Pooling of platelets in the spleen: role in thepathogenesis of "hypersplenic"thrombocytopenia.

R H Aster

J Clin Invest. 1966;45(5):645-657. https://doi.org/10.1172/JCI105380.

Research Article

Find the latest version:

http://jci.me/105380-pdf

Journal of Clinical InvestigationVol. 45, No. 5, 1966

Pooling of Platelets in the Spleen: Role in the Patho-genesis of "Hypersplenic" Thrombocytopenia *

RICHARDH. ASTERt(From the Thorndike Memorial Laboratory and Second and Fourth [Harvard] Medical Serv-

ices, Boston City Hospital, and the Department of Medicine, Harvard Medical School,Boston, Mass.)

Reduction in one or more of the circulating ele-ments of the blood often attends splenic enlarge-ment. Anemia has been attributed to hemolysisplus pooling of red blood cells in the spleen (34).The causes of the leukopenia and thrombocyto-penia seen in splenomegalic subjects are less wellunderstood. Thrombocytopenia has been ascribedboth to bone marrow suppression through a hu-moral mechanism (7-9) and to premature destruc-tion of cells in the enlarged spleen (10-11). Thepresent work explores still a third possibility: thatthe general circulation is depleted of platelets bytheir entry into a large exchangeable splenic pool.

Methods

Labeling of platelets with 'Cr and calculation of plate-let recoveries after transfusion were carried out as de-scribed previously (12-13). The value used for plateletrecovery was the per cent of transfused platelets remain-ing in the general circulation for 2 hours. From 1 to 4hours after infusion of 'Cr-labeled platelets into normalor splenomegalic subjects there was no significant varia-tion in circulating 'Cr. Duplicate determinations of plate-let recovery in the same subject never varied from eachother by more than 10%.

In most studies 'Cr-labeled autogenous platelets wereemployed. Where it was necessary to use homologousplatelets, these were obtained from ABO, Rh-compatible

* Submitted for publication April 2, 1965; acceptedJanuary 4, 1966.

Supported in part by grants HE-07652-02 and Ti-AM-5391-02 from the U. S. Public Health Service and byresearch grant FR-76 from the Division of Research Fa-cilities and Resources, National Institutes of Health,Bethesda, Md.

Part of this work was presented previously in prelimi-nary form (1, 2), and at the Seventy-eighth AnnualMeeting of the Association of American Physicians, At-lantic City, N. J., May 4-5, 1965.

t Medical Foundation research fellow.Address requests for reprints to Dr. Richard H. Aster,

Thorndike Memorial Laboratory, Boston City Hospital,Boston, Mass. 02118.

donors. In the absence of isoimmunity from pregnancyor previous transfusion, homologous platelets behave likeautogenous cells (13-14).

Surface scanning of body organs was done with a di-rectional scintillation counter using a radiation spectrom-eter to reduce background to less than 100 cpm.1 Nor-mal-sized spleens were scanned with the subject lyingon the right side. The orifice of the collimator was placedon the skin surface in the midaxillary line with its centerdirectly over the line of diaphragmatic dullness (midin-spiration). Maximal splenic surface activity was nearlyalways obtained at this site. The solid angle encom-passed by the scanning device was sufficient to includethe entire normal spleen. Deep inspiration reduced sur-face counts per minute over the spleen, but the countingrate was not affected by casual breathing. With enlargedspleens that extended into the abdomen, an anterior sitewas chosen for scanning. The anterior of the liver wasscanned with the medial margin of the collimator at themidline and the superior margin i inch above the area ofdiaphragmatic dullness. The anterior of the lung wasscanned over the right upper lobe. Surface counts perminute were divided by ten times the number of micro-curies of 'Cr given so as to make values independent ofthe dose of radioactivity. The number of counts per min-ute to be expected over various sites on the basis of bloodcontent alone was determined by scanning normal sub-jects 1 hour after transfusion of 'Cr-labeled red bloodcells. With the particular scanning device used, liver,spleen, and lung counts per minute were 50 to 75%o of theprecordial values, whereas spleen: liver ratios averaged1: 1 in agreement with the observations of Jandl, Green-berg, Yonemoto, and Castle (15). In asplenic subjects,"spleen": liver ratios ranged from 0.5 to 0.8. A sig-nificant deviation from these values after transfusion ofplatelets indicated that platelet-5Cr was concentrated(relative to erythrocytes) in a particular site. Previousstudies have suggested that actual total radioactivity innormal-sized organs can be crudely estimated from ex-ternal scanning with various standards for comparison(12, 16-18). Direct assay of 'Cr in three surgically re-moved normal-sized spleens (idiopathic thrombocytopenic

1 Nuclear-Chicago DS-301 scintillation detector with3-inch diameter crystal and collimator 3 inches wide and3 inches deep and 132B spectrometer-scaler with windowat 100 kv. Counts over various body sites could be re-produced with an error of less than 5%c.

645

RICHARD H. ASTER

purpura) showed that estimates of radioactivity obtainedwith external scanning and previously derived conversionfactors were in error by only 15, 20, and 22%, respec-tively (19). The obvious limitations of such measure-ments have been noted by a number of observers (15-19).Jandl and co-workers (15) as well as Hughes Jones,Mollison, and Veall (17) emphasized the geometricalcomplexities that make quantitation difficult. Wolfand Fischer (20), who used a realistic plastic "phantom,"also advised caution in calculating organ radioactivityfrom external scanning data. In the present work,ratios of radioactivity over one site as compared toanother were relied upon in studying organ distribu-tion of platelets. When 'Cr was released from thespleen by epinephrine injection, surface values before andafter drug infusion were compared with each other so asto utilize the patient as his own control. Estimates ofabsolute organ radioactivity made from external scanningwere consistent with, but not crucial for, the conclusionsreached.

Platelets were counted by the method of Brecher andCronkite (21) with blood obtained from an arm vein andanticoagulated with EDTA. When platelet levels wereless than 100,000 per mm8, dilution was 1: 20 rather than1: 100 in order to improve counting accuracy.

Epinephrine infusion. Epinephrine hydrochloride di-luted 1: 125,000 in 5% dextrose was given intravenouslyover a 20- to 25-minute period at a constant rate. Base-line platelet levels were determined in duplicate, andcounts were obtained at 5 minutes, at 15 minutes, and atthe end of the epinephrine infusion. A total of about 7 jig

100-

Iao.

OlbO F

- IXA

- 2O

Z Z

e .Cr 040-

Xi

epinephrine per kg of body weight was given (0.49 mgfor a 70-kg subject), and pulse and blood pressure weremonitored during the infusion. Smaller doses were usedin persons over 50 years of age. Patients with knowncardiovascular disease were excluded from this study.Pallor was observed within 2 to 3 minutes after the startof infusion, and palpitation was noted for the first fewminutes by all the recipients. Systolic blood pressureincreased approximately 30 to 50 mmHg above base-linelevels. Rarely, premature ventricular contractions oc-curred. In general, however, side effects with slow, con-trolled infusion were less than with subcutaneous injec-tions of the same amount of adrenalin and disappearedwithin 3 to 5 minutes after terminating the infusion. Noadverse reactions occurred in any subjects.

Normal subjects comprised medical investigators orinformed convalescent patients hospitalized on the medicalwards for illnesses unrelated to blood or blood-formingorgans.

Results

Transfusion of 51Cr-labeled platelets to normalsubjects. Previous studies have shown that whenplatelets are labeled in acid citrate and transfusedto normal persons, 50 to 80% are recovered in theperipheral blood 1 to 2 hours after transfusion(average, 62%). The remaining platelet-51Cr isconcentrated largely in the spleen (12, 13). Anormal disappearance curve of 51Cr-labeled plate-

0 5 10 15 20 25 30 0 5 D0 15 20 25 30

MINUTES AFTER INJECTION

FIG. 1. ACUTEDISAPPEARANCEFROMTHE CIRCULATION (DESCENDING CURVES) OF 'CR-LABELEDAUTOGENOUSPLATELETS TRANSFUSEDTO A NORMALSUBJECT (LEFT) AND A PATIENT WITHSPLENOMEGALY(RIGHT). Splenic surface radioactivity as per cent of maximal value is shownin the ascending curves.

646

POOLINGOF PLATELETS IN THE SPLEEN

lets typical of four such studies in which multipledeterminations were made during the first 30 min-utes after transfusion is given in Figure 1 (left).The initial rate of platelet disappearance was about5% per minute. In normal subjects, circulatingplatelet 51Cr was minimum at about 10 minutes,after which there was often an increase of about5% to a level that remained steady for severalhours. In all cases, splenic surface radioactivityincreased in direct proportion to the rate at which51Cr-labeled cells disappeared from the blood. Incontrast, there was a decrease in radioactivity overliver, lung, and precordium consistent with theamount of 51Cr leaving the bloodstream duringthe same time interval.

After the platelet-5iCr level had stabilized in theperipheral blood, the ratio of spleen: liver andspleen: precordial radioactivity averaged 5: 1 and4: 1, respectively, as compared to average values ofabout 1: 1 and 2: 3, respectively, after injection of51Cr-labeled red blood cells.

Transfusion of 51Cr-labeled platelets to asplenicsubjects. Survival of autogenous 51Cr-labeledplatelets was studied in six subjects splenectomizedfor a variety of reasons 2 to 12 weeks previously(four for traumatic rupture, one for "spontaneousrupture," one for splenic tuberculosis). Recover-ies of transfused platelets were higher than in nor-mal persons, ranging from 80 to 100% and aver-aging 91% (Figure 2). In the asplenic subjects,labeled platelets disappeared over 8 to 10 days ata rate approximately equal to that of normal sub-jects. Each point on the disappearance curveswas about 50% higher than the correspondingpoint on the curve of an average normal subject.

Transfusion of 51Cr-labeled platelets to spleno-megalic subjects. Platelet survival was studied ineleven patients with congestive splenomegaly sec-ondary to Laennec's cirrhosis, of whom three hadascites or jaundice or both. There was no evi-dence of bleeding at the time of study. Five pa-tients were studied in whom splenic enlargementwas primarily of a proliferative type: two chroniclymphatic leukemia, two polycythemia vera, oneinfectious mononucleosis. In five instances ho-mologous platelets were used. None of the pa-tients receiving homologous platelets had beentransfused more than three times previously, andno significant differences were observed in the re-covery and survival of homologous, as opposed to

100'

o 80

-0z

o 60I~-hiJ

!.CLwi 40-0I-z

:0

hIa.*,

POST-SPLENECTOMY0

0

0

I 0

NORMAL

CD

00000coo

a

8

SPLENOMEGALY

0

0

80

db0

08

FIG. 2. RECOVERY IN THE GENERAL CIRCULATION OF1CR-LABELED PLATELETS TRANSFUSEDTO ASPLENIC, NOR-

MAL, AND SPLENOMEGALICSUBJECTS. "Recovery" is thepercentage of platelets remaining in the general circula-tion 2 hours after injection.

autogenous, cells in the splenomegalic group.Platelet levels ranged from 27,000 to 300,000 per

mm.After transfusion to splenomegalic subjects, most

51Cr-labeled platelets were rapidly cleared fromthe general circulation (Figure 1, right). The dis-appearance rate averaged 13%o per minute for thefirst few minutes after transfusion as comparedwith 5%o per minute in normal subjects. After 8or 9 minutes, a relatively steady level of 51Cr ac-tivity in the peripheral blood was maintained forthe next few hours. After circulating platelet5TCr had stabilized, surface radioactivity over theliver and lung was that which would be expectedfrom normal blood flow through these areas, i.e.,there was no evidence for concentration of 51Cr ineither site. In contrast, 51Cr accumulated in thespleen, the ratio of spleen: liver radioactivity rang-ing from 7: 1 to 15: 1 (range of 20 normal sub-jects, 2:1 to 6:1).

Platelet recoveries (platelet-51Cr remaining inthe general circulation 2 hours after transfusion)were uniformly low in the splenomegalic patients,averaging 23%o (Figure 2). In general, lower

647

RICHARD H. ASTER

060

z

1J

X 2z

I-

w

5-

IL 20-Ua.0

z

0

0-

0

0

*00

0

0

0

0

0

A A

1.0 20 3.0RECIPIENT'S PLATELET LEVEL (X105 PER MM3

FIG. 3. RELATIONSHIP BETWEENPLATELET LEVELS AND

RECOVERYIN THE GENERALCIRCULATION OF 5'CR-LABELEDPLATELETS INJECTED INTO SPLENOMEGALICPATIENTS. 0 =

congestive splenomegaly secondary to cirrhosis, 0 =

chronic lymphatic leukemia, A = infectious mononucleo-sis, and A = polycythemia vera.

recoveries were obtained in patients with more se-

vere thrombocytopenia (platelets less than 60,000per mm3) (Figure 3). Lowest recoveries were

obtained in patients with the largest spleens, as

estimated by palpation. In two splenomegalic pa-

200-

80.

Uj M

~- 2 0.

0

2 4 6 8 10

DAYS

FIG. 4. SURVIVAL OF 'CR-LABELED PLATELETS IN 13SPLENOMEGALICPATIENTS. Platelet 'Cr present in thegeneral circulation 2 hours after infusion of labeled plate-lets was taken as the 100% value, and platelets remainingon subsequent days were related to this. Shaded area

represents the range and horizontal lines the average

values for 30 normal subjects 1 to 10 days after infu-sion of platelets. Although initial platelet recoverieswere low in splenomegalic patients (Figure 3), the sur-

vival time of platelets remaining in the general circula-tion was nearly normal.

tients originally given autogenous cells, repeat ex-

periments with homologous platelets resulted insimilar recoveries: 18%o and 34%o as compared to20% and 30%o, respectively, for autogenous plate-lets. In a third patient labeled platelets were di-vided into two portions: one was returned to thedonor, and the other was transfused into a normalrecipient. Recovery was 15 %S in the splenomegalicpatient and 50%o in the normal. Approximatelythe same value (55%o) was obtained in the normalsubject with autogenous cells.

+120

-j

> 100'-J

z

-j

4 +60

0 +40

0 +40+20z

w 0-

-20-

0

0

0

0

8 0

8

8

ASPLENIC NORMAL SPLENOMEGALY

FIG. 5. PER CENT CHANGEIN PLATELET LEVELS AFTER

INFUSION OF EPINEPHRINE TO ASPLENIC AND NORMALSUB-JECTS AND SPLENOMEGALICPATIENTS WITH THROMBOCYTO-PENIA (PLATELETS 40,000 TO 120,000 PER MM8). The maxi-mumof three platelet determinations obtained during epi-nephrine infusion was used. Dark circle denotes per centincrease in a patient with splenomegaly (infectious mono-

nucleosis) but normal platelet level (300,000 per mm8).Although percentage increase in platelets was greaterin splenomegalic patients than in normal subjects, theabsolute increase was equal to or less than that in nor-

mals, since initial platelet concentrations were lower.

Life-span curves of platelets that remained inthe general circulation are shown in Figure 4. Theaverage time required for circulating 51Cr to di-minish to 50%, 25%yo, and 12.5% of the initial val-ues was 3.65, 5.83, and 7.63 days, respectively.These figures were 85%o, 877o, and 95%o of thecomparable normal values.

Effect of epinephrine on platelet levels in normalsubjects. It is well known that pharmacologicdoses of epinephrine cause an increase in circu-lating platelets (22-24), not -accounted for by thesmall associated changes in blood volume (25, 26),

648

I

0

POOLING OF PLATELETS IN THE SPLEEN

W 1 CIRCULAT/NO

En V

0 \ SPLE/J/C SURFACECr -20 \ RADIOACrIV17Y

-40

20 4A0MINUTES AFTER START OF EPINEPHRINE

FIG. 6. PER CENT CHANGEIN PLATELET LEVELS, CIRCU-LATING PLATELET 'CRI ANDSPLENIC SURFACERADIOACTIVITYAFTrER INFUSION OF EPINEPHRINE TO A NORMALSUBJECT 4HOURS AFTER TRANSFUSION OF LABELED PLATELETS. ASplatelet concentrations and 'Cr increased in the peripheralblood, there was a decrease in splenic surface radio-activity.

but the source of these cells has not been deter-mined. When0.5 to 1.0 mgof epinephrine was in-jected subcutaneously or intramuscularly into nor-mal subjects, platelets usually increased, but theresponses (and the intensity of side effects) werequite variable. With slow intraven ous infusion ofepinephrine (see Methods) a -much more uniformplatelet response was achieved. A steady increasein platelet levels followed infusion of epinephrine,reaching a peak after about 15 minutes. Plateletlevels remained elevated as long as the infusionwas continued and returned to base line or slightlybelow within 10 to 15 minutes after it was termi-nated. After infusion of 4.0 to 9.0 jug epinephrineper kg (average, 6.0 jug per kg), platelet increasesin nine normal subjects (recorded in Figure 5)ranged from 25 to 709o', averaging 459o'. Onfour occasions, epinephrine was given to normalsubjects 2 to 24 hours after transfusion of 51 Cr-

labeled autogenous platelets. Increases in plate-let levels of 27 to 45% above base line were ob-served and were accompanied by proportional in-creases in circulating platelet-51Cr so that the spe-cific activity of circulating platelets remainednearly constant. Splenic surface radioactivity de-clined steadily during the period of epinephrineinfusion (Figure 6). After termination of the in-fusion, radioactivity in the peripheral blood andover the spleen returned to base-line levels in 10to 15 minutes. Figure 6 shows the effect of epi-nephrine in a typical subject. Ten days aftertransfusion, 51Cr-labeled platelets had disappearedfrom the bloodstream, but considerable radioac-tivity remained in the spleen (12). When epi-nephrine was given at this time, no change in sur-face radioactivity or circulating 51Cr resulted.

In three normal subjects, the effects of epi-nephrine on the acute disappearance of transfused51Cr-labeled platelets from the circulation were

EPINEPVIRINE

z

0

4

30

0~~~~~~~~~~~

to.

0J&n 0 0 so 30 40

MINUTES AFTER PLATELET INJECTION

FIG. 7. DISAPPEARANCEFROM THE GENERAL CIRCULA-TION OF 'CR-LABELED PLATELETS TRANSFUSEDTO A NORMALSUBJECT AND ASSOCIATED CHANGESIN SPLENIC SURFACERADIOACTIVITY WITH (0-*) ANDWITHOUT(OS-O)SIMULTANEOUS INFUSION OF EPINEPHRINE. With epi-nephrine there was a delay both in the disappearance of'Cr from the blood and in the expected rise in splenicsurface radioactivity.

649

-RICHARD H. ASTER

studied. When epinephrine infusion was begunbefore platelet transfusion, 51Cr disappearance andsplenic 51Cr uptake curves were radically altered.During epinephrine infusion only about 15% ofplatelets entered the spleen, after which a steadystate was established; 85%o of transfused plateletswere thus left in the peripheral circulation. Afterepinephrine was stopped, there were a further in-crease in splenic radioactivity and a fall in circu-lating platelet-51Cr as additional platelets enteredthe organ. Figure 7 illustrates this sequence in a

typical subject, comparing platelet disappearancecurves obtained with and without epinephrineinfusion.

Epinephrine stimulation in asplenic subjects.On 12 occasions, amounts of epinephrine that in-variably produced significant platelet increases innormal persons were given to asplenic subjects.Indications for splenectomy included hereditaryspherocytosis (two patients) and trauma (six pa-

tients). The time interval between splenectomyand adrenalin testing ranged from 3 days to 12years. All subjects but one had normal or in-creased platelet levels at the time of testing. Dur-ing a 20- to 25-minute period of epinephrine infu-sion during which at least three blood sampleswere obtained, no significant increase in plateletlevels occurred (Figure 5). In three subjects,platelets were labeled with 51Cr before epinephrinetesting. No increase in circulating 51Cr resultedfrom the epinephrine infusion. Total amounts ofepinephrine administered ranged from 3.5 to 7.7,ug per kg (average, 5.2).

Epinephrine stimulation in splenomegalic sub-jects. Epinephrine infusions were given to six

subjects with congestive splenomegaly and throm-bocytopenia (40,000 to 120,000 platelets per mm3).Total amounts of epinephrine administered rangedfrom 3.1 to 6.5 ,ug per kg (average 4.6). Becauseof the age and poor health of some of the patientsstudied, the dose was less than that used in normalindividuals. The percentage increases in plateletsabove base-line levels were greater (average two-fold) than in normal subjects (Figure 5). Sinceinitial blood platelet levels were low, however, theabsolute magnitude of the increase was equal toor less than that observed in normal persons. Inthree of the six subjects, the spleen decreased insize during epinephrine infusion. The estimatedreduction in spleen size did not correlate closelywith the platelet increase observed in the periph-eral blood.

In four of the splenomegalic patients, plateletswere labeled with 51Cr 1 day before epinephrinetesting. As in normal subjects, an increase in cir-culating 5lCr occurred that was proportional to thetotal platelet increase. Simultaneous decreases insplenic radioactivity were observed (15 to 30%below base line).

One patient with splenomegaly secondary to in-fectious mononucleosis who had a normal plateletlevel showed a lesser percentage increase in cir-culating platelets after epinephrine than did spleno-megalic patients with thrombocytopenia (Figure5).

Studies on surgically removed spleens. Plate-let levels were determined on splenic blood fromfive surgically removed spleens. This was doneon blood drained from the spleen or by making in-cisions through the splenic capsule at several sites

TABLE I

Comparison of platelet concentrations in peripheral and splenic blood

Spleen pool

%ofPlatelets total

bodySpleen Peripheral Splenic Total platelet

Patient Diagnosis Weight blood blood platelets* masst

g /mmS1 Hereditary spherocytosis 745 150,000 650,000 2.4 X 1011 262 Hereditary spherocytosis 420 121,000 880,000 1.9 X 1011 243 Hereditary spherocytosis 428 315,000 960,000 2.1 X 1011 174 Idiopathic thrombocytopenic

purpura 155 70,000 1,000,000 7.7 X 101' 315 Congestive splenomegaly 1,800 75,000 466,000 4.2 X 1011 62

* Assuming 50% of spleen weight to represent blood.t Blood volume in general circulation estimated as 7% of body weight.

650

POOLING OF PLATELETS IN THE SPLEEN

and depths and drawing blood into diluting pipettesdirectly from the incision. Table I shows that inall cases, platelet concentration was much higherin splenic blood than in peripheral blood. Theplatelet concentrations in splenic blood probablyrepresent minimal values, because some fibrin for-mation was present. In one patient with throm-bocytopenia and splenomegaly secondary to post-necrotic cirrhosis, the enlarged spleen was injectedwith heparin immediately after surgical removal.Blood was drained from the spleen and coagulationwas minimal. The spleen was then perfused with100 ml isotonic NaCl, and the perfusate was pooledwith the venous drainage; 3.5 X 101" plateletswere recovered from the organ. If transfused tothe peripheral circulation of the same patient, thisquantity of platelets would have been sufficient toincrease her platelet level from 75,000 to 180,000per mm3. Removal of red blood cells from thespleen by drainage and perfusion was about 90%effective. Scrapings taken from cut sections of theperfused organ showed, however, that many plate-lets were not removed. It is probable that thetotal number of platelets in the enlarged spleenwould, if transferred to the peripheral circulation,have increased this patient's platelet level to nor-

mal or above.

Discussion

The human spleen normally contains less than2%o of the red cell mass, although a larger propor-

tion may be pooled there in certain pathologicstates characterized by splenomegaly. The totalred cell mass present in the organ even in extremesplenomegaly rarely exceeds 20%o of the total (5,28-30). On the other hand, the total body plate-

2 Estimates of total splenic platelet content on the basisof platelet counts done on splenic pulp or blood drainedfrom the spleen are probably only rough approximations,since the volume of platelet distribution in the spleen isnot precisely known. The volume of red blood cells con-

tained in the normal spleen has been estimated at 20 to 40ml (5, 27), which would correspond to 27 to 54% bloodby weight in a 150-g spleen. In a patient with congestivesplenomegaly, we were able to recover 50% of organ

weight as whole blood despite probable loss by the organ

of some blood content as a result of tying off the splenicartery before the vein at operation. In calculating totalplatelet content of the enlarged spleens (Table I), it was

assumed that platelets were distributed through 50%o ofthe organ.

let mass is quite small (about 5 to 10 ml) so thatthe spleen could readily retain a large percentageof total platelets if factors operated to concentratethem there. The following considerations suggestthat even the normal spleen does indeed containa sizeable fraction of the total body platelet massin the form of an exchangeable pool and that suffi-cient platelets may be concentrated in enlargedspleens to cause thrombocytopenia.

In both the normal and splenomegalic subjectsthere was a reciprocal relationship between thedisappearance of 51Cr-labeled platelets from thecirculation during the first 10 minutes after trans-fusion and the increase in splenic radioactivity.In contrast, very few platelets (average 9% ofthose infused) disappeared from the general cir-culation acutely after transfusion to asplenic sub-jects (Figure 2). It was apparent from surfacescanning in normal subjects that 51Cr was concen-trated in the spleen far out of proportion to itsvolume of blood, since spleen: liver and spleen:precordium ratios averaged 5: 1 and 4: 1, re-spectively, as compared to average values of about1: 1 and 2: 3, respectively, after infusion of 51Cr-labeled red blood cells. In splenomegalic patients,the discrepancy was even more striking, withspleen: liver ratios ranging from 7: 1 to 15: 1shortly after infusion of 51Cr-labeled platelets. Inboth normal and splenomegalic subjects radioac-tivity of lung and liver relative to precordium wasconsistent with that anticipated on the basis of theblood content of these regions. No concentrationof 51Cr could be detected over any body site otherthan the spleen, nor was any 51Cr released in de-tectable form into the plasma. Estimates of totalradioactivity in the spleens of normal subjects de-rived from previous external scanning data (16),although only approximations, correlated withinlimits of error with the amount of 51Cr that left thebloodstream. As noted earlier (15), total 51Cr isunderestimated by scanning over enlarged spleensbecause a smaller segment of the organ is coveredby the orifice of the scanning device. This factwould tend to decrease surface counts per minuteas compared with the same amount of radioactivityconcentrated in a normal-sized spleen. Thus, thedisproportion between 51Cr in the spleen and liverof splenomegalic subjects after transfusion of la-beled platelets is even greater than the increasedspleen: liver ratio would imply. Taken together,

651

RICHARD H. ASTER

o90-z

,, 80\

a.-

0.

W-0

a.)t70\

60 ,0 2 4 6

MINUTES AFTER INJECTION

FIG. 8. ACUTE DISAPPEARANCEFROMTHE CIRCULATION

OF 'CR-LABELED PLATELETS TRANSFUSEDTO A NORMAL

SUBJECT (S *). Final platelet recovery at 2 hourswas 65%,. The dashed curve represents the expecteddisappearance rate if 35% of platelets were damaged inthe labeling process and destined to be removed fromthe circulation on first passage through the spleen. Inderiving the theoretical curve, it was assumed that car-diac output equals one blood volume per minute andthat splenic blood flow represents 4% of cardiac output.The percentage of platelets remaining in the circulation tminutes after injection is given by 100 (.65 +.35e0°4t).

the data strongly suggest that 51Cr-labeled plate-lets that leave the circulation acutely after transfu-sion enter the spleen.

The rate at which 51Cr-labeled platelets leavethe bloodstream after transfusion is consistent withestimates of splenic blood flow at rest.8 If all theseplatelets do enter the spleen, then during the first8 minutes or so after transfusion most labeled cellsentering the spleen are, at least temporarily, re-

tained there. This could be due either to mixing

3 Splenic blood flow in man has not been measured di-rectly, but observations by Hughes Jones and associates(17) on splenic sequestration of red blood cells sensi-tized with incomplete antibodies suggest a minimal normalvalue of 3 to 4% of cardiac output. Using a similar tech-nique, Kaplan and Jandl (31) showed that 15% of cardiacoutput was cleared of cells by the spleen in a patientwith hypersplenism. Wagner and co-workers foundthat enlarged spleens removed. damaged red blood cellsfrom the circulation three to four times as rapidly as didnormal spleens (32).

of cells with an exchangeable "pool" or to perma-nent (nonexchangeable) sequestration. After 7to 8 minutes, peripheral blood and splenic radio-activity remained relatively constant (Figure 1),implying that destruction of labeled cells stopsrather abruptly or, alternatively, that labeled plate-lets have fully mixed with unlabeled platelets inthe hypothetical pool. The data are- not consistentwith the possibility that splenic concentration of51Cr is due solely to damage of a fixed fraction(say 35%) of cells during labeling so that thesecells are destined to be sequestered when theyreach the spleen, for in such a case, disappearanceof 51Cr from the circulation would follow a simpleexponential curve dependent on the fraction ofcardiac output that comprises splenic blood flow.It may be seen from Figure 8 that the observeddisappearance of platelet 51Cr is much more rapidthan the rate expected on the basis of this type ofdestructive mechanism. For the theoretical curveto approximate the observed one even during thefitst 1 to 2 minutes, normal splenic blood flowmust be assumed equal to 15%o of cardiac output, afigure at least three times greater than even themost liberal estimates. If platelet injury is re-sponsible for splenic 51Cr uptake after transfusion,the rapidity with which labeled cells enter thespleen makes it necessary then to postulate a uni-form injury to all the platelets, which is reversibleafter the cells have spent 7 to 8 minutes in therecipient's circulation. That such is not the caseis suggested by the observation that when entry oftransfused cells into the spleen is delayed 10 to 15minutes by infusion of epinephrine, final plateletrecovery is not increased (four observations), de-spite the fact that transfused platelets circulatefreely from the time of their injection (Figure 7).Moreover, infusion of epinephrine results in re-lease of at least part of the splenic 51Cr into thegeneral circulation (Figure 6), even when given1 day after the original transfusion. Since plate-let levels did not increase in asplenic persons givenepinephrine, it seems probable that virtually all theincrease in circulating platelets after adrenergic-stimulation in normal subjects is derived from thespleen. Less than a 50%o reduction of splenic sur-

face radioactivity could be achieved with epineph-rine, but some of the initial surface activity was.derived from underlying structures such as liver..

652

POOLING OF PLATELETS IN THE SPLEEN

As splenic radioactivity diminished and radioac-tivity in other regions such as the liver increased,the contribution from extrasplenic sites becamerelatively greater. Thus the data are consistentwith the possibility that more than 50% of splenicradioactivity was mobilized, albeit a precise esti-mation is not possible.

These observations are subject to various in-terpretations. For example, it could be postulatedthat platelets are mobilized from the lung (33),liver, or the marginal vascular pool (34) afterepinephrine infusion. It would then be necessary,however, to explain why the platelet levels ofasplenic subjects do not change during adrenergicstimulation. Moreover, the consistent decrease inradioactivity over the spleen after epinephrine andthe small but definite increase over the liver andlungs are inconsistent with these possibilities. Asplenic platelet pool exchanging with the generalcirculation and sensitive to epinephrine can, on theother hand, readily explain the observed data.The finding that asplenic subjects do not increaseplatelet levels in response to epinephrine confirmsthe recent report of Griffoni, Scaltrini, Confalo-nieri, and Conigliaro (35), but contrasts with theearlier findings of Chatterjea, Dameshek, andStefanini (36). Possibly, the indirect countingtechnique used by the latter workers accounts forthe discrepancy. Others have found that indirectplatelet counts may show apparent fluctuations inplatelet levels not confirmed by direct counting ofcells (37).

Disappearance curves of platelet 51Cr from thecirculation (Figure 1) suggest that about 8 min-utes is required for a steady state to be establishedin the normal spleen. In contrast, labeled redblood cells establish equilibrium in the spleen inless than 2 minutes (38, 39). This suggests thatplatelets may enter a compartment of the spleen,pass through the compartment serially, and thenenter the splenic vein. If this is so, then the sizeof the pool would be proportional to the productof splenic blood flow (approximately 4%o of totalblood volume per minute) and platelet transit timethrough the spleen. An average value of 8 min-utes for transit time would produce a "pool" con-stituting about 32%o of the total platelet mass.This figure is consistent with the time required forsplenic radioactivity to stabilize after injection oflabeled cells and with the difference in platelet re-

covery observed between asplenic and normal sub-jects (Figures 1 and 2). It is of interest thatBjorkman has shown that particles less than 3 u indiameter enter the pulp cords rather than sinuseswhen injected into the splenic artery, whereas thereverse is true of larger particles (40). If plate-lets (< 2 ju in diameter) are required to traversethe more complex system of splenic cords, theirtransit time through the organ might well be pro-longed relative to red blood cells, leukocytes, andplasma. Such a process could be compared on amacro scale to the "molecular sieve" type of col-umn utilized for molecular separations in analyti-cal chemistry.

Doubtless the model suggested by the data is anoversimplification, since other types of pools couldbe made to "fit" the observed curves. Moreover,it is probable that some of the 51Cr that accumu-lates in the spleen after transfusion is a result ofpermanent sequestration of cells damaged in vitroor that the entire pool is not mobilized by epineph-rine or both. For these reasons the exact size ofthe splenic platelet pool cannot be estimated withprecision from the available data. It seems highlyprobable, however, that such an exchangeable poolexists and that it constitutes between 20 and 40%oof the total platelet mass. Direct determination ofplatelet numbers in surgically removed normalspleens may be helpful in determining the spleen'splatelet content more precisely. Such spleens arenot in a normal state, however, and it is not pos-sible to be certain from such data that the plateletsencountered are exchangeable with those in thebloodstream.

Since human platelets are present in excess ofthe amount required for hemostasis, it is difficultto assign an important physiologic role to thesplenic platelet pool. Certain implications aresuggested, however. For example, estimates ofhuman platelet production based on platelet life-span and circulating platelet levels should prob-ably be increased. At least part of the thrombo-cytosis that follows splenectomy may be attributedto loss of the splenic pooling function. In experi-mental and clinical studies involving transfusionof fresh or preserved platelets the calculation ofrecovery should take into consideration obligatorysplenic platelet pooling. Thus, an apparent re-covery in venous blood of 65 to 75% of transfusedcells may indicate nearly 100%o viability. In ear-

653

RICHARD H. ASTER

Her studies from this laboratory it was presumedthat the high uptake of 51Cr by the spleen aftertransfusion of labeled platelets represented per-manent sequestration by this organ of cells injuredin the labeling process (12). During the 8- to 10-day period during which 51Cr-labeled platelets dis-appeared from the blood stream, no further rise in51Cr in the spleen was observed. Since eliminationof 51Cr from the spleen after permanent sequestra-tion of platelets was rather slow (about 3% perday), it appeared that relatively few platelets weredestroyed in the organ under normal conditions.With splenic pooling recognized as the primarycause of the initial splenic surface radioactivity, itappears likely that a number of effete plateletsat least equivalent to the pool size are eventuallyremoved from the circulation by the spleen anddestroyed therein.

Splenic pooling as a possible cause of hyper-splenic thrombocytopenia. If the mechanism forsplenic platelet pooling postulated above is cor-rect, it is to be expected that factors which increasesplenic blood flow or prolong platelet transit timethrough the spleen will increase the pool size. Insplenomegalic patients platelet recoveries in theperipheral blood were markedly reduced (Figure3). For reasons stated above, apparently all ornearly all of the 51Cr leaving the bloodstreamacutely after transfusion accumulated in the spleensof these individuals. It is unlikely that any defectintrinsic to platelets themselves accounted for thereduced recoveries, for similar results were ob-tained with homologous and autogenous cells.Moreover, platelets from one splenomegalic sub-ject gave 15% recovery in the donor but normalrecovery in a normal recipient. It is also improb-

TABLE II

Maximal platelet mass calculated from dilution of injected5ICr-labeled platelets

Maximal platelet mass*No. of

Diagnosis subjects Average Range

platelets/kg body wtNormal 30 2.97 X101' 1.80- 5.20Congestive splenomegaly 11 2.75 Xl O° 2.17- 6.12Chronic lymphatic leukemia

with splenomegaly 2 2.89 X101 1.92- 3.86Infectious mononucleosis 1 6.10 X1010Polycythemia vera with

splenomegaly 2 24.0 X1010 20.8-27.2

* Maximal platelet mass = [platelets per milliliter venous blood/per cent recovery of transfused patients (venous blood)]X(70 ml/kilograms body weight).

able that a so-called "platelet debt" contributedsignificantly to decreased recovery, for normalplatelet recoveries are often observed in patientswith much more severe degrees of thrombocyto-penia (19, 41). Furthermore, transfusion of 4.5X 101 homologous platelets (equivalent to 5 U ofwhole blood) to one of our splenomegalic recipi-ents resulted in about the same recovery (18%) asdid a much smaller number of autogenous cells(1.5 x 1010) (20%o recovery). As with normalindividuals, then, it appears necessary to choosebetween permanent destruction in the spleen ofcells damaged in vitro and spleen pooling or a com-bination of these two factors to explain the verylow platelet recoveries observed in splenomegalicsubjects. If cell injury is responsible, it is likelyfor reasons stated above to be of a reversible na-ture, since peripheral blood 51Cr was relativelystable after about 7 to 8 minutes as with normalpersons (Figure 1). In terms of total plateletsmobilized, splenomegalic patients responded nobetter and usually less well to epinephrine stimula-tion than did normal subjects, although the per-centage platelet increase was greater (Figure 5).If one assumes that splenic pool size is representedby platelets mobilized by epinephrine and that thefraction of platelets "released" by a given dose ofepinephrine is unrelated to spleen size, then thelow recovery of platelets transfused to spleno-megalic subjects cannot be accounted for entirelyby pooling, and the initial high splenic 51Cr uptakemay be due in part to in vitro cell injury. Re-versible platelet injury is, however, conjectural,and there is no information as to the relative ef-fectiveness of epinephrine in pathological spleens.Indeed, there is no reason to think that infusionof epinephrine for 15 to 20 minutes necessarilymobilizes all platelets from an enlarged spleen,and the doses of epinephrine given were less thanthose administered to normal subjects. Eventhough shrinkage of several spleens was observed,the compartment containing platelets, which canoccupy a total packed volume of only 5 or 10 ml,need not necessarily have been affected in propor-tion to the gross reduction in spleen size, whichpresumably reflects mainly a shift of the red cellpool.

The data in Figure 3 relating circulating plate-let levels to the fraction of cells that remains inthe general circulation after transfusion can be

654

POOLINGOF PLATELETS IN THE SPLEEN

used to calculate total body platelet mass by iso-tope dilution. The values obtained are maximalones based on the assumptions that transfusedcells mix completely with the total circulatingplatelet mass and that none are permanently se-

questered in the spleen as a consequence of injurydue to in vitro manipulation alone. Table II com-

pares the platelet mass of splenomegalic patientscalculated in this way with that of normal subjects.The average platelet mass of all splenomegalicsubjects excluding two patients with polycythemiavera was about the same as for 30 normal subjects.The range of values in the two groups was alsocomparable.

Although the sources of error referred to aboveleave some uncertainty as to the exact fraction oftotal body platelets that is present in the spleens ofsplenomegalic patients, it seems evident that theenlarged spleens do concentrate platelets to a de-gree disproportionate to the amount of blood theycontain. Average platelet life-span in the 16splenomegalic patients studied was nearly normal(Figure 4), in confirmation of studies by Cohen,Gardner, and Barnett (14). If the total plateletmass is approximately normal (Tables I and II),it follows that platelet production, which is pro-

portional to the ratio platelet mass: platelet life-span, is also normal in these individuals, as mightbe expected from the number of megakaryocytes intheir marrow. It has been suggested that in hy-persplenic states the margins of the megakaryo-cytes appear smooth and inactive (9), but there isnot uniform agreement on this point. Moreover,it has been shown experimentally that similaralterations in megakaryocyte morphology followacute platelet depletion despite a normal rate ofplatelet production (42-43). The "nonproduc-tive" appearance of the megakaryocyte has beenmost strongly emphasized in idiopathic thrombo-cytopenic purpura (44), a disease now known tobe characterized in almost all instances by normalor even increased platelet production rates (45).Apparently the smooth appearance of megakaryo-cytes is, in most instances, a physiological reflec-tion of immaturity of the cell, rather than of a

pathological failure to function properly.That thrombocytopenia in the peripheral blood

of patients with hypersplenism is chiefly a conse-

quence of splenic platelet pooling was suggestedearlier by Reimann, Erdogan, and Ulagay (46),

who found platelet concentrations to be higher inblood from enlarged spleens than in the generalcirculation. Wright, Doan, Bouroncle, and Zol-linger (11) and Chatterjea and co-workers (36)showed that injection of epinephrine directly intothe splenic artery of patients with congestivesplenomegaly results in greatly increased plateletlevels in the splenic vein, demonstrating concentra-tion of platelets in the enlarged spleens. Penny,Rozenberg, and Firkin, using a perfusion tech-nique, recovered 11 x 101" platelets that appearednormal from a patient with thrombocytopenia andsplenomegaly secondary to cirrhosis (47). Thir-teen other pathologic spleens perfused by theseworkers all contained platelets far in excess of theamount to be expected on the basis of splenic bloodvolume alone. This was true whether or not thesubjects had thrombocytopenia. Our experiencewith a limited number of surgical specimens wassimilar (Table I). On the basis of a limited num-ber of observations (Figure 3) it would appearthat pooling is as pronounced in spleens enlargedby proliferative disease as in spleens enlarged asa result of congestion in the splanchnic vascularbed. More studies of patients with proliferativesplenomegaly will be required, however, particu-larly in subjects with hereditary spherocytosis, adisease which is rarely complicated by thrombocy-topenia and in which selective red cell trapping isa primary event. If the spleens of these subjectspool platelets to a significant degree, as suggestedby Penny and associates (47), then platelet pro-duction must be increased to maintain normalplatelet levels. In our two patients with polycy-themia vera, the pathological increase in plateletproduction known to be characteristic of this dis-ease probably accounted for normal platelet levelsdespite the presence of a very large splenic plateletpool.

In conclusion, the evidence indicates that poolingof platelets in an enlarged spleen accounts for thethrombocytopenia of the hypersplenic state. Aslightly increased rate of platelet destruction playsan additive, but relatively minor, role. Until moreprecise estimates can be made of the pool size andits exchangeability with the general circulation,however, it is not possible to state with certaintythat these two mechanisms are totally responsiblefor reduced platelet levels in splenomegalic pa-tients.

655

RICHARD H. ASTER

Even though platelet production appears to benormal in most individuals with hypersplenism,the possibility that enlarged spleens exert humoralcontrol over the marrow is not altogether excluded.Indeed, the fact that most patients with congestivesplenomegaly appear unable to increase plateletproduction so as to compensate for platelet poolingcould be attributed to an inhibitory effect. On theother hand, the failure of these subjects to com-pensate may be a consequence of the limited ca-pacity of normal marrow to increase platelet pro-duction in response to platelet depletion (42-43).Alternatively, the apparent "failure" of marrow torespond to the low blood platelet levels that resultfrom pooling raises the prospect that the marrowreacts to a product of net platelet destruction ratherthan to absolute platelet levels in the peripheralblood.

Summary

The distribution of human platelets in the bodywas studied with 51Cr-labeled platelets, externalscintillation scanning, epinephrine stimulation, andplatelet counts on the blood of surgically removedspleens.

Evidence is presented which indicates that nor-mally about one-third of the total platelet mass isconcentrated in the spleen, exchanging with theremaining two-thirds of platelets, which are evenlydistributed throughout the rest of the vascularsystem. Pulmonary or "marginal" pools do notappear to contain a significant fraction of totalplatelets in man.

In splenomegaly, the splenic platelet pool may begreatly increased, so that 50 to 90%o of plateletsare in the spleen at any given time. This re-distribution of cells from the peripheral circula-tion to the spleen appears sufficient to producethrombocytopenia despite unimpaired platelet pro-duction, normal total body platelet mass, andnearly normal platelet life-span.

The protracted thrombocytosis that usually fol-lows splenectomy may be due, in large part, to re-moval of the splenic platelet pool.

The present studies are compatible with thepossibility that platelet production is governed bythe rate of platelet destruction rather than by theplatelet concentration of the blood.

AcknowledgmentsI am indebted to Dr. James Jandl for reviewing the

manuscript and for many helpful suggestions made dur-ing the course of this study, and to Dr. William Mc-Dermott for providing a surgically removed spleen andseveral patients for investigation. The technical as-sistance of Miss Nancy Mann, Miss Barbara Mink, andMiss Hollister Nash is gratefully acknowledged.

Addendum

Since this manuscript was prepared, Penny, Rozenberg,and Firkin have determined the platelet content of 14surgically removed spleens by a direct perfusion tech-nique (Blood 1966, 17, 1). Estimates of splenic poolsize obtained by this method and the data here presentedon the magnitude and dynamics of the pool appear tocorrelate closely.

References

1. Aster, R. H. Effect of anticoagulant, ABO incom-patibility and spleen size on recovery and survivalof transfused human platelets (abstract). Clin.Res. 1964, 12, 219.

2. Aster, R. H. Studies on the pathophysiology ofthrombocytopenic disorders using external scintil-lation scanning (abstract). Xth Congress of theInternational Society of Hematology. Stockholm,Ljungl6fs Lithografiska AB, 1964, p. N7.

3. Jandl, J. H. The anemia of liver disease: observa-tions on its mechanism. J. clin. Invest. 1955, 34,390.

4. Giblett, E. R., A. G. Motulsky, F. Casserd, B.Houghton, and C. A. Finch. Studies on the patho-genesis of splenic anemia. Blood 1956, 11, 1118.

5. Motulsky, A. G., F. Casserd, E. R. Giblett, G. 0.Broun, Jr., and C. A. Finch. Anemia and thespleen. New Engl. J. Med. 1958, 259, 1164, 1215.

6. Harris, I. M., J. McAlister, and T. A. J. Prankerd.Splenomegaly and the circulating red cell. Brit.J. Haemat. 1958, 4, 97.

7. Frank, L. Die hamorrhagische Diathese in Schitten-helm: Handbuch des Blutes und der blutildendenOrgane. Berlin, Springer, 1925, p. 432.

8. Heilmeyer, L. Rapports physiologiques entre la rateet al moelle osseuse. Rev. Hemat. 1954, 9, 267.

9. Dameshek, W. Hypersplenism. Bull. N. Y. Acad.Med. 1955, 31, 113.

10. Doan, C. A. Hypersplenism. Bull. N. Y. Acad. Med.1949, 25, 625.

11. Wright, C.-S., C. A. Doan, B. A. Bouroncle, andR. M. Zollinger. Direct splenic arterial and ve-nous blood studies in the hypersplenic syndromesbefore and after epinephrine. Blood 1951, 6, 195.

12. Aster, R. H., and J. H. Jandl. Platelet sequestrationin man. I. Methods. J. clin. Invest. 1964, 43, 843.

13. Aster, R. H. Effect of anticoagulant and ABO in-compatibility on recovery of transfused human plate-lets. Blood 1965, 26, 732.

656

POOLINGOF PLATELETS IN THE SPLEEN

14. Cohen, P., F. H. Gardner, and G. 0. Barnett. Re-classification of the thrombocytopenias by the Crl-labeling method for measuring platelet lifespan.New Engl. J. Med. 1961, 264, 1294.

15. Jandl, J. H., M. S. Greenberg, R. H. Yonemoto, andW. B. Castle. Clinical determination of the sitesof red cell sequestration in hemolytic anemias. J.clin. Invest. 1956, 35, 842.

16. Aster, R. H., and J. H. Jandl. Platelet sequestrationin man. II. Immunological and clinical studies.J. clin. Invest. 1964, 43, 856.

17. Hughes Jones, N. C., P. L. Mollison, and N. Veall.Removal of incompatible red cells by the spleen.Brit. J. Haemat. 1957, 3, 125.

18. Heidelberg, J. G., P. G. Sirota, W. C. Dewey, andR. G. Rose. Determination of the distribution ofp' in phantoms and in patients by detection ofBremsstrahlung radiation with a scintiscanner.Amer. J. Roentgenol. 1963, 90, 325.

19. Aster, R. H. Unpublished observations.20. Wolf, R., and J. Fischer. Untersuchungen am Phan-

tom mit 'Cr zur Frage der Beurteilung der Ergeb-nisse von Organaktivititsmessungen. Strahlen-therapie 1962, 119, 226.

21. Brecher, G., and E. P. Cronkite. Morphology andenumeration of human blood platelets. J. appl.Physiol. 1950, 3, 365.

22. Frey, W. Zur Frage der funktionellen Milzdiagnostikmittels Adrenalin. Z. ges exp. Med. 1914, 3, 416.

23. Krumbhaar, E. B. Functions of the spleen. Physiol.Rev. 1926, 6, 160.

24. Benhamou, E. L'exploration functionelle de la rite:diagnostic et traitment des syndromes splenique.Paris, Masson, 1933, p. 33.

25. Ebert, R. V., and E. A. Stead, Jr. Demonstrationthat in normal man no reserves of blood are mo-bilized by exercise, epinephrine or hemorrhage.Amer. J. med. Sci. 1941, 201, 655.

26. Parson, W., H. S. Mayerson, C. Lyons, B. Porter,and W. V. Trautman, Jr. Effect of the administra-tion of adrenalin on the circulating red cell vol-ume. Amer. J. Physiol. 1948, 155, 239.

27. Crosby, W. H. Normal functions of the spleen rela-tive to red blood cells: a review. Blood 1959, 14,399.

28. Rothschild, M. A., A Bauman, R. S. Yalow, and S. A.Berson. Effect of splenomegaly on blood volume.J. appl. Physiol. 1954, 6, 701.

29. Fudenberg, H., M. Baldini, J. P. Mahoney, and W.Dameshek. The body hematocrit/venous hemato-crit ratio and the "splenic reservoir." Blood 1961,17, 71.

30. Tizianello, A., and I. Pannacciulli. The effect ofsplenomegaly on dilution curves of tagged erythro-cytes and red cell volume. Comparative studies onnormal anaemic and splenomegalic patients. Actahaemat. (Basel) 1959, 21, 346.

31. Kaplan, M. E., and J. H. Jandl. The effect of rheu-matoid factors and of antiglobulins on immune he-molysis in vivo. J. exp. Med. 1963, 117, 105.

32. Wagner, H. N., Jr., M. A. Razzak, R. A. Gaertner,W. P. Caine, Jr., and 0. T. Feagin. Removal oferythrocytes from the circulation. Arch. intern.Med. 1962, 110, 90.

33. Bierman, H. R., K. H. Kelly, F. L. Cordes, R. L.Byron, Jr., J. A. Polhemus, and S. Rappoport.The release of leukocytes and platelets from thepulmonary circulation by epinephrine. Blood 1952,7, 683.

34. Vejlens, G. The distribution of leukocytes in thevascular system. Acta path. microbiol. scand. 1938(suppl. 33).

35. Griffoni, V., G. C. Scaltrini, C. Confalonieri, and S.Conigliaro. La prova dell adrenalina in ematologia.Haematologica 1959, 44, 393.

36. Chatterjea, J. B., W. Dameshek, and M. Stefanini.The adrenalin (epinephrine) test as applied tohematologic disorders. Blood 1953, 8, 211.

37. Wilson, S. J., G. Eisemann, and J. H. Chance. Theeffect of the "thrombocytopenic factor" of idio-pathic thrombocytopenic purpura on platelet levelsas measured by direct and indirect methods. J.Lab. clin. Med. 1952, 40, 498.

38. Harris, I. M., J. M. McAlister, and T. A. J. Prankerd.The relationship of abnormal red cells to the nor-mal spleen. Clin. Sci. 1957, 16, 223.

39. Toghill, P. J., and B. N. C. Prichard. A study of theaction of noradrenaline on the splenic red cellpool. Clin. Sci. 1964, 26, 203.

40. Bj6rkman, S. F. The splenic circulation. Withspecial reference to the function of the spleen sinuswall. Acta med. scand. 1947 (suppl.), 191.

41. Levin, R. H., J. H. Pert, and E. J. Freireich. Re-sponse to transfusion of platelets pooled from mul-tiple donors and the effects of various technics ofconcentrating platelets. Transfusion (Philad.)1965, 5, 54.

42. Craddock, C. G., Jr., W. S. Adams, S. Perry, andJ. S. Lawrence. The dynamics of platelet pro-duction as studied by a depletion technique innormal and irradiated dogs. J. Lab. clin. Med.1955, 45, 906.

43. Matter, M., J. R. Hartmann, J. Kautz, Q. B. De-Marsh, and C. A. Finch. A study of thrombo-poiesis in induced acute thrombocytopenia. Blood1960, 15, 174.

44. Dameshek, W., and E. B. Miller. The megakaryo-cytes in idiopathic thrombocytopenic purpura, aform of hypersplenism. Blood 1946, 1, 27.

45. Najean, Y., N. Ardaillou, J. Caen, M.-J. Larrieu, andJ. Bernard. Survival of radiochromium-labeledplatelets in thrombocytopenias. Blood 1963, 22,718.

46. Reimann, F., G. Erdogan, and I. Ulagay. Untersuch-ungen uiber die Genese des Syndromes der "Hyper-splenic" bie portalem Hochdruck. Acta hepato-splenol. (Stuttg.) 1960, 7, 230, 285.

47. Penny, R., M. C. Rozenberg, and B. G. Firkin. Thesplenic platelet pool. Medical Res. 1964, 1, 102.

657