Blood volume and red cell mass in children with moderate and severe malaria measured by chromium-53 dilution and gas chromatography/mass spectrometric analysis


  • Derek C. Macallan,

    Corresponding author
    1. Division of Cellular and Molecular Medicine, Centre for Infection, St. George's Hospital Medical School, London SW17 0RE, UK
    • Centre for Infection, St George's, University of London, Cranmer Terrace, London SW17 0RE. UK.
    Search for more papers by this author
  • Daniel A. Abaye,

    1. Division of Cellular and Molecular Medicine, Centre for Infection, St. George's Hospital Medical School, London SW17 0RE, UK
    Current affiliation:
    1. Stable Isotope Biochemistry Laboratory, Scottish Universities Environmental Research Centre, Rankine Avenue, East Kilbide G75 0QF, UK.
    Search for more papers by this author
  • Simone Dottin,

    1. Division of Cellular and Molecular Medicine, Centre for Infection, St. George's Hospital Medical School, London SW17 0RE, UK
    Search for more papers by this author
  • Myriam Onanga,

    1. Département de Parasitologie, Mycologie et Médecine Tropicale, Faculté de Médecine et des Sciences de la Santé, Libreville, Gabon
    Search for more papers by this author
  • Maryvonne Kombila,

    1. Département de Parasitologie, Mycologie et Médecine Tropicale, Faculté de Médecine et des Sciences de la Santé, Libreville, Gabon
    Search for more papers by this author
  • Arnaud Dzeing-Ella,

    1. Département de Parasitologie, Mycologie et Médecine Tropicale, Faculté de Médecine et des Sciences de la Santé, Libreville, Gabon
    Search for more papers by this author
  • Peter G. Kremsner,

    1. Medical Research Unit, Albert Schweitzer Hospital, Lambaréné, Gabon
    2. Sektion Humanparasitologie, Institut für Tropenmedizin, Universität Tübingen, Wilhelmstraße 27, 72074 Tübingen, Germany
    Search for more papers by this author
  • Sanjeev Krishna,

    1. Division of Cellular and Molecular Medicine, Centre for Infection, St. George's Hospital Medical School, London SW17 0RE, UK
    2. Medical Research Unit, Albert Schweitzer Hospital, Lambaréné, Gabon
    Search for more papers by this author
  • Timothy Planche

    1. Division of Cellular and Molecular Medicine, Centre for Infection, St. George's Hospital Medical School, London SW17 0RE, UK
    2. Medical Research Unit, Albert Schweitzer Hospital, Lambaréné, Gabon
    Search for more papers by this author

  • Presented at the 2nd Joint European Stable Isotope User Meeting (JESIUM), Presqu'île de Giens, France, 31 August–5 September, 2008.


Understanding blood volume changes in children with malaria is important for managing fluid status. Traditionally, blood/red cell volume measurements have used radioactive chromium isotopes. We applied an alternative approach, using non-radioactive chromium-53 labelling and mass spectrometry to investigate red cell volume (RCV) in Gabonese children with malaria. Nineteen children with malaria participated (10 severe, 9 moderately severe; ages 15 months to 7 years). Blood labelled with 53Cr-chromate exvivo was re-injected, then sampled 30 min later. Pre- and post-injection 53Cr content were measured by gas chromatography/electron ionisation mass spectrometry of the chromium-trifluoroacetylacetone (TFA) chelate, calibrated against 50Cr standards. Blood and red cell volumes were calculated from isotopic dilution in 15 of 19 children (in four, insufficient signal mitigated analysis). In this small pilot study, there were no significant differences between moderate and severe cases. Including all subjects, the mean RCV was reduced compared with predicted values (184 vs. 269 mL; p = 0.016) but blood volume, 71 ± 33 mL/kg (normalised for weight), was close to predicted, ∼77 mL/kg, commensurate with reduced haematocrit. Blood lactate concentration correlated negatively with RCV/weight (r = −0.56, p = 0.028), consistent with anaemia. In one case, sequential samples over 42 days gave an estimated rate of 53Cr disappearance of 1.4%/day (equivalent half-life: 70 days). 53Cr-labelling of red cells may be used to estimate blood and red cell volumes and can be used as an investigative tool in situations such as childhood diseases and resource-constrained settings. Although the red cell mass is depleted in malaria, the blood volume appears relatively well preserved. Copyright © 2009 John Wiley & Sons, Ltd.

Appropriate fluid management of children with malaria is critical for optimal supportive care and depends on knowledge of the changes in fluid balance and red cell mass that occur during disease. Clinical signs in children frequently suggest volume depletion,1, 2 in the acute phase of infection and, aggressive volume replacement/support is advocated in management,3 especially in febrile children with poor fluid intake. However, direct measurements to guide fluid replacement strategies are lacking.

The primary goal of optimising fluid volume status is to maintain good tissue perfusion. However, hypovolaemia is only one contributor to tissue hypoxia and thus acidosis. Local hypoperfusion may also result from microvascular changes such as obstruction from infected erythrocytes4, 5 and acidosis may have other causes.6 Local hypoperfusion and acidosis may therefore be misleading if used as markers of the degree of hypovolaemia.

Fluid volume changes in malaria are complex and estimating fluid volumes in sick children is very challenging for clinicians who frequently rely on indices such as tachycardia, prolonged capillary refill times, central venous pressure (if available) and decreased urine volume to guide them.2, 7–9 However, clinical assessment of fluid status is difficult and imprecise. Central venous pressure (CVP) measurements in children are inconsistent; some suggest that blood volume is relatively well preserved in severe malaria,10–12 whilst others suggest a reduction in CVP.13 Recent studies by Planche et al. using heavy water and sodium bromide dilution to measure total body water (TBW), extracellular water (ECW) and intracellular water (ICW) volumes suggest that the degree of depletion of TBW in severe childhood malaria is surprisingly small (mean loss 6%) and that ECW volumes may be relatively well conserved compared with convalescent values.11 However, such measurements do not directly reflect changes in the distribution of fluid between intravascular and extravascular compartments which are important determinants of tissue perfusion.

Within the intravascular compartment, blood volume represents the sum of red cell volume and plasma volume, and these are affected differentially by malaria.14 Red cells are sequestered and destroyed and may decrease red cell mass sufficiently to cause anaemia. Blood volume may be maintained by an increase in plasma volume, but only at the expense of reduced haematocrit. Plasma volume itself may, however, be lost by transcapillary leakage but the pathophysiology of malaria does not follow the classical model of sepsis in which dramatically increased capillary permeability results in loss of intravascular volume and expanded ECW. Although there is some evidence from measurements of urinary albumin/creatinine ratios and transcapillary escape rates for radiolabeled albumin for an association between a generalsed increase in systemic capillary permeability and markers of disease severity,10 such changes are relatively small and these data, along with data from fluorescein angiography,15 suggest that microvascular changes may be more important than volumetric changes in the generation of complications such as coma.

Direct estimation of blood volume is difficult but combining measurements of red cell mass and haematocrit allows such an estimation to be made. Previous methodologies have relied on radioactive tracers.16–19 More recently, measurements of red blood cell mass and blood and plasma volume have been made using non-radioactive stable isotopes of chromium measured by isotope ratio mass spectrometry.20–22 This approach involves no exposure to radioactivity and, therefore, has major advantages in terms of patient safety and reagent handling which mean that it can be applied to vulnerable groups such as pregnant women21 and children, and access to radio-isotope sources is not an issue.

We applied this technique to measure red cell mass and blood volume in children with moderate and severe malaria in Gabon. Our aims were four-fold; first, to develop and optimise this approach for use in such a setting; secondly, to evaluate changes in these parameters in children with malaria; thirdly to determine if there was an association between severity of disease and changes in red cell and blood volumes; and, finally, to test if longitudinal studies might allow investigation of the disappearance kinetics of labelled red cells.


Subjects and clinical assessment

The study was conducted at the Albert Schweitzer Hospital, Lambaréné, Gabon. All study procedures were approved by the ethics committees of the International Foundation of the Albert Schweitzer Hospital and the Gabonese Ministry of Health and informed consent was obtained from all parents.

Nineteen children (aged 1 to 8 years) with parasitologically confirmed malaria (defined as the presence of asexual forms of Plasmodium falciparum in thick or thin blood films) entered into the study. They formed a subgroup of a previously described cohort.11 Severe and moderate cases were defined according to blood lactate and glucose concentrations, together with conscious level; severe malaria was defined by one or more of: blood lactate >5 mmol/L, blood glucose ≤2 mmol/L, Blantyre coma score (BCS) ≤2, or repeated, observed seizures. Moderate malaria was malaria without any of the features of severe malaria as above but with a requirement for parenteral treatment because of one or more of the following: a history of frequent (>2) and recent vomiting (within 12 h), drowsiness, obtundation, or prostration.8 Alternative diagnoses were excluded clinically. Children were admitted and managed for malaria according to standard protocols as described elsewhere.11

Study procedures

Labelling studies were performed as follows: On admission, a 1 mL sample of blood was taken and stored for analysis of baseline chromium-53 (53Cr) enrichment. A further aliquot of about 3–5 mL (0.2 mL blood/kg body wt + 1 mL) was mixed with acid citrate dextrose (ACD) then slowly injected into a bottle containing approximately 0.5 mL solution of 53Cr-labelled hexavalent sodium dichromate (prepared with 5–6.5 µg of 53Cr Namath imageCr2O7; isotopic purity = 98.23%; Medical Isotopes Inc., Pelham, NH, USA). The blood-Namath imageCr2O7 mixture was incubated at ambient temperature for 30 min, mixing at 5 min intervals, then washed three times with sterile 0.9% saline before being re-suspended in 3.5 mL saline. An aliquot of this fraction of labelled red cells (1 mL) was retained for analysis and the rest re-injected into the subject. After 30 min for equilibration, a further 1 mL sample of blood was taken from the subject for the measurement of post-labelling 53Cr enrichment in peripheral blood. Samples were stored at −20°C until analysis.


The analysis of 53Cr content was based on the approach described by Veillon and co-workers.20, 22 Briefly, blood samples were thawed then aliquoted into washed glass tubes containing magnesium nitrate (0.186 g/mL, 200 µL) and a ‘spike’ of internal standard 50Cr-labelled chromium chloride 50CrCl3 (50Cr 94% purity, Oak Ridge National Laboratory, Oak Ridge, TN, USA; a gift from Claude Veillon, Human Nutrition Research Center, Beltsville, MD, USA). In order to calculate accurately the volume of blood added to the spike, tubes were weighed before and after addition of blood and the weight converted into volume using the formula:

equation image

where hct is the measured haematocrit, and drbc and dplasma are the average densities of red cells and plasma (taken to be 1.095 and 1.020 g/mL, respectively23). The amount of internal standard was adjusted according to the expected amount of 53Cr being analysed. In validation experiments, 10, 100 or 1000 ng aliquots were added; for the labelled red cell fraction (pre-injection), 1000 ng was used, and for baseline and post-injection peripheral blood, 100 ng was used. Samples were freeze-dried, dry-ashed in a muffle furnace and then wet-ashed with nitric acid, hydrogen peroxide and hydrochloric acid as described by Silver et al. and by Veillon et al.20, 22 Upon cooling, ammonium acetate buffer and trifluoroacetylacetone (TFA; alternative name, 1,1,1-trifluoro-2,4-pentanedione; Sigma, Gillingham, UK) were added and the samples incubated at room temperature overnight to chelate iron. The iron-TFA complex was removed to avoid interference with the chromium chelate by extraction with hexane;20, 22 this was repeated until the brick-red colour had disappeared, optimum results being obtained after six 0.5 mL washes, Chromium was then chelated by addition of a further aliquot of TFA and incubation at 70°C for 90 min. The chromium-TFA chelate was extracted in hexane, washed in dilute NaOH to remove any acid residue, dried in nitrogen and re-suspended in hexane for analysis by gas chromatography/mass spectrometry (GC/MS).

GC/MS analysis and mass isotopomer calculations

Analysis of isotope abundance was performed using a single-quadrupole GC/MS instrument (Agilent 6890/5793 GC/MS, AgilentTechnologies, Bracknell, UK) with a non-polar column (ZB-5; 5% polysilarylene, 95% polydimethylsiloxane, 30 m × 0.25 mm × 0.25 µm; Phenomenex, Torrance, CA, USA), in electron ionsation (EI) mode, monitoring ions at m/z 356, 358, and 359 by selective ion monitoring (SIM). These three ions represent the dominant ions from the chromium-TFA derivative containing 50Cr, 52Cr and 53Cr, respectively. Relative ion abundances were resolved to yield the relative abundances of natural chromium (N), ‘spike’ (internal standard) chromium (S), and ‘labelled’ chromium (L) by means of an inverse matrix, similar to the approach described by Bluck et al.24 This was achieved as follows. First, the theoretical isotopomer distribution for each mixture (N, S, and L) was calculated as the sum of two components:

  • (1)the isotopomer distribution of the molecular moiety excluding the chromium ion (C10H8O4F6), assuming natural isotopic abundances for carbon, hydrogen, oxygen and fluorine, and
  • (2)the abundance of each chromium isotope, taken as:
    • (i)for natural chromium: 50Cr 4.35%, 52Cr 83.79%, 53Cr 9.50% and 54Cr 2.37%;25
    • (ii)for the spike chromium: 50Cr 96.79%, 52Cr 2.98%, 53Cr 0.18%, 54Cr 0.05%;26
    • (iii)and for the label: 50Cr 0.0008%, 52Cr 1.6%, and 53Cr 98.23% (Medical Isotopes Inc.), as shown in Table 1(a).
Table 1. Calculation of Isotope abundances for chromate isotopomers
(a) Isotopic composition of chromate
Source of chromiumProportion
 Chromium 500.04350.96790.0008
 Chromium 520.83790.02980.0164
 Chromium 530.09500.00180.9823
 Chromium 54*0.02370.00050.0005
(b) Theoretical proportion of each isotopomer generated
Source of chromiumNaturalSpikeLabel
Isotopomer (m/z)
(c) Conversion factors: composition of chromium from ion abundances [inverse matrix of (b)]
Ion (m/z)356358359
Source of chromium
(d) Calculation of proportion of source of chromate from distribution of ions measured
  • *

    54Cr is included for completeness in the composition data (a), but does not contribute to ions detected below m/z 360 (M+4), so does not impact the calculations in (b)–(d) which only include the three ions shown. If data were collected over an extended range (e.g. m/z 356–360 inclusive), an extended inverse matrix could be used, obtaining the least-squares solution to an over-determined set of equations. The parameters in (d) represent the values given in (c). A worked example is given in the Supporting Information.

Ions (m/z)358356359 
Measured abundance of ionsxyzCalculated composition of chromium
Source of chromium    
 Naturalabc= ax + by + cz
 Spikedef= dx + ey + fz
 Labelledghi= gx + hy + iz

Combining these abundances allows estimation of the theoretical proportion of each ion generated by EI from each isotope of chromium when ionised as the TFA-chelate (Table 1(b)). Table 1(b) forms a 3 × 3 matrix, giving nine constants which together describe the contribution of each mixture to each ion. In order to make the reverse calculation, i.e. to estimate unknown proportions of chromium isotopic mixtures from the measured abundances of three ions, an inverse matrix was calculated using the MINVERSE array function (Excel, Microsoft, Seattle, WA, USA) as shown in Table 1(c). This gives multiplication factors from which the relative contribution of the three mixtures can be calculated (Table 1(d)).

This approach was validated using a series of test solutions consisting of the spike (50CrCl3), the label (Namath imageCr2O7) and natural chromate (Na2Cr2O7.2H2O) prepared singly at concentrations of 10, 100 and 1000 ng/mL and in combinations designed to estimate the sensitivity of detection of 53Cr and the confounding effect of variations in the background.

Measurement of plasma and red cell volumes

From the relative abundances of the three ions m/z 356, 358 and 359, the relative molar abundances of the three chromium mixtures (natural, spike and label) were calculated for each sample using the matrix above. Since the amount of 50Cr spike mixture was known, the other two components could be calculated from their respective ratios. The enrichment of the 53Cr label in each labelled sample was then corrected for baseline by subtraction of the amount of 53Cr label calculated to be present in the pre-labelling blood sample from the same subject. This was then expressed as concentration of 53Cr in the blood sample by dividing by the volume of blood to which the spike had been added.

Red cell and blood volumes were calculated as follows: Let L be the amount of 53Cr label; C, the concentration of label on red cells; V, the volume of red cells; and hct haematocrit, where subscript inf refers to infusate, wb to whole body blood and pb to peripheral blood. The amount of label, L, remains constant, i.e. Linf = Lwb.

The first of these terms, Linf, is given by: Linf = Vinf × Cinf = [Vinf/hctinf] × [Cinf × hctinf]. [Vinf/hctinf] is the total volume of the infusate, which was derived from the weight (g) of the syringe before and after injection divided by blood density, adjusted for haematocrit (hct), and [Cinf × hctinf] is the concentration of 53Cr in whole infusate, which was measured.

Since the second term, Lwb, is given by: Lwb = Vwb × Cwb, then, substituting, red cell volume (RCV), Vwb = Vinf × Cinf/Cwb.

We assume that red cells are fully mixed. Thus, Cwb = Cpb, which was measured, either directly by analysing packed cells from centrifuged post-labelling blood or by analysing whole peripheral blood and dividing by hctpb; in some children both analyses were performed and the mean used.

Blood volume (BV) may be estimated from RCV and hct but consideration must be given to the fact that the haematocrit of blood averaged throughout the whole body is less than that of peripheral blood,27–29 the ratio being known as the F-cell ratio (F), thus hctwb = hctpb × F. Hence BV = RCV/[hctpb × F].

The values of F that we used (0.83 for boys and 0.85 for girls) were derived from the largest cohort of children studied to date,29 although the assumptions inherent in correcting using F may not be directly applicable in malaria, because of sequestration.14 Results were expressed as means and standard deviations and groups compared by unpaired, two-tailed Students t-test. A worked example is given in the Supporting Information.


Validation and calibration of isotopic labelling

Although this approach has been used previously,20, 22 in view of the substantial modifications needed for application in the context of children with malaria in Gabon, it was deemed necessary to validate the measurement of 53Cr enrichment by this approach. When standard solutions containing varying amounts of 53Cr-chromate were analysed it was found that there was good agreement between the measured and actual amounts of 53Cr (Fig. 1). 53Cr levels could be reliably measured down to amounts of 10 ng with a coefficient of variation (cv) of less than 10% with either 1000 ng or 100 ng internal standards; as expected, reliable measurements of smaller amounts of 53Cr were best achieved with 100 ng of 50Cr internal standard (Fig. 1(b)), rather than 1000 ng (Fig. 1(a)), the cv at levels of 2 ng using a 100 ng spike being about 12% (n = 10).

Figure 1.

Validation of 53Cr estimates in labelled samples. Estimation of 53Cr-chromate concentrations in standard solutions using a 1000 ng (a) and 100 ng (b) 50Cr internal standard. Scales are logarithmic. (c) Effect of adding increasing amounts of natural chromate as background on estimation of 53Cr-chromate content of standard samples at four different levels of 53Cr enrichment.

As one of the major difficulties inherent in these analyses is the varying abundance of background chromium in the samples, standard curves were repeated with differing amounts of unlabelled, natural chromate added to investigate any confounding effect that this might have. It was found that, with the use of an internal standard and the inverse matrix approach discussed above, the amount of 53Cr could be measured independently of the amount of natural chromate present (Fig. 1(c)).

Subject characteristics

Nineteen subjects entered the study, of whom ten (3M: 7F) were designated as having severe malaria (FSS) and nine (3M: 6F) as having moderate disease (FSM) (Table 2). The ages ranged from 1 year 3 months to 7 years 4 months. Demographics suggested that children with moderate and severe malaria were drawn from similar groups in terms of height, weight and age. As expected, children in the severe malaria group had a significantly lower Blantyre coma score (BCS), median score 2, than those in the moderate group, median 5 (p ≤ 0.001, Mann-Whitney test). Similarly, the mean temperature and pulse rate were higher in these children, consistent with more severe disease (Table 2), although higher parasitaemia, lower packed cell volume and higher glucose and lactate were not significantly different between groups containing relatively few subjects (Table 2).

Table 2. Clinical details of subjects
IdentifierAgeSexHeightWeightBCSLactateGlucoseTempHeart rateSys BPDia BPPCVParasitaemia
mo cmkg mmol/Lmmol/L°Cmin−1mmHgmmHg% 
  • Abbreviations: mo, months; Sys BP, systolic blood pressure; Dia BP, diastolic blood pressure; PCV, packed cell volume.

  • Means were compared by Students t-test (two-tailed) except for non-parametric data,

  • *

    where median is shown and comparison was by Mann Whitney test.

FSM1142f9212.752.35.138.712060 1981,138
Mean moderate36 9011.85*2.45.337.8135864926127,062
SD22 153.6,538
Mean severe29 8811.52*4.57.738.9161985521343,102
SD12 82.3,812
P value0.41 0.660.84<0.01*
Mean all32.32 8911.643.56.538.4149925223240,767
SD all16.95 122.922.73.51.12414117313,533

Red cell and blood volumes

Estimation of blood and red cell volumes was possible in eight children with severe and seven children with moderate malaria as shown in Table 3. In four others technical difficulties, inadequate amounts of incorporated 53Cr and/or missing samples prevented complete analysis. The average concentration of 53Cr in the infused red cells was 124 ng/mL, corresponding to re-injection of about 400 ng on average (mean injection volume was 2.9 mL). Retrospective analysis of an aliquot of labelled chromate yielded a concentration of 1.4 µg/mL, which was considerably less than expected (samples were prepared with 5–6.5 µg in 0.5 mL so it should have contained about 10 µg/mL). As a consequence, only about 300 ng (range 177–435 ng) was re-injected and this resulted in labelling rates in post-infusion blood samples much lower than expected with a mean value of ∼0.7 ng/mL (range 0.2–2.2 ng/mL).

Table 3. Blood and red cell volumes from 53Cr abundance in children with malaria
IdentifierPeripheral bloodBody weightInfusatePost-infusionRed cell volumesBlood volumes
PCVkg53Cr concPCVVolumeAmount 53Cr53Cr concTotalPredictedper unit weightTotalPredictedper unit weight
ng/mL mLngngmLmLmL/kgmLmLmL/kg
  • *

    p < 0.05 versus predicted values by paired Students t-test.

  • missing sample or data; mean group value used.

    Abbreviations: conc, concentration; PCV, packed cell volume or haematocrit.

Moderately severe malaria
 Mean moderate0.2612.31230.182.843200.4923228517.595393475
  SD moderate0.093.3400.041.19880.271936810.857723730
Severe malaria
 Mean severe0.2110.931220.133.033570.80142*25512.474083167
  SD severe0.072.2150.040.55510.58110428.442215738
  P value (t-test)0.220.340.940.0520.700.330.220.280.310.320.420.330.66
 Mean all0.2311.581220.152.943400.66184*26914.883987971
  SD all0.082.76300.050.88710.48156569.649419833

The mean red cell volume (RCV) for all children with malaria was estimated to be 184 mL, equivalent to 14.8 mL/kg (Table 3). Although there was a large degree of inter-individual variation, compared with the predicted values (269 ± 56 mL),29 red cell volumes were significantly depleted in subjects with malaria (p = 0.016, paired t-test). When analysed by subgroup, this difference was significant for those with severe malaria (p = 0.011; Table 3) but not for those with moderate disease (p = 0.38), although the direct comparison for RCV between severe cases (12.4 mL/kg) and moderate cases (17.5 mL/kg) was not significant in this small sample (Fig. 2(a); p = 0.32). The main predictor of RCV (normalised for weight), not surprisingly, was PCV (Fig. 2(b); r = 0.81). Interestingly, there was very little correlation between RCV and TBW (r = 0.24, p = 0.42) or ECW (Fig. 2(c); r = 0.15, p = 0.61).

Figure 2.

Estimates of blood volume and red cell volume in children with moderate and severe malaria. For all graphs open diamonds represent children with moderate malaria and closed diamonds represent severe cases. (a) Effect of malaria on measured red cell volume, normalised for body weight (RCV) in children with moderate and severe disease; difference is not significant (p = 0.33 by two-tailed t-test); (b) Haematocrit (PCV) is a strong predictor of RCV, r = 0.81, but (c) there is no significant correlation between RCV and extracellular water volume (ECW, from Planche et al.,11 r = 0.15, p = 0.61). (d) Comparison of the effects of moderate and severe malaria on estimated blood volume, normalised for body weight, shows no significant difference between groups. (e) Relationship of measured to predicted blood volume29 shows significant correlation (r = 0.69, p = 0.005; dashed line shows line of unity). (f) There is a significant negative correlation between blood lactate on admission and RCV (r = 0.56, all subjects; p = 0.028).

The estimated mean blood volume for the whole group of subjects was 71 mL/kg body weight (Table 3). Although there was a large degree of inter-individual variation (SD 33 mL/kg), these values were close to predicted values using gender-specific values of 78.5 and 74.9 mL/kg body weight.29 Values in moderate and severe cases overlapped (Fig. 2(d)) and there was no significant difference between the groups in terms of blood volume (p = 0.66). As expected, there was a correlation between measured and predicted values (Fig. 2(e), r = 0.69, all subjects), although some subjects did appear to have very low red cell and blood volumes. Those with the lowest red cell volumes tended to have the highest lactate concentrations (Fig. 2(f), r = −0.56, p = 0.028), whereas no significant correlation was found between lactate concentrations and blood volume, or with TBW or ECW.

Kinetics of erythrocyte loss

In one subject with severe malaria (FSS2), it was possible to obtain and analyse repeated blood samples over a 6-week period. Enrichment of red cells with 53Cr was found to fall progressively with time. The loss of 53Cr modelled best as a mono-exponential fall with a disappearance rate constant of −0.014 ± 0.004 day−1 (Fig. 3), after correcting for an assumed elution rate for 53Cr chromate of about 1%/day based on observations with 51Cr chromate.30 This is equivalent to a mean cell survival of about 70 days.

Figure 3.

Disappearance of 53Cr-labelled red blood cells following severe malaria. Amount of 53Cr in peripheral blood red cell fractions following labelling during an episode of severe malaria in one subject (FSS2). Data are expressed as log10 of 53Cr-chromate concentration; error bars represent analytic variance of triplicate measurements of isotopic enrichment expressed as ± 1 SD, except for day 14 where only one measurement was available. The slope (−0.0139 day−1) gives the disappearance rate constant, equivalent to a mean cell survival of about 70 days.


We have shown in this study that mass spectrometric analysis of 53Cr-chromate-labelled red blood cells allows estimation of blood and red cell volume in children with malaria. This approach is adapted from one previously used in adults.20, 22 Its main advantages over previous methods are (1) safety for the subject and (2) ease of handling, since no radioactivity is involved. The procedure is relatively non-invasive, involving cannulation for venesection and re-injection of labelled blood; less than 5 mL blood is adequate for analysis.

In terms of clinical implications, because of large inter-individual variations and a relatively small sample size, this study should be considered a pilot study and cannot guide therapeutic decisions. However, several observations can be made. First, red cell volumes were lower than predicted values; secondly, comparing children with moderate and severe disease, the main difference appeared to be in red cell volume, in contrast to blood volume; thirdly, blood volumes appeared relatively well preserved, consistent with a compensatory increase in plasma volume; fourthly, there was a correlation between low red cell volume and raised blood lactate concentration, consistent with the known association between anaemia and hyperlactataemia; finally, red cell survival may be shortened in the recovery phase from malaria, although this was only measured in one subject.

The main limitation of this study was large inter-individual variation. Some of this is, of course, physiological, but some variability also arose because the low levels of 53Cr labelling achieved compromised the analytical precision; further studies should therefore be performed with increased doses of 53Cr. Accuracy is also highly dependent on the measured haematocrit. Residual plasma in packed cells would bias results slightly. Some authors have corrected for ‘trapped plasma’, using a factor of 0.96–0.97;31 this is not always done27, 32 and may be less reliable at low haematocrits. We chose not to apply such an uncertain correction factor in this study. In some subjects blood was spun and plasma removed prior to 53Cr analysis; this may also have introduced some variability. When we analysed the same sample as both whole blood and packed cells agreement was generally good (n = 4; r2 = 0.84) but it is probably better to analyse only whole blood for 53Cr content. Inclusion of magnesium nitrate solution with analytical samples is important to optimise chromate yield. Adsorption onto glass may explain the surprisingly low amounts of chromate actually infused, which were lower than intended; some loss by adsorption onto the glass vial may have occurred as no magnesium nitrate was included in the vial containing the chromate prior to and during blood labelling. The use of an internal standard (50Cr) is critical to quantify background natural chromate, which contains 9.5% 53Cr, and the combination of analysis of three ions with a mathematical matrix to deconvolute the isotopomer distribution enables compensation for varying amounts of chromate across a very wide range (Fig. 1(c)). This paper describes a novel application of such reverse matrix modelling, although similar matrix approaches have been validated elsewhere.24

The procedure is safe in terms of potential toxicity; the dose of chromium given, <1 µg, is well below the range of potential toxicity, being at least one order of magnitude less than the estimated normal dietary intake of chromium in the USA (range 25–224, mean 76 µg/day33). Although hexavalent compounds may be more toxic, chronic oral intake of 3 µg chromium(VI)/kg/day is considered safe.34, 35 Previous studies have safely used much higher doses; for example, in one study, 1200 µg of 53Cr was given over three days to lactating women.36 Of course, the absence of radioactivity is the main advantage over 51Cr-based approaches.

The intention of this study was not to make comparisons with normal values and no ‘control’ (non-parasitised) group was included. Normative data for children are sparse37 and may be limited by the use of assumed F-cell ratios from adults. Our predicted data and the F-cell ratio that we used to calculate blood volume were derived from a cohort of ‘healthy’ children in Belgium (in remission from nephrotic syndrome) in whom both 51Cr- and 131I-albumin studies had been performed.29 This is probably the best available dataset but it may not be directly comparable with African children in Gabon, particularly as malaria itself may influence whole body haematocrit values by increasing the volume of sequestered red cells.14 Extrapolating from red cell volume to blood volume, as we have done, introduces further uncertainties but avoids the administration of substances such as Evans blue or 131I-albumin for the direct measurement of plasma volume. In addition, as falciparum malaria is a disease which causes sequestration of red cells, there may be a difference in values for F between severe and moderate cases of malaria. In terms of normalisation, we used body weight as, unlike Raes et al.,29 we did not find such good correlations with lean body mass (LBM) or body surface area, possibly because LBM algorithms do not apply well to this age/population group.

This approach also gives an opportunity to investigate red-cell kinetics from post-labelling dynamics of 53Cr-labelled red cells. Our data from one child gave a mean life-span of about 70 days. It is not possible to comment on a value from a single subject; the data is shown to demonstrate feasibility. However, previous estimates are in the same range: in one group of adult/adolescent patients in Thailand with falciparum malaria, a mean value of 57 days was obtained (range 30–66),38 while more severe patients had shortened survival, 44 ± 22 days;39 modelled estimates, accounting for sequestration, gave a value of about 62 days,14 while in healthy controls, the mean lifespan was 90 days.38


This study demonstrates that the use of 53Cr to measure red cell and blood volume is feasible as an investigative tool in children with malaria. It is safe, has good acceptability and involves no radiation exposure, although the analysis is considerably more complex and time-consuming than for radioactive 51Cr. Because of these complexities, it would not be suitable for use as a routine diagnostic test, but is useful for investigating disease pathophysiology, particularly in situations where radiation is undesirable. In this study, large inter-individual variation may reflect analytical variance; a larger study with more subjects using a larger dose of 53Cr for labelling would yield very useful data to guide therapy of childhood malaria. Comparable studies in convalescent or healthy children might be considered and further studies of the disappearance kinetics of labelled cells would also be of considerable interest in understanding the residual effects of post-malaria hypersplenism and how the red cell mass is repleted in the recovery phase. Initial results suggest that the major depletion in childhood malaria occurs in the red cell compartment and that compensatory increases in plasma volume largely make up for this volume deficit. In view of this, adequate but careful fluid resuscitation is clearly of primary importance in the management of this disease.


Additional supporting information may be found in the online version of this article.


We are grateful to the patients and their families at Albert Schweitzer Hospital, Lambarene, Gabon, who willingly participated in this study. We thank Katja Engel, Jean Francois Faucher and other member of staff for their work on this project. We acknowledge financial support from the Charitable Trustees of St George's Hospital (to DM, SD and DA). We also thank Dr Andy Jewell for assistance in processing samples and David and Jan Iliffe for advice on use of inverse matrix calculations.