Journal of Immunological Methods 311 (2006) 207–219www.elsevier.com/locate/jim

Research paper

Simultaneous determination of membrane CD64 and HLA-DRexpression by blood neutrophils and monocytes using themonoclonal antibody fluorescence capability of a routine

haematology analyser

Wim van der Meer a, Ludi van Dun b, Jacqueline Klein Gunnewiek a,Bodo Roemer c, Colin Stephen Scott c,⁎

a Department of Clinical Chemistry, Radboud University Nijmegen Medical Center, P.O. Box 9101, 6500 HB Nijmegen, The Netherlandsb Abbott BV, Siriusdreef 51, NL2132 WT Hoofddorp, The Netherlands

c Abbott GmbH and Co KG, Diagnostics Division, Max-Planck-Ring 2, 65205 Wiesbaden–Delkenheim, Germany

Received 26 April 2005; received in revised form 6 February 2006; accepted 15 February 2006Available online 13 March 2006

Abstract

This study reports the design of an immunofluorescent method for the co-determination of neutrophil CD64 (PMN-CD64),monocyte CD64 (MON-CD64) and monocyte HLA-DR (MON-Ia) expression with the Cell-Dyn CD4000 haematologyanalyser. Normal PMN-CD64, MON-CD64 and MON-Ia expression, defined as the mean±2SD of 25 healthy adults aftercorrection for isotype control staining, corresponded to 17–67, 515–1045 and 170–670 AFU respectively. Analyticalreproducibility determined by duplicate analysis of 12 random samples revealed good assay consistency for all three analysedantigens, with day to day variation in normal subjects being relatively minor in significance. CD4000 PMN-CD64 and HLA-DR values showed good inter-method correlation with flow cytometry although short term (12 h) stability studies suggestedan in vitro trend for increasing PMN-CD64 and variable HLA-DR antigen expression with progressive storage. Observedranges of PMN-CD64, MON-CD64 and MON-Ia for 109 randomly-selected clinical samples were 31–1058, 307–2843 and10–876 AFU. Abnormal PMN-CD64 and MON-CD64 shared the same trend (upregulation) while abnormal monocyte MON-Ia was characterised by declining expression. Normal PMN-CD64 was only seen with normal (45/52) or intermediate (7/52)MON-CD64, while high PMN-CD64 was mostly associated with intermediate (18/22) or high (3/22) MON-CD64. MON-Iaexpression was largely independent (p=0.04) of PMN-CD64 although marked decreases in MON-Ia were invariablyassociated with intermediate or high PMN-CD64. MON-Ia expression was inversely related (p<0.0001) to absolutegranulocyte counts, and patients with high PMN-CD64 were more likely (8/25) to have in excess of 10% Band Cellscompared to samples with normal/intermediate PMN-CD64 (0/84). When compared to C-reactive protein (CRP), high PMN-CD64 and MON-CD64 were always associated with an increased CRP concentration, but minor proportions of samples withnormal PMN-CD64 (11/52) or normal MON-CD64 (11/65) could also have an increased CRP. The procedures described inthis communication overcome a number of limitations associated with flow cytometry, and co-determination of CD64 and

⁎ Corresponding author. Tel.: +44 1278 671165.E-mail address: steven.scott@abbott.com (C.S. Scott).

0022-1759/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.jim.2006.02.007

208 W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

HLA-DR antigen expression may provide complimentary insights into patient heterogeneity in the assessment of suspectedsepsis compared to CD64 analysis alone.© 2006 Elsevier B.V. All rights reserved.

Keywords: Neutrophils; Monocytes; CD64; HLA-DR; C-reactive protein; Sepsis

1. Introduction

There are many reports regarding the potential ofmeasuring neutrophil membrane CD64 (PMN-CD64)for the diagnostic assessment of sepsis (Guyre et al.,1990; Davis et al., 1995; Hirsh et al., 2001; Naccashaet al., 2001; Layseca-Espinosa et al., 2002; Ng et al.,2004). The membrane molecule defined by monoclo-nal antibody CD64 is a high-affinity receptor (FcγRI)found on normal monocytes and is only expressed atlow levels by normal neutrophils. Upregulation andincreased expression of PMN-CD64 appears to be asensitive marker for early-onset clinical infection innewborn children (Fjaertoft et al., 1999; Ng et al.,2004), and in adults its use has been variouslysuggested for differentiating systemic infection fromactive inflammatory disease (Allen et al., 2002),monitoring γ-interferon therapy (Davis et al., 1995)and as an indicator for initiating or discontinuingantibiotic treatment (Ng et al., 2004). PMN-CD64analysis may be particularly useful for assessing sepsisin young children and elderly patients, wherehaematological parameters such as leukocyte/granulo-cyte counts, or the presence of immature granulocytesand band cells are relatively uninformative (Crocker etal., 1985; Werman and Brown, 1986; Shapiro andGreenfield, 1987), and for investigating patients withprimary disturbances in neutrophil numbers associatedwith haematological malignancies or myelosuppressivetherapies.

All studies to date of PMN-CD64 expression havebeen performed by flow cytometry using fluorochome–monoclonal antibody conjugates. However, a moreefficient approach would be to develop a procedurewhereby EDTA–anticoagulated blood samples submit-ted for full blood count (FBC) analysis could beadditionally processed for PMN-CD64 as a supplemen-tary procedure by the haematology laboratory. The firstaim this study was therefore to examine whether or not amethod for the simultaneous measurement of PMN-CD64 and monocyte CD64 (MON-CD64) could bepotentially implemented on a routine haematologyanalyser (Cell-Dyn CD4000; Abbott Diagnostics,

Santa Clara, CA, United States). As part of the methoddesign, a second marker (HLA-DR) was includedbecause (in contrast to CD64) its expression bymonocytes is reportedly decreased in patients withsepsis (Muller-Kobold et al., 2000; Le Tulzo et al.,2004; Sedlackova et al., 2005). Supplementary aims ofthis study were therefore to determine relationshipsbetween relative PMN-CD64 expression and the levelsof MON-CD64 and MON-Ia in a randomly-selectedseries of patient samples where clinical requests for C-reactive protein (CRP) had been made, and to examinepattern heterogeneity of CD64/HLA-DR antigen ex-pression in comparison with CRP concentration andabsolute granulocyte count.

2. Materials and methods

2.1. CD4000 analysis of neutrophil and monocytemembrane CD64/HLA-DR

In addition to the standard FBC configuration, theCD4000 optical bench has the capability of measuringFL1, FL2 and FL3 fluorescence. These are integratedinto a specific assay processing routine for theenumeration of CD3+CD4+ and CD3+CD8+ popula-tions (Marshall et al., 2000), and the method used for theco-determination of PMN-CD4, MON-CD64 andMON-Ia expression described in this study was basedon an adaptation of this procedure. In essence, the tworeaction tubes (CD3/CD4 and CD3/CD8) normally usedfor the CD3/4/8 counts of a single sample weresubstituted with non-anticoagulated Vacutainer tubescontaining 100 μl whole blood plus 20 μl anti-CD64/FITC (IgG1 subclass; Becton Dickinson) and 15 μl anti-HLA-DR/PE (IgG2a subclass; Becton Dickinson).These were processed using the CD4/CD8 assay modeafter preliminary incubation at room temperature for10 min. Antibody–blood ratios were obtained followingpreliminary titration studies to confirm optimal antibodyconcentrations, and in this assay configuration the twotubes corresponded to parallel CD64/HLA-DR analysesfor two different patient samples. On completion ofinstrument blood sampling and data acquisition (List

209W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

Mode enabled), raw data files were downloaded to a PCand batch-converted to standard FCS2.0 format withCell-Dyn Clinical Data Standard (CDS) convertersoftware (CDS Office Suite; Abbott Diagnostics Cell-Dyn R&D Department, Europe) prior to populationanalysis. This ‘in-house’ software was designed togenerate single-dilution FCS files, readable by standardFCS software/freeware, because Cell-Dyn systemsproduce multiple files (for each part of instrumentanalysis) that are stored within a single List Mode file.Converted List Mode files from the CD4000, typicallycomprising 12 to 20 k individual events, allowgeneration of multiple plot types with parameter optionsbeing 7° Intermediate Angle Scatter (cell complexity),0° Axial Light Loss (cell size), FL1, FL2 and FL3.

2.2. Samples studied

To determine control ranges for neutrophil andmonocyte antigen expression for the CD4000 method,25 samples were collected from normal healthy adultvolunteers. Supplementary studies were also undertakento define methodological limitations. Assay reproduc-ibility was determined by duplicate analysis of 12randomly selected samples, and stability of PMN-CD64antigen expression during sample storage was assessedby analysing 12 EDTA–anticoagulated samples at 1, 6and 12 h after collection. In addition, day to dayvariation in individual normal subjects was determinedby examining neutrophil and monocyte antigen expres-sion with samples from 9 healthy volunteers collectedthree days apart, and the CD4000 method for PMN-CD64 antigen quantitation was compared with a flowcytometry procedure by analysis of 12 further samples.The flow cytometry method (Beckmann Coulter EpicsXL) involved incubation of EDTA anticoagulated bloodsamples with monoclonal antibodies for 15 min,erythrocyte lysis with 0.15 M ammonium chloride for10 min, washing and cell resuspension. Using 90° side-scatter and CD45-ECD/CD14-PE/Cy5 fluorescence(Trillium Diagnostics LLC, Maine USA) granulocyteswere gated and monocytes excluded. By means ofparallel analysis with CD64-FITC and CD163-PE(Trillium Diagnostics), granulocytes were defined asCD64+ when fluorescent staining exceeded the first logdecade. Similarly, the CD4000 method for HLA-DRpopulation discrimination was assessed by parallelanalysis of lymphocyte populations versus flow cyto-metry (14 samples) as previously described (Molero etal., 2005).

To further understand the relationships betweenPMN-CD64, MON-CD64 and MON-Ia expression, a

series of 109 EDTA-anticoagulated patient sampleswere analysed within 6 h of venesection (after storage atambient temperature). These were randomly selectedfrom the routine haematology workload and comprisedpatients for whom turbidimetric measurements of C-reactive protein (CRP) had been clinically requested onthe same day. Full blood counts were obtained with theCD4000 analyser, and stained blood films werereviewed to determine the morphological differential(including percentages of Band Cells and ImmatureGranulocytes). In addition, leukocyte integrity wasassessed using the automated CD4000 propidium iodidestaining procedure (Erwa et al., 1998) and all samplesshowed > 98% viability.

2.3. Statistical analysis

Relationships between individual measured para-meters were variously analysed by logarithmic plots,Passing and Bablok (1983) agreement and the Chi-square test (Analyse-It Software Ltd, PO Box 103,Leeds, England).

3. Results

3.1. Methodological procedures

Gating of neutrophil and monocyte populations(Figs. 1 and 2) with WINMDI flow cytometry software(http://facs.scripps.edu/software.html) was facilitated byprimary separation of all collected leukocyte events intoHLA-DR+ and HLA-DR− fractions (FL1 versus FL2plot). Subsequent region setting for the neutrophilpopulation was based on morphological discriminationwhen HLA-DR− events were redisplayed in theCD4000 optical (0° versus 7°) plot, while monocytescorresponded to the HLA-DR+ subpopulation located inthe upper area of a second 0° versus 7° plot (Figs. 1 and2). After region setting, gated neutrophil and monocytefractions were analysed in FL1 and FL2 histograms (4-decade log) to obtain the median and geometric meansof CD64 and HLA-DR staining intensities (Fig. 2). Asthese were numerically similar for all tested samples, themedian value, expressed as arbitrary fluorescent units(AFU) was subsequently used for all comparativemeasurements. Isotype controls were used to determine‘background’ fluorescence for each cell population andthis was deducted from the median values of samplefluorescence prior to comparative analysis. In addition,FL3 staining (propidium iodide) automatically includedin the analytical process by the CD4000 (Erwa et al.,1998) was assessed to confirm that all studied samples

1a 1b

2a 2b

3a 3b

CD64

HLA-DR

1c

2c

3c

1d

2d

3d

CD64

HLA-DR

CD64

HLA-DR

104

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E

255

00

SIZ

E25

50

0 S

IZE

255

00

SIZ

E

25507 COMPLEXITY

25507 COMPLEXITY

25507 COMPLEXITY

25507 COMPLEXITY

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E

25507 COMPLEXITY

103

102

101

100

104103102101100

104

103

102

101

100

104103102101100

104

103

102

101

100

104103102101100

210W.van

derMeer

etal.

/Journal

ofIm

munological

Methods

311(2006)

207–219

211W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

had high (>98%) leukocyte viability and no evidence ofsample deterioration.

Preliminary assessment of FL1 and FL2 data for fileswith and without fluorescent channel compensationrevealed that crossover of FL1 fluorescence (CD64) intothe FL2 channel (HLA-DR) had no significant effect onsemi-quantitative levels of HLA-DR staining. Thereforein order to simplify the procedure, non-compensatedfiles were used for statistical determinations of medianantigen staining.

3.2. Control neutrophil and monocyte antigenexpression

Histogram profiles of PMN-CD64 staining inten-sity showed a normal distribution, and analysis of the25 normal samples revealed a mean PMN-CD64 of98 AFU (observed range 83–126). Parallel analyseswith an IgG1 isotype control reagent indicatedbackground levels of 56 AFU and this value wassubsequently deducted from all PMN-CD64 determi-nations. Based on the corrected control values(observed minus background), the normal range forPMN-CD64 defined as the mean±2SD (standarddeviation) corresponded to 17 to 67 AFU. A highlevel of PMN-CD64 expression (> 135 AFU) wastherefore defined as a PMN-CD64 value exceedingtwice the upper normal limit, while an intermediateincrease represented PMN-CD64 expression between67 and 135 AFU. Using a similar approach forMON-CD64, normal, intermediate and high levels ofMON-CD64 expression were respectively defined as515 to 1045, 1046 to 2090 and > 2090 AFU aftercorrection for background (isotype control)fluorescence.

In contrast to neutrophil and monocyte CD64,where abnormal trends were associated with antigenupregulation, abnormal monocyte HLA-DR (MON-Ia) was characterised by declining expression. Thus,the isotype-corrected normal range (mean±2SD) for25 normal samples corresponded to 170 to 670 AFU,a marked decrease was defined as less than 50% ofthe lower normal limit (<85 AFU) and an interme-diate decrease as MON-Ia values between 85 and170 AFU.

Fig. 1. Primary gating procedure for separation of leukocyte populations acpatterns of neutrophil and monocyte antigen expression (sample 1— normal PCD64, low MON-Ia; sample 3— high PMN-CD64, normal MON-Ia). All leFL2 plots (a) and a region set for the HLA-DR+ fraction (colour-coded red(complexity) plot (b) to allow subsequent visualisation of HLA-DR− events onused to subsequently set further regions (circled) for neutrophil and monocy

3.3. Methodological variables

Analytical reproducibility was determined by dupli-cate analysis of 12 random patient samples. Of these, allwere evaluable for PMN-CD64 expression and all butone were evaluable for MON-CD64 and MON-Iaexpression. The single sample that could not beanalysed had very high PMN-CD64 expression(580 AFU) with severe monocytopenia and too fewmonocyte events for reliable analysis. The results ofthese duplicate studies (Fig. 3) revealed a high level ofassay consistency for all three analysed antigens.

Stability of PMN-CD64 antigen expression duringsample storage was assessed by analysing 12 EDTA-anticoagulated samples at 1, 6 and 12 h after collection.At 1 h, the mean PMN-CD64 expression was 63.7 AFUand this was normalised to represent a 100% baselinelevel. At 6 h, the mean change was +2.2% (range −3%to +14%) and at 12 h the mean change was +10% (range0% to +30%). Further examination of the data indicatedno significant decrease in neutrophil viability during the12 h period, and this excluded the possibility ofsignificant fluorescence crossover from propidiumiodide staining of non-viable leukocytes in the FL3channel. Parallel assessments revealed good stability ofMON-CD64 expression, with mean changes at 6 and12 h of −5.5% (range −10.3% to 0%) and +5.6% (range−6.9% to +22%) respectively from a baseline of835 AFU. In contrast, changes in MON-Ia expressionwere more variable. Compared to a mean baseline of339 AFU, there was a consistent decrease in expressionafter 6 h storage (average change −19.5%; range −33%to 0%) while at 12 h the mean change compared tobaseline was +20.7% (range −14% to +54%). However,while the magnitudes of in vitro changes for MON-Ia forindividual samples were significant these had littleeffect on interpretation in terms of categorising results asnormal, intermediate decrease or marked decrease.

Day to day variation in normal individuals wasdetermined by examining neutrophil and monocyteantigen expression with samples from 9 healthyvolunteers collected three days apart. The results(Fig. 4) indicated that for PMN-CD64 expression, 8/9 showed similar assay values within the normalcontrol range. Only one subject showed a significant

cording to HLA-DR (Ia) expression for three samples with differingMN-CD64, normal MON-Ia; sample 2— intermediate increase PMN-ukocyte events were initially examined in CD64/FL1 versus HLA-DR/). All events were then redisplayed in a CD4000 0° (size) versus 7°ly (plot c) or HLA-DR+ events only (plot d). These latter two plots werete fractions respectively (Fig. 2).

1f

1e

2e

3e

2f

3f

1g

2g

3g

1h

2h

3h

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E

25507 COMPLEXITY

255

00

SIZ

E25

50

320

0 S

IZE

25507 COMPLEXITY

25507 COMPLEXITY

25507 COMPLEXITY

25507 COMPLEXITY

255

00

SIZ

E25

5

512

0 051

20

100 101 102 103 104100 101 102 103 104

320

100 101 102 103 104

320

100 101 102 103 104

100 101 102 103 104

100 101 102 103 104

1024

0

0 S

IZE

Fig. 2. Following the primary gating of leukocyte populations according to HLA-DR (Ia) expression with the three samples in Fig. 1, secondary regions were applied to define neutrophil and monocytefractions for subsequent measurements of membrane antigen expression. The HLA-DR− components displayed in the 0/7° plot were gated for neutrophils (plots e) while the HLA-DR+ componentswere gated for monocytes (plot g). After gate setting, histograms of antigen expression were used to determine median levels of staining. These were then corrected for control staining as discussed inMaterials and methods. Pre-correction (control) values for neutrophil CD64 (PMN-CD64) expression in the three examples corresponded to 979 (plot 1f), 198 (2f) and 505 (3f) AFU respectively.Similarly, the expression of monocyte CD64 (MON-CD64) and HLA-DR (MON-Ia) was determined from histogram displays of the monocyte fraction. For the three examples, the MON-Ia valueswere 264 (1 h), 75 (2 h) and 340 (3 h) AFU respectively. Although not shown here, MON-CD64 expression was determined using an identical approach.

212W.van

derMeer

etal.

/Journal

ofIm

munological

Methods

311(2006)

207–219

0

100

200

300

400

500

600

0 100 200 300 400 500 600

PMN-CD64 (Duplicate 1)

PM

N-C

D64

(D

up

licat

e 2)

500

700

900

1100

1300

1500

500 700 900 1100 1300 1500MON-CD64 (Duplicate 1)

MO

N-C

D64

(D

up

licat

e 2)

0

50

100

150

200

250

300

350

400

0 50 100 150 200 250 300 350 400MON-Ia (Duplicate 1)

MO

N-I

a (D

up

licat

e 2)

(a) (b) (c)

R2 = 0.99 R2 = 0.99 R2 = 0.99

Fig. 3. Duplicate analysis of 12 randomly selected samples for PMN-CD64, MON-CD64 and MON-Ia expression. For each tested sample, twoseparate tubes were stained and processed, with the levels of antigen expression (AFU) being determined as detailed in the text. The pairedmeasurements for MON-CD64 and MON-Ia expression exclude one sample with severe monocytopenia where there were too few events for reliableanalysis.

213W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

difference with a PMN-CD64 of 139 AFU (inter-mediate increase) on day 1 compared to 67 AFU(normal) on day 3. Similar day to day consistencywas also seen with MON-CD64 and MON-Ia, withonly one individual in each group showing higherexpression on day 1 compared to day 3.

Correlation between the CD4000 PMN-CD64method and flow cytometry was assessed by parallelanalysis of 12 randomly selected samples. For thiscomparison, CD4000 PMN-CD64 results were deter-

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

0 100 200 300 400 500 6

Day 1

Day

3 A

nal

ysis

Fig. 4. Day to day consistency of PMN-CD64 (solid circles), MON-CD64subjects were venesected on two separate occasions three days apart and assashown as boxed rectangles, with the data indicating generally good stability oin each of the assay groups, are seen to slightly exceed normal limits on day

mined as percentages CD64+ neutrophils where thefluorescence threshold used to discriminate betweennegative and positive was set at the upper 95%percentile of PMN-CD64 expression by 10 normalsamples. This revealed excellent agreement betweenthe two methods (Fig. 5a). We also confirmed that thesemi-quantitative (AFU) method of CD64 antigenexpression used in this study correlated with the moregeneral estimation of percentages positive cells (Fig.6). A comparative analysis between the CD4000 and

00 700 800 900 1000 1100 1200

Analysis

(open circles) and MON-Ia (solid triangles) expression. Nine healthyyed for all three antigens with the CD4000 method. Normal ranges aref antigen expression. Outliers, corresponding to one of the nine subjects1 only of the comparative analyses.

00

10

20

20

30

40

40

50

60

60

70

80

80

90

100

100 0 20 40 60 80 100

Flow Cytometry % CD64+ Neutrophils

CD

4000

% C

D64

+ N

eutr

op

hils

0

10

20

30

40

50

60

70

80

90

100

Flow Cytometry % Ia+ Lymphocytes

CD

4000

% I

a+ L

ymp

ho

cyte

s

y = 1.00x -0.6 y = 1.09x + 1.23

(a) (b)

Fig. 5. Comparison of the CD4000 method for PMN-CD64 and leukocyte HLA-DR (Ia) expression with flow cytometry. For the PMN-CD64comparison (a), CD4000 results for 12 samples were expressed as the percentages of CD64+ positive neutrophils where the fluorescence thresholdused to discriminate between negative and positive was set at the upper 95% percentile of PMN-CD64 expression by 10 normal samples (see Fig. 6for relationships between % CD64+ neutrophil and AFU values). For the comparison of CD4000 and flow cytometry HLA-DR analyses, percentagesof HLA-DR+ lymphocytes (B-cells and activated T-cells) were determined in 14 samples, with the discrimination threshold being set by reference tocoexisting HLA-DR− leukocyte populations (predominantly neutrophils and non-activated T-cells).

214 W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

flow cytometry for HLA-DR staining, undertaken bydetermining the percentages of HLA-DR+ lympho-cytes in a series of 14 samples, also showed a highdegree of inter-method equivalence (Fig. 5b).

3.4. Patterns of neutrophil and monocyte CD64 andHLA-DR antigen expression

Observed ranges of PMN-CD64, MON-CD64 andMON-Ia for the 109 clinical samples were 31 to 1058,307 to 2843 and 10 to 876 AFU respectively.Membrane antigen data was obtained for all butthree samples with severe monocytopenia. PMN-CD64 and MON-CD64 expression shared the same

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000 1200

CD4000 PMN-CD64 (AFU)

CD

4000

PM

N-C

D64

(%

Po

siti

ve P

MN

)

(a)

Fig. 6. Comparison of alternative reporting methods for PMN-CD64 expressiothe CD4000 to determine both the quantitative level of antigen expressionfluorescence threshold used to discriminate between negative and positivenormal samples. Plot (a) shows the general comparison for all samples whilePMN-CD64 expression (<200 AFU).

overall trend although relative increases in MON-CD64 expression were higher than PMN-CD64 (Fig.7a). More detailed examination of individual samples(Table 1) showed that normal PMN-CD64 expressionwas only seen with normal (45/52) or intermediateMON-CD64 (7/52), while high PMN-CD64 wasusually associated with intermediate (18/22) or high(3/22) MON-CD64. By comparison, MON-Ia expres-sion appeared to be largely independent (p=0.04) ofPMN-CD64 (Fig. 7b) although a marked decrease inMON-Ia was always associated with intermediate orhigh levels of PMN-CD64 (Table 1). Interestingly,marked decreases in MON-Ia were not seen whenMON-CD64 was expressed at high levels.

0

10

20

30

40

50

60

70

80

90

100

0 20 40 60 80 100 120 140 160 180 200

CD4000 PMN-CD64 (AFU)

CD

4000

PM

N-C

D64

(%

Po

siti

ve P

MN

)

(b)

n. The 109 clinical samples analysed in this study were processed with(AFU) and the percentages of CD64+ neutrophils. For the latter, thewas set at the upper 95% percentile of PMN-CD64 expression by 10plot (b) shows a specific comparison for samples in the lower range of

0

500

1000

1500

2000

2500

3000

0 200 400 600 800 1000 1200

Median Neutrophil CD64 (AFU)

Med

ian

Mo

no

cyte

CD

64 (

AF

U)

0

100

200

300

400

500

600

700

800

900

0 200 400 600 800 1000 1200

Median Neutrophil CD64 (AFU)

Med

ian

Mo

no

cyte

Ia (

AF

U)

(a) (b)

Fig. 7. Individual sample (n=106) relationships between median neutrophil CD64 and (a) monocyte CD64 and (b) monocyte HLA-DR (Ia). Plottedtrendlines are derived from logarithmic plots.

215W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

3.5. Patterns of antigen expression and granulocytecounts

Evaluated relationships with absolute granulocytecounts suggested (Table 2) some association with PMN-CD64 expression but none with MON-CD64. There wasalso an inverse relationship (p<0.0001) between MON-Ia and the absolute granulocyte count, with a trend forhigher counts with decreasing membrane HLA-DRexpression. Examination of other haematological para-meters revealed no significant association betweenneutrophil and monocyte antigen levels and percentages

Table 1Relationships between PMN-CD64, MON-CD64 and MON-Iaexpression for 106 randomly selected clinical samples

Monocyte CD64

Normal Intermediate High

Neutrophil CD64Normal 45 7 0Intermediate 19 13 0High 1 18 3

X 2 statistic, 48.9 (p<0.0001)

Monocyte Ia

Normal Intermediate decrease Marked decrease

Neutrophil CD64Normal 46 6 0Intermediate 23 5 4High 12 4 6

X 2 statistic, 15.7 (p=0.04)Monocyte CD64

Normal 57 5 3Intermediate 21 10 7High 3 0 0

X 2 statistic, 15.0 (p=0.005)

Normal, intermediate and high antigen levels as defined in Results.Data excludes three samples with severe monocytopenia.

of Immature Granulocytes (data not shown) althoughsamples with high PMN-CD64 were more likely (8/25)to have in excess of 10% Band Cells compared tosamples with normal or intermediate PMN-CD64 (0/84).

3.6. Patterns of antigen expression and CRPconcentration

The range of CRP concentrations for the 109 clinicalsamples analysed in this study was <10 to 460 mg/l.There were 57 samples with a normal CRP (<10 mg/l),49 between 10 and 100 mg/l, and 3 with concentrations

Table 2Relationships between neutrophil (CD64) and monocyte (CD64 andIa) antigen expression and absolute granulocyte counts for 109randomly selected clinical samples

Absolute granulocyte count (× 109/l)

<8.0 8.0–16.0 >16.0

Neutrophil CD64Normal 42 10 0Intermediate 17 14 1High 16 5 4

X 2 statistic, 17.0 (p=0.002)Monocyte CD64

Normal 51 13 1Intermediate 20 14 4High 2 1 0

X 2 statistic, 9.2 (p=0.06)Monocyte Ia

Normal 62 19 0Intermediate decrease 10 5 0Marked decrease 1 4 5

X 2 statistic, 55.0 (p<0.0001)

Ranges for absolute granulocyte counts represent multiples of theupper normal range (8.0×109/l). Normal, intermediate and highantigen levels as defined in Results; monocyte antigen data excludesthree samples with severe monocytopenia.

Table 3Relationships between neutrophil (CD64) and monocyte (CD64 andIa) antigen expression and C-reactive protein (CRP) concentrations for109 randomly selected clinical samples

CRP (mg/l)

<10 10–50 51–100 >100

Neutrophil CD64Normal 41 8 3 0Intermediate 16 6 10 0High 0 2 20 3

X 2 statistic, 62.1 (p<0.0001)Monocyte CD64Normal 54 8 3 naIntermediate 3 8 27 naHigh 0 0 3 na

X 2 statistic, 67.3 (p<0.0001)Monocyte IaNormal 52 11 18 naIntermediate decrease 4 4 7 naMarked decrease 1 1 8 na

X 2 statistic, 19.8 (p<0.001)

Ranges for CRP concentrations correspond to mild, moderate andsevere inflammatory processes as defined by Whicher (1998). Normal,intermediate and high antigen levels as defined in Results; monocyteantigen data excludes three samples with severe monocytopenia andCRP levels exceeding 100 mg/l.

216 W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

higher than 100 mg/l. When neutrophil and monocyteantigen expression was compared to CRP (Table 3),high degrees of overall association were observedalthough individual sample variation was also evident.For example, while high PMN-CD64 and high MON-CD64 expression were always associated with anincreased CRP concentration, this could also be seenwhen PMN-CD64 (11/52) and MON-CD64 (11/65)were normal. Similarly, 11/15 and 9/10 of the patientsamples with intermediate or marked decreases inMON-Ia had increased CRP levels. Of note with respectto the analysis of monocytes was that the three sampleswith severe monocytopenia (and insufficient populationevents to determine antigen expression) all had CRPconcentrations exceeding 200 mg/l.

4. Discussion

The primary aim of this study was to evaluate a two-colour immunofluorescent assay for leukocyte CD64and HLA-DR membrane markers on a routine haema-tology analyser with fluorescent capabilities. These twoantigens are of particular interest because of accumu-lated evidence suggesting that upregulation (neutrophiland monocyte CD64) and downregulation (monocyteHLA-DR) have consistent associations with sepsis.While further substantiating these associations was notthe purpose of this study, our evaluation of antigen

expression in 109 randomly-selected samples withclinical requests for CRP did allow us to undertakesupplementary studies into patterns of antigen changeand their relationships with both CRP concentration andabsolute granulocyte count. The value of these addi-tional analyses was to highlight the existence ofheterogeneous patterns of antigen expression comparedto these widely-used empirical assessments of inflam-mation/infection.

The methodological procedure evaluated in thisstudy was analogous to standard flow cytometry inthat EDTA-anticoagulated blood samples were pre-incubated with FITC/PE-labelled monoclonal antibo-dies, analysed in a laser-illuminated optical system withfluorescent detectors, and the resulting raw-file infor-mation processed with cytometry software to determinethe nature of specific cell populations. Compared toflow cytometry, the CD4000 analysis is simpler in thatinstrument calibration and gain setting is continuouslymaintained for the purposes of routine blood countanalysis. Processing of pre-stained samples, subsequentdata acquisition (up to 20,000 leukocyte events) and redcell lysis are also part of the automated procedure. Thereis also no need for washing and the method has theadditional advantage of simultaneous leukocyte viabil-ity measurements (Erwa et al., 1998) provided by theCD4000 which minimise potential inconsistencies ofantigen measurements associated with leukocyte popu-lation deterioration. Data processing using a PC isrelatively straightforward and can be achieved invarious ways. In this study, downloaded CD4000 ListMode files were converted to standard FCS2.0 formatand analysed using WINMDI software. More recently, asoftware programme has been developed (FCS Expressv3; http://www.denovosoftware.com) for instrument usewhich allows the opening (of CD4000 and CD-Sapphire) files directly without the need for FCSconversion.

Normal ranges for PMN-CD64, MON-CD64 andMON-Ia were determined by analysis of 25 normaladults, and the CD4000 method was shown to havegood duplicate precision and agreement with flowcytometry procedures. Some day to day variation washowever noted in some individuals but observedquantitative differences were of minor interpretativesignificance. One potentially important observationhowever was that storage of EDTA–anticoagulatedsamples at room temperature could lead to changes inneutrophil and monocyte antigen expression. Althoughthere was significant individual sample variation, thechanges were nevertheless relatively small for PMN-CD64 and MON-CD64 in the first 6 h of storage. In

217W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

contrast, the expression of MON-Ia was more unstable,with changes from baseline estimates being −20% at 6 hand +21% at 12 h. Although awareness of this isimportant, the practical implications of short-termstorage changes with regards to interpretation (i.e.distinguishing between normal and abnormal) were ofminor significance for the different ranges of antigenexpression defined in this study.

Membrane CD64/HLA-DR antigen levels weresemi-quantitatively expressed as arbitrary fluorescentunits (AFU). This approach was taken as the alternativepractice of reporting antigen expression as percentagesof cells exceeding a defined fluorescence threshold hassome limitations. This is illustrated in Fig. 6 wherePMN-CD64 results of the 109 clinical samples as bothAFU and percentages CD64+ positive neutrophils aredirectly compared. While there is a very good overallcorrelation, the method of determining AFU appears tobe more informative at low and high levels ofexpression. Determining differences in these rangesmay be important for serial patient monitoring.

While the CD4000 method of antigen quantitationdescribed in this study was clearly able to provideconsistent relative comparisons of antigen density, weaccept that this may not be ideal with respect to theneed for standardising inter-institutional reportingpractices. In this context, improvements can beachieved by quantifying antigen molecules per cell(Ng et al., 2004; Poncelet, 2004) or by reference tointernal fluoresceinated control (calibration) particles(Gratama et al., 1998; Vogt et al., 2000). With respectto neutrophil CD64 measurements, this latter principleis utilised by the Leuko64™ assay kit (TrilliumDiagnostics LLC, Maine USA). However, while thisapproach ensures greater consistency of antibodyreagent source and fluorochrome–protein characteris-tics of monoclonal antibody conjugates it has littlevalue in confirming whether or not leukocytes areoptimally stained.

As indicated earlier, there is considerable support andclinical evidence for using measurements of neutrophilCD64 expression in assessing patients with suspectedsepsis (Guyre et al., 1990; Davis et al., 1995; Fjaertoft etal., 1999; Hirsh et al., 2001; Naccasha et al., 2001;Layseca-Espinosa et al., 2002; Allen et al., 2002; Ng etal., 2004). This is substantiated by in vitro observationsshowing mediated upregulation of PMN-CD64 bylipopolysaccharides (Wagner et al., 2003), γ-interferon(Schiff et al., 1997) and G-CSF (Kakinoki et al., 2004).PMN-CD64 also has a number of potential practicaladvantages in that it is expressed at very low levels bynormal neutrophils and is relatively insensitive to

sample manipulation (Davis et al., 1995; Davis, 1996).Similarly, the measurement of monocyte HLA-DRexpression has also been widely evaluated and shownto decrease in sepsis (Muller-Kobold et al., 2000;Drossou-Agakidou et al., 2002; Sedlackova et al., 2005)and be associated with transient immunosuppression (LeTulzo et al., 2004). Consequently, as CD64 and HLA-DR show opposite trends of upregulation and down-regulation in sepsis, their combined analysis asproposed in this communication might be advantageouscompared to single antigen assessments. Moreover, themethod can still provide PMN-CD64 interpretive resultswhen blood samples are severely monocytopenic andhave too few events for reliable statistical analysis ofHLA-DR (and MON-CD64) expression.

To analyse the potential of the simultaneous CD64/HLA-DR method further, a series of 109 patients forwhom clinical requests for CRP had been made wererandomly selected from the haematology workload. Theresults showed that while neutrophil and monocyteCD64 upregulation followed a similar trend, relation-ships with MON-Ia expression were more complex.Significant reductions in MON-Ia are characteristicallyseen in early sepsis, with progressive return to normalityas sepsis-associated transient immunosuppression isovercome (Drossou-Agakidou et al., 2002). Our obser-vations showed that while markedly depressed mono-cyte HLA-DR expression was associated with highPMN-CD64 this was in contrast to MON-CD64 whichwas either normal or only moderately increased. Thepossibility that these different trends was to some extenta reflection of differences in cell population kinetics,where the half-life of neutrophils (6–9 h) is considerablyshorter than monocytes (8–72 h), and differentialcellular responses to inflammation/infection is sup-ported by a number of further observations. Forexample, all five patients with absolute granulocytecounts exceeding 16.0×109/l showed markedly reducedMON-Ia expression. Four of these showed a consistentpattern of high PMN-CD64, intermediate MON-CD64and high (>100 mg/l) CRP concentrations. In contrast,the fifth patient showed an intermediate increase inPMN-CD64, normal MON-CD64 and a normal CRP.Interestingly, this particular patient who was beingartificially ventilated and had a long history ofintermittent septic episodes showed a significantincrease in CRP in the immediate period followingantigen testing. An additional finding was that allsamples with profound monocytopenia had very highPMN-CD64 expression and CRP concentrations.

In summary, this study has demonstrated the designfeasibility of a fluorescent method for the semi-

218 W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

quantitative determination of neutrophil (CD64) andmonocyte (CD64 and HLA-DR) antigens associatedwith the assessment of sepsis. The combined use ofCD64 and HLA-DR is considered more informativethan using CD64 alone, and while standardisation of theprocess could be improved by method modificationswith respect to fluorescent quantitation, the procedurenevertheless remains simple and relatively straightfor-ward to implement. We believe that the methoddescribed in this communication could overcome anumber of limitations associated with flow cytometryand provide a potential indicator of sepsis at relativelylow cost and minimum technical expertise.

Acknowledgements

We would like to thank Mrs Jolanda Terwal andNataša Pranjic-Tumenko for excellent technical assis-tance, and Mr Paul Brans (Department of Haematology,Radboud University Nijmegen Medical Center) forundertaking the flow cytometry comparative studies.

This study was supported in part by AbbottDiagnostics.

References

Allen, E., Bakke, A.C., Purtzer, M.Z., Deodhar, A., 2002. NeutrophilCD64 expression: distinguishing acute inflammatory autoimmunedisease from systemic infections. Ann. Rheum. Dis. 61, 522.

Crocker, P.J., Quick, G., McCombs,W., 1985. Occult bacteremia in theemergency department: diagnostic criteria for the young febrilechild. Ann. Emerg. Med. 14, 1172.

Davis, B., 1996. Quantitative neutrophil CD64 expression: promisingdiagnostic indicator of acute inflammation or systemic acuteinflammatory response. Clin. Immunol. Newsl. 16, 121.

Davis, B.H., Bigelow, N.C., Curnutte, J.T., Ornvold, K., 1995.Neutrophil CD64 expression. Potential diagnostic indicator ofacute inflammation and therapeutic monitor of interferon-γtherapy. Lab. Hematol. 1, 3.

Drossou-Agakidou, V., Kanakoudi-Tsakalidou, F., Sarafidis, K.,Tzimouli, V., Parakou, A., Kremenopoulos, G., Germenis, A.,2002. In vivo effect of rhGM-CSF and rhG-CSF on monocyteHLA-DR expression of septic neonates. Cytokine 18, 260.

Erwa, W., Bauer, F.R., Etschmaier, R., Steiner, U., Scott, C.S.,Sedlmayr, P., 1998. Analysis of aged samples with the AbbottCD4000 hematology analyzer. Eur. J. Lab. Med. 6, 4.

Fjaertoft, G., Hakansson, L., Ewald, U., Foucard, T., Venge, P., 1999.Neutrophils from term and preterm newborn infants express highaffinity Fc-gamma-receptor I (CD64) during bacterial infections.Pediatr. Res. 45, 871.

Gratama, J.W., D'Hautcourt, J.L., Mandy, F., Rothe, G., Barnett, D.,Janossy, G., Papa, S., Schmitz, G., Lenkei, R., 1998. Flowcytometric quantitation of immunofluorescence intensity: pro-blems and perspectives. European working group on clinical cellanalysis. Cytometry 33, 166.

Guyre, P.M., Campbell, A.S., Kniffin, W.D., Fanger, M.W., 1990.Monocytes and polmorphonuclear neutrophils of patients with

streptococcal pharyngitis express increased numbers of type I IgGFc receptors. J. Clin. Invest. 86, 1892.

Hirsh, M., Mahamid, E., Bashenko, Y., Hirsh, I., Krausz, M.M.,2001. Overexpression of the high-affinity Fcgamma receptor(CD64) associated with leukocyte dysfunction in sepsis. Shock16, 102.

Kakinoki, Y., Kubota, H., Yamamoto, Y., 2004. CD64 surfaceexpression on neutrophils and monocytes is significantly upregu-lated after stimulation with granulocyte-stimulating factor duringCHOP chemotherapy for patients with non-Hodgkins lymphoma.Int. J. Hematol. 79, 55.

Layseca-Espinosa, E., Perez-Gonzales, L.F., Torres-Montes, A.,Baranda la Fuente, H., Rosenstein, Y., Gonzalez-Amaro, R.,2002. Expression of CD64 as a potential marker of neonatal sepsis.Pediatr. Allergy Immunol. 13, 319.

Le Tulzo, Y., Pangault, C., Amiot, L., Guilloux, V., Arvieux, C.,Camus, C., Fauchet, R., Thomas, R., Dernous, B., 2004. Monocytehuman leukocyte antigen-DR transcriptional downregulation bycortisol during septic shock. Am. J. Respir. Crit. Care Med. 169,1144.

Marshall, P., Hung, D., Yuan, J., Kim, Y.R., 2000. Rapid,automated, closed-tube quantitation of CD4+ and CD8+ T-cellpopulations on the Cell-Dyn 4000 hematology analyzer. Lab.Hematol. 6, 137.

Molero, T.L., Roemer, B., Del Mar Perera Alvarez, M., Lemes, A., DeLa Iglesia Inigo, S., Palacios, G., Scott, C.S., 2005. Analysis andenumeration of T cells, B cells and NK cells using the monoclonalantibody fluorescence capability of a routine haematology analyser(Cell-Dyn CD4000). Clin. Lab. Haematol. 27, 224.

Muller-Kobold, A.C., Tulleken, J.E., Zijlstra, J.G., Sluiter, W.,Hermans, J., Kallenberg, C.G., Tervaert, J.W., 2000. Leukocyteactivation in sepsis; correlations with disease state and mortality.Intensive Care Med. 26, 883.

Naccasha, N., Gervasi, M.T., Chaiorapongsa, T., Berman, S., Yoon,B.H., Maymon, E., Romero, R., 2001. Phenotypic and metaboliccharacteristics of monocytes and granulocytes in normalpregnancy and maternal infection. Am. J. Obstet. Gynecol. 185,1118.

Ng, P.C., Li, G., Chui, K.M., Chu, W.C., Li, K., Wong, R.P., Chik,K.W., Wong, T.F., 2004. Neutrophil CD64 is a sensitive diag-nostic marker for early onset neonatal infection. Pediatr. Res. 56,796.

Passing, H., Bablok, W., 1983. A new biometrical procedure fortesting the equality of measurements from two different analyticalmethods. J. Clin. Chem. Biochem. 21, 709.

Poncelet, P., 2004. Microbeads and flow cytometry: how and why putthe “-metry” in immuno-cytometry? Ann. Biol. Clin. 62, 53.

Schiff, D.E., Rae, J., Martin, T.R., Davis, B.H., Curnette, J.T., 1997.Increased phagocyte Fc gammaR1 expression and improved Fcgamma-receptor-mediated phagocytosis after in vivo recombinanthuman interferon-gamma treatment of normal human subjects.Blood 90, 3187.

Sedlackova, L., Prucha, M., Dostal, M., 2005. Immunologicalmonitoring of sepsis using flow cytometry — quantitation ofmonocyte HLA-DR expression and granulocyte CD64 expression.Epidemiol. Mikrobiol. Immunol. 54, 54.

Shapiro, M.F., Greenfield, S., 1987. The complete blood count andleukocyte differential count. An approach to their rationalapplication. Ann. Intern. Med. 106, 65.

Vogt, R.F., Whitfield, W.E., Henederson, L.O., Hannon, W.H., 2000.Fluorescence intensity calibration for immunophenotyping by flowcytometry. Methods 21, 289.

219W. van der Meer et al. / Journal of Immunological Methods 311 (2006) 207–219

Wagner, C., Deppisch, R., Denefleh, B., Hug, F., Andrassy, K.,Hansch, G.M., 2003. Expression patterns of the lipopolysaccharidereceptor CD14 on the Fcgamma receptors CD16 and CD64 onpolymorphonuclear neutrophils: data from patients with severebacterial infection and lipopolysaccharide-exposed cells. Shock19, 5.

Werman, H.A., Brown, C.G., 1986. White blood cell count anddifferential count. Emerg. Med. Clin North Am. 4, 41.

Whicher, J., 1998. C-reactive protein (CRP). In: Thomas, L.(Ed.), Clinical Laboratory Diagnostics. T-H Books, Frankfurt,pp. 700–706.