Technical Notes

Enumeration of Peripheral Lymphocyte SubsetsUsing 6 vs. 4 Color Staining: A ClinicalEvaluation of a New Flowcytometer

Claude Lambert,* Iobagiu Cristina, and Genin ChristianImmunology Laboratory, University Hospital of St Etienne, France

Technological advances in instruments allow the evaluation of many lymphocyte subsets in one step. Theaim of this study was to evaluate the new FACSCanto flowcytometer in routine conditions, using a 6 color combi-nation, single platform, whole blood, lysis, no wash protocol.Methods: Two systems were simultaneously compared on 67 blood samples and external quality controls,using CD3,CD4, CD8, CD19, CD16/56, and CD45 in one tube TRUCOUNT beadsTM (BD Biosciences) or twotubes (TetraChromeTM and FlowcountTM, Beckman-Coulter and DakoCytomation).Results: The day-to-day instrument detection but automatic compensations were stable. Manual compen-

sation settings were satisfactory using available facilities. Commercial and UK NEQAS quality control resultswere acceptable. The intra-experiment reproducibility was good (coefficient of variation (CV) <3%) buthighly operator-dependant (CD41 T cell count CVs from 1.2 to 9.7, six operators). Storage of samples wasacceptable, but storage of stained samples altered absolute count reliability. Serial dilutions show a goodcount accuracy. The FACScanto T subsets and B cell data were highly correlated with our reference values(r2 > 0.87) and absolutes values were very close (slopes > 0.89).

The gating strategy, fluorochrome choice, and compensation setting are discussed. A few improve-ments are expected (sample loader, data management, auto-gating, acquisition parameters, sample mix-ing, absolute values calculation, etc).

In conclusions, despite its complexity, 6 color staining is a reliable, stable, and highly informative tech-nique for lymphocyte subset monitoring but remains to be optimized. q 2005 International Society for Analytical

Cytology

Key terms: T cell subsets; 6 color flowcytometry; quality assurance; routine analysis; immune monitoring

Evaluating the cognate immune response is usuallyrestricted to the CD4þ and CD8þ T(CD3) cells count inroutine practices (1). Because the use of CD45 is recom-mended (2–4) to improve leukocyte identification (5,6),analyzing more subsets with the 4 colors availablerequires a second combination including CD3, CD19, andCD56 (7), unless mixing antibodies with improbable co-expression (8). This increases cost and time consumption.Furthermore, recently described unconventional T cellsubsets (9–19) are not available for clinical evaluation.

In routine practice, processes are reduced to a one-stepprocedure using whole blood, premixed antibodies, inter-nal standard, red blood cell lysis, and no wash significantlyimpairs the quality and reproducibility of labeling. Stand-ardization of instrument settings and gating strategy havebeen achieved for 4 color systems as assessed by inde-pendent (UK NEQAS Assessment Schemes for LeukocytesImmuno-phenotyping (20–23)) or commercial (Interlabora-tory Quality Assurance Program; Beckman-Coulter (24)) in-ternational quality control surveys.

The recently available 5/6 color systems give muchmore information in one single tube test but have notyet been optimized, standardized, and validated in rou-tine conditions (25).The aim of this study was to evaluate the new FACS-

Canto in routine conditions, for T, B, and NK lympho-cytes evaluation, using a 6 color labeling, lysing, nowash, and single platform procedure.

PATIENTS AND METHODS

The FACScanto (BD Biosciences, San Jose, CA) instru-ment stability was analyzed daily using 7-color SetUpTM

Beads (BD Biosciences).

*Correspondence to: Dr. Claude Lambert, Immunology Laboratory, Hopi-tal Bellevue, CHUSt Etienne, F 42055St Etienne Cedex 2, France.E-mail: Claude.lambert@univ-st-etienne.frReceived 24 August 2004; Accepted 20 June 2005Published online 13 December 2005 Wiley InterScience (www.

interscience.wiley.com).DOI: 10.1002/cyto.b.20072

Cytometry Part B (Clinical Cytometry) 70B:29–38 (2005)

q 2005 International Society for Analytical Cytology

The exactitude and stability of cell count was testedusing international quality controls purchased from UKNEQAS LI (Sheffield, UK) and Immuno-TrolTM (Beckman-Coulter Fullerton, CA). The day-to-day reproducibility,sample stability, and count accuracy were evaluatedusing fresh blood from routine practice (21F/46M agedfrom 16 to 79 (50 6 14.7)). Patients came from HIV(n ¼ 22), renal transplantation (n ¼ 42), pediatric (16 yearold), or dermatology (n ¼ 2) departments.

Lymphocytes were immunolabeled using a combina-tion of CD3-fluorescein isothiocyanate (FITC, IgG1, clonesk7), CD16-phycoerythrin (PE, IgG1, Clone B73.1), CD56-PE (IgG1, Clone NCAM 16.2), CD45-peridin chlorophylprotein (PerCP, IgG1, clone 2D1), CD4-PE-Cyanin7 (PE-Cy7, IgG1, clone SK3), CD19-allophycocyanin (APC,IgG1, clone SJ25C1), and CD8-APC-Cyanin7 (APC-Cy7,IgG1, clone SK1) and TRUCOUNT beadsTM (BD Bioscien-ces), according to manufacturer’s recommendations.From 2 to 10 mL of each antibody were distributed inTrucount tubes. Then, an exact volume (i.e. 50 mL) of(EDTA added) fresh peripheral blood (within 4 h afterwithdrawal) was added, gently mixed, and incubated for20 min at room temperature in the dark. The lysis andfixation were obtained by adding 500 mL of FACSLysingsolutionTM (BD Biosciences). Samples were analyzedwithin 1 h on a FACScanto (BD Biosciences).

The new system was compared with our referencesystem by analyzing simultaneously the same samplesusing CD3-PE-Cyanin 5 (PE-Cy5, clone UCHT1), CD4-PE-texas red (PETR, clone SFCI12T4D11), CD8a-PE (cloneSFCI21thyD3), and CD45-FITC (clone B3821F4A) (Tetra-ChromeTM, Beckman-Coulter) and CD16-FITC (IgG1,clone DJ130c, DakoCytomation, Copenhagen, Denmark),CD56-PE (IgG1, clone N901, Beckman-Coulter), CD19PE-TR (clone 89B Beckman-Coulter) and CD3 PE-Cy5(clone UCHT1, DakoCytomation), 10–20 mL, on 100 mLof blood in the same conditions. Lysis and fixation wasperformed using ImmunoprepTM on T-Q prepTM (Beck-man-Coulter) before adding FlowcountTM beads (Beck-man-Coulter) and analysis on EPICS XL flowcytometer(Beckman-Coulter).

Instrument setting on FacsCanto were performed usingfresh blood in nine tubes (unlabeled, six single labeled,six isotype controls and the mixed antibodies), with Fac-slyse/no wash procedure using FACS DivaTM calculationfacilities (BD Biosciences). Detection and compensationstability were tested daily using SetUpTM beads and FACS-CantoTM Software (BD Biosciences). On Epics XL system,compensations were recalculated weekly, using a standard-ized procedure (Cyto-compTM, Beckman-Coulter).

Dot plots were analyzed using FACS DivaTM (BD Bio-sciences) or Coulter Epics System IITM (Beckman-Coulter)softwares, respectively, using similar gating: CD45 vs. Sidescatter identification, except for B and NK cells on XL sys-tem where a (Forward/Side scatter). On XL system (Tetra-oneTM), lymphocytes were further selected on FS/SSexcluding dead lymphocytes, which number could be sig-nificant in delayed analyses. T and B cells were selected onCD3þ or CD19þ/SS lymphocytes. CD4þ and CD8þ were

analyzed on CD3þ and NK (CD16/CD56) on CD3-lympho-cytes. On Epics XL, all the gates were mathematicallydefined by the software.Absolute values were calculated manually from FACS-

Canto (10,000 lymphocytes events) and automatically onEpics XL (minimum of 7,500 lymphocytes and 1,000bead events, maximum 300 s analysis).Statistical analysis was performed using paired or

unpaired student’s t test, linear regressions, and Bland-Altman plot on Excel (Microsoft, USA) and expressed asmean 6 1SD. Coefficient of variations (CVs) were calcu-lated as 1SD/mean.

RESULTS

The mean waiting time for the sample analysis was3 min (i.e. 27 min for tube number 10), sufficient toallow cell sedimentation. Because samples are not prop-erly vortexed just before acquisition, less than 10 sam-ples were analyzed in each run.

Stability of the Instrument

Analysis of standard beads differed significantly fromthe manufacturer target values (first data on Fig. 1a) butwere highly stable over a period of 3 months (CVs from0.50 to 0.90% for the blue laser and 1.4 to 1.9% for thered laser) after exclusion of four aberrant values becauseof transient fluidic problems. Results were not influ-enced by operators.

Six Color Compensation Setting

The automatic compensation settings available onFACScanto software appeared not stable on daily settingof SetUpTM Beads (Fig. 1b) over 3 months (CV from 2.4to 22%) and occasional aberrant settings occurred. Fur-thermore, unexpected compensations were required (i.e.between two lasers; Table 1).

Lymphocyte 6 Parameter Protocol Setting

Sample analyses with automatic compensation settingswere not satisfactory. Manual compensations were setup using DIVA software facilities, but this was difficultbecause the FACSlysis/no wash procedure gave bigdebris and poor signal-over-noise ratios, making lympho-cytes difficult to select on a FS/SS dot plot. Alternativeprocedures, such as lyse-and-wash or anti-IgG coatedbeads devoted to compensation settings (i.e. Comp-BeadsTM, BD Biosciences), were not acceptable for lyse/no wash protocols. Final settings gave good definition ofleukocytes (Fig. 2a), B cells (Fig. 2b), CD3þ T cell sub-sets (Fig. 2c), and NK cells (Fig. 2d).To simplify and to secure the antibody distribution,

premixing of antibodies for 50 samples was found to bestable for a few days without any reagent interferencewhen compared with antibodies distributed individuallyin the tube (results not shown) and were subsequentlyused.

30 LAMBERT ET AL.

Exactitude Using International Quality Controls

The analysis of ImmunoTrol (lot 758734) results were goodfor T cells and poor for B and NK cells when compared with

the XL results and target values provided by the manufacturer

(Table 2). Overall, the FACSCanto absolute valueswere higher

than Epics XL values as reported in the NEQAS survey.

FIG. 1. Standardized 7 color set-up beads (BD Biosciences) wereanalyzed everyday for 3 monthsusing FACSCanto Software. One outof three results is plotted, showinghighly stable results except on flu-idic instability events that wererepaired. Automatic compensationsanalyzed everyday for 3 months werenot very stable for unexplained rea-sons. [Color figure can be viewed inthe online issue, which is availableat www.interscience.wiley.com.]

Table 1Compensation Matrix: Mean of Compensations Automatically Calculated Using 7 Color Setup Beads and

FACSCanto Software, Over a Period of 3 Months

FITC PE PerCP PerCP Cy5.5 PE-Cy7 APC APC-Cy7

FITC 0.96 6 0.16 0 0 0.13 6 0.02 0 0PE 18.8 6 0.44 0.07 6 0.05 0 1.6 6 0.14 0 0PerCP 2.4 6 0.30 14.7 6 1.2 4.5 6 0.37 0.9 6 0.10 0.5 6 0.14PerCP Cy5.5 2.4 6 0.30 14.7 6 1.2 4.5 6 0.37 0.9 6 0.10 0.5 6 0.14PE-Cy7 0.27 6 0.02 1.4 6 0.11 8.7 6 0.52 21.2 6 1.3 0.16 6 0.02 5.6 6 0.75APC 0.01 0.09 6 0.08 6.8 6 0.82 3.9 6 0.8 0.05 6 0.03 21.1 6 3.6APC-Cy7 0.0 0.01 6 0.02 0 2.9 6 0.27 2.4 6 0.24 2.8 6 0.61

The highest corrections were FITC on PE; PR on PerCP; PerCP on PE-Cy7 and APC; PerCP-Cy5.5 on PE-Cy7 and even APC andAPC-Cy7; and APC-Cy7 on APC. Results were not much stable with high standard deviations (in bold), especially for PE on PerCP;PerCP on PE-Cy7, APC and APC-Cy7; PE-Cy7 on APC-Cy7, APC on APC-Cy7 and APC-Cy7 on PE-Cy7 and APC.

31LYMPHOCYTE MONITORING WITH 6 COLOR FLOWCYTOMETRY

Similarly, two (number 154 and 155) NEQAS sampleanalyzes were graded satisfactory (within 2 standarddeviations) even if lower than Trucount mean values(FACSCalibur) and closer to the Flowcount (Epics XL)mean values (Table 3). B cell data were in line with therest, but NK cell data were much lower than what wasreported on all systems. The global data had a largeintercenter variability (global CV > 20%).

Reproducibility of 6 Color Lymphocyte Analysis

Because sample preparation steps are crucial for abso-lute value determination, the reproducibility of the sys-tem was tested on 5 repeats, on the same day, by thesame operator. A poor reproducibility was obtained onthe two (154 and 155) NEQAS samples (Table 3; Fig. 3a)possibly because of difficulties to gate leukocyte popula-tions after stabilization process of the samples. Resultson fresh samples were better (Fig. 3b), especially afterimproving operator-dependant factors (specially byproper sample mixing, wiping the tip, reverse pipetting,and using a calibrated, fixed volume, electronic pipette).The CVs dropped from 12 to 3% and were correct for Band NK even in one severe CD4þ lymphopenia (3 cells/mL,CV¼ 17.9%; patient 6).The importance of managing operator-dependant

parameters was confirmed by analyzing one fresh sampleby six different operators (one biologist, two residents,two technicians, and one student technician) in tripli-cates. The results show good reproducibilities: CD4þ T(CV from 1.6 to 9.7%), CD8þ T (0.7 to 8.1%), B (0.5 to6.3%), and NK (1.2 to 8.3%) cells (Fig. 3c). Three opera-tors had CVs below 2% for CD4þ T and B cells.The reproducibility over the time was good on NEQAS

samples one day after the first analysis (Fig. 3a). On alonger term, Immuno-Trol (lot 758734) daily analysis bydifferent operators (26 repeats over a 8-week period;Fig. 4) was highly stable (CD4þ T cells CV ¼ 4.5%;CD8þ T cells CV ¼ 8.8%) excluding one value (per-formed by a neophyte operator). Results were compara-ble to our previous data (Flowcount system, 25 repeats:CVs 8.6% for CD4þ T cells and 9.3% for CD8þ T cells).The B and NK cell counts had poor reproducibility (CV:21 and 11.1%, respectively; tested on Trucount only).

Stability of Samples

Because samples may not always be analyzed within4 h of withdrawal, we analyzed the stability of sampleskept at 48C over 6 consecutive days. The cell numberson day 0 (973 and 660 CD4þ T cells/mL and 774 and

FIG. 2. Representative dot plots; lymphocytes were selected onCD45 vs. side scatter dot plot (a). T cells and B cells were electroni-cally selected on CD3þ (not shown) or CD19þ (b) vs. side scatter, fromgated lymphocytes. CD4þ and CD8þ were analyzed on CD3þ gatedlymphocytes (c). CD16/CD56 were selected on CD3 negative lym-phocytes (d). Unconventional T cells were occasionally observed:CD4þCD8dim (e), CD4dimCD8þ (f), CD8dimþ (g), CD3 bright (h),which may express CD8 (i). CD56 was occasionally expressed by CD8þ

T cells (j) or CD3 bright (k), while NK cells could express low levels ofCD8 (m). [Color figure can be viewed in the online issue, which isavailable at www.interscience.wiley.com.]

<

32 LAMBERT ET AL.

158 CD8þ T cells/mL, in patients a and b, respectively)were quite stable, even for NK cells (Fig. 5).

Similarly, because the processed sample may notalways be analyzed soon after it is prepared, the stabilityof the preparations (sample with beads) was tested on5 consecutive days. The absolute values gradually de-creased over the time, using FACSLysing solution (whichcontain a preservative; Figs. 5c and 5d) and even morerapidly using a fixative-free lysing solution (PharM Lyse,BD Biosciences; Fig. 5e).

Lymphocyte Count Accuracy

Washed mononuclear cells purified by gradient densitywere diluted 1 in 2 repeatedly in their own washed pelletthat contained red blood cells and granulocytes but werepartially depleted in mononuclear cells. Results from twopatients (Fig. 6) were close to the linear regressioncurves on log/log graph over the ranges tested (patient a:CD4þ T cells from 502 to 23; CD8þ T cells from 810 to31; patient b: CD4þ T cells from 491 to 14; CD8þ T cellsfrom 1850 to 66 cells/mL), even for B cells and NK cells,as long as the concentration was higher than 10 cells/mL.

Comparison With Our Reference Technique

The two systems (antibodies, lysis, and instruments)were compared on 67 consecutive fresh samples, ana-lyzed simultaneously. The cell counts were widely dis-tributed: CD4þ T cells from 10 to 1333 cells/mL, CD8þ

T cells from 71 to 1965 cells/mL, B cells from 22 to 698cells/mL, and NK cells from 2 to 872 cells/mL (Epics XLdata). The FACScanto data were highly correlated (Fig. 7)

with Epics XL values (r2: CD4þ T cells 0.906 and CD8 Tcells 0.873). Absolute values were very close (slopes:CD4 T cells 1.005 and CD8 T cells 1.009). Correlationswere not so good for B cells (r2 ¼ 0.87; slope ¼ 0.89)and specially NK cells (r2 ¼ 0.54, slope 0.809) becauseof the differences of gating strategies as confirmed byBland-Altman graphs (Fig. 7).

Unconventional T Cells

Unconventional T cell subsets were observed on theFACScanto system as expected from previous studies(from 5 to 10% of T cells). Among the 67 patients wehave tested, 6 patients had a clear unusually raised(range: 20–140 cells/mL) CD4þCD8dim T cells (Fig. 2e)and 3 had a raised (range: 20–39 cells/mL) CD4dimCD8þ

T cells (Fig. 2f). Absolute values of these rare eventswere correlated between the two techniques(CD4þCD8dim T cells r2 ¼ 0.745; CD4dimCD8þ T cells:r2 CD4dimCD8þ T cells 0.76) but were found to be dis-similar (slopes: 2.29 and 0.87, respectively) probablybecause of different gating strategies. Moreover, sevenpatients had a raised (range: 20–150 cells/mL)CD8dimCD4neg T cells (Fig. 2g) with good correlationbetween the two techniques (r2 ¼ 0.669; slope ¼0.875). Finally, three patients had a raised CD3 bright Tcell population (Fig. 2h), suggesting a rise of circulatinggdT cells. In some instance, these CD3 bright T cellcould express low levels of CD8 (Fig. 2i).

Further T Cell Subsets

A consistent subset of T cells (in 15/67 patients) brightlyexpressed the CD56 (range: 10–71 cells/mL; Fig. 2d). These

Table 2Quality Control Analysis Reproducibility During an Eight and Four Week Survey on Each System

Cells/mL Target values Epics XL CV (%) FACSCanto CV (%)

CD4þ T cells 499 508 6 43.8 8.6 537 6 24.4 4.5CD8þ T cells 252 250 6 23.3 9.3 235 6 22.6 9.6B cells 148 NA 177 6 37.1 21NK cells 118 NA 172 6 19.0 11.1

Exactitude of lymphocyte count (cells/mL) was evaluated daily by comparing external quality con-trol (ImmunoTrolTM, lot 758734) target values with our results on Epics XL (n ¼ 25; 8 weeks) andlater, on FACSCanto 6 color (13 repeats during 4 weeks). (NA: non available).

Table 3Results Obtained from Repeated Analyses of Two UK NEQAS Quality Control Samples (Number

154 and 155), Compared with the Global Results from the International Survey

Flowcount survey(n ¼ 58)

Trucount survey(n ¼ 95)

Our data(5 repeats) CV (%)

Sample 154 (cells/mL)CD4þ T cells 980 6 170 1120 6 170 1004 6 112 11CD8þ T cells 220 6 40 290 6 90 216 6 26 12B cells 240 6 80 320 6 80 289 6 29 10NK Cells 140 6 140 150 6 60 121 6 16 13

Sample 155 (cells/mL)CD4þ T Cells 780 6 70 940 6 60 836 6 60 5.9CD8þ T Cells 480 6 60 590 6 20 459 6 43 7B cells 110 6 30 180 6 20 2066 13 6NK Cells 120 6 50 130 6 10 80 6 7 6.1

Our data were lower than the Trucount international means but close to the Flowcount data andto our previous results.

33LYMPHOCYTE MONITORING WITH 6 COLOR FLOWCYTOMETRY

are mainly CD8þ T cells (Fig. 2j) or CD3 bright (gdTcells; Fig. 2k). The CD56 expression on FACSCanto wasnot univocal, as it was mixed with CD16, but CD16 was

rarely expressed on T cells when used separately on EpicsXL. Conversely, NK cells frequently (22/67 of patients)expressed CD8 (Fig. 2m) (range: 20–458 cells/mL; 6–94%

FIG. 3. Reproducibility (5 repeats) was tested on two NEQAS samples (a) with CV ranging from 6 to 12% (arrows represent respective target val-ues); on six fresh samples (b) with improving CV (<3%) using reverse pipetting, new calibrated, fixed volume, electronic pipette. One fresh sample(c) was prepared by six different operators (global CV < 6.2%). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

FIG. 4. Daily analysis of a quality control (Immuno-Trol; lot 758734) by different operators shows good stability on FACSCanto (26 repeats during8-week period), compared with the 25 last analyses performed on EPICS XL (during 8 weeks). Lines indicate the respective target values. [Color fig-ure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

34 LAMBERT ET AL.

of NK cells). The CD8 was then at lower density whencompared with CD3þ T cells. In summary, CD8 dim ex-pression can be observed on CD3þCD4þCD8dim, CD3þCD4-CD8dim, CD3 bright gdT, and CD3 negative NK Cells.

Polynuclear Leukocytes

Finally, the use of CD45 made possible to detect a raise inpolynuclear basophils (low CD45/low side scatter (Fig. 2a))to be differentiated from blast and in polynuclear eosino-phils (CD45 bright/high side scatter (Fig. 2a)).

DISCUSSION

Our results show that this new 6 color system can beused reliably for routine analysis and that new, poten-tially relevant, information is emerging.

FIG. 5. The stability (a, b) was tested on samples kept at 48C during 6consecutive days before preparation and on two processed sample (c, d)mixed with the beads in FACSlysingTM solution. The values were muchless stable in fixative-free PharMLyseTM solution (BD Biosciences; e).[Color figure can be viewed in the online issue, which is available atwww.interscience.wiley.com.]

FIG. 6. The count accuracy was checked by serial dilution of twowashed peripheral mononuclear cells in their own red blood cell pellet(last values). Values were perfectly linear as far as they were higherthan 10 cells/mL. [Color figure can be viewed in the online issue,which is available at www.interscience.wiley.com.]

35LYMPHOCYTE MONITORING WITH 6 COLOR FLOWCYTOMETRY

FIG. 7

36 LAMBERT ET AL.

The automatic compensation procedures available arenot yet optimized for 6 colors and not secure becauseno real time data (dot plots) or error message are acces-sible to the operator. However, manual compensationsetup is accessible to any user, provided recommenda-tions are respected and assistance facility is used. Set-tings must be performed using the sample process thatwill be ultimately used because lysis/wash methodsstrongly interfere with compensation settings. This is amajor concern for mild/dimmed fluorescence intensitysuch as CD56 or unconventional T cell subsets.

Using the new fluorochromes brings about new com-pensation problems. Because of its spectral overlap onPE-Cy7, PerCP-Cy5.5 (proposed by the manufacturer) isless convenient than PerCP for a pan leukocyte (CD45)labeling. Interlaser compensations were needed evenwith separated laser beans possibly due to direct activa-tion of cyanins (used in PerCP-Cy5.5 and PE-Cy7 tan-dems) by the red laser. Inversely, APC-Cy7 excited bythe red laser was detected by PE-Cy7 detector despitelaser separation. Thus, CD8-APC would be more appro-priate than CD8-APC-Cy7 for reliable determination ofunconventional CD4þ/CD8þ populations.

Several points remain to be defined by users consensus.The compensation settings were highly stable and daily re-calculation could even introduce errors. A weekly checkmay be reasonable. The real positive signal threshold shouldbe determined on unlabeled cells (considering only auto-fluorescence) but non-specific binding must be excluded,especially on pathologic samples. The use of internal con-trols (e.g. binding of CD4/CD8 on CD19þ cells) do notexclude all bias on pathologic cells occasionally observed.Full minus one strategy must then be performed (26).

The absolute count based on a prepared fixed numberof beads (Trucount) depended much on the operator.This is even more crucial because the recommendedsample volume is very small (50 mL) for the viscous andheterogeneous whole blood. A correct sample mixing iscrucial before pipetting. Similarly, excessive blood, lyingalong the outside of the tip that was dipped in the sam-ple, has to be wiped so as to minimize influences on thefinal count. The alternative method is less operator-dependant (FlowcountTM, CytocountTM from Dakocyto-mation) because any manipulating error is reproducedfor the pipetting of the sample and the beads in solutionby the same operator, same pipette.

Bead gating may also be a critical point. CD45/Sidescatter plot must be avoided because of wide heteroge-neity (Fig. 1a). The proportion of doublets (that shouldbe considered as singlets) may vary with bead preserva-tion and sample interaction. This strategy or exclusionof dead cells on XL system may explain an overestima-

tion of international quality controls (i.e. NEQAS) whencompared with most of other systems. Monitoring ofbead flow rate during acquisition would be of interest.Six color analysis has several advantages on clinical point

of view. Our and previous works (9–16) show additional Tcell subsets at highly variable levels with probable clinicalrelevance that have to be clarified. The CD4þCD8dim Tcells are most probably derived from the CD4þ T cells butexpress the CD8aa homodimeric isoform suggesting theycould be related to the mucosal immunity (9,10; personaldata) with preferentially a Th1 activity (28). On the otherhand, the CD4dimCD8 T cell has CD8ab heterodimeric iso-form (related to the systemic immune system; 11–13) andpreferentially Th2 activity (121). The less known CD8 dim Tcells derive from the CD8aa T cell subset that share proper-ties with gdT cells (14–16,28). Each subset can haverestricted diversity (11, personal data). A rise of gdT cellscan also be indirectly detected because these T cells expresshigher levels of CD3 and occasionally CD8 (29). A role ofgdT cells has been described in asthma, cancer, auto-immune diseases, mucosal homeostasis, and wound healing(cited in 29). Expression of CD56 on CD8þ T could be asso-ciated with cytotoxic activity (19), while the significance ofthe CD8 expression on NK cells (CD3neg, CD16/CD56þ) isnot elucidated yet. Finally, basophilia and hypereosinophiliahave been clearly associated with parasitic infestation,allergy, or drug reaction (30).The combination proposed by the manufacturer may

not be optimum and using TCRab/TCRgd would be moreinformative than CD19/CD3 except maybe in B lymphocy-tosis related to Epstein Barr virus infections (7), wheremore than CD19 B cell markers would be relevant.A few technical features of that system have to be

improved rapidly: the forward scatter has poor definition;the data management system is neither safe nor userfriendly; automatic gating and ‘‘Snap-to-gate’’ are not work-ing properly; sample identification and mixing, acquisitionmonitoring (such as minimal count of events, maximumtime in acquisition), reports, calculation of absolute value.

ACKNOWLEDGMENTS

We thank Pascale Saby, Martine Celle, and Nadine Bar-del for skillful technical assistance and Clare Walker forthorough proofreading.

LITERATURE CITED1. Helbert M, Breuer J. Monitoring patients with HIV disease. J Clin

Pathol. 2000;53:266–272.2. 1997 revised guidelines for performing CD4þ T-cell determinations

in persons infected with human immunodeficiency virus (HIV).Centers for Disease Control and Prevention. MMWR Recomm Rep1997;46(RR-2):1–29.

3. Members of the General Haematology Task Force of BCSH. Guide-lines for the enumeration of CD4þ T lymphocytes in immunosup-pressed individuals. CD4þ T lymphocyte Working Party. Clin LabHaematol 1997;19:231–241.

4. Mandy F, Bergeron M, Houle G, Bradley J, Fahey J. Impact of theinternational program for Quality Assessment and Standardizationfor Immunological Measures Relevant to HIV/AIDS: QASI. Cytome-try 2002;50:111–116.

5. Lacombe F, Durrieu F, Briais A, Dumain P, Belloc F, Bascans E, Reiffers J,Boisseau MR, Bernard P. Flow cytometry CD45 gating for immunophe-notyping of acute myeloid leukemia. Leukemia 1997;11: 1878–1886.

FIG. 7. Two system comparisons: 67 consecutive fresh samplessimultaneously analyzed on FlowcountTM, ImmunoprepTM, Epics XL(Beckman-Coulter) and TrucountTM, FACSlysingTM, and FACSCantoTM

systems by the same operator were highly correlated. Bland-Altmancurves show systemic bias for B and NK cells. [Color figure can beviewed in the online issue, which is available at www.interscience.wiley.com.]

<

37LYMPHOCYTE MONITORING WITH 6 COLOR FLOWCYTOMETRY

6. Festin R, Bjorkland A, Totterman TH. Multicolor flow cytometricanalysis of the CD45 antigen provides improved lymphoid cell dis-crimination in bone marrow and tissue biopsies. J Immunol Meth-ods 1994;177:215–224.

7. Morrissey PE, Lorber KM, Marcarelli M, Bia MJ, Kliger AS, LorberMI. Immunologic monitoring after organ transplantation: relation-ship between Epstein-Barr virus infection and CD19þ B cells meas-ured by flow cytometry. Transplant Proc 1995;27:1428–1430.

8. Bellido M, Rubiol E, Ubeda J, Estivill C, Lopez O, Manteiga R, Nom-dedeu JF. Rapid and simple immunophenotypic characterization oflymphocytes using a new test. Haematologica 1998;83:681–685.

9. Ortolani C, Forti E, Radin E, Cibin R, Cossarizza A. Cytofluorimetricidentification of two populations of double positive (CD4þ,CD8þ)T lymphocytes in human peripheral blood. Biochem Biophys ResCommun 1993;191:601–609.

10. Luhtala M, Lassila O, Toivanen P, Vainio O. A novel peripheralCD4þ CD8þ T cell population: inheritance of CD8alpha expressionon CD4þ T cells. Eur J Immunol 1997;27:189–193.

11. Lambert C, Iobagiu C, Genin C. Persistent Oligoclonal CD4dimCD8þTT cells in peripheral blood. Cytometry 2005;66B:10–17.

12. Zloza A, Sullivan YB, Connick E, Landay AL, Al-Harthi L. CD8þ Tcells that express CD4 on their surface (CD4dimCD8bright T cells)recognize an antigen-specific target, are detected in vivo, and can beproductively infected by T-tropic HIV. Blood 2003;102:2156–2164.

13. Sullivan YB, Landay AL, Zack JA, Kitchen SG, Al-Harthi L. Upregulationof CD4 on CD8þ T cells: CD4dimCD8bright T cells constitute an acti-vated phenotype of CD8þ T cells. Immunology 2001;103: 270–280.

14. Suni MA, Ghanekar SA, Houck DW, et al. CD4(þ)CD8(dim) T lym-phocytes exhibit enhanced cytokine expression, proliferation andcytotoxic activity in response to HCMV and HIV-1 antigens. Eur JImmunol 2001;31:2512–2520.

15. Sun J, Kavathas PB. Comparison of the roles of CD8 alpha alphaand CD8 alpha beta in interaction with MHC class I. J Immunol1997;159:6077–6082.

16. Gao GF, Willcox BE, Wyer JR, Boulter JM, O’Callaghan CA, MaenakaK, Stuart DI, Jones EY, Van Der Merwe PA, Bell JI, Jakobsen BK.Classical and nonclassical class I major histocompatibility complexmolecules exhibit subtle conformational differences that affect bind-ing to CD8alphaalpha. J Biol Chem 2000;275:15232–15238.

17. Steele CR, Oppenheim DE, Hayday AC. Gamma(delta) T cells: non-classical ligands for non-classical cells. Curr Biol 2000;10:R282–R285.

18. Carding SR, Egan PJ. Gammadelta T cells: functional plasticity andheterogeneity. Nat Rev Immunol 2002;2:336–345.

19. Pittet MJ, Speiser DE, Valmori D, Cerottini JC, Romero P. Cytolyticeffector function in human circulating CD8þ T cells closely correlateswith CD56 surface expression. J Immunol 2000;164:1148–1152.

20. Whitby L, Granger V, Storie I, Goodfellow K, Sawle A, Reilly JT,Barnett D. Quality control of CD4þ T-lymphocyte enumeration:results from the last 9 years of the United Kingdom National ExternalQuality Assessment Scheme for Immune Monitoring (1993–2001).Cytometry 2002;50:102–110.

21. Barnett D, Granger V, Mayr P, Storie I, Wilson GA, Reilly JT. Evalua-tion of a novel stable whole blood quality control material for lym-phocyte subset analysis: results from the UK NEQAS immune moni-toring scheme. Cytometry 1996;26:216–222.

22. Baudouin F, Sarda MN, Goguel A, Bene MC. Multicenter study of refer-ence stabilized human blood for lymphocyte immunophenotypingquality control in flow cytometry. GEIL. Cytometry 1999;38: 127–132.

23. Van Blerk M, Bernier M, Bossuyt X, Chatelain B, D’Hautcourt JL,Demanet C, Kestens L, Van Bockstaele D, Crucitti T, Libeer JC.National external quality assessment scheme for lymphocyte immu-nophenotyping in Belgium. Clin Chem Lab Med 2003;41:323–330.

24. Bergeron M, Shafaie A, Ding T, Phaneuf S, Soucy N, Mandy F, Brad-ley J, Fahey J. Evaluation of stabilized blood cell products as candi-date preparations for quality assessment programs for CD4 T-cellcounting. Cytometry 2002;50:86–91.

25. Luider J, Cyfra M, Johnson P, Auer I. Impact of the new BeckmanCoulter Cytomics FC 500 5-color flow cytometer on a regional flowcytometry clinical laboratory service. Lab Hematol 2004;10:102–108.

26. O’Gorman MR, Thomas J. Isotype controls—time to let go? Cytome-try 1999;38:78–80.

27. Roederer M. Spectral compensation for flow cytometry: visualiza-tion artifacts, limitations, and caveats. Cytometry 2001;45:194–205.

28. Yamada K, Kimura Y, Nishimura H, Namii Y, Murase M, Yoshikai Y.Characterization of CD4þ CD8alphaalphaþ and CD4-CD8alphaalpha-þintestinal intraepithelial lymphocytes in rats. Int Immunol 1999;11:21–28.

29. Lambert C, Genin C. CD3 bright lymphocyte population reveal gd Tcells. Cytometry B Clin Cytometry 2004;61B:45–53.

30. Brito-Babapulle F. The eosinophilias, including the idiopathic hyper-eosinophilic syndrome. Br J Haematol 2003;121:203–223.

38 LAMBERT ET AL.