Hematopathology / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

Automated Reticulocyte Counting and Immature

Reticulocyte Fraction Measurement
Comparison of ABX PENTRA 120 Retic, Sysmex R-2000, Flow
Cytometry, and Manual Counts
Francis Lacombe, MD, PhD, Laurent Lacoste, Jean-Philippe Vial, MD, Alex Briais,
Josy Reiffers, MD, Michel R. Boisseau, MD, and Philippe Bernard, MD

Key Words: Blood cell count; Reticulocyte; Reticulocyte count; Reticulocyte percentage; Immature reticulocyte fraction; Flow cytometry;
Automated hematology analyzer

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

Abstract An automated reticulocyte count is now considered as a
We evaluated reticulocyte counting and measurement necessary blood parameter. During the last 10 years, flow
of immature reticulocyte fraction (IRF) with the ABX cytometry has become the reference technique for measuring
PENTRA 120 Retic blood analyzer on 300 blood samples. absolute reticulocyte counts and parameters of reticulocyte
Reticulocyte counts were compared with those obtained maturation.1 5 Compared with manual reticulocyte counting,
by visual counting of 2,000 RBCs, by the TOA (Kobe, these automatic techniques have led to a decrease in laboratory
Japan) Sysmex R-2000 and a flow cytometry method. The labor costs,6 a substantial improvement in the accuracy and
parameters analyzed were the percentages of reticulocytes precision of reticulocyte counts,7-9 and the acquisition of new
on all analyzers and the IRF with different modalities. The parameters for measuring reticulocyte maturation.10" These
Retic Count kit (Becton Dickinson, San Jose, CA) was new parameters, which are essentially based on the measure-
used with the Coulter (Hialech, FL) XL, and a mean ment of RBC RNA content, are now termed the immature retic-
channel of fluorescence (MCF) was calculated tofitthe ulocyte fraction (IRF) and can be obtained by using several
reticulocyte maturation. Reticulocyte counting with the methods of RNA staining with fluorescent nucleic acid reagents.
ABX (Montpellier, France) PENTRA 120 Retic showed Despite the lack of agreement about IRF parameters, many
excellent precision and linearity with no significant authors consider them of great clinical significance.12"14 Flow
carryover. Reticulocyte counts were stable after blood cytometers and dedicated reticulocyte analyzers have been used,
storage for 72 hours at 4°C but not at room temperature but in a hematology laboratory, it is preferable to perform reticu-
(RT). IRF parameters values were stable for only 8 hours locyte analysis and the CBC count on the same instrument.15
at 4°C and 6 hours at RT. Comparisons of the methods We evaluated the ABX PENTRA 120 Retic (ABX, Mont-
showed good intraclass correlation (Rl)for reticulocyte pellier, France) blood analyzer for reticulocyte counting (carry-
percentages between ABX PENTRA 120 Retic and over, precision, linearity, stability) and IRF parameters. The
Sysmex R-2000, ABX PENTRA 120 Retic and flow results obtained with this analyzer were compared with those
cytometry, Sysmex R-2000 and flow cytometry, and ABX obtained with flow cytometry, Sysmex R-2000 (TOA, Kobe,
PENTRA 120 Retic and manual counting. IRF values Japan), and manual counting in 300 blood samples from
were correlated between fluorescence rates and RNA healthy individuals or patients with blood disease.
content, but in each case, low RI values were found,
showing that Sysmex and ABX IRF values were not
concordant. We obtained a significant correlation
between mean fluorescence index and the MCF measured Materials and Methods
by flow cytometry, but the 2 methods were not concordant
using the RI. The ABX PENTRA 120 Retic is a good Patients
instrument for analyzing reticulocyte count and
Reticulocyte counts and indices were measured within a
percentage and allows a good analysis of IRF with
maximum of 6 hours in 300 normal and pathologic samples
several modalities.
from the blood collected daily for routine CBC counts.

© American Society of Clinical Pathologists Am J Clin Pathol 1999;112:677-686 677

Lacombe et al / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

Pathologic samples were essentially from adult patients with performed on whole blood using auramine-0 staining of
hematologic disease (chronic and acute leukemia), and reticulocyte RNA.
normal samples were from healthy volunteer donors.
Precision
Reticulocyte Analysis With the ABX PENTRA 120 Retic We performed studies on 3 patients with low (mean,
Reticulocyte analysis was performed with the ABX 0.33%), medium (mean, 2.22%), and high (mean, 12.5%)
PENTRA 120 Retic using the 3.0 software version. On percentages of reticulocytes. For each sample, mean values
whole blood, this analyzer gives a measurement of propor- and coefficients of variation (CV) were calculated using 18
tional reticulocyte count and absolute reticulocyte count. It determinations.
also provides an evaluation of reticulocyte maturation
according to 3 classes, low RNA content (RetL), medium Linearity
RNA content (RetM), and high RNA content (RetH), and a A blood sample with a high percentage of reticulocytes
quantification of global reticulocyte RNA content with a (25.1%) was diluted in ABX Diluent (ABX) from 0.9 to
mean fluorescence index (MFI). Briefly, 0.8 uL of whole 0.1. Assuming the linearity of the RBC count, we calcu-
blood is mixed with 2.5 mL of a proprietary formulation of lated an expected value of reticulocyte count corrected
thiazole orange (under Becton Dickinson license, San Jose, from the RBC value for each dilution and compared it with
CA) and incubated at 35°C for 25 seconds. An aliquot of the actual value.

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

the dilution is transferred to the optical bench, and cells are
analyzed sequentially to determine the true volume by Carryover
resistivity and the fluorescence under a 20-mW argon ion One blood sample from a patient with a high reticulo-
laser. A maximum of 32,000 cells are analyzed, and the cyte count (Sh) (9.1%) and another from a patient with low
ABX PENTRA 120 Retic, using customized gating for reticulocyte count (SI) (0.3%) were successively run 3 times,
each sample, separates reticulocytes from mature RBCs, and the carryover was calculated as reticulocyte count of SI
WBCs, and platelets. The results are displayed on a reticu- run just after the run of Sh minus the reticulocyte count of SI
locyte matrix with RNA content on the y-axis and cell determined as the mean of 3 consecutive runs.
volume on the x-axis IFigure II.
Sensitivity
Reticulocyte Analysis With Alternative Methods Some samples from patients in postinduction treatment
The same samples were analyzed using 3 methods. of acute myeloid leukemia and, thus, presumed to have no
reticulocytes were analyzed at the nadir of chemotherapy
Visual Counting of Reticulocytes effect for reticulocyte percentages.
The National Committee for Clinical Laboratory Stan-
dards new methylene blue method was used, and reticulo- Reticulocyte Sample Stability
cytes were counted by a well-trained operator in 2,000 RBCs.5 Blood from 10 donors was analyzed for percentages and
absolute reticulocyte counts, RetL, RetM, RetH, and MFI on
Flow Cytometry With Thiazole Orange16 the day of collection and after 2, 4, 6, 8, 24,48, and 72 hours
We used the Retic Count (Becton Dickinson) kit for of storage at room temperature (RT) and 4°C.
analyzing reticulocytes according to the manufacturer's
recommendations. An XL flow cytometer (Coulter, Data Analysis
Hialeah. FL) was used after daily checking of settings. Data are expressed as mean ± SD. Paired Student /
Moreover, we calculated an index of reticulocyte matura- tests were used to compare data within groups, and
tion based on the geometric RNA fluorescence mean unpaired Student / tests were used to compare data
provided by the flow cytometer (mean channel fluores- between groups. Comparison of reticulocyte counts and
cence [MCF]). indices obtained with the different methods was performed
using the Pearson coefficient of correlation. The degree of
Sysmex R-2000 Analysis agreement between 2 different methods also could be visu-
The same samples were analyzed using a Sysmex R- alized by using a plot of the mean reading of the 2
2000 analyzer, which gave proportional absolute counts methods against the difference in reading between the 2
of reticulocytes and a quantification of reticulocyte matu- methods. To determine the degree of concordance of 2
ration using 3 classes: high fluorescence rate (HFR), methods, we used the intraclass correlation coefficient
medium fluorescence rate (MFR), and low fluores- (RI), 1819 instead of the Pearson coefficient of correlation,
cence rate (LFR).17 The reticulocyte measurement was which is just an estimation of trend. It is generally

678 AmJCIinPathol 1999;112:677-686 © American Society of Clinical Pathologists

 Hematopathology / ORIGINAL ARTICLE

80
0»
y = 0.987x + 0.009
O 70-
Z. r = 0.999
**§ 60-
0
» 50-
§.
g 4<H
3
•a 3 0 -
0)

• :.

i
V 20-
a
§ 10-
V
a. -
U'-.-~.h Ml 0 1 1 1—
1 • ••— 10 20 30 40 50 60 70 80
i ii • Obtained Reticulocyte Count (109/L)

Volume •Figure 21 Linearity of reticulocyte counting with ABX

•Figure I I Reticulocyte analysis with the ABX PENTRA 120 PENTRA 120 Retic. A blood sample with a high percentage
Retic analyzer. Scatter plot of RNA content (y-axis) vs Coulter of reticulocytes (25.1 %) was diluted in ABX Diluent (ABX)
volume (x-axis). Four regions are individualized: RBC, low from 0.9 to 0.1. The expected value of reticulocyte count

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

RNA content (RetL), medium RNA content (RetM), and high corrected from the RBC value for each dilution is plotted on
RNA content (RetH) regions. the y-axis, and the obtained value on the x-axis.

• Table II
Within-Run Precision for Reticulocyte Counting and Immature Reticulocyte Fraction Parameters (RetL, RetM, RetH, MFI)*
Reticulocyte Count

Low Mediu n High

Mean ± SD CV (%) Mean ± S D CV(%) Mean ± SD CV (%)

Reticu locytes (%) 0.33 ±0.04 12.8 2.1 ±0.2 9.4 12.5 ± 0.19 1.52
RetL 87.1 ±5.2 6.0 74.0 ±5.8 7.9 84.5 ± 1.7 2.0
RetM 8.5 ±4.5 52.5 19.0 ±3.6 18.8 12.8 ± 1.3 10.6
RetH 4.3 ±3.5 81.5 7.1 ±3.4 48.1 2.7 ± 0.6 21.1
MFI 13.1 ±2.9 22.2 22.8 ±3.6 15.9 16.9 ± 1.0 6.21

RetL = low RNA contenl; RetM = medium RNA content: Ret H = high RNA content; MFI = mean fluorescence index; CV = coefficient of variation.
" Values are the mean ± SD of 18 determinations each for I patient with low. 1 with normal, and 1 with high reticulocyte count.

assumed that a good agreement between 2 methods is cells found in these conditions. Also as expected, the CVs
achieved when the lower limit of the 95% confidence became larger with the immature reticulocyte RetH class.
interval of RI is at least 0.75.
Linearity
Expected values of reticulocyte count are plotted
against obtained values of reticulocyte count in IFigure 21.
Results An excellent correlation was observed (r = 0.999), the
equation of the regression curve showing a slope very close
Within-Run Precision to 1 and a y-axis interception at 0.009. The linearity of the
Results from 3 samples with low, medium, and high ABX PENTRA 120 Retic was considered excellent.
reticulocyte counts are shown in (Table II. Each value was
calculated from 18 separate measurements performed on Carryover
the same blood sample. The coefficient of variation (CV) No significant carryover could be found (<0.001%)
of the reticulocyte percentages for low, medium, and high using manual or automatic run for the samples.
reticulocyte counts was 12.8%, 9.4%, and 1.52%, respec-
tively (Table 1). As expected, the CVs of the different para- Sensitivity
meters describing the reticulocyte maturation were larger, In all samples analyzed with an expected percentage of
especially for the low reticulocyte counts, owing to the rare reticulocytes of 0%, we found no percentages of reticulocytes

© American Society of Clinical Pathologists Am J Clin Pathol 1999; 112:677-686 679

Lacombe et al / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

superior to 0.2%. Nevertheless, in the same cases, it was variations of reticulocyte percentages for each time after
impossible to obtain a zero value. In these clinical condi- collection and 2 temperature storage conditions. A significant
tions, the sensitivity of the ABX PENTRA 120 Retic could decrease in reticulocyte percentages appeared after 48 hours of
be assumed as superior to 0.2%. storage at RT compared with 4°C storage (P < .01). The same
decrease was observed for absolute reticulocyte count (data
Reticulocyte Sample Stability not shown). There were no statistically significant differences
The stability of reticulocyte measurements was evalu- after storage at 4°C for percentage or absolute count of reticu-
ated after blood storage at RT and 4°C. Ten blood samples locytes between day 0 and any time measurement.
were analyzed at day 0 and 2, 4, 6, 8, 24, 48, and 72 hours ITable 21 shows the statistical differences obtained by
after collection. IFigure 31 shows the normalized percentage paired t test between day 0 and 2, 4, 6, and 8 hours after

B 15
I15

10

1=

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

:> o
o
a
<o _ c°
*J
c

1-1°
Q.
-15-

-20 -20
A2 A4 A6 A8 A24 A48* A72* A2 A4 A6 A8 A24 A 48 A72
Reticulocyte (%) Stability After A(h) at RT Reticulocyte (%) Stability After A(h) at 4°C

IFigure 3INormalized percentage of variation of reticulocyte percentages after blood storage. Ten blood samples were
analyzed at day 0, then 2 (A2), 4 (A4), 6 (A6), 8 (A8), 24 (A24), 48 (A48), and 72 (A72) hours after collection with 2 blood storage
conditions. A, At room temperature (RT). B, At 4°C. *P < .01.

B 15 B 6
10

5-I 4

0 2

-6H 0
sS » - i o i > -2

i
CO

I -15

-20
- -•- - RetL RT
--a--RetL4°C
- • - RetM + RetH RT
-4

-6
MFI RT
-25 - O - RetM + RetH 4°C -8 MFI 4°C
-30 -10

-35 -12
A2 A4 A6 \8 \2 A4 \6 \8
IFigure 41 Normalized variation percentages of immature reticulocyte fraction values after blood storage. Ten blood samples
were analyzed at day 0, then 2 (A2), 4 (A4), 6 (A6), 8 (A8) hours after collection with 2 blood storage conditions: room
temperature (RT) and 4°C. A, low RNA content (RetL), and medium RNA content plus high RNA content (RetM + RetH).
B, mean fluorescence index (MFI).

680 Am J ClinPathol 1999:112:677- © American Society of Clinical Pathologists

 Hematopathology / ORIGINAL ARTICLE

blood collection for the IRF parameters and RT and 4°C Comparisons between IRF parameters are fastidious
storage conditions. At RT, a significant decrease in the and open to controversy. In the present study, we first
youngest reticulocytes (RetM and RetH) and a significant compared values of HFR from the Sysmex R-2000 and
increase in the oldest (RetL) could be seen at 8 hours. More- RetH from the ABX PENTRA 120 Retic, which are
over, a significant decrease in RetM and an increase in RetL supposed to represent the youngest reticulocytes. In IFigure
was noted at 8 hours at 4°C. Cold storage of reticulocytes at 7AI, it is obvious that these parameters do not measure the
4°C did not prevent modifications of IRF parameters and same entity, especially regarding low percentages of reticu-
only delayed their maturity. A reliable analysis of such para- locytes (<0.5%). In this case (equation 2), the coefficient of
meters implies a blood analysis no more than 6 hours after correlation was only 0.26. When percentages of reticulo-
blood collection. cytes greater than 0.5% were considered (equation 1), the
IFigure 4AI shows the normalized percentage varia- coefficient of correlation was better (r - 0.57). Then, we
tions of RetL and RetM + RetH values for 2, 4, 6, and 8 compared MFR + HFR values with the Sysmex R-2000
hours after blood collection and 2 temperature storage and RetM + RetH values with the ABX PENTRA 120
conditions. On IFigure 4BI, it is clear that the MFI values Retic. In IFigure 7BI, as in Figure 7A, if samples with
significantly decreased from 6 hours after blood collection reticulocytes less than 0.5% were taken considered (equa-
at RT (P < .01). At 8 hours, the MFI values were signifi- tion 2), the coefficient of correlation was low (r = 0.375)
cantly lower at RT (P < .003) and 4°C (P < .05), but were and became better for samples with reticulocytes of 0.5%

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

more so in the former. or more (equation 1, r = 0.66). In any respect, in all these
analyses, it was impossible to find a good concordance for
Agreement Between Methods IRF parameters between the 2 instruments.
We compared values obtained with the ABX PENTRA In a final experiment, we compared 2 other IRF parame-
120 Retic and 3 other methods. Results are given in liable 31 ters: MFI with the ABX PENTRA 120 Retic and the MCF
and IFigure 51 and IFigure 61. The Pearson product moment with the Coulter XL. Initially, the Retic Count kit (Becton
(r) values were always greater than 0.906, and intraclass corre- Dickinson) was not designed to provide an estimation of
lation studies showed similar trends, with RI always greater reticulocyte immaturity. However, from raw data recorded in
than 0.896. Therefore, a very good concordance between all list mode with the XL flow cytometer, it was easy to calcu-
methods, including manual counting, was found. Moreover, late the MCF of reticulocytes stained with thiazole orange,
the analysis of slope and intercept values showed no biases. bearing in mind that the fluorescence intensities were

•Table 21
Probability of a Significant Difference Between Immature Reticulocyte Fraction Values for 10 Samples at the
Time of Blood Collection Compared With 2,4,6, and 8 Hours Later*

RetL RetM RetH RetM + RetH MM

2h NS/NS NS/NS NS/NS NS/NS NS/NS
NS/NS NS/NS NS/NS NS/NS NS/NS
6h .04/NS NS/NS NS/NS .02/NS .01/NS
8li 007/.002 .003/006 .005/NS .0003/0003 .003/.05

RetL = low RNA content; RetM = medium RNA content; Ret H = high RNA content; MFI = mean fluorescence index; NS = not significant.
* Values are given for 2 blood storage conditions (room temperature/4°C). The paired / test was used.

•Table 31
Comparison of 4 Methods for Reticulocyte Counting, Intraclass Correlation Coefficients (RI), Pearson Product Moment Values
(r), Slopes, and Regression Intercepts for 300 Samples
Pearson Product Moment Correlation

RI Slope Intercept
Man/R2000 0.935 0.937 0.921 0.0255
Man/Flow 0.896 0.906 1.051 -0.0508
Man/PENTRA120 0.932 0.945 1.047 -0.2623
R2000/Flow 0.919 0.935 1.103 -0.0266
R2000/PENTRA120 0.942 0.952 1.065 -0.2035
Flow/PENTRA120 0.933 0.946 0.900 -0.0473
Man = manual counting; R2000 - = Sysmex R-2000; Flow =ttowcytometry; PENTRA 120= A B X PENTRA 120 Retic.

© American Society of Clinical Pathologists Am J Clin Pathol 1999;112:677-686 681

L a c o m b e e t al / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

y=1.103x- 0.0266
r = 0.935

i 1

t 6 10 12
Coulter XL (%) Mean Reticulocyte (%)

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

y=1.0665x-0.2035
r= 0.952
Rl = 0.942

4 6 8 10 12
PENTRA 120 Retic (%) Mean Reticulocyte (%)

: 0.900x - 0.0473
r= 0.946
e 0.933

i —%
8 10 12

4 6
Mean Reticulocyte (%)
PENTRA 120 Retic (%)
IFigure 51 Agreement between 3 automated reticulocyte counting methods: Sysmex R-2000, ABX PENTRA 120 Retic, and flow
cytometry (Coulter XL). A, C, E, Scatter plots and regression equations for reticulocyte counting. B, D, F, Relation between
mean values and difference in values for reticulocyte counting. Rl = intraclass correlation.

682 Am J Clin Pathol 1999:112:677-686 CD American Society of Clinical Pathologists

 Hematopathologv / ORIGINAL ARTICLE

y = 0.9208x + 0.0255
r= 0.937
I Rl = 0.935

4 6 8 12 Mean Reticulocyte (%)

Sysmex R-2000 {%)

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

Rl = 0.896

I -1
10 12
4 ° 0 O

4 6 8 Mean Reticulocyte (%)

Coulter XL (%)

e : 0.932

<
EC
K
2
UJ

O)
(I!

4 6 8 12
1 10 J

PENTRA 120 Retic (%) Mean Reticulocyte (%)

•Figure 61 Agreement between 3 automated reticulocyte counting methods: Sysmex R-2000, ABX PENTRA 120 Retic, and flow
cytometry (Coulter XL) and manual counting. A, C, E, Scatter plots and regression equations for reticulocyte counting. B, D, F,
Relation between mean values and difference in values for reticulocyte counting. Rl = intraclass correlation.

© American Society of Clinical Pathologists Am J Clin Pathol 1999;112:677-686 6 8 3

 Lacombe et al / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

5 10 15 20 25 30 35 40 45 50
HFR MFR + HFR

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

IFigure 71 Comparison of immature reticulocyte fraction parameters: values for high fluorescence rate (HFR; RNA content) and
medium fluorescence rate (MFR; RNA content) obtained on the Sysmex R-2000; high RNA content (RetH) and medium RNA content
(RetM) obtained on the ABX PENTRA 120 Retic. A, HFR and RetH values are compared with 2 modalities: blood samples with less
than 0.5% reticulocytes (Ret %) and with 0.5% or more reticulocytes. B, HFR + MFR and RetH + RetM values are compared with
2 modalities: blood samples with less than 0.5% reticulocytes and with 0.5% or more reticulocytes. The corresponding regression
equations are indicated for each: equation 1 for 0.5% or more reticulocytes, and equation 2 for less than 0.5% reticulocytes.

45 T Equation 1 represented on a logarithmic scale. On IFigure 81, it may be

y = 22.53x +7.0311 noted that MFI and MCF were significantly correlated,
40 r=0.75
especially when a logarithmic regression was used (equa-
35 tion 2, r = 0.79).
30
25
20
Discussion

15H
Our study of the operating characteristics of the ABX
PENTRA 120 Retic analyzer demonstrates that it has good
10 Equation 2 precision and linearity and no significant carryover. We studied
y = 14.73Ln(x) + 28.82
5- r=0.79 in depth the stability of reticulocyte counts and IRF parameters
for several times of conservation (2 to 72 hours) with different
0
0.5 1.5 storage conditions for blood samples: RT and 4°C. There was a
XL(MCF) significant decrease in the percentage and, thus, the absolute
reticulocyte count when samples were stored at RT. This differ-
IFigure 81 Comparison of immature reticulocyte fraction
ence disappeared at a 4°C storage condition. In our hands,
parameters: values for mean fluorescence index (MFI)
reticulocyte analysis must be performed before 24 hours at RT
obtained on the ABX PENTRA 120 Retic and mean channel
and may be delayed 72 hours at 4°C storage conditions. The
fluorescence (MCF) obtained on the Coulter XL (XL). Values
same recommendations do not hold if IRF parameters are to be
for MFI and MCF were compared by using 2 regression
studied. In this case, at RT and 4°C, it was impossible to obtain
equation modalities: equation 1, linear regression; equation
reliable and reproducible results after 24 hours of storage (data
2, logarithmic regression.
not shown). For conservation times between 2 and 8 hours, it is
obvious in Table 2 that a maturation of reticulocytes occurs in
vitro and at a greater rate with RT storage conditions. The
number of the youngest reticulocytes (RetH + RetM)
decreased from 6 hours, and an increase in more mature

684 Am J Clin Pathol 1999; 11X677-686 © American Society of Clinical Pathologists

 Hematopathology / ORIGINAL ARTICLE

reticulocytes (RetL) was noted at the same time. The findings Some interferences are well known to perturb good
were similar when considering MFI as a measurement of IRF. reticulocyte counting (eg, nucleated RBCs, RBC inclusions,
In all cases, the maturation rate was slower at 4°C, as shown cold agglutinins, hemoglobinopathy, high WBC count).5 In
in Figure 4 (P < .03). the present series, we verified that the reticulocyte count was
The in vitro stability of the reticulocyte count was not affected by nucleated RBCs and high WBC counts (data
checked by Cavill et al20; in their collaborative study, no not shown), but more cases are needed for analyzing other
significant decrease of reticulocyte counts was found at RT interferences.
or at 4°C. However, RT samples with high reticulocyte From this comparative study of 300 blood samples, we
counts decreased during the first 24 hours but not at 4°C. No conclude that the new ABX PENTRA 120 Retic reticulocyte
systematic study of IRF parameter evolution with time of method provides excellent results near those with flow
conservation has been published, so it should be done to fix cytometry and the Sysmex R-2000. Furthermore, the ability
the necessary storage conditions for blood samples in routine of the ABX PENTRA 120 Retic to provide a reticulocyte
clinical diagnosis. count and several modalities for IRF parameters in a fully
We performed a comparison of 4 methods for reticulo- automated and integrated way is a labor-saving advantage
cyte counting and 3 for IRF parameter evaluation. Consid- and represents a way of improving clinical diagnoses.
ering the former, with the classic Pearson product-moment
correlation (r), we found excellent agreement between all From the Laboratory of Hematology and the Blood Disease
Department, Hopital Haut-Leveque, CHU Bordeaux, France.

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

methods (Table 3 and Figures 5 and 6). To estimate the level
of agreement, we used the intraclass correlation coefficient Address reprint requests to Dr Lacombe: Laboratoire
d'Hematologic Hopital Haut-Leveque, 33604 Pessac, France.
and classically considered that a satisfactory level was
achieved when the lower limit of the 95% confidence
interval was at least 0.75. !8J9 Unlike Brugnara et al,6 we did References
not find poor correlations between manual and automated
1. Tanke HJ, Rothbarth PH, Vossen JMJJ, et al. Flow cytometry
counting; all the RI values were more than 0.890. This of reticulocytes applied to clinical hematology. Blood.
difference could be explained by a higher number of RBCs 1983;61:1091-1097.
counted by the operator (2,000 in our study). In our hands, 2. National Committee for Clinical Laboratory Standards.
all the methods gave good concordance. Methods for Reticulocyte Counting: Proposed Standards.
Villanova, PA: National CommitteeforClinical Laboratory
Unfortunately, comparisons of IRF parameters did not Standards; 1985. NCCLS document H16-P.
give the same good results. Even if Sysmex R-2000 and ABX 3. Davis BH, Bigelow NC. Flow cytometric reticulocyte
PENTRA 120 Retic did not use the same staining method, quantification using thiazole orange provides clinically useful
HFR, MFR, and LFR with the Sysmex R-2000 and RetH, reticulocyte maturity index. Arch Pathol Lab Med.
1989;113:684-689.
RetM, and RetL with the ABX PENTRA 120 Retic are compa-
4. Fergusson DJ, Lee SF, Gordon PA. Evaluation of reticulocyte
rable parameters to measure young reticulocytes. In Figure 7, it counts by flow cytometry in a routine laboratory. Am J
is obvious that even if a significant correlation could be found Hematoi. 1990;33:13-17.
for each parameter, the concordance between the instruments is 5. National CommitteeforClinical Laboratory Standards.
poor, and the calculation of RI showed a lack of concordance Methods for Reticulocyte Counting (Flow Cytometry and
Supravital Dyes); Approved Guideline. Villanova, PA: National
between the methods: LFR/RetL (RI = 0.573), MFR/RetM (RI CommitteeforClinical Laboratory Standards; 1997. NCCLS
= 0.577), and HFR/RetH (RI = 0.387). This became particu- document H44-A.
larly clear when considering data with low percentages of retic- 6. Brugnara C, Hipp MJ, Irving PJ, et al. Automated reticulocyte
ulocytes (<0.5%), and IRF parameters are without a doubt counting and measurement of reticulocyte cellular indices:
unreliable in this case. Similarly, MFI values seemed to be well evaluation of the Miles H*3 blood analyzer. Am J Clin Pathol.
1994;102:623-632.
correlated, but the RI calculation (RI = 0.70) showed that there
7. Bowen D, Bentley N, Hoy T, et al. Comparison of a modified
was not a good concordance between the methods. However, thiazole orange technique with a fully automated analyser for
although it is understandable that MFI was not as sensitive as reticulocyte counting. J Clin Pathol. 1991;44:130-133.
RetH for detecting the youngest reticulocytes, its repro- 8. Schimenti KJ, Lacerna K, Wamble A, et al. Reticulocyte
ducibility from one machine to another seems to be greater, and quantification byflowcytometry, image analysis, and manual
counting. Cytometry. 1992;13:853-892.
its systematic use could be of interest for better standardizing
9. Van-Houte AJ, Bartels PC, Schoorl M, et al. Methodology-
IRF parameters. Moreover, in Table 1, it seems that CVs for dependent variations in reticulocyte counts using a manual
MFI are better than those for RetH or RetM for any given and two different flow cytometric procedures. Eur] Clin Chem
instrument. Since the usefulness of IRF parameters in clinical Clin Biochem. 1994;32:859-863.
decisions is now well documented,1215 an international stan- 10. Davis BH, DiCorato M, Bigelow NC, et al. Proposal for
standardization of flow cytometric reticulocyte maturity index
dardization should be established. (RM1) measurements. Cytometry. 1993;14:318-326.

© American Society of Clinical Pathologists Am J Clin Pathol 1999; 112:677-686 685

Lacombe et al / ABX PENTRA 120 RETIC AND RETICULOCYTE ANALYSIS

11. Davis BH. Immature reticulocyte fraction (IRF): by any name, 16. Lee LG, Chen C-H, Chiu LA. Thiazole orange: a new dye for
a useful clinical parameter of erythropoietic activity. Lab reticulocyte analysis. Cytometry. 1986;7:508-517.
Hematol. 1996;2:2-8. 17. Tichelli A, Gratwohl A, Driessen A, et al. Evaluation of the
12. Davis BH, Bigelow N, Ball ED, et al. Utility of flow Sysmex R-1000®: an automated reticulocyte analyzer.
cytometric reticulocyte quantification as a predictor of Am J Clin Pathol. 1990;93:70-78.
engraftment in autologous bone marrow transplantation. 18. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing
Am J Hematol. 1989;32:81-87. rater reliability. Psychol Bull. 1979;86:320-428.
13. Watanabe K, Kawai Y, Takeuchi K, et al. Reticulocyte 19. Lee J, Koh D, Ong CN. Statistical evaluation of agreement
maturity as an indicator for estimating qualitative abnormality between two methods for measuring a quantitative variable.
of erythropoiesis. J C!in Pathol. 1994;47:736-739. Comput Bio! Med. 1989;19:61-70.
14. Lesesve JF, Lacombe F, Marit G, et al. High fluorescence 20. Cavill I, Kraaijenhagen R, Pradella R, et al. In vitro stability
reticulocytes are an indicator of bone marrow recovery after of the reticulocyte count. Ciin Lab Haematol. 1996; 18
chemotherapy. Eur J Haematol. 1995;54:61-63. (suppll):9-ll.
15. Corberand JX. Reticulocyte analysis using flow cytometry.
HematolCellTher. 1996;38:487-494.

Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 8, 2016

686 Am J Clin Pathol 1999;112:677-686 © American Society of Clinical Pathologists