Download Automated Reticulocyte Counting and Measurement of Reticulocyte

HEMATOPATHOLOGY
Original Article
Automated Reticulocyte Counting and
Measurement of Reticulocyte Cellular Indices
Evaluation of the Miles H*3 Blood Analyzer
CARLO BRUGNARA, MD, MARY JO HIPP, MT (ASCP), CLS (NCA), PATRICIA J. IRVING, MT (ASCP),
HAZEL LATHROP, MT (ASCP), CLS (NCA), PATRICIA A. LEE, MT (ASCP), SH,
ELAINE M. MINCHELLO, MT (ASCP), AND JAMES WINKELMAN, MD
This study evaluated reticulocyte counting and measurement of reticulocyte cellular indices with the Miles H*3 blood analyzer, a new instrument that combines the Technicon/Miles technology for blood cells
counting with a staining technique allowing counting of reticulocytes,
quantification of staining intensity and measurement of reticulocyte cellular indices. Reticulocyte counts obtained with the Miles H*3 analyzer
were compared with those obtained by manual counting, flow cytometry
(thiazole orange method) and by the Sysmex R-3000 (Baxter Diagnostics) reticulocyte analyzer. Reticulocyte counting with the Miles H*3
showed excellent precision, and linearity in the range tested (1.1-49%
and 1-72% reticulocytes, respectively, with two different protocols)
with no significant carryover. Reticulocyte counts were stable after
storing blood samples for 72 hours at 4 °C. Comparison of the four
different methods, showed an acceptable intraclass correlation between
Miles H*3 and Sysmex R-3000 (intraclass correlation coefficient, [r,] =
.952), Miles H*3 and flow cytometry (r, = .922), and Sysmex R-3000
and flow cytometry (r, = .938). There was no satisfactory correlation
between any of the three automated methods and the values obtained
with manual counting of reticulocytes (r, = .538-.7S5), consistent with
the well known imprecision of the manual technique. For a group of
normal pediatric subjects, age 1-10, we obtained the following values
(±SD) of reticulocyte indices: mean corpuscular volume 97.6 ± 4.7 fL;
cell hemoglobin concentration mean 28.2 ±1.4 g/dL; cell hemoglobin
content 26.7 ±1.6 pg. We determined the direct cost, including depreciation, of the manual and instrumental methods. Cost/test varied from
$1.61 for manual method to $6.03 for the Sysmex R-3000. Cost/test for
flow cytometry and Miles H*3 were $3.34 and $3.49, respectively. (Key
words: Anemia; Blood cell count; Erythrocyte count; Erythrocyte indices; Erythrocytes; Erythrocyte volume; Hematologic tests; Reticulocyte volume distribution width; Reticulocyte; Reticulocyte counting)
Am J Clin Pathol 1994;102:623-632.
laboratory to have the option of performing CBC and reticulocyte counting on the same instrument.
The new H*3 (Miles, Diagnostics Division, Tarrytown, NY)
blood analyzer is a further refinement of the established Technicon H* 1 /H*2 technology9"12 with the addition of reticulocyte
counting capability. This is based on a new reticulocyte staining method, which utilizes light absorptive properties of the
nucleic acid dye Oxazine 750. This instrument is the first to our
knowledge which combines the capabilities of a routine CBC
and 5 part differential blood analyzer with those of a reticulocyte analyzer. Reticulocyte cellular indices such as mean cell
volume (MCVr), cell hemoglobin concentration mean
(CHCMr), cell hemoglobin content (CHr), and their respective
distribution widths (RDWr, HDWr, and CHDWr) can also be
measured. Reticulocyte indices change considerably in normal
subjects receiving recombinant human erythropoietin subcutaneously. The decrease in the CHr induced by subcutaneous
recombinant human erythropoietin therapy has provided evireplete normal
From the Departments of Laboratory Medicine and Pathology, The dence for iron-deficient erythropoiesis in iron
13
Children's Hospital; the Clinical Laboratories, Brigham and Women's subject, despite oral iron supplementation.
Hospital; and Harvard Medical School, Boston, Massachusetts.
We tested the Miles H*3 blood analyzer operating capabili-
Automated reticulocyte counting has become an essential component of the hematology laboratory. Automated reticulocyte
counting can be carried out by either a dedicated instrument,
(eg, Sysmex R-3000; Baxter Diagnostics', McGraw Park, IL)
or by a flow cytometer (eg, thiazole orange method2"7). In the
latter case, reticulocyte counting is performed in a batch mode
that is scheduled among other tests such as cell surface markers
for lymphocytes or other cells, platelet antibody testing or cell
cycle analysis. These automatic techniques have lead to significant savings in labor costs and significant improvement in the
accuracy and precision of the reticulocyte enumeration compared with manual counting. 8 A dedicated reticulocyte analyzer or a dedicated flow cytometer adds a significant expense
to the laboratory capital budget and also have substantial maintenance costs. Thus, it would be desirable for a hematology
ties for reticulocyte counting (carryover, precision, linearity,
stability, and time dependence of the staining reaction) and
compared the values obtained with this instrument with those
obtained by flow cytometry, Sysmex R-3000 automated reticulocyte analyzer and manual counting in a large number of pa-
Manuscript received July 20, 1993; revision accepted November 15,
1993.
Address reprint requests to Dr. Brugnara: Department of Laboratory
Medicine, The Children's Hospital, 300 Longwood Avenue, Bader
760, Boston MA 02115.
623
HEMATOPATHOLOGY
624
Original Article
TABLE 1. WITHIN RUN PRECISION FOR RETICULOCYTE
COUNTING, STAINING INTENSITY,
AND RETICULOCYTE INDICES
£
50 -i
4-)
Normal Control
Reticulocylosis
Mean ± SD CV (%) Mean ± SD CV (%)
% Reticulocytes
Staining intensity (%)
Low
Medium
High
Reticulocyte indices
MCVr
RDWr
CHCMr
HDWr
CHr
CHDWr
1.55 ± 0.15
9.7
8.8 ± 0.59
6.7
86.2 ± 1.8
11.8 ± 1.5
1.94 ±0.69
2.0
12.7
35.7
55.2 ± 1.71
24.9 ± 0.78
19.9 ± 1.7
3.1
3.2
8.6
C
3
O 40
O n
oo>*30
O 20
4->
91.8
18.0
29.5
3.5
26.2
4.1
± 1.5
± 1.7
±0.35
±0.27
±0.26
±0.36
1.6
9.2
1.2
7.6
1
8.6
126.9 ± 1.23
18.4 ±0.5
25.1 +0.37
4.8 ±0.19
30.5 ± 0.25
4.2 ±0.10
0.9
2.8
1.5
3.9
0.8
2.5
0)
Tjl"
™ 0
^5
0
10
20
Expected
Values are lhe mean ± SD of 21 determinations each for one normal control and one
patient with reliculocytosis. Similar results were obtained in another normal subject and in
one additional patient with reticulocytosis. CV = coefficient of variation; MCVr = reticulocyte mean cell volume: RDWr = reticulocyte distribution widths; CHCMr = reticulocyte cell
hemoglobin concentration mean; HDWr = reticulocyte hemoglobin distribution widths;
CHr = reticulocyte cell hemoglobin content; CHDWr = reticulocyte cell hemoglobin content
distribution widths.
30
40
50
reticulocyte count (%)
tients. Since the inclusion of reticulocyte testing by instrumental methods has potentially significant operational consequences, we determined and compared the true direct costs,
including depreciation, of the several methods evaluated in this
study.
MATERIALS A N D METHODS
Reticulocyte
Analyzer
analysis
with the Miles H*3
Hematology
Reticulocyte indices were measured on a Miles H*3 RTX
Hematology Analyzer. Whole blood, collected in potassium
ethylenediaminetetraacetic acid (EDTA), was diluted 1:1000
with Reticulocyte Reagent T03-3392-50 (Miles Diagnostics).
The reagent spheres the red cells using the zwitterionic detergent N-tetradecyl-N,N-dimethyl-3-ammonio-1 -propansulfonate and selectively stains the reticulocytes using the dye Oxazine 750. The stained reticulocytes were discriminated from
mature red blood cells by their increased absorption of light.
The prepared samples were analyzed on the Miles H*3 and the
reticulocyte indices were measured as described previously for
red blood cells.9"12 The software used was version V1.01 for the
.o
o
T
20
40
60
80
Expected reticulocyte count (%)
5n
£
4-
V)
3-
+->
>>
o
o
FIG. 1. Linearity of reticulocyte counting with the Miles H*3 blood
analyzer. Top, a reticulocyte-rich fraction was added in different ratios
to a reticulocyte-poor fraction of blood from the same patient. Middle,
blood from a patient with pyruvate kinase deficiency (95.3% reticulocytes by manual counting) was added in various ratios to an ABO-compatible blood sample (.85% reticulocytes) diluted with saline to a similar number of red cells per microliter. Bottom, the constancy of
percentage of reticulocytes at varying hematocrits was examined in
blood from a normal subject.
A.J.C.P. • November 1994
2-
—i
r
1y - 1.34 + 0.003x
20
40
r - 0.890
60
Hct (%)
80
BRUGNARA ET AL.
625
Miles H*3 and Reticulocytes
TABLE 2. CARRYOVER STUDIES FOR RETICULOCYTE
COUNTING ON THE MILES H»3 BLOOD ANALYZER
% Reticulocytes
Method A
(1)
Method B
(0
(2)
% Carryover
n
Range
Mean
Range
Mean tSD
12
1.3-1.8
1.49
0.09-0.37
0.218 ± 0.083
10
10
12.1-16.1
95
0-0.068
0.061-0.2
0.019 ± 0.025
0.126 ± 0.014
14.3
—
Method A = blood from norma] control was prepared with 77.1% hematocrit. Cell-free
sample was prepared from spun plasma from the same patient. Method B (1) = blood from a
patient with autoimmune hemolytic anemia (hematocrit: 26%, 14.3% reticulocytes). Method
B(2) = blood from a patient with pyruvate kinase deficiency (hematocrit: 21%. 95% reticulocytes). In Method A. one aspiration of sample was followed by three aspirations of cell-free
plasma for a total of 12 cycles. In Method B, each sample was run in triplicate followed by
three cycles of diluent, for a total often cycles.
seven tubes was then run three times with the Miles H*3 analyzer in the reticulocyte mode. Mean values and coefficients of
variation (CV) were calculated based on 21 determinations for
each subject.
Imprecision due to sample preparation. Studies were carried
out in one normal subjects and in one patient with reticulocytosis. Five separate vials were prepared for each subject, each
containing 3 /xL of blood added to 3 mL of reticulocyte reagent.
After incubation at room temperature (15-90 minutes), each
of the 5 vials was run in triplicate for reticulocyte counting. The
data were analyzed with nested analysis of variance14 using the
SAS (SAS Institute Inc., Cary, NC) software package. The total
standard deviation for each run was broken down into its
within preparation and between preparation error components. A F test for a significant preparation effect was performed and the P value calculated. P values greater than .05
5-
Miles H*3 RTX system. The Miles H*3 includesflowcytometric analysis of cells, with laser light scattering to quantify cell
volume, hemoglobin concentration, and the light absorbance
of cells stained with Oxazine 750 to detect reticulocytes and
distinguish them from mature cells. A total of 20,000 cells are
counted for each sample. Light absorbance is measured over
100 discrete channels. The gate for reticulocytes is established
based on the distribution of the negative cells. To establish the
three levels of staining intensity, the number of channels between the threshold for reticulocytes and channel 80 (d) is divided by 3. Low-staining- and medium-staining-intensity reticulocytes comprise all the positive cells between threshold,
threshold + l/id, and threshold + 2/id, respectively. High-staining-intensity reticulocytes comprise all the positive cells between threshold +2/id and channel 100. The amount of light
absorbed by the reticulocyte is directly proportional to its RNA
content.
For every red cell assessed, the cell volume, hemoglobin concentration, and presence or absence of residual mRNA can be
determined. After the volume and hemoglobin concentration
of individual mature red blood cells and reticulocytes is measured, the hemoglobin content of individual cells is calculated
as volume X hemoglobin concentration. Histograms are generated for each of the measured indices and distribution widths
are calculated for each one of the three histograms of reticulocyte indices: RDWr (for MCVr), HDWR (for CHCMr), and
CHDWr for CHr.
Two cytograms are generated, one plotting cell hemoglobin
concentration (scatter high) versus staining intensity (absorbance), the other cell volume (scatter low) versus cell hemoglobin concentration. Mature red cells are displayed in red and
reticulocytes in blue.
4-
I
3-
*
* *
T
2"
,,
" ^ — < l
*
II
T
*
*
I
,1
II
,,
I -
0
i
1
>
i
1
1
•
1
•
1
•
1
•
r
30
Time (min)
Precision
Within-run precision. Studies were carried out in 2 normal
subjects and in two patients with reticulocytosis. For each subject, 5 separate vials were prepared, each containing 3 mL of
reticulocyte reagent and 3 /*L aliquot of blood collected in potassium EDTA. After incubation at room temperature (15-90
minutes), the five vials were pooled in a 25 mL Erlenmeyer
flask, and aliquoted into seven glass tubes, each containing
approximately 2 mL of the reaction medium. Each of these
Time (min)
FIG. 2. Stability over time of the percentage of reticulocytes and staining intensity after addition of blood to the reticulocyte reagent. Top,
mean ± SD for reticulocyte counts in 15 subjects is plotted as a function of incubation time in the reticulocyte reagent. Bottom, the percentage of reticulocytes with low, medium, and high staining intensity is
plotted as a function of incubation time in the reticulocyte reagent.
•Statistically significant (P < .05) difference (paired /-test) compared
with the 15-minute values.
Vol. 102 • No. 5
626
HEMATOPATHOLOGY
Original Article
f^+H+1
15
30 45
60
75
Time
15
90 105 120
30
45
60
75
90 105 120
Time (min)
(min)
3
a
x 2
:-T-H+++i
plasma. The hematocrit of this pool was 5%. The cells collected
from the bottom of the conical tube were resuspended in the
cell free plasma at the same hematocrit of the reticulocyte pool.
The reticulocyte pool was then serially diluted with the bottom
pool at 20% intervals. Method B: Blood from a splenectomized
patient with red cell pyruvate kinase deficiency (95.3% reticulocytes, determined with manual counting) was combined in different ratios with ABO compatible blood from a normal control previously diluted with physiologic saline to a similar
number of cells/VL.
Absolute reticulocytes. Blood (150 mL in EDTA) was collected from a normal donor, and spun down for 25 minutes at 4
°C at 750 g. Plasma was removed and spun again for 10 minutes at 750g. This supernatant was used as 0% pool. Buffy coat
was removed and the packed cells were pooled (100% pool,
hematocrit 77.1%). Mixtures of the 0% and 100% pools were
prepared to obtain intermediate values (80, 60, 40, and 20%).
Carryover
15
15
30 45
30
60
75
90 105 120
Time
(min)
45
75
60
90 105 120
Time (min)
15
30
45
60
75
90 105 120
Time (min)
30
45
60
75
105 120
90
Time (min)
FIG. 3. Time course of changes in reticulocyte cellular indices and their
relative distribution widths after addition of blood to the reticulocyte
reagent. Top, mean reticulocyte cell volume and volume distribution
widths are plotted versus incubation time. Middle, reticulocyte cell
hemoglobin concentration and distribution widths are plotted versus
incubation time. Bottom, reticulocyte hemoglobin content and distribution widths are plotted versus incubation time. *Statistically significant (P < .05) difference (paired Mest) compared with the 15-minute
values. Values are the mean ± SD for 15 subjects.
indicate that the preparation error is not statistically significant.
Method A. The 100% pool used for determination of linearity for absolute reticulocyte measurements (see above, hematocrit 77.1 %) was also used for carryover studies. After the appropriate dilution with the reticulocyte reagent, the specimen was
run and the plasma used for the 0% pool was measured consecutively in triplicate. This was repeated 12 times. Carryover was
determined as (D1-D3/S-D3) X 100, where S is the number of
cells counted in the 100% pool and D are then numbers of cells
counted in the respective runs of 0% pool.1
Method B. Bloods from one patient with a moderately increase reticulocyte count (14.3%) and one patient with pyruvate kinase deficiency (95.3% reticulocytes) were used for this
study. For each patient, 10 vials were prepared, each containing 3 ML of blood and 3 mL of reticulocyte reagent. Each vial
was run in triplicate followed by three runs of diluent, and the
carryover calculated as detailed in ref. 1, based on a total of 10
cycles.
Reticulocyte
Percentage of reticulocytes. Method A: Thirty-five 7-mL potassium EDTA tubes were collected from a single patient undergoing therapeutic phlebotomy for idiopathic hemochromatosis. Blood was pooled into a 500 mL bag and filtered with a
Fenwall filter to remove leukocytes. Blood was then centrifuged at 2,000 g for one hour at 4 °C using 15 mL conical tubes.
The cell free plasma was collected and reserved for later use.
The upper 2-3 mm layer of red cells was collected from each
tube, as well as one fraction of cells from the bottom part of one
conical tube. The collected upper layer was recentrifuged using
Wintrobe tubes for 30 minutes at 2,000g at 4 °C. The uppermost layer of pale red cells was removed from each Wintrobe
tube and diluted to a final volume of .5 mL with the cell free
Stability
Blood from 30 donors was drawn into EDTA tubes and analyzed in duplicate for percentage of reticulocytes on the day of
collection and after 72 hours storage at 4 °C. This time was
chosen to replicate working conditions in the Clinical Laboratories, with specimens being drawn on Friday and run the following Monday.
Reticulocyte
Linearity
Sample
Prepared
Sample
Stability
The protocol was designed to evaluate the stability of the
prepared sample over a period of 15-90 minutes from the time
of preparation. Blood from 15 different donors was added to
the reticulocyte reagent and assayed at 15-minute intervals
over a period of two hours.
Agreement
Between
Methods
Values for reticulocyte counts obtained with the Miles H*3
system were compared with those obtained with the following
methods:
Manual counting of reticulocytes. The NCCLS new methylene blue method was used and reticulocyte were counted with
a Miller disk by two different technologists.15 Average values of
A.J.C.P. • November 1994
BRUGNARA ET AL.
Miles H*3 and Reticulocytes
627
TABLE 3. COMPARISON OF FOUR METHODS FOR RETICULOCYTE COUNTING: INTRACLASS CORRELATION (r,)
COEFFICIENTS, PEARSON PRODUCT-MOMENT (r) VALUES, REGRESSION INTERCEPTS, AND SLOPES
Intraclass Correlation
95% ConfidenceLimits
H*3/SYS
H*3/FLOW
MAN/SYS
MAN/FLOW
MAN/H*3
SYS/FLOW
Pearson s Product-Moment Ct jrrelation
n
rt
Lower
Upper
r
Intercept
Slope
175
158
184
176
158
175
0.952
0.922
0.538
0.755
0.610
0.938
0.938
0.883
0.447
0.487
0.533
0.922
0.973
0.957
0.620
0.861
0.677
0.950
0.974
0.938
0.980
0.940
0.940
0.954
-0.326
1.082
0.211
1.440
0.665
1.195
1.014
1.170
1.050
1.307
0.977
1.242
H*3 = Miles H*3; SYS => Sysmex R-3000; FLOW == flow cytometry MAN •= manual counting.
these two counts were used for comparison with the other
methods.
Flow cytometry with ihiazole orange. The method used was
based on that described by Lee and Chiu.2 A FACS/SCAN
(Becton Dickinson) flow cytometer was used.
Sysmex R-3000. Reticulocyte were counted on blood samples collected in EDTA with a Sysmex R-3000' in the manual
mode, following manufacturer's specifications.
Studies in Pediatric Normal Controls and Hematologic
Patients
Reticulocyte count and red cell and reticulocyte indices were
measured in 110 hematologic normal controls (51 males and
59 females, age 1-10 y), using "leftover" samples from the
blood collected for routine CBC measurement. All the subjects
studied were outpatients with normal hematological parameters, according to our established, age-adjusted, normal range.
Various pediatric and adult patients with hematologic and
non-hematologic disorders were also studied.
Data Analysis
Group data are expressed as mean ± standard deviation
(SD). Paired Student's Mests were used to compare data within
groups and unpaired Student's Mests to compare data between
groups. Correlation studies used standard linear regression analysis with the Pearson's product-moment correlation coefficient.
Comparison of reticulocyte counts obtained with different
methods was carried out with the intraclass correlation coefficient using SAS.16"18 To determine if two methods can be used
interchangeably, the Pearson's product-moment correlation (r)
is customarily used. However, r is just an estimation of trend
rather than concordance.17 The intraclass correlation (r,)
quantitates the extent of agreement between methods. Meaningful agreement is achieved if the lower limit of the 95% confidence interval of the intraclass correlation is at least .75.17 The
pattern of agreement between any methods can also be visually
analyzed by plotting the mean reading of the two methods for
each subject against the difference in reading between the two
methods.16
RESULTS
Within Run Precision
Table 1 presents data on the intraassay precision for one
subject with normal reticulocyte counts and one with an increased reticulocyte counts. Precision was estimated from the
CV of 21 separate measurements performed on the same blood
as detailed in Methods. The CV was 9.7% for the normal control and 6.7% for the sample with high reticulocyte counts. We
also examined the precision of the values obtained for staining
intensity of reticulocytes. As expected, in normal blood, there
is a relatively larger variability in estimating the percentage of
high intensity staining reticulocytes, due to the small number
of cells present in normal conditions. When the reticulocyte
count increases, the CV for the medium and high intensity
staining reticulocytes decreases significantly. The CVs for the
three classes of staining intensity in samples with increased
reticulocyte content ranged from 3.1% to 8.6% (Table 1).
Excellent precision was found in the measurements of the
reticulocyte cellular indices in both the normal control and the
patient with reticulocytosis (Table 1). The CVs for measurements of MCVr, CHCMr, and CHr were .8% to 1.6%; CVs for
the distribution widths of these indices also were excellent
(range, 2.5% to 9.2%).
Imprecision Due to Sample Preparation
Interassay precision was evaluated by comparing reticulocyte counts from the same specimen run asfivetriplicate specimens. Nested analysis of variance was used to identify the two
components of the total standard deviation for each run,
namely the within preparation and between preparation standard deviations. An F test for a significant preparation effect
was performed. P values greater than .05 indicate a statistically
not significant preparation error. For a normal control blood
(% reticulocytes mean: 1.70, n = 15) the total SD was .18, the
within preparation SD. 16 and the between preparation SD. 14,
with P = .3261 (difference not significant). For a patient with
reticulocytosis (% reticulocytes mean: 8.6, n = 15) the total SD
was .71, the within preparation SD and the between preparation SD were .71 and .00, respectively with P = .8408 (difference not significant). Thus, no statistically significant preparation error could be identified.
Vol. 102 • N o . 5
628
HEMATOPATHOLOGY
Original Article
Linearity
Linearity was established by obtaining a reticulocyte-rich
fraction from a subject with moderately increased reticulocyte
counts and diluting it with a fraction with low reticulocyte
content from the same subject. As shown in Figure 1, excellent
linearity was obtained in the range tested (1-49%). Linearity
was also determined by mixing blood of a patient with pyruvate
kinase deficiency (reticulocytes = 95.3%) with an ABO compatible normal control, diluted to a similar number of cells/^L. As
shown in Figure 1B, linearity was established in the range of 1%
to 73% reticulocytes. Undiluted blood of this patient could not
be counted with any of the automatic instruments.
The effect of varying hematocrit on the percentage of reticulocytes was studied by counting reticulocytes in blood from a
normal subject for hematocrit values of 14.9% to 77.1%. As
shown in Figure 1C, there was no significant effect of changing
TABLE 4. RETICULOCYTE AND RED CELL INDICES
OBTAINED IN A PEDIATRIC POPULATION WITH
THE MILES H«3 HEMATOLOGY ANALYZER
Age (years)
% Reticulocytes
Red cell indices
MCV (fL)
CHCM (g/dL)
CH (pg)
RDW (%)
HDW (g/dL)
Reticulocyte
MCVr (fL)
CHCMr (g/dL)
CHr(pg)
RDWr (%)
HDWr (g/dL)
CHDWr (pg)
Mean ± SD
Range
4.5 ± 2.6
1.5 ± 0 . 6
1-10
0.4-4.0
79.3
34.9
27.7
14.0
2.6
± 3.7
± 1.0
± 1.4
±0.9
± 0.2
72.0-91.2
31.9-36.8
24.4-31.6
12.4-17.8
2.1-3.1
97.6
28.2
26.7
14.3
3.3
3.3
± 4.7
± 1.4
± 1.6
±2.0
±0.7
± 0.4
87.7-116.5
23.3-31.4
21.4-30.5
10.9-22.0
2.5-7.5
2.6-4.8
Data were collected from 110 normal controls (51 males. 59 females, age 1 -10 years), with
normal CBC values as defined by age-adjusted normal range values. MCV = mean cell
volume: RDW = red cell distribution widths; CHCM = cell hemoglobin concentration mean;
HDW = hemoglobin distribution widths; CH = cell hemoglobin content; CHDW = cell
hemoglobin content distribution widths.
H*3 - SYSMEX
this variable on the values obtained for the percentage of reticulocytes.
10
20
30
10
Mean value (%)
H*3
(%
20
30
reticulocytes)
Carryover
FLOW
0
10
0
20
10
20
30
H*3 (% reticulocytes)
Mean value (%)
Results of carryover studies are presented in Table 2. Two
different methods were used to evaluate carryover. In the first
one, a suspension of cells with 77.1% hematocrit was used to
evaluate carryover into a cell free plasma from the same subject. This process was repeated 12 times.
In the second method, blood from a patient with moderately
increased (14.3%) and markedly increased (95.3%) reticulocyte
counts were run in triplicate followed by three runs of diluent
(as described in ref. 1) for a total often cycles each. In all three
cases, carryover was minimal, with average values of .218, .019
and .126%, respectively.
MAN - H*3
o
o go
Reticulocyte
°
Sample
Stability
a 0
5
%3°o
o
°
0
o
0
10
0
o
20
Mean value (%)
30
0
10
Manual (%
20
30
reticulocytes)
FIG. 4. Left column, relation between mean values and difference in
values of reticulocyte counting with the Miles H*3 and the other three
methods. Right column, scatter plots and regression equation of reticulocyte counting with the Miles H*3 and the other three methods. Top
row, comparison between Miles H*3 and Sysmex R-3000. Middle row,
comparison between Miles H*3 and flow cytometry. Bottom row, comparison between Miles H*3 and manual reticulocyte counting. H*3 =
Miles H*3; SYSMEX = Sysmex R-3000; FLOW = flow cytometry;
MAN = manual counting.
To evaluate stability of the reticulocyte measurements over
time, 30 blood samples were measured at day 0 and again after
72 hours of storage at 4 °C. Data were then compared with the
paired Mest. There were no statistically significant differences
after 72 hours of cold storage for percentage of reticulocytes,
and MCVr. The CHCMr index was significantly decreased
after storage (average decrease of 1.2 ± 1.2 g/dL; P < .01) and
this change was accompanied by a significant increase in
RDWr (average increase 2.0 ± 4.6%; P < .05) and in HDWr
(average increase .5 ± . 5 g/dL; P<. 01). There was also a significant decrease in CHr (average decrease 1.8 ± 4.5 pg; P < .05)
with no significant changes in CHDWr. Stability of reticulocyte indices at room temperature or at different times of storage was non evaluated.
A.J.C.P. • November 1994
BRUGNARA ET AL.
629
Miles H*3 and Reticulocytes
A. NORMAL CONTROL
RETICULOCYTE INDICES
MCVr
CHCMr
RDWr
HDWr
CHr
CHDWr
112.2
27.8
16.9
3.68
30.2
3.6
fL
g/dL
%
g/dL
pg
pg
B. IRON-DEFICIENT ANEMIA
FIG. 5. Reticulocyte analysis
with the Miles H*3 analyzer.
Left column, plot of staining
intensity (x axis) versus red
scatter high (5°-15°; y axis),
which is proportional to cell
hemoglobin concentration.
Right column, plots of hemoglobin concentration (x axis)
versus cell volume (y axis).
Red dots = erythrocytes; blue
dots = reticulocytes; white
dots outside of the red cell
area = platelets, mononuclear
cells, or scattered-light coincident events. A, Normal control, 1.5%. B, Iron deficiency,
reticulocytes 1.6%. C, Sickle
cell anemia, reticulocytes
18.1%. D, Hereditary spherocytosis, reticulocytes 11.7%.
RETICULOCYTE INDICES
MCVr
CHCMr
RDWr
HDWr
CHr
CHDWr
84.7
20.8
21.3
3.70
16.9
3.6
fL
g/dL
%
g/dL
pg
pg
C. SICKLE CELL ANEMIA
RETICULOCYTE INDICES
MCVr
CHCMr
RDWr
HDWr
CHr
CHDWr
106.2
29.0
23.3
6.06
29.2
4.5
fL
g/dL
%
g/dL
pg
pg
D. HEREDITARY SPHEROCYTOSIS
RETICULOCYTE INDICES
MCVr
CHCMr
RDWr
HDWr
CHr
CHDWr
Reticulocyte
Prepared Sample
100.6
27.7
20.0
4.74
26.7
4.4
fL
g/dL
%
g/dL
pg
pg
Stability
Reticulocyte counts and indices were measured in 15 subjects at 15-minute intervals for 120 minutes after the addition
of blood to the reticulocyte reagent. Changes in the variables
measured were plotted against time and compared with the
values obtained at 15 minutes. As shown in Figure 2, values for
percentage of reticulocytes were essentially stable over two
hours. With paired /-test analysis, in comparison with the value
obtained at 15 minutes, a statistically significant decrease in the
percentage of reticulocytes was observed after 45 minutes.
Staining intensity showed a small but significant increase in the
percentage of reticulocyte in the low intensity fraction starting
at 90 minutes. This was associated with a significant decrease
in the percentage of reticulocytes in the medium and high intensity staining fractions at the same time points.
Study of the reticulocyte cell indices revealed important
changes during the 2 hours of incubation with the reticulocyte
reagent. Paired /-test analysis showed a significant increase of
MCVr beginning at 60 minutes, which was associated with a
significant decrease in the CHCMr values (Fig. 3). This indicates that the reticulocytes swell over time in the reticulocyte
reagent. A significant increase was also observed in the volume
of red cells between 15 and 120 minutes of incubation (P <
Vol. 102 • No. 5
630
HEMATOPATHOLOGY
Original Article
Studies in Pediatric Normal Controls and Hematologic
Patients
Sysmex
Manual FACS/SCAN R-3000 MilesH*3 Values for red cell and reticulocyte indices obtained with the
Miles H*3 in a normal pediatric populations are presented in
Direct labor
Table 4. Included in these group were 110 subjects, age 1-10,
Specimen preparation
0.18
0.13
—
0.13
with normal hematologic parameters (hematocrit, hemogloTesting procedure
1.23
0.36
0.39
0.22
bin, red cell indices) based on age-adjusted normal range valConsumables
ues. We found no significant differences in the measured paReagents
0.02
1.20
0.89
2.00
rameters between males and female. The data indicate that,
Other
0.18
0.05
—
0.05
compared with mature red cells, reticulocytes have a larger
Calibrators, quality control —
0.75
0.47
0.49
Proficiency testing
—
0.02
0.02
0.02
volume, a reduced cell hemoglobin concentration and a similar
Depreciation
—
0.53
2.88
0.58
hemoglobin content.
Service contracts
—
0.30
1.38
—
Figure 5 presents data on reticulocyte analysis of a normal
control and 3 patients with iron deficiency, sickle cell anemia
Total cost/test
1.61
3.34
6.03
3.49
and hereditary spherocytosis, respectively. As shown in Figure
Labor cost is based on time studies of a batch of 10 samples for reticulocyte analysis (3
5, staining intensity of reticulocytes can be plotted versus cell
minutes 45 seconds for preparation and 6 minutes 10 seconds run time for Miles H*3, 11
hemoglobin concentration and this variable can also be plotted
minutes 7 seconds for Sysmex R-3000) and a batch of 12 samples for FACS/SCAN (4 minutes
against cell volume. These plots show the presence of hy35 seconds for preparation and 12 minutes 10 seconds run lime) at the hourly rate of $21
pochromic reticulocytes in iron deficient anemia, and a
inclusive of fringe benefits. Manual reticulocyte counting: 30 seconds for preparation and 3.5
minutes for reading (Miller's disk technique). If CAP workload units (9 minutes) were used,
marked heterogeneity of reticulocytes volume and hemoglobin
$3.15 can be attributed as labor cost for manual counting. List prices were used for reagent
concentration in sickle cell anemia and hereditary spherocytocosts. Data for Sysmex R-3000 are based on yearly usage in our laboratory (manufacturer's
sis, with the presence of dehydrated reticulocytes.
total is $0.55/test). Quality control expenses: FACS/SCAN. yearly usage of 12 kit of ReticCheck (Streck Lab. Inc; $85/kit) and 30 control runs/week; Sysmex R-3000, 9 control kits/
We found no interferences with the Miles H*3 method for
year (Baxter B3I60/90/AP. $268.80/kit) and 24 control runs/week; Miles H*3, 12 kit/year
reticulocyte counting in patients with malaria infection (2 pa($85/kit) and 24 control runs/week; CAP survey, $ 123/year. Instrument list price: FACS/
tients tested, on with P. Falciparum and the other with P. FalciSCAN, $99,500; Sysmex R-3000, $109,000; Miles H*3, $212,000; Miles H , 2, $189,000; Cost
parum plus P. Malariae infection), high WBC counts, high
of interface with LIS ($6,000) was added to the capital cost of each instrument. Depreciation
is calculated as a straight line in 5 years. Service contracts are for 7 days/week coverage:
platelets counts, alpha or beta thalassemia, sickle cell anemia,
FACS/SCAN $ 12,000; Sysmex R-3000 $ 11,000; Miles H*3 or H*2. $ 13,990. Incremental cost
renal failure, bone marrow transplant and neonatal blood. The
attributable to reticulocyte counting for service contract and depreciation was based for
H*3 Miles could not gate on the negative, mature cells in two
FACS/SCAN on the current utilization of 5 of 40 working hours/week (1/8 of total cost) and
on an annual test volume of 5,000. For Miles H*3, the incremental cost was based on the price
transfused patients, one with Hb H/CS (Constance Spring) and
difference compared with Miles H*2 ($23,000). An alternative analysis for Miles H*3 based
one with severe Hb H disease. Presence of nucleated red cells
on the total instrument cost and annual billable volume of 90,000 CBCs and 8,000 reticulo(one patient, WBC 232,850/>L, corrected WBC 18,310/>L)
cyte counts, yielded a cost/reticulocyte test of $0.44 for depreciation and $0.14 for service
was associated with overestimation of reticulocyte counts
contracts.
(2.5% manual, 19.6% Miles H*3, 12.7% Sysmex R-3000).
TABLE 5. COST ANALYSIS OF RETICULOCYTE COUNTING
.001; paired Mest). A comparison of MCVr/red cell volume
ratios at 15 and 120 minutes still showed a statistically significant increase (P < .001, paired Mest), suggesting that the swelling was proportionally greater in reticulocytes than mature
cells. These changes did not produce any significant variation
in CHr. Thus, CHr values can be compared independently of
the reaction time used. The CHr index has been shown to be
useful in detecting iron-deficient erythropoiesis.13
Agreement Between
Methods
Comparison of reticulocyte counts obtained with the Miles
H*3, Sysmex R-3000, flow cytometry and manual counting is
presented in Table 3 and Figure 4. Pearson's product moment
(r) values greater than .93 were obtained for all comparisons,
indicating that the different methods had similar trends (Table
3). However, intraclass correlation studies yielded unsatisfactory results when manual counting was compared with the
three automated methods, since in all three cases the lower 95%
confidence limit for r{ was lower than .75 (Table 3). Good intraclass correlation coefficients (r; for the lower 95% confidence
limit > .75) were obtained for the comparisons among the
three automated methods, indicating that there is satisfactory
concordance among the three instruments.
DISCUSSION
Our studies of the operating characteristics of the Miles H*3
analyzer indicate that reticulocyte counting with this instrument has good precision, excellent linearity in the range tested
(1-72%), and insignificant carryover. Reticulocyte counts obtained with the Miles H*3 instrument are stable when blood is
stored for 72 hours at 4 °C. Comparison of four different methods for reticulocyte counting was carried out with two different
methods of statistical analysis. With the classic Pearson's product-moment correlation r, there was relatively good agreement
between any of the two methods (Table 3 and Fig. 4). However,
this kind of analysis only indicates a linear trend for both variables to change in the same direction," and does not permit
determination of agreement between any of the two methods.
With the intraclass correlation coefficient, the level of agreement can be estimated; a satisfactory level is achieved when the
lower limit of the 95% confidence interval is at least .75." Using these criteria, manual counting of reticulocytes can not be
considered interchangeable with the three automatic methods.
This reflects the significant imprecision of manual reticulocyte
counting, mostly based on the interobserver variation in defining a reticulocyte and on the small number of cells counted.8
Excellent level of agreement was achieved with any combination of the three automated methods for reticulocyte counting.
As shown in Figure 4, there was a tendency for the flow cytometry method to overestimate counts compared with the H*3, but
A.J.CP.-November 1994
BRUGNARA ET AL.
Miles H*3 a,
the overall agreement was well above the acceptable values.
Some of the differences might be due to the fact that the FACS/
SCAN method includes "shift" reticulocytes which are not
captured by either the Sysmex R-3000 or the Miles H*3. Autofluorescence and interference by WBC and platelets are a significant interference for the flow cytometry method, but not for
the Miles H*3 method. We found no significant interfering
factors in the reticulocyte counting with the Miles H*3 analyzer, with the exception of nucleated red cells. In particular,
no significant interference was observed in the presence of high
WBC counts or platelets counts.
Use of red cell indices has become an essential part of the
differential diagnosis of anemias.' 1,19'2° The Miles H*3 hematology analyzer has the capability of measuring the values of
MCVr, CHCMr, and CHr, and the distribution widths RDWr,
HDWr, and CHDWr. One interesting potential clinical application of the reticulocyte indices is detection of iron-deficient
erythropoiesis induced by recombinant human erythropoietin13,21,22). Reticulocyte indices and CHr in particular could
allow a "real time" evaluation of the appropriateness of the
recombinant human erythropoietin-induced erythropoiesis
and provide the bases for a more rationale and cost-effective
usage of this expensive recombinant red cell growth factor. In
our evaluation, we observed that the reticulocyte indices have
excellent precision and reproducibility (Table 1). Whereas
MCVr and CHCMr have a tendency to change over time when
the cells are incubated in the reticulocyte reagent (Fig. 3), CHr
remains stable for 2 hours in the reagent buffer. Thus, time of
incubation in the reticulocyte reagent is not a critical variable
for this reticulocyte parameter.
The availability of reticulocyte cell indices with the Miles
H*3 is an important advance compared with the currently
available methods, which quantify reticulocyte staining intensity, without providing any information on volume, hemoglobin concentration and content of the cells examined. The staining intensity of reticulocytes has been used to obtain insight
into erythropoiesis in a variety of clinical conditions. 4,7 It will
be important to test whether similarly useful information can
be gathered from the study of the reticulocyte indices and CHr
in particular, in settings such as bone marrow transplantation,
differential diagnosis of anemias and sickle cell anemia. It certainly represents a powerful and unique feature of this instrument as compared with all the other reticulocyte analyzers.
The cost-effectiveness for the clinical laboratories of combining blood cell counting and reticulocyte counting in one
instrument versus two separate instruments deserves careful
consideration. An analysis based exclusively on capital cost
indicates that the purchase of the Miles H*3 versus the purchase of an H*2 plus a Sysmex R-3000 is financially very attractive. Capital cost would be $218,000 versus $310,000 (including $6,000 of cost for each interface with LIS), with an
annual depreciation expense (5 year, straight line) of $43,600
versus $62,000. Other purchasing options such as leasing or
reagent rentals were not considered. Expenses for service contracts are also significant lower in the Miles H*3 ($13,990)
compared with the H*2 plus Sysmex R-3000 ($24,990).
We analyzed the cost of reticulocyte counting with a manual
technique and with three automated reticulocyte analyzers.
This analysis is based on our experience at Children's Hospital
and Brigham and Women's Hospital (Table 5). In this analysis,
we did not consider discounts that might be obtained in purchasing instruments, service contracts, or supplies, and we itemized only the direct components of cost.22 An open tube testing
Vol. I'
631
Reticulocytes
mode for all the instruments was assumed. Time spent by technologists in starting up and shutting down instruments was not
included.
Depreciation and service contract costs were estimated for
the FACS/SCAN based on the relative utilization of the instrument for reticulocyte counting (one eighth of total time). For
the Miles H*3, two different approaches were used, with similar outcomes: the first one, based on the additional cost of an
H*3 compared with H*2 system ($23,000 for purchase price,
and $0 for service contract) yielded a cost per reticulocyte
count of $.58 for depreciation and service contracts. The second one used the purchase price and service contract cost for
an H*3 instrument, and calculated the cost per test based on a
volume of 90,000 billable CBCs and 8,000 billable reticulocyte
counts per year. This second approach yielded a cost per reticulocyte test of $.44 for depreciation and $.14 for service contracts.
Our analysis indicates that the cost for reticulocyte counting
with a dedicated analyzer (Sysmex R-3000) is significantly
higher compared to instruments such as a flow cytometer or
the Miles H*3 analyzer that perform other, additional, tests.
Changes in annual volume of reticulocyte and CBC billable
tests may greatly affect this comparison, which is based on the
actual volumes at our two institutions. The manual counting
technique is significantly cheaper that the automated methods,
due to the absence of depreciation and maintenance expenses
and to the low cost of consumables. However, the labor component of this method is significantly higher (3-4 fold) than the
other three methods. A 6-8 fold higher value than the automated methods can be obtained for the labor cost of the manual method, if CAP workload units (9 minutes) are used rather
than the value obtained in our time studies (4 minutes).
For laboratorians facing the constant challenge of containing
expenses, the availability of a combined blood and reticulocyte
analyzer adds an interesting option regarding the choice of the
appropriate hematology analyzer. The availability of reticulocyte cellular indices with the Miles H*3 and the possibility that
they may become useful indicators of the appropriateness of
erythropoiesis in a variety of clinical conditions also represents
a significant advantage compared with other automated reticulocyte analyzers.
Acknowledgments. We thank the staff of the Hematology Laboratory
at Brigham and Women's Hospital and Beth Israel Hospital for providing samples from patients. We thank Nina M. Wagner and Kristen
Controne for manual reticulocyte counting and William J. Canfield for
statistical analysis.
REFERENCES
1. Tichelli A, Gratwohl A, Driessen A, et al. Evaluation of the Sysmex R-1000®: An automated reticulocyte analyzer. Am J Clin
Pathol 1990;93:70-78.
2. Lee LG, Chiu LA. Thiazole orange: A new dye for reticulocyte
analysis. Cytometry 1986;7:508-517.
3. Davis BH, Bigelow NC. Flow cytometric reticulocyte quantification using thiazole orange provides clinicall useful reticulocyte
maturity index. Arch Pathol Lab Med 1989; 113:684-689.
4. Davis BH, Bigelow N, Ball ED, Mills L, Cornwell GG III. Utility of
flow cytometric reticulocyte quantification as a predictor of engraftment in autologous bone marrow transplantation. Am J
Hematol 1989;32:81-87.
5. Ferguson DJ, Lee SF, Gordon PA. Evaluation of reticulocyte
counts by flow cytometry in a routine laboratory. Am J Hematol
1990;33:13-17.
•No. 5
632
HEMATOPATHOLOGY
Original Article
6. Schimenti KJ, Lacerna K, Wamble A, et al. Reticulocyte quantification byflowcytometry, image analysis and manual counting.
Cytometry 1992;13:853-862.
7. Davis BH, Bigelow NC, Koepke JA, et al. Interlaboratory comparison of flow cytometric reticulocyte analysis. Blood
1992;80:280a. Abstract.
8. Savage RA, Skoog DP, Rabinovitch A. Analytic inaccuracy and
imprecision in reticulocyte counting: A preliminary report from
the College of American Pathologists reticulocyte project. Blood
Cells 1985; 11:97-112.
9. Fossat G, David M, Harle JR, et al. New parameters in erythrocyte
counting. Arch Pathol Lab Med 1987; 111:1150-1154.
10. Mohandas N, Kim YR, Tycko DH, et al. Accurate and independent measurement of volume and hemoglobin concentration of
individual red cells by laser light scattering. Blood 1986;68:506513.
11. Mohandas N, Johnson A, Wyatt J, et al. Automated quantitation
of cell density distribution and hyperdense cell fraction in RBC
disorders. Blood 1989;74:442-447.
12. Tycko DH, Metz MH, Epstein EA, Grinbaum A. A flow-cytometric light scattering measurement of red blood cell volume and
hemoglobin concentration. J Appl Opt 1985;24:1355-1365.
13. Brugnara C, Colella GM, Cremins J, et al. Effects of subcutaneous
recombinant human erythropoietin in normal subjects: Development of decreased reticulocyte hemoglobin content and
iron-deficient erythropoiesis. J Lab Clin Med 1994; 123:
660-667.
14. Schaffe H. The Analvsis of Variance. New York: John Wiley,
1959, pp. 248-258.'
15. National Committee for Clinical Laboratory Standards. Methods
for Reticulocyte Counting: Proposed Standards. Document
H16-P. Villanova, PA: National Committee for Clinical Laboratory Standards, 1985.
16. Stacy DL, Ping H. Serum ferritin measurements and the degree of
agreement using four techniques. Am J Clin Pathol
1992;98:511-515.
17. Lee J, Koh D, Ong CN. Statistical evaluation of agreement between two methods for measuring a quantitative variable. Comput Biol Med 1989; 19:61-70.
18. Shrout PE, Fleiss JL. Intraclass correlations: Uses in assessing rater
reliability. Psychol Bull 1979;86:420-428.
19. D'Onofrio G, Zini G, Ricerca BM, Mancini S, Mango G. Automated measurement of red blood cell microcytosis and hypochromia in iron deficiency and /3-thalassemia trait. Arch
Pathol Lab Med 1992; 116:84-89.
20. Bunyaratvej A, Butthep P, Fucharoen S, Saw D. Erythrocyte volume and haemoglobin concentration in haemoglobin H disease: Discrimination between the two genotypes. Acta Haematol 1992; 87:1 -5.
21. Brugnara C, Chambers LA, Malynn E, Goldberg MA, Kruskall
MS. Red cell regeneration induced by subcutaneous recombinant erythropoietin: Iron-deficient erythropoiesis in iron-replete subjects. Blood 1993;81:956-964.
22. Macdougall IC, Cavill I, Hulme B, et al. Detection of functional
iron deficiency during erythropoietin treatment: A new approach. Br Med J 1992;304:225-226.
23. Tirabassi CP, Winkelman JW. Incremental cost basing of laboratory test bid proposals. Lab Med 1988; 19:111-115.
A.J.C.P. • November 1994