Investigation patients haemolytic provisionboth, bound to the red cells.3 Cold auto-antibodies are usually IgM, the notable ex-ception being the IgG class Donath-Landsteiner antibody;

62 Clin Pathol 1995;48:602-610

ACP Broadsheet No. 145 July 1995

Investigation of patients with autoimmunehaemolytic anaemia and provision of blood fortransfusion

R J Sokol, D J Booker, R Stamps

IntroductionAutoimmune haemolysis may be defined as areduced red cell lifespan due to the productionof antibodies which react with antigens carriedon the individual's own red cells. When the rateof red cell destruction exceeds the regenerativecapacity of the bone marrow, anaemia results.Autoimmune haemolytic disease may be con-veniently classified into warm, cold and mixedtypes reflecting the thermal optima of the auto-antibodies responsible.' It may occur as a prim-ary condition or be associated aetiologicallywith other diseases, notably lympho-proliferative disorders, other autoimmuneconditions (for example, systemic lupuserythematosus, ulcerative colitis), myelodys-plastic syndromes or certain drugs-methyl-dopa being the classic example.' A rare typeof autoimmune haemolysis can occur if thered cell membrane is altered to expose certaincryptantigens (for example, T and Tk) whichthen react with specific antibodies present inthe serum of most individuals.2Red cell autoantibodies are of IgG, IgM and

IgA classes. Warm autoantibodies are mainlyIgG, although recent studies using sensitivetechniques have shown that some 37% of casesalso have increased amounts of IgM or IgA, orboth, bound to the red cells.3 Cold auto-antibodies are usually IgM, the notable ex-ception being the IgG class Donath-Landsteiner antibody; however, examplesof cold reacting IgA autoantibodies (for ex-ample, anti-Pr) have also been reported.4

In patients with autoimmune haemolysis,the coating of autoantibodies per se does notdamage the red cells but causes haemolysisvia complement activation and/or by inducinginteractions with effector cells, mainly in themononuclear phagocyte system.' Haemolysismay be extravascular (more common) or intra-vascular (more spectacular).

Extravascular haemolysis occurs when auto-antibodies or C3 components bind to red cellsand react with specific receptors on mono-nuclear phagocytes (or other white cells) re-sulting in red cell destruction throughphagocytosis, spherocyte formation or antibody

dependent cellular cytotoxicity. The differentimmunoglobulin classes and complement com-ponents may act synergistically in bringingabout red cell destruction.3

Intravascular haemolysis occurs when com-plement is fully activated. It is rare (being seenin less than a fifth of patients with autoimmunehaemolytic anaemia), though it is more com-mon in those cases occurring in childhood.'Autoantibodies which trigger intravascularhaemolysis are mostly of IgM class, but itcan also be caused by some warm reactingantibodies ofIgGl and IgG3 subclass; the mostimportant IgG example, however, is the coldreacting Donath-Landsteiner antibody.' Tra-ditionally, IgA class red cell autoantibodies arethought not to activate complement, but thismay not always be the case as there are scatteredreports of patients with autoimmune haemo-lytic anaemia and intravascular haemolysiswhere only IgA class antibodies were detected6;a suggestion has been made that IgA antibodiescan activate complement when they are in anaggregated form.67 Complement activation isoften triggered, but rarely proceeds beyond theC3 stage because of the presence of regulatoryinhibitors. Red cells circulating with C3bbound to their surface are removed by phago-cytosis, mainly by the liver macrophages. Thenaturally occurring regulatory factors act onany red cells not engulfed, cleaving the C3b toC3d,g; these cells survive normally and areidentified in vitro with anti-C3d reagents.'

Diagnosis of autoimmune haemolysis isbased on demonstrating that autoantibodies orcomplement components, or both, are boundto the red cells and are associated with ashortened red cell lifespan. Diagnosis may beeasy or, on occasions, be extremely difficult-for example, when it is a minor part of a chronicdisorder or serious condition where the effectsofhaemorrhage, treatment, blood transfusions,and other causes of anaemia, have also to betaken into account.'

This broadsheet is based on the standardoperating procedures used at the Trent BloodTransfusion Centre (Sheffield) for investigatingsuspected cases of autoimmune haemolytic

This Broadsheet has beenprepared by the authors at theinvitation of the Association ofClincial Pathologists whoreserve the copyright. Furthercopies of this Broadsheet maybe obtained from thePublishing Manager, 3rournalof Clinical Pathology, BMAHouse, Tavistock Square,London WClH 9_tR.

Trent BloodTransfusion Centre,Longley Lane,Sheffield S5 7JNR J SokolD J BookerR Stamps

Correspondence to:Dr R J Sokol.

Accepted for publication20 December 1994

602

on May 23, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.48.7.602 on 1 July 1995. Dow

nloaded from

http://jcp.bmj.com/

The investigation of autoimmune haemolytic anaemia

Investigation ofpatients with autoimmune haemolytic anaemia. * Tests considered to be more appropriate for specialist laboratories.

anaemia. It aims to provide guidance to im-munohaematological tests which can be carriedout in a hospital blood bank, as well as givingan overview of some of the techniques whichcan be performed in a specialist laboratory.Using these guidelines, summarised in the fig-ure, even serologically complex cases can beunravelled and the patients safely transfused.

Samples requiredThe investigation of autoimmune haemolyticanaemia can be complicated and therefore it isimportant that adequate blood samples areobtained from the patient. Full testing can thenbe performed without delay, thus allowing anyblood cross-matched to be transfused with con-fidence that it will not harm the recipient.Our ideal samples are, firstly, serum from

30 ml clotted blood; this should be separatedat 37°C to permit accurate assessment of thethermal amplitude and titration score of anycold autoantibodies present. Both amplitudeand score will be artefactually reduced if thesamples are allowed to clot at room temperatureor placed in a refrigerator at 4°C as the coldautoantibodies will be adsorbed on to the redcells.

Secondly, we like 20 ml blood collected withEDTA as anticoagulant-these samples are re-quired for grouping, direct antiglobulin tests,the production of eluates, and for auto-absorption studies. Their use for direct anti-globulin testing is strongly recommended as

EDTA inhibits in vitro complement activationand therefore any complement components de-tected on the red cells will be the result ofactivation in vivo.8

Finally, an "Investigation Request Form" isrequired; this should give full personal andclinical details, including drug therapy, pre-vious pregnancies and transfusions and theresults of any relevant laboratory tests.

Tests carried out on patients' red cellsABO GROUPING AND Rh GENOTYPINGABO grouping is part of any standard ser-ological investigation of autoimmune haemol-ysis. It becomes particularly important iftransfusions are required, or if rare A or Bspecific autoantibodies are suspected. Althoughstandard saline agglutination procedures areused, ABO grouping may be difficult in thepresence of large amounts of autoantibodiesand control tests using inert AB serum andmonoclonal control media must be included.If group A subtyping is required, it is bestperformed using the lectin Dolichos biflorus(which shows anti-Al specificity). Cold auto-agglutinins are a particular problem and thepatient's red cells may have to be washed severaltimes in phosphate buffered saline (PBS) (pH7 0) prewarmed to 37°C before they are suit-able for grouping; it may even be necessary toABO group at 37°C as well as at 18°C. Reverseor serum grouping with group A1, A,, B, and0 cells as well as with the patient's own red

603



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/

Sokol, Booker, Stamps

cells is an important part of the procedure, andmay also have to be performed at 37°C.

Full Rh genotyping is carried out so thatsuitable blood can be selected for transfusionand also to permit any red cell antibodies show-ing Rh specificity to be identified as either auto-or alloantibodies. It is important that "salineagglutinating" reagents are used as patients'red cells which are heavily coated with im-munoglobulins may spontaneously agglutinateif suspended in solutions of bovine albumin orproteolytic enzymes. The red cells should bewashed four times in PBS (pH 7 0) (prewarmedto 37°C if cold autoagglutinins are suspected)before Rh typing. A 3-5% suspension of redcells is tested for the C, D, E, c, e, and Cwantigens. Suitable control cells must be usedfor each reagent, and inert AB serum or mono-clonal control media, or both, must be includedto identify any autoagglutination. Com-mercially available IgM monoclonal Rh anti-sera give fast and dependable results; theyusually require a short incubation at room tem-perature followed by gentle centrifugation. Theresults are read microscopically.

In some patients with autoimmune hae-molytic anaemia the cells are so heavily coatedwith immunoglobulin that spontaneous ag-glutination makes it impossible to type thecells using the above methods. In these caseschloroquine can be used to remove red cellbound immunoglobulin.9 Suitable control cellsmust also be treated as prolonged exposure tochloroquine may damage the red cell antigens.ZZAP, a mixture of dithiothreitol and cysteineactivated papain,10 has been recommended asan alternative to chloroquine, but has the dis-advantage of denaturing Kell, Duffy and MNSantigens.

Method using chloroquine-The patient's andcontrol cells are washed three times in PBS(pH 7 0); 5% suspensions in chloroquine di-phosphate solution (200 mg/ml in PBS(pH5-0)) are incubated either at 30°C for90 minutes or at 37°C for a maximum of30 minutes, washed twice and typed.9

DIRECT ANTIGLOBULIN TESTSDirect antiglobulin tests are used to detectincreased amounts of red cell bound im-munoproteins. A positive test result may bedue to (i) autoantibodies and complementcomponents, (ii) non-specific adsorption ofprotein-for example, in patients with my-eloma or in those with a general increase iny-globulin concentrations, (iii) immune com-plexes, (iv) drug induced antibodies, and (v)alloantibodies-for example, in haemolytic dis-ease of the newborn or when the transfusionof incompatible blood has resulted in a delayedtransfusion reaction.

In the standard agglutination direct anti-globulin test a range of antisera is employed.Broad spectrum reagents are valuable for initialscreening and must contain at least anti-IgG,-IgA and -C3d. Monospecific reagents are usedto identify which immunoproteins are coating

the red cells: anti-IgG, -IgA, -IgM, -C3d,-C4c, and -C3c, as well as IgG subclass anti-sera, are now commonly used. The anti-immunoglobulin reagents must be heavy chainspecific. In our experience a good anti-IgGwill start to give positive results at about 150molecules per red cell.

Method-Red cells are washed at least fourtimes in PBS (pH 7 0) and a 2% suspensionis tested using a spin tube technique (onedrop of red cell suspension to two drops ofantiglobulin reagent). The results are readmicroscopically. Where low affinity antibodiesare suspected, ice-cold PBS (pH 7 0) is usedto wash the red cells before testing.

Direct antiglobulin tests utilising enzymelinked reagents may also be useful in the in-vestigation ofpatients with autoimmune haemo-lytic anaemia. "-13 These methods are extremelysensitive, being able to detect the small quant-ities of immunoglobulins found on normal redcells (about 33 molecules per cell in the caseof IgG) 13; they can thus recognise slightlyincreased amounts of cell bound autoanti-bodies (particularly of IgM and IgA class)which, although undetectable by agglutinationtechniques, may have a significant clinicaleffect.3 2 1

Method-Alkaline phosphatase linked anti-IgG, -IgA and -IgM (50 jil) (Sigma, Poole,Dorset, UK), appropriately diluted in PBS (pH7-0), are mixed with 25 ,ul of a 25% suspensionof washed test and control red cells in a U-wellmicrotitre plate. After incubation at 370C, thecells are washed six times in PBS. Substrate(200 ,ul) (P-nitrophenyl phosphate 1 mglml incarbonate buffer (pH 9 8)) is added to eachwell and the plate incubated at 25°C for afurther 25 minutes. The plate is centrifugedand 100 jlI of supernatant from each well istransferred to a flat-well microtitre plate, thereactions are stopped using 50 ,ul 3 M NaOHand the optical densities read at 405 nm. Redcell counts on the buttons of cells left in theU-well plate allow the optical density readingsto be adjusted to a standard cell count. A resultis considered to be positive when the opticaldensity value is more than 3 SDs above themean value obtained for a series of healthysubjects. 13

A number of promising new methods whichutilise column technology are coming on to themarket. In the best known example (DiaMedAG, Cressier sur Morat, Switzerland) cards aremanufactured containing a set of monospecificantiglobulin reagents in gel. The red cells(which require no washing as serum or plasmacannot enter the gel and neutralise the reagents)are diluted in low ionic strength saline (LISS),added to wells at the top of each column andcentrifuged. Agglutinated red cells are held inthe gel matrix whilst non-agglutinated cellsform a button at the bottom of the column.The results are read visually and can be pho-tocopied as a record. The lack of a washing

604



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


step improves the sensitivity of the direct anti-globulin test and also permits detection of lowaffinity autoantibodies. At the time of writing,our experience with the DiaMed system is thatthe anti-IgG is excellent (with a sensitivitybetween the agglutination and enzyme linkedmethods), but the anti-C3d and -IgM are lesssatisfactory. In fact, the latter is now beingreplaced by an anti-IgA reagent.

TESTS ON RED CELL ELUATESThese are carried out to elucidate the cause ofa positive direct antiglobulin test. Auto-antibodies, for example, rebind to normal redcells and the immunoglobulin concentrationwhich occurs during preparation makes eluatesespecially suitable for determining the im-munoglobulin class, subclass and any bloodgroup specificity. Direct antiglobulin testswhich show large amounts of cell bound IgG,but where no antibody is found in the eluate,may be due to drugs such as penicillin. Otherexamples of a positive direct antiglobulin testand no elutable antibody are seen in patientswith increased amounts of immune complexesor those with paraprotein or y-globulins non-specifically coating the red cells. If a specificantibody is eluted, the patient's red cells shouldbe checked for the blood group concerned.The presence of alloantibodies in an eluatesuggests a delayed transfusion reaction (orhaemolytic disease of the newborn), though"mimicking" autoantibodies4 15 and theMatuhasi-Ogata phenomenon"6 where allo-antibodies can be found non-specifically as-sociated with autoantibodies in red cell eluatesshould not be forgotten when interpreting theresults.There are many different techniques for pro-

ducing eluates; two which we have foundparticularly useful will be described. The chloro-form/trichloroethylene method'7 is suitablefor red cells which are strongly IgG sensitised(that is, those giving a visible reaction in thestandard direct antiglobulin test).

Method-Packed red cells (1 ml) are washedfour times and resuspended to 50% in PBS(pH 7 0). Two volumes of chloroform/tri-chloroethylene mixture (1:1) are added, thetube is sealed and the contents mixed vig-orously. The tube is then incubated at 37°Cfor 10 minutes with occasional mixing and thencentrifuged at high speed for five minutes. Theeluate is the haemoglobin stained supernatant.We recommend using acid elution methods

if the direct antiglobulin test is only weaklypositive (suggesting a small increase in cellbound immunoglobulins). These techniqueshave the advantage of causing minimal damageto the red cell membrane. They are based onthe principle that reducing the pH breaks downthe bonds between antibody and red cell. Beingfree from haemoglobin, acid eluates can beconcentrated (for example, with a Minicon-CS 15 spinal fluid concentrator) to permit de-tection of weaker antibodies. The ELU-PLUS(Dominion Biologicals, Dartmouth, Nova

Scotia, Canada) is an excellent commercial redcell acid elution kit.

Method-Packed cells (1 ml) are washed fourtimes in the "wash solution" provided and 1 mlglycine HCI buffer solution (pH 3 0) in-corporating a pH indicator is added. The tubeis sealed, mixed and centrifuged immediatelyfor one minute, the supernatant eluate beingtaken off without delay. The pH is adjusted to6-5-7-5 (shown by blue colour development)by dropwise addition of the "base solution";any precipitate which forms is removed bycentrifugation.

Eluates prepared using either of the methodsdescribed above are tested by an indirect anti-globulin technique for antibody specificityagainst a short panel of group 0 red cellsfully typed for the most common blood groupantigens. The immunoglobulin class and IgGsubclass of the autoantibody are determinedby incubating the eluate with red cells for onehour at 37°C, washing and then testing withmonospecific anti-IgG, -IgM, and -IgA as wellas with anti-IgG subclass reagents.

TESTS FOR POLYAGGLUTINABLE RED CELLSPolyagglutination2 has on rare occasions beenassociated with autoimmune haemolytic an-aemia. Many of the acquired forms, such at T,Tk, acquired B, Th, and VA, are produced bythe action of microbial enzymes on the red cellmembrane, altering its structure and exposingcryptantigens. Naturally occurring IgM anti-bodies in normal adult (and in certain animal)serum may react with these cryptantigens caus-ing red cell destruction,218 however, some au-thorities believe that in most, if not all cases ofpolyagglutination, the haemolysis is caused bythe action of the microbial enzymes rather thanthe antibodies.'9

Polyagglutinable cells are agglutinated by alarge proportion of ABO compatible adulthuman serum samples, but not by cord or(usually) by autologous serum, and sometimesgive variable non-specific reactions with anti-human globulin reagents. Suspected poly-agglutinable cells should therefore be mixed(as a 3-5% suspension in saline) with severalgroup AB adult and cord serum samples. Ifagglutination is observed with the majority ofthe adult samples, but not with the cordsamples, polyagglutination is likely. The varioustypes of polyagglutination can be easily differ-entiated by use of a small panel of lectins(Gamma Biologicals, Houston, Texas, USA)which includes extracts of Arachis hypogoea,Salvia sclarea, S horminum, and Glycine S6ja.

FUNCTIONAL CELLULAR ASSAYSThese tests can be useful in evaluating theclinical significance of autoantibodies althoughthey are rarely carried out routinely.The monocyte monolayer assay has been

used most often.20 It involves producing a layerof monocytes by adherence, incubating with asuspension ofpatient's red cells and, after fixing

605



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


and staining, counting the percentage ofmono-cytes with phagocytosed or adherent red cells,or both. The test is simple to perform; eitherdonor or autologous monocytes can be used,with perhaps the patient's own monocytes beingmore representative of the in vivo situation.Antibody dependent cellular cytotoxicity as-

says have rarely been used in the investigationof autoimmune haemolysis.2' They assess theamount of cytotoxic damage to 51Cr labelledred cells by measuring the amount of isotopereleased by effector cells, usually monocytes(or lymphocytes). The test is quantitative,though has the disadvantages of being difficultand expensive to perform because of the use ofradioisotopes. Introduction of a method usingenzyme linked labels would be an interestingfuture development.The chemiluminescence test has not been

evaluated in patients with autoimmune haemo-lytic anaemia as yet.22 It assesses the mono-cyte response of immunoglobulin coated redcells during phagocytosis: the monocytes pro-duce oxygen radicals which react with luminolgenerating light emission which is measured ina luminometer. The test is simple, rapid andobjective, and would seem to have potentialvalue in assessing the clinical significance ofred cell autoantibodies.

Tests carried out on patients' serumANTIBODY INVESTIGATIONSThe purpose of these tests is to detect auto-and alloantibodies in the patients' serum, todetermine whether the autoantibodies are ofwarm or cold type and to identify blood groupspecificity. Each patient's serum is testedagainst their own red cells (auto-control) anda panel of group 0 cells which have been fullytyped for the common blood group systems(Rh, Kell, Duffy, Kidd, Lewis, MNS, and P)using saline agglutination at 18°C and 37°Cand albumin, papain and indirect antiglobulintechniques at 37°C. Although these are stand-ard serological methods, it is most importantto pay meticulous attention to detail. Serumand red cell suspensions should be warmedbefore mixing. The results should be readmicroscopically, making careful note of re-action strengths as variations may indicate amixture ofauto- and alloantibodies. Ifcold auto-agglutinins with a wide thermal range arepresent, the results of tests performed at 37°Cshould be read on slides prewarmed to thattemperature, although the cold antibodies maynot be active at 37°C. If the tests are allowedto cool before being read, then agglutinates willform which make the interpretation ofthe 37°Cresults inaccurate. Because of the problemswith cooling, indirect antiglobulin tests maybenefit from suspension in PBS, prewarmed to37°C, before the normal wash cycle.The albumin displacement technique using

20% bovine albumin, although outmoded bymore sensitive methods, is included both forthe detection of alloantibodies when the auto-antibody reactions with enzyme and indirectantiglobulin techniques are very strong and alsofor identifying gross rouleaux formation whichcan be mistaken for autoantibody.

Indirect antiglobulin tests using LISS areutilised as they are sensitive and give rapidresults. They are based on the principle thatdecreasing the ionic strength of a reaction mix-ture increases the rate at which antigen-antibody complexes are formed. Some patients,however, have non-clinically significant auto-antibodies which react only when using theLISS method, giving negative reactions withindirect antiglobulin techniques utilising iso-tonic saline. In our experience these cases,which usually have a negative or weakly positivedirect antiglobulin test, often have increased,y-globulin concentrations, and show weak roul-eaux formation in the saline and papain tests.

EXAMINATION OF THE SPECIFICITY ANDTHERMAL RANGE OF COLD AUTOAGGLUTININSIn cases where there is a cold autoagglutininof possible clinical significance the serum istitrated in saline (doubling dilutions from neatto 1 in 512 are usually sufficient) and testedwith pooled OI, Oi, papainised OI, and thepatient's own red cells at 1 8°C and 30°C (Oicordcells are used because of the extreme rarity ofOiadult cells). The tests at 30°C should alsoinclude the albumin displacement techniquewith OI, Oi and patient's cells as it has beenshown that activity in albumin at this tem-perature is a reasonable indicator that a coldantibody is of clinical significance.23 Careshould be taken when reading these tests tomaintain the microscope slides at 30°C.The majority of patients suffering from cold

haemagglutinin disease (CHAD) have autoanti-I in their serum, although some have autoanti-i or an antibody showing no obvious speci-ficity. Anti-I and anti-i may be associated withMycoplasma pneumoniae and infectious mono-nucleosis, respectively. Rarely, anti-Pr is found;this reacts equally well with OI and Oi cells,but weakly or not at all with the papainised cellsas the Pr antigen is destroyed by proteolyticenzymes.

HAEMOLYSIN TESTSThe potential for autoagglutinins (usually coldreacting ones) to cause complement activationand intravascular haemolysis in vivo is assessedin vitro by performing haemolysin tests. Aspreviously mentioned, collection of samplesand separation of serum for these tests mustbe carried out strictly at 37°C, otherwiseautoabsorption of antibody may occur-thisis particularly important in patients suspectedof having paroxysmal cold haemoglobinuriawhere the levels of autoantibody may be lowto start with.

Method-As the patient's serum is likely to becomplement deficient, complement must beadded to the tests by preparing doublingdilutions (from 1 in 2 to 1 in 16) of patient'sserum in pooled group 0 serum less than24hours old; 10 drops of each dilution areplaced in two rows of 10 x 75 mm tubes. Asbetter correlation with the in vivo effect of the

606



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


antibody is given if the serum is first acidifiedto a pH of about 6-8, one drop of 0-2 M HClis carefully added to each tube. The acidifiedserum dilutions and a 50% suspension ofpooled group 0 red cells (made up in thecomplement rich serum) are prewarmed to37°C, one drop of the red cell suspension isadded to each tube and after mixing, one rowis removed and incubated at 18°C for twohours, the other row remaining at 37°C. Thered cells are resuspended, centrifuged and thesupernatants examined for haemolysis. It isimportant to compare the colour of the testsupernatant with that of the original serumdilution, particularly if the latter already showsevidence of lysis. The acid haemolysin tests at18°C are usually more strongly positive thanthe ones at 37GC.

Haemolysins are infrequent in warm typeautoimmune haemolytic anaemia. They mayvery rarely cause severe intravascular haemol-ysis, but usually they have minimal clinicaleffects and are only detected using papainisedcells. 524

Method-Three drops of a 5% suspension ofprewarmed, papainised pooled 0 cells areadded to three drops of prewarmed serumdilutions (prepared as above but not acidified),mixed and incubated at 37°C for one to twohours before centrifugation and examinationof the supernatants.

In patients with unexplained haemolysis orin those with suspected paroxysmal coldhaemoglobinuria an indirect Donath-Landsteiner test25 is performed.

Method-Doubling dilutions of the patient'sserum (from 1 in 2 to 1 in 16) and a 50%suspension of pooled 0 cells are prepared infresh pooled 0 serum as before. Ten drops ofeach dilution are placed in pairs of 10 x 75 mmtubes set in two rows and prewarmed to 37°C;one drop of the red cells is then added andmixed. One set of dilutions is left at 37°C fortwo hours while the other is placed in meltingice (0°C) for one hour before being transferredback to 37°C for a further hour; the contentsof the tubes are then mixed and centrifuged.A positive result is indicated by lysis in the teststhat have been cooled and re-warmed, withno lysis in the tests which remained at37°C throughout. Classically, the Donath-Landsteiner antibody shows P blood groupspecificity and, if possible, this should be con-firmed by repeating the test using pp cells andobtaining a negative result.

If paroxysmal cold haemoglobinuria is sus-pected but the indirect Donath-Landsteinertest is negative, further investigations are car-ried out. Repeating the test using papainisedred cells sometimes gives a positive result incases where the level of autoantibody is par-ticularly low. A two-stage procedure in whichthe complement is added after the tests havebeen at 0°C for one hour overcomes the prob-lem of inhibition of the autoantibody by glo-boside present in the fresh serum used as asource of complement.

ABSORPTION OF AUTOANTIBODIESStrongly reacting serum autoantibodies maymask the presence of alloantibodies and thusput the patient at risk of a transfusion reaction.The recent suggestion that the risk of allo-immunisation is generally overstated26 wasquickly refuted27 and in our experience allo-antibodies are found in about 14% of patientswith autoantibodies.28 Both cold and warmautoantibodies may hamper the detection ofalloantibodies, especially when using indirectantiglobulin techniques, and several methodsfor absorbing autoantibodies are available toovercome this difficulty.Cold autoagglutinins can be either auto-

absorbed or in the case of anti-I (but not anti-ior anti-Pr) absorbed using rabbit erythrocytestroma (Organon Teknika, Boxtel, The Neth-erlands). Rabbit red cells possess structuressimilar to the human I, H and HI antigens andthe stroma will remove these cold agglutininsbut not most clinically significant allo-antibodies. (Note that anti-B and anti-P, arealso removed by the stroma, and the absorbedserum should therefore not be used for ABOgrouping or for routine compatibility testing.)

Method for autoabsorption-One volume ofpacked patient's red cells (which have beenwashed four times in PBS prewarmed to 37°C)is added to one volume of patient's serum. Thecells and serum are mixed and placed at 4°Cfor at least one hour (or overnight if timeallows) and centrifuged; the serum is now readyfor testing. Further absorptions may be carriedout if necessary.

Method for absorption using rabbit erythrocytestmma-Patient serum (1 ml) is added to a tubeof stroma (from which the preservative hasbeen removed), mixed well, placed at 4°C fora minimum of 30 minutes (or for 60 minutesif time allows) and centrifuged. The absorbedserum is removed for testing or for furtherabsorption if required.Warm autoantibodies can be removed from

a patient's serum using either differential orautoabsorption techniques. If sufficient patientred cells are available and the patient has notbeen transfused within the last four months,then absorption with autologous red cellstreated with dithiothreitol and papain (ZZAP)may be used.29 The dithiothreitol removessome, if not all, of the autoantibody alreadybound to the red cells and the papain, byexposing further antigen sites, renders the cellmore efficient at absorbing further antibody.

Methodfor autoabsorption-To absorb 0-5 ml ofpatient's serum, at least 1 ml of packed, treatedautologous red cells is required; as some ofthe cells lyse during the dithiothreitollpapaintreatment, it is advisable to process at least1-5 ml of red cells. The patient's red cells arewashed three times in PBS (pH 7 0), packed,mixed with two volumes of dithiothreitol/papain working solution (2-5 ml 0-2 M dithio-

607



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


threitol plus 05 ml 1% papain solution plus2 ml PBS (pH 7 0)), incubated at 37°C for30 minutes, washed three times (removing asmuch saline as possible after the last wash) anddivided into two or three equal volumes. Onevolume of patient's serum is added to onealiquot of packed cells, mixed well, incubatedat 37°C for 30 minutes, centrifuged at highspeed, and transferred to the second aliquot ofred cells, mixed well and incubated at 37°C asbefore, repeating the absorption for a third timeif sufficient cells are available. The absorbedserum is tested against a comprehensive panelof fully typed red cells using a LISS indirectantiglobulin technique.

In practice, patients with strong auto-antibodies that require absorption are oftenanaemic and, because of the requirements forgrouping, direct antiglobulin testing and elu-tion studies, there are usually insufficient redcells available for successful autoabsorptioneven when several anticoagulated bloodsamples have been collected. In addition, trans-fusion within the previous four months canmean that the circulating transfused red cellsmay complicate the procedure by regaining thepotential to absorb weak alloantibodies aftertreatment with dithiothreitol and papain. Forthese reasons, we tend to carry out differentialabsorptions in preference to autoabsorption.

Differential absorption uses carefully se-lected group 0 red cells to identify or excludethe presence of alloantibodies in the Rh, Kell,Duffy, Kidd, Lewis, and MNS systems. Bloodof suitable type can be aliquoted into vials andstored at 4°C for up to three weeks, givingsufficient red cells for multiple tests. Examplesof two cell types suitable for absorption studiesare Orr, kk, Jka+b- and 0 R,R1, kk, Jkab+, oneof them also being Lea-b if possible; the Duffyand MNS antigens are not important as theyare destroyed by the enzyme treatment used inthe procedure and would therefore not absorbthe corresponding alloantibodies. (Note thatanti-Lewis antibodies should already have beendetected in the 1 8°C saline agglutinationpanel.) The potential disadvantage of differ-ential absorption is the possibility that an anti-body to a high incidence antigen may beremoved by the absorbing cells, but in practicesuch antibodies are so rare that this is not amajor problem.

Method for differential absorption-Approx-imately 3-4 ml of each of the absorbingcells is washed three times in PBS (pH 7 0)and an equal volume of 1% papain solution isadded. After incubation at 37°C for 15minutes, the cells are washed a further threetimes (ensuring that they are well packed andall saline is removed after the final wash) andseparated into three or four aliquots. To thefirst aliquot of each absorbing cell, an equalvolume of patient's serum is added, mixed,incubated at 37°C for 15-30 minutes and cent-rifuged. The serum is then transferred to an-other aliquot of the same absorbing cell andthe procedure repeated. In most cases, re-

peating the procedure up to four times is ad-equate to remove strong autoantibodies andpermit the detection of concomitant allo-antibodies. The two absorbed serum samplesare tested against a comprehensive panel ofred cells using a LISS indirect antiglobulintechnique.

ESTIMATION OF SERUM HAPTOGLOBINS ANDSERUM PROTEIN ELECTROPHORESISHaptoglobins are oc-2-glycoproteins which havethe property of combining with haemoglobin;the normal serum range is 0A4-2 0 g/l. Theirconcentration is regarded as a reasonably sens-itive indicator of haemolysis, although in con-ditions where there is an increased rate ofsynthesis (for example, inflammatory disease,malignancy and steroid administration) greaterrates of red cell destruction may be necessaryto depress the concentration, and lower con-centrations may be found normally duringpregnancy. If a patient's serum haptoglobinconcentration is <0 1 g/l, then increased redcell breakdown is occurring, but this is notnecessarily because ofautoimmune haemolysis.For example, if other tests do not supportan immune aetiology, it may be advisable toperform Ham's test in case the patient is suffer-ing from paroxysmal nocturnal haemo-globinuria.Serum protein electrophoresis is sometimes

helpful in the investigation of patients withsuspected autoimmune haemolysis by revealinggeneral protein abnormalities. Paraproteins ora general increase in y-globulins can give riseto rouleaux formation (which can be confusedwith agglutination) and to positive direct anti-globulin tests due to non-specific adsorption.

Supplying blood for transfusionBlood to be given to patients suffering fromautoimmune haemolytic anaemia should havethe same ABO group as the potential recipient,except in the rare instances where the auto-antibodies show specificity for the A or Bantigens; in these cases, group 0 blood ispreferred.28 To avoid stimulating the pro-duction of Rh alloantibodies and to preventtransfusion reactions from such antibodies al-ready present but masked by the auto-antibodies, the blood to be transfused shouldhave the same Rh antigens as the patient (table).It is preferable to select blood using this prin-ciple even if the patient's autoantibody showsspecificity within the Rh system.28 The onlyexception would be if the haemolysis was ful-minating and the specificity simple, in whichcase it might be considered that the advantageof possible increased red cell survival wouldoutweigh the potential for stimulating allo-antibody production. In our experience suchcircumstances are exceedingly rare. As 99-8%of the population possess the k antigen (91%being kk), Kell negative blood (that is, kk)should also be routinely chosen to avoid stim-ulating the development of anti-Kell (and toprevent a reaction if it is present but masked).

608



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


Rh genotypes of blood usually selected for patients with redceUl autoantibodies

Patient's Rh genotype Rh genotype of suitable blood

R,r R,R,, R,r or rrR,R, R,RR,R2 Any Rh genotyperr rr

R2r R2R2, R2r or rrR2R2 R2R2

The blood selected should, of course, not pos-

sess antigens to any other alloantibodies pres-

ent. The most important non-Rh alloantibodiesto identify are those in the Duffy, Kidd, Lewis,and MNS (and Kell) systems.The units ofblood chosen are cross-matched

against the patient's serum using saline ag-

glutination methods at 30°C (to avoid problemswith clinically insignificant cold antibodies) andsaline, albumin, papain, and LISS indirect anti-globulin techniques at 37°C. In cases whereabsorption procedures have identified low in-cidence antibodies, cross-matching is also car-

ried out with the absorbed serum. Red cellunits which have been concentrated with theaddition of "saline, adenine, glucose-mannitol" (that is, SAG-M cells) are givenby choice.28 If these are not available, thenconcentrated red cells are suitable. Previously,it was thought necessary to wash the cells toremove plasma (which may be rich in com-

plement) before transfusing patients with dem-onstrable haemolysins-for example, thosewith chronic cold haemagglutinin disease or

with the Donath-Landsteiner antibody. Ourexperience has shown that this is unnecessaryif SAG-M units are selected.28The blood is almost always incompatible

by at least one technique and these units are

therefore issued to the patient with a labelstating "not compatible but considered suitablefor".28 In the rare cases where absorption ofthe autoantibody is unsuccessful, or where in-sufficient time is available to rule out the pres-ence of masked alloantibodies, a note is issuedwith the blood warning the clinician of theincreased risk of an adverse reaction and ad-vising that the patient is kept under close su-

pervision during the transfusion. If strong coldautoagglutinins are present in the patient'sserum, labels stating the advisability of trans-fusing through a warmer are attached to theblood packs.No critically ill patient with autoimmune

haemolytic anaemia should die through lack ofblood and in extreme circumstances we wouldempirically transfuse group Orr Kell negativeSAG-M blood, possibly also giving steroidsor intravenous immunoglobulin,30 or both, toprevent or reduce possible untoward con-

sequences. The use ofmonoclonal antisera, theLISS antiglobulin technique and rapid ab-sorption procedures allow suitable blood to beselected and cross-matched in under two hours,a fraction of the time taken only a few yearsago. Since January 1983, we have issued morethan 21 500 units of blood to over 5000 re-

cipients with red cell autoantibodies without,as far as we are aware, any serious ill effects.

Investigation of drug induced immunehaemolytic anaemiaFinally, many drugs and chemicals can bringabout red cell destruction in vivo by an immunemechanism and although many cases are notautoimmune in nature, it is convenient to out-line briefly the three main types (drug ad-sorption, immune complex and autoimmune)and their investigation in this broadsheet. Druginduced immune haemolysis is very complexand for an in-depth consideration ofthe subjectthe reader is recommended to consult one ofthe excellent recent review articles.3' 34 Eventhough such haemolysis is rare, occurring inabout one per million of the population,33 itmay cause significant problems clinically.The drug adsorption (or hapten) type is

usually associated with large doses of med-ication. The drug (for example, penicillin,cephalosporin or carbimazole) coats the patient'sred cells and leads to the development of IgGclass antibodies against a combination of drugand red cell membrane, resulting in the red cellsbeing destroyed by the mononuclear phagocytesystem. The immune complex type of immunehaemolysis can occur when small doses of drug(for example, quinine, rifampicin and tol-butamide) are being taken. Antibodies are pro-duced against the drug and form immunecomplexes which attach to the red cells andactivate complement; the resulting intra-vascular haemolysis may be severe and lead torenal failure or may even be fatal.The serological investigation of both these

types of drug induced haemolysis is fraughtwith difficulties. For example, certain drugs donot readily dissolve in aqueous solutions so thatpreparing an isotonic medium can be difficult.This can be further complicated by the presenceof inert filler in some tablets and capsulesmaking control of the amount of drug in so-lution almost impossible. Other problems in-clude the likelihood that the antibodies aredirected against metabolites of the drug32 orthat the patient may be taking several differentmedicines. Consulting the established lit-erature is invaluable in deciding the approachto the investigation. When a solution of drughas been prepared, an attempt is made to coatit onto group 0 red cells, initially by incubationof drug solution and red cells at 37°C. If thisfails, it may be necessary to vary the pH or usecross-linking techniques. The coated cells arethen used in serological tests (with and withoutcomplement) utilising the patient's serum anda red cell eluate; lysis and agglutination re-actions in saline and with indirect antiglobulintests (using monospecific anti-IgG and anti-C3 reagents) are noted. Positive results withdrug coated cells but negative reactions withuncoated red cells indicate the drug adsorptiontype of immune haemolysis. An immune com-plex mechanism is suggested if positive resultsare obtained with untreated group 0 red cellsand mixtures of drug solution and patient'sserum. The direct antiglobulin tests often showcomplement coating the red cells, but variousclasses of immunoglobulin can sometimes bedetected, particularly if sensitive enzyme linkedmethods are used.

609



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/


There are no laboratory tests to identify thethird type ofdrug induced immune haemolysis.Patients treated with methyldopa-for ex-ample, can develop autoantibodies in-distinguishable from those found in warm typeautoimmune haemolytic anaemia. In mostcases there is little or no clinical effect, but inothers overt haemolysis develops. It has beenpostulated that the drugs affect T suppressorlymphocyte control of B lymphocytes, per-mitting the production of autoantibodies.Methyldopa is prescribed less frequentlynowadays, and these cases are seen less often,whereas a few years ago methyldopa accountedfor some two thirds of all patients with a druginduced haemolysis.35 Currently, most cases ofthe autoimmune type of drug haemolysis arecaused by levodopa or non-steroidal anti-in-flammatory agents (for example, ibuprofen,mefenamic acid, naproxen).

We thank Mrs M E Grayson, Mrs A Steward and Miss NColton for secretarial assistance.

1 Sokol RJ, Booker DJ, Stamps R. The pathology of auto-immune haemolytic anaemia. JClin Pathol 1992;45:1047-52.

2 Levene C, Levene NA, Buskila D, Manny N. Red cellpolyagglutination Transfusion Med Rev 1988;2:176-85.

3 Sokol RJ, Hewitt S, Booker DJ, Bailey A. Red cell auto-antibodies, multiple immunoglobulin classes, and auto-immune hemolysis. Transfusion 1990;30:714-17.

4 Roelcke D, Haik H, Kreft H, Macdonald B, Pereira A,Habibi B. IgA cold agglutinins recognise Pr and Sa an-tigens expressed on glycophorins. Transfusion 1993;33:472-5.

5 Sokol RJ, Hewitt S. Autoimmune hemolysis: a critical re-view. Crit Rev Oncol Hematol 1985;4:125-54.

6 Sokol RJ, Booker DJ, Stamps R, Murphy F, Booth JR.Severe autoimmune haemolytic anaemia (AIHA) due tostrong IgA class autoantibodies showing Rh specifity.Transfusion Med 1994;4(Suppl 1):47.

7 Griffiss J McL. Biologic function of the serum IgA system:modulation ofcomplement-mediated effectormechanismsand conservation of antigenic mass. Ann NY Acad Sci1983;409:697-707.

8 Petz LD, Garratty G. Acquired immune hemolytic anemias.New York: Churchill Livingstone, 1980:159.

9 Beaumont AE, Stamps R, Booker DJ, Sokol RJ. Animproved method for removal of red cell boundimmunoglobulin using chloroquine solution. Immuno-hematology 1994;10:22-4.

10 Branch DR, Petz LD. A new reagent (ZZAP) having mul-tiple applications in immunohematology. Am Jf Clin Pathol1982;78:161-7.

11 Sokol RJ, Hewitt S, Booker DJ, Stamps R. Enzyme-linkeddirect antiglobulin tests in patients with autoimmunehaemolysis. Clin Pathol 1985;38:912-14.

12 Sokol RJ, Hewitt S, Booker DJ, Stamps R. Small quantitiesof erythrocyte bound immunoglobulin and autoimmunehaemolysis. Clin Pathol 1987;40:254-7.

13 Sokol RJ, Hewitt S, Booker DJ, Stamps R, Booth JR. Anenzyme-linked direct antigloblin test for assessing eryth-rocyte bound immunoglobulins. J ImmunolMethods 1988;106:31-5.

14 Domen RE, Clarke A. Case reports: red blood cell auto-antibodies mimicking alloantibodies. Immunohematology1991;7:98-101.

15 Engelfriet CP, Overbeeke MAM, von dem Borne AEGKr.Autoimmune hemolytic anemia. Semin Hematol 1992;29:3-12.

16 Allen FH, Issitt PD, Degnan TJ, Jackson VA, Reihart JK,Knowlin RK, et al. Further observations on the Matuhasi-Ogata phenomenon. Vox Sang 1969;16:47-56.

17 Massuet L, Martin C, Ribera A, Argelagues E, Duran-Suarez JR, Triginer J. Antibody elution from red bloodcells by chloroform and trichloroethylene. Transfusion1982;22:359-61.

18 Sayas MJ, Pastor E, Casanova C, Lurbe A, Carbonell F.Severe intravascular haemolysis associated with poly-agglutinability. Br J Haematol 1990;75:433-9.

19 Issitt PD. Applied blood group serology. 3rd edn. Miami:Montgomery Scientific Publications, 1985;456-76.

20 Zupanska B, Sokol RJ, Booker DJ, Stamps R. Erythrocyteautoantibodies, the monocyte monolayer assay and in vivohaemolysis. Br J Haematol 1993;84:144-50.

21 Yust I, Frisch B, Goldsher N. Antibody-dependent cell-mediated cytotoxity and phagocytosis of autologous redblood cells in alphamethyldopa-induced haemolysis. ScandJ Hematol 1986;36:211-16.

22 Zupanska B. Cellular immunoassays and their use for pre-dicting the clinical significance of antibodies. In: GarrattyG, ed. Immunobiology of transfuision medicine. New York:Marcel Dekker, 1993:465-91.

23 Garratty G, Petz LD, Hoops JK. The correlation of coldagglutinin titrations in saline and albumin with haemolyticanaemia. BrJ Haematol 1977;35:587-95.

24 Engelfriet CP, Ouwehand WH, van't Veer MB, BeckersDO, Mass NEL, von dem Borne AEGKr. Autoimmunehaemolytic anaemias. Ballieres Clin ImmunolAllergy 1987;1:251-67.

25 Heddle NM. Acute paroxysmal cold hemoglobinuria. Trans-fusion Med Rev 1989;III:219-29.

26 Salama A, Berghofer H, Mueller-Eckhardt C. Red bloodcell transfusion in warm-type autoimmune haemolyticanaemia. Lancet 1992;340:1515-17.

27 Garratty G, Petz LD. Transfusingpatients with autoimmunehaemolytic anaemia. Lancet 1993;341:1220.

28 Sokol RJ, Hewitt S, Booker DJ, Morris BM. Patients withred cell autoantibodies: selection of blood for transfusion.Clin Lab Haematol 1988;10:257-64.

29 James P, Rowe GP, Tozzo GG. Elucidation of alloantibodiesin autoimmune haemolytic anaemia. Vox Sang 1988;54:167-71.

30 Kohan AI, Niborski RC, Rey JA, Amerise G, Vazquez MI,Zani N, et al. High-dose intravenous immunoglobulin innon-ABO transfusion incompatibility. Vox Sang 1994;67:195-8.

31 Petz LD. Drug-induced immune hemolytic anemia. In:Nance SJ, ed. Immune destruction ofred blood cells. Arlington,Virginia: Amercian Association of Blood Banks, 1989:53-75.

32 Mueller-Eckhardt C, Salama A. Drug-induced immunecytopenias: a unifying pathogenetic concept with specialemphasis on the role of drug metabolites. Transfusion MedRev 1990;IV:69-77.

33 Garratty G. Drug-induced immune hemolytic anemia. In:Garratty G, ed. Immunobiology oftransfusion medicine. NewYork: Marcel Dekker, 1993:523-51.

34 Petz LD. Drug-induced autoimmune hemolytic anemia.Transfusion Med Rev 1993;VII:242-54.

35 Petz LD, Garratty G. Acquired immune hemolytic anemias.New York: Churchill Livingstone, 1980:29.

610



j.com/

J Clin P


nloaded from

http://jcp.bmj.com/

Documents

Investigation patients haemolytic provisionboth, bound to the red cells.3 Cold auto-antibodies are usually IgM, the notable ex-ception being the IgG class Donath-Landsteiner antibody;