Macrocytic Hyperchromic Anremia on the Witwatersrand

568 S.A. ME DIC AL JO U R NAL August 9, 1947 Macrocytic Hyperchromic Anremia on the Witwatersrand PART 1. By G. W. H. SCHEPERS, D.Sc., JOHANNESBURG. T HE "idiopathic" macrocytic anmi a s have lately gained considerable attention in hmatological literature; but despite much discussion, it is by no means certain that more than a few of the mysteries surrounding this di ease have been solved. It is therefore very necessary to continue the investigation of these anmias in the hope that more light may be shed not only on the nature of pernicious anmi a, but also on its less well understood etiology. As long, however, as a blood count is carried out only when a nmia is suspected, these important diseases will remain obscure, for the evidence will be biased and will represent a limited selection rather than a random sampling. To ignore the blood count in a single case is also tantamount to overlooking a whole physiologico-anatomical system of the body. The student is taught to examine the lungs and the heart stethoscopically even when no pulmonary or cardiac disease is suggested by the history. Unfortunately, it is not sufficiently emphaised how remunerative the blood count may be. The technical issue should only be a handicap where insufficient time is available, though a clinician who has made a habit of routine blood examinations realises how little time is really occupied in the procedure. There is considerable scope for the study of hmatologica l problems on the Witwatersrand. Its altitude is approximately 6,000 feet above sea-level, and many mine-workers daily pass up and down mine shafts of up to 11,000 feet in depth. The features of the "normal" blood count here have been exhaustively dealt with by Symons (1939), Stammers (1933), Liknaitzky (1935), Emmerson (1933), and Schepers (1947). In the present paper I propose to discuss only the an mi as of residents without mining service. Those men who are daily subjected to varying barometric conditions and simultaneously are exposed to silica dust, thereby slowly developing silicosis, constitute an entirely separate hm a tological problem. Before the effects of silica exposure and deep-level mining can be appreciated it will be necessary fi rst to become acquainted with the anmias of the non-minincy population at this altitude. T he subject constituting the case material of the present analysis were mostly members of Reef Benefit Society Panels. From a potential reservoir of approximately 50,000 individuals I discovered 300 cases of blood-picture variations in the direction of anmia macrocytosis or of hyperchromatosis. This gives an incidence rate of at least 0.6 per cent., which. is considerably in excess of that for pernicious a nmla. C RITERIA FOR DIAGNOSIS OF AN MIA. There are roug hly three features commonly accepted as evidence of anmia, viz. deficiency of erythrocytes ; diminution of hmoglobin; and the clinical syndrome. But even for these the standards vary much, in addition to the apparent variation from centre to centre of the criteria necessary for the diagnosis of a nmi a. Erythrocytes: The normal red cell count is taken to be 5.0 million ; but few physicians will diagnose anmia before the count in an individual case has fa llen below 4.0 million. Samson Wright (1942) has taken a more progressive view by suspecting an mia with counts of less than 4.5 million for males and less than 4.0 million for females. The argument seems to be that a deficiency of 0.5 million cells reduces the count to the female level only, and reduction by a further 0.5 million may be due to technical defects, or is within the range of statistical, theoretical or observed normality. This attitude is radically wrong. One may lose the greater part of one's stomach, kidneys, suprarenals or lungs, and still function "normally". These organs appear to have a " reserve " of functional tissue for compensatory adaptation. But the blood has very little reserve in individual cases, even though there is evidence of statistical variations. And the life of an erythrocyte being short, its retention in the circulating blood is not of much use for emergency purposes (besides those in the spleen). The blood must remain at a " normal " level to pr'ovide the necessary constant " breathing surface" which metabolism requires. All the blood cells in circulation at one time are used for this purpose, while only 1/ 25 of the liver, for instance, is normally in use at anyone time. The extent of the loss sustained when one million cells are short is only fully realised when the loss of total surface area is calculated. According to Lovatt Evans (1941), the total surface area of the blood in a man weighing 70 kilos may be about 3,000 square metres, or 1,500 times his body surface. A deficiency of one million cells may reduce this surface area by 600 square metres. No one would willingly allow his domestic premises to be encroached on to this extent without feeling suffocated. Fortunately there are other, more progressive, h matologists who stress the importance of correcting any deficiency, howeve r sm all. Murphy (1939) states that he has had most gratifying rsponses to treatment of very small deficiencies. Having had the same results, I can fully endorse his statement. One is not justified always in accepting 5.0 million cells per cubic millimetre as " normal", for though it may do for low-lying countries like England and parts of Western Europe where the altitude above sea-level is seldom more than 500 feet, it cannot apply universally. Lovatt Evans gives SA million as an average normal count (England); Price-J ones (193 1 ) obtained averages in London of 5.428 for males, and 5.012 for females ; Foster and Raymond J ohnson (193 1) reported 5.26 million cells as an

Augtlstus 9 1947 S.A. T Y DS K R IF VI R G ENEESKUN D E 1569 average for American men students. The weighted mean value for the red cell count, according to the values obtained by these and other authors, is 5.5 million per c.mm. (cf. Table I). If 5.0 million be taken as the minimum requirement in European countries of moderate altitude, more patients are likely t o obtain relief from distressing symptoms. But even with a count of 5.5 mijiion a man could be relatively" amemic " if his metabolism had previously been adjusted to, say, 6.0 or 6.5 million celis. Lovatt Evans gives 114 per cent. and 100 per cent. for males and females respectively (Haldane Standard); Price-Jones obtained 105.4 per cent. and 98 26 per cent (Haldane) for London. Foster and Raymond Johnson give a hcemoglobin value of 15.63 grams per cent. for normal American male students (i.e. approximately II4 per cent. Haldane), and Whitby (1936) obtained normal values of II4 per cent. and 100 per cent. (Haldane) or 90 per cent. and 80 per cent. (Sahli). From Table I, it appears that the weighted mean for the hcemoglobin value in respect TABLE 1. NORMAL BLOOD COUNTS FOR :MALE ADULTS AT SEA-LEVEL AJ."'ID WITW ATERSRAND. Author. No. of Subjects. Erythrocytes (millions per c.mm.). Range. Mean. Hremoglobin (grams %). Colour Index. 100% Hb.=14 5 =20 grns.% Range. Mean. 100% R.b.c.=5 0 = 6 0 million. Altitude. Schepers. 100 4 94-7 4-6 03 Symons 46 5 1-7 0 6 02 Liknaitzky.. 60 5 1 - (i 8 5 99 E mmerson.. 118 4 7-7 6 6 04 Stammers.. 14 - - - 17 0-24 7 20 0 1 12 1 02 I: - d'". Cl) f 12 7 17 2 15 0 0 91 0 75 12 2-16 3 14 5 0 88 0 72.80 do o - - - - -.-:: 13 8-16 6 14 8 - -.,; Weighted m ean.. - - - 6 04 - - 17 2 1 00 0 83 Price J ones et al... 96 4 5-6 5 5 5 McGeorge. 50 4 9-6 4 5 5 Whitby and Hynes. 50 - - 5 7 F oster and R aymond J ohnson 115 - - 5 3 Weighted mean. - - - 5 5 12 4-17 0 14 6 0 96 o 0 13 1-17 3 15 2 1 00 o 3 13 8-17 4 15 5 0 98 0 82 Oi > - - 15 6 1 08 O 7 d en - - 15 2 1 00 0 83 On the Witwatersrand 6.0 miliion cells per c.mm. may be regarded as the basic requirement, judging by the figures of Emmerson (1932), Stammers (1933), Liknaitzky (1935) and Symon (1939). The present writer obtained an average erythrocyte enumeration of 6.03 million cells as the result of counts personaliy conducted on a hundred selected healthy medical students, bus and t ram drivers, traffic policemen and labourers not in the mining service (Schepers, 1947). The weighted mean value of the erythrocyte count in males on the Witwatersrand thus appears to be constant at 6.0 million cells. In this paper I therefore take 5.5 miliion as the mi nimum requirement for men, and 5.0 million as the minimum for women. H <Emo globin: Remarks similar to the preceding apply also with respect to the hcemoglobin levels. of males at sea-level may be approximately 15.2 grams per cent. (i.e. 110 per cent. H aldane or 8g per cent. Sahli). Symons (1939) tried to show that though there may be a moderate erythrocytosis normaliy on the Witwatersrand, the hcemoglobin levels here are relatively low, so that the colour index i low. Liknaitzky and Stammers previously reported relatively low hcemoglobin values. So far I have not been able to confirm this relative normo-chromia or hypochromia. On the contrary, the hcemoglobin appears to be raised out of proportion to the increase in the red cell count. Figures obtained from my series of selected robust local residents sometimes gave re:narkably high hcemoglobin value ( chepers, 1947). Compounding my averages with those of

570 S.A. ME D I C AL JO U R NAL August 9, 1947, tammer and Liknaitzky, I obtained a weighted mean value of 17.2 grams per cent. The colour index on sea-level tandards then works out at unity. If 6.0 million cells per c. mm. and 20 g rams h;emoglobin per cent. be taken as the normal values for the Witwatersrand area, my colour index calculation of approximate unity also remains directly comparable to those obtained in other countries where 5.0 million cells and 14.5 g rams h;emoglobin per cent. are normal. The colour indices published by Liknaitzky and Symons provide no local neutral point for comparison. C ell Size: Though an;emias are classifiable on the basis of mean cell diameter, this measurement is not yet much used as a diagnostic criterion. It should be so employed. In my normal male subjects the m ean erythrocyte diameter was found to be approximately 6.9 microns. i.e. 0.3 micron less than the m ean established by Price-Jones. I found such significant variations from this mean in certai n of my patients with uggestive symptom that I have included in this serie those case in which the mean red cell diameter measurement fell beyond the standard deviation of my normal eri es, even when the red cell count and h;emoglobin level were approximately normal. Expressing these diffe rences of cell size in terms of the cell surface area brings them into prominence. But calculations ordinarily recommended fo r cell volume determination provide no advantage. The significance of the relative " macrocytosis" of my cases will emerge presently. Clinical F eatwres: To diagnose an;emia without a blood count is a dangerous but unfortunately popular clinical practice. One of my non-an;emic patients told me that she had received weekly " injections for an;emia " for a period of three years without a sing le blood examination befor or during the therapy. She merely suffered from slight myxcedema. It requires great experience of correlation between the blood coun t and the clinical picture to acquire a " clinical sense" for the recognition of an;emia. When this individual knowledge has been attained one realises that the so-called classical descriptions of an;emia may be most misleading. To wait for the classical signs of Addisonian a n;emia to appear before deciding that a blood count and treatment are indicated is to wait until the harm done to the heart, endocrines and nervous system has gone too far. There are, however, certain clinical features which are highly suggestive of an underlying an;emia. When these were present, notwithstanding an apparently normal blood count, patients were ubjected to anti-an;emic therapy which often resulted TABLE ll. STATISTICAL FEATURES OF MACROCYTIC 1L}EMIAS. Range. Averages. Dispersion. Items. Standard Coefficient Min. Max. Mean. Mode. Median. Deviation. of Variation. P er cent. E rythrocytes (millions per c.mm.). 1 0 7 0 4 04 4 0 4 0 0 908 22 50 H rem oglobin (grams per cent.). 11 9 24 7 18 0 17 8 18 73 2 283 12 69 Colour lndex..... 0 7 2 0 1 37 1 30 1 35 0 224 16 40 (100% Rb. = 20 gms. % 100% R.b.c.= 6 million per c.mm.) Mean Erythrocyte Diameter... 4 0 11 0 7 8 8 0 7 5 1 29 16 70 (lin ear microns) Mean Erythrocyte Surface Area. 25 0 225 0 102 0 100 0 108 0 39 0 39 80 (square microns) Leucocytes (thousands per c.mm.). 5 0 8 0 7 0 S O 11 0 15 3 9 0 22 0 6 6 43 10 11 0 40 0 23 0

Augustus 9 1947 S.A. TYDSKRI F VIR GE NEESKUNDE 571 in a remarkable preliminary fall in the blood count, followed by restoration of the blood level and alleviation of symptom s. This observance suggests that in estimating the presence or absence of an<emia, one is concerned not only with the normality of the anatomical blood picture. but also with the vitality of the individual cells. We know that effete erythrocytes are disposed of in the spleen. But what we do not know is what proportion of circulating red cells, though already physiologicaliy useless, continue circulating until such time as they can be disposed of by the reticulo-endothelial system. Present-day physiology tacitly assumes that all circulating erythrocytes are functioning cells, and that a cen count plus h;:emoglobin estimation is an index of the" vital capacity " of the blood. Oxygen-carrying capacity is of course determined separately. But O 2 and CO 2 carriage are not the only functions of red celis. Once it is granted that red cells m ay continue in circulation, although their vital functions are no longer adequate, it follows as a corollary that an<emic and other symptoms m ay result in proportion to the number of such" effete " cells com posing the blo od count. Perhaps som e new modification of the Romanovsky stains can be devised to differentiate between normal and spent cells ; perhaps vital and supra-vital staining in suspension will be necessary. But the significan ce of such an hypothesis is that it emphasizes the role of the clinical observer in the diagnosis of an;:emia, and stresses the danger of ignoring slight modifications of the blood picture. HAOMATOLOGY : In order to simplify discussion, the histographs obtainable from an analysis of the blood counts in the present series of cases are reproduced in Figs. I, 2, 3, 5 and 9. The features of the. an<emias are more or less self-evident from these and from Table II. Erythrocytes: Symptoms of an;:emia were present in cases with erythrocyte counts ranging from 1.0 million cells per c.mm. to 7.0 million cells per C.mm. (Fig. I, T able II). The m odal mean is seen to occur at 4.0 million cens and the median and arithmetic mean approximately corresponded to this figure. While the tendency persists only to speak in terms of an<em ia with red cell counts below 4.0 minion, ju st about half of my ca es w ould be regarded as such, still yielding a remarkably high incidence. But to a more enlig h tened view-point it must be evident that a large number of patients with counts above 4.0 million mu t be included unde r the headi n g macro cy tic ana?/nia. The evenness of the histographic patterns suggests that these an<emias are all related in cytological character. Tho u gh the sta ndard deviation i apparently low, the coefficient of va r iation is relatively large (22.5 per cent. ; Table I ). H cem oglobin: In Fig. 2 the h <emoglobin levels are g iven as ahli perce ntages. Turning these percentage to ab olute weights ha no e ffect on the pattern of the hi tograph. On comparing Figs I and 2, it is evident that the range o f va ria tion in h<emo g lobin levels is g reater, thoug h the actual disper ion. as expre ea by the tandard deviation and coeffici ent of variation, i relatively small (Table If). There is also a relatively close clu tering around the modal mean of 105 per cent. ( ahli), which again coincided fairly sati factorily with the a:i -(.) Z p Cj p:; 10,. eo,. 10. -. I I _ o o : - I o.;., 0,;.n n 0." 004 FIG. L ERYTHROCYTE COU T (Millions per c.mm.) FIG. 2. HlEMOGLOBI LEVELS. (SahJi percentages.). I I. arithmetic mean. P articula r attention is to be paid to the h<emog lobin level to the rig ht of this modal mean. T o rely on T alq vist tandards only for the recognition of an<emias would mean mi ing about half the ca e of macrocytic a noemia. C olollr Jndex: The degree of hoemog\obin concentration per cell is reaecte d in the colour indices graphed in Fig. 3. The modal mean of ca. 1.3 wa

572 S.A. MEDICAL JO U R NA L August 9, 1947 a little less than the arithmetic mean (Table Ill ). The amemias are evidently also markedly hyperchromic, though the range of variation is considerable. The dispersion is low, however, and the values are closely clustered round the high modal mean. With such hyperchromia it is surprising that symptoms of anremia were still in abundance. The clue to this paradoxical state of affairs is to be found in the accompanying macrocytosis. Though these cells are over-saturated with hremoglobin, the distribution of hremoglobin is relatively ineffective. It may even be that this apparent high saturation is not compensatory for the cell defect, but indicative of the loss of efficiency of cells due to be destroyed by the reticulo-endothelial system. 70 the writer places much g r eater reliance than on Eve's halometer, though the latter admittedly gives a quicker result. Bu cell diameter mensuration seldom needs to be done in a hurry. ai,.:j - >.n '" z1i HO j:qs 0'" H. Cl :;:; Oc:: ::S r--------------------------------- u o 100 00 so 70 '" -., zo I I FIG. 3. COLOUR I NDICES. (Normal range : 0.9 1.1. ) I FIG. 4. ERYTHROCYTE COU TS. (In millions per c. mm.) The modal mean coincides with a mean diameter of approximately 8.0 linear microns (Fig 5, Table II), though m easurements of 7.5 and 8.S microns were almost. as frequent. There is less kurtosis in this histograph than that given in Price-Jones's r--------------------------------, Correlation between Red Cell Count and H cen1 0 - globin Levels: Though the colour index may express the degree of hremoglobin saturation per cell in individual cases, the relationship between h<emoglobin and cell count is only established for the whole series in a graph of correlation such as in Fig. 4. Roughly speaking, the h<emoglobin concentration increases with increase in the cell count. But the rate of increase shows a diphasic (perhaps triphasic) trend. Thus, for counts between 4.0 and 5.0 million cells the rise in h<emoglobin is very steep (intermediate almost vertical sector of the curve), increasing from ca. 80 per cent. to ca. 120 per cent. (Sahli). Above and below these limits the curve shows deceleration, which is progressive as its lower and upper limits are approached. It is obvious that within this series of anremias the changes in the cell count and h<emoglobin level obey fundamental physiological rules. M ean E rythrocyte Diameter: These were established for the most part by means of the elaborate micrometer method on stained smears, on which '" FIG. 5. MEAN ERYTHROCYTE DIAMETERS. (In linea r microns.)

Augustus 9 1947 S. A. TYDS K RIF V IR GE NEE S KUN DE 573 curve for pernicious anremia. The biometrical dispersion is rem arkably low for such a great range of variation in cell size (Table II). These anremias are therefore undoubtedly macrocytic. In many of the small-celled anremias anisocytosis was m arked and opaque megalocytes were interspersed between excessively small cells, so that, though the mean had to be established at a low measurement, the curves showed marked kurtosis compared to similarly established Price-J ones his tographs for individual cases of hyperchromic anremia. In these cases, too, the colour index was relatively high, so that the anremias qualified for inclusion within the same classification as the remaining macrocytic series. Moreover, on treatment these cases often showed an initial exce sive increase in cell diameter, just as the large-celled series frequently passed through microcytic crises with numerous filaments and granules before reverting to normal diameters. In these respects the present series of amemias behaved as do the pernicious anremlas. FIG. 6. MEAN ERYTHROCYTE DIAMETERS. (In linear microns.) Correlation between Erythrocyte Counts and Mem Erythrocyte Diameters: This correlation, depicted in Fig. 6, was most suggestive. With cell counts falling from 5.0 million to 3.0 million the cell diameter evidently increases pari passl from.a mean of approximately 6.0 microns to a men. diameter of circa 10.0 microns. Beyond these limits the curve shows divergent trends. Thus, as the cell count rises the rate of decrease in cell volume falls less steeply; and as the cell count falls below 3.? mil}ion, the mean erythrocyte diameter falls less Invanably within the higher range, as could be expected f:om the intermediate trend of the curve of correlation. But the graph strongly confirms the evidence for a responsible physiological principle uggested by the correlation established between cell count and hremoglobin level. Correlation between Hamwglobin Levels and M ean Erythrocyte Diameters: The curve obtained by comparing these value in the pre ent series of anremias (Fig. 7) is the direct contrary of that resulting from correlation between the cell count and cell diameters (Fig. 6). My curve could perhaps have been a more decidedly sigmoid one, with a straighter 1 r--------------------------------- 140 130 -eo e..... -.- :..... :::.-. :. :.:-''': e, :... -. -.- e. :.: :1..._. :.-. :... e.:. :.::.: : :.. o!-' -:; """":: :-- ""'!"'" -- --_.J 9 ::: l\1ea FIG. 7. ERYTHROCYTE DIAMETER (In linear microns.) intermediate sector. But on the whole the main trend appears to be for cell size to increase proportionately to increase in the hremoglobin level. This rate of increase is somewhat more erratic in the microcyte cases (though the result may be modified by the fact that fewer of the ca es fell within this group). At the other extreme the cells are less constantly ultra-hyperchromic when megalocytosis predominated. Again, the result may have to be modified when larger series of these extreme cases become available. But abou the correlation between hremoglobin levels and cell measurements of intermediate range there can be no doubt. Correlation between Mean Erythrocyte Surface Area and Colour Index: Probably the most interesting correlation of all those characterising this series of anremia is the curve obtained by plotting the colour index again t the mean erythrocyte surface area of individual ca e (Fig. 8). Unlike the previously discussed curve, the present one possesses n diphasic cha.racter. It describes a single phase WIth a convexity upwards. The first portion is almost straight; the latter half curves more sharply. 1.'his shows hat. the colour index of the present senes of anremlas Increases pari passu with increase of cell surface area up to the normal mean of approximately 75 sq. microns per cell. Beyond this

574 S.A. MEDI CAL JOURNAL August 9, 1947 range the curve hows progressive deceleration, i.e. the rate of increase in colour index is no longer proportionate to the rate of increase in cell sur face area. The general character of the curve suggests that, beyond a certain surface area, not much in exce s of approximately 250 sq. microns, there would be no further increase in the colour index. And also, in an;emias with excessively ma rked microcytosis the colour index would probably tend to fall disproportionately rapidly. surface. When we see a macrocyte on a slide we are really assessing its surface area rather than its volume. But though an experienced h;ematologist may recognise degrees of h;emoglobin concentration in such cells by observi ng the extent of visible hypoor hyperchromia. thi ronc:pntration can be more directly e tablished arithmetically. TABLE Ill. A. CO CENTRATION OF H lemoglobin PER UNIT AREA OF ERYTHROCYTIC SURFACE. ui r:.l Q >-< A Z >-< p::;, P 0 H 0 Q Mean Erythrocytic Colour Index (C.!.) Hremoglobin Sur. Surface Area (sq. (Hb.=120% Sahli. ) face Concentration Microns) (M.E.S.A.) R.b.c.=6 0 million Index (H.S.C.!.). per c.mm.) 25 0 70 2 80 50 0 90 1 80 75 1 20 1 60 100 1 35 1 35 125 1 ' 50 1 20 150 1 60 1 04 175 1 68 0 96 200 1 76 0 88 225 1 80 0 80 FIG. 8. B. VARIATIONS I N THE HlEMOGLOBIN SURFACE CONCENTRATION INDEX ON THE WITWATERS. RAND (Alt. 6,000 feet ). MEA ERYTHROCYTE S RFACE AREAS. (In square microns. ) H.S.C.!'=C.I.+ Standard Coefficient 100 M.E.S.A. Mean Range. Deviation. of Variation. Hce'/ll oglobill utface Concentration Index: To express the h;ematological feature revealed in the preceding correlation more precisely, I have devised a H cemoglobin SlIrface COllcentration Index, which is obtained thus : Normal values ],57 } 13-2 40 0 290 Per cent. 18 6 H'amoglobin SUT/ace _ OolOUT I ndex X 100 Concentration I ndex - M ean Erythrocyte Sur/ace Area Anremia 1 06 0'!l0-2'80 0 337 31 8 C. Z. or: H.S.C.I. = ---- X 100. M.E.. A. To the biometrical purist it may seem distasteful to interming le indices with absolute measurements. But the advantages obtained in doing so are particularly great in the present case. We know that 9/ ro of a normal blood corpuscle i compo ed of h;emoglobin. We also know that thls h;emoglobin i di tributed to the inner surface of the erythrocyte pellicle. The cell also provides a maximum of urface a rea for this purpose by virtue of its bi-concave discoidal shape. We know that the h;emoglobin must function through the cell pellicle, and its molecules must intermittently" attract" and " discharge " the oxygen and carbon dioxide m olecules. The O'reater the surface area provided for this purpo e the better the " breathing capacity" of the cell becomes. When the cell alters in size, the ignificant change i not the volumetric cytoplasmic increase or decrea e, but the effect on cell The colour index is the fir st step in t his direction. Having provided a clue concerning the h;emoglobin concentration per cell, it has not yet determined what the effective di tribution of thi h;emogjobin is. What really matters is not the h;emoglobin concentra tion per cell, but the breathing surface of the blood. The m ean erythrocyte surface area is a further step in the direction. If now.the colour index be divided by the mean erythrocyte surface area, we obtain the desired result; for we take into consideration not <?nly the cell surface a rea, but also the h::emoglobin level, the red cell count and the standards existing for the individual geographic circumstances. Biirker (quoting from Lovatt Evans) established that there is a close relationship between the surface area of the erythrocyte and its h;emog lobi n content, and obtained a ratio constant at (32 x ro 14 ) grams h;emog lobin per square micron of corpuscle surface in different animals. This factor presupposes the

Augustus 9 1947 S. A. T YDSKRIF VI R G ENEESKUN D E 575 normality of the individuals and ignores the total cell count. T he H ;emoglobin urface Concentration Index aims at expressing the same ratio m ore simply (but taking into account the geog raphic blood standard and the red cell count). R eference to T able III will show how this index behaves in the present series of macrocytic an;emias. F rom this index it is evident that the actual concentration of h;emoglobin per unit area of erythrocytic surface area decreases in inverse ratio to the increase in the colour index and the cell surface (and thus the cell diam eter). The physiological efficiency of such a cell is thus g rossly m odified. The most sig nifi cant change in the e an;emias is therefore obviously macrocytosis. Cytological F eatures : The macrocytosis, and occasionally extreme mic rocyto is, of these cases have already been stressed. These macrocytes were most topically hypochromic. But many of the larger cells tended to be very opaque with a granular appearance. Basophilism and punctate basophilia were much les common. Reticul ocytes we re infrequently seen in cases not yet under treatment. R ouleaux formation was poor. Anisocytosis was almost always demonstrable and often present to a marked degree. Poiki locytosis was less commonly present. 1 ormocytes, normoblasts and even meg al oblasts were often observed but showed no special characters. T hey often became most num erous soon after treatment had been instituted, disappearing later. r----------------------------------------- 10 )(I R SD'SIS o 0 10 " FIG. 9. LEUCOCYTE COU T B. (Thousands per c. mm.) L eucocy tes: R eference to T able II and F ig. 9 wi ll show that the range of variation in the leucocyte count is considerable, and di sper ion between these li mits is O"reat, so that the median value. mode a nd arithmetic mean do not coincide. F rom Fig. 9 it is evident that there are three di stinct moieties wi thin this series, which accounts for this excessive range and inordinate dispersion. Correlation with clinical symptom sug g ests that the leucopenic g roup (7.0, range 5.0-8.0) occurs in the a febrile cases or where no active infective element could be demonstrated. In the e case there wa a g eneral trend for a hi ft to the rig ht in the rneth count. Immature leucocyte were fe w ; and neutrophil polymorphonuclear predominated, except in certain ca e when a relative lymph ocyto si existed. ma ll Iymphocyte comprised mo t of the agranular cells, except in the latter erie when large and immature lymphocytes were more common. In the intermediate series (mode 9.0, range 8.0-11.0) minor septic complications, espe cially in the urinary tract, were suffi ciently frequently pre ent to account for sli g ht leucocyto is. Also in the e cases the tendency wa toward the appearance of immature leucocytes (megalocytes and even myeloblasts). with a relati ve prepond erance of large lymphocytes, relatively hi g h monocytosis and a tendency to slight eo inophilia. In the leucocyte group (mode 23.0, range 11.0-4 0.0 ) py;emic complication may have been the underlying cause. In these ca e the g ranular cells again predominated, with a light hift to the left in the Arneth count. It is to be noted that the leucopenia re ponded a r eai1 y to parenteral live r therapy as the erythrocytopel1la. THERAPY. In each case the type of therap'y was adjusted to the attendant circumstance. R eliance wa mainly placed on pa renteral liver preparation. (The work was done before folic acid came into vogue.) Where necessary, oral ferric preparation w ere used a adjuvants, particularly where the patient wa of the psychological type, responding better to m ixture and pills swallowed than to injecti o ns. But th injection worked Ollt cheaper in the long run. Sig ns of intolerance were ra re. In addition to the anti-a n;emic trea tment attention was paid to all systemic accompaniments o f the disea e. P art of the recovery in the blood was in most case attributable to thi form of the rapy. R esponse to therapy was relatively rapid. A few cases re i ted all manner of therapy. The improvement in the red cell count was in ma ny cases di- proportionate to the reticul ocyte re ponse. P erhap part of the recovery wa then due to correcti on of underlying relative hydr;emia. The respon e involved not only re toratioll of the red cell count, which could usually be ent up to well over 6.0 million cell per c. mm., in from three to four m c n hs, but also wa refl ected in the cell ize. which returned to normal (tholw h in ome ca es howing intermediate cr-i es of exce ive microcyto is or macrocyto i ). Cell shape re pond ed rapidly to tr-ea tment. H ;emoglobin wa often le rapidly re tored, in ome ca e fa ll in O' to lower level at fir t. The colour index fe ll relatively sharply. In addition the leucocyte count improved peedily. Most g ratifying wa the clinical and subjectiye im provement. Many patie nt re ponded within a week, and were evidently 0 encourao"ed that they di d not eem to mind the painful reaction ometimes following on li ver in jections. (T o be continued.)