Micromethod for demonstrating increased platelet surface immunoglobulin G Findings in acute chronic and human immunodeficiency virus-1-related immunologic thrombocytopenias.код для вставкиСкачать
American Journal of Hematology 34:275-282 (1990) Micromethod for Demonstrating Increased Platelet Surface Immunoglobulin G: Findings in Acute, Chronic, and Human Immunodeficiency Virus-I -Related Immunologic Thrombocytopenias Dahlia Sthoeger, Michael Nardi, Susan Travis, Margaret Karpatkin, and Simon Karpatkin Departments of Pediatrics and Medicine, New York University Medical Center, New York A method has been developed for the demonstration of increased platelet surface IgG that uses 1 ml of blood regardless of the platelet count. Platelets are gel filtered to remove plasma and contaminating lymphocytes. They are then reacted with fluorescein-conjugated antihuman IgG and analysed by flow cytometry. Percent positive staining cells vary from 10% to 80% of total cells examined. A platelet antibody index is derived from the product of percent positive staining cells X mean fluorescence intensity of positive staining cells. All patients studied with chronic idiopathic thrombocytopenia purpura (ITP) or human immunodeficiencyvirus-1 (HIV-1)-related thrombocytopenia had increased platelet surface IgG. Twelve acute children and 11chronic children had indices averaging 3.5and 8.9-fold greater than 12 normal children, respectively. Five of 12 children with acute ITP had normal platelet IgG. There was no linear correlation between the platelet antibody index and platelet count. Platelets of patients with acute, chronic, or HIV-1-related ITP displayed autofluorescence. In chronic ITP, the percentage of platelets displaying autofluorescence had a significant negative correlation with the platelet count. This technique will be a valuable diagnostic tool in the pediatric population. Key words: ITP fluorescence-activated flow cytometry, acute ITP, chronic ATP, HIV-1related ITP, autofluorescence INTRODUCTION Idiopathic thrombocytopenic purpura (ITP), also known as autoimmune thrombocytopenic purpura (ATP), is an acute or chronic disorder in which IgG is bound to the platelet membrane in most patients. The opsonized platelets are removed by macrophages of the reticuloendothelial system. Demonstration of increased amounts of IgG on the platelet may be helpful in establishing the diagnosis of ATP [ 11. Numerous techniques for measurement of platelet-bound IgG have been described [2,3]. Many of these require large volumes of blood, particularly if the platelet count is very low, and therefore are not useful in infants and young children. Here we describe a method for demonstrating increased platelet surface IgG that requires no more than 1 ml of blood regardless of the platelet count, does not utilize radioactive isotopes, and eliminates contaminating lymphocytes, which can give spurious results with low plate0 1990 Wiley-Liss, Inc. let counts. Gel-filtered platelets are exposed to fluorescein-labeled antihuman IgG and then analyzed by fluorescence-activated flow cytometry to determine the number of platelets that fluoresce and the intensity of their fluorescence. Received for publication September 22, 1989; accepted February 15, 1990. Address reprint requests to Margaret Karpatkin, MD, Department of Pediatrics, NYU Medical Center, 550 First Avenue, New York, NY 10016. Dahlia Sthoeger is now at Kaplan Hospital, Rehovath, Israel. Susan Travis is at the University College of Medicine and Dentistry, New Jersey-Robert Wood Johnson Medical School, Camden, New Jersey. 276 Technique: Sthoeger et at. Tyrode’s buffer containing 1% bovine serum albumin (BSA). A second 1 ml aliquot of gel-filtered platelets is Patient Population incubated with Tyrode’s buffer plus 1% BSA without The patient population consisted of 12 children with addition of antibody. The platelets are washed four times acute ITP (the term is used for the acute childhood dis- in Tyrode’s buffer 1% BSA, and each aliquot is then order for which the etiology is poorly understood), 11 resuspended in 50 pl of 1% paraformaldehyde. One milwith chronic ATP, and five who were thrombocytopenic liliter of isoton (Coulter Diagnostics, Hielaeh, FL) is and had antibodies against HIV- 1. The 23 ITP-ATP chil- added, and the platelets are resuspended by gentle tapdren presented with purpura, did not have lymphade- ping and then analyzed by flow cytometry (FACScan; nopathy or splenomegaly , and had negative antinuclear Becton Dickinson). antibodies and no risk factors for HIV- 1 . All but one had undergone bone marrow aspirations, which revealed nor- Fluorescence-Activated Cell Scanning (FACScan) mal to increased megakaryocytes. The disease was conThe FACScan is equipped with an argon laser operatsidered to be acute if it resolved within 6 months of ing at 488 nm. Data acquisition of 10,000 events was diagnosis (platelet count 150,000/mm3 or more on no gated on a cytogram of forward and right-angle scatter to therapy). Disease lasting longer than 6 months was coninclude platelets and exclude cellular debris (Fig. 1A). sidered chronic. In addition, 21 adults with chronic ATP Under these conditions, 98% of gated cells of healthy were studied. controls stain with 3B2, a fluorescein-labeled monoclonal antibody for platelet GPIIb . The ungated parCollection of Blood ticles contain -1% of the total positive staining cells. Blood specimens were obtained from discarded sam- Under these conditions, less than 10% of lymphocytes ples routinely employed for relevant clinical tests. are gated as determined from an analysis of lymphocytes Venous blood was drawn into 0.04 ml of 7.5% potasium separated by Ficoll-Hypaque from thrice-defibrinated EDTA (Vacutainer Systems, Becton Dickinson, Ruther- blood (less than 5% of these particles stain for GPIIb). ford, NJ) and centrifuged at lOOg for 10 min to obtain This represents 9% of total “lymphocyte” IgG (Ficollplatelet-rich plasma (PRP). Acetyl salicyclic acid was Hypaque fraction). Prior gel filtration of PRP eliminates added to the PRP to a final concentration of 0.25 mM to virtually all lymphocyte and red blood cell contaminaprevent clumping. tion. In severely thrombocytopenic patients (platelet counts < 30,00O/pl) platelet and red blood cell fragments are often present . Therefore, fewer intact Separation of Platelets From Plasma platelets are gated (Fig. 1B). Approximately 80% of the and Lymphocytes gated cells of this patient with HIV- 1 thombocytopenia Depending on the volume of PRP available, between (platelet count 30,0OO/pl) stain positively for GPIIb. The 0.1 ml and 1.0 ml is applied to a disposable 200 mm ungated “cellular debris” contains less than 2% of total column (Kontes, Vineland, NJ) of sepharose 2B in Typositive staining cells. rode’s buffer, pH 7.4. The column is washed with TyA fluorescence index or platelet antibody index is obrode’s buffer; the first 1 ml of effluent is discarded, and tained by multiplying the mean intensity of fluorescence the next 2 ml, which contains the platelets, is collected. of antibody-treated cells (x axis) by the number of plateTo verify the absence of lymphocytes in the platelet fraclets that fluoresce (y axis) beyond the gate. For each tion, PRP known to contain HIV-1 was subjected to gel patient, the value obtained for the nonantibody-incubated filtration. The platelet fraction was amplified by the platelets is subtracted from that obtained for the antipolymerase chain reaction using Taq 1 polymerase and body-treated platelets (Fig. 2). This index, rather than oligonucleotide primers for the gag gene of HIV-1. percent positive staining cells or mean intensity, was Southern blotting of six amplified samples did not reveal noted to give the largest differences and least overlap gag, whereas unfiltered samples were positive for gag between control and patient samples. Control studies 141. with irrelevant fluorescene-conjugated anti-CD4 antibody revealed no change in fluorescence before and after Incubation With Fluorescein-Conjugated Antibody addition of antibody to the platelet sample (data not One milliliter of gel-filtered platelets is centrifuged at shown). Unprocessed blood samples may be stored for 1 week 1,OOOg for 6 min at room temperature, the supernatant is after collection without change in the antibody index. discarded, and 50 p,l of 125 ng of fluorescein-conjugated After fixation with paraformaldehyde, platelets can be goat F (ab’), antihuman 1gG (Tago Inc., Burlingame, stored for a least 1 month. CA) is added to the platelet pellet for 30 min at 4” in MATERIALS AND METHODS Technique: Micromethod for Platelet Surface IgG 277 DOT PLOT OF LIGHT-SCATTER RESULTS Percent Positive Cells 4 The percent positive staining cells was assessed under instrument conditions in which less than 1% of unstained normal cells fluoresced (Fig. 2). The mean percent positive staining cells (after addition of fluorescent antibody) for 12 healthy children aged 9 months to 16 years was 5.1 +. 2.4 (S.D.). The mean percent positive cells for 20 healthy adults was 4.9 2.9 (S.D.). This was not significantly different from the case in healthy children. We therefore pooled the adult and child controls, obtaining a mean percent staining of 5.0, with a 2 S.D. upper limit of 10.4. The percent positive staining cells for chronic children averaged 35.4 -+ 20.8 S.D. (range 8.0-78.8) compared to 12.4 +- 8.9 for acute children (range 2.728.4) (Fig. 3). This 2.9-fold difference was statistically significant ( P = 0.002), possibly due to the presence of seven ITP children with normal values. HIV-I-ITP children had a value of 28.8 2 12.9. Chronic adults had a value of 24.3 -+ 17.4, which was not significally different from the value in chronic children (Fig. 3). In severely thrombocytopenic patients (< 30,000 plateletslpl) percent positive platelets may represent minimal values because of the presence of nonplatelet particles in the "platelet" gated area (see Discussion). Linear regression analysis of platelet count vs. percent positive staining cells was performed for acute and chronic children as well as chronic adults. No correlation was noted for any of the groups. r .< .' I . +_ + u cn h ssc B r 7 ' .: ... . . . Platelet Antibody Index The mean index for 12 healthy children was 217 ? 82 (S.D.), with a range of 54-259. The mean index for 20 healthy adults was 137 & 84, with a range of 23-282. This was not significantly different from that for the u healthy children. The pooled adult and child controls v3 . . . . .: gave a mean index of 166 & 9 1. This provides a + 2 Er ; . . . . ... . :. . .. ,. . . . . .., . . :._ S.D. upper limit of 348. The index obtained in each of J ... .. '.. . . $ ,.: . four children who were thrombocytopenic due to nonim. . . . , .,. . . . .. . , _ . . munologic diseases was within the normal range (Table I). The platelet antibody index in 21 adults and children, thrombocytopenic for a variety of causes, was compared with the level of platelet-associated IgG measured by our standard radioimmune assay [ 11,121. Comparison by regression analysis was 0.6, with a P value of < 0.01. Fig. 1. Dot plot of light scatter properties from a health! Nine patients were negative according to both methods control subject (A) and a patient with a platelet count o 30,0001p.1 and cellular debris (6). The area in brackets is and 12 were positive. gated for fluorescence evaluation. x axis depicts sidewards Seven out of the 12 patients with acute ITP had an light scatter, y axis depicts forward light scatter. elevated platelet antibody index and five were normal (Fig. 3). There is no apparent difference between these two groups of patients with respect to age, platelet count, and time elapsing between diagnosis and remission. In ..'I,. ,, ' I , ' . I , ~ I.,.. , ssc 278 Technique: Sthoeger et al. 3813 CONTROL K W m z 3 z J J W - 0 D PATIENT + Ab 0 1 FLUORESCENCE Fig. 2. Fluorescence-activated flow cytometry analysis of control and patient's platelets. A: Autofluorescence of control platelets. Vertical line represents the marker set at < 1% autofluorescence. 6: Fluorescence pattern after addition of fluoresceinatedgoat antihuman IgG antibody to gel-filtered platelets. C: Autofluorescence of patient's platelets. D: Fluorescence after addition of fluoresceinated goat antihuman IgG to gel-filtered platelets. two patients in whom follow-up samples were obtained after remission, the index had returned to normal. All of the children with chronic ATP had elevated indices, as did the children who were HIV-1 positive. The mean index in the children with chronic ATP, 3,093 +- 2,509 was 2.5-fold higher than in those with acute ITP, 1,228 +- 1,520 ( P < 0.04 unpaired t test; Fig. 3 ) . This is due to the presence of five ITP children with normal values. All five children with HIV-l-related ITP had elevated indices that averaged 6,910 2 5,017 (Fig. 3 ) . Their mean index was 5.6-fold greater than acute ITP ( P = 0.002) and 2.2-fold greater than chronic ATP, ( P = 0.06). Nineteen adults with chronic ATP were also studied. The platelet antibody index was elevated in all of them, with a mean value was 1,484 -+ 945 (Fig. 4). This was significantly lower than the value for childhood chronic ATP ( P < 0.02). The relationship between the platelet count and the antibody index was evaluated by linear and by exponential regression analysis in the children with acute ITP, and in the children and adults with chronic ATP. There was no correlation in any of these groups. Percent Autof luorescence The percent platelets that fluoresced without the addition of antibody (percent autofluorescence) was surprisingly greater in patients than in normal subjects: acute childhood ITP, 5.6 -+ 0.28 vs. control 0.83 +- 0.28, P < 0.09; chronic childhood ATP, 8.5 5 8.6 vs. control 0.83 -+ 0.28, P < 0.005; chronic adults ATP, 4.1 ? 4.9 vs. control 0.44 +- 0.28, P < 0.002; HIV-1-ITP, 10.6 ? 6.5 vs. control 0.83 ? 0.28, P < 0.001 (Fig. 5). The percent autof luorescence correlated inversely with the platelet count in chronic childhood ATP employing a log/log plot (r = 0.86, P < 0.001, n = lo). Autofluorescence was not observed in all four nonimmunologic thrombocytopenic patients (range 0.4-1.3%). We ex- Technique: Micromethod for Platelet Surface IgG % Ab Staining h i e Chronic Mean Intensity HIV Platelet Ab - 279 Index lormoi Acute 304 15 9 8 7 120 65 5 9w 100 0 4 0 0 0 8 40 * Fig. 3. Scattergram of fluorescent antibody-treated platelets of children with immunologic platelet disorders: Percent positive fluorescent platelets, mean intensity of fluorescent platelets, and mean platelet antibody index (percent Platelet countlkl Positive cells (%) Platelet Ab index 1 fluorescent cells x mean intensity of fluorescent cells) for controls, acute childhood ITP, chronic childhood ATP, and childhood HIV-1-related ITP. TABLE 1. Percent Positive Cells and Platelet Antibody Index in Children With Nonimmunologic - Thrombocytopenia . Acute myeloblastic leukemia in remission, on chemotherapy (12 years, M) May-Heggelin anomaly (2 weeks, F) Post bone marrow transplant for acute lymphocytic leukemia (10 years, M) Anorexia nervosa (15 years, M) Control children (n = 12) 2 0 20 Diagnosis (age, sex) 3 0 . Chronic Normol 180 Chronic 0 'latelet Ah Index Normol 1 9 8 74,000 8.6 29 1 40,000 11.5" 435a 7 0 100 5.0 114 115,000 10.0 305 > 150,000 < 10.4 < 381 3 59, 80 30,000 6 m 4 0 60 40 LU 8 0 8 0 .-- -__ _. + 3 2 1 "The normal values for children under 6 months have not been established. plored the possibility that 1% paraformaldehyde fixation or IgG bound to the membrane could increase autofluorescence. Stained normal platelets, with or without paraformaldehyde fixation, demonstrated no autofluorescence (data not shown). Neither did PLAl-positive platelets incubated with polyclonal anti-PLA1 antibody, washed, and then analyzed with FACScan. No increase Fig. 4. Scattergram of fluorescent antibody-treated platelets of adults with chronic autoimmune thrombocytopenic purpura: Percent positive fluorescent platelets, mean intensity of fluorescent platelets, and mean platelet antibody index. in percent autofluorescence occurred, under conditions where the second antibody-stained sample fluoresced intensely (data not shown). The intensity of autofluores- Technique: Sthoeger et al. 280 36 F - Normal dormol Child Adult Acute Child - :hronic Child :hronic Adult HIV Child 0 0 0 8 b c ’“i 4 0 0 b 0 0 0 8 B 0 0 0 0 8 -4-- Fig. 5. Scattergram of percent positive autofluorescent platelets of nonfluorescent antibody-treated platelets of children and adults with immunologic platelet disorders: Acute ITP, chronic childhood ATP, chronic adult ATP, HIV1-related ITP of childhood. cence was no greater in any of the patients with ITP compared to controls. DISCUSSION We have developed a method for demonstrating increased IgG on the platelet surface using no more than 1 ml of whole blood, with a platelet count as low as lO,OOO/pl. Studies of acute childhood ITP have been hampered by the need to draw unacceptably large volumes of blood from small children in order to obtain a sufficient number of platelets. Results of the test correlate well (r = 0.6, P < 0.01) but not perfectly with those obtained in the same patients using an established radioimmunoassay that utilizes much larger volumes of blood [ 11,121. The lack of a perfect correlation may be because the radioimmunoassay measures released IgG [ 13,141 as well as surface IgG, whereas fluorescence flow cytometry measures cell surface IgG. Other groups employing washed platelets have developed similar methods demonstrating the usefulness of flow cytometry for the measurement of antiplatelet antibody [7-lo]. Our method differs from the others in the use of gel-filtered platelets, which eliminates lymphocytes and lymphocyte fragments and the use of a platelet antibody index that evaluates both percent positive platelets as well as fluorescence intensity of the positive platelets. This antibody index was found to be more useful than percent positive platelets alone in providing the largest differences and least overlap between control and patient samples. Our study provides new information on the utilization of flow cytometry to evaluate and compare platelet antibody of acute, chronic, and HIV-1 immunologic thrombocytopenic children as well as chronic ATP adults and on the presence of autofluorescence in unstained platelet samples of acute, chronic, or HIV-1 ITP patients. The platelet antibody index was compared in children with acute ITP and chronic ATP. The mean value was higher in those with chronic ATP, probably due to the presence of normal values in five of the ITP children (see below). This is in contrast to two other pediatric studies that have reported the level of platelet-associated IgG to be higher in acute ITP at time of presentation than in chronic ATP [15,16]. It has been suggested that this finding may have been an artefact due to the fact that platelet counts in acute ITP at onset are frequently lower than in chronic, as was the case in one of these studies . Because the subjects were young children, a smaller volume of blood than usual was drawn, providing too few platelets, perhaps resulting in a spuriously low total platelet count. In addition, it is likely that a proportionately higher number of lymphocytes would have been present in the PRP that was tested. The IgG on these lymphocytes may have resulted in an artefactually high reading of IgG in the specimen. In the studies reported herein, there was no lymphocyte contamination and no relationship between the patients’ antibody index and the platelet counts, indicating that no such artefact was present, Ware et al.  have suggested that the higher level they noted in acute ITP was present only in those children tested at the onset of disease, before therapy. However two of our five acute ITP patients with normal levels of platelet IgG were tested before therapy. All of the children with chronic ATP had increased antiplatelet antibody indices, as did the adults. In five of 12 patients with acute ITP, the index was within the normal range. No difference between these five patients and the others could be determined with respect to platelet count, age, sex, or duration of disease. Cheung et a1 [ 171 also noted normal levels of platelet IgG in four of 26 acute ITP patients, with elevated values in all 22 pediatric chronic ATP patients studied. One can speculate that the clinical entity of acute ITP of childhood has a number of etiologies that do not all result in the presence of increased IgG on the platelet surface. These may be the children who do not develop chronic ATP. An alternative explanation could be that some patients have IgM exclusively bound to their platelet membrane, which would not have been detected with the second antibody employed in this test. In the two patients who had raised indices and were available for follow-up in remission, the index had returned to the normal range. When patients with ITP or chronic ATP were com- Technique: Micromethod for Platelet Surface IgG pared to controls, there were a significantly larger number of platelets displaying autofluorescence; i.e., some platelets fluoresced without exposure to antibody. In addition, there was a significant inverse relationship between platelet count and percent autofluorescence in chronic childhood ATP. The reason for this is not clear. It does not appear to be due to antibody binding to the surface of the platelets as judged from experiments with anti PLAl antibody, or is it due to the fixative employed. Of particular interest was the observation of the heterogeneous distribution of antibody in the patients studied, in that only a fraction of the patient’s platelets examined had antibody detectable with this method. This varied from 10% to 80% positive staining cells in children as well as adults, with a mean of approximately 25% positive staining cells for all four groups studied. Similar results have been reported by others [8,10]. It is conceivable that some of the negative-staining cells represent IgM or complement bound to platelets since their presence was not examined in this study. However, Ault [lo], who examined the interrelationship between platelet IgG, IgM, and C3 by flow cytometry, found strong correlations between the presence of IgG and IgM on the platelet surface, and no cases of C3 alone on the platelet surface, ruling out this possibility. Another possibility could be heterogeneity of antigen density or incomplete saturation by circulating antiplatelet antibody. A third possibility could be the presence of nonplatelets in the gated area. This appears to be the case in patients with severe thrombocytopenia, as was reported by Ault [lo], who noted nonplatelet fragments within the gated platelet region. We have confirmed this observation in three severely thrombocytopenic ATP patients with platelet counts of 6,0OO/p1, lO,OOO/pl, and 25,OOO/ul. Utilizing our monoclonal antibody against platelet membrane GPIIb, these three patients had 33%, 12%, and 25% GPIIb staining particles, respectively, in the gated area. However, we cannot rule out the possibility that the antiGPIIb monoclonal antibody employed to detect “true” platelets was not blocked by IgG bound to the GPIIbGPIIIa epitope [ 5 ] . This method will be of value in studying thombocytopenia in sick premature infants. Low platelet counts occur in up to 20% of this group of patients. Some studies have suggested that increased platelet destruction may pay a role [ 18-201. In one of these studies, elevated PAIgG was demonstrated in some patients . The level of platelet surface IgG in healthy newborn infants has not been established. The method described here will allow the determination of a normal range in nonthrombocytopenic infants of different gestational ages, allowing a more meaningful evaluation of thrombocytopenic infants. In conclusion, a method has been developed for dern- 281 onstrating increased PAIgG that has the following advantages. 1) It requires a small volume of blood. 2) It measures surface platelet IgG. The platelets are not manipulated by washing and multiple centrifugations, with consequent activation and release of granule IgG on their surface. 3) There is no lymphocyte contamination that could spuriously affect low platelet counts. 4)It does not require radioisotopes. 5) The test is less labor intensive than most other procedures, requiring 2.5 hr of working time compared to 7 hr for the radioimmunoassay or ELISA. ACKNOWLEDGMENTS This study was supported by a grant from the National Heart, Lung and Blood Institute (HL-13336-18). REFERENCES I . Karpatkin S : Autoimmune thrombocytopenic purpura. Semin Hematol 22:289-302, 1985. 2. Kelton JG, Denomme G, Lucarelli A, Garvey J, Powers P, Carter C: Comparison of the measurement of surface or total platelet associated IgG in the diagnosis of immune thrombocytopenia. Am J Haematol 18:l-5, 1985. 3. Schwartz K: Platelet antibody: Review of detection methods. Am J Haematol 29:106, 1988. 4. Karpatkin S, Nardi M, Lennette ET, Byrne B, Poiesz B: Antihuman immunodeficiency virus type 1 antibody complexes on platelets of seropositive thrombocytopenic homosexuals and narcotic addicts. Proc Natl Acad Sci USA 85:9763-9767, 1988. 5 . Varon D, Karpatkin S: A monoclonal anti-platelet antibody with decreased reactivity for autoimmune thrombocytopenic platelets. Proc Natl Acad Sci USA 80:6992, 1983. 6. Zucker-Franklin D, Karpatkin S: Red cell and platelet fragmentation in idiopathic autoimmune thrombocytopenic purpura. N Engl J Med 297517, 1979. 7. Lazarchick J , Hall SA: Platelet-associated IgG assay using flow cytometric analysis. J Immunol Methods 87:257, 1986. 8 Rosenfeld CS, Nichols G, Bodensteiner DC: Flow cytometric measurement of antiplatelet antibodies. J Clin Pathol 87518, 1987. 9 Heim MC, Peterson BH: Detection of platelet-associated immunoglobulin in immune thrombocytopenia by flow cytometry. Diagn Clin Immunol 5:309, 1988. 10 Ault KA: Flow cytometric measurement of platelet-associated immunoglobulin. Pathol Immunopathol Res 7:395, 1988. 11 Faig D, Karpatkin S: Cumulative experience with a simplified solidphase radioimmunoassay for the detection of bound antiplatelet IgG, serum auto-, allo-, and drug dependent antibodies. Blood 60:807813, 1982. 12 Walsh CM, Nardi MA, Karpatkin S: On the mechanism of thrombocytopenic purpura in sexually active homosexual men. N Engl J Med 31 1:635-639, 1984. 13. Pfueller SL, David R: Liberation of surface and internal plateletassociated IgG (PA-IgG) during platelet activation. Br J Haematol 63:785-794, 1986. 14. George JN, Saucerman S: Platelet IgG, IgA, IgM and albumin: Correlation of platelet and plasma concentrations in normal subjects and in platients with ITP or dysproteinemia. Blood 72:363, 1988. 282 Technique: Sthoeger et al. 15. Lightsey AL, Koenig HM, McMillan R , Stone JR: Platelet-associated immunoglobulin G in childhood idiopathic thrombocytopenic purpura. J Pediatr 94:201, 1979. 16. Ware R , Kinney TR, Friedman HS, Falletta JM, Rosse WF: Prognostic implications for direct platelet associated IgG in childhood idiopathic thrombocytopenic purpura. Am J Pediatr Hematol Oncol 8:32, 1986. 17. Cheung NV, Hilgartner MW, Schulman 1, McFall P, Glader BE: Platelet associated immunoglobulin G in childhood idiopathic thrombocytopenic purpura.J Pediatr 102:366, 1983. 18. Mehta P, Vasa R , Naoman L, Karpatkin M: Thromhocytopenia in the high risk infant. J Pediatr 97:791, 1980. 19. Castle V, Andrew M , Kelton J, Carter C, Giron D: Frequency and mechanisms of neonatal thrombocytopenia. J Pediatr 108:749, 1986. 20. Castle V, Coates G, Kelton JG, Andrew M: '"In-oxine platelet survivals in thrombocytopenic infants. Blood 70:652, 1987.