AJCP / Original Article
Do the Flags Related to Immature Granulocytes
Reported by the Sysmex XE-5000 Warrant a
Microscopic Slide Review?
Heidi Eilertsen, MSc,1,2 and Tor-Arne Hagve, MD, PhD1,3
From the 1Division of Diagnostics and Technology, Akershus University Hospital, Lrenskog, Norway; 2Faculty of Health Sciences, Oslo, and Akershus
University College of Applied Sciences, Oslo, Norway; and 3Institute of Clinical Medicine, University of Oslo, Akershus University Hospital, Oslo, Norway.
Key Words: Sysmex XE-5000; IG count; Imm Gran? flag; Analytical quality
Am J Clin Pathol October 2014;142:553-560
DOI: 10.1309/AJCP4V4EXYFFOELL
ABSTRACT
Objectives: The Sysmex XE-5000 instruments (Sysmex,
Kobe, Japan) count immature granulocytes (IGs) and use
the Imm Gran? flag to signal unreliable results. This
study investigated the usefulness of the Imm Gran? flag
and the analytical and diagnostic performance of the IG
measurements in a side-by-side evaluation.
Methods: In total, 408 samples were analyzed on three
XE-5000 instruments. The IG count and the Imm Gran?
flag reports from all three instruments were used for
reproducibility studies. The diagnostic performance of the
automated IGs and the Imm Gran? flag were studied by
comparing the XE-5000 results with the results of the manual
differential.
Results: The reproducibility of the Imm Gran? flagging
between instruments was poor (, 0.75-0.80). The most
significant contributor to the report of the Imm Gran?
flag was bands, and the flag played a minor role in detecting
blasts. The interinstrument reproducibility of the IG counts
was high (intraclass correlation, 0.99). The IG count
reported by XE-5000s was higher than the manual IG count
(36%-55%), and the difference and the variability tended to
increase with increasing levels of IGs.
Conclusions: The Imm Gran? flag has a poor analytical
quality and gives no substantial information on the presence
of blasts in the sample. We therefore suggest reporting the
automated IG count without initial microscopic slide review.
American Society for Clinical Pathology
Modern automated hematology instruments, in addition
to a fully five-part differential, also provide characterization
and counting of abnormal cells in the peripheral blood, such as
blasts, atypical lymphocytes, and immature granulocytes (IGs).
The Sysmex XE-5000 (Sysmex, Kobe, Japan) is a
fully automated hematology instrument that counts IGs
(metamyelocytes, myelocytes, and promyelocytes) in the
differential (DIFF) channel based on flow cytometry.1 In
addition, information on immature cells is derived from
the immature myeloid information (IMI) channel based on
radiofrequency and direct current measuring principles.1,2 In
the IMI channel, not only IGs but also bands and myeloblasts
are detected. If a difference occurs between the numbers
of cells detected in the DIFF channel and the IMI channel,
a suspect IG flag (Imm Gran?) is reported in addition to
the immature count derived from the DIFF channel.1 A flag
report indicates the presence of abnormal cells, a finding that,
according to the current recommendations, has to be verified
by a manual evaluation of cell morphology.3
According to our standard procedure, a blood smear
shall be examined for (1) all samples with an Imm Gran?
flag and (2) all samples with a relative IG count of 5%
or more (presented as an IG Present flag). The purpose
of the examination is to verify the presence of IGs and to
identify or exclude the presence of other pathologic cells
in peripheral blood such as blasts. The microscopic slide
review is not supposed to replace the automated IG count
with a manual IG count.
To our knowledge, no studies have investigated the
quality and the usefulness of the IG-related flags. The aim of
this study was thus to evaluate the analytical and diagnostic
performance of IG measurements as well as the usefulness of
Am J Clin Pathol 2014;142:553-560 553
553 DOI: 10.1309/ AJCP4V4EXYFFOELL 553
Eilertsen and Hagve / Usefulness of Immature Granulocyte Flags on the Sysmex XE-5000
the IG flag reports on the Sysmex XE-5000, questioning the
need for a manual slide review or a manual count of samples
with a flag report. Three XE-5000 instruments were evaluated
side-by-side. The interinstrument agreement and the accuracy
of the IG counts, as well as the usability of the IG flag reports,
were studied by comparing with a manual differential.
Materials and Methods
Sample Selection and Study Design
At Akershus University Hospital, Oslo, Norway, the
routine hematology samples are analyzed randomly on one
of three Sysmex XE-5000 instruments (hereafter, XE1, XE2,
and XE3) integrated onto a track-based automation system.
The instruments were purchased at the same time, calibrated
and harmonized by the manufacturer, and used with identical
software packages, including firmware upgrades. During a
period of 5 months, 408 samples from the daily workload
were selected and reanalyzed on all three instruments. For
the purpose of this study, all samples were kept anonymous.
Only samples reported initially with one or more of the four
suspect flags, blasts, IGs, atypical lymphocytes, and abnormal
lymphocytes/lymphoblasts were included. The triggering
thresholds for the suspect flags were all factory settings. The
IG count and the flag report results from all three instruments
were used for the study of interinstrument agreement. The
diagnostic performance of the IG count and the Imm Gran?
flag were determined by comparing the results reported by the
instruments with the results from a blood smear examination.
Blood Sampling and Analytical Conditions
Venous blood samples were collected in Vacuette tubes
(Greiner Bio-One, Frickenhausen, Germany) containing
dipotassium EDTA. The specimens were transported to
the laboratory by a pneumatic tube transporting system
and processed on the Sysmex XE-5000 instruments using
software version 00-04 in closed mode within 4 hours
after collection. The instruments were continuously involved
in the routine workload during the study period, and their
performance was monitored with an internal quality control
(QC) system, including Sysmexs QC material e-Check and
an external quality surveillance program. The QC system,
based on the e-Check, includes evaluation of the parameter IG
number and IG percentage but not the instruments flagging
performance.
Sysmex XE-5000
The Sysmex XE-5000 provides WBC differential counts
based on an optical principle, including forward, side
scatter, and side fluorescence. The XE-5000 has a DIFF
554 Am J Clin Pathol 2014;142:553-560
554 DOI: 10.1309/AJCP4V4EXYFFOELL
channel, which identifies and counts IG cells in addition to
the standard five-part differential populations. The number
of cells the instrument routinely counts in each sample to
generate a differential cell count depends on the total number
of WBCs, the dilution ratio, and the counted volume. At a
WBC count value of 5.0 109/L, 3,922 cells are counted
in the DIFF channel. The IG counts derived from the
DIFF channel include metamyelocytes, myelocytes, and
promyelocytes. The software reports both the absolute and
the relative IG counts.4 To alert crucial levels, one can flag
the samples with a user-defined flag, termed IG Present,
which occurs when the absolute or relative IG count exceeds
the user-defined threshold. In a separate IMI channel, not
only IGs but also bands and myeloblasts are detected. The
number of cells the instruments routinely counts using
radiofrequency in the IMI channel, at a WBC count value of
5.0 109/L, is 5,000 cells. The instrument software examines
the scattergrams from both the DIFF and the IMI channels.
If there are any disagreements in the information derived
from the scattergrams, the IG flag algorithm comes into
operation and the Imm Gran? flag will be triggered. The
IG count derived from the DIFF channel will, however, still
be reported. A Q value provides information on the degree of
positivity or negativity of the IG flag on a scale of 0 to 300,
with increments of 10 arbitrary units. At a given threshold,
the instrument will report the Imm Gran? flag. The factory
setting threshold is preset to an arbitrary value of 100 and is
user adjustable. In this study, the factory setting threshold
setting was 100 for trigging the Imm Gran? flag, and the
user-defined IG percentage threshold for triggering the IG
Present flag was 5%.
Manual Differential Leukocyte Count
For each sample, two blood smears were prepared
and stained with May-Grnwald-Giemsa using the Sysmex
SP-1000. A total of 200 cells were counted by two highly
trained technicians, each counting 100 cells. The total count of
metamyelocytes, myelocytes, and promyelocytes represented
the IG count. The mean IG percentage was calculated for
each sample based on the two manual differential counts. The
absolute count was computed by the WBC count obtained
from the XE-5000. To determine a positive smear finding,
we used the criteria listed by the International Consensus
Group for Hematology Review.3 The presence of a single
myelocyte/promyelocyte (0.5%) and/or two metamyelocytes
(1%) qualified for a true-positive smear finding.
Statistical Analysis
Statistical analyses were performed using the SPSS
software version 16 (SPSS, Chicago, IL) for Windows
(Microsoft, Redmond, WA) and Analyse-it version 2.26
(Analyse-it Software, Leeds, England). A two-tailed P value
American Society for Clinical Pathology
 AJCP / Original Article

of less than .05 was considered statistically significant. To
describe the data, the IG counts were summarized as median
and range before the analysis of concordance. The intraclass
correlation coefficient (ICC [2.1], two-way random, single
measures) was calculated to quantify the interinstrument
reliability of the IG counts, and values were used to determine
interinstrument reliability for the Imm Gran? flag.5 The
diagnostic performance of the automated IG counts was
examined by receiver operating characteristic (ROC) curve
analyses and compared for all three instruments with respect
to smear findings. Passing-Bablok regression was used for
the method comparison between the Sysmex XE-5000s
and the manual differential counts.6 Bland-Altman analysis
was used to determine the degree of agreement between the
Sysmex XE-5000s and the manual differential IG counts.7
Pearson correlation coefficient was used to determine the
association between the two variables. Logistic regression
was used to assess the association between the predictors;
band, metamyelocytes, myelocytes, promyelocytes, and
blasts analyzed by manual microcopy of the smears; and
the IG flag generated by the Sysmex XE-5000 as the
dichotomous dependent outcome.8 All significant variables
in the univariate analysis were included in the multivariate
analysis to adjust for confounders.
Results
The results of the leucocyte counts of the 408 samples
performed on the three Sysmex instruments varied from
leukopenia to severe leukocytosis (0.5-207 109/L) with a
median of 9.9 109/L and mean of 20.3 109/L. A manual
differential count of the samples showed metamyelocytes
(1% cells) in 133 samples, myelocytes (0.5% cells) in
280, promyelocytes (0.5% cells) in 41, and blasts (0.5%
cells) in 48. No IGs were found in 31% of the samples, while
one, two, or three of the populations of myeloid cells were
found in 34%, 38%, and 7% of the samples, respectively.
The agreement between the two manual differential IG cell
counts in the 408 cases, presented as an ICC, was 0.903 (95%
confidence interval [CI], 0.884-0.920).
Table 1
Interinstrument Variability for the Imm Gran? Flag
Reports Generated by Three Sysmex XE-5000 Instruments
Pair of 95% Confidence P Value
Instruments a Interval
XE1/XE2 0.79 0.71-0.87 <.001
XE1/XE3 0.75 0.67-0.83 <.001
XE2/XE3 0.80 0.73-0.88 <.001
a The extent of agreement between two instruments (n = 408).
Samples With the Imm Gran? Flag
Based on the factory setting thresholds for flagging, the
XE1, XE2, and XE3 instruments reported the Imm Gran?
flag for 81, 80, and 77 samples, respectively. The values for
agreement between the pair of instruments were 0.79 (XE1/
XE2), 0.75 (XE1/XE3), and 0.80 (XE2/XE3) Table 1. One
or more instruments flagged for Imm Gran? in 102 samples.
The three instruments showed full agreement in reporting the
flag in 60 (59%) samples. Blast cells were detected by manual
counting in 17 (XE1), 17 (XE2), and 18 (XE3) of the samples
reported with an Imm Gran? flag. For most of these samples
(16, 15, and 16, respectively), a Blast? flag was also reported.
Only one to two of the 48 samples with blasts present in the
blood smear were reported with an Imm Gran? flag as a lone
flag. For these samples, one of the technicians observed one
blast by the 100-cell manual differential count, corresponding
to the presence of 0.5% circulation blasts in peripheral blood.
In addition, in one to two of the 48 samples with blasts present
in the blood smear, neither the Imm Gran? flag nor the
Blast? flag was reported by any of the instruments. For one
of these samples, one of the technicians observed one blast by
the 100-cell manual differential count, corresponding to the
presence of 0.5% circulation blasts in peripheral blood. Case 2
was a leukopenic sample (0.9 109/L).
The odds ratios (ORs) for association between the
myeloid cells observed in the blood film and the Imm
Gran? flag reported by XE1 are presented in Table 2. The
most significant contributor to the report of the Imm Gran?
flag is bands. The odds for a flag report increase by 8.2% per
percentage increase in the number of bands. The increase in
the odds per percentage increase in myeloblasts was 5.4%.

Table 2
Logistic Regression Analyses of the Risk of the Imm Gran? Flag Generated by Sysmex XE-5000 (XE1) Based on the Myeloid
Cell Proportions (%) in Blood Smears

Unadjusted Effect Adjusted Effect

Variable OR 95% CI P Value OR 95% CI P Value

Bands 1.082 1.046-1.119 <.001 1.082 1.041-1.124 <.001

Immature granulocytes 1.103 1.049-1.160 <.001 1.056 0.999-1.115 .053
Myeloblasts 1.044 1.019-1.070 <.001 1.054 1.028-1.081 <.001

CI, confidence interval; OR, odds ratio.

American Society for Clinical Pathology Am J Clin Pathol 2014;142:553-560 555

555 DOI: 10.1309/ AJCP4V4EXYFFOELL 555
Eilertsen and Hagve / Usefulness of Immature Granulocyte Flags on the Sysmex XE-5000
The results from the other two instruments were similar
(results not shown).
An IG count was reported for 46% to 52% of the samples
with an Imm Gran? flag by the XE-5000 instruments. On
the XE1, the IG count was 55% higher than the smear-based
count (y = 1.55x [95% CI, 1.00 to 3.15] + 0.00 [95% CI, 0.09
to 0.01]). Similar results were found for the XE2 and the XE3
(data not shown). The difference between the two methods as
well as the variability tended to increase with the level of the
IG counts (data not shown).
Samples With No Imm Gran? Flag
For the 306 samples with no Imm Gran? flag, an IG
count result was reported for 289, 295, and 279 samples
analyzed by XE1, XE2, and XE3, respectively. The IG counts
ranged from 0 to 4.7 109/L with a median of 0.2 109/L
and a mean of 0.4 109/L Table 3. The interinstrument
reproducibility of the absolute IG counts between the three
instruments, presented as an ICC, was 0.988 (95% CI, 0.985-
0.990). The interinstrument variability of the absolute IG
counts between XE1/XE2 and XE1/XE3 as measured by the
coefficient of variation was 14% and 19%, respectively.
Of the 306 samples with no Imm Gran? flag report,
the XE1, XE2, and XE3 instruments reported the flag IG
present (IG >5%) in 95, 90, and 96 samples, respectively.
Blast cells were detected by manual review in 14 (XE1), 12
(XE2), and 13 (XE3) of the samples reported with an IG
Present flag. For most of these samples (13, 12, and 12,
respectively), a Blast? flag also was reported. Consequently,
only one of the samples with blasts present in the blood smear
was reported with IG present as a lone flag. In addition,
in nine to 10 of the 48 samples with blasts present in the
smear, neither the IG Present flag nor the Blast? flag was
reported by any of the instruments.
Figure 1 presents the regression analysis of the results
of the manual IG count vs the automated count on the three
XE-5000 instruments. The coefficient of correlation varied
between 0.82 (manual count and XE3) and 0.89 (manual
count and XE2). The automated IG count was 36% to 40%
higher than the manual smear count. The difference between
the two methods as well as the variability tended to increase
with increasing levels of IGs Figure 2. The agreement
between the IG count from technician 1 and the XE1 and
technician 2 and the XE1, presented as an ICC, was 0.815
(95% CI, 0.751-0.861) and 0.738 (95% CI, 0.619-0.814),
respectively. Similar results were found for the XE2 and the
XE3 (data not shown).
Figure 3 shows the ROC curves for the overall
performance of the IG count on the Sysmex XE-5000s to
predict IGs when we used the criteria for a positive smear
established by the International Consensus Group.3 The areas
under the curve (AUCs) were 0.918 (95% CI, 0.886-0.949),
556 Am J Clin Pathol 2014;142:553-560
556 DOI: 10.1309/AJCP4V4EXYFFOELL
Table 3
Mean, Median, and Range for Immature Granulocyte (IG)
Counts Reported by Three Sysmex XE-5000 Instrumentsa
Instrument Mean, 109/L Median, 109/L Range, 109/L
XE1 0.44 0.15 0-4.28
XE2 0.43 0.15 0-4.59
XE3 0.46 0.18 0-4.72
a In total, 306 samples were measured on each instrument (XE1, XE2, and XE3).
0.931 (95% CI, 0.903-0.959), and 0.922 (95% CI, 0.891-
0.953) for XE1, XE2, and XE3, respectively.
Discussion
The main purpose of the present study was to evaluate the
usefulness of the IG flags generated by the Sysmex XE-5000.
The Sysmex XE-5000 reports two flag types that are related
to the presence of IGs in peripheral blood: the IG Present
flag, indicating the potential presence of IGs, and the Imm
Gran? flag, indicating the presence of suspect cells such as
blasts. Based on the established review criteria, blood smear
examination is performed to assess the validity of the reported
flags. This practice is time-consuming and expensive. Thus,
further research is needed to address the clinical significance
of the flagging system and the need for follow-up.
The main finding in this study is the insufficient
analytical performance of the Imm Gran? flag, indicating
that the Imm Gran? flag is not a sufficient criterion
for a microscopic review. The analysis of the same
samples on three instruments revealed low concordance in
generating the Imm Gran? flag. In only 60% of cases,
all three instruments reported the IG flag on the same
samples. These findings are difficult to explain because
all samples were analyzed at the same time and under the
same conditions on all three instruments to reduce the
factors that usually influence comparative evaluations.9,10
In addition, the instruments were under a continuous QC
system. It is, however, a problem that the commercial
control programs do not include an assessment of clinically
important cell types such as blasts or the instruments
flagging performance. With a common Q value cutoff
at 100, the observed value (0.75-0.80) for agreement
between pairs of instruments indicates poor reproducibility
of the Imm Gran? flag, supporting the conclusion that
the analytical performance of the flag is low. In general,
the usefulness of instrument-generated flags is evaluated
by their sensitivity and specificity.1,11-13 However, on
the basis of the present findings, we suggest that the
interinstrumental reproducibility in reporting flags also
has to be taken into account when the usefulness of flag is
evaluated.
American Society for Clinical Pathology
 AJCP / Original Article

A B

6 6
Sysmex XE-5000 IG Count (109/L) XE1

Sysmex XE-5000 IG Count (109/L) XE2

5 5

4 4

3 3

2 2

1 1
Identity Identity
Passing-Bablok Passing-Bablok
0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Manual IG Count (10 /L)
9
Manual IG Count (10 /L) 9

Figure 1 Passing-Bablok regression analysis of the absolute

C
immature granulocyte (IG) count determined by manual
6 differential and three Sysmex XE-5000 instruments (Sysmex,
Sysmex XE-5000 IG Count (109/L) XE3

Kobe, Japan). A, XE1. y = 1.36x (95% CI, 1.27-1.47) + 0.01

5 (95% CI, 0.00-0.01). B, XE2. y = 1.38x (95% CI, 1.29-1.50) +
0.00 (95% CI, 0.00-0.01). C, XE3. y = 1.40x (95% CI, 1.31-
4 1.49) + 0.00 (95% CI, 0.00-0.01).

1
Identity
Passing-Bablok
0
0 1 2 3 4 5 6
Manual IG Count (109/L)

An interesting observation in this study is that the most
important contributor to the report of the Imm Gran? flag
is the presence of band cells. It may thus be questioned
whether the Imm Gran? flag also is a measure of left-
shift. Consequently, depending on the actual guidelines, the
presence of band cells may result in a significant delay in the
report of results.
The present data also show that the Imm Gran? flag
plays only a minor role in detecting blasts. In samples with
an Imm Gran? flag, blasts were detected by slide review in
approximately 20% of the samples. In 88% to 94% of these,
a Blasts? flag was reported together with the Imm Gran?
flag. In accordance with previous results, in approximately 5%
of all samples with blasts, no Blasts? flag was reported.14
The IG Present flag is reported when the IG count
from the DIFF channel is increased above a given threshold
and thus depends only on the analytical quality of the
IG count and the level of the threshold. The present data
clearly demonstrate that the IG Present flag gives almost
no information on the presence of blasts in addition to the
Blasts? flag. In samples with blasts detected by a manual
slide review, a Blasts? flag in 93% to 100% of the cases
accompanied the IG Present flag. It is thus questionable
whether the presence of the Imm Gran? or IG Present
flag provides sufficient information to warrant a slide review.
In this study, up to two of the 48 samples with blasts detected
by a manual differential were reported with Imm Gran? as
the only flag. The consequence of omitting a slide review of
samples with Imm Gran? is the risk of missing myeloblasts
in peripheral blood, which may have an important clinical
impact. However, in the two present cases, the smear review
revealed only a single blast, observed by one of the two
technicians. This discordance could be due to the limitations
of the 100-cell manual differential or the fact that one of the

American Society for Clinical Pathology Am J Clin Pathol 2014;142:553-560 557

557 DOI: 10.1309/ AJCP4V4EXYFFOELL 557
Eilertsen and Hagve / Usefulness of Immature Granulocyte Flags on the Sysmex XE-5000
2.0
Difference Sysmex XE-5000 IG Count
1.5
Manual IG Count (109/L)
1.0
0.5
0 No discrimination
0.5
1.0
0.5 0.5 1.5
Mean of All IG Count (109/L)
Figure 2 Bland-Altman plot for the absolute immature
granulocyte (IG) counts determined by manual differential and
the Sysmex XE-5000 (XE1; Sysmex, Kobe, Japan). The 95%
limits of agreement are 0.525 to 0.797.
technicians overlooked the presence of blasts. It is well known
that manual cell classification is subjective and has limited
application in low-frequency cell populations.15,16 Thus, it is
difficult to assess the relevance of the single blast detection.
In a previous study, we found that the miss rate of the
Sysmex XE-5000 platform in detecting circulation blasts
was high, with an observed sensitivity of the Blasts? flag
from 0.54 to 0.75.14 In this context, the loss of clinical value
by disregarding the Imm Gran? flag seems insignificant.
However, the disregard of Imm Gran? as the sole reason
for manual review will result in a reduced number of smears,
simplified routines, reduced use of resources, and a shorter
turnaround time. Consequently, there is no obvious rationale
for manual examination of blood smears in cases flagged
solely for Imm Gran?.
The comparison of the Sysmex XE-5000 count with
the manual count of IGs (metamyelocytes, myelocytes,
and promyelocytes) in samples with no Imm Gran? flag
revealed a correlation coefficient of 0.82 to 0.89 for the
absolute count depending on the three instruments. This is
in agreement with previous studies.17,18 The comparison
between the manual count and the XE-5000 method
shows a linear proportional bias, with XE-5000 reporting
approximately 35% higher IG counts than the manual method.
Similar results were found in samples with an Imm Gran?
flag. This is in agreement with several previous studies
demonstrating that the manual count method underestimates
558 Am J Clin Pathol 2014;142:553-560
558 DOI: 10.1309/AJCP4V4EXYFFOELL
Identity 1.0
Bias (0.136)
95% Limits of
agreement
0.8
True-Positive Rate (Sensitivity)
0.6
0.4
XE1
0.2
XE2
XE3
0.0
0.0 0.2 0.4 0.6 0.8 1.0
False-Positive Rate (1 Specificity)
Figure 3 Receiver operator characteristic curves for overall
2.5 3.5 4.5 performance of the immature granulocyte count on the
Sysmex XE-5000 (Sysmex, Kobe, Japan) to predict immature
granulocytes in blood samples.
the counting of IGs compared with counting on the Sysmex
XE-2100 and by flow cytometry.17-19 High imprecision for
the manual method at low counts of cells is well known.15
However, the present study shows a negative bias as well
as a significantly higher variability for the manual method
at IG counts of 0.5 109/L or more. This may be explained
by the counting statistics as suggested by Koepke et al,16 but
it may also be a result of a nonhomogeneous distribution
of leukocytes on the blood smears. Wedge techniques for
preparing smears may cause nonhomogeneous distribution
of leukocytes in the films. Large cells as IGs may be pushed
to the edge and to the sides of the blood film and thereby
disappear from the counting area.20 Another possible reason
is that of misclassification. Inadequate classification of
metamyelocytes as bands excludes the metamyelocytes from
the manual IG count and thereby produces inconsistency
between automated and manual IG counting.21 It is difficult
to assess the significance of the inconsistency since the
comparison is halting for many reasons. The instruments
have an assumed statistical advantage over the manual
count by the larger number of cells counted.15 In this study,
the agreement observed between the IG counts from the
instruments (0.99) was greater than the agreement of IG
counts from the two manual counts (0.90). In addition, the
XE-5000 measurements had a smaller range of uncertainty.
Therefore, although quantitative discordant results are
difficult to evaluate, the IG counts from the XE-5000 are
supposed to be more accurate than the manual cell counts.
Consequently, there is no reason for not using the instrument
counts. However, a manual review is needed, which would
American Society for Clinical Pathology

yield quantitative results of the separate nonblastic IG types
(metamyelocytes, myelocytes, and promyelocytes).
The analytical accuracy of IG counts observed in
samples with and without an Imm Gran? flag was
comparable. These findings indicate that the IG count might
be reliable even when accompanied by an Imm Gran? flag
and can be reported.
The reproducibility of the absolute IG count obtained
from the three instruments evaluated by the ICC value was
high (0.99). The interinstrument variability between pairs of
instruments measured as percent coefficient of variation was
also acceptable (14%-19%). These findings are comparable
to those of Fernandes and Hamaguchi17 and Briggs et al22 for
within-run reproducibility of IGs on the Sysmex XE-2100
and indicate that random use of different instruments seems
to have a minor influence on the analytical performance with
respect to IG counts.
The ability of the instrument count to predict the
presence of IGs in smears is very good, with AUCs ranging
from 0.918 to 0.931. Manual examination of blood smears
does not necessarily count IGs with high accuracy depending
on the number of IGs in the sample.23 In previous studies,
difficulties in identifying IG populations by morphology due
to a low number of cells have been reported.24 The manual
count of IGs may thus not be the perfect gold standard in
an ROC curve test. In this study, however, the relatively high
numbers of IGs in the samples support the validity of the
ROC curve test.
Taken together, the present data confirm a systematic
bias in IG counting between the manual slide method and the
Sysmex XE-5000 method and that the analytical quality and
the diagnostic accuracy of the instrument counting are high.
This is an important finding since it has been demonstrated
that the IG count may have a central role as an inflammation
marker, especially in differentiating between infection and
inflammation of other causes.25-28
The agreement between the IG counts from the
XE-5000 instruments is good. The IG count from the
XE-5000 is higher than that from the manual smear count,
and the difference between the two methods increases
with an increasing number of IGs. The lack of agreement
between the manual IG count and the IG count from the
XE-5000 may be linked to errors in the manual counting.
The Imm Gran? flag has poor analytical quality and
gives no substantial information on the presence of blasts
in the sample. Therefore, we suggest reporting the IG
count from the XE-5000 without smear confirmation.
Address reprint requests to Ms Eilertsen: Faculty of Health
Sciences, Oslo and Akershus University College of Applied
Sciences, PO Box 4, St Olavs plass, N-0130 Oslo, Norway; heidi.
eilertsen@hioa.no.
American Society for Clinical Pathology
AJCP / Original Article
References
1. Briggs CJ, Linssen J, Longair I, et al. Improved flagging rates
on the Sysmex XE-5000 compared with the XE-2100 reduce
the number of manual film reviews and increase laboratory
productivity. Am J Clin Pathol. 2011;136:309-316.
2. Cimenti C, Erwa W, Herkner KR, et al. The predictive
value of immature granulocyte count and immature myeloid
information in the diagnosis of neonatal sepsis. Clin Chem
Lab Med. 2012;50:1429-1432.
3. Barnes PW, McFadden SL, Machin SJ, et al; International
Consensus Group for Hematology. Review: suggested criteria
for action following automated CBC and WBC differential
analysis. Lab Hematol. 2005;11:83-90.
4. Briggs C, Harrison P, Grant D, et al. New quantitative
parameters on a recently introduced automated blood cell
counter: the XE-2100. Clin Lab Haematol. 2000;22:345-350.
5. Shoukri MM. Measures of Interobserver Agreement. Boca
Raton, FL: CRC Press; 2004:39-61.
6. Bilic-Zulle L. Comparison of methods: Passing and Bablok
regression. Biochem Med. 2011;21:49-52.
7. Bland JM, Altman DG. Statistical methods for assessing
agreement between two methods of clinical measurement.
Lancet. 1986;1:307-310.
8. Altman DG. Practical Statistics for Medical Research. Boca
Raton, FL: CRC Press; 1999.
9. Imeri F, Herkloz R, Risch L, et al. Stability of hematological
analytes depends on the hematology analyser used: a stability
study with Bayer Advia 120, Beckman Coulter LH 750 and
Sysmex XE 2100. Clin Chim Acta. 2008;397:68-71.
10. de Baca ME, Gulati G, Kocher W, et al. Effects of storage
of blood at room temperature on hematologic parameters
measured on Sysmex XE-2100. Lab Med. 2006;37:28-36.
11. Kang SH, Kim HK, Ham CK, et al. Comparison of four
hematology analyzers, CELL-DYN Sapphire, ADVIA 120,
Coulter LH 750 and Sysmex XE-2100, in terms of clinical
usefulness. Int J Lab Hematol. 2008;30:480-486.
12. Ruzicka K, Veitl M, Thalammer-Scherrer R, et al. The
new hematology analyzer Sysmex XE-2100: performance
evaluation of a novel white blood cell differential technology.
Arch Pathol Lab Med. 2001;125:391-396.
13. Stamminger G, Auch D, Diem H, et al. Performance of
the XE-2100 leucocyte differential. Clin Lab Haematol.
2002;24:271-280.
14. Eilertsen H, Vllestad NK, Hagve TA. The usefulness of
blast flags on the Sysmex XE-5000 is questionable. Am J Clin
Pathol. 2013;139:633-640.
15. Rumke CL. Statistical reflections on finding atypical cells.
Blood Cells. 1985;11:141-144.
16. Koepke JA, Dotson MA, Shifmann MA. A critical
evaluation of the manual/visual differential leukocyte
counting method. Blood Cells. 1985;11:173-186.
17. Fernandes B, Hamaguchi Y. Automated enumeration of
immature granulocytes. Am J Clin Pathol. 2007;128:454-463.
18. Field D, Taube E, Heumann S. Performance evaluation of
immature granulocyte parameter on the Sysmex XE-2100
automated hematology analyzer. Lab Hematol. 2006;12:11-14.
19. Cherian S, Levin G, Wing YL, et al. Evaluation of an 8-color
flow cytometric reference method for white blood cell differential
enumeration. Cytometry B Clin Cytom. 2010;78:319-328.
20. Stiene-Martin EA. Causes for poor leukocyte distribution
in manual spreader-slide blood films. Am J Med Tech.
1980;46:624-632.
Am J Clin Pathol 2014;142:553-560 559
559 DOI: 10.1309/ AJCP4V4EXYFFOELL 559
Eilertsen and Hagve / Usefulness of Immature Granulocyte Flags on the Sysmex XE-5000
21. Novak RW. The beleaguered band count. Clin Lab
Med.1993;14:895-903.
22. Briggs C, Harrison P, Grant D, et al. Performance evaluation
of the Sysmex XE-2100, automated haematology analyser.
Sysmex J Int. 1999;9:113-119.
23. Valenstein PN. Evaluating diagnostic tests with imperfect
standards. Am J Clin Pathol. 1990;93:252-258.
24. Fujimoto H, Sakata T, Hamaguchi Y, et al. Flow cytometric
method for enumeration and classification of reactive immature
granulocyte populations. Cytometry. 2000;42:371-378.
25. Nierhaus A, Klatte S, Linssen J, et al. Revisiting the white
blood cell count: immature granulocytes count as a diagnostic
marker to discriminate between SIRS and sepsisa
prospective, observational study. BMC Immunol. 2013;14:8.
560 Am J Clin Pathol 2014;142:553-560
560 DOI: 10.1309/AJCP4V4EXYFFOELL
26. Briggs C, Kunka S, Fujimoto H, et al. Evaluation of immature
granulocyte counts by the XE-IG master: upgraded software
for the XE-2100 automated hematology analyzer. Lab
Hematol. 2003;9:117-124.
27. Ansari-Lari AM, Kickler T, Borowitz M. Immature
granulocyte measurement using the Sysmex XE-2100:
relationship to infection and sepsis. Am J Clin Pathol.
2003;120:795-799.
28. Nigro KG, ORiordan MA, 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.
American Society for Clinical Pathology