LETTERS TO THE EDITOR

Complications of plasma exchange in after percutaneous insertion of a subclavian central ve- thrombotic thrombocytopenic nous catheter from pneumothorax and hemorrhage. Two purpura-hemolytic uremic syndrome: patients suffered cardiac arrest with pulseless electrical a study of 78 additional patients activity: one from an anaphylactic reaction to plasma and The frequency of patients treated with plasma exchange the other from pericardial hemorrhage and tamponade, (PE) for thrombotic thrombocytopenic purpura- presumably due to cardiac perforation by an internal hemolytic uremic syndrome (TTP-HUS) increased seven- jugular catheter insertion guidewire. fold from 1981 to 1997.1 Therefore, the morbidity and Other major catheter-related complications included mortality due to PE is an increasingly important consid- one patient with a retroperitoneal hemorrhage following eration in management decisions for patients with clini- femoral catheter insertion and seven patients with cath- cally suspected TTP-HUS. Some studies have described eter thrombosis that prevented PE and/or required place- few complications associated with PE,2 but our previous ment of a new central venous catheter; two of these seven report on 71 consecutive patients with clinically sus- patients had catheter-related venous thrombosis requir- pected TTP-HUS treated with PE from 1996 to 1999 dem- ing systemic anticoagulation. Ten patients developed sys- onstrated a major complication rate of 30 percent, in- temic infection: eight had cultures positive for the cluding two deaths.3 This report describes our experience presence of bacteria (Staphylococcus aureus [five], Staph- during the subsequent 3 years, 1999 to 2002, with 78 con- ylococcus epidermidis [three]); the two patients with secutive patients. negative blood cultures were treated with parenteral an- From June 25, 1999, to June 25, 2002, a total of 81 tibiotics for presumed sepsis. consecutive patients were referred to the Oklahoma Other major plasma-related complications included Blood Institute for PE treatment of their first episode of in two patients requiring dopamine, acute clinically suspected TTP-HUS. Three patients were ex- hypoxia in two patients, serum sickness in one patient cluded because they died before a central venous cath- requiring systemic glucocorticoids, and severe vomiting eter for PE was inserted. in one patient that prevented completion of PE. Twenty-one of 78 patients (27 percent) had 27 major Twenty-seven (35 percent) patients developed minor complications (Table 1). One patient died immediately complications, including 10 patients who also had major complications. The majority of the minor complications were urticaria (22 patients); other minor complications TABLE 1. Major complications of PE in 78 included vomiting, tetany, and hypotension responding consecutive patients treated for clinically to intravenous fluids. suspected TTP-HUS* These data confirm and extend our previous report.3 Number Complications of patients In summary, over 6 years we have observed 54 major Catheter-related complications complications, including 3 deaths, related to PE in 42 of Pulmonary hemorrhage and pneumothorax 149 (28 percent) consecutive patients treated for clini- resulting in death 1 Cardiac tamponade with arrest 1 cally suspected TTP-HUS. These observations are essen- Retroperitoneal hemorrhage 1 tial to understand the risks of PE when evaluating the Thrombosis 7 management of patients who may have TTP-HUS. Catheter obstruction† 5 Venous thrombosis systemic anticoagulation 2 J.R. McMinn, Jr., MD Systemic infection 10 Ira A. Thomas, BS Documented bacteremia 8 Deirdra R. Terrell, MPH Suspected bacteremia‡ 2 Deanna Duvall, BSN Plasma-related complications Sara K. Vesely, PhD with cardiac arrest 1 Hypotension requiring dopamine 2 James N. George, MD Serum sickness 1 -Oncology Section Hypoxia§ 2 The University of Oklahoma Health Sciences Center Vomiting࿣ 1 PO Box 26901 * The classification of complications as central venous cath- eter-related or plasma-related and the definition of major Oklahoma City, OK 73190 complications has been previously described.3 e-mail: [email protected]. † Required placement of a new catheter and/or prevented PE. ‡ Negative blood cultures but systemic symptoms present and treated with a full course of parenteral antibiotics for pre- REFERENCES sumed sepsis. 1. Clark WF, Rock GA, Buskard N, et al. Therapeutic plasma § Required treatment with glucocorticoids and/or stopping PE. ࿣ Required stopping PE. exchange: an update from the Canadian Group. Ann Intern Med 1999;131:453-62.

Volume 43, March 2003 TRANSFUSION 415 LETTERS TO THE EDITOR

2. McLeod BC, Sniecinski I, Ciavarella D, et al. Frequency of immediate adverse effects associated with therapeutic TABLE 1. Flow cytometric data of fluorescence of RBCs labeled with a panel of blood apheresis. Transfusion 1999;39:282-8. group 3. Rizvi MA, Vesely SK, George JN, et al. Complications of Median fluorescence plasma exchange in 71 consecutive patients treated for Clone (MoAb) Control RBCs Patient’s

clinically suspected thrombotic thrombocytopenic specificity or serum (O R1R2 or O rr) RBCs purpura-hemolytic uremic syndrome. Transfusion 2000; Negative control 4 5 D FITC-BRAD-3*† 191 167 40:896-901. D BRAD-5*† 743 661 C MS257* 57 36 C HMR7† 118 109 Quantitation and phenotyping of fetal RBCs c MS37* 29 19 c MS51* 85 93 in maternal blood by flow cytometry E HIRO-18* 79 76 Flow cytometry may be used to detect and quantitate E HIRO-91† 34 32 fetal D+ RBCs in fetomaternal hemorrhage (FMH) to cal- E 907* 41 51 e MS70* 4 5 culate the dose of Rho (D) immune globulin required to K Serum 43 54 and 4 prevent maternal alloimmunization and protect against k Serum 12 27 a subsequent HDN. Fetal RBCs may be detected with an- Fy Serum 33 50 Fyb Serum 19 44 tibodies to fetal Hb (anti-HbF) or to the D (anti- Jka Serum 7 12 D), either by indirect labeling or, more recently, by using Jkb Serum 12 5 directly conjugated MoAbs.1,2 Antibodies to additional * The MoAbs were submitted to the Third International Work- RBC blood group have been used to phenotype shop on Monoclonal Antibodies Against Human Red Blood Cells and Related Antigens held in 1996. 3 RBCs after transfusion and to monitor chimerism after † The MoAbs were submitted to the Fourth International Work- marrow transplantation.4,5 shop on Monoclonal Antibodies Against Human Red Blood We report the case of a woman whose fetus died in Cells and Related Antigens held in 2001. utero at 35 weeks of gestation following trauma. The woman’s RBCs were Group A, D+, and the antibody screen was negative. The Kleihauer-Betke test demon- Using adult RBCs alone or with added cord RBCs, a strated a 50-mL FMH. Cord blood was not available for marker was set to encompass strongly fluorescent fetal phenotyping. It was therefore not known whether she RBCs on histograms of FITC-anti-HbF–stained RBCs. The could be immunized to antigens (other than D) expressed percentage of events under this marker in the adult on fetal RBCs. Fetal and maternal RBCs were phenotyped sample (0.09%) was subtracted from that of the patient’s by flow cytometry, the large FMH was confirmed, and an sample (2.59%), that is, 2.5 percent (Fig. 1A). Using the antigen mismatch was identified. formula FMH (mL of packed fetal RBCs) = percentage of the FMH in the patient’s sample 22,6 ן RBCs from the patient, two adult donors (controls, positive events

phenotypes O R1R2 and O rr K+ k+ Fy(a+b+) JK(a+b+)), was calculated to be 55 mL. and cord blood were washed, and a mixture of cord and The median fluorescence of RBCs obtained with the adult RBCs was made; 20-␮L aliquots of 5 percent sus- blood group-specific antibodies is listed in Table 1. O ␮ pensions were labeled with 10 L of FITC-conjugated R1R2 RBCs were tested with the MoAbs and O rr RBCs anti-HbF (Silenus, Boronia, Victoria, Australia), 5 ␮Lof with the antisera; the patient’s RBCs were tested with all FITC-conjugated anti-D (BRAD-3),2 200 ␮L of MoAb, or antibodies. All samples, except three, exhibited fluores- 50 ␮L of (Table 1). RBCs were incubated either cence, as the patient’s RBCs were negative with anti-Jkb at room temperature for 30 minutes (anti-HbF) or at 37ЊC and anti-e did not show binding. For most antibodies, the for 1 hour (blood group antibodies). Before labeling with median fluorescence of RBCs was similar between the

anti-HbF, RBCs were fixed (in 1 mL of 0.05 percent glu- controls exhibiting a single dose of antigen (O R1R2 and O taraldehyde in PBS), washed, permeabilized (in 0.5 mL of rr) and the patient, indicating that similar amounts of 0.1% Triton X-100 in PBS with 0.1% BSA), and washed.1 antibodies were bound. Only one sample showed two Unlabeled primary antibodies were detected with 100 ␮L populations; this occurred when the patient’s RBCs were of 1-in-100 FITC-conjugated anti-human IgG (F(ab); incubated with anti-K. The major population (97.7%) was Jackson ImmunoResearch, West Grove, PA). Samples K+ (median fluorescence, 54), but a minor population of were analyzed with a flow cytometer (FACSCalibur, Bec- 2.2 percent of events was K– (median fluorescence, 4) ton Dickinson, Oxford, UK),6 collecting 50,000 events and (Fig. 1B). Subtracting the background of unlabeled events placing markers as appropriate on histograms of log- under this marker (0.2%) gave 2 percent K– cells or a integrated FL1 fluorescence. The percentage of gated 44-mL FMH. events and the median fluorescence of events under the Some examples of anti-C, anti-c, and anti-E gave markers were computed. greater fluorescence than others (Table 1). It was ob-

416 TRANSFUSION Volume 43, March 2003 LETTERS TO THE EDITOR

binding fluorescent microspheres10 (sensitive but te- dious). Reports of flow cytometric detection and quanti- tation of minor cell populations when using one to three antibodies have been extensively reviewed.11 The case study detailed here demonstrates the utility and rapidity of flow cytometry, both for quantitation of FMH and for extensive phenotyping of mixed RBC populations.

REFERENCES 1. Davis BH, Olsen S, Bigelow NC, Chen JC. Detection of fetal red cells in fetomaternal hemorrhage using a fetal hemoglobin monoclonal antibody by flow cytometry. Transfusion 1998;38:749-56. 2. Lloyd-Evans P, Kumpel BM, Bromelow I, et al. Use of a directly conjugated monoclonal anti-D (BRAD-3) for quantification of fetomaternal hemorrhage by flow cy- tometry. Transfusion 1996;36:432-7. 3. Griffin GD, Lippert LE, Dow NS, et al. A flow cytometric method for phenotyping recipient red cells following transfusion. Transfusion 1994;34:233-7. 4. Hendriks ECM, de Man AJM, van Berkel YCM, et al. Flow cytometric method for the routine follow-up of red cell populations after bone marrow transplantation. Br J Hae- matol 1997;97:141-5. 5. David B, Bernard D, Navenot JM, et al. Flow cytometric monitoring of chimerism after bone mar- row transplantation. Transfus Med 1999;9:209-17. Fig. 1. Histograms of events stained with anti-HbF (A) and 6. Kumpel BM. Analysis of factors affecting quantification of anti-K (B). Data for the patient’s RBCs are in gray (A and B); fetomaternal hemorrhage by flow cytometry. Transfusion adult RBCs are outlined in A and K+ RBCs are outlined in B. 2000;40:1376-83. The sample from the patient comprised 2.5% of RBCs that 7. Le Pennec P-Y, Gane P, Noizat-Pirenne F, et al. Flow cy- were strongly fluorescent with anti-HbF (A, events under the tometric evaluation of non anti-Rh1 (D) monoclonal Rh marker M1) and 2.0% of RBC that were unlabeled with antibodies. Transfus Clin Biol 1996;6:433-5. anti-K (B, events under the marker M1). Most of the pa- 8. Petx LD, Yam P, Wallace RB, et al. Mixed hematopoietic tient’s RBCs, and the control K+ RBCs, bound anti-K chimerism following bone marrow transplantation for he- (B, events under the marker M2). matologic malignancies. Blood 1987;70:1331-7. 9. Zago-Novaretti MC, Dulley FL, Dorlhiac PE, Chamone served previously that some IgG non-anti-D MoAbs gave DAF. Use of the gel test to detect mixed red blood cell relatively low fluorescence in flow cytometry, compared populations in bone marrow transplantation patients. Vox to agglutination.7 IgM antibodies against Rh C, c, E, and Sang 1993;65:161-2. e may be more effective.5 10. De Man AJM, Foolen WJG, van Dijk BA, et al. A fluores- On the basis of the median fluorescence values, the cent microsphere method for the investigation of erythro- most likely phenotypes of maternal and fetal RBCs in the cyte chimerism after allogenic bone marrow transplanta- patient’s sample were maternal, C+ , c+ , D+ , E+ , K+ , k+, tion using antigenic differences. Vox Sang 1988;55:37-41. Fy(a+b+), Jk(a+ b–); and fetal, C+ , c+ , D+ , E+ , K–, k+, 11. Garratty G, Arndt P. Applications of flow cytofluorometry Fy(a+b+), Jk(a+b–). The only difference detected between to transfusion science. Transfusion 1995;35:157-78. maternal and fetal RBCs was that fetal RBCs were K–. Belinda M. Kumpel, BSc, PhD Thus, using this panel of antibodies, there were no blood International Blood Group Reference Laboratory group antigens detected on fetal RBCs that could immu- Southmead Road nize the mother. Bristol BS10 5ND, UK Flow cytometry can reliably detect minor RBC popu- e-mail: [email protected] lations of 1 percent or less, depending on the antibody Ann P. MacDonald, BSc, FIBMS used for identification, and has advantages over other North of Scotland Service techniques such as differential agglutination8,9 (rapid but Raigmore Hospital less sensitive) and microscopic enumeration of RBC- Inverness IV2 3UJ, UK

Volume 43, March 2003 TRANSFUSION 417 LETTERS TO THE EDITOR

Ten-year survival of transfusion recipients cent for recipients of Յ10 RBC units and 31.0 percent for identified by hepatitis C lookback recipients of more than 10 RBC units. Therefore, if the NYUMC 10-year probabilities of survival were applied to Empirical data from population-based studies on the a transfused community population (with 10% of the long-term (Ն10-year posttransfusion) probability of sur- members of that population receiving more than 10 RBC vival of transfusion recipients are needed for public units in the course of 1 year), a 10-year survival of 40 policy decisions,1 but no such data currently exist for percent would be calculated for that community popula- patients transfused after 1981. The 10-year probabilities tion. of survival of Olmsted County, Minnesota residents For patients transfused in 1993 in two Swedish coun- transfused in 19812 were often used previously, but these ties, Tynell et al.6 reported 1-year probabilities of survival figures most likely do not reflect the survival of patients that were 13 percent lower than those reported from transfused in the 1990s. Five- to 10-year posttransfusion Olmsted County,2 and 40-month probabilities of survival probabilities of survival that range from 263 to 41 per- that were 20 percent lower than the Olmsted County fig- cent4 can be inferred from HCV lookback studies.1 These ures. For patients transfused in 1994 in northern England, studies were not designed to calculate posttransfusion Wells et al.7 reported 2- and 5-year probabilities of sur- survival, however, and it is difficult to infer the length of vival that were, respectively, 15 percent and 22 percent recipient follow-up from the presented data.1,5 The New lower than the Olmsted County figures.2 Thus, when data York University Medical Center (NYUMC) HCV lookback on posttransfusion survival are needed as input for future study was designed to calculate posttransfusion survival, studies, posttransfusion survival probabilities of 66,6 60,7 and it reported 4-year posttransfusion probabilities of 47,7 and 40 percent, respectively, at 1, 2, 5, and 10 years survival for patients transfused in 1988-96.5 Of 695 pa- posttransfusion can be used until definitive empirical tients from that study,5 227 had received transfusion(s) in data become available. A population-based study that in- 1988-90 and had thus been followed until death or for 120 cludes several US counties has to be conducted to gen- months by the time of the study. Table 1 shows the 10- erate the information needed for public policy decisions year survival of these subjects. in the future. The 26.0 percent overall survival is 46.2 percent Eleftherios C. Vamvakas, MD, PhD lower than the 48.3 percent survival calculated for Olm- Medical, Scientific, and Research Affairs sted County residents transfused in 1981.2 Of the NYUMC Canadian Blood Services patients, 63.4 percent (Table 1) had received more than 1800 Alta Vista Drive 10 RBC units in the course of 1 year, as compared to 8.6 Ottawa, ON K1G 4J5, Canada percent of the Olmsted County residents.2 Even in the e-mail: [email protected] 1990s, probably less than 10 percent of the members of an unselected, transfused, community population re- ceive more than 10 RBC units in the course of 1 year. REFERENCES When the NYUMC (Table 1) and Olmsted County2 study 1. Vamvakas E. Evidence-Based Practice of Transfusion Medi- populations were stratified by transfusion dose, the re- cine. Bethesda: AABB Press, 2001. duction in survival between the two studies was 14.7 per- 2. Vamvakas E, Taswell HF. Long-term survival after blood transfusion. Transfusion 1994;34:471-7. 3. Long A, Spurll G, Demers H, et al. Targeted hepatitis C TABLE 1. Ten-year posttransfusion probabilities lookback: Que´bec, Canada. Transfusion 1999;39:194-200. of survival 4. Smith BC, Chapman CE, Burt AD, et al. Outcome of post- Sample Survival transfusion hepatitis C: Disease severity in blood- size (%) component recipients and their implicated donors. All patients 227 26.0 Men 140 26.4 Q J Med 1997;90:587-92. Women 87 25.3 5. Vamvakas E, Goldstein R. Four-year survival of transfusion Young (under 41) 39 46.2 recipients identified by hepatitis C lookback. Transfusion Middle-aged (41–65) 73 35.6 Senior (over 65) 115 13.0 2002;42:691-7. Medical* 107 15.9 6. Tynell E, Norda R, Shanell A, et al. Long-term survival in Surgical* 120 35.0 transfusion recipients in Sweden, 1993. Transfusion 2001; Recipients of Յ10 RBC units 83 43.4 Recipients of >10 RBC units 144 16.0 41:251-5. 7. Wells AW, Chapman CE, Stainsby D, et al. Long-term sur- * Patients transfused for treatment of a “medical” or “surgical” disease.2,5,6 vival after transfusion in the North of England: a popula- tion-based study. Vox Sang 2002;83 (Suppl. 2):219.

418 TRANSFUSION Volume 43, March 2003 LETTERS TO THE EDITOR

SUBMISSION OF LETTERS

Instructions for submission of letters can be found in the Detailed Instructions for Authors published on pages 128 to 133 of the January issue. Submit letters to: S. Gerald Sandler, MD Department of Laboratory Medicine/M-1306 Georgetown University Medical Center 3800 Reservoir Road, NW, Washington, DC 20007 fax (202)784-2440 e-mail: [email protected] EDITOR’S NOTE: To permit timely publication of correspondence, the references have not been verified as they are for articles appearing in TRANSFUSION, and, therefore, the accuracy of references cited in Letters to the Editor is the sole responsibility of the authors. Payment is not required for submission of Letters to the Editor.

Volume 43, March 2003 TRANSFUSION 419