A new mechanism for immune destruction of red blood cells?

In 1997, we reported three cases of autoimmune hemolytic anemia (AIHA) associated with immunoglobulin (Ig)M warm autoantibodies directed at determinants on or associated with glycophorins.1 The antibodies were anti-Ena, -Wrb, -Pr. All three patients had severe hemolytic anemia with evidence of hemoglobinemia and hemoglobinuria. Two of the three patients (associated with the anti-Ena and -Wrb) died as a result of the hemolysis. In this article we emphasized how the serology did not correlate well with the degree of in vivo hemolysis. The direct antiglobulin tests (DATs) were 3+, 1+, ½+ respectively, due to red blood cell (RBC)-bound C3; IgM was detected (1+) on the RBCs of one patient; and no RBC-bound IgG was detected in any of the cases. The sera of all three patients agglutinated RBCs at 37°C, but the anti-Ena and anti-Wrb were dependent, or enhanced by, a low pH (6.5) or the presence of 30% albumin. The antibodies showed another unusual characteristic in that they reacted better at 30°C than at 37°C. In 2009, we published serologic findings in 49 cases of AIHA associated with IgM warm autoantibodies; 4 of 17 (24%), studied for specificity were antibodies directed against epitopes on or associated with glycophorins (anti-N, -Pr, -Ena, -Wrb).2 Because of the above poor correlations with the serology and clinical status of the patients we studied, we were very interested to read two publications by Brain and colleagues in 20023 and 2004.4 Like us, Brain and colleagues3 were also puzzled by the unimpressive serologic findings in a patient with severe AIHA associated with an IgM anti-Pr autoantibody. The DAT was weakly positive due to RBC-bound C3; the serum contained only an anti-Pr cold agglutinin of low titer and high thermal amplitude (titer of 32 at 4°C, titer of 1 at 30°C, and nonreactive at 37°C). The patient had a life-threatening hemolytic anemia (HA) with a 3-day t1/2 of 51Cr-labeled RBCs. Dr Brain is an expert in the nonimmune hemolytic anemias and so approached the problem in a different way than a serologist. As Pr is present on glycophorins (mainly glycophorin A [GPA], but also present on glycophorin B [GPB] and possibly glycophorins C and D), he wondered if the anti-Pr was affecting GPA because, as earlier work had shown, attachment of bifunctional ligands (lectins and monoclonal antibodies) to GPA make RBCs more resistant to shear forces and reduce the mobility of Band 3 in the RBC membrane.5,6 Brain and coworkers3 went on to show that GPA-reactive lectins (Maclura pomifera and wheat germ), antibodies to GPA (human anti-Pr and mouse monoclonal anti-GPA), would cause hemolysis of human RBCs in vitro, without involving complement. The lysis appeared to be due to sodium, and probably calcium, entering the RBCs. They suggested that antibodies directed at epitopes on GPA might aggregate the GPA causing destabilization of the phospholipid bilayer of the RBC membrane, leading to an increase in cation permeability. In a later publication Brain and coworkers4 showed that calcium entered approximately 28% of the RBCs when the RBCs were under stress; they suggested that mechanosensitive channels could be activated by the changes in lateral pressure within the phospholipid bilayer of the RBC membrane, such as those caused by GPA–anti-GPA interactions. In this issue of TRANSFUSION, Brain and colleagues7 continue the story. They had another patient, a 29-year-old man, with a HA associated with cold agglutinins. The DAT was weakly positive with anti-C3 and negative with anti-IgG. The cold agglutinin titer (CAT) was abnormal, but not very high (titer of 256 at 4°C); the cold agglutinin reacted up to 30°C, thus fulfilling the criteria for cold agglutinin syndrome (CAS).8 The antibody had several unusual characteristics; it was a polyclonal IgG cold agglutinin with anti-Pr specificity. The case was very similar to one described by Curtis and coworkers,9 which stimulated Brain and coworkers7 to treat the patient with cyclophosphamide, as used by Curtis and coworkers,9 when they had little success with transfusions, plasma exchange, steroids, IVIG, and rituximab; Curtis and coworkers9 also used splenectomy. Both patients responded well and had complete remissions. Brain and colleagues7 confirmed that calcium entered RBCs when the RBCs were treated with the patient's IgG anti-Pr and a rabbit IgG anti-GPA; high-titer IgM anti-I from a patient with CAS did not cause the same phenomenon. The authors added a new finding that after exposure to antibodies to epitopes on GPA, phosphatidylethanolamine (PE) was exposed on the RBC membrane. These changes appeared to occur only in a subpopulation of the RBCs. Nevertheless, Brain and colleagues7 feel that this antibody-induced mechanism could cause HA independent of complement-mediated hemolysis. There are many publications concerning anti-Pr in the literature; many of these were found in patients with CAS, but unfortunately, many of them concentrate on serologic and biochemical characteristics of the antibodies.10-13 Nevertheless, on reviewing the literature on association of anti-Pr with HA, I found 27 reports with reasonable evidence for an associated HA.1-3,7,9,14-35 Anti-Pr was first described by Marsh and Jenkins in 1968;36 they called it anti-Sp1 (Sp for species) because all human RBCs reacted. Similar antibodies were described by Roelcke37 in 1969; he named his specificity HD for Heidelberg. Later, it was suggested that both terms be replaced with Pr (protease sensitive); anti-Pr do not react with protease-treated RBCs and most do not react with sialidase-treated RBCs. There are at least eight subspecificities (e.g., Pr1h, Pr1d, Pr2, Pr3h, Pr3d, Pra, PrM, PrN; h = human, d = dog).11,12 Anti-Pra do not react with protease-treated RBCs, but will react with sialidase-treated RBCs and can easily be confused with anti-Ena.8 Of the 27 cases of anti-Pr associated with HA, 14 (52%) were associated with an IgM anti-Pr cold agglutinin; 12 (44%) were associated with an IgG agglutinin; and one was associated with an IgA Pr agglutinin. Of the 14 patients with IgM anti-Pr, five (36%) had a low CAT (4°C) of 32 or less, and nine (64%) had a CAT of 128 to 32,000. All of the antibodies with a low CAT reacted up to 30 or 37°C. One report contained no DAT result, three were DAT-negative, and 10 of the 13 (77%) tested had RBC-bound C3 but no IgG detected. Of eight antibodies tested for clonality, all were monoclonal IgM (seven were κ and three were λ). Six of the 12 (50%) patients with IgG anti-Pr had abnormal CATs (80-2048), one had no cold agglutinins, and five (42%) had titers of 4 to 32. Two of 12 reports did not give DAT results; 3 of 10 tested were DAT-negative; four had only RBC-bound C3, two had RBC-bound IgG and C3, and one had only RBC-bound IgG. Four of the 12 anti-Pr were tested for clonality, all were monoclonal (three κ and one λ). The CAT was 256. The DAT showed RBC-bound C3, no IgG but RBC-bound IgA was not measured. The IgA anti-Pr was monoclonal (κ). Unfortunately, the clinical details in many of these reports were often brief/incomplete. Sometimes a diagnosis of CAS was given, but no supporting data (clinical or serologic), except the CAT, were given. Seven (26%) of the 27 HAs were associated with vaccination or a defined infection. Among the IgM anti-Pr were two cases associated with varicella/chicken pox25,29 and one case associated with DPT vaccination.33 Among the IgG anti-Pr were two cases of rubella,27,32 one case of varicella,21 and one case after DPT vaccination.34 There are other cases of these associations without HA in the literature.10,11 The above data on the 27 reports of HA associated with anti-Pr can be used to support or question the hypothesis by Brain and colleagues.3,4,7 Although 10 (77%) of 13 cases associated with IgM anti-Pr, that had DAT results, had RBC-bound C3, 3 of 13 had negative DATs with anti-IgG and anti-C3; only 5 of 14 (36%) suffered with severe HA (e.g., hemoglobinemia/hemoglobinuria). The other 64% had only mild to moderate HA. Most of the patients would be classified as CAS. I do not think that there is much doubt that complement played a major role in the HA of the five severe cases, but it is uncertain how much a role it played in the other nine cases. Most of these cases were typical of CAS (mild to moderate HA with no acute hemoglobinemia/hemoglobinuria, consistent with extravascular complement-mediated destruction by macrophages, mainly in the liver), 3 of the 13 had negative DATs, which either means that the anti-C3 were poor quality or possibly that the hypothesis of Brain and colleagues was causing the HA. After reviewing the literature, I was surprised to find that 44% of the HAs were associated with IgG anti-Pr; all but one of the five tested for clonality were monoclonal (three of the four were κ and one was λ). Two reports did not give DAT results. Of the 10 patients tested, four had RBC-bound C3 (without IgG), two had IgG plus C3, one had RBC-bound IgG with no C3, and three were negative. Thus, 50% had RBC-bound C3. Six of the 12 had IgG cold agglutinins with normal CATs of less than 64 and six had CATs of 128 to 2048. As might be expected, there appeared to be less complement involvement than the IgM group (77% vs. 50%), and RBC-bound IgG was detected only in 3 of 10 (30%) patients. Thus, in many cases, it was difficult to explain the sometimes severe hemolysis by IgG/C3 interactions. My personal opinion is that individual patients might have one or both mechanisms as causative mechanisms for hemolysis. This concept would explain the varied serologic findings associated with HA and antibodies to determinants on or associated with glycophorin A (e.g., anti-M, -N, -Pr, -Ena, -Wrb), including apparent intravascular lysis with minimal serologic signs of complement activation. All of these antibody specificities have been described as autoantibodies and all, except anti-Pr, have been found as alloantibodies.8 Most anti-Pr, -Ena, -Wrb have been shown to be capable of activating complement and, on occasion, have been associated with acute intravascular hemolysis.8 Anti-M and -N rarely activate complement and are far less common than the others as the cause of AIHA.8 We presented data in 2008 on a patient with a warm-type AIHA associated with anti-Ena who died after sudden intravascular hemolysis (hemoglobinuria and hemoglobinemia), renal failure, and disseminated intravascular coagulation.38 The patient's RBCs were strongly sensitized with IgG and C3; her plasma contained an anti-Ena, strongly reactive by the antiglobulin test. This antibody appeared to activate complement, causing the fatal chain of clinical reactions known to be associated with complement activation. Although calcium/PE-mediated destruction may have been occurring, I would not think that this mechanism was the main offender in this particular patient. There are some interesting articles in the literature showing a relationship of calcium loading of the RBCs and sensitivity to complement. Test and coworkers39 showed that calcium loading of RBCs caused a preferential loss of RBC membrane proteins having a glycosylphosphatidylinositol anchor, thus making them similar to RBCs from patients with paroxysmal nocturnal hemoglobinemia, which have increased sensitivity to complement. This observation suggests a possibility for the hypothesis by Brain and coworkers7 to be combined with complement-mediated hemolysis, which may not be apparent by routine in vitro tests. The emphasis of this editorial has been on anti-Pr because the patients from Brain's group had anti-Pr, but it must be remembered that M, N, Mg, Mc, Ena, T, and Tn are also on GPA; Pr, T, Tn, “N,” He, Mv, U, S, and s are on GPB; and Ge is on GPC and GPD. It will be interesting to see if the same events associated with GPA antigen-antibody reactions occur with the other glycophorins. It would also be of interest to explain why anti-M and -N rarely activate complement or cause HA. Is it because of the position of the epitopes on GPA (M and N are furthest away from the RBC membrane; Pr is distributed along the GPA structure, and Ena is close to the membrane)? In 2007 I chose to be iconoclastic in the choice of my Blundell Award Lecture title. I entitled it “Do we really understand immune RBC destruction?”40 After more than 30 years of research in this area, I emphasized that there is so much we do not understand. I said that we must encourage investigators to think “outside of the box” (e.g., rather than trying to explain in vivo events using serologic results). In the publication of this lecture,40 I commended Michael Brain (ex–Chief of Hematology at McMaster University, Canada), who is an expert on nonimmune hemolytic anemias, for doing just that in his 2002/2004 articles3,4 and now I commend him again for continuing the research in his “retirement.” I have always taught my students that antibodies do not harm RBCs directly; they only cause damage indirectly, by activating complement efficiently enough to cause direct lysis of the RBCs, or RBC-bound IgG and/or complement can interact with receptors on macrophages in the reticuloendothelial system. I was wrong. The articles by Brain and coworkers3,4,7 suggest an antibody-mediated mechanism that can sometimes be independent of complement or macrophage classical mechanisms, or possibly be part of a combined effect. The author declares no conflict of interest.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.