HUMORAL AND CELLULAR IMMUNOSUPPRESSION
In healthy individuals, microorganisms only rarely cause pathological manifestations. The majority are destroyed within a few hours in a nonspecific manner by the host's innate immune system, which is the first line of defence. It is only if a microorganism passes these nonspecific defences that a specific immune response to a pathogen is elicited, generating memory cells that will prevent reinfection from the same microorganism (adaptive immunity and delayed response).
During the adaptive immune response, various cytokines and Th subgroups elicit specific effector functions. Thus, T-cells can be grouped into various subsets based on their effector functions and molecular phenotype. Distinct T-cell subsets promote different types of immune response. For example, cytotoxic T-cells kill infected or transformed cells and regulatory T-cells dampen excessive inflammation. Upon activation and expansion, CD4+ T-cells develop into different Th subsets with different cytokine profiles and distinct effector functions. Until recently, T-cells were divided into Th1 or Th2 cells, depending on the cytokines they produce. A third subset of IL-17-producing effector T helper cells (Th17) has now been characterized. Therefore, cytokine profiles (including key cytokine receptors) and activated effector Th subgroups vary depending on the type of infection (e.g. intracellular bacteria, fungi, viruses, extracellular bacteria, or parasites) .
Antibody-mediated (humoral) immunity is responsible for the neutralization of bacterial exotoxins, such as tetanus, diphtheria, and botulism; neutralization of enzymes that facilitate bacterial diffusion (staphylococcus, streptococcus); opsonization of bacteria, enabling their phagocytosis by binding to the Fc receptor on the phagocyte or through the binding of C3b to the phagocyte surface via specific receptors; and bacterial lysis, either directly (mainly via IgM) or via activation of the classical complement pathway. Antibody-dependent cell cytotoxicity is a cell-mediated immune defence whereby effector cells, mainly natural killer (NK) cells and macrophages of the immune system actively lyse a target cell, whose membrane-surface antigens have been bound by specific antibodies. In addition, secreted antibodies play a role in local immunity through their antibacterial and antitoxic actions (Table 2).
Immunosuppression reflects a poor response of the immune system. Immunodeficiency is defined as primary when it is due to a genetic defect in one or several components of the immune system. It may also be secondary to ongoing disease such as AIDS or lymphoma . Immunosuppression may also be caused by exogenous agents, such as drugs or chemicals. As presented in the previous chapter, despite the lack of recent randomized controlled trials (RCTs), secondary immunodeficiencies constitute one of the major uses for immunoglobulin use in France, in terms of quantity of IVIg consumption. Several recommendations support the use of IVIg or SCIg in secondary immune deficiency observed in multiple myeloma or chronic lymphoid leukaemia (CLL) [28,32]. Prior to IVIg substitutive therapy, prophylactic antibiotics should be proposed in patients with secondary immunodeficiencies, and IVIg should only be prescribed in case of failure of prophylactic antibiotic therapy, which is not the case in PID patients in whom IVIg is proposed as a first-line therapy.
When evaluating a participant with a suspected immunodeficiency, clues suggestive of an immune deficiency may be related to aspects of infection (unusual frequency, severity, duration, complications, and pathogens) or may be noninfectious (e.g. granulomatosis, lymphoma or solid tumours) . Indeed, haematological malignancies, immunosuppressants, and biological agents, such as rituximab (affecting B-cells), infliximab, etanercept, adalimumab, anakinra (affecting cellular immunity), are common causes of secondary immune deficiency . In a context of recurring bacterial infection, quantification of immunoglobulins, serum electrophoresis, Bence–Jones protein, and identification of antibodies against vaccine antigens can be performed to investigate humoral immune deficiency. In patients with meningococcal meningitis, investigation of haemolytic activity of the classical complement pathway (CH50) may help identify a defect in the final common pathway, whereas analysis of the haemolytic activity of the alternative (AP50) complement pathway and quantification of properdin may allow identification of a defect in the alternate complement pathway. In the case of opportunistic infections, possible lymphocytopenia should be investigated, with phenotypic analysis of T-lymphocyte subpopulations, particularly CD4+ T lymphocytes. In this setting, HIV serology should be performed.
Hypogammaglobulinaemia refers to a deficit of circulating gammaglobulins and antibody production, and may be either primary or acquired. Acquired hypogammaglobulinaemia can be caused by a number of haematological malignancies including non-Hodgkin's lymphomas, CLL, or multiple myeloma. In these conditions, high-dose chemotherapy, fludarabine, rituximab, stem cell transplantation and/or radiotherapy may contribute to low serum immunoglobulin levels. Hypogammaglobulinaemia may also be the consequence of treatments such as glucocorticoids alone or more often in association with immunosuppressants, rituximab, plasma exchanges, D-penicillamine, phenytoin, or gold compounds for the treatment of systemic inflammatory or autoimmune diseases [34–37]. Thus, hypogammaglobulinaemia may occur in patients with rheumatoid arthritis (RA) treated with rituximab. In the presence of hypoalbuminaemia, acquired hypogammaglobulinaemia may also be the consequence of protein loss through nephrotic syndrome, exudative enteropathy, or hypercatabolism (third-degree burns and myotonic dystrophy). Moreover, hypogammaglobulinaemia may be associated with thymoma, and may be found in patients with cryoglobulinaemia [38,39].
Primary hypogammaglobulinaemia is much less frequent. Presentation is obviously different, due to the absence of any underlying cause, to a usually younger age and to a possible familial context. It can result from a number of disorders, including CVID, IgA, or IgG subclass deficiency, hyper-IgM syndrome, XLA (with or without growth hormone deficiency), Purtillo syndrome (susceptibility to Epstein–Barr virus), combined B-cell and T-cell immunodeficiencies, Wiskott–Aldrich syndrome, or ataxia telangiectasia.
Infections in multiple myeloma
Multiple myeloma is a haematological malignancy characterized by the development in the bone marrow of clonal plasma cells producing a monoclonal immunoglobulin (also known as paraprotein, an abnormal immunoglobulin produced by the tumour clone). In theory, multiple myeloma can produce all classes of immunoglobulin, but IgG paraproteins are most common, followed by IgA . In multiple myeloma patients, protein electrophoresis of the blood and urine may display the presence of a paraprotein band, with or without reduction of the other (normal) immunoglobulins (known as immune paresis). Multiple myeloma almost consistently results in decreased production of polyclonal immunoglobulins, leading to increased rates of infection. A recent study has shown that antibody titres are depressed in patients with multiple myeloma, monoclonal gammopathy of undetermined significance (MGUS), and Waldenstrom's macroglobulinaemia. Humoral immunity to 24 different pathogens in elderly patients with multiple myeloma (n = 25), Waldenstrom's macroglobulinaemia (n = 16), and MGUS (n = 18) was compared with age-matched controls (n = 20). Although multiple myeloma patients displayed the most depressed humoral immunity, significantly decreased antibody levels were also evident in patients with Waldenstrom's macroglobulinaemia and MGUS, particularly against Staphylococcus aureus, pneumococci, and varicella .
Multiple myeloma is associated with an increased incidence of early infection (5–15%)  that is related to defects in both humoral and cellular immunity, reduced mobility and performance status, caused by the disease and its treatment, although neutropenia is not usually a factor in early infection. In haematological malignancies, it may be difficult to separate the causes for infection risk between the disease itself and the disease treatment. Studies have observed differences in the type and timing of infections in the course of the disease. Indeed, bacterial infections (mostly affecting the respiratory and urinary tracts) are more frequent in the first 3 months after initial chemotherapy [42,43]. Savage et al.  showed that Gram-negative organisms caused infection at times of neutropenia, whereas encapsulated organisms were responsible for infections at other times. Up to 10% of multiple myeloma patients die of infective causes within 60 days of diagnosis . Early mortality in multiple myeloma patients was studied in a series of large multicentre trials. Bacterial infection was directly responsible for the majority (45%) of early deaths. Of these, 66% were due to pneumonia . Savage et al.  described a biphasic pattern of bacterial infection in multiple myeloma, whereby episodes of infection with Streptococcus pneumoniae and Haemophilus influenzae occurred early in the disease course, and in patients responding to chemotherapy. Gram-negative bacilli and Staphylococcus aureus were responsible for 80% of infections reported after diagnosis and for 92% of deaths from infection. Infection with Gram-negative bacteria occurred in patients with active and advancing disease and in those responding to chemotherapy in case of neutropenia. The authors concluded that in multiple myeloma patients, humoral immune deficiency may be the major defect leading to S. pneumoniae and H. influenzae infections, whereas additional factors related to disease activity appeared to predispose to Gram-negative infection.
Supportive care in multiple myeloma
In multiple myeloma, supportive care is important, because multiple myeloma is incurable and most patients have to contend with both disease-related and treatment-related adverse effects. Thus, maintenance of quality of life is a key factor in multiple myeloma management. There are various national guidelines for immunoglobulin use in these patients, which despite some discrepancies and varying recommendation grades and evidence levels, agree that prophylactic immunoglobuline treatment is not routinely recommended, but is useful in a small subset of patients with severe, recurrent bacterial infections and hypogammaglobulinaemia (due to antibody failure associated with haematological malignancies or in patients undergoing haemopoietic stem cell transplantation) [28,45–47]. Importantly, other contributing infection risks that are not preventable by IVIg (e.g. neutropenia or chemotherapy-related mucositis) must be considered in these patients when determining the likely role of hypogammaglobulinaemia in the development of infections. Recurrent or severe bacterial infections include infections that have not recovered despite continuous oral antibiotic prophylactic therapy for 3 months. Other selection criteria for IVIg in the treatment of hypogammaglobulinaemia linked with haematological malignancy include IgG lower than 5 g/l (excluding paraprotein), and a documented failure of serum antibody response to unconjugated pneumococcal or other polysaccharide vaccine challenge . Guidelines recommend using the lowest possible immunoglobulin dose to achieve appropriate clinical outcomes for each patient. Maintenance doses are in the region 0.4 g/kg every 4 weeks, modified to achieve IgG trough levels of at least the lower limit of the age-specific serum IgG reference range [45,46]. An additional (loading) dose may be administered at treatment initiation if the serum IgG level is lower than 4 g/L. Alternatively, suggested SCIg dose is 0.1 g/kg lean body mass every week .
Intravenous immunoglobulins in the treatment of infections
In antibody-deficient patients, IVIgs contain important quantities of opsonizing antibodies against a wide array of bacteria. IVIgs bind and neutralize microbial agents, preventing their binding to the surface of target cells. The antibacterial efficacy of IVIgs is dependent on the number of donors contributing to the pool (the more the better, >1000 according to recommendations) . The principal specificities of neutralizing antibodies, antitoxins, and opsonizing contents within available IVIg preparations are as follows:
1. Viruses: Hepatitis A, B, Epstein–Barr, cytomegalovirus, poliomyelitis, measles, mumps, myxovirus influenza, adenovirus, herpes simplex, varicelle-zona, coxsackie, and rubella;
2. Bacteria and bacterial toxins: Staphylococcus aureus, group B streptococcus, S. pneumoniae, H. influenzae, Pseudomonas aeruginosa, Salmonella, Shigella, Klebsiella pneumoniae, Escherichia coli, Bordetella pertussis, diphtheria toxin, and tetanus toxin.
Guidance on the clinical investigation of human IVIg is shown in Table 3.
CLL, chronic lymphoid leukaemia; XLA, X-linked agammaglobulinaemia.
In conclusion, infectious complications of malignant lymphomas, multiple myeloma, and CLL represent a major prognostic factor. In this setting, hypogammaglobulinaemia is an important risk factor in the development of infections, in particular of pyogenic bacterial infections. IVIgs contain large amounts of neutralizing antibodies, antitoxin, and opsonizing antibodies, which may confer benefit in these patients. Multiple myeloma, CLL with severe hypogammaglobulinaemia, and recurrent infections are validated indications for the use of IVIg in the context of infection prevention during secondary immunodeficiency disorders, although the studies included relatively few patients and were published in the 1990s, without comparison of the efficacy of IVIg and antibiotic prophylaxis for the prevention of infection.
IMMUNE DEFICIENCIES AND LYMPHOID DISORDERS: MANAGEMENT OF INFECTIOUS COMPLICATIONS
Lymphoproliferative disorders are a heterogeneous group of diseases characterized by the expansion of clonal lymphoid cells. These disorders are commonly associated with immune deficiency, in which humoral immunity usually predominates. Secondary immune deficiencies can also involve cellular immunity, particularly in some lymphoid diseases (e.g. Hodgkin's lymphoma). In addition, they can be exacerbated by lymphoid malignancy treatment, including splenectomy, anti-B-lymphocyte monoclonal antibody (mAb) treatment, and medullary transplantation, which greatly impacts on the infectious risk. This risk may be further increased by the recent introduction of kinase inhibitors that target signalling pathways essential for B-cell development, in particular those targeting the B-cell receptor, such as Ibrutinib, which is emerging as a very promising targeted therapy of CLL and lymphomas .
The principal pathogens involving humoral immune responses are encapsulated microorganisms, dominated by S. pneumoniae, H. influenzae type b, or Neisseria meningitidis. In most cases, anatomical barriers such as the skin and mucous barriers of the intestinal and respiratory tracts stop infectious agents. However, in the event of invasive infection, the first line of defence is provided by innate immunity. If an infectious agent manages to penetrate the anatomic defences, acute inflammation occurs via humoral factors, including activation of the complement system, recruitment of phagocytic cells, and marking of pathogens for destruction. Part of the inflammatory response involves recruitment of cells of the innate immune system, such as neutrophils, macrophages, dendritic cells, or NK cells. The invading microorganism is directed at the spleen, where a subpopulation of B-lymphocytes reacts to antigens in order to opsonize and eliminate the infectious agent. Thus, in an immunodeficient setting, infection with encapsulated microorganisms may occur due to a complement deficiency (hereditary and congenital), innate or humoral immune deficiency, or asplenia (responsible for more frequent and severe infections) .
Anti-B-lymphocyte monoclonal antibody treatment
Treatment with anti-B-lymphocyte (anti-CD20) mAb (rituximab), often combined with chemotherapy, leads to an increased risk of bacterial and viral infections in patients with blood disorders. Serum IgG levels are not heavily affected by rituximab, possibly because CD20-negative long-lived plasma cells maintain antibody production . However, serum IgM has been shown to decrease after rituximab treatment . In addition, the humoral response to vaccination might be delayed until recovery of B-cells [51,52].
Sustained B-cell lymphopenia has been recorded with single-use rituximab 6 months after injection. However, hypogammaglobulinaemia was rare (except in newborns). No increase in incidence or in severity of infectious complications was observed . Following repeated treatment, the proportion of patients with circulating IgM and IgG levels below the lower limit of normal increased with the number of cycles. IgMs decreased by 10.3, 18.5, and 23.5% after the first, second, and third cycles, respectively. IgG and IgA levels also decreased to a lesser extent . Importantly, cases of progressive multifocal leukoencephalopathy (PML) after rituximab treatment have been reported. In these patients, the main indications for rituximab treatment included CLL (24.6%), follicular lymphoma (19.3%), and diffuse large B-cell lymphoma (12.3%). Of note, all patients who developed PML had previously been treated with therapies that also affect immune function, including purine nucleoside analogues, alkylating agents, or corticosteroids .
The long-term effect of rituximab therapy on the immune system has not been studied in detail. Although numerical normalization of B-cell numbers takes place in the majority of patients 1–2 years after a single course of rituximab, the mediated B-cell depletion appears to yield a more durable effect on peripheral blood B-cell subpopulations with a long-term delay in memory B-cell subsets [56,57]. Rituximab maintenance, which is increasingly proposed in the treatment of indolent lymphomas, is associated with a higher rate of infectious complications compared with observation alone .
Immune deficiencies may also be related to the underlying disease. Hypogammaglobulinaemia is the principal risk indicator of deficiency in humoral immunity. Hypogammaglobulinaemia (or decrease in polyclonal uninvolved immunoglobulin serum levels in the presence of a monoclonal spike, as evaluated by serum electrophoresis or by specific dosage of the immunoglobulin class other than that of the monoclonal immunoglobulin) is particularly frequent in multiple myeloma (>95%) and CLL (# 60%). It can also occur in other B-cell lymphoproliferative disorders such as MGUS (# 30%), Waldenstrom's macroglobulinaemia and some non-Hodgkin's lymphomas. It is very rare in Hodgkin's lymphoma and has been observed in a few cases of T-cell lymphoma.
Humoral immune deficiency can also occur in lymphoproliferative disorders in the absence of hypogammaglobulinaemia. The evaluation of antibody responses to protein and polysaccharide antigens, either spontaneously or following vaccination, is likely to be a better risk indicator. In CLL, patients with hypogammaglobulinaemia and a satisfactory rate of antipneumococcal antibodies postvaccination (approximately 20%) were observed and, conversely, approximately 5% of patients had normal IgG levels and a low rate of antipneumococcal antibodies .
Hypogammaglobulinaemia (or low postvaccination or postinfection antibody levels) has important prognostic implications. Indeed, the estimated incidence of serious infections in multiple myeloma is much higher than for control populations, as in CLL, Waldenstrom's macroglobulinaemia, and MGUS patients . Moreover, mortality related to infections is also higher in these patients. In multiple myeloma, it is estimated to vary between 15 and 80% [61–63].
During the disease course, the main infectious risk is due to encapsulated bacteria, particularly to S. pneumoniae. The frequency and type of infections vary: during periods of treatment, the drug-related deficit in innate and cellular immunity (including neutropenia) may cause fungal and viral infections in addition to bacterial infection. When they are achieved, remissions are accompanied by a decreased infection rate with possible correction of hypogammaglobulinaemia, but the risk unfortunately reappears upon relapse. Of note, novel therapeutic strategies (such as high-dose therapy and autologous transplantation) and novel agents (e.g. bortezomib, which produces a specific risk of herpetic infection) are likely to have modified the profile of infectious complications. Indeed, a recent large population-based study from Sweden  showed that bacterial and viral infections represent a major threat to multiple myeloma patients who had a risk of specific infections like pneumonia and septicaemia over 10 times higher than in controls during the first year after diagnosis. In the same time, they had an 18-fold risk of developing a viral infection, particularly influenza and herpes zoster infections.
In patients with MGUS, uninvolved (normal, polyclonal, or background) immunoglobulin concentrations have a low prognostic significance for progression to multiple myeloma or other lymphoid disorders. In the Mayo Clinic study of more than 1300 MGUS participants followed for 20–30 years, the risk of progression was estimated at about 1% per year. The initial concentration of serum monoclonal protein was a significant predictor of progression, unlike the concentration of uninvolved immunoglobulins (which was reduced in 38% of patients). Patients with IgM or IgA monoclonal protein had an increased risk of disease progression, compared with patients who had IgG monoclonal protein (P = 0.001) .
Humoral immunodeficiency in lymphoproliferative disorders: therapeutic implications
For patients with lymphoproliferative disorders and secondary humoral immunodeficiency, the increased risk of infections, particularly from encapsulated organisms, has therapeutic implications regarding antibiotic prophylaxis, vaccination, and IVIg therapy.
Taking into account the dominant incidence of infection with encapsulated microorganisms, the specific targeting of these pathogens is preferable (S. pneumoniae in particular). For this purpose, penicillin derivatives appear to be a good choice, while taking individual data (allergy) and local epidemiology (e.g. incidence of penicillin-resistant bacteria) into consideration. In spite of their relatively large spectrum, fluoroquinolones are proposed in some countries. Macrolides (e.g. azithromycin three times per week) are also used in some centres, highlighting variation in practice in the absence of a clear evidence base.
In multiple myeloma patients, only one randomized study has investigated whether morbidity and mortality of early infection can be prevented by prophylactic administration of antibiotics. The study used trimethoprim-sulfamethoxazole (TMP-SMX), which is not a first-line choice for encapsulated microorganisms. Bacterial infection (including severe infection) was shown to be significantly less frequent in patients assigned TMP-SMX compared with controls. The difference in infection-related deaths between groups was not significant, but the number of enrolled patients was small (n = 54) .
Similarly to the antibiotic prophylaxis issue, the question of whether or not vaccination is indicated in patients with lymphoproliferative disorders has not been extensively investigated. Due to secondary humoral immunodeficiency, the efficacy of vaccination may be limited, although it may confer partial protection. Biological studies have shown variable antibody response data, including low antibody response, especially after immunochemotherapy [51,66,67]. Importantly, vaccine efficacy trials with clinical endpoints (such as infections) are lacking.
The principles that apply to vaccination of immunosuppressed cancer patients and to patients with primary immunodeficiency  also apply to patients with lymphoid disorders, with particular emphasis on the following key points: vaccination with inactivated vaccines is considered well tolerated, but live vaccines should be avoided; vaccination of close contacts with inactivated vaccines is strongly recommended; and there is no optimal time at which patients with lymphoproliferative disorders would most benefit from vaccination. The clinical decision to vaccinate must be made individually, based on evaluation of infection risk, and institution-specific and country-specific guidelines. However, whenever possible, early vaccination is recommended, preferably before introducing any therapy, and vaccination should not be performed during immunosuppressive treatment (glucocorticoids or chemotherapy) [56,69]. Indeed, response to vaccination is usually markedly reduced after chemotherapy, although response after treatment with novel agents remains to be well assessed (e.g. bortezomib, and the so-called immunomodulatory drugs in particular). The remission status of the underlying disease and the therapies applied (particularly high-dose steroids and high-dose chemotherapy with autotransplantation) determine the patient's ability to develop a protective response. Patients should, therefore, be vaccinated at earlier stages of the disease, such as MGUS or smouldering myeloma, and when in remission (at least 6 months after the end of treatment). It has been suggested that absolute lymphocyte and CD4+ cell counts and the plasma levels of uninvolved immunoglobulins may be considered when planning the timing of vaccination, although there are few data in support of this approach . Among patients scheduled to receive therapy, vaccination should be performed at least 14 days before initiation of chemotherapy, before stem cell mobilization among patients undergoing this procedure, upon achievement of best response, 3–6 months after completion of chemotherapy, or 6–12 months after receiving high-dose chemotherapy. If given between cycles of chemotherapy, the effectiveness of the procedure is likely to be dramatically reduced .
Once the decision to vaccinate a patient with a lymphoid disorder is made, antipneumococcal vaccine is certainly the first to be considered. The 23-valent pneumococcal polysaccharide vaccine (PPSV23) contains 23 of the 90 known polysaccharide serotypes, which are involved in 85–90% of adult invasive infections . PPSV23 in lymphoproliferative disorders such as multiple myeloma has shown some biological efficacy, most often incomplete and not durable in the long term (<5%) . Even if they include fewer serotypes, the more recent glycoconjugate vaccines, such as Prevenar 7 or Prevenar 13, are likely to be more effective. Importantly, their mode of action has been recently elucidated, and new designs may improve their efficacy in future . Sinisalo et al.  compared antibody response rates to vaccine antigens from conjugated pneumococcal vaccination in 52 CLL patients and age-matched and sex-matched controls. Compared with controls, antibody response rates were lower in CLL patients. However, significant vaccination response to at least six antigens was obtained in almost 40% of the patients. The study confirmed that early administration of conjugate vaccine, prior to chemotherapy and hypogammaglobulinaemia, should be advocated, as the least severely affected patients responded.
Further studies with clinically relevant endpoints are required for a better evaluation of both polysaccharide and conjugated pneumococcal vaccines in patients with secondary humoral immune deficiency. In addition, these studies should determine the optimal vaccination strategy. For allogeneic stem cell transplantation recipient, the Centers for Disease Control and Prevention recommends sequential administration of three doses of pneumococcal conjugate vaccine, beginning 3–6 months after the graft, followed by a dose of PPSV23 and by subsequent revaccination [70,74]. Regardless of the modality of vaccination, subsequent development of infection should prompt further recommendation of antibiotic prophylaxis .
Although significant amounts of infection are due to nontypeable Haemophilus, vaccination options against H. influenzae type b and meningococcal infection may be considered in some cases. In a recent small study, nine patients with indolent B-cell lymphoma treated with rituximab were revaccinated both with PPSV23 and the conjugated H. influenza type b vaccine (Act-Hib). The latter was more effective in creating an immune response than the former, providing further evidence for the value of conjugated vaccine compared with polysaccharide vaccine .
Regarding viral infections, particularly with influenza, vaccination is recommended in patients with lymphoid disorders, as in all immunosuppressed cancer patients . However, its efficacy cannot be fully guaranteed. It is therefore strongly recommend to vaccinate close contacts, using inactivated vaccines .
Further therapeutic implications in patients with lymphoproliferative disorders and secondary humoral immunodeficiencies include the administration and timing of IVIg (or SCIg). IVIgs have been shown to prevent infections in these patients . Their administration depends on catabolism and serum levels of IgGs. As a reminder, the FcRn receptor is a ubiquitous class I-like major histocompatibility complex (MHC) molecule that binds IgG in a pH-dependent saturable manner. FcRn has a role in IgG transplacental and transepithelial transport, and in protecting/recycling IgG, by regulating IgG catabolism and homeostasis. In other words, in polyclonal or monoclonal hypogammaglobulinaemic patients, beyond a threshold of circulating IgGs (approximately 30 g/l), their catabolism is accelerated, and their protective effect is reduced. In practical terms, IVIg or SCIg may accelerate IgG catabolism.
In a randomized, double-blinded, placebo-controlled, multicentre prophylactic IVIg trial, patients with plateau-phase multiple myeloma (n = 82) were randomized to IVIg 0.4 g/kg per month versus placebo for 1 year. Thirty-eight serious infections occurred in 470 patient-months on placebo, compared with 19 in 449 patient-months on IVIg (P = 0.019) . More studies are available on CLL patients [77–82]. There is agreement that IVIg treatment in patients with plateau-phase multiple myeloma (in which patients are in remission) and CLL is effective in reducing the frequency of infections, although the CLL studies contained varying numbers of advanced-stage patients . In addition, these available studies did not use antibiotic prophylaxis; one might imagine that added antibiotic prophylaxis could confer further protection against infection in these individuals, especially in advanced-stage patients.
The protective effect of IVIg prophylaxis versus placebo or no treatment in hypogammaglobulinaemic patients with CLL and multiple myeloma was confirmed in a meta-analysis. A significant decrease in the occurrence of major infections was observed, relative risk (RR) 0.45 [95% confidence interval (CI) 0.27–0.75, three out of nine trials] and a significant reduction in clinically documented infections, RR 0.49 (95% CI 0.39–0.61, three out of trials) . The analysis did not, however, report a survival benefit: RR 1.36 (95% CI 0.58–3.19, two out of nine trials). The authors concluded that, based on available data, prophylactic IVIg cannot be routinely recommended in hypogammaglobulinaemic patients with lymphoproliferative disorders. Again, treatment decisions should be made on an individual basis .
Other reasons hindering routine use of IVIg in these patients include treatment costs and impact of treatment on quality of life. Following the pivotal RCT of IVIg in CLL patients with hypogammaglobulinaemia demonstrating statistically significant reductions in the rate of bacterial infections among patients who received IVIg , Weeks et al.  reported on the cost-effectiveness of prophylactic IVIg treatment. The analysis found that IVIg therapy was associated with decreased quality-adjusted life expectancy, due to inconveniences associated with treatment. Even when disregarding these inconveniences, IVIg therapy resulted in a modest gain of 0.8 quality-adjusted days per patient per year of therapy, at a considerable cost of $6 million per quality-adjusted life-year gained. This cost-effectiveness study was, however, challenged. Appropriate cost-effectiveness studies should prospectively eliminate patients who have had no previous moderate or major infections. Cost-effectiveness of IVIg prophylactic use in patients with CLL and hypogammaglobulinaemia should be analysed in a subset of patients at high risk for infection .
In conclusion, despite the heavy impact of infectious complications on survival and quality of life in patients with secondary immune deficiencies, particularly in patients with lymphoproliferative disorders, many uncertainties remain. There currently are no well designed efficacy trials using clinical endpoints (such as infections) for addressing antibiotic prophylaxis, vaccination, and to a lesser extent, immunoglobulin replacement issues. In addition, indicators reliably identifying patients who are most at risk for infection are still lacking, and patient monitoring also needs to be improved. For this purpose, assessing response to vaccination may serve as a surrogate marker, but this approach has limitations, particularly because serological response to an antigen, either a polysaccharide or a protein, does not automatically imply responsiveness to all antigens of the same type. Moreover, potential loss of response following additional immunosuppressive therapies may occur. Accordingly, a pragmatic approach, in which treatment is decided on an individual basis, is currently often adopted. This approach must be mainly directed against infections with encapsulated organisms, particularly S. pneumoniae, and should use vaccination at earlier stages of any lymphoid disease and in persons in close contact with the patients. Because hypogammaglobulinaemia does not necessarily imply an increased infection risk, immunoglobulin replacement should only be considered in those patients who present with recurrent or severe infections that occur despite appropriate antibiotic prophylaxis and vaccination.
TREATMENT-RELATED IMMUNE DEFICIENCY: FOCUS ON AUTOIMMUNE DISEASES
Acquired immunodeficiencies of iatrogenic origin (treatment-related) are common and can result in severe clinical complications. They are the result of therapy that suppresses immune response, either as a goal or as a side-effect. Increased susceptibility to infections is an important consequence of immunosuppression, which must always be considered and balanced against the therapeutic benefits of a particular treatment. The most common treatments associated with iatrogenic immunodeficiencies are glucocorticoids, cytotoxic drugs, antilymphocyte antibodies, or ionizing radiation. There is a tendency of treating physicians to forget how frequently immunosuppressive treatments are used. The discussion below describes the role of novel biological therapies in the development of iatrogenic immunodeficiency.
Biological therapies are often misleadingly referred to as targeted therapy. Indeed, when cytokines and chemokines (e.g. TNF-α) are blocked in a targeted fashion, a cascade of downstream consequences is initiated, in addition to a direct effect on cytokine levels. Several lymphocyte populations will therefore be affected. For example, psoriasis treatment with ustekinumab blocks p40 (the shared subunit of IL-12 and IL-23) and blocks both the Th1 and Th17 inflammatory pathways . Moreover, multiple drug-dependent effects have been observed by blocking the same cytokine, for instance by blocking TNF-α with soluble TNF receptor (TNF-R) or with mAbs. This results in different efficacy and tolerability profiles. One may think that the mechanisms involved in these biological therapies are better understood than those of older immunosuppressors [such as methotrexate (MTX)], but it is not the case. Even the simplest scenarios, such as B-cell depletion with rituximab, also impact on T-cells . Similarly, after rituximab treatment, the recovery of B-cells may vary from patient to patient, which also impacts on treatment efficacy and safety [51,52].
Limitations of available evidence
Even though the effects of biological therapy on the immune response may not be fully understood, animal models are available to address outstanding questions and to possibly predict tolerability issues in humans. Importantly, ‘knockout’ animal models have limitations. Indeed, a study reported that TNF-α(−/−) mice were resistant to experimental skin tumours induced by 9,10-dimethyl-1,2-benzanthracene and 12-O-tetradecanoylphorbol-13-acetate . However, it has since transpired that nonmelanoma skin cancer incidence may be exacerbated by anti-TNF-α treatment [89–91]. Clinical trials also have limitations, in that they are carried out in strictly selected patients, and follow-up is usually short. Tuberculosis (TB) risk was discovered in patients undergoing anti-TNF-α treatments after the drug was commercialized, even though preclinical arguments were available [92,93].
Biological therapy-induced iatrogenic immune deficiencies
Biological therapies induce three types of consequences of iatrogenic immunodeficiency: infections, cancers and lymphomas, and induced autoimmunity. Regarding all these complications, the use of common immunosuppressive drugs over many years needs to be taken into account as well as the effects of the most recent biological therapy being prescribed. Some therapies first used for malignant disease, such as rituximab, and now also used in autoimmune disease, might have different complications in this latter setting. With a focus on rituximab and anti-TNF-α therapies, the discussion below will review whether infections in patients prescribed biological therapies are more frequent, whether different opportunistic pathogens are involved, and whether biological therapies induce hypogammaglobulinaemia.
Infections in patients receiving TNF-α antagonists
Several clinical trials (open-label follow-up at the end of the randomized controlled phase), meta-analyses, and data collected from registries have investigated whether infections are more frequent in patients treated with biological therapies. Data from insurance registers are often less precise, but are compensated for by large patient numbers. Unlike clinical trial populations, the appeal of data from patient databases is that they contain heterogeneous patient populations. Approximately half of the data available reported a two-fold increased risk of developing severe infection (requiring hospitalization and/or intravenous antibiotics) in patients with treatment-refractory RA receiving TNF-α antagonists. However, some data from registries do not show an increase in infection risk (Table 4).
A more recent meta-analysis of six RCTs in early RA patients treated with TNF-α antagonists [without prior disease-modifying antirheumatic drugs (DMARDs) or MTX treatment] did not report an increased infection risk in these patients. From a total of 2183 patients receiving biological therapy and 1236 patients receiving MTX, the pooled odds ratio (OR) for serious infections was 1.28 (95% CI, 0.82–2.02) , although the duration of the trials included was relatively short (6–12 months). Therefore, whereas other analyses have shown an increased risk of serious infection in patients receiving anti-TNF-α therapy, more recent data show that this is not the case in early RA patients with no prior DMARDs and/or MTX treatment . Importantly, infectious risk is not only dependent on the type of biological therapy, but also on patient comorbidities and underlying disease.
Wallis et al. [100,101] presented a profile of opportunistic infections in RA patients treated with TNF-α antagonists, mainly consisting of intracellular pathogens, indicative of a deficiency in cellular immunity. The study reported that the risk of TB reactivation was greater with the infliximab mAb than with the etanercept soluble receptor. Moreover, the British and French registries [British Society for Rheumatology Biologics Register and Research Axed on Tolerance of Biotherapies respectively] reported an increased risk of TB in RA patients treated with anti-TNF-α compared with patients with active RA treated with traditional DMARDs. Interestingly, these studies also pointed to a higher rate of TB in patients receiving infliximab and adalimumab compared with etanercept [102,103].
TNF-α antagonists: monoclonal antibodies versus soluble receptor
The differences in mechanisms of action between the two approaches used to block TNF-α are not fully understood. Membrane-bound TNF-α is important in immunosurveillance against intracellular pathogens because it more selectively activates one of the two TNF-α receptors involved in T-cell apoptosis and in granuloma formation against intracellular pathogens. It is thought that etanercept has a reduced affinity for membrane-bound TNF-α [104,105]. Bruns et al.  investigated the mechanisms by which anti-TNF-α treatment impairs T-cell-directed antimicrobial activity. Effector memory T-cells (TEMRA), which have elevated expression of membrane-bound TNF-α and express large amounts of perforin and granulysin, are particularly involved in Mycobacterium tuberculosis surveillance. Mean expression of membrane-bound TNF-α on CD8+ TEMRA cells is higher in RA patients than in healthy participants (2.1 versus 0.3%). In-vitro infliximab treatment provoked TEMRA-cell apoptosis, thus reducing protection against M. tuberculosis. Infliximab did not affect (at 24 and 96 h) the rate of IFN-γ produced by the peripheral blood mononuclear cells of patients treated with infliximab, but lymphocyte perforin and granulysin expression were both reduced 2 weeks after infliximab. In other words, anti-TNF-α therapy triggered a reduction of CD8+ TEMRA cells with antimicrobial activity against M. tuberculosis.
In another example of cellular immune deficiency, anti-TNF-α therapy has also been reported to increase herpes zoster infections in RA patients enrolled in the German biologics register (RABBIT) . Again, mAb use was associated with a higher risk compared with etanercept, possibly for the same reasons outlined above.
Effects of biological therapies on humoral immunity
Although there is a wealth of data on TNF-α antagonists and cellular immune deficiency against intracellular pathogens, data on humoral immunity are very limited. The prevalence of hypogammaglobulinaemia in RA patients treated with either conventional glucocorticoid therapy or biological therapies, and the respective impact of the underlying disease on humoral immunity have not been presented to date. There are few data on serum immunoglobulin levels in RA patients treated with TNF-α antagonists. Etanercept may inhibit B-cells because it not only blocks TNF-α but also lymphotoxin α, which has an important role in the homeostasis of germinal centre memory B-cells . A cross-sectional study of RA patients treated with MTX (n = 17), etanercept (n = 11) or MTX and etanercept (n = 17) analysed the effect of etanercept on B-cells in blood and tonsils. The study provided preliminary evidence supporting the hypothesis that TNF-α antagonists disrupt germinal centre reactions via effects on follicular dendritic cells in humans. Anolik et al.  demonstrated that RA patients on etanercept had reduced serum memory B-cells (CD27+), and disorganized lymphoid follicles with increased naive B-cells and reduced centroblasts in tonsils.
In addition to TNF-α antagonists, abatacept, which modulates CD28-mediated T-cell costimulation, has been described as affecting antigen-specific T-cell and B-cell responses in vivo. Abatacept is associated with a failure of antigen-specific T-cells to acquire the T follicular helper cell phenotype, leading to a failure of these cells to enter B-cell follicles, and resulting in reduced specific antibody responses, despite normal B-cell clonal expansion . Therefore, without direct action on B-cells, abatacept may also affect immunoglobulin levels. Similarly, in systemic lupus erythematosus (SLE), the recently approved belimumab mAb against BAFF reduces B-cells, and serum levels of IgM slowly decrease over time, with only a modest effect on serum levels of IgG .
Data from patient registries
Patient registers offer the opportunity to study large numbers of heterogeneous patient populations treated with biological therapies, including rituximab, abatacept, or tocilizumab. For example, the anti-CD20 rituximab AutoImmunity and Rituximab French Society of Rheumatology registry (n = 2600) patients are very different to those included in clinical trials, in that 13% have a history of cancer, and only two third received concomitant immunosuppressive therapy (e.g. MTX). Analyses of severe infections that led to hospitalization or to intravenous antibiotic therapy in this population showed that 200 severe infections occurred per 3683 patient-years, equivalent to 5.4 severe infections per 100 patient-years , which is higher than the rates recorded in rituximab clinical trials in RA patients.
Serum electrophoresis data were available in some patients prior to rituximab treatment, which showed that 5.4% of RA patients had hypogammaglobulinaemia (<6 g/l), and 4.1% had low IgG levels (<6 g/l). In keeping with reported literature, hypogammaglobulinaemia and hypo-IgG before rituximab therapy was associated with increased age and higher doses of oral glucocorticoids. When analysing the risk of severe infection, the patients who had hypogammaglobulinaemia before rituximab therapy were more likely to have a severe infection in multivariate analysis . The prevalence of hypogammaglobulinaemia (approximately one in 20 patients) in RA patients and the association with the risk of serious infections shows the relevance of monitoring gammaglobulins prior to rituximab therapy . In a recent study evaluating the long-term safety of rituximab in RA, 22.4% of rituximab-treated patients developed low IgM and 3.5% low IgG levels for at least 4 months after at least one treatment course. Patients with low IgG after rituximab were also more prone to develop serious infections .
Thus, the risk–benefit ratio of rituximab treatment must be evaluated for each patient with a hypo-IgG . No studies have demonstrated that immunoglobulins should be administered to patients who develop hypogammaglobulinaemia with rituximab. Several therapeutic options are available for RA patients who develop a hypo-IgG, so other therapies may be considered. As discussed, strategies to prevent the increased risk of infection include vaccination of patients and of close contacts prior to treatment with a biological therapy (against influenza and pneumococcus).
Cancer and lymphoma risk and TNF-α antagonists
A meta-analysis of double-blinded, placebo-controlled randomized trial recorded rates of malignancy in treatment-naive patients with early RA prescribed TNF-α antagonists. From a total of 2183 patients on biological therapy, the pooled OR for malignancies was 1.08 (95% CI 0.50–2.32), showing no significant difference in the rate of malignancies between the anti-TNF-α therapy group and the control group .
Longitudinal data from the US National Data Base for Rheumatic Diseases allowed the investigation of cancers in RA patients and their relationship with biotherapies. The analyses included 13 869 RA patients with at least 1 year of follow-up, including 6597 treated with anti-TNF-α therapy, and recorded the incidence of treatment-emergent cancers. The only cancers for which TNF-α antagonists significantly increased the incidence were nonmelanoma skin cancers OR = 1.5 (95% CI 1.2–1.8) . Therefore, it is advisable to refer patients receiving TNF-α antagonists to a dermatologist. Of note, all immunosuppressors, including glucocorticoids, are known to increase skin cancer risk .
Increased lymphoma risk compared with the general population is more difficult to assess, because RA is by itself a risk factor. Indeed, risk of lymphoma is substantially increased in a subset of RA patients with severe disease and elevated inflammatory activity . The role of underlying disease in the increased lymphoma risk therefore should not be underestimated. Several studies have reported increased lymphoma risk in patients receiving TNF-α antagonists [119–122]. However, no increased risk of lymphoma in RA patients has been observed with anti-TNF-α compared with conventional immunosuppressors [123,124]. Further, a Cochrane meta-analysis of 163 RCTs including 50 010 participants did not show an increased risk of lymphoma in RA patients treated with anti-TNF-α therapy: OR 0.53 (95% CI 0.17–1.66) .
In RA patients treated with anti-TNF-α (and possibly other immunosuppressors), two opposing effects are proposed to modulate lymphoma risk. First, treatment efficacy reduces disease activity, and consequently lymphoma risk. Second, immunosuppressive treatments may have a detrimental effect by reducing T-cell immunosurveillance against cancers, thereby potentially increasing the risk of certain lymphomas (e.g. Epstein–Barr virus associated). Which of these two effects predominates over the other, and whether these effects depend on the nature of the anti-TNF agent used, remains to be seen.
Patients should be followed closely in order to survey the occurrence of malignancies. Standard screening and referral to a dermatologist are recommended, and considered decisions should be made as to the choice of biological therapy, depending on patient comorbidities, age, disease activity, and associated treatments.
Biological therapies and autoimmunity
Despite a broad array of rare observations reported mainly in clinical case reports, such as anti-TNF-α-induced lupus-like syndrome, granulomatosis, vascularities, or demyelinating reactions, there is no strong clinical evidence to support TNF-α antagonists as causing these disorders . For instance, in the case of lupus, these often are cutaneous hypersensitivity reactions with anti-double-stranded DNA antibodies [127,128]. Some often relatively benign case reports of ‘true’ SLE have been observed.
A paradoxical clinical response to anti-TNF-α therapy is the development of psoriasis. Several hundreds of authentic cases have been described . The potential mechanisms involved in this response appear to involve a disrupted cytokine balance following inhibition of TNF-α, which causes upregulation of plasmacytoid dendritic cells and production of IFN-α in predisposed individuals [129–131]. A second hypothesis is a ‘superagonist TNF’ effect of circulating anti-TNF/TNF complexes .
One may surmise that these paradoxical effects may not be specific to Fab or TNF-α itself, but could also be related to the common Fc fraction of immunoglobulins because psoriasis cases have also been described with rituximab therapy .
In summary, the increased risk of infection in RA patients receiving biological agents is established. These treatments do not seem to confer an increase in cancer risk except for skin cancer. The impact of these therapies on induction of autoimmune reactions is limited. Despite potential complications of biological therapies in autoimmune diseases, the risk–benefit ratio remains very satisfactory. For instance, TNF-α antagonists improves survival of RA patients by reducing cardiovascular mortality .