erythropoiesis with particular reference its study … · j. clin. path. .(1949), 2, 1....

J. clin. Path. .(1949), 2, 1.

ERYTHROPOIESIS WITH PARTICULAR REFERENCEITS STUDY BY BIOPSY OF HUMAN

BONE MARROW: A REVIEWBY

J. V. DACIE AND J. C. WHITEFrom the Department of Pathology, Postgraduate Medical School of London

- (RECEIVED FOR PUBLICATION, JANUARY 5, 1949)

PageMETODS OF BONE MARROW BioPsY .. .. 1

Sternal puncture .. .. .. .. .. 2Other sites for needle biopsy; the tibia in child-

hood, the iliac crest, and vertebral spines .. 2EXAMNATION OF ASPIRATED BONE MARROW .. 3Preparation and staining of bone marrow films 3Concentration of bone marrow by centrifugation 3Quantitative cell counts on aspirated bonemarrow .. .. .. .. .. .. 3

Preparation of histological sections of aspiratedbone marrow .. .. .. .. .. 4

ExrENT OF NORMAL MARROW AT DIFFERENT AGES 4The site of development in the bone marrow .. 5

MORPHOLOGY OF NORMAL ERYTRHROPOIESIS .. 5Nomenclature of the erythrocyte precursors .. 5Synonyms for the pronormoblast and normoblast 5Formation in intra-uterine life .. .. 6Normoblastic erythropoiesis in the adult .. 6Differential cell counts on marrow films .. 8

GROWTH AND DIFFjRENiATION OF ERYTHROBLASTS 10Mitosis during normal erythropoiesis .. .. 11Life span of the erythroblasts .. .. .. 11Derivation of reticulocytes from normoblasts .. 13Loss of nucleus of normoblasts .. .. .. 14Regulation of erythropoiesis .. .. .. 14

PageCYTOCHEMISTRY OF ERYTHROCYrE DEvELOPMENT.. 15Formation of haemoglobin; siderocytes .. 21

ABNORMAL ERYTHROPOIESIS.. .. .. .. 22Megaloblastic erythropoiesis .. .. 22Morphology of the megaloblasts .. 22Intermediate megaloblasts .. .. 22The nature ofmegaloblastic erythropoiesis .. 23The variability ofmegaloblastic bone marrow.. 24The effect on megaloblastic bone marrow of

treatment with tile liver anti-anaemic prin-ciple .. .. .. .. .. 24

The megaloblast "problem" .. .. 24

OTHER TYPES OF ABNORMAL ERYTHROPOIESIS .. 25Macronormoblastic erythropoiesis .. 25Erythropoiesis in iron deficiency .. 26Erythropoiesis in refractory anaemia .. 27Erythropoiesis in haemolytic anaemia.. -.. 28Erythropoiesis in leukaemia and allied disorders 28Erythropoiesis in carcinomatosis of the bone

marrow.. .. .. .. .. .. 29Cellular gigantism in human erythropoiesis .. 29Abnormalities of mature erythrocytes .. 30

CONCLUSIONS.. .. .. .. .. .. 30

In this review we are attempting to present apicture of the morphology, physiology, and patho-logy of erythropoiesis in man. The ease with whichthe blood cells and their precursors may be examinedprovides an almost unique opportunity for theirstudy and greatly adds to the charm and interest ofhaematology; and the results obtained thereby areprobably 4ignificant for the growth ofhuman tissuesas a whole. Despite great progress in recent years,however, many gaps in knowledge and anomaliesstill exist, particularly in respect of the factors

A

underlying the growth and differentiation of bloodcells and the effects ofpathological states upon them.Because ofthe very great importance of technique,

we are prefacing the main discussion with-sectionson the methods of bone marrow biopsy.

Methods of Bone Marrow BiopsyThe haemopoietic function of the red marrow has

been realized since the description by Neumann in1868 of the origin therein of the erythrocytes. Theearly studies on human material were carried out

on March 30, 2020 by guest. P

rotected by copyright.http://jcp.bm

j.com/

J Clin P

athol: first published as 10.1136/jcp.2.1.1 on 1 February 1949. D

ownloaded from

http://jcp.bmj.com/

J. V. DACIE AND J. C.

on post-mortem issue, however, and it is onlycomparatively recently that biopsy of marrow hasbeen widely employed and has thus renderedpossible the examination of fresh material instained smears. A big step forward in the techniqueof examination was made by Seyfarth (1923), whointroduced a method of trephining the sternum bymeans of a small 3 mm. straight trephine. Thepreparation of histological sections from suchmaterial was admirably described by Custer (1933).However, although trephine biopsy provides excel-lent material for sections and smears, it has dis-advantages; it is a minor surgical operation, whichcan be safely carried out only under aseptic condi-tions in the operafing theatre, and which may bedangerous in the presence of a bleeding tendency orof granulocytopenia. In addition, the skin incisiontakes several days to heal, and the trephining canseldom be repeated. The introduction by Arinkin(1929) of the method of needle puncture biopsy ofthe sternum has, therefore, given a great impetusto the study of the marrow tissues during life. Therelative values of the trephine technique and ofneedle biopsy are discussed by Dameshek and others(1937) and by Osgood and Seaman (1944). Stainedsmears ofmaterial aspirated after death are generallyunsatisfactory, for the degenerating cells are easilybroken up by the process of spreading (Rohr andHafter, 1937; Leitner (Leitner and others, 1949a)).

Sternal punctures-Arinkin's method of ster-nal puncture provides material which can notonly be spread on slides, dried in the air, andstained like blood films, but which may also beexamined supravitally by various optical techniques,and in addition provides marrow particles whichmay be sectioned or subjected to chemical micro-analysis. The results obtained by Arinkin's tech-nique in health and in many blood diseases haveappeared in numerous papers and several mono-graphs (for example, Segerdahl, 1935; BodleyScott, 1939; Leitner, 1941; Rastelli, 1943).A variety of needles exists. Some sort of adjustable

guard is essential, and the needle designed by Salah(1934) is widely used in this country. In addition, needle-trephines have been designed; that of Turkel andBethel (1943) is useful, yielding a small piece of corticalbone and underlying marrow. The trephine is passedthrough a hollow needle, and no skin incision isrequired .

In careful hands and using an aseptic techniquesternal puncture is a safe procedure in the greatmajority of patients and may often be performed onout-patients. However, fatalities have resulted,mostly through complete trnsfixion of the stemum(Lancet, April 10, 1948), but this should not occur

if a guarded needle is employed. It should not belightly undertaken in the presence of a bleedingtendency. Although, as a rule, little haemorrhageresults in patients with a prolonged bleeding time,it should never be performed in haemophilia or inpatients with similar coagulation defects.The usual site for sternal puncture is the manu-

brium sterni or the first or second piece of the bodyof the bone; in our experience rather more cellularmarrow with a smaller admixture of blood isobtained from the manubrium, although in this siteit is rather difficult to be sure that the needle hasentered the medullary cavity. The sternum may berepeatedly punctured; a different site should beselected for each puncture in order to avoid marrowpossibly disorganized by haemorrhage resultingfrom previous punctures.

Other sites for needle biopsy; the tibia in childhood,the iliac crest, and vertebral spines.-In the youngestchildren, from birth to four or five years, the medialaspect of the head of the tibia may be punctured andactive marrow withdrawn. This procedure is certainlyless dangerous and less upsetting to the child than issternal puncture. In older children the tibial corticalbone is usually too dense and the marrow within normallyless active. Better samples are obtained by sternalpuncture (Kato, 1937).

Recently the iliac crest (Rubinstein, 1948) and spinousprocesses of the vertebrae have been punctured and foundin adults to yield good samples of marrow. Puncture ofthe spinous processes of the vertebrae is particularlyconvenient, as they lie superficially and the overlyingtissues are not highly sensitive. The needle is passedvertically into the spines of the lumbar or lower thoracicvertebrae of the sitting or reclining patient (Loge, 1948).Rather more pressure is required than for sternal punc-ture. An added advantage of spinous process punctureis that the patient cannot see what istappening and thatseveral attempts at puncture and aspiration can be madein the same anaesthetized area if necessary.

It is an advantage to have a choice of several sites,both in the same region, for example the sternum andvertebral spines, and between these areas, as repeatedly" dry " or non-cellular punctures are sometimes encoun-tered.* In these cases, selection of a different site mayeither yield a cellular marrow or strengthen suspicionof a widespread pathological change affecting themarrow. The questions of different sites and of variousforms of puncture needles are discussed at length byRastelli (1943a).

* Repeated failure to withdraw marrow is disconcerting. In ourexperience " dry " punctures or aspirates containing only blood orvery few marrow cells lave been due, in cases where we have beenconfident that the needle has penetrated the cortical bone, to marrowaplasia or hypoplasia, to marrow fibrosis, to secondary carcinoma-tosis, and sometimes to leukaemia, particularly of primitive cell types(see also page 4). Dameshek and others (1937) and Da ek(1948) have had similar experiences. In these cases a surgical trepbmeoperation may be required before the diagnosis can be made withcertainty.

-2



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

Although there is normally a considerable variationin the composition of cellular marrow withdrawn fromadjacent or different sites, the general trend and typeand maturity of haemopoiesis and the balance betweenerythropoietic and leucopoietic activity is similar(Pizzolato and Stasney, 1947). This essentially similaryet variable pattern -is well shown when sections ofaspirated material are studied (J. N. Davidson andothers, 1948).

The Examination of Aspirated Bone MarrowThe preparation and staining--of bone marrow

films-The volume of marrow which may beaspirated by puncture is limited, and the morematerial aspirated the greater the proportion ofcontaminating blood. Osgood and Seaman (1944)recommend that 1 ml. should be withdrawn as aroutine, but we prefer not to take more than 0.2 ml.

Most, if not all, workers employ a Romanowskydye for the routine staining of bone marrow films;many use a panoptic method, at a controlled pH(about 6.8) using May-Grunwald-Giemsa or Jenner-Giemsa stains. Both these techniques give excellentresults, but in our experience a relatively longfixation with methanol (3-5 min.) is essential(Turnbull, 1948).*A well spread and well stained cellular film of

marrow can be a delightful object, but the methodof preparation of the film is as important as thestaining technique. It is desirable to concentratethe marrow cells at the expense of the blood bywhich they are inevitably diluted, and variousmethods of concentration have been suggested.Davidson (1941) for instance expresses the contentsof the aspiration syringe into a watch glass, picksout with forceps the fragments of marrow whichadhere to the glass surface, and spreads themindividually on slides. This technique providesmarrow fragments with little contaminating bloodand gives a better idea of the cellularity of the tissuethan can be obtained by merely spreading drops ofthe aspirated material directly on to slides. Otherauthors (Limarzi, 1947) use more complicatedmethods (vide infra). We ourselves employ a verysimple concentrating device which is generallysatisfactory. A series of drops of syringe coptentsis delivered on to slides and most of the blood isthen sucked off with a fine Pasteur pipette appliedto one edge of the drop; the irregularly shaped

* Although Romanowsky staining is undoubtedly the best singlestaining method, there are simple supplementary techniques of value.By wet-fixing films in Susa fluid, pyronin-methyl green may be usedto demonstrate nucleoli and cytoplasmic basophitia (White, 1947).Similarly Feulgen's staining reaction gives a good picture of thechromatin and chromosomes; light green yellowish S is a particularlysuitable counter-stain. Phase-contrast techniques on unstained ceUsmounted in fonnalin have been used by Jones (1948) in the study ofmitochondria.

marrow fragments tend to adhere to the slide andare left behind. A film is then made of the marrowfragments and the remaining blood by means of asmooth-edged glass spreader of not more than 2 cm.in width; the marrow fragments are draggedbehind the spreader and leave a trail of cells behindthem. It is on these trails that differential countsshould be done, commencing from the marrowfragment and working back towards the head of thefilm; in this way smaller numbers of cells from theperipheral blood become incorporated in thedifferential count.

Concentration of marrow by centrifugation.-Severafworkers have used centrifugation techniques in an attemptto assess the relative distribution of marrow cells, peri-pheral blood, and fat in aspirated material. Limarzi(1947) has devised a method for centrifugingheparinized aspirated marrow in a Wintrobe haematocrittube. Yellow fat, red fat, plasma, marrow cells, andmature erythrocytes separate from above downwards-The marrow cell layer is used for preparing films andfor differential cell counts. Limarzi (1947) claims thathis method gives a relatively accurate indication ofmarrow hypo- or hyperplasia and that the amount offat present can be gauged. He also claims that theseparation of fat from the marrow cells facilitates theirproper staining and that numbers of uniform prepara-tions can be obtained from the sample. Berman andAxelrod (1947) have described a comprehensive methodfor treating aspirated marrow so as to obtain volumetricreadings of the constituent fractions, as well as smears,imprints, and histological sections.The data obtained by concentration techniques

employing centrifugation are useful in group studies, butare less valuable in individual cases on account of thewide range of values encountered even in the normal.

Quantitative cell counts on aspirated marrow.-Anumber of figures for the cell content of aspirated normalmarrow have been given in the literature (Osgood andSeaman, 1944; Vaughan and Brockmyre, 1947). Thevariation is extremely wide; this is hardly surprising inview of the uncontrollable factor of dilution with peri-pheral blood and the tendency of the marrow cells toadhere together in clumps of varying size. Perhaps thebest method of enumeration is that of Isaacs (1937a), inwhich the particles of aspirated marrow are mixed withand broken up in normal serum. He attributed thedispersing action of serum to enzyme activity (Isaacs,1930).Tentative normal standards are given by Osgood and

Seaman (1944). Fieschi, quoted by Rastelli (1943b), withwhom we largely agree, states that no information is tobe obtained from cell counts on the marrow that cannotbe derived from assessment of simultaneously preparedfilms and sections. Clearly, the value of the quantitativemethod depends upon details of technique and is ofgreatest use in serial studies (see Stasney and Pizzolato,.1942).

3



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. DACIdE AND ). C.. WHITE

prrat f histological setion Ofa1ad ebomarow.-Although the operation of trephining the-sternum provides excellent material for imprint prepara-tions and for sections, the latter require decaldficationwhich impairs staining, particularly by the Romanowskydyes. It is, however, possible to prepare sections of thesmall fragments of marrow, free from bone, which may-be obtained by simple aspiration through an ordinarypuncture needle. These fragments of marrow, whichSchleicher (1944a) has termed " marrow units," arenaturally much smaller than the material obtained bysurgical trephining, but they can be stained readily andthey then provide a picture in miniature of the cellularity,proportion of fat spaces, and general architecture of thenuarrow. It is their small size, usually up to 1 mm. indiameter, which limits the value of the technique, forthere is always some uncertainty as to how representativeof the marrow the fragments are.

A number of techniques for sectioning these fragmentshave been described (for example, Davidson, 1941;Mertens, 1945; White and others, 1946; Cappell, 1947;Weisberger and Heinle, 1948), which differ in the details ofhandling the fragments and fixing and embedding. Innone of these methods is it nicessary to decalcify thetissue. These histological sections are quite easy toprepare; do they yield reliable and helpful information?The doubt as to whether the small fragments are reallyrepresentative has already been referred to. In practicewe find that although they are a useful adjunct to theexamination of films, and- reveal details of structureotherwise not easily appreciated, they do not often revealmore of diagnostic importance than do film preparations;indeed the latter are made from fragments similar tothose sectioned. They certainly are less satisfactory forthe study of subtle differences in haemopoiesis than are

well stained films, particularly as the cytoplasmicstructures are less well preserved. On the other handthey do iadicate the relationships of the marrow cellsone to another and to the parent reticulum cells and theyshow the proportion of the marrow occupied by fat andthe relationships between sinusoids and the marrow cells.In addition it is easy to recognize cells such as phagocyticreticulum cells, which are difficult to identify in smears,and easily broken in the spreading of the film. Similarlymegakaryocytes, which are only seen in smears in smallnumbers, are more prominent in sections, and occa-sionally a satisfactory and decisive view of tumour cellsMay be obtained.Whilst relatively large -fragments of marrow are

obtained by simple aspiration on most occasions, wehive sometimes been unsuccessful just when it seemedmost important to obtain material. In these patients,mostly suffering from marrow aplasia or hypoplasia,marrow fibrosis (" myelosclerosis "), secondary carcino-matosis, or aleukaemic leukaemia, the films are similarlyuinformative. In myelosclerosis and in carcinomatousinfiltration the tissue is presumably too firmly organized,with a thickened reticulum and perhaps an increase incollagen also, to be disrupted by suction. In the leuk-aemias and-in " aplastic" anaemia it is more difficult tounderstand why aspiration should fail; sometimes it is

diffiult even to -obtain blood. Possibly the ease withwhich material is withdrawn depends -not only on thetexture of the marrow, in particular the volume ofdeveloping marrow cells compared with- the basic reti-cular structure, but also on its blood supply. If theblood circulating through the bone marrow is restricted,as it might be, due to proliferation of leukaemic cellsor perhaps to marrow aplasia, difficulty in aspirationmight result; for it is clear that, when one is aspiratingfrom a small enclosed area, blood must enter toreplace the material withdrawn. In practice, if theoperator fails to obtain marrow from the sternum,it is advisable to try another site such as the lumbarspine. If aspiration of the spinq is unsatisfactory,trephining through an introducer needle (Tuirkel andBethel, 1943) or surgical trephining of the sternum maybe needed.

In Plate I are illustrated photomicrographs of sectionsof fragments of marrow from patients suffering from avariety of blood disorders. The specimens were obtainedby aspiration of the manubrium of the sternum andprepared by the method of White and others (1946).

The Extent of the Normal Marrow at DifferentAges

As the normal distribution of the bone marrowhas a bearing on the selection of sites for biopsy,this question will be briefly considered. At birth,active red marrow is found throughout all theossification centres in the skeleton, and puncture ofthe tibia medial to the tubercle will yield activemarrow in young children. The regression ofactive marrow from the long bones in later child-hood and adolescence and the partial regression inthe flat bones is well illustrated by Custer andAhlfeldt (1932). The active red marrow is replacedby fat in which the potentially haemopoietic reti-culum cells normally lie dormant. Under abnormal.conditions, such as in many anaemias and in theleukaemias, these reticulum cells once more becomeactive and form haemopoietic stem cells; the fatcells niay almost completely disappear. There doesnot seem to be any striking difference between theextent of the marrow once the growth period iscom.pleted. There is no significant difference indistribution and cellularity between the marrows ofelderly subjects andnormal adults (Reichand others,1944;' Leitner and others, 1949b).The cause of the apparent increase in volume of the

marrow in childhood is uncertain-; it may in fact be moreapparent than real. It is not certain that the volume ofmarrow related to the child's total blood-cell -volume,which it constantly maintains, is any greater than themarrow-total-blood-cell, volume relationship in theadult. A possible cause of a hyperplastic marrow wouldbe that the erythrocytes (and leucocytes) in cidhohad a shorter life than in the adult. Mollison's (1948)

4-



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

observations survival of the blood of newborn*infafits1other normalnfants suggests, however, thatpostnatal exythrocytes are likely to survive for a normallength of time. Moreover, the total granulocyte countin childhood and adolescence is also very similar to thatof the adult, which suggests that the demand on themarrow for granulocytes is similar throughout life andis not likely to be a cause of marrow hyperplasia.Normal erythropoiesis: site of development.-The

exact site of erythropoiesis in the normal marrow andthe problems connected with the haemopoietic potenciesof the more primitive marrow cells have been a sourceof controversy for many years. Although it is not ourintention in this paper to review in detail the manytheories of haemopoiesis that have been suggested, alimited treatment is required. A cardinal point in theeolyphyletic theories of Sabin, Cunningham, and Doanand co-workers has been that, whereas granulocytes areformed extravascularly, the erythrocytes arise intravas-cularly from the lining endothelium of inter snusoidalcapillaries (Doan, 1923y ldbaiaid others, 1925). Theire0idei3as based partly upon the study of regenerationoccurring in the hypoplastic fatty marrow of pigeonspreviously starved. However, granulocytes and erythro-cytes do not seem to develop in this way in man and inother mammals, and an extravascular origin for both hasbeen claimed by the majority of workers (Bunting, 1906;Maximov, 1910; Drinker and others, 1922; Bloom,1938a). More recently Hamre (1947) has described theextravascular formation of erythrocytes and granulocytesin rats and the subsequent migration of the more maturecells across the intact sinusoidal endothelium.

Study of our own sectioned material has convinced usof the correctness of the conception of extravascularhaemopoiesis in man. The more primitive haemopoieticcells in both normal and abnormal marrows alwaysseem to be in the stroma of the marrow and outside themarrow sinusoids and capillaries.* In the sinusoids,however, there may be seen a few partly mature forms,such as polychromatic normoblasts and metamyelocytes,in addition to mature erythrocytes and leucocytes.

If it is admitted that haemopoiesis is extravascular,the problem of the migration of the developing bloodcells from the marrow stroma into the blood streamremains to be solved. In the case of granulocytes, withtheir known capacity for amoeboid movement, themigration of the more mature forms across an endothelialbarrier presents little difficulty and is generally admittedto occur, but the movement of the non-motile normo-

blasts is more difficult to understand. In Drinker andothers' (1922) view it was the pressure of expandingcollections of maturing normoblasts that forced thelatter through the sinusoidal endothelium. Maximov(1910) and Key (1921) postulated temporary deficienciesin the continuous endothelium but did not produceconclusive evidence of this. Hamre (1947) has failed todemonstrate in rats any such breaches in continuity. He

* Silver impregnation of the marrow, by demonstrating the reti-cular framework, helps to differentiate the marrow stroma with itsnetwork of fine fibrils from the sinusoids which lie on a layer oflongitudinal argyrophil fibrils (see Plate I, Figs. 5 and 6).

observed the passage of both granulocytes and erythro-cytes through the sinusoidal walls, and considered thatthe cells passed between the margins of the lining endo-thelial cells and that pressure from the growing mass ofcells in the stroma might well help them to do so. Ourevidence from human material points in the samedirection; in hyperplastic marrows such as are seen inAddisonian pernicious anaemia we have observed groupsof cells bulging into the lumina of the sinusoids. Isaacs(I933) has called attention to the mutual adhesiveness ofthe more primitive marrow cells. As the cells maturethey become free from one another through liquefactionof the intercellular material and come to lie in " lake-like " spaces. This separation is a necessary preliminaryto a cell's entry into the vascular sinusoids proper. Therelationship between erythrocyte or normoblast migra-tion through the sinusoidal walls and the formation ofpoikilocytes will be considered in a later section.Although it can be accepted that the erythrocytes andgranulocytes are normally formed extravascularly in themarrow, this is not necessarily true of haemopoiesis inother situations. In the spleen, for instance, activehaemopoiesis in dilated sinusoids may sometimes beseen in pathological material (see also the section onfoetal erythropoiesis).

The Morphology of Normal ErythropoiesisNomenclature of the erythrocyte precursors.

-For the sake of simplicity the vexed questionof nomenclature* has so far not been considered,but the subject cannot be entirely avoided. Thenormal form of erythropoiesis gives rise to thenormal erythrocyte (the normocyte) by a processbest referred to as normoblastic erythropoiesis. Theearliest definite erythrocyte precursor is the pronor-moblast, and later nucleated forms a-re normoblasts.In later sections we shall refer to and describe othertypes of formation, megaloblastic, intermediate,and macronormoblastic, etc.The term " erythroblast " should be reserved for

any nucleated erythrocyte precursor without refer-ence to the form of development, in accordance withthe original usage of Ehrlich (Ehrlich and Lazarus,1898, 1900).Synonyms for the pronormoblast and normoblast.-

Ferrata (Ferrata and Storti, 1948) called the earliestrecognizable erythrocyte precursor the pro-erythroblast,derived from the haemocytoblast and developing througherythroblast stages to the erythrocyte. This terminologyis used by the Italian school of haematology generally(Ferrata and Storti, 1948; Rastelli, 1943c). Naegeli(1931) has used the term macroblast (makroblast) torefer to the younger and larger erythroblasts, particularlyin embryonal organs and in the marrow in states of rapid

* In this paper we discuss the orthodox and traditional nomen-clature only. Osgood and Ashworth (1937) introduced an entirelynovel system which has not been generally accepted. The wholequestion is now once more under review by the International Societyof Haematology.

S



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. DACIE AND J. C. WHITE

blood regeneration, and Swiss authors such as Rohr(i94)) and Leitner (1941) have used the term similarly,but this usage is confusing and better avoided (see thediscussion on macronormoblasts in a later section).Similarly, the use by Sabin and co-workers of the term"megaloblast " for young basophilic red cell precursorsunder normal conditions (Doan and others, 1925) hasno justification and has generally been abandoned.

Schulten (1937) derived the normal erythrocyte fromthe pro-erythroblast through macroblast and normoblaststages, and Dameshek and Valentine (1937) a series ofnormoblasts, A, B, and C, from haemohistioblaststhrough the erythrogone. Isradls (1939), whilst derivingthe erythrocytes from multi-potent haemocytoblasts,called the earliest recognizable stage the pro-erythroblast,and the subsequent stages of normoblasts A, B, and C,in increasing order of maturity.Turnbull (1936a), basing his observations on sectioned

material, described the derivation of normoblasts frompluripotent haemocytoblasts through a series of "primaryerythroblasts." He recogniized three stages of normo-blasts of similar size; the earliest with basophilic cyto-plasm, then the polychromatic stage, and finally ripeorthochromatic normoblasts of the final stage.

Formation in intra-uterine life.-Details of the forma-tion of the erythrocytes in intra-uterine life are of morethan academic interest, for there are certain interestingcontrasts and analogies between foetal erythropoiesisand that of the adult in health and in disease. Ehrlich<1880; Ehrlich and Lazarus, 1898, 1900) was the first tocall attention to the similarity between the megaloblastsof pernicious anaemia and the haemoglobinized*primitive erythroblasts developing within the primitiveblood vessels of the presomite human embryo of 2 to5 mm. in length. These primitive foetal erythroblastsdo not lose their nuclei and are very large cells. Theyare replaced by the first " definitive " erythroblasts,which in size and nuclear structure are intermediatebetween the cells of the primitive generation and thoseof adult life.t These cells lose their,uclei when matureand give rise to erythrocytes (macrocytes) whose volumeis greater than those of the normal adult. The nucleatedprecursors may be termed macronormoblasts4 (Jones,1943). The liver is the chief site of their formation inmid-foetal life.§ From the fifth month of intra-uterinelife onwards erythropoiesis commences within themarrow cavities of the ossification centres and at birthintramedullary formation is of prime importance,extramedullary foci having almost entirely disappearedby this time. The type of formation at birth is normo-blastic and similar to that of the adult.

It appears that the type of haemoglobin produced at all stagesof intra-uterine life is the same-that known as foetal haemoglobin(H. M. Jope, 1948; Hoch and others, 1949).

t The primitive generation of erythroblasts is formed intravascu-larly, as are the definitive erythroblasts (macronormoblasts) in earlyfoetal life. Later, erythropoiesis in the liver is at first both intravas-cular and extravascular, but eventually is entirely extravascular, as itis in the bone marrow from the start (Gilmour, 1941).

t Some details of ma&ronormoblastic development are given laterin this paper.

Gilmour's (1941) paper is an important contribution to the studyof intra-uterine haemopoiesis. It is based on 57 human embryosand foetuses. See also Bloom and Bartelmez (1940).

The cause of this variation in cell morphology atdifferent age periods is far from clear. There is a gradualdiminution in erythrocyte diameter and volume as thefoetus matures,* but there is little evidence to prove thatthe macrocytosis is dependent upon an insufficient supplyof anti-anaemia factors (Wintrobe, 1946a). Possibly thepresence of the primitive persistently nucleated erythro-blasts recalls phylogenetic development. That foetal

A macrocytes may have a shortened life in vivo is suggestedby Mollison's (1948) transfusion experiments. In thisconnexion it is interesting to recall that macrocytosis(and macronormoblasts) are seen in adult life in certainanaemias associated with rapid blood formation.

It is well known that foci of extramedullary haemo-poiesis may be met with in adults in a variety of blood

Vdiseases, particularly where the bone marrow is widely,invaded by carcinoma or where the marrow is under-going fibrosis as in myelosclerosis. Less frequentlyextramedullary haemopoiesis is seen in haemolytic

.anaemia or even in pernicious anaemia. The foci occurmost frequently in the liver and spleen, sites of vigorousintra-uterine formation, but are occasionally found inlymph glands or as heterotopic foci (see Vaughan,1936a and b).Normoblastic erythropoiesis in the adult.-

The irregular lattice-like pattern of cellular andfatty areas in the adult human marrow has alreadybeen referred to, and close inspection of the cellularareas shows that focal areas of erythropoiesis aresurrounded by larger zones of leucopoiesis.The basic potentially haemopoietic cell in the

marrow is a reticulum cell. Maximov (1927) hasdescribed the appearance of these cells in sectionsas flattened, pale-staining, and inconspicuous, andhe considered them to be less differentiated than thedye-storing and phagocytic cells of the marrowstroma or the littoral cells of the marrow sinusoids.The question as to whether these more differentiatedphagocytic cells can form haemocytoblasts or moremature forms of haemopoietic cells has not beensettled and is discussed by Bloom (1938b). Ferrataand Storti (1948) recognized the haemohistioblast asan intermediate form between the primitive reticulumcells and haemocytoblasts. Clearly, irrespective ofnomenclature, there is continuity of developmentbetween the -most primitive and the more matureforms. Normally, these primitive cell types arepresent in small numbers and are somewhat difficultto identify in marrow smears. The cells are fragileand easily broken up by spreading, and cytoplasmicoutlines may be indistinct. In pathological statessuch as Addisonian pernicious anaemia, and incertain refractory anaemias where there is hyper-plasia of primitive cells, they are more frequentand may be seen in groups or even as syncitia (seeSchleicher, 1945).

Wintrobe (1946, p. 32) gives interesting data from thirty humanfoetuses; see also Wintrobe and Shumacker, 1935.

6



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

In film preparations the marrow reticulum cellsvary between 20 and 30 Cu. The cytoplasm is palelybasophilic and abundant, and may contain a fewazurophil granules; fine spherical mitochondria arenumerous and evenly distributed. The nucleus islarge, round, or oval, with a pale-staining* finechromatin pattern and one or more round or ovalnucleoli.

The haemocytoblastt is a large cell of somewhatvariable outline, in size averaging about 20 ,u indiameter. The cytoplasm is moderately deeplybasophilic, and contains numerous mitochondriabut no granules. The nucleus is round, averagingabout 15 ,u in diameter; it stains palely withRomanowsky dyes and presents a finely strandedand stippled (leptochromatic) chromatin pattern.Multiple nucleoli are present or a single elongatednucleolus of complex form bounded by a very finelayer of nucleolus-associated chromatin.

The pronormoblast is a slightly smaller cell thanthe haemocytoblast and usually of a roundedoutline, although often showing fine pointedprocesses at the periphery of the cytoplasm. Theaverage diameter is less than 20 p and the nucleusless than 15 [L. The cytoplasm is more basophilicbut less extensive than in the haemocytoblast;mitochondria are numerous. The chromatin patternof the nucleus is more condensed and the nucleoliare smaller and are less easily seen, being surroundedby denser nucleolus-associated chromatin. Theterm normoblast refers to the developing erythro-blast from the stage when the nucleoli have dis-appeared to the loss of the nucleus itself. As the cellmatures there is a progressive shrinkage in size toabout 8 [Land a parallel decrease in size ofthe nucleus.The cytoplasm becomes less basophilic as the cellmatures, and increasingly acidophilic, this latterproperty being considered to be due to the forma-tion of globin or haemoglobin.Thus normoblasts may be described as basophilic,

polychromatic, or orthochromatic, although in ourexperience fully orthochromatic cells are rarely

* The descriptions given hereafter refer to Romanowsky-stainedmethanol-fixed dry film preparations. The appearances in sections orin wet fixed films differ due to the preservation of a more sphericalshape of the cell and to the use of different fixatives. A descriptionis outside the scope of this article.

t The haemocytoblast of Ferrata is the marrow stem cell of theerythrocytes, granulocytes, and megakaryocytes; it is the mostorganized stem cell capable of developing and differentiating to morethan one type of blood cell (Ferrata and Storti, 1948; Rastelli, 1943c).Downey (1938a) discusses at length the potency of the haemocytoblast.He agrees that it is multi-potent, but unfortunately confuses the issueby referring to it as a " myeloblast." Helly's (1910) " erythrogone "is probably the same cell (see also Dameshek and Valentine, 1937).The multi-potency of the haemocytoblast is not universally accepted.Rohr (1940), for instance, derives the erythrogone (uni-potent)directly from the reticulum.

T,he development ofthe granulocytes from haemocytoblasts proceedsin a parallel fashion to that of the erythroblast. Mitosis occurs onlyin the earlier stages, and as the cell matures the degree of cytoplasmicand nucleolar basophilia diminishes and at the same time the azurophiland later the specific granulations appear.

encountered in normal subjects-that is to say, cellsin which the cytoplasm is as acidophilic as that ofadult erythrocytes. Such cells are, however, oftenillustrated (Wintrobe, 1946, plate 2; Whitby andBritton, 1946. plate 4). Everything depends uponthe staining. It is possible to stain films to showorthochromatic normoblasts, but they are seldomseen if staining is controlled in such a way thatpolychromasia is well demonstrated. Indeed it isdifficult to imagine an orthochromatic normoblast(having presumably lost its cytoplasmic basophilia)becoming a polychromatic reticulocyte and regainingcytoplasmic basophilia when its nucleus is disposedof. This would only be possible if the desoxy-ribonucleic acid of the normoblast nucleus contri-buted to the basophilic staining of the cytoplasm ofthe non-nucleated cell which developed from it, andfor this there is no evidence. Brachet (1946) states,however, that Dustin suggested such a possibility.Reticulocytes are Feulgen-negative. If ribonucleicacid had been formed from desoxyribonucleic acid,an increase of both the reticular-filamentousmaterial and diffuse basophilia removable by ribo-nuclease would be expected in the reticulocytes ascompared with the normoblasts, but this does notseem to be so. A point that should be stressed isthat the progressive diminution in size and pyknosisof a cell's nucleus is a more reliable guide to thatcell's maturity than is the state of ripening of thecytoplasm, and that even in the normal the relativeprogress of nuclear and cytoplasmic ripening variesappreciably from cel to cell.The areas of deeply staining chromatin, which

first appear in the vicinity of the dwindling nucleoliand near the nuclear membrane, are brought togetherby shrinkage of the nucleus and ultimately fuse. Inits final form the nucleus is entirely pyknotic andabout 4-6 tL in size. That the changing pattern ofthe nuclear chromatin is not an artifact producedby fixation has been shown by studies with theelectron microscope (Rebuck and Woods, 1948).

A few additional details as to the cytoplasm oferythroblasts may be added. Although the earlier formspossess a cytocentrum, Golgi apparatus, and mitochon-dria (Maximov and Bloom, 1948), these structures arelost in the course of differentiation. Mitochondria maybe demonstrated by the supravital use of Janus green(Key, 1921; Sabin, 1928), and Jones (1947, 1948) hasshown that, due to their content of lipoids, they can bestained by Sudan black by the method of Baker (1945)in air-dried films fixed in formol-calcium chloride (seePlate IV, Fig. 5). Jones has also used phase contrasttechniques. He has demonstrated that in air-driedunstained films mounted in formalin, mitochondria lieabove and below the nucleus, and that in this way theymust contribute to the chromatin pattern of the nucleus

7



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

1-. V. D EANJi. C. W- ITE

as reveaed by saining with komanowaky dyes. Hebasalso shown that the clear perinuclear area represents theunstained negative images of mitochondria and nothyaloplasm (see Wilson, 1928). More cytoplasmicparticles are revealed by phase contrast microscopy thancan be stained supravitally by Janus green.

Differental ceil couts on marrow films.-Most workers perform differential cell counts onmarrow films, and by presenting the data in theform of a " myelogram " express the incidence of

A B

7

C

I2

the various cell types as per . Such figures,unfortunately, are less accurate than they appear tobe at first sight. Not only is there an unknownadmixture with cells from the peripheral blood inwhich the marrow cells are diluted, but in additionthere is a tendency for the more primitive cells andrelatively fixed marrow cells such as megakaryocytesto remain behind in the marrow. Ideally, differen-tial counts should be performed on sections ofaspirated material, but here, unfortunately, there is

D E

19'9(i

10II

2.0

24

2S20

TExT FIG. 1.-Series of scale drawings of maturing erythroblasts. Series A shows representative mnegaloblats(Figs. 1-5) from a patient with severe Addisonian anaemia, erythrocytes 1,300,000 per c.mm. In series B(Figs. 6-10) are illustrated intermediate megaloblasts of comparable maturity from the marrow of a patientwith steatorrhoea, erythrocytes 3,500,000 per c.mm. In series C (Figs. 11-15) the cells (normoblasts) werefrom the marrow of a normal adult man. Series D (Figs. 16-20) shows cells (macronormoblasts) of sinilarage groups from a patient with a macrocytic haemolytic anaemia (nocturnal haemoglobinuria), erythocytes1,500,000 per c.mm. In series E (Figs. 21-25) the cells were drawn from the marrow of a patient withwasever iron deficiency erythrocytes 3,000,000 per c.mm.

In each series the topmost cell is a primitive cell (promegaloblast or pronormoblast).

8

23



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

RE HAE PR N B.N PN PN NEAR'LY LATE.

difficulty in the proper identification of cells.Dameshek and others (1937) give illuminatingfigures for parallel observations made on sectionsand on smears made from puncture fluid, and otherobservations are reviewed by Osgood and Seaman(1944) and by Leitner and others (1949). Osgoodand Seaman, in an excellent review, also discuss theunavoidable statistical errors involved in makingdifferential counts and the necessity for countinglarge numbers of cells, particularly if an. attempt isbeing made to assess the frequency of cells presentin only small numbers. A further difficulty is theimpossibility of accurately dividing into arbitraryclasses cells ofwhich every gradation ofdevelopment

TEXT FIG. 2.-Curves of maturation and mitotic activityduring erythropoiesis.

Upper figure: Maturation curves.Maturation curve from calculated means

offigures from 15 normal adult marrows (6 x 1,000;9 x 500 cell counts). The observed range is indi-cated in the stippled area.Maturation curve for erythropoiesis in Addi-

sonian pernicious anaemia.C G- Normoblastic development.

< *- * Megaloblastic development.--- -Maturation curve in iron deficiency

anaemia........Maturation curve in congenital haemo-

lytic anaemia.Ordinates: Per-cent of cell types.Abscissae: Stages of erythropoiesis RE,

reticulum cells. HAE, haemocytoblasts. Pr.N.,pronormoblasts. B.N., basophilic normoblasts.P.N. early and late,- early and late polychromaticnormoblasts. N., pyknotic normoblasts.

Lowerfigure: Mitosis during erythropoiesis.Distribution by developmental stages of67 erythro-

blasts in mitosis, encountered in 100 mitoses fromnormal marrows.

Ordinates: Per cent of the mitotic cells.Abscissae: Stages of erythropoiesis, as in upper

figure.

may be seen; and the fact that different authorsnaturally use different criteria to separate, say,"early " from " late " normoblasts makes compari-son between their figures still more difficult.For all the above reasons the figures given in the

literature for the "normal " myelogram differwidely, as is illustrated in the tables of BodleyScott (1939), Osgood and Seaman (1944), andLeitner and others (1949c).Pontoni (1936) and Rastelli (1943d) have also

discussed the analysis of differential cell counts.Pontoni correctly uses the term " haemomyelo-gram " for the ordinary differential count, andrestricts the use of the term " myelogram " to counts

9



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/


made on the residuaum of cells after excluding thefully ripened ones, that is the segmented granulo-cytes, lymphocytes, and monocytes, etc. Thecommonly employed leucocyte-erythroblast or leuco-cyte-erythroid ratio, being based on the " haemo-myelogram," is of limited usefulness; in practiceit gives a very wide range in normal subjects (2: 1 to12: 1). That dilution with peripheral blood is amajor factor in leading to a high leucocyte-erythroidratio is shown by our lymphocyte values in a seriesof fifteen normal volunteers. The lymphocytepercentage varied between 13 and 31 per cent,and although some of these cells may have beenderived from isolated lymph follicles, in mostinstances it is likely they came from the peripheralblood.

Pontoni's " leuco-erythrogenetic " ratio is abeter expression ofthe relative proportions ofleuco-poiesis to erythropoiesis. Mature' leucocytes areexcluded from the count, and in our series ofnormalsubjects the ratio varied from 0.56: 1 to 2.67: 1. Inmost anaemias the ratio is below unity.

Fortunately, the factor of dilution with peripheralblood does not affect differential counts on marrowerythroblasts, unless there is a marked peripheralerythroblastaemia. These differential counts aremoresatisfactorilyexpressedbyrelating the numbersof the different classes of eryrhroblasts to the totalnumber of erythroblasts rather than expressingthem as percentages of the total nucleated cellspresent.

Because ofdifferences in classification it is difficultto find from the literature normal ranges for theproportion of erythroblasts of different maturities.There is, however, wide agreement that in health thepronormoblasts and basophil normoblasts are fewin number and that normoblasts ofmedium maturitywith various grades ofacidophilia of their cytoplasmand increasing density of the nucleus predominate.Fully pyknotic cells are less numerous, and pyknoticcells with fully orthochromatic cytoplasm are rarelymet with. In our series of fifteen normal subjects,2 per cent ofthe erythroblasts were prononnoblasts,5 per cent basophilic normoblasts, 51 per cent earlyand late polychromatic normoblasts, and 42 per centpyknotic normoblasts. In disease these proportionsmay be much altered.

Maturationcurves(Pontoni, 1936; Baserga, 1939)have been used to illustrate the changes in propor-tion of thevarious classes ofcells in different diseasesor at different stages in the course of an illness(Text Fig. 2). Cotti and others (1938) haveused a rather different way of expressing theproportion of primitive to maturing cells, and bytheir maturation index indicate the numbers of

TEXT FIG. 3.-Densitometer records across spectrogramof frog red cell at three wavelengths, obtained fromreflecting microscope. (Figure kindly supplied byDrs. R. Barer, E. R. Holiday, and E. M. Jope.)

N., cell nucleus. C., cytoplasm.405 mu. In the intense Soret band absorption

due to haemoglobip.312 mp. A waVelength at which little specific

absorption occurs.265 mjs. In the strong absorption band due to

purine and pyrimidine residues of the nucleic acids.

polychromatic and orthochromatic erythroblastsrelative to the basophilic erythroblasts. We haveemployed a similar device to express the marrowimmaturity in Addisonian and other megalocyticanaemias (Text Fig. 4).

Growth and Differentiation of Erythroblast

In the previous sections we have described thestages in development from the haemocytoblast tothe pyknotic normoblast. It is almost superfluousto state that this separation into stages is artificial;

I

t-Ct-.Cw

S t~~~~405ryi

512m5

265m

1-0



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

60

40

20

ERYTHROID-LEUCOCYTERATIO

RATIO OF PRMATURING

0

200

0

*

*

o

0.* 00o

0 0

0of I0 *

0 .* 0

* 0

0

0 a

2 3 4 1ERYTHROCYTES PER C.MM.

TEXT FIG. 4.-Relationship between the peripheral erytterythroid-leucocyte ratios (left-hand chart) and thecell ratios (right-hand chart) in thirty-five patientsmarrows. Twenty-six of the patients had Addisonianfour patients had steatorrhoea, 0; in three patients thewith pregnancy, o; in two with malnutrition, *;anaemia had followed a gastro-enterostomy.

in reality there is a smooth gradation-from the mostprimitive to the most mature forms. This processof development is a dual one, involving growth andcell division and differentiation. According toCowdry's (1942) classification of levels of cellularactivity the haemopoietic reticulum cells and haemo-cytoblasts are classed as vegetative intermitotic cellsand the earlier erythroblasts as progressively differen-tiated intermitotic cells. In the final stages ofdevelopment, when differentiation is completed, thecells are fixed post-mitotic cells.Mitosis during erythropoiesis.-A proportion of the

haemopoietic cells are always in mitosis. Japa (1942)studied the sternal marrows from three non-anaemicsubjects and found 0.5 per cent of mitoses in fixed andstained films, but in aceto-carmine squash preparationsfrom the same material there were 1.5 per cent, about45 cells of the 100 in mitosis being erythroblasts. Hethought that some of the dividing cells were disruptedwhen films were made. Mitoses may be seen at allstages, from haemopoietic reticulum cells to the poly-chromatic erythroblasts. Mitoses are not observed inlatest polychromatic and pyknotic normoblasts, which isnot surprising in view of the nuclear structure. Normallythe absolute numbers of mitoses increase from the more

primitive to the more mature stages, and it is rare to finda primitive cell in mitosis without extensive search.However, as the numbers of intermitotic cells alsoincrease at each stage, the relative number of cells inmitosis at each stage is not so variable (see Text Fig. 2).

This suggests an orderly increase in cell numbers by

mitosis throughout the growthand earlier differentiation

tIMITIVE TO phases of erythropoiesis.CELLS The products of the final

mitoses then completedifferentiation irreversibly.Abnormalities in mitosisleading to the formation ofmultinucleated cells are notinfrequently found, even innormal erythroblasts and

* their precursors (see p. 30).i Fieschi (1938) represented* the incidence of each stage* a of the mitotic cycle graphic-o 0 ally and found that meta-

* phaseswerethemostnumerous0 o 0o in normal erythropoiesis, and

- our results and those of Japa2 3 4 (1942) also indicate this.

Increased erythropoiesis,irocyte counts and the however, leads to a higherprimitive cell-maturing proportion of prophases.who had megaloblastic Life span of the erythro-pernicious anaemia, 0; blasts.-The exact number ofanaemia was associated mitoses that take place inand in one patient, the development of a normal

erythrocyte and the lengthof the mitotic and inter-

mitotic periods at each stage ofdevelopment are unknown.The whole process probably occupies several days.Ponder (1948) calculates that in the rat the lifetimes ofthe haemocytoblast, " erythroblast," normoblast, andreticulocyte are 0.23, 0.35, 2.3, and 3.6 days respectively.Hevesey and Ottesen (1945) have shown that if phos-phates containing p32 isotope are fed to hens, isotopeappears in the desoxyribonucleic acid of the nuclei ofcirculating erythrocytes after five days (see also Hevesey,1948a). Shemin and Rittenberg (1946) have shown thatin normal man NL5-containing haem appears in circulat-ing erythrocytes within two days of the ingestion ofN'5-glycine. This represents the time between theformation of haem and the delivery of the erythrocytesinto the blood stream, and is only part of the total timerequired for erythropoiesis (see also p. 21).

In man there is other indirect evidence. The responseto potent liver extract in a severe case of Addisonianpernicious anaemia indicates that the reticulocytes aredelivered into the peripheral blood stream at the maxi-mum rate at about four to six days after an injection ofliver. This period of maximum reticulocyte productioncorresponds presumably to the final maturation of thepredominating age group of marrow cells, which in asevere case are the abnormal haemocytoblasts andbasophilic promegaloblaW. Owren's (1948) studies onthe mechanism of the production of crises in congenitalhaemolytic anaemia also provide some evidence aboutthe rate of development of the maturing erythroblast.In his case 4, regeneration of " pro-erythroblasts'"(pronormoblasts) in the marrow had started on the ninthday of the crisis. Reticulocytes were appearing in the

11



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. DiSACE'. W11

.- . 1.I

-- oo

so 4 B

s x

xe to)0

Xt 00n

ttz,

X*

t_

°. ')X

en -Ij

_ _

E° x'ds-SX M.

C oRS

> <,, bo E S-, .Sb

Stzoo w S

o S zt ;e>, .Dt

co

t.oo 114)

X

Nbc

CLE bo

x

0

cloo

O._r

.ox0c

mtl:.,-I

* (A

a =0-

.0 :io-W)

CO 00

b-..o0E :s

.1"

E-I.. :

r_

1._

0,-

" -

0

O0 -

a;~

rCL, ° . 0 0

er- o *VaX oQ

. Ae.oo u E

L~~~~~C U

I. ..

CQ6

co (L

bo

$. 4)4- 0

>E -0s¢-

E UM. - sv) _ P.

.wtce *_

.Q}Y

oR ts^- bk s:Cs osbr @-

s o s

ot:Fm st

* c#o-- -b

= rA

=.U 2.Uu

E r.

O)4- ('4

_c

-.

_1.

=, X

en",c) S.U ot

U00

0 V).-O

.i *,-.- PeC.) :'I

iv0* C_

6 X

0010c efn

0

._

S~O -

i.-

tadb

C xs

t-11 %6

v t-

:3,~C

*U--:.. *

ot

x

I'D

s -r

.o_:

6 EXL

CZ CZS

E 3 CEa

E-

E,6

00

xI.qC71,

a;+.a

u0tv

rE0mlaE00.-j

r.0-"

-4cq

611,

r-

x

_rIn ::t,

o xo

1. x

. --

rA

Cq-~

as

bo

03.4 3

X

-

CACZ

_ z

bow)

, X-un

la

_ rA

!U v

1.0V1

0U)ula04099X

94w

.4

2040zlm

0zt

2990z

z_-

9.U

L L.

O Z

_Xo

,_e

_4Q

4);:CZs* S

0,6=

W3 a

4 o °°

C4.__

,-7Cr

E t

03E, - u

x *tZ 0

x;nE*

vofio = ° °t o 0

v~~~~~~~~- cL

xx o (U CZ

0 CZ 10 >I

va,0 bo(U

bo- 0o 0 -ba C.- a.0Cl r..- CZE o

. Ol >,0 E 10

'tj OJ

C= Cd CaCd 't= ".,

c E 4)., rA-- rA E=

to 06CA Cd 0 0

0 0 E 'CA*

0 0 -4 rAen Ca Cdu 0 7a ---

Cd

x



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

K4-

(N4(N

0

L)(NJ

C)N

e..1 1

.S., '.f

-xx

.1i

(Nmn

-0

0C,

i.

,_AI,.\

enen

N COD

_N _e

V-

'4' ( ,/-4m so rs oo

#hIc'0~~~~~~~~~~~~~~t

(0CrN

re)

0

(N4

en

%ON

f



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

peripheral blood at a maximum rate between the twelfthand fourteenth days, giving a maturation time of threeto five days.The orthodox view that a single normoblast gives rise

to a single erythrocyte has not passed unchallenged.Plum (1947) has maintained that the numbers of erythro-blasts in mitosis are too small to yield the requisite dailyquota of erythrocytes necessary to maintain equilibrium,and has published details of studies in vitro in support ofhis contention. He believes that several erythrocytesmay be derived from each normoblast by budding off(gemmation) from the cytoplasm. Bostrom (1948)supports Plum's view and illustrates schematically howthe budding process is supposed to take place (see alsop. 30). These interesting and revolutionary views needconfirmation.* Ponder (1948) on theoretical groundsdoes not see the necessity for gemmation, and it cannotnecessarily be assumed that " erythroblasts " behave invivo as they do in vitro in artificial culture.

In our experience study of biopsy films of normal orhyperplastic normoblastic marrows yields no supportfor the mechanism of gemmation. The general appear-ance of films seems to indicate that ripe normoblastsbecome reticulocytes after the nucleus is lost. If, forinstance, an actively erythropoietic marrow film from apatient with haemolytic anaemia is examined, satisfactoryevidence of budding from the cytoplasm is extremelydifficult to find. On the other hand pyknotic normo-blasts with well formed evenly rounded polychromaticcytoplasm are frequent. Some normoblasts with scantyirregularly -c'ontoured polychromatic or slightly baso-philic cytoplasm may be seen, and at first sight it wouldappear possible that these were cells which had alreadybudded off cytoplasm. Against this hypothesis is theserious objection that the only non-nucleated masses ofcytoplasm usually seen are fully formed polychromaticerythrocytes, which seem to be clearly derived frompyknotic nucleated precursors. Basophilic and irregu-larly formed cytoplasmic masses are rarely found innormal or hyperplastic normoblastic marrows. Whathappens to the normoblasts with scanty cytoplasm is farfrom clear: possibly they never mature at all. Thepossibility of limited gemmation appears more plausiblein marrow films from patients with severe megalocyticanaemias. Promegaloblasts and polychromatic megalo-blasts of irregular contour are not infrequent, 'andseparated fragments ofcytoplasm may be found; whetherthese fragments can develop into mature erythrocytes isunknown.

Derivation of reticulocytes from normoblasts.-It isgenerally considered that reticulocytes are direct suc-cessors of ripe normoblasts, after loss of the nuclei.The evidence for reticulocytes being younger than thebulk of the circulating erythrocytes is firmly based onthe constant finding of increased numbers of reticulo-cytes whenever other evidence indicates an increasedproduction of erythrocytes, as well as upon the fact thatblood from the bone marrow contains a higher pro-portion of reticulocytes than is found in the peripheral

* Several other heterodox hypotheses have been put forward (seeLancet, March 29, 1947).

blood (Ungricht, 1938). Moreover, culture in vitro(Jacobsen, 1947) and survival studies on transfused bloodcontaining a large number of reticulocytes have demon-strated loss of basophilic material and transformationinto non-reticulated erythrocytes (Young and Lawrence,1945).

Reticulocytes are on the average probably a littlelarger in diameter, and possibly a little thinner, than areadult erythrocytes, and it is presumed, therefore, that aslight alteration in corpuscular shape takes place afterthe cells have left the bone marrow. It is possible thatit is the spleen which is largely responsible for themodification of corpuscular shape (Miller and others,1942); however, the evidence can be interpreted in morethan one way.*The characteristic feature of reticulocytes is their

ability to react supravitally with certain basic dyes;they contain a component which is precipitated inunfixed corpuscles by brilliant cresyl blue as a bluegranular network, and which in methanol-fixed air-driedfilms gives a diffuse basophilic staining and an appearanceof " polychromasia " to the whole field of corpuscles.Dustin (1944, 1946) has shown that this basophilicmaterial contains ribonucleic acid, and it seems verylikely that it represents the last remnants of the basophilicribonucleoproteins so abundantly present in the primitivecells from which the reticulocyte has been derived.Indeed, if material aspirated from the bone marrow istreated with brilliant cresyl blue before smears are madeand fixed, staining with Romanowsky dyes will thenshow that the basophilic material in the primitive cellsand in the developing normoblasts has been aggregatedby the dye in much the same way as in the reticulocytesthemselves. Another apparent difference betweenreticulocytes and adult erythrocytes is that reticulocytescontain more protoporphyrin than do the adult cor-pusclest (Watson and Clarke, 1937; Watson and others,1944). Haemoglobinizing normoblasts also appear tocontain free protoporphyrin (Stasney and McCord,1942).

It is as yet uncertain whether all the erythrocytes thatenter the blood stream are unripened reticulocytes orwhether a proportion are fully ripened adult erythrocytes.Heath and Daland (1930), Young and Lawrence (1945),and Wintrobe (1946) believe that a proportion of maturecorpuscles are delivered. Baar and Lloyd (1943) andNizet (1946) disagree with this. Nizet's view that all thecells delivered are reticulocytes is based upon a con-sideration of the constant proportion of the most maturetype of reticulocyte in the blood stream compared withmore immature types, and upon cross-circulationexperiments with marked Heinz-body-containing

* There is little question that reticulocytes seen in pathologicalstates, such as in recovery frcm haemorrhage and in some haemolyticanaemias, are macrocytes with an increased diameter and volume(Dameshek and Schwartz, 1940; Wintrobe, 1946b). It is probably afair inference that normal reticulocytes are also a little larger thanadult corpuscles, but more evidence is required on this point. Thecontroversial question as to how the spleen influences erythropoiesisis discussed by Doan and Wright (1946) and Dameshek and Estren(1947).

t The reticulocyte protoporphyrin cannot be held responsible forthe reaction with cresyl blue as suggested earlier by Watson andClarke (1937).

13



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. DACIE AND 1. C. WHITE

erythrocytes. The question is important in relation tocalculations of the longevity of the erythrocytes based onreticulocyte ripening times, and cannot yet be con-sidered as finally settled.Loss of the nucleus of the normoblast.-All authors

agree that the nucleus of the normoblast disappearsafter it becomes pyknotic* and the cell is ripe. Themechanism of its disappearance has been the subject ofmuch discussion. There is no unanimity; loss by expul-sion, karyorrhexis, or karyolysis have from time to timebeen thought to play a part. Naegeli (1931) consideredthat karyolysis took place, and states that this also wasthe view of Koellicker and Neumann. Amongst othersin agreement with this conception are Cooke (1930),Davidson (1930), Habelman (1940), Kracke (1941), andFerrata and Storti (1948). Howell (1890) thought onthe other hand that expulsion was the mechanism, andmore recently Maximov (1927), Turmbull (1936b), ahdWintrobe (1946) support this point of view. Fieschiand Astaldi (1946) claim to have observed this in theirmarrow cultures.

It has been argued that the formation of Howell-Jollybodies indicates that karyorrhexis does in fact takeplace, and that in certain circumstances it is incompleteand that a remnant of the nucleus persists. However,study of the bone marrow in those cases where Howell-Jolly bodies are being formed shows that they are separ-ated off from the nucleus at a relatively early stage,before the nucleus as a whole becomes pyknotic. More-over, this form of karyorrhexis is not normally seen, andit seems, therefore, unlikely that the nucleus of the normalnormoblast is disposed of by this mechanism.La Cour (1944) and Discombe (1946) have observed

budding of nuclear material into the cytoplasm of animaland human polymorphonuclear leucocytes, and La Courhas suggested that the normoblast nucleus may lose itscapacity to stain owing to loss of desoxyribonucleic acidduring haemoglobin formation. It is interesting to notethat in 1912 Schilling-Torgau suggested that it was onlythe chromatin ofthe normoblast nucleus that was lost. Itseems almost unnecessary to point out that the presenceor absence of a nucleus is usually judged by the presenceor absence of chromatin staining and that loss of thenucleus itself would be inferred if for any reason itscharacteristic staining reactions were lost. In vitro, forexample, depolymerization of nuclear desoxyribonucleicacid by the enzyme, desoxyribonuclease, and loss ofstaining has been shown by Brachet and Shaver (1948) totake place very rapidly.

It is true that in most smears of bone marrow cellsnaked pyknotic nuclei are occasionally found. Becauseit is impossible to exclude mechanical trauma as thecause, this observation by itself cannot be used tosupport the view that loss by expulsion occurs in vivo,especially in view of the small number of free nucleiencountered.

It is clear that no answer can yet be given as to howthe normoblast loses,its nucleus. The absence of transi-

* Pyknosis of the nuclus denotes in general an injured or dyingcell. M. J. D. White (1947), discussing pyknosis, suggests that thisprocess may be .a modified non-functional mitosis in which thechromosomes have become fused.

tion stages between pyknotic normoblasts and reticulo-cytes seems to us significant; whatever the process,it is clearly a rapid one.

Regulation of erythropoiesis.-In health,erythropoiesis is nicely balanced to compensate forthe normal daily loss of corpuscles, amounting toabout 1/120 of the total of the circulating erythro-cytes. Biopsy of the bone marrow gives a stillpicture of this normal compensatory erythropoiesis,but tells little of the mechanisms by which it iscontrofled and adjusted to the needs of the body.The very many haemopoietic factors which seem tobe required are well reviewed by Cartwright (1947).These include several metallic elements (Schultze,1940), amino-acids, vitamins, and growth factors,many of them being nutrients in a general sense.Knowledge of their importance is based uponcritical experiments with laboratory animals andobservations on man. A more mysterious controlmay be effected by endocrine glands and growth ormaturation substances of intrinsic origin. Clinicalevidence suggests that anaemia may develop inhypothyroidism (Bomford, 1938), in hypopituitarism(Snapper and others, 1937; Watkinson and others,1947), in hypogonadism (McCullagh and Jones,1942), and in Addison's disease (Wintrobe, 1946c).In rats, hypophysectomy regularly results in anaemia(Crafts, 1946). Finkelstein and others (1944) haveclaimed that male hormones stimulate and femalehormones depress erythropoiesis. It seems reason-able to conclude that the endocrine glands controlerythropoiesis to some extent, perhaps largely bytheir action on the growth and metabolism df thebody as a whole. The subject is reviewed byDaughaday and others (1948).The reticulocyte-ripening factor of Plum (1942)

has already been mentioned, and Jacobsen (1947)refers to unpublished work suggesting that theripening ofnormoblasts may be controlled by similarhumoral factors. These interesting studies requireconfirmation.A lowered oxygen tension has for long been

considered an important stimulus to erythroxoiesis,and the polycythaemia which develops at highaltitudes is evidence of this relationship. The recentstudies of Hurtado and others (1945) indicate thatalthough there is a wide variation in response, themajority of people residing at high altitudes developpolycythaemia. There seems a direct relationshipbetween the degree, duration, and continuity of theanoxia and the rise in the erythrocyte count.

The exact way in which oxygen-lack stimulateserythropoiesis is obscure. Grant and Root (1947),working with dogs, observed increased erythropoiesisto last for three weeks following loss of 30 per cent

14



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

of the blood volume, but-the lowered oxygen tensionand saturation in the bone marrow blood lastedthree to five hours only; and the same degree ofincreased erythropoiesis resulted when the bloodloss was brought about by smaller successivebleedings, which did not result in any diminutionin the oxygen tension (see also Grant, 1948;Berk and others, 1948). Again the studies madeby Rosin and Rachmilewitz (1948) on marrow

cultures in vitro indicated depression of erythro-poiesis at oxygen percentages below 12 per centand increased activity when the oxygen percentagewas raised to 50 per cent. These results are not,however, supported by the clinical observationsof Reinhard and others (1944), who observed adiminution in erythropoiesis, resulting in a fall inreticulocyte percentage and a fall in the erythrocytecount, when patients with sickle-cell anaemiainhaled 70 to 100 per cent of oxygen for eight to20 days. Tinsley and Moore (1948) reported anextension of this work. They observed depressionof erythropoiesis in patients with sickle-cell anaemiaand with congenital haemolytic anaemia who,inhaled 50 to 95 per cent oxygen, and a rapid rise inreticulocytes when oxygen inhalation was discon-tinued. The reticulocyte response of patients withAddisonian pernicious anaemia treated with liverand inhaling 70 to 80 per cent oxygen was alsopartly suppressed; a second and higher responsefollowed the cessation of oxygen inhalation.The picture is thus complex; reduced oxygen

tension stimulates and increased oxygen tensioninhibits erythropoiesis. Increased erythropoiesisregularly follows haemorrhage, although (in dogs)no reduction in oxygen tension -may be demon-strable, and in marrow cultures reduced oxygentension is a depressant rather than a stimulant.

It is obvious that further information is requiredon the relationship between plasma- oxygen tension,the respiration of erythroblasts, and erythropoiesis.Possibly the actual flow of blood through themarrow is a factor which affects erythropoiesis, bycontrolling the delivery of erythrocytes into thecirculation from the sinusoids, and perhaps even byaffecting erythropoietic activity itself.

Cytochemistry of Erythrocyte Development

Before discussing what is known of the chemicalchanges which take place in the developing normo-

blast, it is necessary to refer briefly to recent know-ledge in general cytochemistry, for in many respectsthe developing haemopoietic marrow behaves in thes ame way as do other actively growing tissues.

Although the factors which control the growth anddifferentiation* of cells remain obscure, more is known ofsome of the chemical substances involved, and of thechemical changes which take place during differentiation.The nucleic acidst play an important role in this process.They are biologically distributed in two chemical forms,the desoxyopentose polynucleotides or " desoxyribonu-cleic acid" and pentose polynucleotides or " ribonucleicacid." Caspersson and his school (consult Caspersson andSantesson, 1942; Caspersson, 1946; Thorell, 1947a) havestudied the distribution of these two forms of nucleicacids by means of their ultra-violet spectral absorption.This is revealed from a sequence of monochromaticphotographs or photometer readings obtained from aquartz microscope illuminated through a monochro-mator.Both cell nuclei and the cytoplasm of young, actively

growing cells can be shown to contain nucleic acids inhigh concentrations (Caspersson and Schultz, 1939,1940). Both forms of nucleic acid are found. Thenucleolus and cytoplasm contain ribonucleic acidsassociated with proteins; the Feulgen reaction* isnegative in these areas of the cell. The nucleolus mayalso contain protein of the histone type.t The chromatinand chromosomes contain, however, desoxyribonucleicacid, and are Feulgen-positive; they also containproteins.

Caspersson (1946) has suggested a number of principleswhich he believes to be of wide applicability: thatprotein synthesis needs the presence of nucleic acids;tha, the nucleus is the centre of the cell for the formationof proteins and that the nucleolus-associated chromatinsecretes into the nucleolus substances of protein naturewhich diffuse outwards to the nuclear membrane; andthat it is on the outer surface of the nuclear membranethat ribonucleoproteins are synthesized.t§

Brachet and his school have approached the subjectindependently from a different angle. They have utilizedenzymes (nucleases) to depolymerize nucleic acids fromfixed cells and have judged their effects by alterations inthe staining reactions, particularly with pyronin-methylgreen. Purified ribonuclease has been used by Brachet(1942, 1946) and Brachet and Shaver (1948), anddesoxyribonuclease (McCarty, 1946) has been employedto depolymerize desoxyribonucleic acid (Brachet, 1942;Sanders, 1946; Brachet and Shaver, 1948; Catchside and

* The Shorter Oxford Dictionary defines " differentiation " (biol.)as: "The process, or the result of the process, by which in thecourse of development a part, organ, etc., is modified into a specialform, or for a special function; specialization; also the gradualproduction of differences between the descendants of the sameancestral types." (See also Bloom, 1937.)

t Extensive treatment of the whole subject of the role of the nucleicacids is given in two recent symposia (1946, 1947) and in the reviewsof Greenstein (1944) and Hevesey (1948b).

* The Feulgen reaction (Feulgen and Rossenbeck, 1924) is generallyheld to be a specific cytochemical test for desoxyribonucleic acid(Dodson, 1946; Stowell, 1946), although this has been questioned(Carr, 1945; Stedman and Stedman, 1947). The chemistry of thereaction has been investigated by Staeey and others (1946).

t The view that histones are important nucleolar constituents hasrecently been challenged (Pollister and Ris, 1947; see also Stedmanand Stedman, 1947).

t Spiegelman and Kamen (1947) have warned against necessarilyaccepting the conception of a particularly close association betweenthe formation of protein and the metabolism of nucleic acids.

§ Hinshelwood (1946) also reviews the conditions under whichproteins are synthesized.

15



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. -DAC1E AND- J. C. WHITE

LEGENDS, PLATES I TO III

PLATE I3 ,U SECTIONS OF ASPIRATED STERNAL MARROW

FIG. 1.-Normal adult man. Iron haematoxylin andeosin, x 580.

FIG. 2.-Aplastic anaemia in man, with severe normo-cytic anaemia and thrombocytopenia purpura.Marrow hypoplastic, with many fat spaces and fewcellular areas. Iron haematoxylin and eosin, x 280.

FIG. 3.-Idiopathic hypochromic anaemia in female.General hyperplasia, with numerous erythroblasts.Delafield's haematoxylin and Giemsa, x 580.

FIG. 4.-Addisonian pernicious anaemia. Very hyper-plastic marrow. Iron haematoxylin and eosin, x 580.

FIG. 5.-Subacute combined degeneration. No anaemia.Marrow of normal structure. Gordon and Sweet'smodification of Wilder's silver impregnation, to showargyrophil reticulin fibrils outlining sinusoids andfat spaces and permeating stroma, x 250.

FIG. 6.-Addisonian pernicious anaemia. Hyperplastic,megaloblastic marrow; stained by silver impregna-tion to show outlines of dilated sinusoids and increasein reticulin of stroma, x250.

PLATE II

FIGS. 1 AND 2.-Bone marrow from two patients withsevere Addisonian pernicious anaemia. Many primi-tive promegaloblasts are present. Erythrocyte countapproximately 1,000,000 per c.mm.

FIG. 3.-Typical megaloblasts from the bone marrowof a patient with Addisonian pernicious anaemia.Erythrocyte count 1,500,000 per c.mm.

Holmes, 1946). The results of this work are very similarto those of Caspersson. Brachet (1946) considers,however, that chromatin, especially the heterochromatin,*contains a proportion of ribonucleic acid (see alsoKauffman and others, 1948) and that the nucleus is ofminor importance in relation to the synthesis of cyto-plasmic protein compared with the "cytoplasmicparticles " (mitochondria and microsomes) with whichthe ribonucleoproteins are largely associated (Bensley,1942; Claude, 1943; Brachet, 1946).

The cytochemistry of the developing bone marrowcells has been studied by several authors. Thorell(1947b, etc.) has developed Caspersson's techniquesof ultra-violet absorption microspectrophotometry.Brachet (1942) has applied his ribonuclease test tovarious amphibian and mammalian bone marrows.White (1947) has extended Brachet's work to normaland pathological human marrow cells, and J. N.Davidson and others (1948) have used the ribo-nuclease test as well as chemical microanalysis ofmarrow nucleic acids. Dustin (1946, etc.) has em-

Heterochromatin (Heitz, 1929) is chromatin closely associatedwith the nucleolus and is thought to possess important metabolicfunctions (see Schultz, 1947). It is probably identical with thenucleolus-associated chromatin of Casperson, which Thorell (1947b)believes may regulate the growth ofmarrow cells, as it does ofcelluargrowth in general.

FIG. 4.-Intermediate megaloblasts from the bonemarrow of a patient with mild Addisonian perniciousanaemia. Erythrocyte count 3,400,000 per c.mm.

FIG. 5.-Intermediate megaloblasts from the bonemarrow of a patient with steatorrhoea. Erythrocytecount 4,300,000 per c.mm.

FIG. 6.-A primitive marrow from a patient withtropical nutritional anaemia. Erythrocyte count1,690,000 per c.mm.

FIG. 7.-The megaloblastic bone marrow of a patientwith pernicious anaemia of pregnancy. Erythrocytecount 2,800,000 per c.mm.

All the preparations in Plates II and III were madefrom bone marrow aspirated from the sternum. Stainedby May-Grunwald-Giemsa, x800.

PLATE IIIFIG. 1.-Normoblasts from the bone marrow of a

normal adult man.

FIG. 2.-Micronormoblasts from ani iron-deficient bonemarrow.

FIG. 3.-Macronormoblasts from the bone. marrow ofa patient with a macrocytic haemolytic anaemia(nocturnal haemoglobinuria).

FIG. 4.-Abnormal megaloblasts from the bone marrowof a patient with pernicious anaemia of pregnancy.

FIG. 5.-Megaloblast-like cells from the bone marrowof a patient with leukanaemia. (We are indebted toDr. Sheila Newstead for this film.)

FIG. 6.-A megaloblast-like cell amongst normoblastsin a bone marrow film from a patient with carcino-matosis of the bone marrow.

ployed ribonuclease in his researches on reticulocytesand Wislocki and Dempsey (1946) have subjectedsamples of Rhesus monkey marrow to ribonucleasein addition to other histochemical techniques.

Thorell's work, recently summarized in a mono-graph (Thorell, 1947b), deserves further considera-tion. He worked with the marrow of rats andrabbits, and with normal and pathological humanmarrow. Individual living cells were subjected toanalysis at different growth stages and the concen-trations and proportions of nucleic acids, proteins,and haemoglobin were determined by means oftheir characteristic absorption maxima at particularwavelengths of the ultra-violet. *

Thorell's work has provided examples of the moregeneral views of Caspersson. The primitive pro-normoblasts contain the maximum amount (5 percent approximately) of ribose polynucleotides inthe cytoplasm and nucleoli. Desoxyribonucleic acidis confined to the nuclear chromatin and nucleolus-associated chromatin. The concentration of ribose

* The nucleic acids give a maximum extinction with light of wave-length 257 m,p. Proteins were estimated by absorption maxima at280 mp, attributable to their content of tyrosine and tryptophane(Holiday, 1936), and hnoglobin was estimated by absorptionmaxima at 404.7 Tnp and 435.8 mni (the Soret band).

16



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

. ,t- IT,

I..I'/I

.

Fig. 1.

Fig. 3. Fig. 4.

Fig. 5.PLATE I.

B

.

.I.

St* 1 to

Fig. 2.

17

..

I., .." *14

.,, 41.I

.IA



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

-- -

s.

J<;o::

e <

_.

:: K .- r v

.: . sqv5

_hi_::rod';s_>

,,,j=,,3, a

*.'_9f

_B__

__i.§ako .o.j!U11 _!F e.. < /,.^: ................. w.,

*,.* ".

_'Fig. 1.

_s-' £ .#!g_ _ t.

E :

FE_ sEi:4

s-be^---

1S

Fig. 4.

SFig. 3.

Fig. 5.

..---..w~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

._.

'14,

.i

Fig. 6.

:.:.,. _^

Fig. 2.

S .4@

O...

-A

PLATE II.

Fig. 7.

w. t

i.

.Ab

9

l.".4



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

e

Fig. 1.

0

'Ur

eS-f

.-

9

0

Fig. 2.

I_- !:I

00a

Fig. 3.

*, ...

. A

.e:. _ _

1-

,A

.-It

Fig. 4.

Fig. 5.

PLATE 1II.

Fig. 6.

4 0 m

At' ,14

't 1; " ill,,, 0I*k .-;Z"



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

V o

*.N

i,~~~~~~~~~~?

jr..

di

Fig. 2.

Fig. 4.

FIGS. 1-4.-Bone marrow obtained by sternalpuncture biopsy from patients who died ofsevere refractory anaemia. In Fig. 1 themarrow cells are particularly large and primi-tile. Fig. 2 shows maturing megaloblasts aswell as primitive cells, some of them megalo-blasts; also cells with pyknotic nuclei anddegenerating cytoplasm. Fig. 4 shows primi-tive cells and degenerating cells with basophiliccytoplasm.

FIG. 5.-Megaloblasts from the bone, marrow ofa patient with Addisonian pernicious anaemia,stained by Sudan black by the method ofBaker (1945)- The mitochondria are chieflyfound in the perinuclear zone. Note thedeeply stained granulocytes at the top of thefigure.

Fig. 1.

Fig. 3.

." ..

{ E . '.

pr:rs

Fig. 5.

PLATE IV.

..N. 4X..W; "-

.1.

.s

PI



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

polynucleotides in the cytoplasm diminishes moder-ately in partially differentiated cells but markedlyso in the nucleolus. By the time the cytoplasm ispolychromatic the concentration of ribonucleic- acidhas decreased to one-tenth of its maximum valueand has completely disappeared in orthochromaticcells. This work, therefore, supports the impressionthat ribonucleic acid concentration is paralleled bythe depth of cytoplasmic basophilia.Absorption attributable to substances of por-

phyrin type was not observed in the cytoplasm ofthe most primitive cells; the earliest cells in whichit was first detected were of moderate maturity withpolychromatic cytoplasm, at a concentration corre-

sponding to 3 ,iuCg. of haemoglobin per cell. Thehaemoglobin content increases to 28 ,p,vg. in themost ripened cells with fully pyknotic nuclei, at a

concentration of 20 to 25 per cent. The final risein concentration of haemoglobin to 33 per cent isattributed by Thorell to shrinkage of the cell whenit changes from a rounded form as a normoblastto a disc-like erythrocyte.On the basis of his cytochemical analysis Thorell

refers to four phases of activity in the developmentof the normoblast: (1) a phase of growth wherethe concentration of ribonucleic acid is maximal,(2) a phase of declining growth where the ribonucleicacids are diminishing and basic cellular proteins likeglobin are being synthesized, (3) a phase of differen-tiation when haemoglobin is being formed rapidly,and (4) a phase of declining differentiation.Although the globin necessary for haemoglobin

synthesis is probably formed at an early stage in thelife of the erythroblast (through participation ofribose polynucleotides as for protein synthesisgenerally) the formation of haem compounds andtheir coupling with globin appears to take place ata later stage, after the ribopolynucleotides havesubstantially disappeared (see also Shemin andRittenberg, 1946).The complexity of the apparatus and methods

used by the Caspersson school is a disadvantage.Recently E. M. Jope (1949) and Barer and others(1949) have outlined a simplified technique using thereflecting microscope (Burch, 1947). By this methodsimultaneous monochromatic cross-sections of thecell structures may be studied (Text Fig. 3). Theintracorpuscular haemoglobin of individual erythro-cytes produces the same intensity of absorption atthe Soret band as was expected from studies ofhaemoglobin in simple solution.

Formation of haemoglobin.-Knowledge of themechanism of haemoglobin synthesis is still very

incomplete. Duesberg in 1938 published a goodreview of the position up to that time, and Bur-

c

mester (1937), in a chemical study ofpigment forma-tion in avian erythroblasts, included figures for theiron content of cells at different stages of develop-ment. The precursors of the porphyrin ring havefor long remained mysterious, but recently Sheminand Rittenberg (1946) have shown that in rat andman glycine labelled with N15-isotope is built intoprotoporphyrin and ultimately appears in thehaemoglobin of the mature erythrocytes. A similarsynthesis has been demonstrated in vitro when duckerythrocytes and erythrocytes from patients withsickle-cell anaemia were incubated with N15-con-taining glycine (London and others, 1948). Glycinecontaining C'4-isotope in the ac-position has alsobeen used in similar experiments; it is incorporatedinto both haem and globin moieties of haemoglobin(Altman and others, 1948).

Thorell's work, already referred to, indicates therelatively early diminution in cytoplasmic ribopoly-nucleotides in polychromatic erythroblasts beforethe start of haemoglobin formation. H. M. Jope(1948) points out that at this stage the cytoplasmicprotein is probably related to the globin of haemo-globin, and that, later, haemoglobin itself is elabor-ated from this globin, which either forms haemgroups on its surface or combines with them aftertheir formation. The cytoplasmic acidophilia ofthe early polychromatic normoblast is thus due toits content of globin rather than of haemoglobin,which is developed later.Siderocytes.-An interesting abnormality of the distri-

bution of iron in erythrocytes and their precursors hasrecently been described. Gruneberg (1941, 1942)observed siderocytes, or erythrocytes with granules givingthe Prussian-blue reaction for iron in the blood ofnormal rat, mouse, and human embryos, and in largenumbers in mice with a congenital anaemia. Later theywere observed in adult human blood by Doniach andothers (1943), and further details have been provided byPappenheimer and others (1945), Case (1946), McFadzeanand Davis (1947), and Dacie and Doniach (1947).Siderotic granules are to be found also in polychromaticand pyknotic normoblasts, usually as a few isolatedgranules, but sometimes confined to a zone around thenuclear membrane (Dacie and Doniach, 1947).

Siderotic granules of the type described by Griinebergseem not to be formed in damaged or aged erythrocytesas has been claimed by Case* (1945), but seem rather tobe produced at the same time as haemoglobin is elabor-ated. Possibly a proportion of the iron which normallyenters the haem molecule is concentrated in loci in thecytoplasm of the developing normoblast in a relativelyfree form. It is noteworthy that when many granules

* Case (1946) has used either aa'-dipyridyl-thiocyanate or acid-potassium ferrocyanide after a preliminary unmasking treatment withammonium sulphide to demonstrate siderocytes. By these techniques,siderocytes (Case) are present in normal blood and increase innumbers in stored blood. Case believes siderotic granules indicatean ageing cell.

21



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

2 A CW

J. V. -IDACIE AND J. C. WAITE

are present the rest of the cytoplasm of the cell stainspalely with acid dyes. When stained with basic dyes thesiderotic granules appear as basophilic granules (Pappen-heimer bodies). The incidence of siderocytes in humanpathology has not yet been fully worked out. They arenot present in the peripheral blood of normal adults, butconstantly appear in small numbers after splenectomy;in certain haemolytic anaemias almost all the corpusclesmay be affected.

Abnormal ErythropoiesisMegaloblastic erytlropoiesis. - The term

"megaloblast" was used by Ehclich (1880) todescribe the type of nucleated erythroblast found inthe bone marrow of patients with perniciousanaemia in relapse. Since that time the terms" megaloblast " and " megaloblastic erythropoiesis"have been widely used in descriptive haematology,but their correct usage and limits, and their signi-ficance, have proved to be a controversial problem.Morphology of the megaloblasts.-Megaloblasts

differ from normoblasts in several ways; they arelarger cells with increased cytoplasm and nuclearsize and at every stage in development they have adifferent and more open nuclear chromatin pattern,and a tendency in the later types for the cytoplasmto manufacture haemoglobin at stages when thechromatin of the nucleus is still ananged in an openmanner. We should like to emphasize severalgeneral points before describing the cells themselves.First, that the "megaloblastic" change may beappreciated at all stages in development from themost primitive types to the most mature ortho-chromatic megaloblasts with pyknosis of thenucleus; secondly, that the degree of abnormalityvaries greatly between cells of the same apparentage and that all grades of change may be seenbetween grossly abnormal cells through intermediatetypes to cells which are almost if not quite normal;and thirdly, that " megaloblasts " are not normallypresent in the marrow. However, although theyare essentially pathological cells, megaloblasts arenot a race of cells apart but a type of abnormality*affecting cell growth and maturation, which developsto a greater or lesser extent as the result of deficien-cies in growth factors of which the anti-perniciousanaemia factor is an important example.These points will now be considered in more

detail. The earliest megaloblasts, the promegalo-blasts (Ferrata and Negr iros-Rinaldi, 1914), arelarger cells than the pronormoblasts, which are theirnormal counterparts. In size, they range from 20 to

Erythrocyte precursors are not the only developing cells affectedby deficiency of tho anti-pernicious anSiemia factor. Abnormallylarge and atypical granulocytes are quite as characteristic, but theirdescription is beyond the scope of this article. As in the megaloblastsa characteristic feature is an asynchronism between the growth anddevelopment of the nucleus and the cytoplasm.

25 t± in diameter in stained dried films. The cytoplasmis deeply basophilic and relatively abundant, withnumerous mitochondria. The chromatin of thenucleus is arranged in a characteristically welldefined finely stippled manner, and one or morebasophilic nucleoli can be identified. From thesebasophilic promegaloblasts a series of ripeningmegaloblasts develop. These cells remain largerthan normal throughout the ripening process, andthe cytoplasmic outline is usually less round thanin the normal series. When the cytoplasm attainsa polychromatic staining reaction, this is often farin advance of that expected on the basis of nuclearstructure, and mitochondria are still numerous.Thus polychromatic megaloblasts with small nuc-leoli are often observed, and the cytoplasm may evenbecome orthochromatic before pyknosis of thenucleus has been completed. As the cells maturethe nuclei become smaller and the chromatin morecondensed; nevertheless, the chromatin pattemremains more open than in developing normoblasts.The chromatin takes the form of short strands andsmall nodes, but there is sufficient " parachromatin "to give an overall mottled effect. The nucleolus-associated chromatin remains finer and morediscrete from the rest of the chromatin than innormal development. As the cel matures furtherthe chromatin strands thicken and the nodes enlargeand tend to fuse, but complete homogeneous pyk-nosis is rarely seen. The nucleus is at first evenlyrounded or slightly oval in shape, but often becomesdistorted and has an irregular outline before it isfully pyknotic. The mitochondria finally disappear.Abnormal mitosis and nuclear fragments (Howell-Jolly bodies*) are quite frequently seen. Thedifference between cells developing normally andthose with well marked megaloblastic changes sillustrated in Figs. 1-8 and 22-29 of the ColouredPlate and in Text Fig. 1.

Intermediate megaloblasts.-The above descrip-tion and the illustrations apply to megaloblasts asseen in well stained smears from patients withsevere Addisonian pernicious anaemia. The degreeof the abnormalities and the extent to which thecells deviate from the normal are closely linked withthe severity of the lack of haemopoietic factors.This is illustrated in the photomicrographs ofPlate II and the Coloured Plate, and in the TextFig. 1, in which are shown a series of scale drawings

* Howell-Jolly bodies are small round Feulgen-positive bodies,approximately 0.5 to 2 I in size. They are derived from nuclearchromatin and may be seen in polychromatic and orthochromaticerythroblasts. It is not certain whether they are produced by nuclearkaryorrhexis or are formed from chromnosomes, which have beenisolated during mitosis. They are found in many types of anaemiaand are especially frequent in Addisonian pernicious ahaemia, insevere haemolytic anaemias, in idiopathic steatorrhoea (Engelt 1939),and after splenectomy (Singer and others, 1941).

22



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

of megaloblasts and a series of normoblasts forcomparison. The megaloblasts in the four photo-micrographs of marrow films (Plate II, Figs. 1-4)differ not only in size but in the extent to whichthey deviate from normal cells. These four patientsA, B, C, and D had anaemia of varying severity;A and B were the most severe cases, with approxi-mately 1,000,000 erythrocytes per c.mm.; case Cwas less severe, with 1,800,000 erythrocytes perc.mm.; and case D the least anaemic with 3,400,000erythrocytes per c.mm. It is evident that not onlyis the degree of megaloblastic change intimatelylinked with the seriousness of the lack of haemo-poietic factors and that it is hence proportional tothe severity of the anaemia in the patient, but thatthe relative proportion of primitive to maturingmegaloblasts is also affected. It is in the severelyanaemic patient that the primitive cells are mostabundant, and in the mildly anaemic the leastabundant. Bilaki-Pasquier (1948) refers to stainedmarrow containing many primitive megaloblasts as" moelle bleue," and notes that the primitive cellsare most numerous when the peripheral erythrocytecount is lowest. See also Leitner (Leitner andothers, 1949d).

The nature of tho megaloblastic change.-Megalo-blasts are produced when there is a deficiency ofcertain haemopoietic factors, which normally arenecessary for effective growth and division and fornormal differentiation. These factors may resembleenzymes or coenzymes in their action, and onlyminute amounts may be required; in the case of therecently purified liver factor (Lester Smith, 1948),it has been calculated that I ug. per day is sufficient(Rickes and others, 1948). * The effect ofa deficiencyof the liver factor is that haemopoiesis becomes dis-ordered -and ineffective. The exact details of howthis is brought about are still obscure; certainlymitosis is deranged to some extent (Japa, 1945),and may occasionally be multipolar or incomplete,with variation of chromosome numbers in thedaughter cells (La Cour, 1944). There seems to beno deficiency in the synthesis of ribonucleic acid as

judged by the amount of cytoplasmic and nucleolarbasophilia removable by ribonuclease (White, 1947),or by microanalysis (J. N. Davidson and others,.1948)--indeed, the content of both types of nucleicacid in the marrow is increased. The derangement is

* It is outside the scope ot this ardicle to consider in detail whetherVitamin B,, is but one of several factors, absence of which will caus-megaloblastic change. It is probable that more than one factor isconcerned and that the normal chain of processes can be broken atseveral places. It is generally held that in patients with th- spruesyndrome, in certain relatively refractory megaloblastic anaemias, inpernicious anaemia of pregnancy, and in nutritional megalocyticanaemia, liver alone. even in large doses, may not be wholly effective(Wills, 1948). it may be added that pteroylglutamic acid is notexclusively concerned with erythropoiesis, but is a growth factor in awide biological sense.

a complex one affecting botltgrowth and differentia-tion and has important effects on cell size, chromatinarrangement, and cytoplasmic ripening, and typi-cally results in an asynchronism between nuclearand cytoplasmic maturation, a point recentlyemphasized by Bessis (1946, 1948). It has alreadybeen mentioned that similar changes may beobserved in the developing leucocytes. Thorell(1947b) has studied the distribution of ribose poly-nucleotides and of haemoglobin in the megaloblastsofmarrow from two patients with pernicious anaemiaby the micro-spectrophotometric method. Hefound haemoglobin present in cells that still possessedthe same high concentration of cytoplasmic andnucleolar ribose polynucleotides that normallycharacterize only- the more primitive e ythrocyteprecursors. Use of the ribonuclease test shows,however, that in those megaloblasts which actuallydevelop to erythrocytes there is a steady decreaseof both cytoplasmic and nucleolar ribonucleic acidcontent. In view of this, it seems possible thatThorell has omitted to analyse the most matureforms of megaloblasts.The effects on erythropoiesis of a deficiency of the

liver principle are mainly twofold: primitive pro-megaloblasts are present in far larger proportionthan are primitive cells in normal marrow, and maytotal 50 per cent or more of the erythroblasts. Inaddition there is a variable proportion of maturingabnormal cells (megaloblasts) from which evolvethe megalocytes and poikilocytes of the peripheralblood. In addition to primitive cells which definitelyhave the characters of promegaloblasts, there arealso found increased numbers of undifferentiatedreticulum cells and transitional stages between themost primitive cells and promegaloblasts and myelo-blasts, such cells corresponding to the haemocyto-blasts of normal marrows.As has been mentioned, it is in the most severe

cases that the higher proportion of primitive cellsis found, and it seems clear that in these patientsthere is a severe impediment to successful cellulardifferentiation. It is not so clear what happens tothe primitive cells if they fail to mature; in certainof our marrows it is possible to trace stages inshrinkage of cell size and pyknosis of the nucleus,without cytoplasmic ripening, until a small indeter-minate remnant results. In the less severely affectedmarrows the proportion of primitive cells is less,and degenerating forms are inconspicuous; manymore maturing megaloblasts are present, and it isobvious that whilst erythropoiesis is abnormal andrelatively ineffective, it is at least proceeding. Thisis borne out by the fact that in these cases theperipheral blood count may remain in equilibrium

23



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

DA?2CIE At.IAD iTE

for long periods at a moderate level, or at least thatit only falls slowly.

The variabk nature of megaloblastic marrow.-Inthe preceding sections on megaloblastic erythro-poiesis we have attempted to describe the mor-phology of megaloblasts and their causation, andhave developed the thesis that all grades of changemay be recognized between extremely abnormalcells and cells almost indistinguishable from thenormal. In this section we shall present -furtherevidence as to the variability of the megaloblastsand of the marrow picture as a whole in patientswith Addisonian pernicious anaemia, and willindicate our views as to the abnormal types oferythropoiesis met with in the sprue syndrome and,in pernicious anaemia of pregnancy. In the follow-ing section we shall compare our own viewpointwith that of other workers.

It is well known that in Addisonian perniciousanaemia the volume of red marrow increases andextends out into the long bones of the skeleton andthat there is a diminution or complete loss of thefat cells normally so abundant in the marrow of anadult (see Plate I), and the same is true to a greateror lesser extent in the other allied megalocyticanaemias. Sternal puncture usually provides verycellular material in which marrow particles arereadily seen. If differential cell counts are per-formed an increased erythrogenetic-leucogeneticratio will be found (Schartum-Hansen, 1937). Thisis illustrated from our material in Text Fig. 4;there is a general correlation between the severity ofthe anaemia and the predominance oferythrogeneticcells, but the picture is complicated because of theassociated disturbance in leucopoiesis. Thesemarrows are in fact not as one-sidedly erythrogeneticas in chronic haemolytic anaemias with similardegrees of anaemia. In Text Fig. 4 are also shownthe results of plotting the degree of anaemia of thepatient (a rough indication of the severity of thelack of haenopoietic factors) against the proportionof primitive nucleolated cells with basophilic cyto-plasm* amongst the whole population of nucleatederythrogenetic cells. There is an obvious positivecorrelation between the degree of anaemia and theproportion of primitive cells, and, as has beenmentioned before, the megloblasts become lessand less abnormal and more " intermediate " panpassu with the falling proportion of primitive cellsand reduction in an (Plate UI).Of the thirty-five patients whose data are plotted

in Text Fig. 4, nine were suffering from non-

*Reticulum cells and slightly more mature types corresponding tohaemocytoblasts which we could not definitely identify as erythrocyteprecursors were not counted.

Addisonian megaloqytic anaemias, and it is parti-cularly interesting to note that the data from thesecases fall into line with those obtained from themore numerous patients with typical Addisoniananaemia. In the two severely anaemic patients,one an example of tropical nutritional megalocyticanaemia and the other a patient'with.perniciousanaemia of pregnancy, there was a high proportionof primitive cells and well marked megaloblasticchange; in the less' anaemic patients there werefewer primitive cells and the megaloblasts weremuch more intermediate in type. As far as ourexperience goes the megaloblastic picture of thenon-Addisonian megalocytic anaemias is similar tothat of typical untreated Addisonian perniciousanaemia of comparable severity.

Effect on megaloblastic bone marrow of treatment withthe liver anti-anaemic principle.-It is well known thatthe bone marrow of a typical case of Addisonian perni-cious anaemia rapidly loses its megaloblastic characterand becomes normoblastic if effective doses of liver aregiven. The exact details and mechanism of this changehave been a matter of some controversy (Jones, 1943;Limarzi and Levinson, 1943; Leitner and others, 1949e).It is probable that the ripening megaloblasts completetheir transformation into megalocytes if they have lostthe capacity for further divisions, but that the primitivenucleolated cells respond to the presence of the liverhaemopoietic factor by undergoing normal divisions anddifferentiating as normoblasts.The earliest changes are appreciable within twenty-four

hours of an injection of liver extract, when basophilicpronormoblasts and normoblasts at an early stage ofdevelopment may be seen in addition to ripening megalo-blasts. During the next forty-eight hours many moreripening polychromatic normoblasts appear and themegaloblasts are less conspicuous, but the latter- do notentirely disappear for at least five days, even if the doseof liver produces a maximal response.* If the liverextract is not fully potent a very mixed picture results,"intermediate " megaloblasts replacing to a greater or lessextent the expected normoblasts. The completeness orotherwise of the transformation- towards normoblasticformation (and the disappearance of abnormal leuco-cytes) is, naturally, the most delicate index of theadequacy of the liver extract.A discussion on the megaloblast problem.-The

preceding sections have been a presentation of ourown point of view, based on our own material. Apoint of great controversy is the question of " inter-mediate" forms of megaloblasts; whether theyoccur at all, what is their significance, and what istheir relationship to macronormoblasts. Theconception of " internediate " forms is by no means

* The change from an inhibited marrow with a high proportion ofprmite cells to an activdy maturing one is acenoompanied by string reductions in the content of both rbnucleic anddesoxyrihonucleic acids (J. N. Davidson and others, 1948).

24-



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

new, but relatively few authors have paid muchattention to it. In pre-war Continental literaturethese forms were referred to under a variety ofnames, as "macroblasts" by Schartum-Hansen(1937), as "megalo-normoblasts" by Hotz andRohr (1938) in cases of sprue, and as "intermediateerythroblasts" by Lambin and de Weert (1938,1939). They have been described by Trowell (1942-3)in tropical dimorphic anaemia and by Zuelzer andothers (1947) in megaloblastic anaemia in infancy.Bessis (1946, 1948) in his review of the megaloblastquestion admits their existence, and Davidson andothers (1947) refer to their occurrence in twopatients with sprue, and they have been describedas " atypical normoblasts " by Cooke and others(1948) in cases of idiopathic steatorrhoea, and as

"intermediate erythroblasts" by Innes (1948) insprue. Our own views have been set out in twoprevious preliminary communications (Dacie andWhite, 1947; Dacie, 1948). Other workers havenot considered the problem at all or have decidedagainst the existence of intermediate forms (Jones,1943), and have stressed the striking differencesbetween normoblastic erythropoiesis and gross

megaloblastic change rather than the similaritieswhich may be seen between the two series (Israels,1939; Leitner and others, 1949f). Most authors,however, including Dameshek and Valentine (1937)and Jones (1938), admit that mixed pictures occur

and that a variable proportion of normoblasts maybe found in a predominantly megaloblasticmarrow.

Not only are we convinced that intermediateforms occur, but we believe that they have a specialsignificance; they appear when the marrow suffersminor deficiencies ofhaemopoietic principles. Theyare seen, in our experience, in mild Addisoniananaemia before liver treatment, in more severe

examples of Addisonian anaemia that have receivedsuboptimal doses of liver, and in cases of the sprue

syndrome and other megalocytic anaemias, whereanaemia is relatively mild and response to liver

unsatisfactory; presumably in these cases some

substance other than the purified liver principle isrequired, or utilization by the marrow is impaired.We feel that these types of cell deserve a widerrecognition and should be referred to as "inter-mediate megaloblasts" rather than as atypical or

intermediate erythroblasts or normoblasts, becausetheir pathogenesis seems to be the same as that ofmore typical megaloblasts. The difference ismerely-quantitative and the cells are less definitelyabnormal. The question of their relation tomacronormoblasts and heteroplastic-erythroblastsis discussed in later sections.

Other Types of Abnormal ErfthropoIeIMacronormoblastic erythropoiesis.-The term

"macronormoblastic erythropoiesis " has beenemployed by Jones (1943) to describe the developingerythroblasts in conditions where there is a peri-pheral macrocytosis not dependent upon a deficiencyof the liver principle. This type of blood picture isseen in some rhronic haemolytic anaemias (Dame-shek and Schwartz, 1940), in liver disease (BodleyScott, 1939), recovery from haemorrhage (Win-trobe, 1946d; Leitner and others, 1949g), and inpatients responding to iron therapy where therehas been iron deficiency. It is also found in latefoetal life. The term macronormoblast refers tothe nucleated precursors of the peripheral macro-cytes. Jones, unfortunately, includes an increasedproportion of " early forms (pronormoblasts andbasophilic normoblasts)" in his conception of amacronormoblastic marrow. To us it seemspreferable to use the term solely to describe normo-blasts which are larger in size than their normalcounterparts of similar age, and not to complicatematters by referring to an increased proportion ofearlier types which, even if perfectly normal cells,would increase the average size of the normoblasts.Few authors seem, however, to have demonstratedthe abnormal size of "macronormoblasts" bydirect measurement. Dameshek and Schwartz(1940) measured the diameters of the erythroblastsin the marrow film of a patient with acute haemo-lytic anaemia and found the cells to be from 1 to1.3 i. larger than normal cells of the same age group.We have also made some measurenments, and con-trasted the maturing polychromatic and pyknoticnormoblasts in normal subjects with those frompatients with macrocytic haemolytic anaemia andiron deficiency anaemia (Text Fig. 5). The resultsof such measurements as these are, however, opento question, as an increased proportion of moreprimitive cells produces an apparent macrocytosis,and it is very difficult to measure only cells of strictlycomparable age groups. Our figures show, however,that whereas in the five normal marrows 11 to 26per cent of the cells measure 8 V. or less, in themacronormoblastic marrow only 2 per cent of thecells are within this range and in the iron-deficientmarrow there are as many as 70 per cent. As thesesmall cells are all mature or almost mature types itseems reasonably certain that real differences exist(see also Plate III, Figs. 1-3).Do macronormoblasts differ from normoblasts

in any respect other than size? Their nuclearstructure resembles that of normoblasts and thusdiffers from that of intermediate megaloblasts.Israebs (1941) has stated that premature haemoglo-

25



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

2. V. DACIE AND J. C. WHITE

binization of tht cytoplasm of erythroblasts (pre-sumably macronormoblasts) often occurs in hyper-plastic marrows with normal maturation. Thischange seems to us to be a slight one and unreliableas a diagnostic aid. We often experience somedifficulty in the recognition of individual macronor-moblasts, but believe that this is only to be expectedbecause both macronormoblasts and intermediatemegaloblasts merge into the normal. The generaltrend of both abnormal types is, however, quitedistinctive if the abnormalities are marked.

Little is known as to why macronormoblastsdevelop or about the mechanism of their formation.In most instances they are found when haemopoiesisis proceeding rapidly and when there is a raisedproportion of reticulocytes in the peripheral blood.Possibly in a hyperplastic marrow unripened cellsare forced into the blood stream because the activegrowth of the marrow as a whole outstrips chematuration of individual cells. Abnormal size andoccasionally the persistence of a nucleus, as well ascytoplasmic basophilia, may thus all be signs ofimmaturity. An alternative hypothesis is that largeripened normoblasts are produced as the result of adiminished number of cell divisions through whichthey have been derived from the pronormoblasts.The differences between the abnonnal and normalare relatively slight, however, and it seems unlikely-that a stimulated, actively erythrogenetic marrow

-J-JLU

doI'D 10z

5

would respond to the demand for more erythrocytesby reducing the number of cell divisions.Another hypothetical possibility is that there is

normally some humoral mechanism which controlsthe development and maturation of the erythro-blasts, similar to that described as controlling thematuration of reticulocytes (Jacobsen, 1947), andthat in conditions of rapid erythropoiesis there maybe a relative deficiency of ripening factors. At themoment all that can be said is that none of thesehypotheses has been proved.

Erythropoiesis in iron deficiency (Micronormo:blastic erythropoiesis).-The effect ofiron deficiencyon-the erythrocytes of the peripheral blood is wellrecognized, but rather less is known about themorphology of the erythrocyte precursors in themarrow, and as to how deficiency of iron results indisordered erythropoiesis. Bodley Scott (1939) hasreviewed the early literature and described his ownfindings in twenty-three patients. He found anincreased cellularity of the marrow and an increasedproportion of erythrogenetic cells roughly parallelto the severity of the anaemia. He described thepredominant cells as polychromatic normoblasts" with an irregular and jagged cell outline and onlya small rim of slate-grey cytoplasm around thepyknotic nucleus." Since Bodley Scott's paper thedisordered erythropoiesis of iron deficiency hasreceived wider recognition (Wintrobe, 1946e; Whitby

6 7 8 9 10 I 1 12 13 14DIAMETERS IN p

TEXT FIG. 5.-Diameter distribution curves of 100 polychromatic and pyknotic normoblasts, measured insternal marrow ifims from five normal subjects x-x, a patient with a macrocytic haemolytic anaemia

nfo, and from a patient with an iron deficiency anaemia o, o. (For interpretation, see text.)

26



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

and Britton, 1946), and might well be referred to as"micronormoblastic" in type (Plate III, Fig. 2, andText Fig. 1). The haemocytoblasts and pro-erythro-blasts in iron deficiency seem normal in size andappearance and are usually present in a normal orslightly increased relative proportion, but there isusually an increase in the numbers of basophilicnormoblasts (Leitner and others,1949h). Our ownobservations and measurements (Text Fig. 5)confirm the small size of the more mature normo-blasts and are in agreement with Bodley Scott'sdescription. Cytoplasmic ripening seems to lagbehind nuclear condensation, and fully ortho-chromatic cells are not seen.The presence of a hyperplastic erythrogenetic

marrow in association with peripheral anaemia hasled to the view that there is an arrest of maturationsimilar to that found in Addisonian perniciousanaemia. However, there is little obvious mor-phological evidence for this; not only are morenormoblasts present than in the normal, but a highproportion seem to be maturing. It might bethought that the malformed hypochromic erythro-cytes and poikilocytes are unduly sensitive to thenormal process of wear and tear and that thiscontributes significantly to the anaemia, which is ineffect partly haemolytic* in type, and that thepresence of so many normoblasts indicated a rapidcell production. However, the level of reticulocytesin the peripheral blood in a case ofuntreated anaemiadue to iron deficiency is rarely above 2 or 3 per cent,and the serum bilirubin is low. The fact thateffective iron therapy leads to an outpouring ofreticulocytes is in favour of the previous existenceof some sort of retardation of development. It ispossible that the effect of a lack of a sufficient levelof plasma iron is a failure of the last stages ofnormoblast differentiation in addition to inadequatesynthesis of haem, and that the ragged-borderedpolychromatic but almost pyknotic normoblastsare dying cells, the most severely affected of whichnever become erythrocytes at all. The reduced sizeof these normoblasts is presumably directly due tothe reduced synthesis of haemoglobin, resulting in-decreased globin formations from its precursors,and hence a small cytoplasmic mass.

Erythropoiesis in refractory anaemia.-The refrac-tory anaemias (Bomford and Rhoads, 1941) aredyshaemopoietic anaemias often referred to by thedescriptive titles aplastic, hypoplastic, or pseudo-aplastic, or as panmyelophthisic anaemias. Forconvenience we are also including "refractory

* It is possible that this mechanism may contribute to the anaemiaof Mediterranean anaemia (Cooley's anaemia). The hypochromicerythrocytcs are extremely deformed, and there is evidence suggestingincreased haemolysis (Dameshek, 1943).

megaloblastic " and achrestic anaemias. The wholegroup is probably heterogeneous, but has in commona resistance, more or less complete, to any treatmentnow known. It is because we do not think thatthe descriptive titles mentioned above necessarilyindicate distinct-entities that we hesitate to discardthe more general term " refractory anaemia." Asa rule leucopenia or thrombocytopenia accompaniesthe anaemia, which is normocytic or macrocytic intype.The only aspect of this group that we are now

considering is the cytology of erythropoiesis. ITherehave been several important studies in recent years.Israels and Wilkinson (1940) described as sufferingfrom " achrestic anaemia " six patients in whom thebiopsied marrow was found to be megaloblastic ormixed megaloblastic and normoblastic. Bomfordand Rhoads (1941), basing their account on sec-tioned biopsy or autopsy material, reported themarrow to be partly mature and cellular, orhypoplastic, or immature and cellular; andDavidson and others (1943) in a smaller series ofcases reported the marrow, as studied by sternalpuncture, to be hypocellular and normoblastic, orhypercellular and megaloblastic. Other patients inwhom the marrow appearances greatly varied aredescribed by Leitner and others (1949i).In our own material we have observed great variation

in the type of erythropoiesis. This variability is illus-trated by the details of the following seven patients. Inthe first case there was a primitive marrow containingnumerous haemocytoblasts and increased numbers ofreticulum cells, with but few ripening cells, mostly inter-mediate megaloblasts, and a normocytic peripheralpicture (Plate III, Fig. 4). In the second patient themarrow was extremely primitive, with some differentia-tion to typical megaloblasts, and a megalocytic peri-pheral blood picture (Plate III, Fig. 1). In the thirdpatient the marrow was hyperplastic; primitive cellswere present in moderately large numbers, and therewas a moderate proportion of differentiating cells, somebeing megaloblasts. In addition, degenerating cells withpyknosis of the nuclei but without cytoplasmic ripeningwere conspicuous. The peripheral blood picture was ofa megalocytic anaemia (Plate III, Fig. 3). In the fourthpatient the marrow, although hypoplastic at autopsy, wasshown during life by sternal puncture to be primitive,but to contain differentiating cells definitely megaloblasticin type (Plate III, Fig. 2). In the marrow of the fifthpatient found to be hypoplastic at autopsy, there was inlife, on the other hand, very little evidence of a megalo-blastic change; most of the nucleated erythroblasts werenormoblasts, a few were intermediate megaloblasts. Thesixth patient, less anaemic than the foregoing, had amoderately hyperplastic and partly primitive marrow;once again the differentiating cells were mostly megalo-blasts of intermediate type. In the seventh patient themarrow at biopsy appeared to be hyperplastic but not

27



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

J. V. bACIE -AND J. C. -WHTE

primitive; eyQthrooii wa normoblastic and normalexcept for some macronormoblasts with almost pyknoticbut distorted nuclei. Anaemia was, however, severe andslightly macrocytic in type.

It is difficult to understand the m nism ofanaemia in those patients who have a hyperplasticand differentiating marrow (pseudo-aplastic anae-mia). Presumably in these cases the final matura-tion of the normoblasts is unusually slow or theripe cells may die before reaching the circulation.Possibly also, in some cases, hyperplastic areasalternate with areas of hypoplasia. In the patientswhose marrows are predominately primitive intype, the mechanism of the failure to deliveradequate numbers of erythrocytes into the peri-pheral circulation is probably similar to that whichoperates in Addisonian pernicious anaemia. Theprimitive cells fail, to a greater or less extent, tocomplete their differentiation, and may be seen toundergo pyknosis of the nucleus and to degeneratewithout differentiating at all. Of the maturing cellsa variable proportion have characters more or lesstypical of megaloblasts. As with our own series ofpatients with typical Addisonian anaemia, the moreseriously anaemic the patient the more inhibitedand primitive the marrow is likely to be.Where aplasia seems to follow a proliferating

marrow of primitive type, this change may well bedue to an intensification of the factors causing theanaemia, whether they be " toxins," lack of growthfactors, or the presence of a growth antagonist.This change from hyperplasia to hypoplasia of themarrow was observed by Bomford and Rhoads(1941) in some of their patients. Marrow cellu-larity, diminishing as the disea progresed, seemedto be a characteristic feature of the group of patientswhose marrows were shown to be hypercellular and" partly mature " at biopsy; in five patients,however, the change was from a "partly mature"marrow to an immature one.

Erythropoiesis in haemolytic anaemia.-Increasedhaernolysis leads almost invariably to increased.erythrepoiesis,* and, in mazrow biopsy smears ofpatients with an active haemolytic anaemia, erythro-poietic cells are usually extremely numerous. Thetype- of erythropoiesis has almost always beendescribed as normoblastic (Leitner and others,1949j). Turnbull (1936c) has, however, identifiedmegaloblasts in addition to normoblasts in post-*mortem sections of marrows from several patientswith acholuric jaundice who died of haemolyticcrises, and he has suggested that in these cases

* Own (1948) has demonstrated tt hypoplasia of the erqto.marrow cells may be a cause of severe cries m consenitalytic anaemia; lack of mauration of primitive erythroblasts

may aLso be a cause (Dameshek and Bloom, 1948).

there ma-y have been a cy of haemopoieticfactors. We have not yet encountered typicalmegaloblasts in narrow smears from our ownpatients; the normoblasts, however, tend to be alittle larger than normal and may thus be describedas macronormoblasts. This increase in size is mosteasy to recognize in the most mature cells. Inaddition to this we have noted, in patients wherehaemolysis is severe, that complete nuclear pyknosisis infrequent and that the nucleus of the almostmature cells with polychromatic cytoplasm may beslightly more " open " than in normal normoblasts,in relation to the acidophilia of the cytoplasm.Occasionally, the most mature nuclei may be irre-gular in shape. The whole picture is, however,unmistakably normoblastic rather than megalo-blastic.

Erythropoiesis in leukaemia and allied disorders.-The association of anaemia with leukaemia isalmost invariable, except in the earliest stages. Inthe peripheral blood the presence of undue aniso-cytosis and poikilocytosis and of erythroblastssuggests that erythropoiesis itself is disordered.Occasionally the disorder of erythropoiesis isdominant and the leucocyte abnormalities subsidiaryor even subsequent; such cases, in which a severemacrocytic anaemia with evidence of abnormalerythropoiesis precedes the development of indubit-able leukaemia, have been referred to as "len-mia" (Leube, 1900; Forkner, 1938; Foy andothers, 1946). In additio'n to this variety of leukae-mia there are patients in whom the proliferativeprocess predominately and persistently affectserythrogenetic cells. This rare disorder was firstdescribed by -di Guglielmo* as acute erythraemiaor erythraemic myelosis. " Leukanaemia " seemsto occupy an intermediate position between ery-thraemia and leukaemia.

In myelogenous, lymphogenous, and monocyticleukaemia, erythropoiesis is predominately normo-blastic, but a small proportion of cells simulatingintermediate megaloblasts may be found; occa-sionally, and often towards the termination of theillness, these cells are present in higher proportions."Megaloblasts" have been reported more fre-quently in cases of leukanaemia (Penati, 1937;Foy and others, 1946; Collins and Rose, 1948). Inacute erythraemia, di Guglielmo has described as"paraerythroblasts " the atypical cells which arefrequently encountered, and whose origin in somecases may be traced from the proliferating reticulumcells (haemohistioblasts). Atypical mitoses andirregular and multilobed nuclei are not infrequently

In di Guglielmo's (1946) review in French there is a full biblio-graphy of his earlier papers.

28



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

seen in more mature forms of these cells, and theremay also be unusually early acidophilia of the cyto-plasm. In general, however, the type of erythro-poiesis has been described as normoblastic, althoughamongst the " paraerythroblasts " there are somecells which closely resemble megaloblasts; thesehave been called "megaloblastiform" cells byHeilmeyer and Schoener (1941), and this term hasbeen used by di Guglielmo himself and by Bessis(1948). Presumably also the patient described byDowney (1938b) as suffering from leukaemicreticulo-endotheliosis, in which megaloblast-likecells were observed to be developing heterotopicallyfrom reticulum cells, and the first patient describedin Schleicher's (1944b) paper represent examples ofa further rare variant of the leukaemia-erythraemiagroup.We have observed cells rather similar to inter-

mediate megaloblasts in leukaemia (Plate IlI,Fig. 5) and in erythraemia. These cells are ratherlarger than normal for their apparent maturity,and the acidophilia of the cytoplasm is in advanceof the form of the nucleus. In general, we believethat this change towards a " megaloblastiform"type of erythropoiesis cannot be influenced by liver

or folic acid therapy. In all probability it indicatesmore than a deprivation of growth factors by therapid growth of leukaemic tissues, and is rather a

reflection of the disordered growth caused by the" leukaemic " stimulus itself affecting erythropoiesis,and resulting in changes in morphology not unlikethose caused by minor deficiencies in the liverhaemopoietic principle (see also Schwarz, 1946).

Erythropoiesis in carcinomatosis of the bonemarrow.-The leuco-erythroblastic anaemia of car-

cinomatosis of the bone marrow has been well

described by Vaughan (1936a and b) and Turnbull(1936d). They have reported the presence of megalo-blasts ofEhrlich with premature ripening of the cyto-plasm in the bone marrow and in the peripheralblood. BothVaughan and Turnbull remark,however,on the small size of the megaloblasts; in Turnbull'sdescription of the marrow findings " most are

relatively small megaloblasts with scanty cytoplasmand some are not appreciably larger than normo-

blasts. . . Erythropoiesis is, therefore, partlymegaloblastic, partly normoblastic, the normoblasticusually predominating." Bessis (1946) has alsoreferred to intermediate types of "erythroblasts"when the bone marrow is invaded by metastaticcarcinoma. In our view it is uncertain whetherthese abnormal cells should be considered as

macronormoblasts or megaloblasts; further studyis required. In a recent case erythropoiesis was

predominantly normoblastic, but isolated cells

resembling intermediate megaloblasts could befound (Plate III, Fig. 6). Possibly both types ofreaction occur. The cause -of these changes isobscure. Vaughan concluded that the growingtumour cells interfered with the metabolism of theerythrogenetic cells; this may well be the case.

Cellular gigantism ii. numan erythropoiesis.-Theoccurrence of cellular gigantism has recently beenreviewed by Schwarz (1946) and by Berman (1947).Binucleated "erythroblasts " may be found in bothnormal and abnormal bone marrows and they probablyarise by endomitosis.* Schwarz (1946) discusses atlength the abnormalities in mitosis which give rise tothese multinucleated cells and describes the very largeerythrocytes (gigantocytes), 18-25 ,u in size, to whichthey give rise. He stresses the fact that no intermediarylinks are found between normal cells and gigantocytesand their multinucleated precursors. He does notconsider that these giant cells are ever produced byamitosis. Berman found in eight normal subjects from1 to 5.1 binucleated cells per 1,000 normal erythroblasts.Multinucleated erythroblasts produced by multipolarmitosis are, however, not found in health; not oneexample was seen by Berman out of 53,167 erythroblastsexamined in his eight normal marrows. In disease,however, they occur not infrequently. Limarzi andLevinson (1943) have described three types of multinu-cleated erythroblasts: the first is produced as the resultof multipolar mitosis without cytoplasmic separation;the second arises by folding and lobulation of thenucleus and finally by separation of the nuclear frag-ments; and the third by complete or incomplete amitosis-a mechanism which has been questioned (Schwarz).The resulting giant cells are well illustrated by Limarziand Levinson (1943), Schleicher (1944b), Schwarz (1946),and Berman (1947).

Multinucleated erythroblasts have been observed in avariety of blood disorders; in Berman's series, in casesof leukaemia, lymphoblastoma, pernicious anaemia,congenital haemolytic anaemia, thrombocytopenic pur-pura, and liver disease. Schleicher (1944b) reported largemultinucleated cells in a patient with reticulum-celledsarcoma (reticulo'endotheliosis) of the bone marrow andin pernicious anaemia. Limarzi and Levinson's case wasconsidered to be an example of erythroblastoma, andwe have recently encountered a good example in a patientwith pernicious anaemia of pregnancy (Plate III,Fig. 4).Berman has stressed the fact that these giant cells

are not specific for any particular disease process, thatthe process may be reversible, as in pernicious anaemia,and that the nuclear and cytoplasrnic patterns conformto the type of the accompanying uninucleated cells.Thus multinucleated normoblasts or megaloblasts maybe found, or giant dysplastic types with some of thecharacters of megaloblasts, as in Linarzi and Levinson'scase.

* Endomitosis is complete nuclear division without division of thecytoplasm.

29



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/


Although, therefore, the exact cause of the aberrantnuclear divisions remains obscure, it is clear that theseprocesses do not necessarily indicate a malignant(" leukaemic ") process. The position seems to be verysimilar to that for the occurrence of giant or multinu-cleated cells in other sites throughout the body. In themarrow they are most frequent during active erythro-poiesis, as they are in other tissues during rapid growthphases (for example, in the testis of the normal mouseHoward, 1948).

Abnormalities of mature erythrocytes.-An almostconstant feature of anaemia is an increased variabilityin the size (anisocytosis) of the mature erythrocytes overand above the normal cell to cell variation. In addition,in most anaemias, corpuscles of abnormal shape andnot strictly round in contour (poikilocytes) are found.The normal variation in cell diameters has been wellworked out by Price-Jones (1933), who has shown thatit conforms to a normal biological distribution, andsimilar variations in volume and thickness, etc., alsoprobably exist. The cause of the exaggerated variationin anaemia is so far unsettled; anisocytosis and poikilo-cytosis are marked in dyshaemopoietic anaemias, parti-cularly in Addisonian pernicious anaemia in severerelapse, in Mediterranean anaemia, and in some examplesof leukaemia and myelosclerosis, and to a lesser extentin iron deficiency anaemia. In haemolytic anaemias andother anaemias with active regeneration, poikilocytosisis much less obvious, and variation in size is partlydue to the large size of the polychromatic corpuscles.Spherocytic microcytes* may also be present and mayadd to the range of cell diameters. Poikilocytosis maythus be taken to indicate disordered erythropoiesis, andthis is probably also true ofmarked anisocytosis, althoughin this case the production of macrocytes may be due torapid rather than to abnormal formation.To authors such as Plum (1947) and Bostrom (1948),

who believe that erythrocytes are produced by buddingfrom the cytoplasm of erythroblasts, the problem ofanisocytosis and poikilocytosis presents no difficulty.The budding process is held to be deranged so that budsof different sizes are formed. Bostrom considers thatpoikilocytes are produced by cytoplasmic budding takingplace through pathologically thickened walls of medullarysinusoids, and schematically illustrates this. However, asalready mentioned, these unusual conceptions are noteasily acceptable. Nevertheless, it is possible that thesmall poikilocytes of pernicious anaemia are cytoplasmicfragments (schistocytes: see Rous, 1928), but whetherthey are formed in the marrow (Habelmann, 1940) orrepresent fragment} broken off larger cells in the peri-pheral blood stream is uncertain. In general we feelthat tge solution of the problem of the formation ofpoikilocytes depends on an understanding of the forces

* "Microspherocytosis" is a change which takes place after theerythrocytes have been delivered from the bone marrow. Themarrow normoblasts are not abnormal; neither, as a rule, are thereticulocytes. The same is true of elliptical corpuscles (see Leitnerand others, 1949k) and of sickle cels (Wintrobe, 1946f ). The antithesisof spherocytosis target-cell formation seems, in Mediterraneananaemia at least, to depend upon an abnormality of formation.

which convert the cytoplasm* surrounding the nucleus ofthbe normoblast into the disc-like form of the reticulocyte.It is easy to imagine that when erythropoiesis is abnormalthis process is deranged and that imperfect erythrocytesresult. Anisocytosis is likely to be a reflection of varia-tions in the size of the normoblasts of the same apparentage. That this variation exists is illustrated in the normo-blast diameter distribution curves (Text Fig. 5) and inPlate III, Figs. 1-3. Very large erythrocytes (giganto-cytes) are occasionally encountered. They arise frommultinucleated erythroblasts produced by abnormalmitosis (Schwarz, 1946).

Conclusions

In this review we have attempted a survey oferythropoiesis in man. The subject has manyramifications and a vast literature. While men-tioning, therefore, some publications of historicalinterest, we have mostly referred to modern work.As our aim has been to present as wide apicture of erythropoiesis as possible, we have notconfined ourselves to morphology and cytogenesis;we have included sections on the growth anddifferentiation of the erythropoietic tissue and adiscussion of the cytochemical aspects of erythro-poiesis, for not only is the mature erythrocyte aperfect example of the differentiation of a cell for ahighly specialized function, but the sequence ofchanges accompanying development affords goodexamples of the processes of tissue growth as awhole.The general pattern of erythropoiesis has been

described in health and in disease. In health, abone marrow biopsy gives a momentary but stillpicture of what is really a carefully regulated dyna-mic process, but even here much of the mechanismof normal erythropoiesis is ill-understood. Indisease, the pattern of erythropoiesis may be greatlyaltered; accelerated, abortive, and aberrant forma-tion may all be seen, sometimes in combination.Here, too, the changes are dynamic. All grades ofchange may be seen between the most abnormalbone marrows and a marrow scarcely distinguish-able from the normal.Bone marrow biopsy provides an almost unique

opportunity for the examination of living humantissue. It is an indispensable adjunct to the properunderstanding of the diseases of the blood.

We are indebted to our medical colleagues for freedomto investigate patients under their care, to Mr. E. V.Wilhmott, F.R.P.S., for the photomicrographs, to ourlaboratory and clerical staff for their assistance, and toother friends for helpful discussions.

* Pronormoblasts and ripening normoblasts are probably almostspherical cells. As the nucleus shrinks and the cytoplasm becomesmore abundant relatively, the cell is roughly elliptical in veriaacross-section. V

30



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

BONE MARROW BIOPSY

REFERENCES

Altman, K. I., Casarett, G. W., Masters, R. E., Noonan, T. R., andSalomon, K. (1948). J. biol. Chem., 176, 319.

Arinkin, M. I. (1929). Folia haemat., Lpz., 38, 233.Baar, H. S., and Lloyd, T. W. (1943). Arch. Dis. Childh., 18, 1.Baker, J. R. (1945). Quart. J. micr. Sci., 85. 1.Barer, R., Holiday, E. R., and Jope, E. M. (1949). To be published.Baserga, A. (1939). Haematologica, 20, 933.Bensley, R. R. (1942). Science, 98, 389.Berk, L., Burchenal, J. H., Wood, T., and Castle, W. B. (1948).

Proc. Soo. exp. Biol, N.Y., 69, 316.Berman, L. (1947). J. Lab. cin. Med., 32, 793.Berman, L., and Axelrod, A. R. (1947). Amer. J. clin. Path., 17, 61.Bessis, M. (1946). Rev. Hdnat., 1, 45.Bessis, M. (1948). " Cytologie Sanguine." Masson, Paris.Bilaki-Pasquier, G. (1948). Ann. med., 49, 417.Bloom, W. (1937). Physiol. Rev., 17, 589.Bloom, W. (1938a). In Downey's "Handbook of Haematology."

New York, Hoeber, vol. 2, p. 908.Bloom, W. (1938b). In Downey's "Handbook of Haematology."

New York, Hoeber, vol. 2, pp. 891, 1367.Bloom, W., and Bartelmez, G. W. (1940). Amer. J. Anat., 67, 21.Bodley Scott, R. (1939). Quart. J. Med., n.s., 8, 127.Bomford, R. R. (1938). Quart J. Med., n.s., 7, 495.Bomford, R. R., and Rhoads, C. P. (1941). Quart. J. Med., n.s., 10,

175.

Bosto.n, L. (1948). Acta med. scand., 131, 303.Brachet, J. (1942). Arch. Biol., Paris, 53, 207.Brachet, J. (1946). In " Symposia of the Society for Quantitative

Biology." No. 1. Nucleic Acid. Cambridge University Press,p. 207.

Brachet, J., and Shaver, J. R. (1948). Stain Tech., 23, 177.Bunting, C. H. (1906). J. exp. Med., 8, 625.Burch, C. R. (1947). Proc. phys. Soc., 59, 41.Burmester, B. R. (1937). Folia haemat., Lpz., 56, 372.Cappell, D. F., Hutchison, H. E., and Smith, G. H. (1947). Brit.

med. J., 1, 403.Carr, J. G. (1945). Nature, Lond., 156 143Cartwright, G. E. (1947). Blood, 2, 111, 256.Case, R. A. M. (1945). J. Path. Bact., 57, 271.Case, R. A. M. (1946). Proc. roy. Soc. B., 133, 235.Caspersson, T. (1946). In "Symposia of the Society for Experimental


Caspersson, T., and Santesson, L. (1942). Acta radiol., Stockh.,Suppl. No. 46.

Caspersson, T., and Schultz, J. (1939). Nature, Lond., 143, 602.Caspersson, T., and Schultz, J. (1940). Proc. nat. Acad. Sci., Wash.,

26, 507.Catchside, D. G., and Holmes, B. (1946). In " Symposia of the

Society for Quantitative Biology." No. 1. Nucleic Acid.Cambridge University Press, p. 225.

Claude, A. (1943). Biol. Svmposia, 10, 111.Collins, D. H., and Rose, W. McI. (1948). J. Path. Bact., 60, 63.Cooke, W. E. (1930). Brit. med. J., 1, 433.Cooke, W. T., Frazer, A. C., Peeney, A. L. P., Sammons, H. G., and

Thompson, M. D. (1948). Quart. J. Med., 17, 9.Cotti, L., Balestrieri, F., and Volta, M. (1938). G. clin. med.,-19, 935.Cowdry, E. V. (1942). In " Problems of Ageing" (Ed. Cowdry),

Second Edit. Williams and Williams, Baltimore, p. 626.Crafts, R. C. (1946). Amer. J. Anat., 79, 267.Custer, R. P. (1933). Amer. J. med. Sci., 185, 617.Custer, R. P., and Ahlfeldt, F. E. (1932). J. Lab. clin. Med., 17, 960.Dacie, J. V. (1948). Brit. rnied. J., 1, 702.Dacie, J. V., and Doniach, I. (1947). J. Path. Bact., 59, 684.Dacie, J. V., and White, J. C. (1947). Lancet, 1, 614.Dameshek, W. (1943). Amer. J. med. Sci., 205, 643.Dameshek, W. (1948). Blood, 3, 209.Dameshek, W., and Bloom, M. L. (1948). Blood, 3, 1381.Dameshek, W., and Estren, S. (1947). "The Spleen and Hyper-

splenism." Grune and Stratton, New York.Dameshek, W., Henstell, H. H., and Valentine, E. H. (1937). Ann.

intern. Med., 11, 801.Dameshek, W., and Schwartz, S. 0. (1940). Medicine, Baltimore, 19,

231.

Dameshek, W., and Valentine, E. H. (1937). Arch. Path., 23, 159.Daughaday, W. H., Williams, R. H., and Daland, G. A. (1948).

Blood, 3, 1342.Davidson, J. N., Leslie, I., and White, J. C. (1948). J. Path. Bact.,

60, 1.

Davidson, L. S. P. (1930). Edinb. med. J., 37, 425.Davidson, L. S. P. (1941). Edinb. med. J., 48, 678.Davidson. L. S. P., Davis, L. J., and Innes, J. (1943). Edinb. med. J..

50. 355, 431.Davidson, L. S. P., Girdwood, R. H., and Innis, E. M. (1947). Lancet,

511.

Discombe, G. (1946). Nature, Lond., 157, 370.Diwany, M. (1940). Arch. Dis. Childh., 15, 159.Doan, C. A. (1923). Proc. Soc exp Biol., N. Y., 20, 260.

Doan, C. A., Cunningham, R. A., and Sabin, F. R. (1925). Contr.Embryol. Carneg. Instn., No. 361, 16, 163.

Doan, C. A., and Wright, C. S. (1946). Blood, 1, 10.Dodson, E. 0. (1946). Stain Tech., 21, 103.Doniach, I., Gru'seberg, H., and Pearson, J. E. G. (1943). J. Path.

Bact., 55, 23.Downey, H. (1938a). " Handbook of Haematology." Hoeber, New

York, vol. 3, p. 1963.Downey, H. (1938b). " Handbook of Haematology." Hoeber, New

York, vol. 2, p. 1321.Drinker, C. K., Drinker, K. R., and Lund, C. C. (1922). Amer. J.

Phvsiol., 62. 1.

Duesberg, R. (1938). Klin. Wschr., 17, 1353.Dustin, P., Jr. (1944). Arch. Biol., Paris, 55, 285.Dustin, P., Jr. (1946). In " Symposia of the Society for Experimental


Ehrlich, P. (1880). In Ehrlich, P., and Lazarus, A. " Die Anaemie.Abt. 1, pp. 36 and 38, Vienna, 1880. Abt. 2, p. 156, Vienna, 1900.

Engel, A. (1939). Nord. Med., 1, 388.Ferrata, A., and de Negreiros-Rinaldi (1914). VirchoKs Arch., 215,

77.Ferrata, A., and Storti, E. (1948). " Le Malattie de Sangue." Societa

Editrice Libraria. Milano. 1st reprinting. (a) p. 40. (b) p. 78and plate opp. p. 64. (c) p. 38. (d) p. 42.

Feulgen, R., and Rossenbeck, H. (1924). Hoppe-Seyl. Z., 135, 203.Fieschi, A. (1938). Haematologica, 19, 539.Fieschi, A., and Astaldi, G. (1946). " La Cultura in Vitro del Midollo

Osseo." Pavia.Finkelstein, G., Gordon, A. S., and Charipper, H. A. (1944). Endo-

crinology, 35, 267.Forkner, C. E. (1938). "Leukemia and Allied Disorders." Mac-

millan, New York, p. 161.Foy, H., Kondi, A., and Murray, J. F. (1946). J. Path. Bact., 58, 157.Gilmour, J. R. (1941). J. Path. Bact., 52, 25.Grant, W. C. (1948). Amer. J. Physiol.,-153, 521.Grant, W. C., and Root, W. S. (1947). Amer. J. Physiol., 150, 618.Greenstein, J. P. (1944). In " Advances in Protein Chemistry."

Academic Press. Vol. 1, p. 210. New York.Grfiieberg, H. (1941). Nature, Lond., 148, 114, 469.Grilneberg,H. (1942). J. Genet., 44, 246.Guglielmo, G. di (1946). Rev. Heinat., 1, 355.Habelmann, G. (1940). Klin. Wschr., 19, 1134.Hamre, C. J. (1947). J. Lab. clin. Med., 32, 756.Heath, C. W., and Daland, G. A. (1930). Arch. intern. Med., 46, 533.Heilmeyer, L., and Schoener, W. (1941). Dtsch. Arch. klin. Med.,

187, 225.Heitz, E. (1929). Ber. dtsch. bot. Ges., 47, 274.Helly, K. (1910). Beitr. path. Anat., 49, 15.Hevesey, G. (1948a). "Radioactive Indicators." Interscience Pub.

New York, pp. 479-481.Hevesey, G. (1948b). In "Advances in Biological and Medical

Physics." Vol. 1. Academic Press. Nest York, p. 409.Hevesey, G., and Ottesen, J. (1945). Nature, Lond., 156, 534.Hinshelwood, C. N. (1946). " The Chemical Kinetics of the Bacterial

Cell." Oxford University Press. London and New York,

pp. 15-22.Hoch, H., Jope, E. M., and Jope, H. M. (1949). To be published.Holiday, E. R. (1936). Biochem. J., 30, 1795.Hotz, H. W., and Rohr, K. (1938). Ergebn. inn. Med. Kinderheilk., 54,

174.Howard, A. (1948). Personal communication.Howell, W. H. (1890). J. Morph., 4, 57.Hurtado, A., Merino, C., and Delgado, E. (1945). Arch. intern.

Med., 75, 284.Innes, E. M. (1948). Edinb. med. J., 55, 282-292.

Isaacs,R.(1930). Foliahaemat.,Lpz.,40,395.Isaacs, R. (1937a). Amer. J. med. Sci., 193, 181.Isaacs, R. (1937b). Physiol. Rev., 17, 291.Isre6ts, M. C. G. (1939). J. Path. Bact., 49, 231.Isreals, M. C. G. (1941). J. Path. Bact., 52, 361.Isra8ls. M. C. G., and Wilkinson, J. W. (1940). Quart. J. Med., n.s. 9,

33, 163.Jacobsen, E. (1947). J. clin. Path., 1, 19.

Japa, J. (1942). Brit. J. exp. Path., 23, 272.

Japa, J. (1945). Brit. J. exp. Path., 26, 111..Jones, O. P. (1938). In Downey's "Handbook of Haematology."

New York, vol. 3, 2045.Jones, 0. P. (1943). Arch. Path., 35, 752.Jones, 0. P. (1947). J. Lab. clin. Med., 3?, 700.Jones, 0. P. (1948). Blood, 3, 967.

Jope, E. M. (1949). In the press.Jope, H. M. (1948). Thesis, Oxford.Kato, K. (1937). Amer. J. Dis. Childh., 54, 209.Kauffman, B. P., McDonald, M., and Gay, H. (1948). Nature,

Lond., 162, 814.Key, J. A. (1921). Arch. intern. Med., 28, 511.

Kirschbaum, A. (1937). Proc. Soc. exp. Biol., N.Y., 35, 542.Kracke, R. R. (1941). " Diseases of the Blood." Lippincott,

Philadelphia. Second Edit. p. 49.

3.1



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/


La Cour, L. F. (1944). Proc. roy. Soc. Edinb. (B)., 62, 73.Lambin, P., and de Weerdt, W. (1938). Rev. belge. Sci. mid., 10, 282.Lambin, P., and de Weerdt, W. (1939). Sang, 13, 928.Leitner, S. J. (1941). Folia haemat., Lpz., 65, 1.

Leitner, S. J., Britton, C. J. C., and Neumark, E. (1949). "BoneMarrow Biopsy." Churchill, London. (a) p. 22. (b) p. 20.(c) pp. 18-19. (d) p. 72. (e) p. 74. (f) p. 25-26. (g) p. 113.(h) p. 100. (i) p. 255-269. (j) pp. 152-153. (k) p. 159.

Lester-Smith, E. (1948). Nature, Lond., 161, 638. 676.Leube, W. von (1900). Berl. klin. Wschr., 37, 851.Limarzi, L. R. (1947). J. Lab. clin. Med., 32, 732.Liznarzi, L. R., and Levinson, S. A. (1943). Arch. Path., 36, 127.Limarzi, L. R., Levinson, S. A., and Jones, R. M. (1942). J. Amer.

med. Ass., 118, 1004.Loge, J. P. (1948). Blood, 3, 198.London, I. M., Shemin, D., and Rittenberg, D. (1948). J. biol.

Chem., 173, 797.McCarty, M. (1946). J. gen. Physiol., 29, 123.McCullagh, E. P., and Jones, R. (1942). J. clin. Endocrinol., 2. 243.McFadzean, A. J. S., and Davis, L. J. (1947). Glasg. med. J., 28, 237.Markoff, N. (1938). Dtsch. Arch. klin. Med., 183, 289.Maximov, A. (1899). Arch. Anat. Physiol., Lpz., Anat. Abt., p. 33.Maximov, A. (1910). Arch. mikr. Anat., 76, 1.Maximov, A. (1927). In Mollendorff, " Handbuch der Mikr. Anat.

d. Mensch." 2 Bnd. I TI, Berlin, Springer. pp. 232, 381, 404,528.

Maximov, A., and Bloom, W. (1948). " A Textbook of Histology."W. B. Saunders. Philadelphia and London. Fifth Edit., p. 89.

Mertens, Elizabeth (1945). Amer. J. med. Sci., 210, 630.Miller, E. B., Singer, K., and Dameshek, W. (1942). Proc. Soc. exp.

Biol., N. Y., 49, 42.Mollison, P. L. (1948). ILncet, 1, 513.Naegeli, O. (1931). "Blutkrankheiten and Blutdiagnostik." Berlin,

Springer. Fifth Edit., pp. 98, 100, 101.Neumann, E. (1868). Zbl. med. Wiss., 6, 689.Nizet, A. (1946). Acta med. scand., 124, 590.Osgood, E. E., and Ashworth, C. M. (1937). "Atlas of Haema-

tology." J. W. Stacey, Inc., San Francisco.Osgood, E. E., and Seaman, A. J. (1944). Physiol. Rev., 24, 46.Owren, P. A. (1948). Blood, 3, 231.Pappenheimer, A. M., Thompson, W. P., Parker, D. D., and Smith,

K. E. (1945). Quart J. Med., n.s., 14, 75.Penati, F. (1937). Minerva med., Torino, 2, 401.Pizzolato, P., and Stasney, J. (1947). J. Lab. cliii. Med., 32, 741.Plum, C. M. (1942). Acta physiol. scand., 4, 259.Plum, C. M. (1947). Blood. Special Issue (1), 33.Pollister, A. W., and Ris, H. (1947). In " Cold Spring Harbor

Symposia on Quantitative Biology." vol. 12, p. 147.Ponder, E. (1945). Science, 102, 257.Ponder, E. (1948). " Haemolysis and Related Phenomena." Grune

and Stratton, New York, p. 155.Pontoni, L. (1936). Haematologica, 17, 833.Price-Jones, C. (1933). " Red Blood Cell Diameters." Oxford

University Press. London.Rastelli, M. (1943). " La Puntura Sternale." Edizioni Italiane.

Rome. (a) pp. 11-38. (b) p. 66. (c) pp. 53-64, (d) pp. 67-83.Rebuck, J. W., and Woods, Helen L. (1948). Blood, 3, 175.Reich, C., Swirsky, M., and Smith, D. (1944). J. Lab. clin. Med., 29,

508.Reinhard, E. H., Moore, C. V., Dubach, R., and Wade, L. J. (1944).

J. cliiz. Inivest., 23, 682.Reisner, E. H. (1943). Arch. intern. Med., 71, 230.Rickes, E. L., Brink, N. G., Koniuszy, F. R., Wood, T. R., and

Folkers, K. (1948). Science, 107, 396, 397, 398.Rohr, K. (1940). Schweiz. med. Wschr., 70, 685.Rohr, K., and Hafter, E. (1937). Folia haemat., Lpz., 58, 38.Rosin, A., and Rachmilewitz, M. (1948). Blood, 3, 165.Rous, P. (1928). Physiol. Rev., 3, 75.Rubinstein, M. A. (1948). J. Amer. med. Ass., 137, 1281.Sabin, F. R. (1928). Physiol. Rev., 8, 191.Salah, M. (1934). J. Egypt. med. Ass., 17, 846.Sanders, F. K. (1946). Quart. J. micr. Sci., 87, 203.Schartum-Hansen, H. (1937). Folia haemal., Lpz., 58, 145.Schilling-Torgau, V. (1912). Folia haemat., Lpz., 14, 129.Schleicher, E. M. (1944a). Amer. J. clin. Path., 14, 370.Schleicher, E. M. (1944b). J. Lab. clin. Med., 29, 127.Schleicher, E. M. (1945). J. Lab. clin. Med., 30, 928.Schulten, H. (1937). Folia haemat., Lpz., 58, 189.

Schultz, J. (1947). In " Cold Spring Harbor Symposia for Quasntita-tive Biology." No. 12, p. 179.

Schultze, M. 0. (1940). Physlol. Rev., 20, 37.Schwarz, E. (1946). Amer. J. Anat., 79, 75.

Segerdahl, E. (1935). Acta med. scand., supp. 64, 1.

Segerdahl, E. (1941). Acta med. scand., 108, 483.

Seyfarth, C. (1923). Dtsch. med. Wschr., 49, 180.Shemin, D., and Rittenberg, D. (1946). J. biol. Chem., 166, 627.

Singer, K., Miller, E. B., and Dameshek, W. (1941). Amer. J. med.

Sci., 202, 171.

Snapper, I., Groen, J., Hunter, D., and Witts, L. J. (1937). Quart. J.

Med., n.s., 6, 195.Spiegelman, S., and Kamen, M. D. (1947). In " Cold Spring Harbor

Symposia on Quantitative Biology." No. 12, p. 211.Stacey, M., Deriaz, R. E., Teece, E. G., and Wiggins, L. F. (1946).

Nature, Lond., 157, 740.Stasney, J., and McCord, W. M. (1942). Proc. Soc. exp. Biol., N.Y.,

51, 340.Stasney, J., and Pizzolato, P. (1942). Proc. Soc. exp. Biol., N.Y., 51,

335.

Stedman, E., and Stedman, E. (1947). In " Cold Spring Harbor

Symposia on Quantitative Biology." Vol. 12: " Nucleic Acidsand Nucleoproteins." p. 224.

Storti, E. (1937). Haematologica, 18, 1.

Stowell, R. E. (1946). Stain Tech., 21, 137.

Symposia of the Society for Experimental Biology. No. I: " NucleicAcid" (1946). Cambridge University Press.

Symposia (1947). Cold Spring Harbor Symposia on QuantitativeBiology. Vol. 12: " Nucleic Acids and Nucleoproteins."New York: Biological Laboratory.

Thorell, B. (1947a). In " Cold Spring Harbor Symposia on Quanti-tative Biology." Vol. 12: " Nucleic Acids and Nucleoproteins."New York. p. 247.

Thorell, B. (1947b). " Studies on the Formation of Cellular Sub-

stances during Blood-cell Production." Acta med. scand., 129,

suppl. 200, and published by H. Kimpton, London.

Tinsley, J., and Moore, C. (1948). Communicated to International

Society of Haematology. Buffalo, New York.

Trowell, H. C. (1942-3). Trans. R. Soc. trop. Med. Hyg., 36, 15 1.

Tiurkel, H., and Bethel, F. H. (1943). J. Lab. clin. Med., 28, 1246.

Turnbull, H. M. (1936). In "The Anaemias." By J. N. Vaughan.Oxford University Press, London. Second Edit. (a) pp. 15-18.

(b) p. 18. (c) pp. 243-246. (d) p. 169.Turnbull, L. H. (1948). Bull. Inst. med. Lab. Tech., 14, 68.

Ungricht, M. (1938). Folia haemat., Lpz., 60, 145.

Vaughan, J. M. (1936a). J. Path. Bact., 42, 541.

Vaughan, J. M. (1936b). "The Anaemias." Oxford Univer-sity Press.

Second Edit., p. 167.

Vaughan, S. L., and Brockmyre, F. (1947). Blood. Special Issue, 1,54.

Watkinson. G., McMen.-rrey, W. H., and Evans, G. (1947). Lanict,1, 631.

Watson, C. J., and Clarke, W. O. (1937). Proc. Soc. exp. Biol., .N.1.,36, 65.

Watson, C. J., Grinstein, M., and Hawkinson, V. (1944). J. cliii.

Invest., 23, 69.

Weisberger, A. S., and Heinle, R. W. (1948). Amer. J. toied. Sdi.,215,170.

Whitby, L. E. H., and Britton, C. J. C. (1946). " Disorders of theBlood." Fifth Edit., p. 188. Churchill, London.

White, J. C. (1947). J. Path. Bact., 59, 223.

White. J. C., Baker, J. R., and Griffin, J. G. (1946). J. Patli. Bact.,58, 155.

White, M. J. D. (1947). "The Chromosomes." Methuen's Mono-

graphs on Biological Subjects. Third Edit., p. 6. Methuen,London.

Wills, L. (1948). Blood, 3, 36.

Wilson, E. B. (1928). " The Cell in Development and Heredity."Third Edit., pp. 31, 59, 64, 70, 77. Macmillan, New York.

Wintrobe, M. M. (1946). " Clinical Haematology." Second Edit.

Kimpton, London. (a) pp. 412-414. (b) pp. 60-61. (c) p. 450.

(d) p. 434. (e) p. 539. (f) p. 513.

Wintrobe, M. M., and Shumacker, H. B., Jr. (1935). J. clin. Invest.,

14, 837.Wislocki, G. B., and Dempsey, E. W. (1946). Anat. Rec., 96, 249.

Young, L. E., and Lawrence, J. S. (1945). J. clin. Invest., 24, 554.

Zuelzer, W. W., Newhall, A., and Hutaff, L. (1947). J. Laob. clin.Med., 32, 1217.

32



j.com/

J Clin P


ownloaded from

http://jcp.bmj.com/

erythropoiesis with particular reference its study … · j. clin. path. .(1949), 2, 1....

Documents