British Journal of Haeniatology, 1976, 32, 403.

Measurement of the Mean Cell Volume using Electronic

Particle Couii ters

J. M. ENGLAND AND M. C. DOWN

MRC Experiwzental Haematology Research Unit, St Mary’s Hospital Medical School, Lotdoon

( R e c h d 3 0 June 1975; accepted for publication IZ August 1975)

SUMMARY. Under carcfully controlled conditions electronic cell counters, for example tlic Coultcr Counter, Model F N , and Channelyzcr, may be calibrated to give MCV values down to as small as 20 fl which agree with those derived from tlie ceiitrifugatioii PCV (corrected for plasma trapping) and thc red cell count.

The MCV values will be too high if the instrument uses a high cell concentration, has a fixed lower threshold, no effective upper threshold and no edit facility. This may partly cxplaiii why the Coulter Counter, Model S, when standardized with 4C cell control gives higher MCV values than the Model F N linked to the Channelyzer. The difference is, on averagc, 2 f l with normal blood samples and 5 f l in cases of niicrocytic anaemia. It is suggested that staiidards of low MCV should be uscd together with thosc of noriiial MCV when calibrating thc Model S.

The mcan cell volume (MCV) was originally calculated from the red cell count (RBC) and the packed cell volume (PCV). However, it was difficult to obtain accurate MCV values becausc of the errors in red cell counting with the haemocytometer (Biggs & MacMillan, 1948a, b). Furtheriiiorc the PCVs were systematically over-estimated because plasma rc- niained trapped between the ccntrifuged red cells though this error was not so serious with thc microliaematocrit as it was with the Wiiitrobe tube (Chaplin & Mollisoii, 1952; Garby & Vuillc, 1961; England ct al, 1972).

The introduction of optical and electronic methods for particle counting iiicreascd the accuracy of the RBC count (Croslaiid-Taylor, 1953 ; Coulter, 1955) so that bctter cstimations ofthe MCV became possible. However, at about this time it was also suggested that electronic cell counters might be used to estimatc directly tlie MCV because the impulses they produce arc proportional to the ccllular volume (Mattern et al, 1957). The MCV determined in this way will not of coursc bc affected by dilution errors or by the problem of plasma trapping. Several workers liavc therefore concluded that the MCV measured electronically will be more accurate. Unfortunately, however, there are niaiiy sources of error which affect elec- tronic estimations of the MCV.

Thc size of the iinpulsc produced by a particle counter is not always proportional to the volume of the cell passing through the orifice becausc the electrical field is not uniform. Cells passing iicar the edge of the orifice, for example, produce impulses with greater amplitude

Correspondence: Dr J. M. England, MRC Experimental Haematology Research Unit, St Mary’s Hospital Medical School, London, W.Z.

403

404 J. M. England and M. C. Down

and duration than those produced by cells going through the axis of the orifice (Thom et al, 1969; Shank et al, 1969). This leads to an over-estimation of the MCV and various attempts have been made to reduce this artefact. The simplest approach is tq use suitable electronic circuits to filter out the impulses derived from cells passing through thc cdge of the orifice, and this is an objective of instruments such as the Coulter Channelyzer (Coulter Electronics, 1975). Alternatively orifice tubes may be devised with special flow characteristics so that nearly all of the cells pass axially through the orifice (Thom et a!, 1969). However, even if the flow pattern is suitably adjusted it may still be necessary to filter the impulses electronically.

There are also further difficulties in estimating the MCV with electronic cell counters, in that the instruments must somehow be adjusted to only measure erythrocytes. It is therefore necessary to have a lower threshold to eliminate platelets and in most instruments this threshold is fixed. However, some of the red cells in cases of microcytic anaemia may be as small as 28 f l and then fall belowwhat might otherwise bc a suitable lower threshold (England & Down, 1974). The exclusion of small cells will tend to over-estimate the MCV-as will the inclusion of large numbers of leucocytes, because many instruments have no effective upper threshold (Hattersley & Ragusa, 1967).

Electronic cell counters actually calculate thc MCV by summating all of the impulses and dividing the total by the cell count, so that the problem of the accuracy of the RBC count also arises. This means that the degrec of dilution of thc blood sample is important because this will determine the coincidence errors and their correction (Cook, ~967).

W e have therefore re-assessed the mcasurcmcnt of the MCV by electronic cell couiitcr~, comparing the results with those obtained from the PCV corrected for plasma trapping. W e have also attempted to assess the effects of the impulses from cells passing near the edge of the orifice tube, the effects of various upper and lower thresholds and the degree of dilution of the blood sample.

MATERIALS AND METHODS

Blood samples obtained from normal subjects, patients, goats, sheep, horses and mice were anticoagulated with KzEDTA (1.5 mg/ml) and studied on the same day. The MCV values were calculated from manual red cell counts and packed cell volumes. The red cell counts were performed in quadruplicate, counting at least 1500 cells from photographs of the count- ing chambers (Dacie & Lewis, 1968). Thc packed cell volumes were measured in Wintrobe tubes centrifuged at 1250 g for 30 min and a correction was made for plasma trapping using 1311-human serum albumin (England & Down, 1975).

Red cell volume distribution curves werc measured on a Coulter Counter, Model FN (orifice IOO pm, attenuation 0.707, aperture 8), connected to a Coulter Channelyzer (base channel threshold 5 , window width 100). With these settings it is necessary to add five to each of the channel numbers to allow for the base channel threshold. The mean channel was calculated from a fitted lognormal distribution as described previously (England & Down, 1974). Volume distribution curves were measured on suspensions which, uiiless otherwise stated, contained 107 cellsll. and were studied with the edit switch ofthe Coulter Channelyzer in the ‘on’ position. Not less than 105 cells were studied in each experiment.

Blood counts were performed on a Coulter Counter, Model S, standardized with Coulter 4C Cell Control.

Mearz Cell Volume 405

RESULTS

A Comparison of the MCV Derivedfroni the P C V and the Red Cell Count with the MCV Obtainedjom the Conlter Channelyzer

The MCVs were calculated from manual red cell counts and spun PCVs for four samples of normal human blood, blood from a subject with p-thalassaemia trait and blood samples from a normal sheep and goat. The red cell volume distribution curves were also obtained for these samples using the Coulter Channelyzer and the mcan channcI number in which the rcd cells were found was calculated.

50 -

40 - d W z 2 30- I 0 2 a 2 0 - Y

0 / 25 50 75 100

MCV (ti)

FIG I . A comparison between the mean channel number for various erythrocytes and their MCV calculated from the RBC and the PCV (corrected for plasma trapping). The line is the best fit for the four nornial human blood samples and has been drawn through the origin.

The mean channel values could then be compared with the MCV results (Fig I). There was a linear relationsliip between the two indicating that the mean impulse (expressed as a channel number) was proportional to the MCV derived from the PCV- (corrected for the trapped plasma) and the manual red cell count. The slope of the line gave the factor, 1.962, by which the chaiinel numbers must be multiplied to coilvert them to volumes in fl.

The MCV Measured on the Coulter Counter, Model S, Compared with the MCV Obtainedjom the Coultcr Channelyzer

The results obtained on 130 blood samples with the Coulter Counter, Model S, have been compared with those obtained from the Coulter Channelyzer using the best thresholds, low cell concentrations (10~11.) and thc edit facility ‘on’. The results are shown in Fig z and the line of equality is indicated. It can be seen that a t around IIO fl both instruments give similar results but that with lower MCV values the Coulter Counter, Model S , gives progressively higher results than the Channelyzer. In severe iron deficiency, for example, the Model S may

406 I. M. Englad and M. C. Down

record 54 fl when the Chaiinelyzer records 46.5 fl, and with goat blood the Model S result was 38 fl when the Channelyzer gave 20 fl.

Various mathematical functions relating these variables have been fitted to the data using the method of least squares. Taking the Model S MCV as s and attempting to predict the Channelyzer MCV, y, the linear relationship was found to be y = - 10-54+ 1 . 0 9 4 ~ . The variance of the points about a fitted parabola was not significantly less than the variance about the line (P> 0.05, F test) and the line did not pass through the origin (P< 0.001, t test to com- pare intercept with 0).

150 -

100

5 3 N

W z z 4 I V

50

MODEL S

FIG 2. MCV valucs on the Coulter Counter, Model S, compared with those fronl the Model FN and the Channelyzer. The open circles represent the aninial blood saniples.

The regression line predicts that an MCVof 89.5 fl on the Model S would be equivalent to an MCV of 87.4 fl on the Channelyzer, an overestimate by the Model S of 2.1 fl. Similarly a microcytic blood sample giving an MCV of 55 f l on the Model S would be 49.6 fl on the Channelyzer, an over-estimate of 5.4 fl. The two instruments would give the same MCV at I12 fl.

There are several possible reasons for the discrepancy between the Model S and the Chan- nelyzer. The Model S has a fixed lower threshold and would automatically exclude some of the very small cells present in cases of microcytic anaemia (Fig 3). This would tend to over- estimate the MCV. Similarly this instrument has no effective upper threshold so that any large extraneous impulses, either due to erythrocytes passing through the edge of the orifice, instrument ‘noise’ or to leucocytes, would also lcad to an increase in the observed value of the

Memi Cell Volume 407 MCV. Tlicse extraneous impulses becoinc particularly noticeable if, as in Model S, a relatively high cell concentration is used and there are no editing circuits (the editing circuits in the Cliannelyzer eliminate long duration impulses originating from cells passing through the edge oftlie orifice tube). This effect is illustrated in Fig 3 wherc the extraneous impulses skew tlie volume distribution curve to tlic right if a higher cell concentration is used or if tlie editing circuits are switched off.

0 L, I I

0 40 80 120

VOLUME (fl)

FIG 3 . Red cell volume distribution curves froin a patient with severe iron deficiency anaemia with the edit switch 'on' at concentration of 10~11. (A) and ~o"/l. (B) arid with the edit switch off at concentra- tions of 1o7/I. (C) and 1oS/1. (D).

The combined effects of these factors can be precisely quantitated by comparing tlie results with the best settings on the Channelyzer (suitable lower and upper thresholds, edit 'on', 107

cellsil.) with those obtained with less suitable settings (lower tlireshold 30 fl, upper threshold 200 fl, edit 'off', 10' cellsil.). Using the less suitable settings it was found that the MCV of normal blood saniplcs were apparently over-estimated by around I 5% whilst the MCV of vcry iiiicrocytic samples were over-estimated by as much as 30%. It follows therefore that any instrument using these less suitable scttings would progressively overestimate the MCV the inore iiiicrocytic the red cells became, and this may explain thc bcliaviour of thc Model S (Fig 2 ) .

DISCUSSION

W e decided to calibrate our electronic cell counter system (tlie Coultcr Countcr, Modcl FN, and the Cliannelyzer) using human and animal blood samples whose MCV was derived from conventional PCV and RBC measurements. A plasma trapping correction was applied to tlie PCV value because plasma may occupy more than 20% of the centrifuged red cell coluiiin when the cells are very small (Chien et a!, 1965; England & Down, 1975). With the Coultcr

408 J. M. England and M. C. Down

Channclyzer the channel number is a measurement of the particle volume, and the mean channel should therefore be an indication of the MCV. 111 fact, there was a direct linear rela- tionship between the mean channel and the MCV down to a volume of 20 f l (Fig I). This demonstrates that it is possible to obtain accurate estimates of the MCV with a conventional Coulter Counter without using special tubes to make the cells pass through the centre of the orifice (Thom ~ ' t al, 1969). Carter et al (1968) also showed that electronic particle counters could provide accurate measurements of the MCV, though they did not study blood samples with MCVs less than 65 fl.

The investigations with the Coulter Counter, Model S, show that when this instrument is calibrated with Coulter 4C cell control it tends to over-estimate the MCV of normal blood samples by 2 fl. This conclusion is in agreement with earlier findings made in this laboratory where Bain & England (1975) found a mean MCV of 89.5 f l for a group of normal subjects when measured on the Coulter Counter, Model S . This value is 1.8 fl higher than the mean MCV value of 87.7 f l which England & Down (1974) had obtained on a similar group of normal subjects using the Coulter Channelyzer. Other workers have also noted that the Model S tends to give a high value for thc MCV of normal subjects (Silver & Frankel, 1971). This may be related to the fact that the Model S was calibrated with 4C cell control and the MCV of this material may be slightly too high if it is determined from a centrifugation PCV without allowance for trapped plasma.

The Coulter Counter, Model S , also progressively over-estimates the MCV as the cell volumes become smaller though the reason for this is not immediately apparent. However, the Model S has several characteristics which could lead to an over-estimation of the MCV particularly when it is low. This instrument has a fixed lower threshold, no effective upper threshold, no edit facility and it uses higher cell concentrations than the 10'11. which we have used on the Channelyzer. All these factors tend to over-estimate the MCV more on microcytic samples than on normal samples. This means that even if the Model S were ad- justed to give the correct value for the normal samples it would still over-estimate the MCV on the microcytic ones. There is, however, a linear relationship between the MCV on the Model S and the MCV on the Channelyzer. It would therefore be very simple to obtain more accurate MCV on the Model S provided the instrument was calibrated with standards of low MCV as well as those of normal MCV.

ACKNOWLEDGMENTS

We thank Professor P. L. Mollison and Dr N. C. Hughes-Jones for their helpful suggestions.

REFERENCES

BAIN, B.J. & ENGLAND, J.M. (1975) Normal haemato- logical values: sex difference in neutrophil count. British Medical Journal, i, 306.

BIGGS, R. & MACMILLAN, R.L. (1948a) The errors of some haeniatological iiiethods as they are used in a routine laboratory. Journal of Clinical Pathology, I, 269.

BIGGS, R. & MACMILLAN, R.L. (1948b) The error of the red cell count. Journal of Clinical Pathology, I, 288.

CARTER, A., CROSLAND-TAYLOR, P.J. & STEWART, J.W. (1968) Measurement of mean corpuscular and packed cell volumes with a Coulter cell counter. Journal of Clinical Pathology, 21,222.

Mean Cell Volume 409 CHAPLIN, H. & MOLLISON, P.L. (1952) Correction for

plasma trapped in the red cell coluninofthe haemato- crit. Blood, 7, 1227.

CHIEN, S., DELLENBACK, R.J., USAMI, S. 8i GREGERSEN, M.I. (1965) l’lasma trapping in heniatocrit deter- mination. Differences among animal species. Proceed- ings of the Soci~ty .for Esprriniciztal Biolo,vy and Medicine, 119. 1155.

COOK, J.S. (1967) Size determination of human erythrocytes with an electronic counter. Joirriial of’ Laboratory and Clinical Ililrdicine, 70, 849.

COULTER ELECTRONICS (197s) Iristrtrction Manual .fir Corr1tt.r Channelyzer C-1000, 3rd cdn. Coulter Electronics, Harpcndeti.

COULTER, W.H. (1955) Instrirction Mantra1 to the Coirlter Coitnter Model A. Coulter Electronics Inc.

CROSLAND-TAYLOR, P.J. (1953) A device for counting small particles suspended in a fluid through a tube. Nature, 171, 37.

DACIE, J.V. 81 LEWIS, S.M. (1968) Practical Harrtiato- logy, 4th edn. Churchill, London.

ENGLAND, J.M. 81 DOWN, M.C. (1974) Red-cell- volume distribution curves and the measurement of anisocytosis. Larzcrt, i, 701.

ENGLAND, J.M. & DOWN, M.C. (1975) Deterinination of the packed cell volume using i311-hunian seruni albumin. British Joitrnal qf’Haematology, 30. 365.

ENGLAND, J.M., WALFORD, D.M. 81 WATERS, D.A.W.

(1972) Re-assessment of the reliability of the haematocrit. British Journal of Haematology, 23, 247.

GARBY, L. & VUILLE, J.-C. (1961) The amount of trapped plasma in a high speed micro-capillary haeniatocrit centrifuge. Scandinavian Journal sf Clinical and Laboratory Investigation, 13, 642.

HATTERSLEY, P.G. 81 RAGUSA, D. (1967) An elec- tronic mean cell volume computer and hematocrit accessory. American Journal of Clinical Pathology, 47, 229.

MATTERN, C.F.T., BRACKETT, F.S. & OLSON, B.J. (1957) Determination of number and size of par- ticles by electrical gating: blood cells. Joumal of Applied Physiology, 10, 56.

SHANK, B.B., ADAMS, R.B., STEIDLEY, K.D. & MURPHY, J.R. (1969) A physical explanation of the bimodal distribution obtained by electronic sizing of erythrocytes. Journal of Laboratory and Clinical Medicine, 74, 630.

SILVER, H. & FRANKEL, S. (1971) Normal values for ineati corpuscular volume as determined by the Model S Coulter Counter. Ainerican Journal o f Clinical Pathology, 55, 438.

THOM, R., HAMPE, A. & SAUERBREY, G. (1969) Die elektronische Volunienbestimniung von Bliitkor- perchen und ihre Fehlerquellen. Zeitschrjfi .fur die Gesamte Esperirrienfelle Medizin, 151, 33 I.