What is the mechanism for acute hemolysis occurring in some patients after intravenous anti-D therapy for immune thrombocytopenic purpura?

In 1983, in the hypothesis section of Lancet, Salama and colleagues¹ suggested that the mechanism involved in the successful treatment of immune thrombocytopenic purpura (ITP) with intravenous immunoglobulin (IVIG) was due to contamination of IVIG with red blood cell (RBC) alloantibodies. Their suggestion was that such antibodies sensitize the recipient's RBCs and that the immunoglobulin (Ig)G-sensitized RBCs compete successfully with the IgG-sensitized platelets (PLTs) for Fc receptor sites on macrophages. In this short report they reported on six D+ ITP patients who were given intravenous (IV) anti-D to produce a mild hemolytic syndrome with a positive direct antiglobulin test (DAT); the PLT counts in three of these patients increased considerably. In 1986,² the same group reported a similar approach to treat 17 ITP patients; significant elevation of the PLT count occurred in 15 of the patients. Other reports from the United States³ and Canada^4,5 followed.

The Rh Institute in Winnipeg, Manitoba, Canada, had produced an intravenous anti-D (WinRho) for prophylaxis of Rh disease. This product was used to treat ITP in the United States³ and Canada^4,5 with good results and few adverse reactions; there was no acute hemolysis reported. The Food and Drug Administration (FDA) licensed Rho(D) immune globulin (anti-D IGIV, then WinRho^SD, currently WinRho^SDF[Cangene Corp., Winnipeg, Manitoba, Canada]) in 1995, for the treatment of ITP in D+, nonsplenectomized children with acute ITP and children and adults with ITP secondary to HIV infection. In 1997, Scaradavou⁶ reported on the successful results of treating 261 D+ ITP patients with WinRho^SD. The mean hemoglobin (Hb) decrease was 0.8 g/dL (SD1.5) 7 days after the first treatment. No patient required a RBC transfusion. Twenty-nine (16%) had Hb decreases of greater than 2.1 g/dL (10 of these had received multiple infusions with a total dose of greater than 60 µg/kg in the first week). No acute hemolysis was noted.

As clinical use widened, a few reports appeared of a more severe hemolytic anemia than expected in some patients. For instance, Barbolla and coworkers⁷ described a patient who received IV anti-D (15 µg/kg body weight) for 3 consecutive days. After the third dose, on the 14th day, the Hb level decreased from 15.4 to 7.2 g/dL, with a reticulocyte count of 12%, and many erythroblasts were noted on the blood film. The bilirubin level was only 1.9 µg/dL and no hemoglobinemia/hemoglobinuria was mentioned. The authors calculated that the dose of IV anti-D (3000 µg) would have been enough to destroy a maximum of 300 mL of D+ RBCs, but it was estimated the patient had destroyed 1600 mL of RBCs. Seven cases of acute onset hemoglobinuria were reported to the manufacturer of WinRho.⁸ In 2000 Gaines⁹ (from the FDA) reported that since 1999 the FDA had reviewed 15 reports of hemoglobinemia and/or hemoglobinuria after administration of WinRho, the only FDA-licensed IV anti-D available at that time. Seven of the patients required transfusion; eight patients developed renal problems (two required dialysis). One patient died as a result of the hemolytic anemia. There have been other reports^10-12 of renal insufficiency after treatment of ITP patients with IV anti-D since the report by Gaines.⁹ A very interesting case report was published in 2004 by Rewald and Francischetti¹³ where they described an ITP patient treated with IV anti-D minidoses for 8 years (4-5.5 µg/kg body weight with intervals of 10-20 days). For 8 years the treatment was uneventful (Hb remained approx. 8 g/dL). The patient then had marked hemoglobinuria each time after receiving anti-D. This continued for 6 months. Changing the lots/brand of anti-D did not prevent the continuing reactions. No renal problems developed. Treatment with anti-D was stopped, and except for occasional transfusions to treat the continuing pancytopenia, the patient kept stable for the 18 months before the report was published.

Mechanisms for the acute hemolysis could not be explained. Three widely different frequency estimates for acute hemolysis with hemoglobinemia and hemoglobinuria were calculated: clinical studies of 528 ITP-treated patients contained no reports; the original clinical trial found 2 in 137 patients (1 in 69); and FDA data (1995-1998) yielded 13 cases in 14,500 treated patients (0.1%, or 1 in 1145 incidence). In 2005, Gaines¹⁴ added to the problems encountered by reporting on patients who received IV anti-D for treatment of ITP and developed disseminated intravascular coagulation (DIC) in addition to acute hemolytic anemia. The FDA received six reports of DIC, through November 30, 2004, five of which involved fatalities.

INTRAVASCULAR VERSUS EXTRAVASCULAR HEMOLYSIS

I was glad to see that Gaines deliberately avoided the term “intravascular hemolysis,” instead using “acute hemoglobinemia/hemoglobinuria” in her 2000 and 2005 publications.^9,14 She discussed her rationale fully in her 2005 article. The distinction between an extravascular and intravascular mechanism of hemolysis was made as early as 1901.¹⁵ Intravascular hemolysis associated with an immune hemolytic anemia (e.g., hemolytic transfusion reactions [HTRs] or autoimmune hemolytic anemia) is where RBCs break down directly within the bloodstream, releasing their Hb into the plasma, leading to a reduction in haptoglobins and the presence of free Hb (hemoglobinemia); this Hb will be converted to the brown pigment methemalbumin. When the haptoglobin system is saturated, the Hb may be excreted in the urine (hemoglobinuria). When this type of hemolytic anemia is induced by antibody, damage to the RBCs is caused by activation of the complement cascade, leading to insertion of C5,6,7,8 through the bilipid layer of the RBC membrane, with subsequent entry of water and ions into the RBC, leading to cell rupture. Complement-associated intravascular lysis is rare. Few blood group antibodies activate complement efficiently enough to complete the complement cascade. The most efficient complement activating antibodies are anti-A and -B. All ABO antibodies are capable of activating complement in vitro, but not all cause intravascular lysis in vivo. Other antibodies that can activate complement are anti-Le^a, -Le^b, -Jk^a, -Jk^b, -Fy^a, -Fy^b, -K, -P, -PP₁P^k, -Vel, and -En^a; powerful anti-H; and anti-I/i. Not all examples are capable of activating complement, and most of them are not efficient enough to cause intravascular hemolysis in vivo. Extravascular hemolysis is caused by interactions of cells (e.g., macrophages) within the reticuloendothelial system (RES), in particular, the spleen and liver, and proteins (e.g., IgG1, IgG2, IgG3, IgA, C3b) on the RBC membrane. If these interactions occur in optimal conditions, then several events can occur: phagocytosis of RBCs, partial phagocytosis of RBCs with release of spherocytes, and cytotoxic attack of the RBCs by macrophages; one or more of these events may occur or dominate. Most blood group antibodies (e.g., Rh) destroy RBCs through this mechanism.

Haptoglobins are found to be decreased in most, if not all, immune hemolytic anemias, probably because Hb can escape from the RES during the above interactions.¹⁶ Thus, Dacie¹⁶ argued, many years ago, that “the concept of an absolute distinction between intravascular and extravascular hemolysis, which implies an absolute difference in hemolytic mechanisms, has therefore to be abandoned.” Nevertheless, the two terms are useful and remain in common use. This is acceptable as long as the users realize that hemoglobinemia and hemoglobinuria do not equate with complement-induced intravascular lysis. These signs only suggest that intravascular lysis might be occurring. As mentioned before, Gaines⁹ obviously concurred with this assessment. Her reports posed several major questions. Is powerful anti-D (e.g., WinRho) sometimes capable of activating complement and causing intravascular hemolysis, renal failure, and DIC? This triad certainly suggests complement activation, but is the culprit anti-D or something else in the product? Why do so few patients receiving this product have such severe hemolysis?

CAN ANTI-D ACTIVATE COMPLEMENT?

In this issue of TRANSFUSION, Gaines and colleagues¹⁷ describe a patient with acute hemoglobinemia/hemoglobinuria after administration of anti-D (WinRho). They try to answer the question of whether the anti-D activated complement, leading to intravascular hemolysis, as may occur in an acute HTR. To test for complement activation, they looked for in vitro hemolysis of D+ RBCs in the presence of WinRho and a source of human complement. They tested seven lots of anti-D, four of which had been implicated in acute hemolytic reactions. Their results were negative; even so, they are to be commended for publishing these negative results as authors usually only rush to publish when obtaining positive results.

The results, of course, fit with what we know about anti-D but not with the clinical results; even powerful anti-D usually does not activate complement. I know of only three reports of Rh alloantibodies that activated complement, detectable by routine methods. The first¹⁸ was an Rh antibody (anti-C+D), famous for its use in the early work on IgG allotypes (it is known as Ripley [Ri]). It was from an immunized male and had a titer of 9 to 18,000 and was found to hemolyze trypsin-treated RBCs. The second¹⁹ was an anti-D made by a person with a weak D phenotype. The antibody only had a titer of 16 and reacted strongly with enzyme-treated RBCs and untreated RBCs by indirect antiglobulin test (IAT) using anti-IgG and anticomplement. The third²⁰ was an anti-Rh27 (cE). The antibody appeared to be a naturally occurring IgM antibody; it reacted strongly with cE+ RBCs using anti-C3 by IAT.

Rh antibodies should be biologically capable of activating complement as they are usually IgG1, IgG3, or a mixture of both. It was suggested that they did not usually do this as IgG antibodies need to fall close enough together on the RBC membrane to form a “doublet” to activate the first component of complement (C1). IgG anti-A and -B can do this as there are 1 to 2 million A/B sites on A or B RBCs, but there are only 20 to 30,000 D sites on RBCs. Using a C1a fixation and transfer test, Rosse and Parker²¹ suggested that Rh antibodies could activate C1 if several Rh specificities were mixed (i.e., yielding more sites), but Freedman and colleagues²² could not confirm these findings using radiolabeled antibodies. I also could not confirm that mixtures of Rh antibodies would sensitize RBCs with complement using anti-C3 in an IAT, and I also could not sensitize D+ RBCs with complement using RhoGam (Ortho Clinical Diagnostics, Raritan, NJ) with a titer of 200,000 (unpublished observations).

Nevertheless, there are reports that, under certain conditions, anti-D can activate complement. In 1960, Klun and Muschel²³ showed that anti-D would fix complement if RBC stroma, rather than intact RBCs, were used. Hidalgo and coworkers²⁴ similarly showed that Rh antibodies would sensitize papain-treated RBC ghosts with complement. They suggested that it was due to aggregation of the Rh sites (yielding “doublets”) under these conditions. Jaffe²⁵ showed that by conventional methods, Rh antibodies showed no complement fixation, but when anti-D–sensitized RBCs were labeled with ⁵¹Cr and injected into patients and normal volunteers, the RBCs survived better in C4-deficient patients. In 1981, Hughes-Jones and Ghosh²⁶ showed that C1q would bind to anti-D–coated RBCs provided the density of anti-D on the membrane was sufficiently great. Thus, the inability to activate complement is more likely to be due to failure of activation stages after C1q (e.g., C1r, C1s, C4/C3 activation).

DO WE ENCOUNTER HTRs WITH HEMOGLOBINEMIA/HEMOGLOBINURIA WITHOUT STRONG EVIDENCE OF COMPLEMENT-ACTIVATING ALLOANTIBODIES IN THE PLASMA, OTHER THAN THE ITP/ANTI-D PROBLEM?

The answer is yes. As mentioned previously, plasma haptoglobins are diminished in almost all hemolytic anemias, including HTRs,²⁷ so small amounts of free Hb must be present in the plasma. Haptoglobins can bind 50 to 150 µg of Hb/dL of plasma. One needs approximately 15 µg Hb/100 mL to observe barely visible hemolysis; 45 µg Hb/100 mL will produce pink plasma, and 60-100 µg Hb/100 mL will produce red plasma. Thus, 5 mL of RBCs would have approximately 1650 µg Hb; in a male patient with a plasma volume of 3000 mL, this would represent 55 µg Hb/100 mL, yielding clearly visible pink to red plasma.²⁸ When RBC ⁵¹Cr survival studies are performed for studying destruction of Rh-incompatible RBCs, one finds minimal amounts of ⁵¹Cr (released from the labeled RBCs) in the plasma, but some (e.g., <5% of the injected amount) is detected.^29,30 The article by Jandl and colleagues in 1957²⁹ is recommended reading for anyone interested in this subject with relation to ABO and D. Although we do not see many HTRs due to anti-D now, many of us have seen patients with hemoglobinemia/hemoglobinuria after a HTR due to anti-D, usually when more than 1 unit of blood is destroyed. The old literature is educational^31,32 as often several units of D+ blood were transfused and hemoglobinuria was quite common. I have never been able to demonstrate complement binding by the anti-D associated with these types of reactions (unpublished observations).

So, how does the Hb get into the plasma/urine mimicking complement-mediated intravascular lysis? Experiments in dogs showed that hemoglobinemia can result from “overloading” the RES with damaged sequestered RBCs.³³ Now that we understand that at least three events (as discussed earlier) occur in the destruction of sensitized RBCs, it seems possible that Hb could be released when the macrophages partially ingest RBCs and/or a cytotoxic attack occurs. Together with breakdown of the RBCs within the macrophage, the system for dealing with free Hb within the RES could be overloaded and Hb may be released into the plasma. These events would occur completely independent of complement activation, but might produce a similar set of laboratory results.

Although the above scenario might explain the hemoglobinemia/hemoglobinuria encountered in HTRs due to non–complement-activating antibodies (e.g., anti-D), I do not think that it would explain the clinical outcomes of the treated ITP patients described by Gaines.^9,14 Many of the observed clinical symptoms after the treatment, and the clinical consequences such as renal failure and DIC, suggest the involvement of complement activation. To determine if the IV anti-D associated with hemoglobinuria/hemoglobinemia was activating complement leading to direct lysis of the patient's RBCs, Gaines and colleagues¹⁷ used a simple assay looking for direct lysis of D+ RBCs in the presence of complement. There are other simple approaches that could be used such as the IAT using a powerful anti-C3 and testing enzyme-treated RBCs in the hemolysin test. I do not agree with Gaines and coworkers¹⁷ when they say that they did not use enzyme-treated RBCs because that would not closely approximate in vivo conditions (“physiologic conditions”). Almost all of the tests we perform in immunohematology do not approximate in vivo conditions. Every day we select blood for transfusion and judge clinical significance by basically adding two drops of serum to one drop of 2% to 5% RBCs, often with potentiators added (e.g., low-ionic-strength saline [LISS], polyethylene glycol [PEG], albumin); these tests are nowhere close to physiologic conditions. When we evaluate drug-induced immune hemolytic anemia, the results correlate well with what we observe in the patients, yet the amount of drug we use in vitro to demonstrate that the patient's antibody causing agglutinates, hemolysis, or sensitization detected by the antiglobulin test in the presence of a particular drug is often nowhere near the amount of drug present in vivo; in fact, penicillin antibodies cannot be detected in vitro if one uses a concentration equal to that in the plasma in vivo, even though the penicillin antibodies are causing the hemolytic anemia and are detected in vitro if much larger quantities of drug are used. “Physiologic” conditions are difficult, if not impossible, to achieve as so many influencing factors are involved in vivo, some of which we do not even know about. I emphasized this in a recent publication³⁴ concerning the gaps in our knowledge of immune RBC destruction. When our routine, or even special approaches fail to explain results in a patient, we must start thinking “out of the box” and try to come up with a mechanism, which is not what we teach or what we learned as a student.

Sometimes antibodies can cause intravascular hemolysis without activating complement. Antibodies to determinants on glycophorin A (Pr, En^a) can cause membrane proteins to aggregate allowing Na²⁺ and Ca²⁺ to enter the RBCs causing lysis.³⁵ Antigen-antibody reactions can sometimes generate reactive oxygen species (ROS), for example, hydrogen peroxide and ozone, which have been shown to sometimes cause lysis of PLTs,³⁶ so why not RBCs? It is of interest to note that Coopamah and colleagues³⁷ showed that when IgG anti-D–sensitized RBCs were incubated with white blood cells, hydrogen peroxide and other ROS were generated by monocytes and granulocytes. Another report by Lee and coworkers³⁸ showed that mice lacking leucine zipper transcription factor (Nrf), which mediates up regulation of antioxidant detoxification and antioxidant genes in apoptosis, led to immune hemolytic anemia. These mice had RBCs of abnormal shape, increased RBC-bound IgG (increased by 120%), and RBCs that were more sensitive to hydrogen peroxide–induced lysis. It was suggested that a chronic increase in oxidative stress from decreased antioxidant capacity sensitizes RBCs with IgG and causes hemolytic anemia in Nr2-1 mice, suggesting a pivotal role for a Nrf-2-antioxidant responsive pathway in the cellular antioxidant system. These two articles,^37,38 especially the one by Coopamah and coworkers,³⁷ suggest a novel mechanism that could operate in vivo with blood group antibody-induced intravascular, complement-independent hemolysis, but may not be obvious using our routine in vitro approaches.

HTRs with no detectable antibodies have been reported since the 1950s.³⁰ In 1996 we reported 71 such patients encountered over a 10-year period.³⁹ Seventy percent of these patients had hemoglobinemia and hemoglobinuria after transfusion of seemingly compatible RBCs. Eighteen percent of the patients had antibodies detectable by a simple unusual procedure, the indirect polybrene test (three anti-C, three anti-Jk^a, two anti-S, two anti-e, one anti-E, and one anti-Jk^b). These antibodies were not detected by other commonly used tests (including LISS additives, enzymes, PEG). In eight patients, antibodies could not be detected, but phenotype-matched RBCs caused no HTRs. Sometimes antibodies (anti-c, -C, and -Vel) became detectable eventually by routine methods. Thus, in some cases, we could detect alloantibodies by nonroutine (and certainly “nonphysiologic”) methods and prove that these antibodies were causing the acute severe HTRs (e.g., blood lacking the pertinent antigens caused no HTRs). In most cases we could not detect the antibodies, although sometimes phenotypically matched blood caused no HTRs. As one or more units of “compatible” RBCs were often destroyed within 24 hours, and no complement activation was obvious (negative DAT with anti-C3, no classical clinical signs), the hemoglobinemia/hemoglobinuria was thought to be associated with extravascular lysis with the RES not being able to cope with the resulting Hb released, and the Hb appearing in the plasma and eventually the urine. We wondered if in such patients, there is an increased rate of cytotoxic attack by macrophages.

CONCLUSIONS

There is still no evidence to explain why a small number of ITP D+ patients receiving IV anti-D have severe acute hemolytic anemia associated with hemoglobinemia/hemoglobinuria, sometimes renal failure, and on rare occasions DIC. Although these all make one suspicious of complement activation, so far there is little in vitro evidence to support this supposition.

Perhaps some of the novel mechanisms discussed in my recent publication³⁴ and by Tarentino and colleagues⁴⁰ might relate to the ITP/anti-D problems. For instance, could the findings be related to the presence of antibodies other than anti-D? IV anti-D, like IVIG, is made by pooling plasma from many donors and can contain antibodies to blood group antigens other than D (e.g., ABO, other Rh, K, Jk^a, Fy^a, etc.).⁴¹ Might these multiple specificities combine to create complement activation as suggested by Rosse and Parker?²¹ What about the role of contaminating antibodies to non-RBC antigens (e.g., HLA, anti-Fc, anti-idiotype, antibodies to allotypic determinants on plasma proteins)? Any of these could activate complement when reacting with the respective antigen encountered in the recipient and cause bystander lysis of RBCs. Patients with ITP receive many transfusions (e.g., PLTs, IVIG, etc.) containing foreign plasma antigens before they receive the IV anti-D. There is good evidence that HLA antibodies have caused HTRs associated with intravascular hemolysis (see Garratty³⁴ for more detail). So, is anti-D the culprit, or is some other “contaminant” in the product causing these reactions? What makes a few patients react to this product, whereas thousands of others show no reactions, and can we find a way to screen for such patients? These questions are very similar to those we have been posing for drug-induced immune hemolytic anemia (DIIHA). Some drugs have become FDA-licensed because clinical trials do not show a significant number of patients who develop IHA after drug administration, yet later we find a relatively small number of recipients who develop severe IHA, sometimes associated with hemoglobinemia/hemoglobinuria, renal failure, and fatalities. A good example are the cephalosporins, which we have found to be the most common cause of DIIHA during the past few decades.⁴² The two main culprits are cefotetan and ceftriaxone,⁴² which were first reported to cause DIIHA (1989 and 1991) and fatalities (1991 and 1992).⁴² Many reports followed, but the drugs are still being used, and although we can show that drug antibodies are involved, we still do not know why individual rare patients have such severe hemolysis after receiving these drugs.

The publications by Gaines and colleagues^9,14,17 add another mystery to the rather long list of features of immune RBC destruction that we still do not understand. It remains a rich area for research!

CONFLICT OF INTEREST

The author declares no conflict of interest.