454 Am J Clin Pathol 2007;128:454-463454 DOI: 10.1309/TVGKD5TVB7W9HHC7

© American Society for Clinical Pathology

Hematopathology / AUTOMATED IMMATURE GRANULOCYTE ENUMERATION

Automated Enumeration of Immature Granulocytes

Bernard Fernandes, MD,1 and Yukio Hamaguchi2

Key Words: WBC differential count; Immature granulocytes; Automated hematology analyzer; Flow cytometry; Reproducibility

DOI: 10.1309/TVGKD5TVB7W9HHC7

A b s t r a c t

The performance characteristics of the XE-2100(Sysmex, Kobe, Japan) automated immaturegranulocyte (IG) count were studied. The automated IG count was compared with the manual morphologycount and with a proposed reference flow cytometriccount. The comparison data were analyzed by bothleast-squares and Passing-Bablok regression analysis.Long-term imprecision using preserved blood qualitycontrol specimens at different levels showed a rangefrom 2.59% to 3.57% coefficient of variation (CV) forwithin-run imprecision and 3.57% to 6.85% CV fortotal imprecision. The within-run reproducibilityperformed using fresh blood on 3 different specimensshowed a range from 5.55% to 8.24% CV. The countswere stable at both room temperature and afterrefrigeration for 24 hours.

Passing-Bablok regression analysis showedexcellent agreement between the proposed referenceflow cytometric IG count and the XE-2100 IG count,while there was less agreement with the manualmorphology count. Our results indicate that theautomated IG count can replace the manualmorphology count for IG counting in the clinicallaboratory. The results also confirm that the flowcytometric IG count is superior to and can replace the manual morphology count as a reference method for IG counting.

The WBC differential count is one of the most useful andfrequently requested tests in the clinical laboratory.1

Originally, it could only be done manually by a skilled tech-nologist examining a stained blood film under the microscope.During the last 50 years, improvements in blood cell countinginstruments have resulted in the majority of differential cellcounts now being done by automated instruments. Theseinstruments are able to provide more accurate and precisecounts than manual methods.2,3

Conventional blood cell counting instruments can enu-merate only the WBCs normally found in the peripheralblood, ie, neutrophils, lymphocytes, monocytes, eosinophils,and basophils. Nevertheless, there are other cells not normal-ly found in the peripheral blood, such as immature granulo-cytes (IGs), blasts, atypical lymphocytes, and nucleatedRBCs, all of which are indicators of different disease states.

The XE-2100 (Sysmex, Kobe, Japan) is an automatedhematology analyzer that can perform a conventional differentialcell count and can also enumerate IGs. We evaluated the per-formance characteristics of the XE-2100 automated IG count.

Materials and Methods

Sample Analysis

Blood samples used for this study were obtained fromresidual blood from specimens sent to the laboratory for routineCBC counts. All blood samples were collected in K3EDTAanticoagulant. A total of 263 samples were analyzed, 203 frompatients with a variety of diseases and conditions, includingacute and chronic inflammation, sepsis, malignancies, hemol-ysis, arthritis, thalassemia, pregnancy, and chemotherapy, and

Am J Clin Pathol 2007;128:454-463 455455 DOI: 10.1309/TVGKD5TVB7W9HHC7 455

© American Society for Clinical Pathology

Hematopathology / ORIGINAL ARTICLE

60 from healthy adult volunteers. The sample from any givenpatient was analyzed only once by each method. All sampleswere analyzed within 8 hours of collection, first on the SysmexXE-2100 analyzer and then by flow cytometry. The sampleswere analyzed by the XE-2100 in the closed mode exceptrarely, when only small quantities were available, in whichcase they were analyzed in open (capillary) mode.

Three levels of commercial quality control specimens (e-Check, Sysmex) were run daily on the XE-2100. The XE-2100 performed a differential count by a combination of lightscatter and fluorescence emission. An aliquot of blood is dilut-ed after lysis of RBCs and incubated with a polymethine-based RNA- and DNA-binding fluorescent dye. The cells areanalyzed in an optical block using a semiconductor laser.Neutrophils, eosinophils, monocytes, and lymphocytes aredifferentiated on the basis of their light scatter and fluores-cence emission characteristics using electronic cluster analy-sis protocols. To enable analysis of IGs, the XE Pro softwaremodule (Sysmex) has to be added to the XE-2100. The basicXE Pro module enhances reagent management, printing ofgraphic data, output of analysis data, and quality management.The XE IG Master submodule (Sysmex) electronically deter-mines the cluster of IGs from the granulocyte cluster on thedifferential histogram. IGs are recognized by their increasedfluorescence emission compared with segmented neutrophilsbecause they contain more RNA and DNA. The IG Pro soft-ware determines the center of the long axis of the granulocytecluster, calculates the lower half of the cluster, and uses a mir-ror image of this to complete the upper half of an ellipse rep-resenting mature neutrophils. All events above this ellipse areenumerated as IGs. These events with higher fluorescencerepresent the higher RNA and DNA content of IGs.

Flow cytometric enumeration of IGs was performedusing the method of Fujimoto et al.4 Briefly, K3EDTA antico-agulated blood was incubated for 30 minutes with CD16–flu-orescein isothiocyanate, CD11b-phycoerythrin, andCD45–peridinin chlorophyll protein for 30 minutes in the darkat room temperature. Flow cytometric analysis was performedwithin 60 minutes of sample preparation on an Epics XL-MCL flow cytometer (Beckman Coulter, Brea, CA). At least10,000 events were counted for each sample. After sequentialgating using CD45/side scatter to separate granulocytes andeosinophils from lymphocytes, monocytes, basophils, anddebris and CD16 to separate neutrophils from eosinophils, IGswere identified by CD16/CD11b gating. IGs were recognizedby the lack of staining for CD16. CD11b was positive in someof these cells (least IGs) but not in the early IGs. The IG countwas expressed as a percentage of the total WBC count. Theabsolute count was derived by using the total WBC countobtained from the XE-2100.

Manual differential cell counting was performed accordingto the National Committee for Clinical Laboratory Standards

(NCCLS) H20-A protocol.5 Three blood films were preparedby a wedge technique and stained with a modified Wrightstain within 2 hours of receipt in the laboratory. The slideswere randomly assigned to 2 experienced laboratory technol-ogists who performed a 200-cell differential count. The thirdblood film was used for arbitration if there were discordantresults between the 2 technologists. The total of promyelo-cytes, myelocytes, and metamyelocytes was included as theIG count. The percentage count was obtained from the com-bined 400-cell count. The absolute count was derived by usingthe total WBC count obtained from the XE-2100.

Within-Run Imprecision (Reproducibility) Using FreshBlood Samples

Within-run imprecision (reproducibility) using freshblood samples was performed in 3 different samples of blood.Each sample was run 10 consecutive times in the open andclosed modes. The mean, SD, and coefficient of variationwere determined for the IG absolute and proportional counts.

Long-Term Imprecision Using Quality Control Material

Long-term imprecision was determined in accordance withthe protocols in the NCCLS document EP5-A6 and using com-mercial quality control materials (e-Check). Three levels of qual-ity control material were run in duplicate 3 times per day for aperiod of 30 days. The first and last runs were used for calculat-ing within-run and total imprecision as described in NCCLSdocument EP5-A. The mean, SD, and coefficient of variation ofthe within-run and total imprecision were determined.

Stability

Short-term stability was assessed by measuring 10 sam-ples immediately after the blood was drawn and after 5, 15,30, and 60 minutes of storage at room temperature. Long-termstability was assessed with samples stored at room tempera-ture and at 4°C through 8°C (refrigerated). Ten samples wereassayed after 4, 8, 12, 24, 36, 48, 56, and 72 hours.

Reference Range

The reference range was generated from 60 samplesobtained from healthy adult volunteers (30 men and 30 women).

Results

The results for within-run imprecision (reproducibility)using fresh blood samples are shown in ❚Table 1❚ and ❚Table 2❚

for 3 samples. Table 1 shows results for samples assayed by theopen mode and Table 2 for samples assayed by the closed mode.

The results for imprecision performed on quality controlmaterial are shown on ❚Table 3❚ and ❚Table 4❚. Table 3 showsthe results for within-run imprecision and Table 4 for totalimprecision.

456 Am J Clin Pathol 2007;128:454-463456 DOI: 10.1309/TVGKD5TVB7W9HHC7

© American Society for Clinical Pathology

Fernandes and Hamaguchi / AUTOMATED IMMATURE GRANULOCYTE ENUMERATION

❚Table 1❚Within-Run Reproducibility (Imprecision) Performed on Fresh Blood Samples (Open Mode)

Specimen No./Parameter Mean SD CV (%) Manufacturer Specifications

1No. of IGs (× 109/L) 0.38 0.03 6.63 SD < 0.12; CV < 25%IGs (%) 6.13 0.37 6.01 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 6,250 (6.3) 0.09 1.46 CV < 3%Neutrophils, % 53.8 (0.54) 1.19 2.22 CV < 8%

2No. of IGs (× 109/L) 0.98 0.06 5.84 SD < 0.12; CV < 25%IGs (%) 8.46 0.47 5.55 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 11,520 (11.5) 0.25 2.17 CV < 3%Neutrophils, % 62.02 (0.62) 0.61 0.98 CV < 8%

3No. of IGs (× 109/L) 0.35 0.03 8.53 SD < 0.12; CV < 25%IGs (%) 2.39 0.20 8.24 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 14,440 (14.4) 0.15 1.04 CV < 3%Neutrophils, % 72.96 (0.73) 0.49 0.67 CV < 8%

CV, coefficient of variation; IGs, immature granulocytes.

❚Table 2❚Within-Run Reproducibility (Imprecision) Performed on Fresh Blood Samples (Closed Mode)

Specimen No./Parameter Mean SD CV (%) Manufacturer Specifications

1No. of IGs (× 109/L) 0.95 0.09 9.26 SD < 0.12; CV < 25%IGs (%) 12.71 1.18 9.32 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 7,480 (7.5) 0.11 1.51 CV < 3%Neutrophils, % 48.91 (0.49) 1.43 2.91 CV < 8%

2No. of IGs (× 109/L) 1.86 0.09 4.96 SD < 0.12; CV < 25%IGs (%) 13.19 0.69 5.20 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 14,100 (14.1) 0.19 1.37 CV < 3%Neutrophils, % 87.13 (0.87) 0.65 0.74 CV < 8%

3No. of IGs (× 109/L) 1.07 0.05 5.11 SD < 0.12; CV < 25%IGs (%) 5.89 0.28 4.83 SD < 1.5; CV < 25%WBC count, /µL (× 109/L) 18,190 (18.2) 0.25 1.35 CV < 3%Neutrophils, % 89.17 (0.89) 0.26 0.29 CV < 8%

CV, coefficient of variation; IGs, immature granulocytes.

❚Table 4❚Total Imprecision: SD and CV on Three Levels of Quality Control Material

Level 1 Level 2 Level 3

Parameter Mean SD CV (%) Mean SD CV (%) Mean SD CV (%)

No. of IGs (× 109/L) 10.1 0.47 4.65 11.2 0.41 3.63 2.23 0.04 3.34IGs (%) 0.28 0.01 3.57 0.77 0.03 3.77 12.1 0.83 0.42

CV, coefficient of variation; IGs, immature granulocytes.

❚Table 3❚Within-Run Imprecision: SD and CV on Three Levels of Quality Control Material

Level 1 Level 2 Level 3

Parameter Mean SD CV (%) Mean SD CV (%) Mean SD CV (%)

No. of IGs (× 109/L) 10.1 0.43 4.27 11.2 0.36 3.32 2.23 0.83 0.42IGs (%) 0.28 0.01 4.43 0.77 0.02 3.15 12.1 0.41 3.35

CV, coefficient of variation; IGs, immature granulocytes.

Am J Clin Pathol 2007;128:454-463 457457 DOI: 10.1309/TVGKD5TVB7W9HHC7 457

© American Society for Clinical Pathology

Hematopathology / ORIGINAL ARTICLE

A total of 10 specimens were analyzed for short-term sta-bility. The results for this analysis are shown in ❚Figure 1❚.

Long-term stability was analyzed in 10 samples kept atroom temperature and 10 samples kept at 4°C (refrigerated).One of the samples at room temperature and 2 of the samplesat 4°C were excluded because data were not available for theentire 72-hour period. The results of the long-term stabilitystudy are shown in ❚Figure 2❚.

Of the 263 samples analyzed for comparison of the XE-2100, manual microscopy, and flow cytometric methods forcounting IGs, 18 were excluded from all 3 data sets becauseincomplete data were available and 1 sample was excluded as anoutlier. Another 15 samples were excluded from the flow cyto-metric data set because the data were incomplete. For compar-isons between the manual count and the XE-2100 and flow cyto-metric counts, only samples in which the XE-2100 count wasmore than 0.25% were used; there would be an artifactual propor-tional bias below this range if all samples were used because a400-cell count can only count down to 0.25%. Thus, 201 sampleswere available for comparison between the XE-2100 and manu-al counts, 229 samples for comparison between the XE-2100 andflow cytometric counts, and 186 samples for comparison betweenmanual and flow cytometric counts. The method comparisonswere performed using linear regression analysis and Passing-Bablok regression analysis. The results of the method compar-isons using linear regression analysis are shown in ❚Figure 3❚. Theresults of the method comparisons using Passing-Bablok regres-sion analysis are shown in ❚Figure 4❚ and ❚Table 5❚.

A total of 60 samples from healthy adult volunteers (30men and 30 women) were analyzed to establish a referencerange. Two samples were excluded because they were neu-tropenic. The results of the reference range study on theremaining 58 samples are shown in ❚Table 6❚.

Discussion

The origins of the differential cell count can be traced tothe pioneering work of Ehrlich and Romanovsky toward theend of the 19th century.7 By using synthetic dyes to stainperipheral blood films, they demonstrated the existence ofmorphologically different WBC populations. By the early tomid 20th century, enumeration of these cells in the form of amanual leukocyte differential cell count became an integralpart of laboratory testing. The introduction of automated cellcounting in the latter part of the 20th century culminated in theroutine use of the automated 5-part WBC differential count inthe clinical laboratory. Because the automated differentialcount is more precise and less labor-intensive than the manu-al differential count,2,8 it has resulted in the twin benefits ofimproved quality and reduced costs in the clinical laboratory.A major limitation of the conventional automated 5-part dif-ferential count is the inability to identify cells not normallyfound in the peripheral blood, such as IGs, or that are presentin such small numbers that they are not detected in the usual100- or 200-cell differential count. The availability of auto-mated identification and counting of IGs offers the possibilityof further improvements in quality and costs in the laboratory.

In this study, we evaluated the performance characteris-tics of the automated IG count performed by the Sysmex XE-2100 analyzer. A comparison of the XE-2100 IG count withthe manual count of promyelocytes, myelocytes, andmetamyelocytes using linear regression revealed a correlationcoefficient of 0.80 for percentage counts. When the manualcount was converted to an absolute count by using the totalWBC count obtained by the analyzer and correlated by usingthe absolute IG count of the XE-2100, the correlation coeffi-cient was 0.82. This indicates a strong relationship between

00 8 16 24

Time (min)

No

. of

IGs

32 40 48 56

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 8 16 24

Time (min)

32 40 48 560

Per

cen

t o

f IG

s

3

6

9

12

15

A B

❚Figure 1❚ Short-term stability study performed on 10 specimens. A, Absolute counts. B, Proportional counts. IGs, immaturegranulocytes.

458 Am J Clin Pathol 2007;128:454-463458 DOI: 10.1309/TVGKD5TVB7W9HHC7

© American Society for Clinical Pathology

Fernandes and Hamaguchi / AUTOMATED IMMATURE GRANULOCYTE ENUMERATION

the 2 methods of counting IGs and validates the replacementof the traditional manual microscopic IG count by the XE-2100 IG count.

The high degree of imprecision of the manual differentialcount has been well documented in the literature.9,10 Thisproblem is magnified when the number of events counted issmall, as is the case with IGs.11,12 The manual count is, there-fore, unsuitable and inappropriate as a reference method forcounting rare events such as IGs. To address this problem,many investigators have attempted to devise alternative refer-ence methods for enumeration of IGs by using flow cytome-try. Pioneering investigations by Terstappen et al13 and Lund-Johansen and Terstappen14 using flow cytometry demonstrat-ed differential expression of antigens on granulocytes duringmaturation. Thus, CD16 is expressed only in mature neu-trophils, not in IGs, whereas CD11b is expressed on some IGsbut not in the very early stages. CD45 is expressed during all

stages of granulocyte maturation. Several workers have usedanti-CD16 in combination with other antibodies againstleukocyte antigens to identify IGs in peripheral blood.15-17

Fujimoto et al4 described a flow cytometric method using anti-CD16, anti-CD11b, and anti-CD45 to identify IGs anddemonstrated good correlation between this assay andmicroscopy of sorted cells. This method represents a potentialreference method for enumeration of IGs.

In our comparison of the flow cytometric IG count withthe XE-2100 IG count using linear regression, there wasexcellent correlation between the 2 (correlation coefficients of0.93 and 0.95 for absolute and proportional counts, respective-ly). This indicates a strong relationship between the XE-2100IG count and the proposed flow cytometric reference methodand validates its use in clinical practice. There was also goodcorrelation between the XE-2100 and manual counts (correla-tion coefficients of 0.82 and 0.80 for absolute and proportional

0 8 16 24

Time (h)

32 40 48 56 64 720

Per

cen

t o

f IG

s

4

8

12

0

0.5

1.0

0 8 16 24

Time (h)

No

. of

IGs

32 40 48 56 64 72

1.5

0 8 16 24

Time (h)

32 40 48 56 64 720

Per

cen

t o

f IG

s

1

2

3

4

5

0

0.1

0.2

0.3

0.4

0.5

0 8 16 24

Time (h)

No

. of

IGs

32 40 48 56 64 72

0.6

A B

C D

❚Figure 2❚ Long-term stability study performed on specimens stored at 4°C (A and B) and room temperature (C and D). Dataare shown for absolute counts (A and C) and proportional counts (B and D). IGs, immature granulocytes.

Am J Clin Pathol 2007;128:454-463 459459 DOI: 10.1309/TVGKD5TVB7W9HHC7 459

© American Society for Clinical Pathology

Hematopathology / ORIGINAL ARTICLE

XE

-210

0 IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

Manual IG Count (x 109/L)

1.5

Flo

w C

yto

met

ry IG

Per

cen

t

20

10

15

5

00 5 10 15

Manual IG Percent

20

Flo

w C

yto

met

ry IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

Manual IG Count (x 109/L)

1.5

Flo

w C

yto

met

ry IG

Per

cen

t

20

10

15

5

00 5 10 15

XE-2100 IG Percent

20

Flo

w C

yto

met

ry IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

XE-2100 IG Count (x 109/L)

1.5

XE

-210

0 IG

Per

cen

t

20

10

15

5

00 5 10 15

Manual IG Percent

20

A B

C D

E F

❚Figure 3❚ Least squares regression analysis between manual and XE-2100 (Sysmex, Kobe, Japan) immature granulocyte (IG)counts (A and B) and XE-2100 and flow cytometric IG counts (C and D), and manual and flow cytometric IG counts (E and F).A, y = 0.8865x + 0.1137; r = 0.8174. B, y = 1.0855x + 1.1477; r = 0.8000. C, y = 1.0087x – 0.0018; r = 0.9305. D, y = 1.0927x –0.1581; r = 0.9535. E, y = 0.8307x + 0.119; r = 0.6982. F, y = 1.2304x + 0.9854; r = 0.7837.

460 Am J Clin Pathol 2007;128:454-463460 DOI: 10.1309/TVGKD5TVB7W9HHC7

© American Society for Clinical Pathology

Fernandes and Hamaguchi / AUTOMATED IMMATURE GRANULOCYTE ENUMERATION

Flo

w C

yto

met

ry IG

Per

cen

t

20

10

15

5

00 5 10 15

Manual IG Percent

20

Flo

w C

yto

met

ry IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

Manual IG Count (x 109/L)

1.5

Flo

w C

yto

met

ry IG

Per

cen

t

20

10

15

5

00 5 10 15

XE-2100 IG Percent

20

Flo

w C

yto

met

ry IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

XE-2100 IG Count (x 109/L)

1.5

XE

-210

0 IG

Per

cen

t

20

10

15

5

00 5 10 15

Manual IG Percent

20

XE

-210

0 IG

Co

un

t (x

109 /

L)

1.5

1.0

0.5

00 0.5 1.0

Manual IG Count (x 109/L)

1.5

A B

C D

E F

❚Figure 4❚ Passing-Bablok regression analysis between manual and XE-2100 (Sysmex, Kobe, Japan) immature granulocyte (IG)counts (A and B) and XE-2100 and flow cytometric IG counts (C and D), and manual and flow cytometric IG counts (E and F).A, y = 1.5466x + 0.0415. B, y = 1.5593x + 0.5407. C, y = 1.0438x – 0.007. D, y = 1.0774x – 0.1295. E, y = 1.7867x + 0.0225. F, y= 1.8648x + 0.2386.

Am J Clin Pathol 2007;128:454-463 461461 DOI: 10.1309/TVGKD5TVB7W9HHC7 461

© American Society for Clinical Pathology

Hematopathology / ORIGINAL ARTICLE

counts, respectively). There was also acceptable correlationbetween the proposed flow cytometric reference method andthe manual method (correlation coefficients of 0.70 and 0.78for absolute and proportional counts, respectively).

However, it is recognized that there are serious flawswhen using least squares linear regression to perform methodcomparisons between 2 laboratory analytic methods.18 Themost serious flaw is that this statistical analysis assumes thatthe reference method is measured without error. This condi-tion patently does not hold for the manual differential count.It also assumes that the SE of the new method is normally dis-tributed and constant for the entire measurement range. Thereare virtually no clinical laboratory tests that can meet theseconditions.

To overcome these difficulties, it has been recommendedthat a more robust statistical procedure, the Passing-Bablokregression analysis, be used for method comparison of clinicallaboratory methods.19 This nonparametric procedure does notmake any assumptions and is not susceptible to the variance ofeither method.

We compared the different methods of counting IGs byPassing-Bablok regression analysis; the results are shown inFigure 4 and Table 5. The comparison between the manual

microscopic and proposed reference flow cytometric methodsshows a discernible linear proportional bias as can be detect-ed by visual inspection of the graphic plot and the fact that thenumber 1.00 is not within the 99% confidence intervals for theslope. There is a similar linear proportional bias in the com-parison between the manual microscopic and XE-2100 meth-ods. The manual microscopic count consistently underesti-mates the IGs at low counts. This bias is most likely due to thehigh imprecision of the manual count and the small number ofcells counted, which compromises accuracy for cells presentin very small numbers.

A notable negative skew is present for the manual count.One possible explanation for this is improper classification ofmetamyelocytes as band cells and excluding them from the IGcount. Band cells are notoriously difficult to classify by man-ual microscopy.20 Another possibility is that the flow cytomet-ric count includes some mature neutrophils, thereby produc-ing the skew. However, experiments using cell-sorting proce-dures have confirmed that the flow cytometric method doesnot include mature neutrophils.4 The comparison between theXE-2100 and flow cytometric methods demonstrates com-plete agreement for the analysis of absolute numbers and anegligible proportional bias for the comparison of percent-ages. This confirms the superiority of the XE-2100 count overthe manual count and validates the use of the flow cytometricmethod as a reference method for counting IGs.

The NCCLS document “Reference LeukocyteDifferential Count (Proportional) and Evaluation ofInstrument Methods”5 has been used for validation of instru-mental leukocyte differential counts for more than 10 years.This procedure uses a 400-cell manual count to reduce thevariance of the count. However, although it may be acceptable

❚Table 6❚Adult Reference Range for the XE-2100 IG Count

Parameter Mean 1 SD 2 SD 3 SD

No. of IGs (× 109/L) 0.01 0.01 0.02 0.03IGs (%) 0.22 0.10 0.20 0.30

IGs, immature granulocytes.

❚Table 5❚Estimate of Agreement Between Different Counting Methods Based on Passing-Bablok Regression Analysis for the ImmatureGranulocyte Count

Methods Compared/Parameter Estimate 99% Confidence Interval

XE-2100* vs manual absolute countSlope 1.55 1.29 to 1.83Intercept 0.04 0.03 to 0.05

XE-2100 vs manual percentageSlope 1.56 1.29 to 1.87Intercept 0.54 0.40 to 0.80

Manual vs flow cytometric absolute countSlope 1.78 1.50 to 2.20Intercept 0.02 0.009 to 0.04

Manual vs flow cytometric percentageSlope 1.87 1.54 to 2.30Intercept 0.24 0.12 to 0.48

XE-2100 vs flow cytometric absolute countSlope 1.04 0.99 to 1.10Intercept –0.007 –0.01 to 0.004

XE-2100 vs flow cytometric percentageSlope 1.07 1.02 to 1.15Intercept –0.13 –0.22 to –0.08

* Sysmex, Kobe, Japan.

462 Am J Clin Pathol 2007;128:454-463462 DOI: 10.1309/TVGKD5TVB7W9HHC7

© American Society for Clinical Pathology

Fernandes and Hamaguchi / AUTOMATED IMMATURE GRANULOCYTE ENUMERATION

for relatively high proportional counts, it is not suitable forcounts of less than 5%. Clearly, as instruments become capa-ble of identifying abnormal cells in the peripheral blood thatare present in very small proportions, an alternative referencemethod for WBC differential counts is required. A flow cyto-metric method is ideal for this purpose. Fluorescence flowcytometric methods using monoclonal antibodies can beexpected to produce the most accurate counts because of theirflexibility and ability to gate around specific cell populations.The proposed flow cytometric reference method for countingIGs is eminently suitable for the purpose. A flowcytometric–based method has also been proposed for countingnucleated RBCs in the peripheral blood,21 and this method hasbeen validated in the field.22

The IG count showed excellent reproducibility in closedand open modes. The coefficient of variation for reproducibil-ity of the IG count compared favorably with the total WBCcount and the absolute and proportional neutrophil counts.

The short-term stability measured for a 1-hour period wasexcellent. The long-term stability was evaluated for a 72-hourperiod with specimens left at room temperature or refrigerat-ed. There was good stability at room temperature and withrefrigeration for a 24-hour period, and, as can be expected,stability deteriorated in blood samples left at room tempera-ture after that period. The stability of the XE-2100 IG countwould be of particular benefit for laboratories with samplesshipped in from a long distance.

The long-term imprecision measured for a period of 30days using a commercial control sample was excellent andcomparable to that of the total WBC and proportional counts.The within-run and total imprecision are superior to the man-ufacturers’ claims. The availability of suitable quality controlmaterial is certainly an asset for introducing this assay into theclinical laboratory.

The manual differential count is time-consuming andexpensive for the laboratory, and laboratories, therefore, relyon instrument “flags” indicating an abnormality may be pres-ent to avoid performing manual reviews or differential countson normal specimens. Analytic instruments provide an “IG”flag to indicate that IGs are potentially present in the periph-eral blood. A recent study indicated that this flag was respon-sible for most of the false-negative (52%) blood film reviewsperformed in hematology laboratories.23 The XE-2100 IGcount can be expected to eliminate this component of falsenegativity and further reduce false positivity related to nonspe-cific flagging. Having an instrument that can accurately countIGs should reduce the unnecessary review of peripheral bloodfilms significantly.

The availability of an automated IG count brings the clini-cal laboratory an additional step closer to the holy grail of auto-mated cell analysis, ie, a complete, automated, extended differ-ential count of all cells in the peripheral blood. An immediate

benefit of this in the clinical laboratory will be the decreasedneed for a manual differential cell count on CBC specimensthat generate an IG flag. Furthermore, the IG count obtainedand reported by the instrument will be more accurate and pre-cise than that from a manual count because it is well estab-lished that the instrument count is more precise than the man-ual count.24

The ability of instruments to count abnormal cells in theperipheral blood rapidly and accurately, even when they arepresent in very small numbers, portends the end of the manu-al differential cell count and a new era in the hematology lab-oratory. The IG count has usefulness in the diagnosis andmanagement of a large number of hematologic and nonhema-tologic diseases. Many of these associations have been estab-lished using manual counting. Further studies using a moreaccurate and precise IG count offer the opportunity to reeval-uate and redefine these associations. Preliminary studies haveshown promising results in using the XE-2100 IG count forscreening for sepsis and infection.25-27

From 1Department of Laboratory Medicine, Mount Sinai Hospitaland University of Toronto, Toronto, Canada; and 2Reagent Group,Sysmex, Kobe, Japan.

Address reprint requests to Dr Fernandes: Dept ofLaboratory Medicine, Mount Sinai Hospital, 600 University Ave,Toronto, Canada, M5G 1X5.

References1. Houwen B. The differential cell count. Lab Hematol.

2001;7:89-100.2. Simson E, Groner W. The state of the art for the automated

WBC differentials. Lab Hematol. 1995;1:13-22.3. Fernandes B, Hamaguchi Y. Performance characteristics of

the Sysmex XT-2000i hematology analyzer. Lab Hematol.2003;9:189-197.

4. Fujimoto H, Sakata T, Hamaguchi Y, et al. Flow cytometricmethod for enumeration and classification of reactiveimmature granulocyte populations. Cytometry. 2000;42:371-378.

5. National Committee for Clinical Laboratory Standards.Reference Leukocyte Differential Count (Proportional) andEvaluation of Instrument Methods; H20-A. Villanova, PA:National Committee for Clinical Laboratory Standards; 1992.

6. National Committee for Clinical Laboratory Standards.Evaluation of Precision Performance of Clinical ChemistryDevices; EP5-A. Villanova, PA: National Committee forClinical Laboratory Standards; 1999.

7. Wittekind D. On the nature of Romanowsky dyes and theRomanowsky-Giemsa effect. Clin Lab Haematol. 1979;1:247-253.

8. Dutcher TF. Leukocyte differentials: are they worth the effort?Clin Lab Med. 1984;4:71-87.

9. Koepke JA, Dotson M, Shiffman MA. A critical evaluation of the manual/visual differential leukocyte counting method.Blood Cells. 1985;11:173-180.

Am J Clin Pathol 2007;128:454-463 463463 DOI: 10.1309/TVGKD5TVB7W9HHC7 463

© American Society for Clinical Pathology

Hematopathology / ORIGINAL ARTICLE

10. Rumke CL. The impression of the ratio of two percentagesobserved in differential leukocyte counting methods. BloodCells. 1985;11:137-140.

11. Rumke CL. The statistical expected variability in differentialleukocyte counting. In: Koepke JA, ed. Differential LeukocyteCounting. Skokie, IL: College of American Pathologists;1978:39-45.

12. Rumke CL. Statistical reflections on finding atypical cells.Blood Cells. 1985;11:141-144.

13. Terstappen LW, Safford M, Loken MR. Flow cytometricanalysis of human bone marrow, III: neutrophil maturation.Leukemia. 1990;4:657-663.

14. Lund-Johansen F, Terstappen LW. Differential expression of cell adhesion molecules during granulocyte maturation.J Leukoc Biol. 1993;54:47-55.

15. Lacombe F, Lacoste L, Briais A, et al. A flow cytometricreference method for quantifying immature granulocytes and blast cells in peripheral blood samples: report from theExtended Differential Panel [abstract]. Lab Hematol.1998;4:104.

16. Lacombe F, Durrieu F. Extended differential in the clinicallaboratory: flow cytometry methods [abstract]. Lab Hematol.2001;7:9.

17. Hübl W, Andert S, Thum G, et al. Value of neutrophil CD16expression for detection of left shift and acute-phase response.Am J Clin Pathol. 1997;107:187-196.

18. Passing H, Bablok W. Comparison of several regressionprocedures for method comparison studies and determinationof sample sizes: application of linear regression procedures formethod comparison studies in clinical chemistry, part II. J ClinChem Clin Biochem. 1984;22:431-445.

19. Passing HJ, Bablok W. A new biometrical procedure for testingthe equality of measurements from two different analyticalmethods: application of linear regression procedures formethod comparison studies in clinical chemistry, part I. J ClinChem Clin Biochem. 1983;21:709-720.

20. Novak RW. The beleaguered band count. Clin Lab Med.1993;14:895-903.

21. Tsuji T, Sakata T, Hamaguchi Y, et al. New rapid flowcytometric method for the enumeration of nucleated red bloodcells. Cytometry. 1999;37:291-301.

22. Fernandes B, Bergeron N, Hamaguchi Y. Automatednucleated red blood cell counting in the perinatal period. Lab Hematol. 2002;8:179-188.

23. Barnes PW, Mcfadden SL, Machin SJ, et al. The InternationalConsensus Group for Hematology Review: suggested criteriafor action following automated CBC and WBC differentialanalysis. Lab Hematol. 2005;11:83-90.

24. Pierre RV. Peripheral blood film review: the demise of the eyecount leukocyte differential. Clin Lab Med. 2002;22:279-297.

25. Briggs C, Kunka S, Fujimoto H, et al. Evaluation of immaturegranulocyte counts by the XE-IG master: upgraded software for the XE-2100 automated hematology analyzer. Lab Hematol.2003;9:117-124.

26. Ansari-Lori MA, Kidder TS, Borowitz MJ. Immaturegranulocyte measurement using the Sysmex XE-2100:relationship to infection and sepsis. 2003;120:795-799.

27. Nigro KG, O’Riordan M, Molloy EJ, et al. Performance of an automated immature granulocyte count as a predictor of neonatal sepsis. Am J Clin Pathol. 2005;123:618-624.