Spurious counts and spurious results on ... - Wiley Online Library

INTERNATIONAL JOU RNAL OF LABO RATO RY HEMATO LOGY

Spurious counts and spurious results on haematology analysers: a review. Part II: white blood cells, red blood cells, haemoglobin, red cell indices and reticulocytes M. ZANDECKI, F. GENEVIEVE, J. GERARD, A. GODON

Haematology Laboratory, University Hospital of Angers, Angers, France Correspondence: M. Zandecki, Haematology Laboratory, University Hospital of Angers, 4, rue Larrey, 49000 Angers, France. Tel.: +33 241 35 53 53; Fax: +33 2 41 35 55 99; E-mail: [email protected] doi:10.1111/j.1365-2257.2006.00871.x

Received 30 January 2006; accepted for publication 30 July 2006 Keywords Haematology analysers, automated count, cell blood count, spurious count, white blood cells, haemoglobin, red blood cells, mean cell volume

SUMMARY

Haematology analysers provide quick and accurate results in most situations. However, spurious results, related either to platelets (part I of this report) or to other parameters from the cell blood count (CBC) may be observed in several instances. Spuriously low white blood cell (WBC) counts may be observed because of agglutination in the presence of ethylenediamine tetra-acetic acid (EDTA). Cryoglobulins, lipids, insufficiently lysed red blood cells (RBC), erythroblasts and platelet aggregates are common situations increasing WBC counts. In most of these instances flagging and/or an abnormal WBC differential scattergram will alert the operator. Several situations lead to abnormal haemoglobin measurement or to abnormal RBC count, including lipids, agglutinins, cryoglobulins and elevated WBC counts. Mean (red) cell volume (MCV) may be also subject to spurious determination, because of agglutinins, excess of glucose or salts and technological considerations. In turn, abnormality related to one measured parameter will lead to abnormal calculated RBC indices: mean cell haemoglobin content (MCHC) is certainly the most important RBC indices to consider, as it is as important as flags generated by the haematology analysers (HA) in alerting the user to a spurious result. In many circumstances, several of the measured parameters from CBC may be altered, and the discovery of a spurious change on one parameter frequently means that the validity of other parameters should be considered. Sensitive flags now allow the identification of several spurious counts, but only the most sophisticated HA have optimal flagging and more simple HA, especially those without a WBC differential scattergram, do not possess the same sensitivity for detecting anomalous results. Reticulocytes are integrated now into the CBC in many HA, and several situations may lead to abnormal counts.

2007 The Authors Journal compilation 2007 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2007, 29, 21–41

21

22

M. ZANDECKI ET AL.

SPURIOUS COUNTS ON HAEMATOLOGY ANALYSERS

scattergrams will be most prone to overlook several factitious counts.

INTRODUCTION Many haematology analysers (HA) that enumerate WBC generate a WBC differential scattergram, which is the basis for the automated WBC differential. Such WBC scattergrams are also pivotal for the generation of flags or alarms, detecting that the WBC count (and the WBC differential) is possibly erroneous and giving some information about the possible origin of the anomaly. As mentioned in part I of this report, the WBC differential scattergram is also of crucial importance to identify anomalies of platelet (PLT) counting, PLT clumps being the most obvious example. Red blood cell (RBC) counts, haemoglobin (Hb), and mean (Red) cell volume (MCV) are other parameters measured by HA, which are also subject to the spurious values in several situations. Red cell (Wintrobe) indices other than MCV and packed cell volume (haematocrit; Hct) are usually calculated from these measured parameters, and in most circumstances erroneous directly measured RBC parameters will subsequently cause in turn erroneous calculated RBC indices. These indices can be used as a flag for abnormal counts. Reticulocytes may be considered now as a part of the CBC, and many HA have integrated reticulocyte counts under a fully automatized method.

WHITE BLOOD CELLS For blood cell analysis, each large sized particle (greater than the size of a PLT) that is not destroyed by haemolytic agents will be identified as a WBC on most HA. After enumeration, and according to the type, impedance with low- and high-frequency electromagnetic or direct current, laser light scattering (at one or at various angles), or peroxidase staining intensity are used, either individually or together, to generate a five-, six-, or even seven- part differential (for review, see Bain & Bates, 2001). It is not in the scope of this report to study how the various WBC are classified but it must be kept in mind that scattergrams generated by the HA to display the WBC differential must be fully understood by operators (Bain & Bates, 2001). In many instances, WBC scattergrams allow the detection of abnormalities related to spurious counts and/or help to explain them. As a rule, instruments that do not generate WBC differential

Spuriously low WBC counts Polymorphonuclear neutrophil aggregates Polymorphonuclear neutrophil (PMN) aggregates may be observed on blood samples drawn into EDTA. The incidence is low but certainly underestimated, corresponding to 2/65 000 full blood counts in USA (Epstein & Kruskall, 1988), 1/9500 in Italy (Bizzaro, 1993), 1/7500 in France (Lesesve et al., 2000). Both male and female show the anomaly. Although no pathology or no specific disease is associated with clustering of PMN, an acute or chronic inflammatory context (Luke, Koepke & Siegel, 1971; Savage, 1989; Kahlil, 1991; Robbins, Conly & Oettinger, 1991; Imbing et al., 1996; Lesesve et al., 2000), liver diseases (Epstein & Kruskall, 1988; Savage, 1989; Kobayashi et al., 1991; Vinatier et al., 1994; Imbing et al., 1996), or circumstances associated with the generation of cold agglutinins (Guibaud, Plumet-Leger & Frobert, 1983; Epstein & Kruskall, 1988; Robbins, Conly & Oettinger, 1991; Deol, Hernandez & Pierre, 1995; Imbing et al., 1996; Lesesve et al., 2000) have been reported in many instances. The phenomenon may or may not be transitory (Antonsen & Beyer, 1989; Bizzaro, 1993; Schinella, Kojikara & Curci, 1995).The decrease may be either moderate or clinically significant, leading at times to suspect agranulocytosis and to generate unneeded investigations (bone marrow aspiration or biopsy) or therapy (antibiotics; Epstein & Kruskall, 1988; Vinatier et al., 1994; Schinella, Kojikara & Curci, 1995). There is no relationship between the phenomenon described here, which is a pure in vitro anomaly related to sampling in EDTA anticoagulant, and the in vivo anomaly occurring in diseases such as adult respiratory distress syndrome or leukostasis, in which PMN tend to aggregate as a consequence of membrane interactions with complement (Jacob et al., 1980). The mechanism leading to agglutination is not fully elucidated (Carr et al., 1996). It is always an in vitro phenomenon, mainly EDTA dependent (Epstein & Kruskall, 1988), although agglutination after the use of sodium citrate or heparin as anticoagulants has also been reported in some instances (Rohr & 2007 The Authors

Journal compilation 2007 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2007, 29, 21–41

M. ZANDECKI ET AL.

Rivers, 1990; Robbins, Conly & Oettinger, 1991; Vinatier et al., 1994; Deol, Hernandez & Pierre, 1995; Carr et al., 1996). Using transfer experiments, either plasma from the affected patients in contact with WBC from normal patients or normal plasma in contact with PMN from the affected patients, the anomaly has proved to be related to a plasma component, and incubation of plasma with an anti-IgM antibody or with dithiothreitol abolished or quantitatively decreased PMN aggregation (Bizzaro, 1993). Using flow cytometry, IgM was found on the PMN surface in a patient (Carr et al., 1996). Some authors have reported that the size of the aggregates was larger at low temperature and that they disappeared at 37 C (Bizzaro, 1993; Carr et al., 1996). Several other authors reported that WBC agglutination in EDTAanticoagulated blood was not corrected by warming the EDTA blood sample, at least in some cases, minimizing the implication of a cold agglutinin in the genesis of the clusters (Robbins, Conly & Oettinger, 1991; Vinatier et al., 1994; Deol, Hernandez & Pierre, 1996). A high level of integrin expression (CD11b– CD18) on PMN membrane was proposed to be related to the generation of PMN clumps (Galifi et al., 1993). White blood cell differential scattergrams (using either size and complexity or size and peroxidase content) may demonstrate anomalies, but it must be kept in mind that the largest aggregates are overlooked, and only the smallest will trigger flagging. On impedance-type HA, the anomaly is suspected when events are present above the area of PMN (top of the WBC graph), or when it is difficult to separate WBC classes (Galifi et al., 1993), and the most frequent flags generated using such HA are ‘immature granulocytes’ or ‘band cells’ (Lippi et al., 1994). On Bayer-Technicon HA, the smallest clusters may appear as a band of dots in the upper right of the WBC differential scattergram: as these aggregates contain many PMN they are reported as peroxidase-rich particles, and an alarm corresponding to high peroxidase content is generated. On some HA, analysis of WBC nuclei is performed on a dedicated channel after the use of a drastic solution ensuring lysis of WBC membranes: number of free nuclei is used as a control for WBC count. Discrepancy between WBC count obtained from that channel and that from the fluorescence or peroxidase channel is observed (Bayer, Tarrytown,

SPURIOUS COUNTS ON HAEMATOLOGY ANALYSERS 23

Figure 1. Aggregate of polymorphonuclear neutrophils observed in an infected patient.

NY, USA; Abbott, Abbott Park, IL, USA). White blood cell aggregates are destroyed after drastic lysis of WBC membranes (Galifi et al., 1993). Blood films may show small (up to five cells), moderate (up to 50 cells), or large (>100 cells) clusters of PMN (Figure 1). Careful morphological examination of these aggregates shows that a few lymphocytes or monocytes may be at times entrapped within the aggregates (Hillyer, Knopf & Berkman, 1990). Immature granulocytes (myelocytes, metamyelocytes) and band cells are not infrequently reported as a part of the clustered cells, and it was proposed that aggregates might develop around myelocytes, whereas PMN alone failed to cluster together (Deol, Hernandez & Pierre, 1995). Aggregates of PMN are devoid of PLT, in contrast to PLT–PMN aggregates (discussed in part I). Eventually, as aggregation of PMN in the presence of EDTA leads to a reduction in their number, HA may generate flags corresponding to spuriously abnormal WBC differentials, e.g. spurious agranulocytosis or spurious lymphocytosis (Epstein & Kruskall, 1988; Vinatier et al., 1994; Schinella, Kojikara & Curci, 1995). Home-made anticoagulants were proposed to overcome the agglutination (Schinella, Kojikara & Curci, 1995). As mentioned above, warming the sample at 37 C may reduce both size and number of clusters in some instances, but full disappearance is far from being a consistent finding, and that method cannot be proposed to overcome the anomaly. Finger prick and immediate dilution of the blood sample prevents the agglutination.


24

M. ZANDECKI ET AL.


Aggregation of WBC other than polymorphs in the presence of EDTA Clusters of normal (nonmalignant) lymphocytes were reported in a patient with urinary tract infection, in a patient with a B-cell lymphoma without bloodstream involvement, in a patient with escarres, and in another one with chronic myelomonocytic leukaemia (Deol, Hernandez & Pierre, 1995; Lesesve et al., 2001). Lymphocyte aggregation was also reported to occur in chronic lymphocytic leukaemia patients, spontaneously when lymphocyte counts are extremely high (>400 · 109/l; O’Flaherty, Kreutzer & Ward, 1978), or not (Bizzaro & Piazza, 1991). Aggregates of three to 50 lymphoma cells were observed in two cases of Splenic Lymphoma with Villous Lymphocytes (SLVL; Juneja et al., 1992; Imbing et al., 1995) and in one case to non-Hodgkin lymphoma mimicking SLVL (Shelton & Frank, 2000). In these situations, the largest aggregates are overlooked by the HA and WBC count is spuriously but variably low. The smallest aggregates may disturb WBC differential and may or may not generate a flag, depending on the type of HA used (Shelton & Frank, 2000). In all cases reported so far EDTA was implicated, although small clusters of cells were also observed on heparinized samples drawn as controls (Deol, Hernandez & Pierre, 1996) and sodium citrate did not appreciably change the clustering tendency of the lymphocytes in one case (Shelton & Frank, 2000). Fingerpricking and immediate dilution of blood seems the best way to avoid aggregates (Lesesve et al., 2001). Heating (37 C) was reported to reduce clumping either partly (Shelton & Frank, 2000) or had no significant effect on the size of clumps (Lesesve et al., 2001). As the number of cases reported is low, only hypotheses on the mechanism(s) leading to lymphocyte agglutination have been proposed, implicating various molecules such as adrenalin (epinephrin), arachidonic acid, or leukotrien B4 (Villa et al., 1984; Shelton & Frank, 2000). In one instance aggregates involving all WBC classes was reported, in a patient with a long-standing history of alcohol abuse and alcoholic cirrhosis (Savage, 1989).

Nature and amount of anticoagulant A decrease in WBC count not related to agglutination was reported in situations corresponding to samples

containing excess of K3-EDTA (but not K2-EDTA) anticoagulant, resulting from insufficient blood drawn after vein puncture (Goossens, Van Duppen & Verwilghen, 1991).

Spuriously high WBC counts PLT aggregates and large platelets Pseudoleukocytosis may be secondary to PLT clumps large enough to mimic WBC size (Solanki & Blackburn, 1983; Cornbleet, 1983; Savage, 1984; Payne & Pierre, 1984; Lombarts & de Kieviet, 1988; Schrezenmeier et al., 1995). All modern HA that analyse WBC subpopulations detect this anomaly: PLT clumps are localized as a rocket shaped area of dots at the lower left hand corner of the WBC differential scattergram, and a flag corresponding to inability to discriminate among the WBC categories, namely lymphocytes, is generated (see Figure 2, part I). Several authors have reported that HA without WBC differential scattergrams are unable to detect such anomaly (Payne & Pierre, 1984; Lombarts & de Kieviet, 1988). Some very large PLT may be encountered in myeloproliferative and in myelodysplastic disorders, whose size and volume may reach those of WBC and which may be enumerated as WBC: flags are usually generated, that mention the presence of pathological particles that cannot be classified as WBC (usually ‘giant PLT’, or ‘PLT aggregates’, depending on the HA).

Nucleated red blood cells (NRBC) They may be found in the blood stream in physiological (newborns) and in pathological circumstances, and at times they may be much more numerous than WBC. Using HA, these NRBC are in contact with lysis agents that destroy their membrane, leaving nuclei free, the latter being responsible for anomalies. Free NRBC nuclei are usually particles 36 g/dl) corresponding to an erroneously high Hb has been reported for patients with severe constitutional or acquired hypertryglyceridemia (Gagne et al., 1977; Mayan et al., 1996), and for patients receiving intravenous administration of fat emulsions (Nosanchuck, Roark & Wanser, 1974; Shah, Patel & Rao, 1975; Nicholls, 1975, 1977; Creer & Ladenson, 1983; Artiss & Zak, 1987; Sandberg, Sonstabo & Christensen, 1989; Cantero, Conejo &


Jimenez, 1996). After the study of various types of hyperlipoproteinemias, it was observed that erroneous Hb and MCHC >36 g/dl were observed in patients with at least 20 g/l of triglycerids, corresponding to type I and part of type V hyperlipoproteinemias (rich in chylomicrons) but not to type IV hyperlipoproteinemias (rich in very low density lipoproteins; Gagne et al., 1977). Samples taken after a meal may at times demonstrate superimposable spurious Hb measurement. Even the most recent HA are sensitive to hyperlipemia, although to a variable extent, as for example Abbott Cell Dyn 4000 that is defined as giving true Hb values for up to 13 and 9 g/l of triglycerid and cholesterol levels, respectively (M.C. Chrieten, personal communication). Although excess of lipids usually spuriously increases Hb, a spurious fall of Hb was also reported once (Savage, 1989). Similar to the presence of cold agglutinins, anomaly related to lipids is suspected when MCHC is >36 g/dl, or when WBC scattergrams demonstrate high number of particles of low to moderate size (Figure 6). For laser-beam HA that measure Hb level within each RBC and generate the so-called « measured MCHC » or CHCM, a difference between the CHCM and the calculated MCHC is observed. Various methods have been proposed to obviate the abnormality on the relevant sample, including isovolumetrical replacement of hyperlipemic plasma with isoosmotic diluent, or ether extraction of lipids. As previously mentioned, such methods may in turn lead to erroneous PLT and WBC counts.

High WBC counts White blood cell may induce excessive turbidity and disturb Hb measurement if their number is sufficiently high. There is no clear-cut threshold for WBC counts associated with spuriously elevated Hb, but one must be careful with all samples with WBC count over 50 or 100 · 109/l (Cornbleet, 1983; Sandberg, Sonstabo & Christensen, 1989), although some HA define threshold at 250 · 109/l or even seem to be fully insensitive to WBC count because of entire WBC lysis performed before Hb measurement (Sysmex). As for lipid disturbance, CHCM determined on some HA may help to find the true Hb value (McVeigh, Faim & van der Weyden, 1989). However, RBC count may be also disturbed by high WBC counts (see later), and one must pay attention to the limits in recalculating


30

M. ZANDECKI ET AL.


RBC indices, and a packed cell volume (centrifuged Hct) should be considered.

2002). However, spurious Hb measurement is far from being a consistent finding in the presence of cryoglobulins (Fohlen-Walter et al., 2002).

Immunoglobulins They have been reported to interfere with a number of clinical laboratory tests (cf., see Roberts, Fontenot & Lehman, 2000). False elevation of Hb measurement by automated methods was observed for several patients with Waldenstro¨m’s macroglobulinemia and monoclonal IgM, or with multiple myeloma and monoclonal IgA or IgG (Wallis & Ford, 1987; McMullin, Wilkin & Elder, 1995; Roberts et al., 2000). This anomaly is related to high levels of Ig that interact with reagents of the lysis solution. For IgM, this phenomenon has been related to the amount of monomeric component within the circulating paraprotein (Goodrick et al., 1993). HA employing Hb conversion to cyanmethaemoglobin seem to be more affected than others, because of addition of surfactants for cyanide-free methods. Red blood cell count and MCV are unaffected in this situation but, as Hb is overestimated, MCHC usually exceeds 36 g/dl. In order to obtain more accurate results in this instance, it was proposed to determine ‘plasma Hb’ after centrifugation of the sample (which gives turbidity because of the paraprotein) and to withdraw it from Hb of the whole blood (Roberts et al., 2000). For Sysmex instruments, it seems that if the sample is half diluted by the operator before analysis on the HA, the phenomenon does not occur (Roberts et al., 2000). For laser-beam HA that determine the measured MCHC (CHCM) a clearcut difference with the calculated MCHC generates an alarm, and accurate Hb from the sample may be obtained from measured MCHC.

Cryoglobulins Spurious Hb values have been reported in some instances, and several mechanisms are proposed to explain these abnormal findings. In some cases spuriously high Hb values were related to a mechanism similar to that described above for immunoglobulins (Taft et al., 1973; Cornbleet, 1983) or to the disturbance of light transmittance, whereas in other cases a slight decrease of both Hb measurement and RBC count was related to a flow anomaly (Taft et al., 1973; Bremmelgaard & Nygard, 1991; Fohlen-Walter et al.,

Haemolysis Hb free within plasma is measured together with that from the RBC, but its amount ranges from 10 to 40 mg/l in normal conditions and does not affect total Hb measurement. However, in situations related to major intravascular haemolysis, including chemicals, mechanical haemolysis associated with heart valves, and haemolytic anaemias associated with blood transfusion, free plasma Hb may be elevated enough to affect total Hb measurement. MCHC may be >36 g/dl. Centrifuged Hct shows a pink or red plasma tinge, namely if free plasma Hb is >200 mg/l, and is, in some instances, the only reliable RBC parameter. Some laser-beam HA directly determine Hb within each RBC, the Cellular Haemoglobin Concentration Mean (CHCM), which allows the accurate Hb value to be calculated. A short time lapse between venepuncture and analysis is of crucial importance because haemolysis may continue in vitro, leading to a spurious decrease of RBC count and total Hb with a spurious increase of free plasma Hb.

Chemical structure of haemoglobin and bilirubin In physiological situations, Hb is more or less coupled to oxygen or to carbon-di-oxide and, according to which molecule is coupled to Hb, the peak of optimal light absorbance differs slightly. The addition of several reagents leave Hb free from coupled molecules and changes it into one stable molecule (cyanmethaemoglobin is an example), the latter demonstrating a narrow peak of light absorbance, which allows accurate determination of Hb concentration. However, high amounts of carbon monoxide coupled to Hb may not be fully transformed, and in such situations a spuriously high Hb concentration is reported (Cornbleet, 1983; Vinatier & Flandrin, 1993). In contrast, sulfhaemoglobin in high amount has been reported as lowering Hb measurement (Cornbleet, 1983). Bilirubin: although one must pay attention to very high amounts of bilirubin within the plasma, most HA do not presently demonstrate any interference with bilirubin, at least for concentrations up to 2007 The Authors


M. ZANDECKI ET AL.


250 mg/l. Above theses values, however, attention is needed.

MCV, and subsequently lead to abnormal calculated MCHC (discussed later).

Spurious RBC counts and red cell parameters

Giant platelets

According to the technology used, spurious results may or may not be observed using some HA, and some knowledge about the methods analysing blood parameters is necessary. On impedance – type HA an aliquot of the blood sample is diluted isoosmotically and the number and height of electric pulses generated by the electrical resistance of RBC that pass through a small orifice allow the determination of both the RBC count and the MCV. Using laser-beam methods, scattering at least at two angles allows the determination of RBC count and MCV. In order to improve accuracy, pretreatment of RBC with a specific reagent that changes them isovolumetrically from discoid to a sphere is performed on some HA. Setting discrimination thresholds is an important consideration: discriminating the smallest RBC from the largest PLT is at times a challenge, already discussed (part I). On the contrary, even in extreme pathological situations, MCV does not exceed 150–160 fl and, as there is no particle above that size in the blood stream in health or in disease, HA do not analyse any particle above 200–300 fl in volume.

High number of giant PLT may lead to spuriously low PLT count (see part 1) but, as they are enumerated as RBC, they may also affect RBC count in a way similar to that for WBC; however, RBC count is usually only slightly affected in this instance (Cornbleet, 1983; Bain & Bates, 2001).

Spuriously elevated RBC counts High WBC counts Most HA enumerate RBC and WBC together within the same channel(s), and the RBC count reported is the sum of both RBC and WBC counts. In physiological conditions, it is not of any importance, as it corresponds to overestimate RBC count by 0.1% (if we consider a WBC count of 5 · 109/l and a RBC count of 5 · 1012/l). However, high WBC counts (>100 · 109/l) may lead to a significant change in the RBC count, particularly if the patient is also anaemic (Bain & Bates, 2001). Moreover, in the latter instances the MCV reported corresponds to the mean volume of RBC and of WBC from the sample and may be spurious, variably according to the nature and the number of WBC from the relevant sample. So, high WBC counts may induce several abnormal findings, including Hb (discussed above), RBC count,

Spuriously decreased RBC counts Cold agglutinins Cold agglutinins aggregate RBC when the temperature is lower than 37 C. Unsurprisingly, peculiar anomalies of RBC parameters in the presence of cold agglutinins were reported first on counters acting at room temperature (Hattersley et al., 1971; Petrucci, Duanne & Chapman, 1971; Bessman & Banks, 1980). According to the HA, the upper threshold that may consider particles as RBC into the RBC channel(s) is located between 200 and 300 fl. So, only particles corresponding either to isolated RBC or to small RBC clumps (two or three RBC), are analysed, whereas large RBC clumps are fully neglected by the HA. This leads to spuriously low RBC counts and to abnormally high MCV (each small RBC clump is considered as one single particle). Haematocrit (RBC · MCV) is erroneous and spuriously low, contrasting with Hb that is measured after RBC lysis and is unaffected by agglutinins. As a rule, the MCHC is spurious, usually >36 g/dl. The peculiar association of low RBC count, high MCV and MCHC >36 g/dl is almost pathognomonic of the cold induced artifact on instruments that work at laboratory temperature, and not infrequently helps to diagnose the cold agglutinin in the patient analysed (Petrucci, Duanne & Chapman, 1971). HA working with reagents at temperatures near 37 C are not fully insensitive to cold agglutinins, however, but changes are less obvious and may remain undiscovered in some instances (Solanki & Blackburn, 1985). HA that measure directly the amount of Hb within each RBC (measured MCHC, or CHCM) usually show discrepancy between the measured MCHC and the calculated one. Whatever the HA used, the common finding is that Hb


32

M. ZANDECKI ET AL.


value is unaffected and that anomalies disappear when the sample is warmed at 37 C and analysed promptly afterwards. Amplification of the anomalies after the sample has been cooled at 4 C for 1 or 2 h reinforces the diagnosis. As viscosity of the sample may be high, leading to inaccurate aspiration, an alarm may be generated on some HA (insufficient sampling, or related). Co-existence of RBC agglutination with EDTA-dependant thrombocytopenia was reported but antibodies directed against RBC and those directed against PLT differed (Bizzaro & Fiorin, 1999).

Warm autoimmune haemolytic anaemia In some instances warm autoantibodies were also reported as inducing RBC agglutination, leading to spurious MCV and RBC count, in a situation superimposable to that observed for cold agglutinins, but not reversible by warming (Weiss & Bessman, 1984).

In vitro haemolysis As mentioned above, in some situations, namely haemolysis related to chemicals or blood transfusion, RBC may continue to lyse within the sample, leading to spuriously low RBC count, together with abnormal Hb measurement and abnormal red cell indices (see Hb measurement).

Mean cell volume and haematocrit Some situations leading to spurious MCV (see Table 1) are already discussed above [high WBC counts, cold (warm) agglutinins]. Spurious Hct is a frequent finding, related to abnormal MCV or/and to abnormal RBC count as it is a parameter that is usually calculated by the HA: situations leading to abnormal Hct, either increased or decreased, will not be discussed in details as they may be deducted from MCV changes (Table 1).

Very small RBC and discrimination with PLT As discussed in part 1, PLT counts may be disturbed by microcytic cells, namely if RBC volume is 250–300 mg/l) MCH› Haemoglobin: spurious decrease Coagulation within the sample All parameters Overfilling vaccum tube All parameters Veinipuncture near a drip MCV› (glucose drip) Sulfhaemoglobin MCV Cold agglutinins, warm agglutinins: › MCH›, RBCfl, PLTfl High WBC counts: › RBC› Hyperglycemia: › MCH› K2 EDTA in excess: › MCHfl Hyper- or hyponatremia: › or fl MCHfl or › Technology: impedance without hydrodynamic focusing MCH› (MCVfl in hypochromic anaemias) MCH > 36 g/dl (not related to spurious counts in some disorders: spherocytosis, xerocytosis, abnormal Hb: see text) Cold agglutinins, warm agglutinins RBC, MCV› Lipids WBC›, Hb› Immunoglobulins Hb› In vivo and in vitro haemolysis Hb altered, Hct Carboxyhaemoglobin (>10–20%) Bilirubin (>300 mg/l, at times less) Immunosuppressive drugs (see also spuriously high Hb, spuriously low MCV) MCH 35 mmol/l may overestimate MCV up to 50 fl (Holt, DeWandler & Arvan, 1982; Planas, Van Voolen & Kelly, 1985; Van Duijnhoven & Treskes, 1996; Figure 7). Increased MCV leads to an increase in the calculated Hct and to a spuriously low MCHC. CHCM measured on some HA is also altered. CBC demonstrating macrocytic hypochromia may be considered first as related to hyperglycemic sample.

Considerations related to the anticoagulant The use of either K2- or K3-EDTA salt does not induce any difference in the CBC, in optimal conditions of sampling. However, when concentration of anticoagulant is increased, because of insufficient volume of blood drawn after venepuncture (or in 2007 The Authors


M. ZANDECKI ET AL.

neonates), some changes may be observed. Decrease of centrifuged Hct resulting from shrinking of RBC was reported in conditions related mainly to excess of K3 salt (Goossens, Van Duppen & Verwilghen, 1991; Hinchliffe, Bellamy & Lilleyman, 1992), in contrast with data obtained on HA: no influence of K3-EDTA concentration was observed on MCV, while K2-EDTA at high concentrations resulted in a slight increase in MCV, and such phenomenon could be observed using several different HA (Goossens, Van Duppen & Verwilghen, 1991).

Hyper- and hypo-natremia Macrocytic and hypochromic RBC changes were observed on blood samples in conditions related to hypernatremia, whereas hyponatremia generated a tendency towards microcytic and hyperchromic RBC (Cornbleet, 1983). Such situations were also reported in animals (Boisvert, Tvedten & Scott, 1999). Hyperand hypo-natremia are both situations also reported as leading to spurious centrifuged Hct (Cornbleet, 1983).

Storage of the sample and MCV EDTA used as anticoagulant allows the accurate determination of the CBC up to 24 h after the sample has been drawn. However, after that time, MCV may increase, namely if the sample is stored at room temperature. This coupled with low haemoglobin could cause the operator to suspect that a slightly microcytic anaemia is normocytic and that a normocytic could be mistaken for a macrocytic anaemia (Cohle, Saleem & Makkaoui, 1981).

Mean cell haemoglobin content (MCHC) Measured parameters allow the calculation of mean cell haemoglobin of individual RBC [MCH as expressed in pg ¼ Hb (g/l)/RBC (1012/l)] and MCHC [MCHC in g/dl ¼ Hb (g/l) · 100/MCV (fl) · RBC (1012/l)]. Some HA measure Hb concentration directly within RBC, named cellular Hb concentration mean (CHCM): discordant values (usually difference over 1.5 g/dl) between MCHC and CHCM allow in many instances the detection of anomalies related to one of the measured RBC parameters.


MCHC >36 g/dl is infrequent on most impedancetype HA, whereas it may be occasionally observed on laser-beam HA, in several constitutive disorders in which RBC are ‘dehydrated’, including hereditary spherocytosis, various haemoglobin disorders (CC, SC, Cb thal), and some rare RBC disorders (xerocytosis). Some acquired conditions mimic constitutive ones, namely acquired immune haemolytic anaemias caused by warm agglutinins, in which RBC coated with warm antibodies may transform gradually into spheres in vivo or in vitro after venepuncture, leading to more or less dehydrated and spherised cells. Many situations corresponding to abnormally high Hb values, and/or abnormally low RBC, and/or spuriously low MCV, also lead to increase MCHC (see the corresponding paragraphs and Table 1). Although the mechanism is unknown, some immunosuppressive drugs may slightly increase MCHC, usually not above 37.5 g/dl (Cornbleet 1983). MCHC