The effect of rituximab therapy on immunoglobulin levels in patients with multisystem autoimmune disease

Background Rituximab is a B cell depleting anti-CD20 monoclonal antibody. CD20 is not expressed on mature plasma cells and accordingly rituximab does not have immediate effects on immunoglobulin levels. However, after rituximab some patients develop hypogammaglobulinaemia. Methods We performed a single centre retrospective review of 177 patients with multisystem autoimmune disease receiving rituximab between 2002 and 2010. The incidence, severity and complications of hypogammaglobulinaemia were investigated. Results Median rituximab dose was 6 g (1–20.2) and total follow-up was 8012 patient-months. At first rituximab, the proportion of patients with IgG <6 g/L was 13% and remained stable at 17% at 24 months and 14% at 60 months. Following rituximab, 61/177 patients (34%) had IgG <6 g/L for at least three consecutive months, of whom 7/177 (4%) had IgG <3 g/L. Low immunoglobulin levels were associated with higher glucocorticoid doses during follow up and there was a trend for median IgG levels to fall after ≥ 6 g rituximab. 45/115 (39%) with IgG ≥6 g/L versus 26/62 (42%) with IgG <6 g/L experienced severe infections (p = 0.750). 6/177 patients (3%) received intravenous immunoglobulin replacement therapy, all with IgG <5 g/L and recurrent infection. Conclusions In multi-system autoimmune disease, prior cyclophosphamide exposure and glucocorticoid therapy but not cumulative rituximab dose was associated with an increased incidence of hypogammaglobulinaemia. Severe infections were common but were not associated with immunoglobulin levels. Repeat dose rituximab therapy appears safe with judicious monitoring.

Rituximab is a chimeric murine/human monoclonal antibody that results in complete peripheral blood B cell depletion for variable time periods; typically 6-12 months. During B cell depletion, new vaccine responses are impaired and a theoretical risk of new infections exists [14]. However, since CD20 is highly expressed on B cells but not on stem cells or mature plasma cells, B cell regeneration from precursors is not directly compromised by rituximab and in the short term established humoral immunity is preserved; usually with stable levels of total IgG and established vaccination antibodies [14]. However, when rituximab is used in cohorts with high prior immunosuppressive exposure, where stem cell and plasma cell compartments maybe compromised, or when rituximab is administered long term using repeat dosing strategies resulting in prolonged B cell depletion, preexisting humoral immunity maybe impaired. Hypogammaglobulinaemia has occurred in more than 50% of non-Hodgkin's lymphoma (NHL) patients, especially in those receiving rituximab in combination with chemotherapy or bone marrow transplantation [15][16][17][18], and less so in RA patients treated with rituximab where 3.5% had IgG levels below the normal range [19]. Hypogammaglobulinaemia has also been reported in small cohorts with primary systemic vasculitis treated with rituximab [20][21][22][23], although the impact of prior high immunosuppression exposure, and repeat rituximab dosing, as is widely used in clinical practice, are unclear.
Immunoglobulin plays a major role in adaptive immunity, and severe depletion of immunoglobulin, as observed in primary immunodeficiency syndromes increases infection risk [24]. The potential development of secondary immunodeficiency due to immunosuppressive medication, including the impact of prolonged B cell depletion on IgG levels and infection risk, is not well studied in patients with autoimmune disease.
We report on the frequency and severity of hypogammaglobulinaemia, and associated infection rates in a large cohort of rituximab treated patients with severe multisystem autoimmune diseases and prolonged follow-up.

Methods
This was a retrospective study conducted at Addenbrooke's Hospital, Cambridge, UK, which serves as a tertiary referral clinic and follows approximately 1000 patients with primary systemic vasculitis and SLE. In accordance with UK National Health Service Research Ethics Committee guidelines, ethical approval and patient consent were not required for this work because it comprises retrospective data and all treatment decisions were made prior to our evaluation.
All patients with multi-system autoimmune disease treated with rituximab between 2002 and 2010 were studied. 104 cases have previously been reported [3,5,[25][26][27] and seven enrolled in a randomized controlled trial of rituximab [1] are also included in this cohort. Patients were excluded if they had fewer than six months followup or required repeated plasma exchange (PLEX).

Clinical and laboratory assessments
The data was collected retrospectively from patient notes and clinical databases. Data collection included pre-rituximab demographics, disease activity assessments, medications, adverse events and immunoglobulin levels at each clinical assessment.

Dose of rituximab
Rituximab induction therapy consisted of either 1000 mg repeated after two weeks or 375 mg/m 2 /week × 4. Patients received further rituximab at the time of relapse or at 6 monthly intervals as a remission maintenance therapy.

Disease activity
AAV disease activity was graded by the Disease Extent Index (DEI score) [28] and by investigators' assessment of disease activity as either full remission (DEI ≤2 and a steroid dose ≤10 mg daily), partial remission (greater than 50% reduction in DEI), ongoing disease/progressive disease (treatment failure), or relapse (increase in disease activity necessitating additional therapy beyond a temporary increase in glucocorticoid therapy).
SLE disease activity was graded by the BILAG disease activity (British Isles Lupus Assessment Group) and by investigators' assessment of disease activity as either full remission (absence of BILAG A, B and C), partial remission (absence of BILAG A and B), ongoing disease/ progressive disease (treatment failure), or relapse (increase in disease activity which required an increase in immunosuppressive therapy).

Infections
Infection was defined as severe when requiring hospitalisation and/or intravenous (IV) antibiotics.

Statistical analysis
Statistical analysis was performed using SPSS version 15 and GraphPad Prism version 5.01 (GraphPad Software, San Diego, CA). Results are expressed as values and proportions for categorical variables and medians and ranges for continuous variables. Clinical assessments and laboratory data were collected on a monthly basis. For missing values (e.g. if a patient did not attend clinic that month) a last value carried forward method was used until end of individual patient follow-up. The median interval between assessments was 3 months, with a median of 14 assessments per patient (1-55) and total patient follow-up of 8012 months. Changes in immunoglobulin levels over time were compared by a Wilcoxon signed rank test and between 2 groups by a Mann-Whitney U test. Proportions of patients were compared using Fisher's exact test or Chi-squared test. Correlations were assessed using Spearman's rank correlation coefficient. Time to first severe infection was analysed using Kaplan-Meier survival analysis with log rank analysis for significance. A familywise p value <0.05 was considered significant for all statistical tests with appropriate adjustments being made for the multiple testing of serial data.

Patient characteristics
One hundred and ninety-one patients received rituximab between 2002 and 2010. Fourteen were excluded; 10 due to less than six months follow-up and four due to repeated plasma exchange (PLEX). One hundred and seventy-seven patients were included ( Table 1). The median age at first rituximab was 47 years (13-82); 31% were male, and the majority had primary systemic vasculitis (56%). Median disease duration before rituximab was 52 months (0-396) including 96% with relapsing/refractory disease. The median number of prior immunosuppressive or immunomodulatory agents excluding glucocorticoids was three (0-14) including prior cyclophosphamide in 121/176 (69%) with a median cumulative dose of 8 g (0-163). At time of first rituximab 72% had active disease and 28% received rituximab for persistent low grade disease activity or as remission maintenance therapy when other drugs were contraindicated. Median follow up was 43 months (2-100). All patients had at least six months of follow-up, except for four who died within six months and were included in the analysis.

Disease response
Rituximab was an effective therapy, with 151/171 patients (88%) achieving complete or partial remission by six months. Complete remission was seen in 117/171 (68%) and partial remission in 34/171 (20%). 20/171 (12%) were considered treatment failures. There was no relationship between overall response (either complete or partial remission) and the presence or absence of hypogammaglobulinaemia (IgG < 6 g/l). In addition, no relationship was identified when complete and partial remission were considered independently, or when hypogammaglobulinaemia was sub-divided into moderate-severe (IgG < 5 g/l) or mild (IgG 5-6 g/l).

Frequency and severity of low immunoglobulin levels
Immunoglobulin data was available for 136/177 patients (77%) at time of first rituximab treatment. 18

Median immunoglobulin levels
Overall, median IgG levels remained relatively stable during follow-up, measuring at 9.3 g/L at time of first rituximab, 8.4 g/L at six months (p = 0.388) and 8.25 g/L at 60 months (N = 52) (p = 0.140). IgM levels fell from 0.8 g/L at time of first rituximab to 0.6 g/L at six months (p < 0.001) and 0.55 g/L at 60 months (p < 0.001). Baseline IgA levels were 1.9 g/L, 1.6 g/L at six months (p = 0.114) and 1.55 g/L at 60 months (p = 0.360) ( Figure 1A). For those with IgG <6 g/L (N = 18) at baseline, there was a trend for levels to increase, from a median of 4.3 g/L at time of first rituximab to 4.7 g/L at six months (p = 0.276) and 6.4 g/L at 60 months (N = 7) (p = 0.099) ( Figure 1B).
Effect of dose of rituximab on IgG levels 39/79 (49%) who received <6 g rituximab and 32/98 (33%) who received ≥6 g rituximab developed IgG hypogammaglobulinaemia for at least 3 months (p = 0.527). Patients who received <6 g of rituximab had stable IgG levels during follow-up with a median of 8.5 g/L at time of first rituximab, 8.1 g/L at six months (p = 0.204) and 8.7 g/L at 60 months (N = 21) (p = 0.178). In patients who received ≥6 g rituximab, IgG levels trended downward from a median of 9.8 g/L at time of first rituximab to 9.3 g/L at six months (p = 0.989) and to 7.5 g/L at 60 months (N = 31) (p = 0.381) ( Figure 1C).

Effect of glucocorticoid exposure on IgG levels
Median cumulative oral prednisolone exposure following the first rituximab infusion was 7.25 g (0-54.5). There was a negative correlation (r = −0.17 (CI −0.31 to −0.02) p = 0.02) between total oral prednisolone exposure after initial rituximab and IgG levels. Data on cumulative glucocorticoid exposure prior to rituximab therapy was unavailable. 57/177 (32%) received at least one dose of intravenous methylprednisolone following rituximab. Of those that received intravenous methylprednisolone 49% (28/57) developed hypogammaglobulinaemia compared to 28% (34/120) of those that did not (p = 0.011).

Effect of disease diagnosis on IgG levels
56% of patients in this cohort had primary systemic vasculitis; 24% SLE and 20% other autoimmune conditions. There was no association between diagnosis and the development of IgG hypogammaglobulinaemia (Chi-square = 3.24; df = 2; p = 0.198).

Severe infections
One hundred and seventy severe infections occurred in 71/177 patients (40%) during 8012 patient-months. Annual incidence rates of infection were 21.5 cases per 100 patient years in year 1; 10.9 in year 2; 5.2 in year 3; 1.9 in year 4 and 2.5 in year 5. The time to first severe infection was not different according to IgG levels ( Figure 2).
Need for IVIg 6/177 patients (3%) received IVIg as replacement. Median age at first rituximab was 43 years (16-67); median rituximab exposure was 5.9 g (4-11.1). The median time from first rituximab treatment to start of replacement was 18.5 months . The median IgG level before first rituximab was 4.8 g/L (4.4-5.8) and IgG level before first IVIg was 3.9 g/L (3.1-4.8). All patients had recurrent infections despite antibiotic prophylaxis and 5/6 had chronic lung disease.

Discussion
Rituximab is an effective therapy for remission induction in patients with multi-system autoimmune disease [1,2,26,29]. However, the majority of patients will relapse after a single course [3,5]. Further courses are effective, although concerns surround the long term effect on humoral immunity and infection risk with repeated rituximab use [30,31]. Rituximab depletes CD20+ B cells for an average of 6-12 months. Mature plasma cells (the source of 95% of circulating IgG) do not express CD20; however, prolonged depletion of plasma cell precursors may reduce replenishment of mature plasma cells leading to hypogammaglobulinaemia and infection risk.
We studied the frequency and severity of hypogammaglobulinaemia and severe infections following repeated rituximab dosing in patients with multi-system autoimmune disease. Low IgG, IgM and IgA levels were more frequent before first rituximab (13%, 10% and 10% of patients respectively) than in NHL or RA [16,17,[32][33][34]. At 60 months, the proportion of patients with low IgG remained stable (14%) however, those with low IgM or IgA rose to 25% (p = 0.018) and 17% (p = 0.211) respectively. The increase in proportion of patients with IgM <0.4 g/L following rituximab, is consistent with previous data in SLE and AAV patients [3,[20][21][22][23]26]. Low levels of IgG, IgM and IgA occurred in 34%, 51% and 23% of patients for at least 3 consecutive months at any time during follow-up. In keeping with the overall stabile proportion of patients with low IgG during follow-up, the majority of episodes of IgG hypogammaglobulinaemia were mild or  transient, and only a minority suffered moderate/severe prolonged IgG hypogammaglobulinaemia. IgG levels were lower at baseline in patients who had received prior cyclophosphamide treatment. However, the development of IgG hypogammaglobulinaemia was not associated with prior cyclophosphamide exposure, but was associated with a higher cumulative corticosteroid exposure during follow-up. Median IgG levels trended downwards in those who received >6 g rituximab, although this did not reach statistical significance. Our study is retrospective and whilst it does represent real life event rates of hypogammaglobulinaemia and severe infections, in patients with complex multisystem autoimmune disease treated with rituximab, outcomes are confounded by clinical strategies to reduce infection risk, and concomitant therapies that contribute to hypogammaglobulinaemia and infection risk. In our practice repeat rituximab dosing is used to treat and prevent disease flare irrespective of peripheral blood B cell count. Immunoglobulins are monitored at each clinic visit. Concomitant immunosuppressive therapies are generally not used alongside rituximab and during follow-up corticosteroids are reduced or withdrawn. In patients with repeated infections, antibiotic prophylaxis is employed; and in patients with infections despite antibiotics and moderate/severe hypogammaglobulinaemia (IgG < 5 g/l), IVIG replacement is used. Discontinuation of further rituximab dosing is considered in patients with IgG < 5 g/l with a downward (rather than stable) trajectory and recurrent infections; however, this decision is balanced against ongoing disease activity, the efficacy benefit derived from rituximab and the availability of other therapeutic options.
Existing data supports the safety of repeated doses of rituximab in RA and NHL, where stable infection rates were observed following initiation of rituximab treatment [33,35,36]. We observed infections in 38% of patients in the first year of treatment; similar to rates seen in other studies in primary systemic vasculitis [1,2,37,38] where active disease, chronic lung and kidney disease, prior cyclophosphamide and high corticosteroid exposure also contribute to infection risk. As in RA and NHL, the occurrence of persistent mild hypogammaglobulinaemia (IgG 5-6 g/l) following rituximab was not associated with a higher infection rate [33,35,36]. Unlike other populations, where IgG levels <5 g/l have been associated with infection risk [34], we did not observe an association; however, regular monitoring, antibiotic prophylaxis and IVIG replacement may have reduced infection rates in this subgroup.
Overall we observed a reduction in severe infection rates over time, along with improved disease control, a minimisation of corticosteroids and avoidance of concomitant immunosuppression. Our data suggest that provided careful monitoring is employed together with judicious use of prophylactic therapies, the IgM hypogammaglobulinaemia, transient mild/moderate IgG hypogammaglobulinaemia or occasional severe IgG hypogammaglobulinaemia observed with repeat rituximab dosing are not dominant factors in infection risk in a severe multisystem autoimmune disease population.

Conclusion
We have reported on the frequency and severity of hypogammaglobulinaemia and potential infective complications following rituximab therapy in a large retrospective survey of patients with multi-system autoimmune disease with long term follow-up. Following rituximab treatment the proportion of patients with IgM and IgA hypogammaglobulinaemia rose, but the proportion with IgG hypogammaglobulinaemia remained stable. The risk of IgG hypogammaglobulinaemia was not increased by low baseline immunoglobulin levels or prior cyclophosphamide therapy or greater cumulative rituximab exposure. However, patients with high concomitant glucocorticoid doses were at risk of developing IgG hypogammaglobulinaemia.
Only a small minority developed severe IgG hypogammaglobulinaemia necessitating IVIG replacement. Severe infections were frequent in this patient population and were associated with higher glucocorticoid exposure but were not associated with IgG levels. Overall, rituximab was a safe therapy in this population. Regular immunoglobulin monitoring is indicated with repeat rituximab dosing to identify the small minority of patients with progressively falling IgG levels.