Journal of Clinical InvestigationVol. 41, No. 5, 1962

IMMUNOREACTIONSINVOLVING PLATELETS. VI. REACTIONSOF MATERNALISOANTIBODIES RESPONSIBLEFOR

NEONATALPURPURA. DIFFERENTIATION OFA SECONDPLATELET ANTIGEN SYSTEM*

By N. RAPHAELSHULMAN,RICHARD H. ASTER, HOWARDA. PEARSONANDMERILYN C. HILLER

(From the National Institute of Arthritis and Metabolic Diseases, and the U. S. Naval MedicalCenter, Bethesda, Md.)

(Submitted for publication November 1, 1961; accepted December 17, 1961)

Although there is some evidence that neonatalthrombocytopenic purpura, which occurs in other-wise normal children born of healthy mothers, iscaused by maternal antibodies formed against fetalplatelets (1-3), lack of reliable serologic tech-niques generally has prevented convincing dem-onstration of the responsible antibodies and hasprecluded thorough investigation of the propertiesand inheritance of specific platelet antigens pre-sumed to be involved. Because experience in thislaboratory has shown that complement fixation isthe most sensitive technique for detecting severaldifferent types of antiplatelet antibodies (4-6),this technique was applied in a search for anti-bodies in the sera of normal mothers who hadgiven birth to thrombocytopenic infants. In thepresent report the isoantibodies which were foundare characterized, and the sensitivity of the com-plement fixation technique is compared with thatof other techniques employed in attempts to detectantiplatelet antibodies.

A well defined platelet antigen system recentlyhas been identified (6, 7) and the nature of theantigen, its reactions with antibody, and its modeof inheritance have been described in detail (6).One of the maternal isoantibodies studied in thepresent work permitted differentiation of a secondplatelet antigen. The properties of the new anti-gen are compared with those of the previously de-scribed antigen.

CASEREPORTS

Sera of four mothers, who had given birth to sixchildren with neonatal purpura, were tested for anti-

* Part of this work was presented at the Meeting ofthe American Federation for Clinical Research, April 30,1961, and appeared as an abstract in Clin. Res. 1961, 9,157.

body. All children were born at full term by normalspontaneous delivery; all developed petechiae and ecchy-moses within 1 to 6 hours after birth and had severethrombocytopenia, but their hemoglobin and leukocytecounts were normal and they had no other congenital de-fects. The four mothers were in good health, had nor-mal platelet levels shortly after parturition, and had nopast history of having idiopathic thrombocytopenic purpuraor of receiving blood transfusions. All of the infantsrecovered as described in the following individual casereports; and, thereafter, none had recurrence of thrombo-cytopenia during follow-up periods ranging from 3 monthsto 9 years.

Family H. A boy, born to a para I gravida II mother,had a platelet count of 12,000 per mm' 2 hours after birth.Cortisone, 40 mg daily, was given for 8 days, then re-duced to 10 mg over a 4-day period. Platelets roserapidly to 180,000 per mm3 on the third day and skinlesions disappeared in 5 to 6 days. Another son, born1 year earlier, had had a large ecchymosis at birth cover-ing most of the lower part of his body. His plateletcount was 5,000 per mm' on the day of birth, 25,000 after3 days, 70,000 after 12 days, 160,000 after 15 days, andnormal thereafter. Purpura persisted for 8 days and wasaccompanied by gross hematuria during the first week.

Family K. A girl, born to a para IV gravida IXmother, had fewer than 50,000 platelets per mm3 1 hourafter birth. Although no steroid therapy was given, herplatelets rose rapidly to normal levels within 4 days, dur-ing which time purpura cleared. A sister, 9 years older,developed severe purpura on the first day of life and atthe time was found to have "very low" platelets. She hadhemorrhagic manifestations for 8 days and required twoblood transfusions; but within 14 days her platelets hadreached normal levels, and within 3 weeks purpura haddisappeared. The mother of Baby K had had four mis-carriages at 3 to 5 months' gestation between the birthof her second and fifth children.

Family C. A boy, born to a para 0 gravida I mother,had 8,000 platelets per mmsand a strongly positive tour-niquet test 6 hours after birth. Tibial bone marrow as-piration done on the day of birth showed almost totalabsence of megakaryocytes. Daily treatment with pred-nisone, 10 mg intramuscularly, was begun immediately.Platelets rose to 80,000 per mm3 by the fourth day and

1059

SHULMAN, ASTER, PEARSONAND HILLER

then rose more gradually to 200,000 over the next 2weeks. Marrow obtained on the tenth day showed anormal number of megakaryocytes.

Family S. A girl, born to a para III gravida IVmother, was found to have only an "occasional" plateleton a blood smear taken several hours after birth. Shewas treated with cortisone, 10 mg daily. Petechiae andecchymosis gradually disappeared and platelets remained"very low" during the first week of life. Steroid therapywas discontinued after 8 days. Platelets rose gradually tonormal levels during the second and third weeks. Bonemarrow obtained on the third day of life was norrnsl,showing an adequate number of megakaryocytes.

MATERIALS ANDMETHODS

Methods of preparing platelet suspensions, countingplatelets, measuring platelet agglutination, complementfixation, inhibition of clot retraction, and nitrogen con-tent of platelet suspensions; treating platelets with tryp-sin; and performing antihuman globulin consumptiontests were as described in previous studies of platelet im-mune systems (4-6). Platelet suspensions generally werestandardized on the basis of N content, 800 ± 50 ltg N perml being equivalent to 2 X 108 platelets per mm3 (6).

Complement (C') activity was measured in terms of a50 per cent hemolytic unit (4), with a single hemolysinpreparation and erythrocytes from a single sheep. Withthe antigen systems studied in this report, the maximalamount of fixation occurred within 1 hour at 370 C. C'fixation was not potentiated by the use of buffers con-taining calcium and magnesium ions. In all reactionmixtures sufficient C' was added so that less than 75 percent would be fixed. Under these conditions the amountof C' fixed was independent of the amount of C' initiallypresent (4).

Additional platelet agglutination tests were performedaccording to the methods described by Dausset (8), andStefanini (9) and Harrington (1) and their co-workers,and the mixed antiglobulin test was done according toChalmers, Coombs, Gurner and Dausset (10). The tannedred cell agglutination test for antiplatelet antibodies wasperformed as described by Kissmeyer-Nielsen (11).

Antibody which was used to phenotype platelets in apreviously described platelet antigen system was obtainedfrom a patient who had developed the antibody afterblood transfusion (6). Methods of measuring this anti-body quantitatively and the definition of an antibodyunit are given in a previous paper (6), and the antigenwhich the antibody identified was labeled PlAl. The ra-tionale for the system of nomenclature used to describethe three phenotypes and genotypes that were recognizedwas presented. It was shown in the PlAl-anti-PlI" systemthat platelets from some individuals fix twice as muchC' per platelet as do platelets from other individuals.Family studies indicated that the doubly reactive plateletswere from individuals homozygous for the gene deter-mining the PIAl antigen [phenotype PlA(1,1)], while lessreactive platelets were from the individuals heterozygousfor that gene [phenotype PlA(l,-) ]. Moreover, it was

found that platelets from homozygous individuals ad-sorbed twice as much antibody per platelet as did plate-lets from heterozygous individuals. Nonreactive plateletswere from individuals lacking the gene [phenotypeP'(-,-) ]. These methods of phenotyping platelets areused in the present work with a maternal isoantibodythat identifies a new platelet antigen (see ExperimentalResults, section 2).

EXPERIMENTALRESULTS

1. Demonstration of maternal-fetal incompati-bility in the PIA- antigen system. The mother ofFamily H had had two thrombocytopenic infants,but serum obtained from her 1 year after the birthof her last child did not contain antibody when wetested it with the father's platelets, using C' fixa-tion and agglutination techniques at the limit oftheir sensitivity for detecting known antiplateletantibodies (4-6). However, when platelets fromthe mother, father, and second child were typedwith anti-PlAl, the mother was found to be a non-reactor, the father homozygous for PJA1, and thechild heterozygous for PlA1 (Figure 1). There-fore, maternal-fetal incompatibility in the PlI1-antigen system was established. Since less than2 per cent of the individuals in the general popu-lation have platelets lacking PlAl antigen (6), theprobability that neonatal purpura was associatedby chance with incompatibility in the plAl systemwas slight.

Serum from the mother of Family K, obtained1 day after the birth of her last thrombocytopenicinfant, showed no activity with the father's plate-lets in the usual C' fixation and direct agglutination

FAMILY H FAMILY K

FIG. 1. PLATELET TYPES OF FAMILIES H AND K DE-TERMINED WITH ANTI-PLA" ANTIBODY. Circles indicatefemales; squares, males. Solid symbols denote individualshomozygous for PIAl [phenotype PlA(1,1) ]; hatchedsymbols, heterozygous individuals [phenotype PlA(l,-)];open symbols, nonreactors [phenotype PI'(-,-) ]. Chil-dren who had neonatal thrombocytopenic purpura aredesignated "P."

1060

NEONATALPURPURA

tests. However, when platelets from this familywere phenotyped with anti-PlAl, again there wasmaternal-fetal incompatibility in the PlAl antigensystem (Figure 1). In view of the infrequency ofPlAl-negative individuals (less than 2 per cent),the probability that this incompatibility would oc-cur in two successive families by chance alone wasless than 0.0004. Therefore, the possibility wasconsidered that the serum of Mother K mightcontain a non-C'-fixing, nonagglutinating antibodywhich, nevertheless, might interfere with activityof the known anti-PlAl in a manner analogous tothat of "blocking" antibodies which develop againsterythrocyte antigens. Table I shows the type ofexperiment used to demonstrate a factor in theserum of Mother K which interfered with C' fixa-tion by anti-PlAl. Prior incubation of plateletscontaining the PlAl antigen with the serum fromMother K markedly reduced the amount of C'fixed by those platelets in the presence of anti-PlAl.When the order of addition was reversed, and theC'-fixing anti-PlAl was added to platelets first,interference with C' fixation was much less. Inter-ference with C' fixation appeared to be specific forthe PlAl-anti-PlAl system, for 75 per cent of theinterfering factor could be removed from 2 ml ofMother K's serum by adsorption with 109 plate-lets of phenotype PlA (1,1), but it remained un-changed after adsorption with the same number ofPlA(_,_) platelets. Moreover, when platelets con-taining PlAi were incubated in the serum of MotherK, centrifuged, washed once in 0.85 per cent NaCl,and resuspended, they remained incapable of fixingC' with anti-PlAl. Furthermore, Mother K's se-rum did not interfere with the C'-fixing activity of

TABLE I

Interference with C' fixation by serumfrom Mother K *

1 2 3Units C'

Prior incubation Later addition fixed

Mother's serum +platelets Anti-PIA2 +C' 0.5

Normal serum +platelets Anti-PI'- +C' 6.0

Anti-PlAl +platelets Mother's serum +C' 2.8

Platelets Mother's serum +C' 0

Anti-PlA1 +platelets C' 5.4

* Reagents in column 1 were incubated for 5 minutes at 37' C, fol-lowed by addition of reagents in column 2 and incubation for 1 hour at37' C. Platelets, type PIA (1,1) (6) at 2.5 X104 per mm2final concentra-tion, 5 U of anti-PI'1 (6), and 0.1 ml of serum from Mother K were usedin the various mixtures, the final volume in each tube being 0.4 ml, ad-justed with 0.85 per cent NaCl.

an isoantibody directed against a platelet antigenwhich differs (see below) from PlAl.

The antiglobulin consumption test, usingMother K's serum alone, was consistently butweakly positive with PlA(1,1) platelets, but nega-tive with PlA(-,-) platelets. The amount of anti-globulin consumed by PlA(1,1) platelets exposedto Mother K's serum was similar to the amountconsumed by the same number of PlA(1,1) plate-lets saturated with the C'-fixing anti-PlAl (6).

The titer of the interfering factor in Mother K'sserum could be estimated by determining the smal-lest amount of serum that would inhibit C' fixa-tion by a given amount of anti-PlA, with a givennumber of platelets. For instance, as little as0.04 ml of Mother K's serum completely preventedC' fixation by 107 PlA(1,1) platelets with 4 U ofanti-PlAl in a 0.4-ml total volume. By this methodof titration, no significant decrease in activity wasfound in Mother K's serum during the first 2months after delivery. Mother H's serum showedno ability to interfere with C' fixation by anti-PlA1,but her serum had been obtained 1 year after thebirth of her last child.

2. Demnonstration of maternal-fetal incompati-bility in a new platelet antigen system. Serumfrom the mothers of Families C and S, when testedwith platelets from both fathers in mixtures simi-lar to those used with anti-PlAl (6), produced C'fixation, as shown in Figure 2. Platelets fromboth mothers gave no C' fixation with their ownor each other's serum, whereas platelets obtainedfrom the affected babies, after they had recovered,reacted with serum from both mothers in a man-ner identical with that of platelets from the fathers.

The factor in the mothers' sera that producedC' fixation is referred to as an antibody because itwas found to be nondialyzable, was recoveredquantitatively in a fraction of serum precipitatedat one-third saturation with ammonium sulfate,and did not lose more than 10 per cent activity onheating for 2 hours at 560 C or after storage forat least 6 months at - 200 C but was rapidly de-stroyed when heated at 700 C.

With the standard curve shown in Figure 2,antibody activity in different serum samples couldbe compared by the amounts of sera required toproduce a given amount of C' fixation in thepresence of excess antigen (4, 6). With any oneserum sample the standard curve was reproducible

1061

SHULMAN, ASTER, PEARSONAND HILLER

from day to day with an error of ± 10 per cent,provided freshly prepared platelets were used.One unit of antibody activity was defined arbi-trarily as that amount required to fix 4 U of C'under conditions shown in Figure 2. The serumsample used in this figure, obtained 1 week postpartum from Mother C, contained 29 ± 2.9 anti-body U per ml. Antibody activity in Mother C'sserum, measured by this method, remained un-changed for approximately 2 months, then decayedgradually to 50 per cent of the initial value withinthe subsequent 2 months. The serum of MotherS contained 20 U of antibody per ml 4 years afterthe birth of her last thrombocytopenic infant, andantibody activity in her serum from the post-partum period contained 25 U per ml, after frozenstorage for 4 years. The frozen serum was theonly sample available on Mother S from the post-partum period.

Serum from the infant of Mother C was theonly infant's serum available from the neonatalperiod. The infant's serum obtained on the firstday of life contained antibody, but the titer wasapproximately one-tenth that of the mother'sserum obtained on the same day.

Sera from 37 mothers of unknown platelettype, obtained at the time they delivered, wereused in C' fixation tests with platelets known toreact with the sera from Mothers C and S. Thesesera were used at one-half dilution with plateletsat concentrations of both 50,000 and 300,000 permm3. None of the sera fixed C'. Previous preg-nancies included, the mothers tested had givenbirth to a total of 90 normal children.

The following observations show that the anti-bodies in sera of Mothers C and S had identicalspecificity for an antigen which differs from PlA.Both mothers' sera were tested against plateletsfrom 49 normal individuals selected at random.For these tests 2.3 U of antibody and a final plate-let concentration of 2.5 x 105 per mm3were usedin mixtures otherwise as in the legend of Figure2. The specificity of reactions with sera fromboth mothers was identical, C' fixation being ob-tained with platelets from 39 per cent of the donors.This frequency of positive reactors is in contrastto the positive reactors of greater than 98 per centfound with anti-PlAl. The platelet antigen iden-tified by antibody in serum from Mothers C and Swas different from the PlAl platelet antigen, as

well as from erythrocyte antigens A1, A2, 0, B,M, N, S P. C, D, E, c, e, Du, K, Fya, Lea andLeb, based on platelet and erythrocyte phenotypesdetermined on 22 individuals. The new plateletantigen will be referred to as PlB1 for reasons dis-cussed previously (6). Anti-PlBl could not bedemonstrated in sera from 12 normal PiB (_-) in-dividuals by C' fixation or agglutination techniques.

As with the PlAi platelet antigen (see Materialsand Methods), the amount of PlBI antigen perplatelet could be determined by the amount of C'fixed per platelet and by the amount of antibodyadsorbed per platelet in the presence of excess anti-body. Figure 3 shows the amount of C' fixed byplatelets from two different reactive donors. Withpreviously described methods for measuring anti-body adsorption (6), the platelets used to obtaincurve 1, Figure 3, adsorbed approximately twiceas much antibody per platelet as did the plateletsused to obtain curve 2, Figure 3. By inference

10 I I I

8

Uj~~~~~~~

0 6

Z _ 0

LU

D o2-

.01 .02 .03 .04 .05 .06 .07 .08 .09SERUM(ML.)

FIG. 2. STANDARDCURVE OF C' FIXATION WITH SERAFROM MOTHERSC AND S. Various amounts of serum,indicated on the abscissa, were incubated with platelets(final concentration 3 X 10' per mm') for 1 hour at 370 Cin the presence of guinea pig C' before residual C' wasmeasured. Total volume of incubation mixture was 0.4ml. Each symbol indicates an experiment done on a dif-ferent day, using a different platelet preparation. Oneserum sample from Mother C was used throughout andall platelets used were phenotype plB (1,-) (see text).Serum from Mother S gave an identical curve, but dif-ferent amounts of serum were required (see text),

1062

NEONATALPURPURA

from knowledge of the PlAl-antigen system, it appeared that platelets doubly reactive with anti-Pliwere from individuals homozygous for PlB1, anthat less reactive platelets were from individualheterozygous for PlBi. Platelets from only onof the positive reactors tested thus far have beeidoubly reactive. However, since positive reactoraccount for 39 per cent of the population testedonly approximately 4.8 per cent of the populationwould be expected (12) to be homozygous for thePIB1 gene. Further evidence that less reactiv(platelets represent the heterozygous state in thlPlB1 antigen system was obtained from the frequency of nonreactive platelets in offspring omatings between nonreactive individuals and individuals presumed to be heterozygous for the pjBgene, on the basis of C' fixation tests done as irFigure 3. Of the 15 offspring tested, 8 were non-reactive with anti-PlBl, and platelets from thosewho were reactive fixed C' quantitatively in thesame manner as did platelets from the reactiveparent. This is consistent with the expected 50-5Cdistribution of the P1Bi gene if the PlBl-positiveparent were, in fact, heterozygous. Phenotypesof Families C and S in the PlBi system are shownin Figure 4; in Table II the observed frequencyof the three possible phenotypes of the PlBi systemare compared with those of the PlAl system.

10

z

-J4

a-

6

0 2 4 6 8

PLATELET CONCENTRATIONX 10-5

FIG. 3. C' FIXATION WITH VARIOUS PLATELET CONCEN-TRATIONS AND A FIXED CONCENTRATIONOF ANTI-PIB". In-cubation mixtures contained 0.06 ml of the same serum

used in Fig. 2 and various concentrations of platelets in-dicated on the abscissa. Conditions otherwise as in Fig.2. Curve 1 was obtained with highly reactive plateletsfrom one individual (see text) and curve 2 was obtainedwith platelets which produced the usual reaction (seetext).

31id

IC

71~e-e

of

FAMILY C FAMILY S

FIG. 4. PLATELET TYPES OF FAMILIES C AND S DETER-MINED WITH ANTI-PLB" ANTIBODY. Symbols are as inFig. 1, hatched symbols representing individuals heterozy-gous for PI"B [phenotype pIB (1,-) ] and open symbols,nonreactors [phenotype P'(- ,)].

1 Although the frequencies of the two genes, PMAand PlBl, are consistent with their being alleles,the following results of phenotyping selected in-dividuals for both antigens indicate that PjBi isnot an allele of PMA'. The one individual who wasapparently homozygous for P1Bi was heterozygousfor PZAi; of seven individuals apparently heterozy-gous for PJBI, five were homozygous and two wereheterozygous for PMA'; and of three PlAl-negativeindividuals, two were also PlB1-negative and onewas apparently heterozygous for PjB1.

3. Properties of PIB1 antigen and PlBl-anti-PIBlcomplex. The C'-fixing activity of platelets con-taining PlB1 antigen was reduced 25 per cent afterheating for 10 minutes at 560 C and 75 per centafter 45 minutes at 560 C, but platelets heatedin this way retained full ability to adsorb anti-body. Platelets heated at 1000 C for 10 minuteslost all C'-fixing activity, as well as the ability toadsorb antibody. Lyophilized platelets reconsti-tuted in 0.85 per cent NaCl did not fix C' withantibody but adsorbed antibody almost as well asan equivalent number of fresh platelets. Plateletssuspended in 0.85 per cent NaCl lost approxi-mately 50 per cent of their C'-fixing activity overa 60-day period at 50 C, whereas platelets keptfrozen at - 200 C retained full ability to fix C'and adsorb antibody for at least 2 months.

Platelets exposed to trypsin for 20 minutes lostapproximately 50 per cent of their C'-fixing activ-ity, but more prolonged exposure to trypsin didnot result in further loss of activity. By an agargel diffusion technique, anti-PlBl did not form aprecipitin line with serum from PlBt-positive per-

1063

I

SHULMAN, ASTER, PEARSONAND HILLER

TABLE II

Frequency of platelet phenotypes identified by anti-PIBl *

PlB system PlA system

Probable Phenotype Number Phenotypegenotype symbol found Frequency symbol Frequency

Homozygousreactor PlB(ll) 1 2 PlA(1,1) 75.7

Heterozygousreactor P1B(1,-) 18 37.5 PlA(l,-) 22.6

Homozygousnonreactor P1B(-,-) 29 60.5 PlA(-,-) 1.7

* PiBi antigen per platelet measured as in Figure 3. Individuals whose platelets conformed to curve 2, Figure 3, wereconsidered heterozygous for the PlB5, while the one individual whose platelets reacted as in curve 1 was considered homo-zygous for PI15. Values for the PIA system are taken from Ref. 6.

sons known to be secretors of A, B, or H sub-stance. The PlB1 antigen was not detected onerythrocytes of P1B1-positive individuals, either byC' fixation or adsorption techniques, and was notdetected on platelets obtained from eight rabbits,seven dogs, and one sheep.

Some indication of the stability of P1B1-anti-Plicomplexes was obtained from the following experi-ments. Platelets saturated with anti-PlB1 by in-cubation in excess antibody were subjected to re-peated washes in 0.85 per cent NaCl and testedfor their ability to fix C' after each wash, as de-scribed in the legend of Figure 5. The Pl5B-anti-PlBi complexes [when we used apparent P153(1,1)or PlB (1,-) platelets] lost much more C'-fixingactivity than did PlAl-anti-PlAl complexes treatedin a similar fashion. Although PlAl-anti-P1Al com-plexes were highly resistant to dissociation by vari-ous forms of treatment (6), P1B1-anti-PlBl com-plexes could be reversed as follows: platelets satu-rated with anti-PlB were washed once quickly withcold 0.85 per cent NaCl, then suspended in 15 percent NaCl and heated at 560 Cfor 30 minutes. Thesupernatant fluid of heated platelets was tested forantibody activity after dialysis for 16 hours against0.85 per cent NaCl at 50 C. Approximately 20per cent of the antibody initially adsorbed on plate-lets was recovered by this technique.

The order of adding antibody, platelets, and C'to reaction mixtures had no effect on the amountof C' fixed by PlAl-anti-PlAl complexes (6); but,as shown in Table III, the amount of C' fixed bymixtures of PlBi and anti-PlBi was influenced bythe sequence of adding reagents. By comparing

results in Table III with results in Figure 2, it isseen that the effect of order of adding reagents onC' fixation was more pronounced when antibodywas in relative excess. Thus, much less C' wasfixed when platelets were preincubated with rela-tively large amounts of antibody before C' wasadded, than was fixed when C' was present in themixture before antigen and antibody were allowedto interact. The higher the ratio of antibody toplatelets, the greater was the discrepancy betweenthe amount of C' fixed when C' was added firstand last, whereas the order of adding C' had littleor no influence on the amount of C' fixed whenantibody was present in relatively small amounts.

4. Serologic techniques other than C' fixationused with anti-PIBl. The following tests, fre-quently used in the study of antiplatelet antibodies,were carried out with serum containing the high-est titer of anti-PlB1 and platelets known to be re-active in C' fixation tests with this antibody.

Direct platelet agglutination tests, performed byseveral methods using twofold serial serum dilu-tions from % to %46 to avoid a possible prozoneeffect, did not give agglutination. Weak agglu-tination was noted frequently with serum concen-trations greater than 50 per cent in experimentaltubes, but control tubes containing normal sera orplatelets nonreactive with anti-PlBi showed thesame agglutination. Specific agglutination, there-fore, could not be evaluated in mixtures contain-ing greater than 50 per cent serum; and withserum concentrations less than 25 per cent, noagglutination occurred in control or experimentalmixtures.

1064

NEONATALPURPURA

The mixed antiglobulin test described by Chal-mers and co-workers (10) and the indirect Coombstest, performed on platelets exposed to antibodyand washed in 0.85 per cent NaCl, did not givespecific agglutination. The antiglobulin consump-

tion test gave the same results with platelets whicheither did or did not react in C' fixation tests withanti-PlBl. The tanned red cell agglutination test(11) gave uniformly negative results, not onlywith anti-PlBl, but also with anti-PlAl.

There was no inhibition of clot retraction when

7 ~ ~ ~

6

LU3 2-

a-

0

'U

NUMBEROF WASHESFIG. 5. EFFECT OF REPEATEDWASHINGON COMPLEMENT

FIXATION BY PLATELETS SENSITIZED WITH ANTI-PL"1.Four ml of anti-PI"1 serum containing 28 antibody U perml was incubated for I hour at 37° C with 2 X 109 plate-lets [apparent phenotype Pl'(1,1)], following whichplatelets were sedimented at 3,000 G and resuspended incold 0.85 per cent NaCl at a concentration of 10' per mm'.An aliquot of the resuspended platelets was saved, and theremaining suspension was brought to 3 ml with cold0.85 per cent NaCI and centrifuged again. This washingprocess was repeated 4 times, and after each wash analiquot of the platelet suspension, containing 2.5 x 107platelets, was mixed immediately with 12 U of C' in afinal volume of 0.4 ml. Residual C' was measured afterincubation for 60 minutes at 37° C, and the symbolsrepresent the amount of C' fixed after each wash. Thedashed line represents results of a similar experimentdone, using platelets sensitized with anti-PI' (6).

TABLE III

Effect of varying order of adding reagentson C'fixation *

Units C'Prior incubation Later addition fixed

5 U anti-PlBl+platelets C' 3.4

Platelets+C' 5 U anti-PlBl 6.51 U antiPlBl+platelets C' 4.0

Platelets+C' 1 U anti-PlBl 4.4

* Prior incubation was carried out for 5 minutes at 370 C;serum was the same as that used in Figure 2; final platelet[apparent phenotype PlB(1,-)] concentration was 2 X 105per mm3, and 10 U of C' was used in mixtures which other-wise were as in Figure 2.

serum containing as much as 29 U of anti-PlBlper ml was mixed with an equal volume of freshblood from individuals whose platelets fixed C'with that antibody.

DISCUSSION

1. Clinical significance. The possibility thatmaternal isoantibodies can cause neonatal throm-bocytopenia has been considered previously (1-3,13, 14) but, in general, adequate serological con-firmation has been prevented by the lack of satis-factory techniques. Moulinier (3) was the onlyinvestigator who attempted to determine the fre-quency of a platelet antigen in the general popula-tion, using serum of a normal mother who hadgiven birth to thrombocytopenic infants. He usedthe antiglobulin consumption test with papaine-treated platelets, had difficulty with false-positivereactions, but concluded that about 22 per cent ofplatelets selected at random were positive for theantigen involved, which was called "duzo." Themode of inheritance of this antigen was not stud-ied. The present work establishes a genetic basisfor maternal-fetal incompatibility in two well de-fined platelet-antigen systems and describes thenature of the reactions of antibodies that can causeneonatal purpura. In one family, evidence forisoimmune origin of the purpuric disease was basedonly on the finding of maternal-fetal incompatibil-ity of PlAl; in another family, on incompatibilityof PlAl as well as the presence of a factor in themother's serum that interfered with C'-fixing ac-tivity of anti-PlAl; and in two other families, onthe presence of a C'-fixing antibody in the mothers'sera and one infant's serum against a newly recog-nized antigen, PlB1.

1065

SHULMAN, ASTER, PEARSONAND HILLER

Since 61 per cent of the population tested lackthe PlB1 antigen, and 39 per cent have it, approx-imately [0.61 x 0.39 = ] 24 per cent of all mar-riages are incompatible for this antigen. Fatherswho carry the antigen are almost all heterozygous;hence there is an opportunity for maternal im-munization by PlB1 in approximately [1/2 x 24per cent = ] 12 per cent of all pregnancies. How-ever, immunization against P1Bl does not appearto occur so frequently as this (see ExperimentalResults, section 2). The true frequency of iso-immune neonatal thrombocytopenia is not accur-ately established, for in most of the patients pur-puric symptoms subside within a few days to aweek (see Case Reports and Refs. 2, 15) and thedisease might not be diagnosed if a platelet countis not done during the first few days of life. Therecent report by Kaplan (16) suggests that neo-natal thrombocytopenia is more common than hasbeen thought. Although erythroblastosis is un-usual in the first-born child, neonatal purpura oc-curred in the first-born child in Families H and C,owing to anti-PlAl and anti-PlBl, respectively.This suggests that platelets may cross the placentaand induce sensitization more easily than do redcells. There may be a number of antigens otherthan PlBl and PlAl that may cause maternal iso-immunization.

Because some thrombocytopenic infants bornto normal mothers have abundant megakaryocytesin their bone marrow (1, 17), while others lackmegakaryocytes (2, 15, 18), neonatal purpura hasbeen classified as megakaryocytic or amegakaryo-cytic (18), implying difference in etiology. Oneof the infants in the present report had normalmegakarocytes (Family S), and one had virtualabsence of megakaryocytes (Family C). Sincethe same isoantibody, anti-PlBl, was no doubt re-sponsible for the thrombocytopenia in both cases,there seems to be no good reason to divide thesyndrome of neonatal thrombocytopenic purpuraarbitrarily into megakaryocytic and amegakaryo-cytic forms. The fact that megakaryocytes maybe absent at birth in isoimmune neonatal purpura,owing to incompatibility of the PlB1 antigen, sug-gests that this antigen may be present on mega-karyocytes or their precursors, but it is not clearas yet why some purpuric infants have adequatemegakaryocytes and others do not. In adults thedestruction of circulating platelets by other well

defined antibodies has not been observed to causehypoplasia of megakaryocytes (6, 19).

Mothers having idiopathic thrombocytopenicpurpura (ITP) frequently give birth to thrombo-cytopenic infants (1, 20). It is interesting, withrespect to duration of thrombocytopenia, that pur-puric infants born of mothers with ITP usuallytake more than 5, and up to 16, weeks to recover(1, 20-26), whereas recovery from isoimmuneneonatal purpura usually occurs in less than 2weeks (see Case Reports and Refs. 2, 15). Thus,it appears that the suspected antiplatelet factor ofITP differs from the observed antiplatelet isoanti-bodies, either in its mode of action or in its abilityto be cleared from the infant's circulation. Al-though ITP has many characteristics of an im-munologic disorder, antibody has not been de-tected in sera of patients with ITP by the generaltechniques used herein (6); but conceivably ITPcould be caused by a "blocking" antibody that isnot detectable by direct measurements (see Dis-cussion, Section 3). If the ITP antiplatelet factoris indeed an antibody, the long period of clearancefrom the infant's circulation may reflect transfer-ence of relatively large amounts of antibody acrossthe placenta.

The appropriate form of therapy to be used incases of isoimmune neonatal purpura is not as yetclear, for effects of steroid hormones, splenectomy,or other forms of treatment on the natural courseof the disease cannot be evaluated without regardto the nature and level of the responsible antibody.With recognition and means of measuring differ-ent specific antibodies and different types of anti-body activity, evaluation of therapy will be sim-plified. In some instances exchange transfusionmay prove to be useful, as it is in erythroblastosisfetalis, for an isoantibody of post-transfusion im-munologic purpura has been successfully removedin this way (6). It should be pointed out, how-ever, that administration of platelets that reactwith a circulating isoantibody may cause severereactions (3, 6).

2. Comparison of the PI-1 and PlB1 antigens.Just as in the case of the PlAl antigen (6), thePlB1 antigen is inherited as a dominant characterand, more specifically, appears to be inherited as aco-dominant character dependent on a gene cap-able of expressing itself in a single or double dose.No evidence was obtained for free PlB1 antigen

1066

1067NEONATALPURPURA

in serum of individuals who had PlBl-positiveplatelets, for plBl antigen on erythrocytes, or for a

naturally occurring anti-PlBl. The same was true

for PJAl antigen. The PlB1 antigen was distinctfrom a number of erythrocyte antigens and fromthe PlAl antigen, and PjBi was not inherited as an

allele of PIA1. Some animal platelets containingPlAl antigen did not contain PlBl antigen.

Platelets containing PlB1 antigen did not losetheir ability to fix C' with antibody as readily afterheating as did PlAl-positive platelets. Severalforms of treatment resulted in loss of the abilityof PlBi-positive platelets to fix C' with anti-PlBl,but not in loss of their ability to adsorb anti-PlBi.This has been observed with PlAlpositive plateletsas well, and suggests that the C'-fixation reactionis dependent on factors in addition to antigen-antibody combination (4). As in the case ofthe PlAl antigen, the PlB1 antigen was resistantto denaturation and to inactivation by trypsin, butwhether the antigenic material is a polysaccharideor protein remains uncertain.

3. The problem of measuring antiplatelet anti-bodies. Detection of a factor in Mother K's serum

which interfered with C' fixation by a known anti-body (see Results, section 1) is analogous to de-tection of the so-called "blocking" antibodies, wellknown to develop against erythrocyte antigens.In the present work, interference with C' fixation,rather than with cellular agglutination, was thebasis for detecting what appears to be an im-mulologically specific nonagglutinating, noncom-

plement-fixing antibody.The C'-fixing reactions of anti-PlBl differed in

some respects from reactions of other platelet anti-bodies which have been described (4, 6). In

assaying anti-PlBl, the general principle of varyingplatelet concentration to find an optimal range fora given antibody concentration (4, 6) applied, butdifferent ranges of platelet concentration were re-

quired in assaying anti-PlBl (Figure 3) and anti-PlAl. This finding indicates that in attempts to

detect unknown antiplatelet antibodies by C' fixa-

tion techniques, platelet concentration in reactionmixtures should be varied over a wide range, at

least from 25,000 to 500,000 per mm3. The order

of adding reagents markedly influenced the amount

of C' fixed by anti-PlBj (see Results, section 3),but had no influence on the amount of C' fixed byanti-PlAl (6). The significance of this phenom-

enon is not clear, but the observation indicates theimportance of varying the order of adding reagentsto C'-fixation mixtures when searching for un-known antibodies.

Various techniques other than C' fixation ap-plied in attempts to detect anti-PlBi were unsuc-cessful (see Results, section 4). We would ex-pect the tanned red cell agglutination tests to benegative in view of the fact that PlB1 antigen isnot present in saline extracts of PlB1-positive plate-lets. The relatively weak association between PlBiand anti-PlBi (see Results, section 3) may accountfor negative results with methods that require re-peated washing of sensitized platelets, such as theantiglobulin consumption test and the mixed anti-globulin test. By contrast, anti-PlA1, which com-bines very firmly with antigen, gave a positiveantiglobulin consumption test (6). Inability ofvarious direct agglutination techniques and the clotretraction inhibition test to detect anti-PlB1 areobservations awaiting explanation, but perhapsfailure of these tests also may be related to theinstability of PlB1-anti-PlBl complexes.

The greater sensitivity and versatility of C'-fixa-tion reactions, compared with other serologicaltests used with the maternal isoantibodies, parallelsobservations made on other anti-platelet antibodysystems (4, 6, 27). This indicates that no searchfor antiplatelet isoantibodies can be consideredcomplete without application of C' fixation tests.

SUMMARY

1. Maternal immunization against antigens onfetal platelets was found to be the basis of mega-karyocytic or amegakaryocytic neonatal thrombo-cytopenic purpura in six children born of fournormal mothers.

2. In two families there was mother-fetal incom-patibility with respect to the previously describedPlAl platelet antigen, and in the other families withrespect to a newly recognized antigen, PiBI. Themode of inheritance and other properties of P1Blantigen are described.

3. Complement fixation proved to be the onlyserologic technique by which anti-PlB1 antibodycould be detected. Characteristics of reactions inthe PlB1-anti-PlBl system are described, and cer-tain factors that influence measurement and detec-tion of antiplatelet antibodies are discussed.

4. One of the maternal isoantibodies could be

SHULMAN, ASTER, PEARSONAND HILLER

measured only by its ability to block the reactionsof a known complement-fixing anti-PlAl antibody.

ACKNOWLEDGMENT

We are indebted to Dr. William J. Harrington, Dr.Sanford Leiken and Dr. Gerald Miller for their assistancein studying two of the reported families. The technicalassistance of Mrs. Winifred Biehl is greatly appreciated,and we wish to thank Mrs. Mary H. McGinniss for de-termining the erythrocyte phenotypes.

ADDENDUM

After this report was submitted, several isoantigens,including the antigen labeled PIBl, were found to bepresent on granulocytes and lymphocytes, as well as onplatelets. A more appropriate descriptive symbol forPlBI would be PlGrLyB` to distinguish it from antigenslimited to one cell type. The report which describes anadditional shared antigen, PlGrLyCl, and an antigenlimited to lymphocytes, LyDi, and explains the inductionof only thrombocytopenia in infants exposed to anti-bodies against antigens shared by platelets and leuko-cytes, will appear shortly (28). The antigen pIAl is onplatelets only.

REFERENCES

1. Harrington, W. J., Sprague, C. C., Minnich, V., Moore,C. V., Aulvin, R. C., and Dubach, R. Immunologicmechanisms in idiopathic and neonatal thrombo-cytopenic purpura. Ann. intern. Med. 1953, 38,433.

2. Schulman, I., Smith, C. H., and Ando, R. E. Con-genital thrombocytopenic purpura: Observations onthree infants born of a non-affected mother; dem-onstration of platelet agglutinins and evidence forplatelet isoimmunization (abstract). A. M. A. J.Dis. Child. 1954, 88, 784.

3. Moulinier, J. Iso-immunisation maternelle anti-plaquettaire et purpura neo-natal. Le systeme degroupe plaquettaire "duzo." Trans. 6th Congress ofEuropean Society of Haematology, 1957. Basel,Karger, 1958, Part 2, p. S817.

4. Shulman, N. R. Immunoreactions involving platelets.I. A steric and kinetic model for formation of acomplex from a human antibody, quinidine as ahaptene, and platelets; and for fixation of comple-ment by the complex. J. exp. Med. 1958, 107, 665.

5. Shulman, N. R. Immunoreactions involving plate-lets. III. Quantitative aspects of platelet agglutina-tion, inhibition of clot retraction, and other reac-tions caused by the antibody of quinidine purpura.J. exp. Med. 1958, 107, 697.

6. Shulman, N. R., Aster, R. H., Leitner, A., and Hiller,M. C. Immunoreactions involving platelets. V.Post-transfusion purpura due to a complement-fixing antibody against a genetically controlledplatelet antigen. A proposed mechanism for throm-

bocytopenia and its relevance in "autoimmunity."J. clin. Invest. 1961, 40, 1597.

7. van Loghem, J. J., Jr., Dorfmeijer, H., van der Hart,M., and Schrender, F. Serological and geneticalstudies on a platelet antigen (Zw). Vox Sang.(Basel) 1959, 4, 161.

8. Dausket, J. Normal and pathological platelet agglu-tinins investigated by means of the shaking method.Vox Sang. (Amsterdam) 1954, 4, 204.

9. Stefanini, M., Plitman, G. I., Dameshek, W., Chatter-jea, J. B., and Mednicoff, I. B. Studies on plate-lets. XI. Antigenicity of platelets and evidence ofplatelet groups and types in man. J. Lab. clin.Med. 1953, 42, 723.

10. Chalmers, D. G., Coombs, R. R. A., Gurner, B. W.,and Dausset, J. The mixed antiglobulin reactionin the detection of human iso-antibodies againstleucocytes, platelets and hela cells. Brit. J. Haemat.1959, 5, 225.

11. Kissmeyer-Nielsen, F. Demonstration of platelet an-tibodies by haemagglutination of antigen coatedtanned erythrocytes. Vox Sang. (Amsterdam)1953, 3, 123.

12. Stern, C. Principles of Human Genetics. San Fran-cisco, Freeman, 1950, p. 152.

13. Mahon, R., Moulinier, J., Cantorne, G., and Lassalle,G. Purpura habituel du nouveau-ne par incom-patibilite foeto-maternelle antiplaquettaire. Gynec.et Obstet. 1957, 56, 517.

14. Garrett, J. V., Giles, H. McC., Coombs, R. R. A., andGurner, B. W. Neonatal purpura with platelet iso-antibody in maternal serum. Lancet 1960, 1, 521.

15. Akerren, Y., and Reinand, T. Thrombocytopenia inthe new-born period. Acta med. scand. 1950,suppl. 246, 281.

16. Kaplan, E. Congenital and neonatal thrombocyto-penic purpura. A review. J. Pediat. 1959, 54, 644.

17. Bluestone, S. S., and Maslow, H. L. Essentialthrombocytopenic purpura in the newborn infant.Report of first case treated by splenectomy. Pedi-atrics 1949, 4, 620.

18. Hugh-Jones, K., Manfield, P. A., and Brewer, H. F.Congenital thrombocytopenic purpura. Arch. Dis.Childh. 1960, 35, 146.

19. Shulman, N. R. Immunoreactions involving plate-lets. IV. Studies on the pathogenesis of thrombo-cytopenia in drug purpura using test doses of quini-dine in sensitized individuals; their implicationsin idiopathic thrombocytopenic purpura. J. exp.Med. 1958, 107, 711.

20. Epstein, R. D., Lozner, E. L., Cobbey, T. S., Jr., andDavidson, C. S. Congenital thrombocytopenicpurpura. Purpura hemorrhagica in pregnancy andin the newborn. Amer. J. Med. 1950, 9, 44.

21. Neville, M. L., and Masterman, L. M. Neonatalthrombocytopenic purpura in two infants. Arch.Dis. Childh. 1954, 29, 163.

22. Killander, A. On the use of exchange transfusion in

1068

NEONATALPURPURA

neonatal thrombocytopenic purpura; report of acase. Acta paediat. (Uppsala) 1959, 48, suppl. 117,29.

23. Morris, M. B. Thrombocytopenic purpura in thenewborn. Arch. Dis. Childh. 1954, 29, 75.

24. Randak, E. F., and Danforth, D. N. Thrombocyto-penic purpura in pregnancy; report of case. Quart.Bull. Northw. Univ. med. Sch. 1951, 25, 199.

25. Vandenbroucke, J., and Verstraete, M. Thrombo-cytopenia due to platelet agglutinins in the new-born. Lancet 1955, 1, 593.

26. Bridges, J. M., and Carre, I. J. Congenital thrombo-cytopenic purpura treated by exchange transfu-sion. Arch. Dis. Childh. 1961, 36, 210.

27. de Nicola, P., Rosti, P., and Zangaglia, 0. Comple-ment fixation test due to the interaction of specificantiplatelet serum and heterologous platelet anti-gen. J. Lab. clin. Med. 1955, 45, 725.

28. Shulman, N. R., Marder, V. J., Aledort, L. M., andHiller, M. C. Complement-fixing isoantibodiesagainst antigens common to platelets and leuko-cytes. Trans. Ass. Amer. Phycns 1962. In press.

IMPORTANTNOTICE

THE EDITORIAL ANDBUSINESS OFFICES OF

THE JOURNALOF CLINICAL INVESTIGATION

WILL MOVEON

MAY 1, 1962

OLD ADDRESS: THE JOURNALOF CLINICAL INVESTIGATION333 CEDARSTREETNEWHAVEN11, CONNECTICUT

NEWADDRESS: THE JOURNALOF CLINICAL INVESTIGATION10 STOUGHTONSTREETBOSTON18, MASSACHUSETTS

1069