Warm-Antibody Autoimmune Hemolytic Anemia
The etiology of AHA is unknown. In warm-antibody AHA, the autoantibodies that mediate RBC destruction are predominantly (but not exclusively) IgG globulins possessing relatively high binding affinity for human RBCs at 37°C. As a result, the major share of plasma autoantibody is bound to the patient’s circulating RBCs. Eluates prepared from the patient’s washed, autoantibody-coated RBCs constitute an important source of purified autoantibody for investigation of specificity, immunoglobulin structure, or other properties. In addition, sera from patients with warm AHA often are used in blood banks for crossmatching and for general screening of antibody specificity. The quantity of such autoantibody in serum may be low and in some cases may not reflect the full spectrum of anti-RBC specificity revealed in concurrently prepared RBC eluates.124
In patients with primary AHA, erythrocyte autoantibodies are the only recognizable immunologic aberration. Furthermore, the autoantibodies of any one patient often are specific for only a single RBC membrane protein (see “Serologic Features” below). The narrow spectrum of autoreactivity suggests the mechanism underlying AHA development in such patients is not secondary to a generalized defect in immune regulation. Rather, these patients may develop warm-antibody AHA through an aberrant immune response to a self-antigen or to an immunogen that mimics a self-antigen.
In patients with secondary AHA, the disease may be associated with a fundamental disturbance in the immune system, for example, when in the setting of lymphoma, CLL, SLE, primary agammaglobulinemia (common variable immunodeficiency), or hyper-IgM immunodeficiency syndrome. In these settings, warm-antibody AHA most likely arises through an underlying defect in immune regulation, although the contribution of an aberrant immune response to self-antigen cannot be excluded. AHA seems especially frequent in patients with low-grade lymphoma or CLL treated with fludarabine 88,89 or 2-chlorodeoxyadenosine (cladribine).90 The T-lymphocytopenia induced by these drugs may exacerbate the preexisting tendency of patients to form autoantibodies.
A long-recognized but poorly understood phenomenon, the development of AHA or a positive DAT following RBC transfusion, has received renewed interest lately.132,133 Although generally transient, the positive DAT may persist for up to 300 days in some transfusion recipients, long after any transfused RBCs have disappeared.134,135 It is not clear whether this represents true autoimmunity or some other mechanism, for example, microchimerism resulting from temporary engraftment of passenger memory lymphocytes from the RBC donor.132
A still unexplained observation is that certain drugs, such as α-methyldopa, can induce warm-reacting IgG anti-RBC autoantibodies in otherwise normal persons. The autoantibodies induced by α-methyldopa have Rh-related serologic and immunochemical136 specificity similar to that of autoantibodies arising in many patients with “spontaneous” AHA. A critical difference is that the drug-associated autoantibodies subside when the drug is discontinued, suggesting that (1) the latent potential to form this type of anti-RBC autoantibody is present in many immunologically normal individuals, and (2) the steps required to generate such autoantibodies do not necessarily create a sustained autoimmune state. On the other hand, maintenance of chronic idiopathic AHA may be either secondary to a continuing (but unknown) stimulus or induced by a short stimulus to which the patient continues to respond.
Normal subjects sometimes have a positive DAT when they volunteer to donate blood.137,138 The positive DAT in these normal donors often results from warm-reacting IgG autoantibodies, similar in serologic specificity131 and in IgG subclass137 to the autoantibodies occurring in AHA. Although many of these donors remain DAT-positive without developing overt hemolytic anemia, a few have been documented to develop AHA.137,138 The prevalence of positive DATs in normal blood donors is approximately 1 in 10,000 donors.137,139 Because blood donation per se likely does not contribute to an increased risk of developing autoantibodies, the 1 in 10,000 proportion likely is the approximate frequency of positive DATs in the entire population. A proportion of patients who present with clinically overt primary AHA may come from a subset of asymptomatic individuals who are DAT-positive, but this notion is not established.
Several concepts have been developed to explain immunologic tolerance to self-antigens.140,141,142,143 Relevant to warm-antibody AHA, membrane-bound antigens expressed in a multivalent array at high concentration may induce tolerance by effecting clonal deletion of autoreactive B cells.144 Both the Rh-related and the non-Rh types of RBC antigens targeted by AHA autoantibodies (see “Serologic Features” below) are expressed normally by human fetal erythrocytes, as early as 10 to 12 weeks of life.145 However, because new B cells develop daily in the marrow throughout life and because B cells may somatically mutate their Ig receptors, self-tolerance in the B cell compartment is never assured. Analogy to observations in NZB (New Zealand black) mice146,147 suggests the peritoneal cavity is a privileged compartment that shelters autoreactive B cells from host RBCs, allowing them to escape deletion, later to produce anti-RBC autoantibodies with appropriate T-cell help. The strong predominance of IgG antibodies in AHA suggests B-cell isotype switching, which is consistent with the idea of an antigen-driven process. Moreover, because T-cell help is necessary for inducing B-cell isotype switching, the pathway(s) to autoantibody induction in AHA also may involve an abnormal or unique mode of antigen presentation to T cells.148
Origin of Cold Agglutinins
A high proportion of monoclonal IgM cold agglutinins with either anti-I or anti-i specificity have heavy-chain variable regions encoded by IGHV4–34 (immunoglobulin heavy chain variable region), formerly designated IGHV4.21.143,149,150,151 IGHV4–34 encodes a distinct idiotype identified by the rat monoclonal antibody 9G4. This idiotype is expressed both by the cold agglutinins themselves and on the surface immunoglobulin of B cells synthesizing cold agglutinins or related immunoglobulins possessing IGHV4–34 sequences.152 Using the 9G4 monoclonal antibody as a probe, this idiotype was found not only in a very high proportion of circulating B cells and marrow lymphoplasmacytoid cells of patients with lymphoma-associated chronic cold agglutinin disease, but also in a smaller proportion of B cells in the blood and lymphoid tissues of normal adult donors and in the spleens of 15-week human fetuses.152 These data suggest B cells expressing the IGHV4–34 gene (or a closely related sequence) are present throughout ontogeny. Therefore, chronic cold agglutinin disease may represent a marked, unregulated expansion of a subset (clone) of such B cells.
Light-chain V-region gene use in anti-I cold agglutinins is highly selective. A strong bias toward use of the κ III variable region subgroup (Vκ-III) is observed.150,151,152,153 Light-chain selection among anti-i cold agglutinins, however, is much more variable and includes the λ type.150,151,152,153,154
Observations that pathologic cold agglutinins are synthesized with distinct and highly selected V-region sequences must be viewed against the background of two other subsequent observations. First, IGHV4–34 or related IGHV genes also may encode the heavy-chain variable regions of other types of antibodies, such as rheumatoid factor autoantibodies and alloantibodies to a variety of blood group antigens, including polypeptide determinants such as Rh.155 Second, normal human antibodies to an exogenous carbohydrate antigen, Haemophilus influenzae type b capsular polysaccharide, also are encoded by a restricted set of IGHV genes156 and Ig light-chain V genes.157 Thus, regulation of Ig gene use for production of anti-I or anti-i cold agglutinins may not differ fundamentally from normal antibody formation to other carbohydrate antigens.
In the setting of B-cell lymphoma or Waldenström macroglobulinemia, cold agglutinins may be produced by the malignant clone itself. Two patients with lymphoma and monoclonal cold agglutinin were identified as having a karyotypically abnormal B-cell clone that produced a cold agglutinin identical to that found in their sera.158,159 Trisomy 3 has been the most frequently observed karyotypic abnormality in patients with non-Hodgkin lymphoma and cold agglutinins.158,160
Normal human sera generally have naturally occurring polyclonal cold agglutinins in low titer (usually 1/64 or less).10 Otherwise healthy persons may develop elevated titers of cold agglutinins specific for I/i antigens during certain infections (e.g., M. pneumoniae, Epstein-Barr virus, cytomegalovirus). In contrast to other forms of cold agglutinin disease, hyperproduction of these postinfectious cold agglutinins is transient. Some evidence indicates postinfectious cold agglutinins may be less clonally restricted than those occurring in chronic cold agglutinin disease,161 but this finding is not universal.162 Whether IGHV4–34 also encodes most heavy-chain variable regions of all naturally occurring or postinfectious cold agglutinins remains to be determined.
The increased production of cold agglutinins in response to infection with M. pneumoniae may be secondary to the fact that the oligosaccharide antigens of the I/i type serve as specific Mycoplasma receptors.163 This process may lead to altered antigen presentation involving a complex between a self-antigen (I/i) and a non–self-antigen (Mycoplasma). Alternatively, the anti-i cold agglutinins may arise as a consequence of polyclonal B cell activation, as occurs in infectious mononucleosis (Chap. 82).
The mechanism(s) whereby dissimilar infectious agents (e.g., spirochetes and several types of virus) induce the immune system to produce Donath-Landsteiner antibodies with specificity for the human P blood group antigen (see “Serologic Features” below) is not known.
Pathogenic Effects of Warm Antibodies
Warm autoantibodies to RBCs in AHA are pathogenic. In contrast to autologous RBCs, labeled RBCs lacking the antigen targeted by the autoantibodies may survive normally in patients with warm-antibody AHA.10,164,165 Furthermore, transplacental passage of IgG anti-RBC autoantibodies from a mother with AHA to the fetus can induce intrauterine or neonatal hemolytic anemia.166 Finally, despite notable exceptions and differences related to IgG subclass of the autoantibody, in general, an inverse relationship between the quantity of RBC-bound IgG antibody and RBC survival is noted in serial studies performed on animals and patients.167,168,169,170,171,172
In warm-antibody AHA, the patient’s RBCs typically are coated with IgG autoantibodies with or without complement proteins. Autoantibody-coated RBCs are trapped by macrophages in the Billroth cords of the spleen and, to a lesser extent, by Kupffer cells in the liver (Chap. 68).164,167,168,170,171,172,173,174 The process leads to generation of spherocytes and fragmentation and ingestion of antibody-coated RBCs.175,176 The macrophage has surface receptors for the Fc region of IgG, with preference for the IgG1 and IgG3 subclasses177,178 and surface receptors for opsonic fragments of C3 (C3b and C3bi) and C4b.179,180,181 When present together on the RBC surface, IgG and C3b/C3bi appear to act cooperatively as opsonins to enhance trapping and phagocytosis.170,171,180,181,182,183,184 Although RBC sequestration in warm-antibody AHA occurs primarily in the spleen,164,171,172,173 very large quantities of RBC-bound IgG167,168,174 or the concurrent presence of C3b on the RBCs167,170,171 may favor trapping in the liver.
Interaction of a trapped RBC with splenic macrophages may result in phagocytosis of the entire cell. More commonly, a type of partial phagocytosis results in spherocyte formation. As RBCs adhere to macrophages via the Fc receptors, portions of RBC membrane are internalized by the macrophage. Because membrane is lost in excess of contents, the noningested portion of the RBC assumes a spherical shape, the shape with the lowest ratio of surface area to volume.175,176,185 Spherical RBCs are more rigid and less deformable than normal RBCs. As such, spherical RBCs are fragmented further and eventually destroyed in future passages through the spleen. Spherocytosis is a consistent and diagnostically important hallmark of AHA,186 and the degree of spherocytosis correlates well with the severity of hemolysis.10
Direct complement-mediated hemolysis with hemoglobinuria is unusual in warm-antibody AHA, even though many warm autoantibodies fix complement. The failure of C3b-coated RBCs to be hemolyzed by the terminal complement cascade (C5–C9) has been attributed, at least in part, to the ability of complement regulatory proteins (factors I and H) in plasma and C3b receptors on the RBC surface to alter the hemolytic function of cell-bound C3b and C4b.187 Glycosylphosphatidylinositol-linked erythrocyte membrane proteins, such as decay-accelerating factor (DAF; CD55) and homologous restriction factor (HRF; CD59), may limit the action of autologous complement on autoantibody-coated RBCs.188,189,190 DAF inhibits the formation and function of cell-bound C3-converting enzyme,188 thus, indirectly limiting formation of C5-converting enzyme. HRF, on the other hand, impedes C9 binding and formation of the C5b–9 membrane attack complex.189
Cytotoxic activities of macrophages and lymphocytes also may play a role in the destruction of RBCs in warm-antibody AHA. Monocytes can lyse IgG-coated RBCs in vitro independently of phagocytosis.191,192 Cell-bound complement is neither necessary nor sufficient for such cytotoxicity, but bound C3b/C3d can potentiate the effects of IgG.192 In one study, cytotoxicity, but not phagocytosis, was inhibited by hydrocortisone in vitro.191 Lymphocytes also can lyse IgG antibody-coated RBCs in vitro.193,194,195 The relative contribution of antibody-dependent monocyte- and lymphocyte-mediated cytotoxicity to RBC destruction in patients with warm-antibody AHA is not known.
Pathogenic Effects of Cold Agglutinins and Hemolysins
Most cold agglutinins are unable to agglutinate RBCs at temperatures higher than 30°C. The highest temperature at which these antibodies cause detectable agglutination is termed the thermal amplitude. The value varies considerably among patients. Generally, patients with cold agglutinins with higher thermal amplitudes have a greater risk for cold agglutinin disease.9 For example, active hemolytic anemia has been observed in patients with cold agglutinins of modest titer (e.g., 1:256) and high thermal amplitudes.196
The pathogenicity of a cold agglutinin depends upon its ability to bind host RBCs and to activate complement.10,182,197,198 This process is called complement fixation. Although in vitro agglutination of the RBCs may be maximal at 0 to 5°C, complement fixation by these antibodies may occur optimally at 20 to 25°C and may be significant at even higher physiologic temperatures.10,196,197 Agglutination is not required for the process. The great preponderance of cold agglutinin molecules are IgM pentamers, but small numbers of IgM hexamers with cold agglutinin activity are found in patients with cold agglutinin disease. Hexamers fix complement and lyse RBCs more efficiently than do pentamers, suggesting that hexameric IgM plays a role in the pathogenesis of hemolysis in these patients.199
Cold agglutinins may bind to RBCs in superficial vessels of the extremities, where the temperature generally ranges between 28 and 31°C, depending upon ambient temperature.200 Cold agglutinins of high thermal amplitude may cause RBCs to aggregate at this temperature, thereby impeding RBC flow and producing acrocyanosis. In addition, the RBC-bound cold agglutinin may activate complement via the classic pathway. Once activated complement proteins are deposited onto the RBC surface, the cold agglutinin need not remain bound to the RBCs for hemolysis to occur. Instead, the cold agglutinin may dissociate from the RBCs at the higher temperatures in the body core and again be capable of binding other RBCs at the lower temperatures in the superficial vessels. As a result, patients with cold agglutinins of high thermal amplitude tend toward a sustained hemolytic process and acrocyanosis.201 In contrast, patients with antibodies of lower thermal amplitude require significant chilling to initiate complement-mediated injury of RBCs. This sequence may result in a burst of hemolysis with hemoglobinuria.201 Combinations of these clinical patterns also occur. Cold agglutinins of the IgA isotype, an isotype that does not fix complement, may cause acrocyanosis but not hemolysis.202 Thus, the relative degree of hemolysis or impeded RBC flow is influenced significantly by the properties and quantity of the cold agglutinins in a given patient.
Complement fixation by cold agglutinins may affect RBC injury by two major mechanisms: (1) direct lysis and (2) opsonization for hepatic and splenic macrophages. Both mechanisms probably operate to varying degrees in any patient. Direct lysis requires propagation of the full C1 to C9 sequence on the RBC membrane. If this process occurs to a significant degree, the patient may experience intravascular hemolysis leading to hemoglobinemia and hemoglobinuria. Intravascular hemolysis of this severity is relatively rare because phosphatidylinositol-linked RBC membrane proteins (DAF and HRF) protect against injury by autologous complement components. Thus, the complement sequence on many RBCs is completed only through the early steps, leaving opsonic fragments of C3 (C3b/C3bi) and C4 (C4b) on the cell surface. The fragments provide only a weak stimulus for phagocytosis by monocytes in vitro.184,203 However, activated macrophages may ingest C3b-coated particles avidly.204 Accordingly, RBCs heavily coated with C3b (and/or C3bi) may be removed from the circulation by macrophages either in the liver or, to a lesser extent, the spleen.171,197,205,206 The trapped RBCs may be ingested entirely or released back into the circulation as spherocytes after losing plasma membrane.
In vivo studies of the fate of 51Cr-labeled C3b-coated RBCs170,197,205,206 indicate many of the erythrocytes trapped in the liver or spleen gradually may reenter the circulation. The released cells generally are coated with the opsonically inactive C3 fragment C3dg. Conversion of cell-bound C3b or C3bi to C3dg results from the action of the naturally occurring complement inhibitor factor I in concert with factor H or CR1 receptors.181 The surviving C3dg-coated RBCs circulate with a near-normal life span170,197,205,206 and are resistant to further uptake of cold agglutinins or complement.197,205,207 However, C3dg-coated RBCs also may react in vitro with anticomplement (anti-C3) serum in the DAT. In fact, most of the antiglobulin-positive RBCs of patients with cold agglutinin disease are coated with C3dg.
In paroxysmal cold hemoglobinuria, the mechanism of hemolysis probably parallels in vitro events (see “Serologic Features” below). During severe chilling, blood flowing through skin capillaries is exposed to low temperatures. The Donath-Landsteiner antibody and early acting complement components presumably bind to RBCs at the lowered temperatures. Upon return of the cells to 37°C in the central circulation, the cells are lysed by propagation of the terminal complement sequence through C9. The Donath-Landsteiner antibody itself dissociates from the RBCs at 37°C. Erythrocyte membrane proteins that restrict C5b–9 assembly (e.g., HRFs) may be less effective in controlling Donath-Landsteiner antibody-initiated complement activation than that initiated by cold agglutinins.
Pathogenesis of Drug-Mediated Immune Injury
Table 54–3 summarizes the three mechanisms of drug-mediated immune injury to RBCs. Drugs also may mediate protein adsorption to RBCs by nonimmune mechanisms, but RBC injury does not occur.
Table 54–3.Major Mechanisms of Drug-Related Hemolytic Anemia and Positive Direct Antiglobulin Tests ||Download (.pdf) Table 54–3. Major Mechanisms of Drug-Related Hemolytic Anemia and Positive Direct Antiglobulin Tests
| ||Hapten/Drug Adsorption ||Ternary Complex Formation ||Autoantibody Binding ||Nonimmunologic Protein Adsorption |
|Prototype drug ||Penicillin ||Quinidine ||α-Methyldopa ||Cephalothin |
|Role of drug ||Binds to red cell membrane ||Forms ternary complex with antibody and red cell membrane component ||Induces formation of antibody to native red cell antigen ||Possibly alters red cell membrane |
|Drug affinity to cell ||Strong ||Weak ||None demonstrated to intact red cell but binding to membranes reported ||Strong |
|Antibody to drug ||Present ||Present ||Absent ||Absent |
|Antibody class predominating ||Immunoglobulin (Ig) G ||IgM or IgG ||IgG ||None |
|Proteins detected by direct antiglobulin test ||IgG, rarely complement ||Complement ||IgG, rarely complement ||Multiple plasma proteins |
|Dose of drug associated with positive antiglobulin test ||High ||Low ||High ||High |
|Presence of drug required for indirect antiglobulin test ||Yes (coating test red cells) ||Yes (added to test medium) ||No ||Yes (added to test medium) |
|Mechanism of red cell destruction ||Splenic sequestration of IgG-coated red cells ||Direct lysis by complement plus splenic–hepatic clearance of C3b-coated red cells ||Splenic sequestration ||None |
Hapten or Drug Adsorption Mechanism
This mechanism applies to drugs that can bind firmly to proteins, including RBC membrane proteins. The classic setting is very-high-dose penicillin therapy,28,29,30,31,32,33,34 which is encountered less commonly today than in previous decades.
Most individuals who receive penicillin develop IgM antibodies directed against the benzylpenicilloyl determinant of penicillin, but this antibody plays no role in penicillin-related immune injury to RBCs. The antibody responsible for hemolytic anemia is of the IgG class, occurs less frequently than the IgM antibody, and may be directed against the benzylpenicilloyl,31 or, more commonly, nonbenzylpenicilloyl determinants.28,29,30,32 Other manifestations of penicillin sensitivity usually are not present.
All patients receiving high doses of penicillin develop substantial coating of RBCs with penicillin. The penicillin coating itself is not injurious. If the penicillin dose is very high (10 to 30 × 106 units per day, or less in the setting of renal failure) and promotes cell coating, and if the patient has an IgG antipenicillin antibody, the antibody binds to the RBC-bound penicillin molecules and the DAT with anti-IgG becomes positive (see Fig. 54–1A).29,31,32,51,208 Antibodies eluted from patients’ RBCs or present in their sera react in the indirect antiglobulin test (IAT) only against penicillin-coated RBCs. This step is critical in distinguishing these drug-dependent antibodies from true autoantibodies.
Not all patients receiving high-dose penicillin develop a positive DAT reaction or hemolytic anemia because only a small proportion of such individuals produce the requisite antibody. Destruction of RBCs coated with penicillin and IgG antipenicillin antibody occurs mainly through sequestration by splenic macrophages.30,209 In some patients with penicillin-induced immune hemolytic anemia, blood monocytes and presumably splenic macrophages may lyse the IgG-coated RBCs without phagocytosis.210 Hemolytic anemia resulting from penicillin typically occurs only after the patient has received the drug for 7 to 10 days and ceases a few days to 2 weeks after the patient discontinues taking the drug.
Low-molecular-weight substances, such as drugs, generally are not immunogenic in their own right. Induction of antidrug antibody is thought to require firm chemical coupling of the drug (as a hapten) to a protein carrier. In the case of penicillin, the carrier protein involved in antibody induction need not be the same as the erythrocyte membrane protein to which penicillin is coupled in the effector phase, that is, when the IgG antipenicillin antibodies bind to penicillin-coated RBCs. In contrast to evidence on the ternary complex mechanism, no evidence indicates the drug-dependent antibodies responsible for RBC injury in this hapten/drug adsorption mechanism also recognize native erythrocyte membrane structures.
Cephalosporins have antigenic cross-reactivity with penicillin211,213 and bind firmly to RBC membranes, as do semisynthetic penicillins.33,34 Hemolytic anemia similar to that seen with penicillin has been ascribed to cephalosporins35,36,37,38,39 and some semisynthetic penicillins.33,34 Tetracycline40,41 and tolbutamide44,45 also may cause hemolysis by this mechanism. Carbromal causes positive IgG antiglobulin reactions by a similar mechanism,43 but hemolytic anemia has not been described.
Ternary Complex Mechanism: Drug–Antibody–Target Cell Interaction
Many drugs can induce immune injury not only of RBCs but also of platelets or granulocytes by a process that differs in several ways from the mechanism of hapten/drug adsorption (see Table 54–3). First, drugs in this group (see Table 54–2) exhibit only weak direct binding to blood cell membranes. Second, a relatively small dose of drug is capable of triggering destruction of blood cells. Third, cellular injury appears to be mediated chiefly by complement activation at the cell surface. The cytopathic process induced by such drugs previously has been termed the innocent bystander or immune complex mechanism. The terminology reflected the prevailing notion that, in vivo, drug–antibody complexes formed first (immune complexes) and then became secondarily bound to target blood cells as “innocent bystanders,” either nonspecifically or possibly via membrane receptors (e.g., Fcγ receptors on platelets or C3b receptors on red cells), with the potential for subsequent activation of complement by bound complexes.
The “immune complex” and “innocent bystander” terminology now seems less appropriate because of models developed from research on analogous drug-dependent platelet injury214,215,216 (Chap. 117) and a series of relevant serologic observations on drug-mediated immune hemolytic anemia. These studies suggest blood cell injury is mediated by a cooperative interaction among three reactants to generate a ternary complex (see Fig. 54–1B) involving (1) the drug (or drug metabolite in some cases), (2) a drug-binding membrane site on the target cell, and (3) antibody. For example, several patients possessed drug-dependent antibodies that exhibited specificity for RBCs bearing defined alloantigens such as those of the Rh, Kell, or Kidd blood groups. That is, even in the presence of drug, the antibodies were selectively nonreactive with human RBCs lacking the alloantigen in question.58,84,217,218,219 In each case, high-affinity drug binding to cell membrane could not be demonstrated. The drug-dependent antibody is thought to bind, through its Fab domain, to a compound neoantigen consisting of loosely bound drug and a blood group antigen intrinsic to the red cell membrane. Elegant studies on quinidine- or quinine-induced immune thrombocytopenia have demonstrated the IgG antibodies implicated in this disorder bind through their Fab domains, not by their Fc domains to platelet Fcγ receptors.220,221
The data elucidate how one patient with quinidine sensitivity may have selective destruction of platelets and another may have selective destruction of RBCs. This process occurs because the pathogenic antibody recognizes the drug only in combination with a particular membrane structure of the RBC (e.g., a known alloantigen) or of the platelet (e.g., α domain of the glycoprotein Ib complex). Therefore, at least in these cases, the target cell does not appear to be purely an innocent bystander. Binding of the drug itself to the target cell membrane is weak until the attachment of the antibody to both drug and cell membrane is stabilized. Yet the binding of the antibody is drug dependent. Such a three-reactant interdependent “troika” is unique to this mechanism of immune cytopenia.
The foregoing discussion depicting drugs as creating a “self + non-self” neoantigen on the target cell applies to the effector phase as opposed to the induction phase of the process. However, the same drug-binding membrane protein appears to be involved in forming the immunogen that induces the antibody, as evidenced by drug-dependent antibodies exhibiting selective reactivity with defined red cell alloantigens (carrier specificity).58,84,217,218,219 How this process is accomplished in the absence of evidence for strong, covalent binding of the drugs in this group to a host membrane protein remains to be elucidated.
RBC destruction by this mechanism may occur intravascularly after completion of the whole complement sequence, resulting in hemoglobinemia and hemoglobinuria. Some destruction of intact C3b-coated RBCs may be mediated by splenic and liver sequestration via the C3b/C3bi receptors on macrophages. The DAT is positive usually only with anticomplement reagents, but exceptions occur. Sometimes, however, the drug-dependent antibody itself can be detected on the RBCs if the offending drug (or its metabolites) is included in all steps of the antiglobulin test, including washing.222
A variety of drugs induce the formation of autoantibodies reactive with autologous (or homologous) RBCs in the absence of the instigating drug (see Tables 54–2 and 54–3). The most studied drug in this category has been α-methyldopa, an antihypertensive agent that no longer is commonly used.73,74,75,76 Levodopa and several unrelated drugs also have been implicated.39,46,59,61,68,77,78,79,80,81,82,83,84,85,86 Patients with CLL treated with pentostatin,87 fludarabine,88 or cladribine90 are particularly predisposed to autoimmune hemolysis, which usually is severe and sometimes fatal.
Positive DAT reactions (with anti-IgG reagents) in patients taking α-methyldopa vary in frequency from 8 to 36 percent. Patients taking higher doses of the drug develop positive reactions with greater frequency.73,75,76 A lag period of 3 to 6 months exists between the start of therapy and development of a positive antiglobulin test. The delay is not shortened when the drug is administered to patients who previously had positive antiglobulin tests while taking α-methyldopa.75
In contrast to the frequent observation of positive antiglobulin reactions, less than 1 percent of patients taking α-methyldopa exhibit hemolytic anemia.74 Development of hemolytic anemia does not depend on drug dose. The hemolysis usually is mild to moderate and occurs chiefly by splenic sequestration of IgG-coated RBCs. α-Methyldopa has been proposed to suppress splenic macrophage function in some patients, and normal survival of antibody-coated RBCs in such patients may be related, in part, to this effect of the drug.223
The DAT reaction usually is positive only for IgG.11 Occasionally, weak anticomplement reactions also are encountered.11 Patients with immune hemolytic anemia resulting from α-methyldopa therapy typically exhibit strongly positive DAT reactions and serum antibody, evidenced by the IAT reaction.11 Antibodies in the serum or eluted from RBC membranes react optimally at 37°C with unaltered autologous or homologous RBCs in the absence of drug (see Fig. 54–1C).74,76,224 Frequently the autoantibodies are reactive with determinants of the Rh complex,74,76,224 and at least some appear to target the same 34-kDa Rh-related polypeptide targeted by the autoantibodies in many cases of “spontaneously arising” AHA.136 Thus, distinguishing these drug-induced antibodies from similar warm-reacting autoantibodies in idiopathic AHA currently is not possible.
The mechanism by which a drug induces formation of an autoantibody is unknown. Radiolabeled α-methyldopa does not react directly with the membranes of intact human RBCs.76,225 However, both α-methyldopa and levodopa reportedly bind to isolated RBC membranes. Binding of the drug to membranes of intact RBCs is inhibited by RBC superoxide dismutase and probably by hemoglobin.225,226 Although not formally demonstrated, these drugs probably bind to membrane antigens of cells that are relatively hemoglobin free, for example, cells at the early proerythroblast stage or RBC stroma. In any case, the resulting altered membrane antigens then may induce autoantibodies. The concept that a drug–membrane compound neoantigen could lead to production of an autoantibody is supported by studies of patients receiving drugs unrelated to α-methyldopa. Patients simultaneously developed a drug-dependent antibody and an autoantibody, both of which showed specificity for the same RBC alloantigen.84 Another hypothesis is that α-methyldopa interacts with human T lymphocytes, resulting in loss of suppressor cell function.227 Subsequent studies, however, have failed to demonstrate any evidence for such a mechanism.228
Patients with CLL treated with the purine analogues fludarabine88,229,230 or cladribine90 may develop AHA. Risk factors for hemolysis include previous therapy with a purine analogue, high β2-microglobulin, a positive DAT prior to therapy, and hypogammaglobulinemia. Purine analogues are potent suppressors of T lymphocytes. These drugs may accelerate the preexisting T-cell immune suppression that normally occurs during progression of CLL, exacerbating the underlying tendency to autoimmunity in CLL. However, the degree of depletion of T-cell subsets is similar in patients who develop hemolysis and in patients who do not.
Nonimmunologic Protein Adsorption
Less than 5 percent of patients receiving cephalosporin antibiotics develop positive antiglobulin reactions11 as a result of nonspecific adsorption of plasma proteins to their RBC membranes.93,94,231 This process may occur within 1 to 2 days after the drug is instituted. Multiple plasma proteins, including immunoglobulins, complement, albumin, fibrinogen, and others, may be detected on RBC membranes in such cases.231,232 Hemolytic anemia resulting from this mechanism has not been reported. The clinical importance of this phenomenon is its potential to complicate crossmatch procedures unless the drug history is considered. Cephalosporin antibiotics also may induce RBC injury by the hapten mechanism, by the ternary complex mechanism, and by the autoantibody mechanism. The latter reactions are more serious but apparently occur less frequently than nonimmunologic protein adsorption.