous immunoglobulin prior to rituximab, and at the time of undergoing this particular study, both of them had normal total IgG. Age at first rituximab and number of rituximab courses demonstrated no evidence of collinearity (Variance Inflation Factor 1.004; tolerance 0.996; maximum condition index 4.95).
A multivariable binary logistic regression model was constructed to examine the association between demographic and treatment-related variables and the likelihood of PH (Table 2). We excluded age of last rituximab and age at time of study in the model, as they closely approximate derivatives of the other variables. Younger age at first dose of rituximab and low serum albumin at time of measuring IgG remained as statistically significant independent predictors.
PH decreased steadily with age at first rituximab from 10/14 (71%) at age 2 to 3.9 years to 6/50 (12%) for aged 8 years and over (p=0.005, figure 1).
To assess the robustness of the multivariable logistic regression findings and to allay concerns of overfitting or that predictors could be related derivatives (eg, age at diagnosis and age of rituximab), we constructed a clinically parsimonious logistic regression model using a limited set of predictors with strong biological plausibility or prior evidence of association (online supplemental file 4). This model retained the key independent association between younger age at first rituximab and low serum albumin supporting the stability of the main findings.
Severe infections
Overall, 12% (13/109) of total cohort required hospitalisation due to severe infections, all received antibiotics for at least 7 days with 10 children among them requiring PICU admission. Underlying infective aetiology was identified in 12 children (three chicken pox with superadded pneumonia, five spontaneous bacterial peritonitis with strep pneumococcus identified in blood culture and four pyelonephritis with E. coli identified in three and Klebsiella in one). Most of the cohort (n=98) were immunised against Haemophilus influenzae, pneumococcus and hepatitis B, but varicella immunisation was up to date in only 28 children.
Percentage of children hospitalised due to infection and rate of hospitalisation per year was significantly higher in the PH group (table 1).
Discussion
In this cross-sectional cohort of children with difficult SDNS/FRNS evaluated at least 12 months after last rituximab exposure, we observed a 25% prevalence of PH. Younger age at first rituximab administration emerged as an independent predictor. PH was also associated with significantly higher infection-related hospitalisation rates; however, this association was based on a small number of retrospectively ascertained events and therefore should be considered exploratory.
Our findings provide data on the prevalence and correlates of low IgG detected at least 12 months after rituximab, but do not establish longitudinal persistence or causality.
As shown in online supplemental file 5, reported prevalence of post-rituximab PH in paediatric NS varies widely (6%–47%), reflecting heterogeneity in definitions, follow-up duration and cohort characteristics.7 11–15 Our estimate falls within this range but is notable for the ≥12 month threshold after rituximab and predominantly MMF-maintained cohort. Unlike some previous reports, we observed a clear association between PH and serious infections requiring hospitalisation, supporting the clinical relevance of sustained IgG suppression in this population; however, given the cross-sectional design, retrospective infection ascertainment and small number of events, this finding should be considered exploratory and requires confirmation in prospective longitudinal studies.
The period of remission post-rituximab is usually time limited, and different strategies are in use to try to maximise the duration of remission post-rituximab. Both continuation of other steroid-sparing immunosuppressants post-rituximab or repeated sequential doses of rituximab and weaning off all other immunosuppressants appear to be equally effective.16 We restricted our use of sequential rituximab to a selected cohort of very difficult SDNS with features of steroid toxicity. Sequential rituximab was administered only after CD19 normalisation and usually after documentation of normal IgG levels.
If total IgG was low, we usually waited to repeat the IgG level after a month or so, and in scenarios with PH but the child definitely needing a change in steroid-sparing agent, we usually switched to CNI. Although sequential rituximab was not an independent predictor in multivariable modelling in our cohort, children receiving multiple courses showed higher PH frequency in univariate analysis, suggesting cumulative exposure may still contribute. Both Chan et al14 and Sinha et al4 reported sustained remission with repeated rituximab dosing but also noted a heightened risk of hypogammaglobulinaemia although no significant increase in risk of serious infection requiring hospitalisation was explicitly stated.
The other strategy of continuation of steroid-sparing immunosuppressant (usually MMF) post-rituximab also appears effective.17–19 In our unit, we usually employ this strategy initially. Most children in our cohort continued MMF following rituximab. MMF can impair B cell proliferation through inosine monophosphate dehydrogenase inhibition and may theoretically contribute to reduced IgG levels.7 However, subgroup analysis was not feasible due to near-universal use; therefore, an independent MMF effect cannot be excluded.
Importantly, continuation of background immunosuppression reflects real-world practice in many centres.12 The ongoing randomised controlled trial by Basu et al20 comparing sequential rituximab with continuation of immunosuppressant post-rituximab may provide more data.
Although reports in children with NS remain limited, our finding of higher infection-related hospitalisation among children with hypogammaglobulinaemia is consistent with previous observations linking rituximab-associated hypogammaglobulinaemia with infectious morbidity.21 Hirano et al reported serious infections requiring hospitalisation in a proportion of children with post-rituximab hypogammaglobulinaemia.22 However, our study cannot establish that hypogammaglobulinaemia directly increases susceptibility to infection. The association may also reflect differences in cumulative immunosuppressive exposure, disease severity or other unmeasured confounders. Furthermore, only 13 infection-related hospitalisations occurred, and events were retrospectively ascertained without a predefined prospective surveillance protocol.
Therefore, the observed association should be considered exploratory and hypothesis-generating rather than confirmatory. Prospective longitudinal studies with systematic infection surveillance are required to determine whether hypogammaglobulinaemia independently contributes to serious infection risk. As rituximab primarily depletes B cells, post-rituximab hypogammaglobulinaemia was initially thought to be unlikely a major concern, and even if it happens, it was postulated to be transient with expected normalisation of IgG with B cell repopulation.
Unfortunately, evidence is currently mounting that in some children post-rituximab hypogammaglobulinaemia might not recover with CD19 normalisation, and this is supported by our data. In a recent retrospective study, Höppener et al23 observed that 51% of paediatric patients treated with rituximab developed PH lasting more than 6 months, and 6.7% remained hypogammaglobulinaemic even after 5 years. While total CD19+B cells reconstituted at a median of 11 months, the recovery of CD19+CD27+IgG+ switched memory B cells was significantly delayed, with a median time of 1.8 years.
Importantly, three patients developed class-switch recombination deficiencies and failed to achieve immunologic recovery. The study highlighted that recovery of total CD19+B cells may not necessarily coincide with recovery of switched memory B cells, suggesting that B cell subset assessment may provide additional information regarding immunological recovery after rituximab.
Similar to our study, increased incidence of post-rituximab PH has also been found in younger children in other reports.12 13 Age-related vulnerability appears biologically plausible.24–26 Younger children have an evolving B cell compartment characterised by a predominance of naïve B cells and relatively limited memory and long-lived plasma cell reserves. Rituximab-mediated CD20 depletion in this context may disproportionately affect immunoglobulin-producing capacity, with delayed recovery of switched memory B cells even after CD19 normalisation. Our finding of a progressive decline in PH prevalence with increasing age at first rituximab exposure supports this developmental susceptibility.
The association between lower recent albumin and hypogammaglobulinaemia should be interpreted cautiously. It may reflect greater cumulative protein loss associated with frequent relapses rather than a direct biological relationship.
Our study has several strengths, including a relatively large cohort, the use of age-specific IgG reference ranges and use of regression modelling to identify independent associated factors. Despite this, our result cannot be extrapolated to depict the true prevalence of post-rituximab PH among difficult NS across the board because of several important limitations. First, the cross-sectional design and absence of IgG longitudinal data precludes causal inference and may even introduce selection bias.
However, restricting inclusion to children ≥12 months post-rituximab reduces misclassification of transient hypogammaglobulinaemia. Second, baseline pre-rituximab IgG levels were unavailable for most participants, preventing exclusion of pre-existing hypogammaglobulinaemia. Third, event numbers were modest, and multivariable modelling should be interpreted cautiously.
We purposefully restricted our definition of infection to those requiring hospitalisation for easy verification through hospital records and reduction of recall bias, the small numbers and their retrospective ascertainment limited assessment of infection risk. Fourth, B cell subpopulation analysis was not available, limiting mechanistic insight. Finally, the predominance of MMF use restricts generalisability to cohorts where all immunosuppression is withdrawn post-rituximab.
Despite these limitations, our study provides clinically relevant data from a setting underrepresented in the literature. The identification of younger age at first rituximab exposure as a consistent risk factor supports cautious use in early childhood and highlights the importance of structured IgG monitoring beyond CD19 recovery. Prospective longitudinal studies incorporating baseline immunoglobulin assessment and B cell phenotyping are needed to better define recovery trajectories and guide management algorithms.
In conclusion, PH occurred in one-quarter of children evaluated ≥12 months after rituximab and was observed to be associated with younger age at exposure and increased infection-related hospitalisation. While these findings require prospective confirmations, until more definitive longitudinal data are available, careful patient selection and post-treatment immunologic surveillance should be considered, particularly in younger children.
Supplementary Material