Volume 44, January 2004

TRANSFUSION 119

Blackwell Science, LtdOxford, UKTRFTransfusion0041-11322003 American Association of Blood BanksJanuary 20044412Original Article

APHERESIS SYSTEM COMPARISONBUENO ET AL.

ABBREVIATIONS:

SDP(s) = single-donor PLT(s); TBV = total

blood volume; WB = whole blood.

From the Blood Donation Center of the Spanish Red Cross and

the Unit of Clinical Epidemiology, University Hospital Puerta de

Hierro, Madrid, Spain.

Address reprint requests to:

José Luis Bueno, MD,

Centro de Donación de Sangre Cruz Roja Española,

C/Juan Montalvo, 3 code 28040 Madrid, Spain; e-mail:

jlbueno@cdscruzroja.infonegocio.com.

Received for publication January 13, 2003; revision received

August 5, 2003, and accepted August 8, 2003.

TRANSFUSION

2004;44:119-124.

H E M A P H E R E S I S

A comparison of PLT collections from two apheresis devices

José Luis Bueno, Luisa Barea, Fernando García, and Emma Castro

BACKGROUND:

The aim of this study was to compare two apheresis devices (COBE Trima and COBE Spectra, Gambro BCT).

STUDY DESIGN AND METHODS:

The study compares 103 Trima procedures with 61 Spectra procedures. The comparison of single-donor PLTs (SDPs) separation parameters and anticoagulant infusion to the donors were the primary targets. Yield and residual amounts of WBCs of SDPs were secondary targets. Residual amounts of WBCs were measured with the Nageotte method.

RESULTS:

Trima and Spectra groups were comparable before apheresis. Two procedures were terminated before completion in the Trima group owing to vein damage. Trima infused more anticoagulant—352

±

104 mL versus 297

±

75 mL (p < 0.01)—and was quicker than Spectra. The time to obtain 3.5

¥

10

11

was (median) 55.8 minutes for the Trima machine and 80.3 minutes for Spectra machine (p < 0.001). Regarding leukoreduction, all the SDPs had fewer than 1

¥

10

6

WBCs per unit except for one in product obtained by the Trima machine.

CONCLUSIONS:

The Trima machine is faster and more efficient than the Spectra machine, and both machines allow standard leukoreduced SDPs to be obtained. Although donors receive a higher anticoagulant infusion with the Trima machine, their tolerance is acceptable.

pheresis shows no advantages for the donor—other than personal satisfaction—when com-pared with whole-blood (WB) donation. Al-though the apheresis technique has been

improved, some problems, such as the duration of theprocedure and donor discomfort owing to the anticoagu-lant, remain to be resolved, especially for plateletpheresis.The improvement of these variables is the current basis forthe development of new apheresis machines.

Recently, a new highly automated apheresis systemwas developed (COBE Trima, Gambro BCT, Lakewood,CO), which allows a combination of homogeneous com-ponents (automated blood collection) to be obtained. Theaim of the present study was to compare this new machinewith a well-known plateletpheresis machine (COBE Spec-tra, Gambro BCT), with respect to separation parameters,anticoagulant infusion, single-donor PLTs (SDPs) yield,and amounts of WBCs.

MATERIALS AND METHODS

Timing of the study

Donations were made during two different time periods.Spectra donations were made between December 1996and February 1997, and Trima between September 1998and August 1999.

Apheresis donors

Volunteer donors complied with the Council of EuropeGuidelines and Recommendations for apheresis

1

and gavetheir written consent before the procedures. Apheresisdonors were recruited from repeat WB donors (with atleast two previous WB donations). Other donor character-istics are shown in Table 1.

Blood cell separators

Trima is a new automated multicomponent blood collec-tion system that allows selective extraction of blood com-ponents. The system uses disposable single-needle setsand an ACD-A solution as anticoagulant and does notrequire the infusion of 0.9 percent sodium chloride orother solutions to the donor. The machines were operatedwith software versions 3.1 and 3.2 and were programmed

A

BUENO ET AL.

120 TRANSFUSION

Volume 44, January 2004

to obtain PLT product targets of 3.5

¥

10

11

for single-doseproduct, 7.1

¥

10

11

for double dose, and 10.5

¥

10

11

for tripledose. The procedures were performed to obtain FFP andRBCs when possible. ACD-A infusion rates and WB flowswere adjusted automatically. The inlet-to-ACD-A startramp ratio was 7:1. This was later automatically adjustedto 11:1. The vein draw and return pressure limits were setat 250 and 310 mmHg, respectively. Trima includes a mod-ified WBC reduction system (LRS) (see below) to obtainleukoreduced SDP and plasma.

The machine software is capable of detecting prob-lems that occur during the procedure that interfere withleukoreduction effectiveness or the product quality, inwhich case an alarm in the display warns that the productshould be checked. If there are no problems the message“Label the component as containing fewer than 1

¥

10

6

WBC content” is displayed at the end of the procedure.COBE Spectra (Gambro BCT) is a well-known apher-

esis machine designed to perform different types of pro-cedures. It is used in our center to obtain leukoreducedSDPs and plasma from volunteer apheresis donors. Theprocedures were performed with computer software (ver-sions 5.1 LRS and Turbo 7.0 LRS). Single-needle sets wereused with ACD-A as anticoagulant, and a 0.9 percentsodium chloride solution was infused to the donor afterthe procedure. The inlet-to-AC ratio was programmed withthe “customized configuration.” This consists in a variablepreset ramp ratio that increases between 7:1 and 13:1 dur-ing the procedure so that the AC infusion decreases fromthe beginning to the end of the process. The machineswere programmed to obtain at least a standard SDP (3.5

¥

10

11

PLTs/unit) and as much plasma as possible. The sys-tem produces leukoreduced SDPs and plasma by a WBCreduction system (LRS). The operational principles of thissystem have recently been explained.

2,3

In the event ofproblems occurring during the procedure that could affectthe separation interface, and therefore compromise theleukoreduction of the product, there is a software alarmthat is similar to that previously described for the Trima

system. Because the total blood volume(TBV) displayed by the Spectra softwareincludes the amount of ACD-A con-sumed, this has been subtracted to allowcomparison with the Trima results.

PLT yield

The SDP yield given by the machinesat the end of the procedure were com-pared with laboratory measurements ofthe products.

4

For laboratory measure-ments, the PLT volume of the productswas automatically obtained from themachine software at the end of the pro-cess and checked by weighing the net

product. No density corrections were made for volumecalculations. After overnight agitation, SDP samples werediluted 1:5 with tempered ACD-A, and then the PLT con-centration and yields were measured with a cell counter(JT, Coulter, Hialeah, FL). The counter was calibrated dailywith appropriate low, medium, and high standard controlsamples.

WBC counting

WBC counting was performed following guidelines byBrandwein and Dickstein

5

but with a 1:5 dilution insteadof 1:10. SDPs were stood at 22

∞

C for at least 2 hours afterthe procedure, and then overnight under continuous agi-tation in a PLT incubator at 22

∞

C (Helmer PLT incubator,Baxter, Deerfield, IL). The following day, 1:5 dilutions ofthe samples were prepared with Türk’s solution (Diagnos-tica, Merck, Darmstadt, Germany). The dilutions werekept in a humid atmosphere at room temperature for atleast 20 minutes. Nageotte chambers (Nageotte modifiedbright-lined, Gambro BCT) were then loaded with 100

m

Lof the dilution in both grids and stored between 10 and 30minutes. The WBCs inside the grid area were counted witha microscope at 200

¥

magnification (Nikon, Japan). Onecell observed in both grids of the Nageotte chamber cor-responds to 0.05 WBCs per

m

L, the lowest detectable limitof the technique accepted. When no WBCs were found, thecalculations were performed as if 1 WBC was seen in thevolume actually counted. This prevented the final resultfrom being biased toward a low count and allowed loga-rithmic analysis.

2

The samples were counted by two differ-ent technicians, one from the Spectra group and the otherfrom Trima. Concordance between the two technicianswas validated through counting 26 additional PLT units.Intraclass correlation was 0.96 IC (0.92-0.98).

Samples before and after apheresis

At the start of the procedures, the average count of the lastfive donations was introduced. Peripheral blood samples

TABLE 1. Donor characteristics before and after apheresis

*

CharacteristicCOBE Spectra LRS

(n = 62)†COBE Trima(n = 117)† p value

Before apheresisSex (% men) 79 76 0.62WB volume (mL) 4803

±

877 (4770) 4865

±

818 (4974) 0.64Hct (%) 43.2

±

4.0 (43.9) 44.5

±

3.3 (45) 0.03Number of PLTs (

¥

10

3

/

m

L) 252

±

44 (247) 241

±

37 (241) 0.09

After apheresisHct (%) 43.6

±

4.2 (44.1) 43.9

±

3.9 (45) 0.58Number of PLTs (

¥

10

3

/

m

L) 191

±

35 (188) 176

±

33 (179) 0.005

* For the Hct comparisons, both before and after apheresis, there were 56 valid cases in the COBE Spectra LRS group. In the PLT comparison after apheresis, there were 61 cases in the COBE and 105 in the Trima groups. In the comparison of WB volume, there were 114 valid cases in Trima.

† Data are reported as mean

±

SD (median).

APHERESIS SYSTEM COMPARISON

Volume 44, January 2004

TRANSFUSION 121

before apheresis were obtained just after starting the pro-cedure to avoid additional venipunctures of apheresisdonors. When the new PLT count was available, usuallywithin 5 to 10 min, this was entered in the machine soft-ware replacing the previous value. Samples after apheresiswere taken immediately after the procedure. To preventsample dilution, 10 mL of blood was discarded from theattached line before taking the sample. An automatedblood cell counter was used for sample counting (JT,Coulter).

Anticoagulant and hypocalcemia symptoms

The Trima software calculates TBV processed, ACD-Ainfused to the donor, and time of the procedure. The Spec-tra software does not show the TBV processed or the ACD-A infused to donors, but both have been calculated toallow comparisons.

For Spectra, TBV processed was calculated by sub-tracting the ACD-A consumed from the total volume pro-cessed by the machine. The ACD-A infused to the donorwas calculated by subtracting the amount of ACD-A in thePLTs and plasma obtained from the total ACD-A used.

The ACD-A volume in the products was calculatedwith the following formula provided by Gambro BCT:

Volume of plasma (mL) + Volume of PLTs (mL)/((R

-

1)(1

-

Hct) + 1).

R = (WB + ACD-A volume)/ACD-A volume.

We evaluated donors’ risk of hypocalcemia with the ACD-A infusion per minute and liter of volemia (mL/(min

¥

L)).When initial hypocalcemia symptoms occurred, theanticoagulant and blood rate was diminished. Donorsreceived a calcium tablet orally (1 mg calcium carbonate,Mastical, BYK ELMU, Madrid, Spain) when symptomswere not relieved.

Evaluation of machines and separation parameters

The following formulae were used for the separationparameters:

Efficiency (%) = PLT yield per SDP (

¥

10

11

)

¥

100/[processed blood volume (mL)

¥

(donor PLTs before count + donor PLTs after count)/2] (

¥

10

9/

L)].

Yield per hour (10

11

PLTs) = Yield

¥

60/minutes of procedure.

Time for a standard dose (min) = minutes of procedure

¥

3.5

¥

10

11

PLTs per unit/laboratory yield (

¥

10

11

PLTs per unit).

Statistical analysis

Homogeneity of proportions was tested with chi-squaretests or the Fisher exact test, and homogeneity of means

was tested with a t test for independent or paired samplesdepending on the cases. Statistical significance was set atprobability values at or below 0.05, and no correctionswere made for multiple comparisons. All comparisonsincluded their 95 percent CIs. Comparison between themeasured PLTs yield and that predicted by the machinesoftware was made with intraclass correlation with a 95percent CI, linear regression analysis, mean comparison,and Bland-Altman plots. The same tests were used forthe correlation evaluation of technicians in the PLT WBCcounting. Analyses of residual WBCs were performedgraphically with lognormal probability after log transfor-mation.

6

Lognormal probability plots of the residual WBCsincluded linear regression estimates of percentiles of thenormal probability distribution (probits) against the loga-rithm of residual WBCs. Those observations which were ator below the minimum detectable level (

±

) were excludedfrom the linear regression analysis but incorporated intothe lognormal probability plot. A computer program file(Excel, Microsoft, Redmond, WA), provided by GambroBCT, was used to make these plots. A Fisher test was usedfor the comparison of hypocalcemia symptoms in donors.

RESULTS

Procedures

For the Trima group, 117 procedures were performed fromwhich only 103 were fully analyzed because some datawere missing from 14 procedures. All the procedureswhich rendered a product ready for distribution wereincluded even though some of them finished before theplanned time. The reasons for terminating the proceduresearlier in the Trima group were damaged veins in 2, vaso-vagal reaction in 1, and at the donor’s request in 1 case. Inthe Spectra group, 62 donations were analyzed with thesame criteria. One procedure was terminated before com-pletion owing to the machine emitting an unrecoverablealarm.

Donors

Gender, PLTs counts before apheresis, and WB volumefrom donors were similar in the Trima and Spectra groups,but the Trima group had a slightly higher Hct before aph-eresis (Table 1). Donor PLT counts after the procedureswere lower in Trima than in Spectra: respective means, 176

±

33 and 191

±

35; difference, 15; and 95 percent CI, 5

¥

10

9

to 26

¥

10

9

PLTs per L.

Hypocalcemia symptoms

One of 50 (2%) donors in the Spectra group and 9 of 110(8%) donors in the Trima group suffered mild hypocalce-mia symptoms (p = 0.174). Adverse effects related to ACwere not collected from 10 donors in the Spectra group

BUENO ET AL.

122 TRANSFUSION

Volume 44, January 2004

and 7 in the Trima group. Three cases in the Trima groupreceived oral calcium because hypocalcemia symptomswere not relieved after reducing the AC infusion rate. Theanticoagulant volume infused to the donors and the infu-sion rate per minute and liter of volemia are shown inTable 3.

Product characteristics are shown in Table 2. Themeasured laboratory PLT yield from the Trima machinecorrelated well with the level predicted by the system(intraclass correlation, 0.71; 95% CI, 0.61-0.79). The mea-sured mean yield was 0.36

¥

10

11

PLTs above the predictedvalue (p = 0.002). Linear regression analysis found a rela-tion consistent with an identity line (measured yield =predicted yield) and 92 and 68.1 percent of the unitswere over 3.0

¥

10

11

and 3.5

¥

10

11

PLTs per unit, respec-tively. Although Trima was set to obtain a dose of 3.5

¥

10

11

per procedure, less than 75 percent of the units fulfilledthis requirement.

Both machines showed good leukoreduction efficacybut the “WBC alarm” was activated during several proce-

dures in our Trima machine. The levelof leukoreduction in Trima machinewas kept below the WBCs per unit limitrequired in all procedures except forone, and 99.0 and 84.1 percent of theunits were less than 1

¥

10

6

and 1

¥

10

5

WBCs per unit, respectively. Figure 1shows the WBC count comparison inSDPs between the two machines. Nosignificant differences were foundbetween the two systems regardingWBC reduction (p = 0.19). With re-spect to PLT concentration, the samepercentage of units in both groupscomplied with the manufacturer’s rec-ommended limits for the type of plasticbag used,

7

but Trima products weremore concentrated than those fromSpectra. Separation parameters areshown in Table 3.

DISCUSSION

This study shows that the Trima apheresis system is quickerand infuses a higher volume of anticoagulant to the donorthan Spectra. Both systems extract PLTs of similar charac-teristics with respect to yield and leukoreduction.

Compared with other publications on Trima,

8-16

ourstudy focuses on separation features. As differences inthe duration of procedures directly concern apheresisdonors,

17

we consider “time for a standard dose” to be anextremely valuable parameter. It represents the inversevalue of “yield per hour.” As Trima works to obtain precon-figured doses of products, this new setting is useful toknow the time taken to obtain a standard unit of PLTs. Itallows optimization of our apheresis schedule, especiallywhen many of our donors are repeat donors with smallvariations in TBV and PLT count before apheresis and alsoevaluates the accuracy of machines in predicting theyields.

TABLE 2. Product characteristics

COBE Spectra LRS (n = 62) COBE Trima (n = 117) Difference (95% CI) p valuePLT volume (mL)* 291

±

48 (280) 283

±

88 (255) 8 (

-

16 to 31) 0.53PLT concentration (

¥

10

9

/L)* 1340

±

342 (1328) 1547

±

654 (1482)

-

206 (

-

381 to

-

32) 0.021000-2100 (

¥

10

9

/L)†‡ 53 (85) 98 (84) 2 (

-

5 to 9) 0.63PLT yield (

¥

10

11

)*§ 3.90

±

1.16 (3.80) 4.22

±

1.70 (3.80)

-

0.32 (

-

0.79 to 0.15) 0.193

¥

10

11

PLTs 55 (89) 95 (92)

-

3% (-9% to 2%) 0.443.5 ¥ 1011 PLTs 39 (63) 70 (68) -5% (-15% to 5%) 0.31

Amount of WBCs (¥106)*§ 0.03 ± 0.02 (0.03) 0.09 ± 0.40 (0.03) -0.07 (-0.16 to 0.03) 0.19Products with <1 ¥ 106 WBCs 62 (100) 102 (99) 1% (-1% to 3%) 0.31

* Data are reported as mean ± SD (median).† Gambro BCT recommendations.‡ Data are reported as number (%).§ In the PLT yield and WBC comparisons there were only 103 valid cases in the Trima group.

Fig. 1. Lognormal plot of WBC counts obtained with Trima (—) and Spectra(– ·· –) aph-

eresis machines.

0.001

0.1

1.0

5.010.0

30.0

50.0

70.0

90.095.0

99.0

99.9

0.01

Per

cent

age

of c

umul

ativ

e di

strib

utio

n

0.1

WBC count (¥ 106)

1 10

APHERESIS SYSTEM COMPARISON

Volume 44, January 2004 TRANSFUSION 123

Trima exposes donors to a higher risk of hypocalce-mia expressed by the AC infusion rate per minute and literof volemia. This higher risk of hypocalcemia events is notsignificant; nevertheless, there is a trend to a higher ratein the Trima group, but this difference could become sig-nificant in a comparison including a higher number ofcases.

At present WBC reduction in blood components rep-resents an important improvement in transfusion.18-20

Although no significant differences were found in WBCcount between the two groups, the only instance of aTrima procedure above the residual WBC limit deservessome comments. A “low-yield” alarm was detected duringthis procedure advising the operator to check the product.The donor was a young woman who is currently an activeapheresis donor and no further problems have beendetected in subsequent donations. Low PLT count beforeapheresis, lipemic plasma, high Hct, or flow disturbancesare some of the problems that supposedly increase theWBC contamination of SDPs in other apheresis sys-tems.2,21 During our study, the same alert advising productchecking occurred with Trima, usually when an eventforced stoppage of the centrifuge, but the WBC count wasabove standard only in the aforementioned procedure.Following these results, and comparing the two machineswith other devices in which the counting of residual WBCsin SDPs is recommended,22 it is believed that the LRS sys-tem provides greater security in this matter and it seemsacceptable to perform WBC counts only in those SDPswhere a WBC alarm is shown.

PLT count after apheresis was lower in Trima donorscompared with those in Spectra. In a previous study com-paring Spectra with Amicus,23 the authors explained thatlower PLT counts after the procedure in Spectra donorswas due to greater hemodilution in this procedure. In ourstudy, Spectra donors received an infusion of 0.9 percentsodium chloride solution after the procedure, and thosesamples taken after the procedure may have been dilutedwhen compared with those from Trima. In this case, dif-ferences between the PLT counts could actually be greater.Both machines show an average product collection of 3.7

¥ 1011 PLTs per unit (data not shown) which means thatPLT extraction was similar in the two machines. We do notthink that differences in efficiency parameters justify thisPLT count difference after apheresis. Perhaps the rinsein the Spectra group after the procedure allows the PLTsthat remain in the set to be returned to the donor, whereasin Trima procedures these could remain in the set. Thishypothesis was verified by checking an “in vitro” Trimaprocedure and conclude that the set can contain approx-imately 1 ¥ 1011 PLTs when discharged. In contrast, whenTrima obtains RBCs in addition to PLTs, PLTs are alsofound in the unit of RBCs.

The main limitation of our study is that the assign-ment of donors to apheresis devices was neither random-ized nor concurrent, which could lead to an interdonorvariability between donors groups; however, measuresbefore donation did not show any differences between thetwo groups except for slight differences in Hct.

In summary, we think Trima shows “time of proce-dure” as the most relevant advantage compared to Spec-tra. No significant differences or advantages are shown inthe other parameters evaluated. Finally, additional studiescomparing new apheresis machines are necessary todetermine the confluence of quick procedures with a lowrate of adverse events and high-quality products to offerdonors and patients.

ACKNOWLEDGMENTS

The authors thank the staff of the Apheresis and Quality Control

Units for their excellence. The authors also thank Martin Hadley-

Adams for his assistance with the English language and prep-

aration of the manuscript and Alicia Saturio for her secretarial

support.

REFERENCES

1. Guide to the preparation use and quality assurance of blood

components. 9th ed. Strasbourg: Council of Europe

Publishing; 2003.

2. Fournel JJ, Zingsem J, Riggert J, et al. A multicenter

TABLE 3. Separation parameters*COBE Spectra LRS (n = 62) COBE Trima (n = 117) Difference (95% CI) p value

Blood volume processed (TBV) (mL) 3545 ± 792 (3487) 3717 ± 1121 (3442) -171 (-490 to 147) 0.29ACD-A infused to donor (mL) 297 ± 75 (299) 352 ± 104 (298.7) 55 (24 to 86) <0.01ACD-A infused to donor per minute and

L of volemia (mL/min ¥ L)0.7 ± 0.1 (0.7) 1.1 ± 0.4 (1.05) 0.39 (0.28 to 0.48) <0.001

Separation time (min) 88 ± 15 (86) 69 ± 20 (65) 19 (13 to 25) <0.0001PLT yield/ hr (¥1011/h) 2.7 ± 0.9 (2.6) 3.8 ± 1.5 (3.7) -1.1 (-1.5 to -0.7) <0.0001Time for a 3.5 standard dose (min) 86.4 ± 34.4 (80.3) 64.6 ± 42.0 (55.8) 21.8 (20.9 to 22.8) 0.001Collection efficiency (%) 50.6 ± 10.9 (51.3) 55.4 ± 11.5 (56.5) -4.8 (-8.4 to -1.2) 0.009

* For the efficiency comparisons there were 61 valid cases in the COBE Spectra LRS group. For blood volume processed, ACD consumption, inlet WB-to-ACD ratio, separation time, efficiency, and collectancy there were 103 valid cases in the COBE Trima group.

† Data are reported as mean ± SD (median).

BUENO ET AL.

124 TRANSFUSION Volume 44, January 2004

evaluation of the routine use of a new white cell-reduction

apheresis system for collection of platelets. Transfusion

1997;37:487-92.

3. Riggert J, Humpe A, Simson G, Kohler M. Quality and safety

of platelet apheresis concentrates produced with a new

leukocyte reduction system. Vox Sang 1998;74:182-8.

4. Burgstaler EA, Pineda AA. Plateletapheresis:

instrumentation validation. Transfus Sci 1999;21:153-61.

5. Brandwein H, Dickstein R. Leukocyte filtration:

understanding counting methods and their implications.

New York: Pall Corporation; 1991.

6. Dumont LJ, Dzik WH, Rebulla P, Brandwein H. Practical

guidelines for process validation and process control of

white cell-reduced blood components: report of the

Biomedical Excellence for Safer Transfusion (BEST) Working

Party of the International Society of Blood Transfusion

(ISBT). Transfusion 1996;36:11-20.

7. Guide to platelet and plasma collections. Lakewood (CO):

Gambro BCT, Inc.; 2001.

8. Elfath MD, Whitley P, Jacobson MS, et al. Evaluation of an

automated system for the collection of packed RBCs,

platelets, and plasma. Transfusion 2000;40:1214-22.

9. Drillat P, Corne C, Martre C. Evaluation of the leucocyte

contamination of plateletpheresis concentrates collected

on Trima separators: a review of three years of use at

fixed sites and mobile units. Transfus Apheresis Sci 2001;25:

63-5.

10. Menichella G, Serafini R, Ciarli M, et al. A new blood

donation strategy: automated blood collection (ABC). Int J

Artif Organs 2001;24:173-7.

11. Murphy MF, Seghatchian J, Krailadsiri P, et al. Evaluation of

Cobe Trima for the collection of blood components with

particular reference to the in vitro characteristics of the red

cell and platelet concentrates and the clinical responses to

transfusion. Transfus Sci 2000;22:39-43.

12. Schooneman F, Claise C. Red cell concentrates from Trima:

quality evaluation during storage. Transfus Apheresis Sci

2001;24:197-8.

13. Soli M, Blanco L, Riggert J, et al. A multicentre evaluation of

a new filtration protocol for leucocyte depletion of high-

haematocrit red blood cells collected by an automated

blood collection system. Vox Sang 2001;81:108-12.

14. Valbonesi M, Florio G, Ruzzenenti MR, et al. Multicom-

ponent collection (MCC) with the latest hemapheresis

apparatuses. Int J Artif Organs 1999;22:511-5.

15. Valbonesi M, Luigi MC, Florio G, et al. The Cobe Trima

system as a tool for optimizing component collection: a

further step towards total apheresis. Int J Artif Organs

1999;22:58-9.

16. Valbonesi M, Bruni R, Florio G, et al. Cellular contamination

of plasma collected with various apheresis systems. Transfus

Apheresis Sci 2001;24:91-4.

17. Burgstaler EA, Pineda AA, Potter BM, Brown R. Platelet-

apheresis with a next generation blood cell separator. J Clin

Apheresis 1997;12:55-62.

18. Nightingale SD. Universal WBC reduction. Transfusion

2001;41:1306-9.

19. Ratko TA, Cummings JP, Oberman HA, et al. Evidence-based

recommendations for the use of WBC-reduced cellular

blood components. Transfusion 2001;41:1310-9.

20. Vamvakas EC, Blajchman MA. Universal WBC reduction: the

case for and against. Transfusion 2001;41:691-712.

21. Fitzpatrick JOFPJ. Residual leukocyte content of platelet-

pheresis concentrates produced on the Haemonetics MCS+:

evaluation of a method to predict high post filter white cell

counts. Transfusion 1995;35:2s.

22. Moog R, Muller N. White cell reduction during platelet-

pheresis: a comparison of three blood cell separators.

Transfusion 1999;39:572-7.

23. Benjamin RJ, Rojas P, Christmas S, et al. Plateletpheresis

efficiency: a comparison of the Spectra LRS and AMICUS

separators. Transfusion 1999;39:895-9.