Transfusion Medicine, 1992, 2,201-205
Platelet counting using plasma platelet concentrate samples R. A. Lord, G. A. Smith, M. J. Nightingale and F. E. Boulton
Wessex Regional Transfusion Centre,
Coxford Road, Southampton, U . K . Received 5 August 1991; accepted for publication I November 1991
900 x 109/1, whereas the E-2500 platelet counts were linear to 2700 x 109/l.A one-in-three pre-dilution was required to obtain accurate, linear counts with the Coulter counter, whereas the E-2500 was accurate without pre-dilution (mean count of 1030.2 x 109/1 compared to 1018.9 x 109/1 counted manually). In conclusion, the method of platelet counting may affect true platelet yields.
S U M M A R Y .Platelet counting using samples of plasma from platelet concentrates prepared for transfusion was assessed. The methods employed included a manual phase-contrast method, and counting with Coulter S Plus and Sysmex E-2500 counters. All methods were reproducible (mean CV of 4.9, 2.2 and 1.4%, respectively). However, neat samples of platelet concentrates analysed by Coulter counter were inaccurate (mean count of 863.8 x 109/1 compared to 1018.9 x 109/1 counted manually). Moreover, the Coulter platelet counts were non-linear above
Key words: Coulter counter, platelet concentrate, platelet count, Sysmex (Toa) analyser.
Automated platelet counting methods are well established in haematology laboratories. Various proven technologies exist, such as light scattering, nephelometry, and impedance (Rowan et af., 1977; 1979; Dalton et af., 1980; Meyer et af., 1980; Bollinger et af., 1987; Payne et af., 1987). Problems with such techniques have been reported, for instance spuriously high platelet counts (Bacus et al., 1980; Meyer et af., 1980; Guthrie et af., 1981), however they are more precise than manual methods (Ashman, 1976; Ardern et af., 1982). Previous reports using these techniques have been concerned with platelet counts within the normal range (i.e. 150-400 x 109/1: Dacie & Lewis, 1984), high counts (i.e. 600-800 x 109/1:Dalton et af., 1980; Payne et af., 1987) or with low platelet counts, due to their clinical significance (Meyer et al., 1980; Cornbleet & Kessinger, 1985). Few reports, to our knowledge, have dealt with the higher platelet counts expected in platelet concentrates prepared for platelet transfusions (Russell & Tunbridge, 1982). Platelet counts in such samples may exceed 800 x 109/1,and in concentrates prepared by plateletpheresis the count may be two to four times higher. The quality assurance Correspondence: Dr R. A. Lord, Quality Assurance Laboratory. WRTC, Coxford Road, Southampton, SO9 SUP. U.K.
20 1
of platelet packs prepared for transfusion requires the use of accurate platelet counting procedures. With the publication of the ‘Guidelines for the Blood Transfusion Services in the United Kingdom’ (Clarke et af., 1986 and the introduction of MCA manufacturers’ licensing, platelet counting under such conditions was re-examined at the Wessex Regional Transfusion Centre (WRTC). MATERIALS A N D M E T H O D S Platelet concentrates are prepared at the WRTC using the method of Slichter & Harker (1971) and the Simplex strategy to optimize platelet yield (Reiss & Katz, 1976). Blood (450 ml) was collected from suitable donors (Aster et al., 1981) into double- or quad-SAGM closed system packs, containing citratephosphate dextrose adenosine-I (CPD-A 1) anticoagulant. Platelets were separated within 6 h of donation using a Beckman J-6B centrifuge with JS 4:2 rotor. Platelet-rich plasma was prepared by centrifuging the packs at 2700 rpm (1200 g) for 3 min at 22°C and slowed using a brake rate of 3. The supernatant thus produced was passed into a platelet satellite pack, leaving 1-2 cm to minimize leucocyte contamination. The packs were re-centrifuged a t 4200 rpm (3000 g)
202
R. A . Lord et al.
and 22°C for 5 min, and slowed using a brake rate of 5. The supernatant consisted of platelet-poor plasma, which was passed back into the main pack, leaving 5070 ml in which to re-suspend the platelet button. The satellite pack containing the prepared platelet concentrate was then sealed leaving a full length of transfer line (‘long tail’) and separated from the main blood pack followed by 45 min equilibration at 22°C and resuspension on a rotary mixer for 15 min. A ‘long tail’ was left in order to obtain approximately 5 ml platelet concentrate to sample for a platelet count. One hundred and twenty samples were randomly collected from such ‘long tails’ into 5 ml EDTA vacutainer tubes (Russell & Tunbridge, 1982) and analysed within 2 h. Dilutions of the platelet concentrates were made in EDTA-PBS or Isoton 111(isotonic solution supplied by Coulter Electronics). All samples were mixed prior to counting to ensure a uniform suspension of cells. Platelet counts were obtained using a manual technique. Briefly, a one in twenty dilution of platelet concentrate was made in ammonium oxalate diluting fluid (Dacie & Lewis, 1984). After mixing for 10 min, an improved Neubauer counting chamber was filled with suspension and placed in a moist chamber. This was left undisturbed for at least 45 min to allow the cells to settle (Ashman, 1976). The counting chamber was examined using a phase-contrast microscope and the numbers of platelets in five 1 mm2 areas counted (always exceeded 400 cells). This procedure was performed by a single experienced technician throughout. Platelet counts were also performed on a Coulter S Plus ( S + ) counter. In addition, the latter 70 platelet concentrate samples were also analysed on a Sysmex E-2500 analyser (Toa Medical Electronics, Japan). Both instruments were calibrated in line with the manufacturers’ specifications, e.g. 4C was used to calibrate the Coulter counter and Cellcheck-400 and Placheck-100 for the Sysmex counter.
and subsequent dilutions made in phosphate-buffered saline (PBS). The platelet count was known for the neat concentrate, theoretical counts were therefore calculated for each dilution. Each dilution was then counted in triplicate using all counting techniques. The mean observed count was plotted against the theoretical count. Cell contamination
To test the hypothesis that cell contamination, particularly cell debris, may interfere with platelet counting, the following preparations were analysed. Platelet-rich plasma was prepared as above (linearity studies). Subsequent dilutions were made in a suspension of whole blood debris prepared by a cell sonicator (MSE Soniprep 150). Each dilution was then counted in triplicate using all counting techniques, the mean observed platelet count was plotted against the theoretical count. Carryover assessment
The high platelet counts obtained using platelet concentrates may affect subsequent counts, therefore, carryover assessments were determined. The Coulter S + and Sysmex E-2500 counters were primed twice with diluent. Platelet counts were performed on plasma samples (manual counts were in excess of 1300 x 109/1)using both instruments. Diluent was then run twice on each instrument and carryover percentages calculated.
RESULTS Comparison of the methods
Ten replicate counts were performed for each technique on each of three platelet concentrate samples. The coefficient of variation (CV%) was calculated.
The results are presented in Table 1. There was a high degree of discrepancy between methods. The platelet counts ranged from 863.8 x 109/1(for samples counted with the Coulter counter) t o 1 3 2 0 . 6 ~109/1 (for 1/3 diluted samples counted with the Coulter counter). The mean platelet counts obtained by manual counting and with the E-2500 counter were comparable (1018-9 and 1030.2 x 109/1, respectively).
Linearity
Reproducibility
Each of five platelet concentrate packs underwent additional preparation. Each pack was spun at 4000 rpm (1OOOg) for 15 min. Three-quarters of the plateletpoor plasma thus produced was removed and the platelets were re-suspended. Manual platelet counts were performed in triplicate on the neat concentrate
The mean coefficient of variation ranged from 1.4% (Sysmex E-2500 counter) to 4.9% (manual technique). The CV for the neat and 1/3 dilution samples counted with the Coulter counter were 2.2 and 2.270, respectively. The samples used covered a range of platelet counts (320 to 1340 x 109/l). The CV for counts with
Reproducibility
Platelet counting 203 Table 1. Comparison of methods Methods
B
A ~~
Mean of differences (A - B)
n
~~~
Significant
t* ~
S+ (neat)
us
S+ (1/3)
us
E-2500 E-2500 E-2500
us us
Phase Phase Phase S+ (neat) S-t (1/3)
us
120 120
-204'6 - 58.0 12.4 +230.0 -79.2
7.61 2.06 1.34 5.73 2.29
+
70 70 70
~~
Yes NO NO Yes NO
* P=O.OOl. critical value of t=2.67.
Linearity was restored if such samples were diluted one in three.
the E-2500 counter improved at the higher range of platelet counts, whilst the CV for counts with the Coulter S+counter (for both neat and 1/3 dilution samples) tended to increase as the platelet count increased.
Cell contamination
Figure 2 shows the effects of cell debris contamination on platelet counting. Using either the Coulter Sf or Sysmex E-2500 counters, the effects of such contamination appear minimal on platelet counts ranging from 100 to 1300 x lo9/].
Linearity
Platelet counts with the manual technique and the Sysmex E-2500 counter showed excellent linearity up to a count of 2700 x I 09/1 (r = 0.98 and 0.99, respectively). However, counts with the Coulter S+counter using neat plasma platelet concentrate samples was non-linear beyond a count of 900x 109/1 (Fig. 1).
Carryover assessment
The percentage carryover for the first diluent run after
2700
2500 2300 2100 X Y
1900
e,
C
3
0
1700
0
1500 i Q,
1300
v)
0 Fig. 1. Platelet linearity studies for a C Coulter S Plus Counter. Using neat m platelet concentrate ( 0 )the correlation coefficient (r) was 0-68, the slope was 0.241, while the y-intercept was 2.98. Using platelet concentrate diluted one-in-three in Isoton 111 (O), . . the slope was 0439, the y-intercept was -0.55 and r was 0.94. Each point represents the mean of five determinations, error bars omitted for clarity.
3
1100
900 700 500
0
I
0
I
500 700
I
1
1100
I
I
1500
I
1
1900
I
1
2300
Theoretical count ( x lo9/ I)
I
I
2700
204 R. A . Lord et al. 1300 1200 -
d
1100
-
1000
-
\
0:
0
900-
7
2
800-
Y
c 3
0 0 TJ
$ & 0 n 0
700
-
600500
-
400-
C
5
300200
-
100
-
0
Fig. 2. Comparison of platelet counts by Coulter S Plus ( 0 ) and Sysmex E-2500 (0)counters using platelet I
I
I
l
l
1
I
I
I
a high platelet count ranged from 0.17 to 0.46% for 1/3 diluted and neat samples respectively, counted with the Coulters counter. The percentage carryover for the Sysmex E-2500 counter was 0.28%. The values obtained for the second diluent run were negligible (0.01-0.08 %). DISCUSSION Various aspects of platelet concentrate preparation have been identified (Aster et al., 1981). However, this study indicates that the method of platelet counting may affect apparent platelet yield. The Coulter S + counter, for example, was precise (CV ranging from 1-1 to 4.6%) but inaccurate (mean platelet count of 863.8 x 109/1 compared to 1018.9 x 109/1 counted manually). Moreover, the counts with the Coulter counter were non-linear above 900 x 109/1,thus confirming the work of Dalton et al., (1980). In addition, high platelet counts were reported as 999 x 109/1or Error code 10 on the Coulter S + counter. The raw results, thus flagged, were then only obtainable by manipulating the calibration switch. Diluting plasma platelet samples (i.e. a one in three dilution) greatly improved accuracy. The mean diluted platelet count was 1 3 2 0 . 6 ~109/1 compared to 1018.9 x 109/1by the manual method. However, during
i
I
1
i
concentrate samples contaminated with cell debris. Each point represents
the course of this study two samples, when diluted, gave spuriously high counts. The results were obviously wrong (5463 and 9876 x 109/1, whilst 1057 and 1 149 x 109/1counted manually) and were repeated. Such problems with the Coulter S + counter have previously been reported (Bacus et al., 1980; Meyer et a/., 1980; Guthrie et at., 1981). The use of a one-inthree pre-dilution of platelet samples on the Coulter S + counter has meant that the WRTC now produces platelet concentrates which are consistently within the absolute platelet yield specifications of the 'Guidelines for the Blood Transfusion Services in the United Kingdom'. In contrast, the Sysmex E-2500counter was precise (CV ranging from 0.9 to 1.4%) and accurate (mean count of 1030.2 x 109/1compared to 1018.9 x 109/1by the manual method). In addition, platelet counts with the E-2500 counter were linear up to a concentration of at least 2700 x 109/1,without the need to pre-dilute the samples, and only when manual counts were above 3400 x 109/1were results unobtainable. Moreover, the E-series of analysers offers a platelet/large cell ratio (P-LCR), which is a ratio of large platelets to total platelets (Payne et al., 1987). This unique feature, in conjunction with the platelet distribution width (PDW) and mean platelet volume (MPV), provides additional information concerning platelet popula-
Platelet counting 205 tions and abnormalities, and shows excellent correlation with platelet morphology studies (Greenwood et al., 1986). Diluting platelet concentrate samples (1/S and 1/10 dilutions were analysed in addition to 1/3 presented here) indicated that the poor performance of the Coulter counter was possibly due to coincidence counting. Other problems reported have included platelets recirculating behind the aperture and platelets passing near the wall of the aperture instead of through its centre axis (Bacus ef al., 1980). The hydrodynamic sheath flow system used in the Toa E-2500helps to eliminate such problems (Clarke et al., 1987; Payne et al., 1987; Warner et al., 1990). Moreover, the use of different sized apertures (50pm in the Coulter S+ counter and 76A2 pm in the E-2500 counter) may also contribute to the differing performances of these analysers. Red cell contamination, particularly cell debris and platelet carryover, were negligible using either the S + or E-2500 counter. In other words, the higher neat platelet counts obtained using the E-2500counter did not appear to be due to cell debris. Moreover, the high counts associated with these products were unlikely to influence subsequent platelet counts. In conclusion, the method of platelet counting may affect true platelet yields. Pre-dilution of platelet concentrate samples may be necessary, unless counts are performed on one of the modern generation of haematology analysers, such as the Sysmex E-series. This study also indicates that the use of traditional haematological equipment by Transfusion Centres may not be straightforward, as their needs are different. ACKNOWLEDGEMENTS We wish to thank the staff of Southampton General Hospital, The Chalybeate Hospital, Southampton and the Royal East Surrey Hospital, Redhill, for their help during this study.
REFERENCES Ardern, J.C., Urmston, A., Hyde, K., Gowenlock, A.H. & Maclver, J.E. (1982) Comparison of materials for quality control of platelet counting using the Coulter Model S Plus. Clinical and Laboratory Haematology, 4, 35-60. Ashman, B. (1976) An assessment of platelet counting methods. Medical Laboratory Sciences, 33, 20 1-208. Aster. R.H., Jenkins, J.W., Tovey, L.A.D. & Kahn, R.A.. (1981) Which are the parameters to be controlled in
platelet concentrates in order that they may be offered to the medical profession as a standardised product with specific properties? Vox Sanguinis, 40,11 5-126. Bacus, J.W., Watt, S. & Trobaugh, F.E. (1980) Clinical evaluation of a new electrical impedance instrument for counting platelets in whole blood. American Journal of Clinical Pathology, 73, 655-663. Bollinger, P.B., Drewinko, B., Brailas, C.D., Smeeton, N.A. & Trujillo, J.M. (1987). The Technicon H*l-An automated hematology analyser for today and tomorrow. American Journal of CIinical Pathology, 87, 7 1-78. Clarke, A., Garvey, B. & Lewis, S.M. (1986) An evaluation of the Sysmex Toa E5000 analyser. Department of Health and Social Security. Cornbleet, J.P. & Kessinger, S. (1985) Accuracy of low platelet counts on the Coulter S-Plus IV. American Journal of Clinical Parhology, 83, 78-80. Dacie, J.C. & Lewis, S.M. (1984) Practical Haematology 6th edn. Churchill Livingstone, London. Dalton, W.T., Bollinger, P. & Drewinko, B. (1980) A sideby-side evaluation of four platelet counting instruments. American Journal of Clinical Pathology, 74, 119-1 34. Greenwood, D., Kingston, P.J. & Tinge, A. (1986) Preliminary evaluations of the Sysmex E-5000 haematology analvser. Gloucester Royal Hospital, Gloucester. Guthrie, D.L., Campbell, S.R. & Maidment, N.J.M. (1981) Platelet counting errors with the Coulter counter model ‘S’Plus. Medical Laboratory Sciences, 38, 385-388. Meyer, K., Chin, B., Magnes, J., Thaler, T., Lotspeich, C. & Baisley, A. (1980) Automated platelet counters. A comparative evaluation of the latest instrumentation. American Journal of Clinical Pathology, 74, 135- 150. Payne, B.A., Pierre, R.V. & Lee, W.K. (1987) Evaluation of the Toa E-5000 automated hematology analyser. American Journal of Clinical Parliology, 88, 5 1-57. Reiss, R.F. & Katz. A.J. (1976) Optimizing recovery of platelets in platelet rich plasma by the Simplex strategy. Transfusion, 16, 370-374. Rowan, R.M., Fraser, C . , Gray, J.H. & McDonald, G.A. (1979) The Colter counter Model ‘S’ Plus-the shape of things to come. Clinical and Laboratory Haematology, 1, 29-40. Rowan, R.M., McDonald, G.A. & Nicoll, W.D. (1977) Automated platelet counting. British Journal of Haematology, 35, 666-667. Russell, W.J. & Tunbridge, L.J. (1982) Platelet counts in stored donor blood. Anaesrhesia Intensive Care, 10,271273. Slichter, S.J. & Harker, L.A. (1971) Preparation and threeday storage of functionally intact platelet concentrates. In: Book of Abstracts 2nd Congress of the International Society of Thrombosis and Haemosfasis, A. Liss, Oslo. Warner, B.A.. Reardon, D.M. & Marshall, D.P. (1990) Automated haematology analysers: a four-way comparison. Medical Laboratory Sciences, 47, 285-296.
.
Platelet counting using plasma platelet concentrate samples R. A. Lord, G. A. Smith, M. J. Nightingale and F. E. Boulton
Wessex Regional Transfusion Centre,
Coxford Road, Southampton, U . K . Received 5 August 1991; accepted for publication I November 1991
900 x 109/1, whereas the E-2500 platelet counts were linear to 2700 x 109/l.A one-in-three pre-dilution was required to obtain accurate, linear counts with the Coulter counter, whereas the E-2500 was accurate without pre-dilution (mean count of 1030.2 x 109/1 compared to 1018.9 x 109/1 counted manually). In conclusion, the method of platelet counting may affect true platelet yields.
S U M M A R Y .Platelet counting using samples of plasma from platelet concentrates prepared for transfusion was assessed. The methods employed included a manual phase-contrast method, and counting with Coulter S Plus and Sysmex E-2500 counters. All methods were reproducible (mean CV of 4.9, 2.2 and 1.4%, respectively). However, neat samples of platelet concentrates analysed by Coulter counter were inaccurate (mean count of 863.8 x 109/1 compared to 1018.9 x 109/1 counted manually). Moreover, the Coulter platelet counts were non-linear above
Key words: Coulter counter, platelet concentrate, platelet count, Sysmex (Toa) analyser.
Automated platelet counting methods are well established in haematology laboratories. Various proven technologies exist, such as light scattering, nephelometry, and impedance (Rowan et af., 1977; 1979; Dalton et af., 1980; Meyer et af., 1980; Bollinger et af., 1987; Payne et af., 1987). Problems with such techniques have been reported, for instance spuriously high platelet counts (Bacus et al., 1980; Meyer et af., 1980; Guthrie et af., 1981), however they are more precise than manual methods (Ashman, 1976; Ardern et af., 1982). Previous reports using these techniques have been concerned with platelet counts within the normal range (i.e. 150-400 x 109/1: Dacie & Lewis, 1984), high counts (i.e. 600-800 x 109/1:Dalton et af., 1980; Payne et af., 1987) or with low platelet counts, due to their clinical significance (Meyer et al., 1980; Cornbleet & Kessinger, 1985). Few reports, to our knowledge, have dealt with the higher platelet counts expected in platelet concentrates prepared for platelet transfusions (Russell & Tunbridge, 1982). Platelet counts in such samples may exceed 800 x 109/1,and in concentrates prepared by plateletpheresis the count may be two to four times higher. The quality assurance Correspondence: Dr R. A. Lord, Quality Assurance Laboratory. WRTC, Coxford Road, Southampton, SO9 SUP. U.K.
20 1
of platelet packs prepared for transfusion requires the use of accurate platelet counting procedures. With the publication of the ‘Guidelines for the Blood Transfusion Services in the United Kingdom’ (Clarke et af., 1986 and the introduction of MCA manufacturers’ licensing, platelet counting under such conditions was re-examined at the Wessex Regional Transfusion Centre (WRTC). MATERIALS A N D M E T H O D S Platelet concentrates are prepared at the WRTC using the method of Slichter & Harker (1971) and the Simplex strategy to optimize platelet yield (Reiss & Katz, 1976). Blood (450 ml) was collected from suitable donors (Aster et al., 1981) into double- or quad-SAGM closed system packs, containing citratephosphate dextrose adenosine-I (CPD-A 1) anticoagulant. Platelets were separated within 6 h of donation using a Beckman J-6B centrifuge with JS 4:2 rotor. Platelet-rich plasma was prepared by centrifuging the packs at 2700 rpm (1200 g) for 3 min at 22°C and slowed using a brake rate of 3. The supernatant thus produced was passed into a platelet satellite pack, leaving 1-2 cm to minimize leucocyte contamination. The packs were re-centrifuged a t 4200 rpm (3000 g)
202
R. A . Lord et al.
and 22°C for 5 min, and slowed using a brake rate of 5. The supernatant consisted of platelet-poor plasma, which was passed back into the main pack, leaving 5070 ml in which to re-suspend the platelet button. The satellite pack containing the prepared platelet concentrate was then sealed leaving a full length of transfer line (‘long tail’) and separated from the main blood pack followed by 45 min equilibration at 22°C and resuspension on a rotary mixer for 15 min. A ‘long tail’ was left in order to obtain approximately 5 ml platelet concentrate to sample for a platelet count. One hundred and twenty samples were randomly collected from such ‘long tails’ into 5 ml EDTA vacutainer tubes (Russell & Tunbridge, 1982) and analysed within 2 h. Dilutions of the platelet concentrates were made in EDTA-PBS or Isoton 111(isotonic solution supplied by Coulter Electronics). All samples were mixed prior to counting to ensure a uniform suspension of cells. Platelet counts were obtained using a manual technique. Briefly, a one in twenty dilution of platelet concentrate was made in ammonium oxalate diluting fluid (Dacie & Lewis, 1984). After mixing for 10 min, an improved Neubauer counting chamber was filled with suspension and placed in a moist chamber. This was left undisturbed for at least 45 min to allow the cells to settle (Ashman, 1976). The counting chamber was examined using a phase-contrast microscope and the numbers of platelets in five 1 mm2 areas counted (always exceeded 400 cells). This procedure was performed by a single experienced technician throughout. Platelet counts were also performed on a Coulter S Plus ( S + ) counter. In addition, the latter 70 platelet concentrate samples were also analysed on a Sysmex E-2500 analyser (Toa Medical Electronics, Japan). Both instruments were calibrated in line with the manufacturers’ specifications, e.g. 4C was used to calibrate the Coulter counter and Cellcheck-400 and Placheck-100 for the Sysmex counter.
and subsequent dilutions made in phosphate-buffered saline (PBS). The platelet count was known for the neat concentrate, theoretical counts were therefore calculated for each dilution. Each dilution was then counted in triplicate using all counting techniques. The mean observed count was plotted against the theoretical count. Cell contamination
To test the hypothesis that cell contamination, particularly cell debris, may interfere with platelet counting, the following preparations were analysed. Platelet-rich plasma was prepared as above (linearity studies). Subsequent dilutions were made in a suspension of whole blood debris prepared by a cell sonicator (MSE Soniprep 150). Each dilution was then counted in triplicate using all counting techniques, the mean observed platelet count was plotted against the theoretical count. Carryover assessment
The high platelet counts obtained using platelet concentrates may affect subsequent counts, therefore, carryover assessments were determined. The Coulter S + and Sysmex E-2500 counters were primed twice with diluent. Platelet counts were performed on plasma samples (manual counts were in excess of 1300 x 109/1)using both instruments. Diluent was then run twice on each instrument and carryover percentages calculated.
RESULTS Comparison of the methods
Ten replicate counts were performed for each technique on each of three platelet concentrate samples. The coefficient of variation (CV%) was calculated.
The results are presented in Table 1. There was a high degree of discrepancy between methods. The platelet counts ranged from 863.8 x 109/1(for samples counted with the Coulter counter) t o 1 3 2 0 . 6 ~109/1 (for 1/3 diluted samples counted with the Coulter counter). The mean platelet counts obtained by manual counting and with the E-2500 counter were comparable (1018-9 and 1030.2 x 109/1, respectively).
Linearity
Reproducibility
Each of five platelet concentrate packs underwent additional preparation. Each pack was spun at 4000 rpm (1OOOg) for 15 min. Three-quarters of the plateletpoor plasma thus produced was removed and the platelets were re-suspended. Manual platelet counts were performed in triplicate on the neat concentrate
The mean coefficient of variation ranged from 1.4% (Sysmex E-2500 counter) to 4.9% (manual technique). The CV for the neat and 1/3 dilution samples counted with the Coulter counter were 2.2 and 2.270, respectively. The samples used covered a range of platelet counts (320 to 1340 x 109/l). The CV for counts with
Reproducibility
Platelet counting 203 Table 1. Comparison of methods Methods
B
A ~~
Mean of differences (A - B)
n
~~~
Significant
t* ~
S+ (neat)
us
S+ (1/3)
us
E-2500 E-2500 E-2500
us us
Phase Phase Phase S+ (neat) S-t (1/3)
us
120 120
-204'6 - 58.0 12.4 +230.0 -79.2
7.61 2.06 1.34 5.73 2.29
+
70 70 70
~~
Yes NO NO Yes NO
* P=O.OOl. critical value of t=2.67.
Linearity was restored if such samples were diluted one in three.
the E-2500 counter improved at the higher range of platelet counts, whilst the CV for counts with the Coulter S+counter (for both neat and 1/3 dilution samples) tended to increase as the platelet count increased.
Cell contamination
Figure 2 shows the effects of cell debris contamination on platelet counting. Using either the Coulter Sf or Sysmex E-2500 counters, the effects of such contamination appear minimal on platelet counts ranging from 100 to 1300 x lo9/].
Linearity
Platelet counts with the manual technique and the Sysmex E-2500 counter showed excellent linearity up to a count of 2700 x I 09/1 (r = 0.98 and 0.99, respectively). However, counts with the Coulter S+counter using neat plasma platelet concentrate samples was non-linear beyond a count of 900x 109/1 (Fig. 1).
Carryover assessment
The percentage carryover for the first diluent run after
2700
2500 2300 2100 X Y
1900
e,
C
3
0
1700
0
1500 i Q,
1300
v)
0 Fig. 1. Platelet linearity studies for a C Coulter S Plus Counter. Using neat m platelet concentrate ( 0 )the correlation coefficient (r) was 0-68, the slope was 0.241, while the y-intercept was 2.98. Using platelet concentrate diluted one-in-three in Isoton 111 (O), . . the slope was 0439, the y-intercept was -0.55 and r was 0.94. Each point represents the mean of five determinations, error bars omitted for clarity.
3
1100
900 700 500
0
I
0
I
500 700
I
1
1100
I
I
1500
I
1
1900
I
1
2300
Theoretical count ( x lo9/ I)
I
I
2700
204 R. A . Lord et al. 1300 1200 -
d
1100
-
1000
-
\
0:
0
900-
7
2
800-
Y
c 3
0 0 TJ
$ & 0 n 0
700
-
600500
-
400-
C
5
300200
-
100
-
0
Fig. 2. Comparison of platelet counts by Coulter S Plus ( 0 ) and Sysmex E-2500 (0)counters using platelet I
I
I
l
l
1
I
I
I
a high platelet count ranged from 0.17 to 0.46% for 1/3 diluted and neat samples respectively, counted with the Coulters counter. The percentage carryover for the Sysmex E-2500 counter was 0.28%. The values obtained for the second diluent run were negligible (0.01-0.08 %). DISCUSSION Various aspects of platelet concentrate preparation have been identified (Aster et al., 1981). However, this study indicates that the method of platelet counting may affect apparent platelet yield. The Coulter S + counter, for example, was precise (CV ranging from 1-1 to 4.6%) but inaccurate (mean platelet count of 863.8 x 109/1 compared to 1018.9 x 109/1 counted manually). Moreover, the counts with the Coulter counter were non-linear above 900 x 109/1,thus confirming the work of Dalton et al., (1980). In addition, high platelet counts were reported as 999 x 109/1or Error code 10 on the Coulter S + counter. The raw results, thus flagged, were then only obtainable by manipulating the calibration switch. Diluting plasma platelet samples (i.e. a one in three dilution) greatly improved accuracy. The mean diluted platelet count was 1 3 2 0 . 6 ~109/1 compared to 1018.9 x 109/1by the manual method. However, during
i
I
1
i
concentrate samples contaminated with cell debris. Each point represents
the course of this study two samples, when diluted, gave spuriously high counts. The results were obviously wrong (5463 and 9876 x 109/1, whilst 1057 and 1 149 x 109/1counted manually) and were repeated. Such problems with the Coulter S + counter have previously been reported (Bacus et al., 1980; Meyer et a/., 1980; Guthrie et at., 1981). The use of a one-inthree pre-dilution of platelet samples on the Coulter S + counter has meant that the WRTC now produces platelet concentrates which are consistently within the absolute platelet yield specifications of the 'Guidelines for the Blood Transfusion Services in the United Kingdom'. In contrast, the Sysmex E-2500counter was precise (CV ranging from 0.9 to 1.4%) and accurate (mean count of 1030.2 x 109/1compared to 1018.9 x 109/1by the manual method). In addition, platelet counts with the E-2500 counter were linear up to a concentration of at least 2700 x 109/1,without the need to pre-dilute the samples, and only when manual counts were above 3400 x 109/1were results unobtainable. Moreover, the E-series of analysers offers a platelet/large cell ratio (P-LCR), which is a ratio of large platelets to total platelets (Payne et al., 1987). This unique feature, in conjunction with the platelet distribution width (PDW) and mean platelet volume (MPV), provides additional information concerning platelet popula-
Platelet counting 205 tions and abnormalities, and shows excellent correlation with platelet morphology studies (Greenwood et al., 1986). Diluting platelet concentrate samples (1/S and 1/10 dilutions were analysed in addition to 1/3 presented here) indicated that the poor performance of the Coulter counter was possibly due to coincidence counting. Other problems reported have included platelets recirculating behind the aperture and platelets passing near the wall of the aperture instead of through its centre axis (Bacus ef al., 1980). The hydrodynamic sheath flow system used in the Toa E-2500helps to eliminate such problems (Clarke et al., 1987; Payne et al., 1987; Warner et al., 1990). Moreover, the use of different sized apertures (50pm in the Coulter S+ counter and 76A2 pm in the E-2500 counter) may also contribute to the differing performances of these analysers. Red cell contamination, particularly cell debris and platelet carryover, were negligible using either the S + or E-2500 counter. In other words, the higher neat platelet counts obtained using the E-2500counter did not appear to be due to cell debris. Moreover, the high counts associated with these products were unlikely to influence subsequent platelet counts. In conclusion, the method of platelet counting may affect true platelet yields. Pre-dilution of platelet concentrate samples may be necessary, unless counts are performed on one of the modern generation of haematology analysers, such as the Sysmex E-series. This study also indicates that the use of traditional haematological equipment by Transfusion Centres may not be straightforward, as their needs are different. ACKNOWLEDGEMENTS We wish to thank the staff of Southampton General Hospital, The Chalybeate Hospital, Southampton and the Royal East Surrey Hospital, Redhill, for their help during this study.
REFERENCES Ardern, J.C., Urmston, A., Hyde, K., Gowenlock, A.H. & Maclver, J.E. (1982) Comparison of materials for quality control of platelet counting using the Coulter Model S Plus. Clinical and Laboratory Haematology, 4, 35-60. Ashman, B. (1976) An assessment of platelet counting methods. Medical Laboratory Sciences, 33, 20 1-208. Aster. R.H., Jenkins, J.W., Tovey, L.A.D. & Kahn, R.A.. (1981) Which are the parameters to be controlled in
platelet concentrates in order that they may be offered to the medical profession as a standardised product with specific properties? Vox Sanguinis, 40,11 5-126. Bacus, J.W., Watt, S. & Trobaugh, F.E. (1980) Clinical evaluation of a new electrical impedance instrument for counting platelets in whole blood. American Journal of Clinical Pathology, 73, 655-663. Bollinger, P.B., Drewinko, B., Brailas, C.D., Smeeton, N.A. & Trujillo, J.M. (1987). The Technicon H*l-An automated hematology analyser for today and tomorrow. American Journal of CIinical Pathology, 87, 7 1-78. Clarke, A., Garvey, B. & Lewis, S.M. (1986) An evaluation of the Sysmex Toa E5000 analyser. Department of Health and Social Security. Cornbleet, J.P. & Kessinger, S. (1985) Accuracy of low platelet counts on the Coulter S-Plus IV. American Journal of Clinical Parhology, 83, 78-80. Dacie, J.C. & Lewis, S.M. (1984) Practical Haematology 6th edn. Churchill Livingstone, London. Dalton, W.T., Bollinger, P. & Drewinko, B. (1980) A sideby-side evaluation of four platelet counting instruments. American Journal of Clinical Pathology, 74, 119-1 34. Greenwood, D., Kingston, P.J. & Tinge, A. (1986) Preliminary evaluations of the Sysmex E-5000 haematology analvser. Gloucester Royal Hospital, Gloucester. Guthrie, D.L., Campbell, S.R. & Maidment, N.J.M. (1981) Platelet counting errors with the Coulter counter model ‘S’Plus. Medical Laboratory Sciences, 38, 385-388. Meyer, K., Chin, B., Magnes, J., Thaler, T., Lotspeich, C. & Baisley, A. (1980) Automated platelet counters. A comparative evaluation of the latest instrumentation. American Journal of Clinical Pathology, 74, 135- 150. Payne, B.A., Pierre, R.V. & Lee, W.K. (1987) Evaluation of the Toa E-5000 automated hematology analyser. American Journal of Clinical Parliology, 88, 5 1-57. Reiss, R.F. & Katz. A.J. (1976) Optimizing recovery of platelets in platelet rich plasma by the Simplex strategy. Transfusion, 16, 370-374. Rowan, R.M., Fraser, C . , Gray, J.H. & McDonald, G.A. (1979) The Colter counter Model ‘S’ Plus-the shape of things to come. Clinical and Laboratory Haematology, 1, 29-40. Rowan, R.M., McDonald, G.A. & Nicoll, W.D. (1977) Automated platelet counting. British Journal of Haematology, 35, 666-667. Russell, W.J. & Tunbridge, L.J. (1982) Platelet counts in stored donor blood. Anaesrhesia Intensive Care, 10,271273. Slichter, S.J. & Harker, L.A. (1971) Preparation and threeday storage of functionally intact platelet concentrates. In: Book of Abstracts 2nd Congress of the International Society of Thrombosis and Haemosfasis, A. Liss, Oslo. Warner, B.A.. Reardon, D.M. & Marshall, D.P. (1990) Automated haematology analysers: a four-way comparison. Medical Laboratory Sciences, 47, 285-296.
.
