Antibodies to Platelet Factor 4–Heparin After Cardiopulmonary Bypass in Patients Anticoagulated With Unfractionated Heparin or a Low-Molecular-Weight Heparin
Clinical Implications for Heparin-Induced Thrombocytopenia
Background—Cardiopulmonary bypass (CPB) induces platelet activation with release of platelet factor 4 (PF4), and patients are exposed to high doses of heparin (H). We investigated whether this contributes to the development of antibodies to H-PF4 and heparin-induced thrombocytopenia (HIT).
Methods and Results—CPB was performed with unfractionated heparin (UFH) in 328 patients. After surgery, patients received UFH (calcium heparin, 200 IU · kg−1 · d−1) (group 1, n=157) or low-molecular-weight heparin (LMWH, Dalteparin, 5000 IU once daily) (group 2, n=171). Eight days after surgery, antibodies to H-PF4 were present in 83 patients (25.3%), 46 in group 1 and 37 in group 2 (P=0.12). Most patients (61%) had IgG1 to H-PF4, but only 8 samples with antibodies induced platelet activation with positive results on serotonin release assay. HIT occurred in 6 patients in group 1, but no thrombocytopenia was observed in subjects receiving LMWH, although 2 had high levels of antibodies with positive serotonin release assay results. When antibodies to H-PF4 were present, mean platelet counts were lower only in patients with FcγRIIA R/R131 platelets.
Conclusions—These results provide evidence that the development of antibodies to H-PF4 after CPB performed with UFH is not influenced by the postoperative heparin treatment. The antibodies associated with high risk of HIT are mainly IgG1, which is present at high titers in the plasma of patients continuously treated with UFH.
Type II heparin–induced thrombocytopenia (HIT) is a severe complication of heparin therapy, often associated with IgG antibodies that bind to FcγRIIA receptors and induce platelet activation in the presence of heparin.1 2 The antigen targets that bind antibodies to the platelet surface are macromolecular complexes of heparin and platelet factor 4 (H-PF4),3 4 5 a C-X-C chemokine present in α-granules. Cardiac surgery with cardiopulmonary bypass (CPB) induces major platelet activation with release of granule contents,6 including PF4, and most patients have been exposed to heparin before CPB. These circumstances in patients exposed to high doses of heparin could increase the incidence of heparin-dependent antibodies. Bauer et al7 recently showed that 50% of patients undergoing CPB had antibodies to H-PF4 on postoperative day 5 after short treatment with unfractionated heparin (UFH).
To study the pathogenicity of antibodies to H-PF4 after CPB and to evaluate the influence of the treatment received in the postoperative period (UFH or low-molecular-weight heparin, LMWH), we studied 328 patients undergoing open-heart surgery. Antibodies to H-PF4 were assayed with ELISA, platelet activation tests (serotonin release assay, SRA) were performed with positive samples, and isotypes with subclasses of antibodies were determined. We also examined whether the evolution of platelet counts varied according to the presence or absence of heparin-dependent antibodies.
Finally, because the FcγRIIA H/H131 genotype was associated with a predisposition to HIT,8 9 we also studied FcγRIIA polymorphism in this prospective cohort of consecutive patients with and without antibodies to H-PF4 complexes.
Patients and Anticoagulation Protocol
A total of 328 patients (94 women and 234 men) 16 to 87 years old underwent open-heart surgery with CPB. All patients were enrolled during the 15 months from May 1996 to June 1997, and informed written consent was systematically obtained according to the principles of the declaration of Helsinki. Of these patients, 156 had CABG and 137 had valve replacement or repair, with CABG in 19. Sixteen other patients underwent CPB for another indication (Table 1⇓).
All patients had previously been exposed to UFH during cardiac catheterization 1 to 3 months before surgery. Anticoagulation for CPB was achieved with sodium UFH (Léo), with an initial intravenous (IV) bolus of UFH, corresponding to 250 IU/kg before aortic cannulation. Additional heparin was then administered with a continuous IV infusion to maintain the activated clotting time (Hemochron) over >600 seconds and plasma heparin levels between 2 and 4 IU/mL. On completion of CPB, the anticoagulant effect of heparin was reversed by slow IV infusion of protamine sulfate until the activated clotting time returned to the preheparinization level.
Patients were then divided into 2 groups according to the type of surgery and the individual assessment of risk factors for thromboembolism. Most patients in group 1 had a valve replacement (n=148), and a few exhibited cardiac failure, arrhythmia, myocardial infarction, and/or a previous history of thrombosis (n=9). They received continuous IV infusion of sodium UFH at a dose of 120 IU · kg−1 · d−1 over 24 hours, followed by subcutaneous calcium heparin (Calciparine, Sanofi) at a dose of 200 IU · kg−1 · d−1 for at least 10 days and then adjusted to maintain the activated partial thromboplastin time in the range of 1.5 to 2.5 times the control range.
The patients in group 2 (n=171) had an estimated lower risk of thromboembolism because most of them had isolated ischemic heart disease and underwent CABG (n=150). A few other patients had surgery for patent foramen ovale (n=13) or mitral valve repair (n=8). They all received LMWH (dalteparin or Fragmine, Pharmacia and Upjohn) with 1 subcutaneous injection of 5000 IU anti-Xa per day for 1 month.
The mean duration of CPB for patients in group 1 was significantly longer than for those in group 2 (Table 1⇑, P<0.0001).
Blood Samples and Platelet Counts
Whole blood was collected on EDTA for platelet counts or on 0.129 mol/L sodium citrate (9:1) for the other biological assays, including DNA analysis. Platelet-poor plasma was isolated by centrifugation of blood samples at 2500g for 15 minutes and stored at −80°C for no longer than 18 months until assay.
Platelet counts were obtained with an automatic counter (Coulter STKS, Coultronics) before surgery, during CPB, and at least 3 times in the postoperative period (on days 3 to 4, 5 to 7, and 8 to 10).
Blood samples for the detection of antibodies to H-PF4 complexes by ELISA were drawn before CPB and twice in the postoperative period (on days 3 to 5 and 7 to 10).
ELISA for Detection of Antibodies to H-PF4 Complexes
H-PF4 ELISA was performed with the Asserachrom HPIA kit (Diagnostica Stago).4 10 For the overall assay measuring IgG, IgA, and IgM isotypes, the result was defined as negative if absorbance at 492 nm (A492) was <0.5 or as positive if A492 was ≥0.5.
In 80 patients with antibodies to H-PF4, isotype distribution was analyzed with monovalent anti-IgG–, anti-IgA–, and anti-IgM–peroxidase conjugates instead of polyclonal anti–human IgG, IgA, and IgM, and A492 values ≥0.20 were considered positive.11 The cutoff values were determined on the basis of the mean value +3 SD of 50 control samples taken from normal subjects and from patients with thrombocytopenia from causes other than HIT.
In 45 patients with IgG antibodies, the subclasses were analyzed with monoclonal mouse anti–human IgG subclass–specific antibodies diluted at 1:5000 for anti-IgG1 (clone MH161-1 ME) and 1:200 for anti-IgG2 (clone MH162-1 ME), anti-IgG3 (clone MH163-1 ME), and anti-IgG4 (clone MH 164-4 ME) (CLB). The cutoff value for each IgG subclass was determined on the basis of the mean of control absorbance values (+3 SD) measured with 37 samples from patients without antibodies to H-PF4. All ELISA values were confirmed with ≥2 separate experiments.
Analysis of FcγRIIA Polymorphism
After isolation of genomic DNA,12 the FcγRIIA-H/R131 polymorphism was analyzed by an allele-specific restriction enzyme digestion method.13 Amplification was carried out with a mutagenic sense primer (5′-GGAAAATCCCAGAAATTCTCGC-3′) that created a BstUI site on the PCR product only if the codon for amino acid 131 resulted in an arginine.
A second mutagenic antisense primer (5′-CAACAGCCTGACT-ACCTATTACGCGGG-3′) introduced an internal BstUI site (in bold type) into every PCR product serving as a control for the restriction enzyme digestion. Reactions were performed in a final volume of 50 μL containing 0.1 to 1 μg of DNA, 1 μmol/L of each primer, 200 μmol/L of each dNTP, 10 mmol/L Tris-HCl (pH 9.0), 50 mmol/L KCl, 0.01% (wt/vol) gelatin, 1.5 mmol/L MgCl2, 0.1% Triton X-100, and 0.5 U of Super Taq DNA Polymerase (ATGC Biotechnologie). Amplification was performed in the GeneAmp PCR system 2400 thermocycler (Perkin-Elmer) programmed for 3 minutes at 94°C, followed by 30 cycles of 15 seconds at 94°C, 30 seconds at 55°C, and 40 seconds at 72°C. A final extension step was performed at 72°C for 7 minutes. PCR products (6 μL) were then incubated for 16 hours at 60°C with 10 U of BstUI (Biolabs) in 50 mmol/L NaCl, 10 mmol/L Tris-HCl (pH 7.9), 10 mmol/L MgCl2, and 1 mmol/L DTT at a final volume of 10 μL. Digestion products were separated by 6% polyacrylamide gel electrophoresis and visualized by ethidium bromide staining. Expected lengths of fragments were 366 bp for undigested PCR products, 343 bp for H/H131, 322 bp for R/R131, and 343 and 322 bp for R/H131.
[14C]Serotonin Release Assays
SRAs were performed as previously described,14 with 80 of the 83 plasma samples containing antibodies to H-PF4. Every plasma sample was tested in duplicate with platelets from 2 different donors having either the FcγRIIA-H/H131 or the FcγRIIA-R/R131 isoform. A test result was defined as positive if release >20% was measured at 0.1 or 1 IU/mL of heparin, with complete inhibition at 10 IU/mL. The result was defined as negative if the release was <20% or not inhibited in the presence of 10 IU/mL of heparin. SRA was also performed on 80 control sera in which no antibodies to H-PF4 were detected, obtained from groups 1 and 2 (40 from each).
Cumulative data (platelet counts, levels of antibodies to H-PF4) were expressed as mean±SEM, and the unpaired t test was used for comparisons between groups. ANOVA with repeated measures was used to compare the evolution of platelet counts. Fisher’s exact test was used to compare the frequency of events between patient groups (eg, thrombocytopenia, development of antibodies to H-PF4), the distribution of FcγRIIA isoforms, and the classes/subclasses of antibodies. A value of P≤0.05 was required for statistical significance.
Development of Antibodies to H-PF4 Complexes After CPB
Significant levels of antibodies to H-PF4 (A492≥0.5) were present in 3 patients in group 1 before surgery (Figure 1⇓). Four additional patients (2 in each group) developed antibodies between day 3 and day 5, but in all cases except 1, levels were stable and remained at <1.0 absorbance unit.
After 1 week of postoperative treatment, 83 patients (ie, 25.3%) had significant titers of antibodies to H-PF4. Forty-six received UFH (29% of 157 patients) and thirty-seven LMWH (21% of 171 patients), but the difference was not significant (P=0.12). Absorbance values measured before CPB in subjects who later developed antibodies to H-PF4 were higher than those of patients with consistently negative ELISA results (P<0.0001, Table 2⇓). However, even in these “negative” patients, absorbance values increased significantly in the second week of heparin treatment (P<0.0001 compared with baseline values).
Evolution of Platelet Counts and FcγRIIA Genotype
Before surgery, 7 patients had thrombocytopenia (<100×109/L), but none developed antibodies to H-PF4, and all recovered a normal platelet count. When patients were considered together, the platelet count fell from 212±3×109/L (mean±SEM) before surgery to 100±3×109/L during CPB. It then remained stable during the first 2 postoperative days, increased to 146±3×109/L on days 3 to 4, and in most patients was completely normal on days 5 to 7.
Mean platelet counts were lower in patients of group 1 as early as the end of CPB (131±3.4×109/L) compared with those of group 2 (149±3.9×109/L, P=0.001), and this difference increased on days 3 to 4 (126±3.1×109/L versus 165±3.7×109/L). In addition, 83 patients in group 1 had thrombocytopenia (<100×109/L) 3 to 4 days after surgery or a decrease in platelet count >40%, compared with only 26 in group 2 (P<0.0001). Mean platelet counts reached values >200×109/L in both groups after day 8 after surgery, with an evolution that was not significantly different in patients with or without antibodies to H-PF4 (P>0.1, repeated-measures ANOVA) (Figure 2⇓). On days 8 to 10, however, patients in group 1 with antibodies to H-PF4 had lower platelet counts than those without antibodies (P=0.03, unpaired t test).
HIT was suspected in 6 patients in group 1 (platelets <100×109/L or decrease >40% on days 8 to 10) and then confirmed by positive H-PF4 ELISA and positive SRA (Table 3⇓). All 6 patients received UFH, and in 2 cases, platelet counts dropped below 50×109/L, particularly in patient 2, who had severe pulmonary embolism. This patient was then treated with danaparoid sodium but died on day 13. Thrombosis did not occur in the other patients, and there was a favorable outcome after heparin withdrawal and initiation of oral anticoagulants (warfarin).
The distribution of FcγRIIA isoforms was studied in patients with and without antibodies to H-PF4 (Figure 3⇓). The H/H131, H/R131, and R/R131 alleles were found in 18.5%, 54.5%, and 27%, respectively, of patients who did not develop antibodies to H-PF4. The corresponding values were 19.7%, 42.2%, and 38.1%, respectively, in patients with antibodies, and the differences observed were not statistically significant. The mean platelet counts in patients with antibodies to H-PF4 were significantly lower on days 8 to 10 of the postoperative period only in those with the R/R131 genotype (Figure 3⇓, P=0.018), and this difference remained significant, even when the values of the 4 patients with HIT were excluded. By comparison, the evolution of platelet counts in patients without antibodies was identical whatever their FcγRIIA genotype.
Class and Subclass of Antibodies to H-PF4 and Platelet Activation
The classes of immunoglobulins specific for H-PF4 were determined in 79 patients. Significant levels of IgG antibodies were measured in 48 patients (24 from each group), with IgA and/or IgM in 29 samples. In 31 patients, only IgA and/or IgM antibodies to H-PF4 were detected. The prevalence of IgG and IgA antibodies and plasma levels were similar in both groups, whereas IgM isotype was more common in group 1 (31 versus 14 in group 2, P=0.003), and titers were higher (0.546±0.049 versus 0.339±0.035, P=0.01).
The predominant subclass of IgG antibodies was IgG1, which was present in 40 of 45 patients (88.8%) (Table 4⇓). An IgG3 response was present in 12 patients, with IgG1 in 8 of them, and no IgG2 or IgG4 antibodies were detected.
SRA was positive in 8 of 80 samples containing antibodies to H-PF4 and negative for all 80 control sera without antibodies. Six samples were from patients treated with UFH, and they all had HIT with antibodies that also cross-reacted with LMWH (dalteparin) on SRA. The other 2 positive samples on SRA were from patients 7 and 8, who were treated with dalteparin (Table 3⇑), and platelets remained >250×109/L in these subjects despite high levels of antibodies to H-PF4. The 8 positive samples contained high levels of antibodies (mean A492=2.43±0.271), with IgG1 in all patients and IgG3 in 2. Interestingly, the R131 gene was expressed by each of these patients, 4 being R131/R131 homozygotes and 4 being H131/R131 heterozygotes. Moreover, SRA was positive with both R131/R131 and H131/H131 [14C]serotonin-labeled platelets. The maximum release induced by antibodies was slightly higher with R131/R131 platelets (74%; range, 54% to 93% versus 58%, 23% to 90% with H131/H131 cells, cumulative data of 2 separate experiments), but the differences were not significant (P=0.075).
In this study, we evaluated the prevalence of antibodies to H-PF4 complexes in patients undergoing cardiac surgery, and we determined the predisposing factors that could favor the development of HIT. Antibodies to H-PF4 were detected in 29% of patients treated with UFH throughout the postoperative period (group 1) and in 21% of those receiving LMWH (group 2). These results are in accordance with those recently reported,15 but 2 previous studies found a higher prevalence of antibodies to H-PF4 after CPB (51% and 61%).7 16 This difference is probably related to the presence of antibodies before surgery in both studies (in 19% and 22% of patients, respectively), whereas only 3 subjects of our cohort had above-normal antibody titers when admitted for cardiac surgery. All our patients had been exposed to heparin 1 to 3 months before CPB, and mean preoperative absorbance levels were higher in those who then developed heparin-dependent antibodies. An anamnestic response to heparin was likely, because 54% (12/22) of patients with absorbance values between 0.25 and 0.5 before CPB later developed antibodies to H-PF4, compared with 22.4% of those with preoperative A492 values <0.25 (68/303, P=0.001).
CPB was always performed with UFH, and this could explain why the prevalence of heparin-dependent antibodies was not different between patients who received UFH or LMWH in the postoperative period. However, it is not certain that the exclusive use of LMWH during and after CPB is less immunogenic than UFH.16 17 PF4 was undoubtedly released in every case during CPB, potentially enhancing the production of heparin-dependent antibodies whether patients were then treated with LMWH or with UFH. In the patients in group 1, most of whom underwent valve replacement, platelet counts were lower after CPB than in those with CABG (group 2), and platelet activation was possibly more intense as a result of longer extracorporeal circulation. In the postoperative period, the patients in group 1 were treated by UFH before warfarin was introduced, because LMWHs are not currently authorized in France after CPB for valve replacement. Most patients in group 2 underwent CABG and received LMWH only for prophylaxis of venous thromboembolism. The patients in groups 1 and 2 were thus different, and in addition to heparin, the underlying conditions might also have enhanced the pathogenicity of heparin-dependent antibodies and hence thrombocytopenia and thrombosis.
Nevertheless, our experience supports the hypothesis that in the presence of LMWH, antibodies to H-PF4 are less likely to induce platelet activation and to provoke HIT. Indeed, HIT occurred only in patients who received UFH postoperatively (incidence of 3.8%), thus explaining why platelet counts were significantly reduced on days 8 to 10, when antibodies to H-PF4 were present only in group 1. In contrast, thrombocytopenia never occurred after 8 days of LMWH in group 2, although 2 subjects also had high levels of IgG to H-PF4, with positive results on SRA. On the other hand, in agreement with the findings of Suh et al,18 the antibodies to H-PF4 that activated platelets were preferentially IgG1, which existed at high plasma levels. However, in 1 patient with definite HIT (patient 6), the antibodies were almost exclusively IgG3, and this also supports the pathogenicity of this subclass. IgG was the most frequent antibody isotype (61%), but IgM (57%) and IgA (31.6%) were also often detected. However, the pathogenicity of IgA and IgM, which has been discussed in previous reports,11 14 18 is probably low, because no thrombocytopenia was observed when they were present alone, and SRA results were negative. High titers of IgM were more frequent in group 1, suggesting a primary antibody response to H-PF4. This is unlikely, however, because all patients had previously been challenged with UFH, and none of the cases with preoperative IgM antibodies later developed IgG to H-PF4.
We also investigated whether the FcγRIIA genotype is a risk factor for developing antibodies to H-PF4 and clinical manifestations of HIT after CPB. First, no difference in gene frequency was found between patients with antibodies to H-PF4 and those without. Thus, as expected, the H/R131 polymorphism of FcγRIIA did not play a role in the immune response to heparin and PF4. Second, when antibodies to H-PF4 were present, mean platelet counts were significantly lower only in FcγRIIA R131/R131 patients, although their antibody titers were identical to those of H131/R131 and H131/H131 patients, and this supports the hypothesis that heparin-dependent IgG1 preferentially activates R/R131 platelets.8
Several studies have been performed on the distribution of FcγRIIA polymorphism and the development of HIT, but with debatable results.8 9 19 20 21 22 In 3 studies, the homozygous H/H131 genotype was shown to be more frequent than R/R131 in patients with HIT.8 9 20 In addition, Denomme et al20 found that IgG1 antibodies to H-PF4 produced higher amounts of membrane-derived microparticles from H/H131 platelets, and it was suggested that platelets with this polymorphism should be chosen for functional assays for HIT.21 Our observations do not agree with these conclusions, because the His131 gene was not overrepresented in patients with HIT, and antibodies to H-PF4 activated both H131/H131 and R131/R131 platelets as assessed by SRA.
Furthermore, our results are in accordance with a recent study of a large population of 389 patients with HIT showing that FcγRIIA-R/R131 was overrepresented, particularly in subjects with thrombotic complications.22 The FcγRIIA-R131 allele is also a recognized risk factor in systemic lupus erythematosus, enhancing the pathogenicity of antibodies by reducing the phagocytosis and clearance of immune complexes.23 Such a process could contribute to HIT, but the platelet activation induced by heparin-dependent antibodies in vivo involves several components, such as ADP receptors, other than FcγRIIA receptors.24 In addition, functionally significant mutations in human Fcγ receptor sequences other than those concerning the amino acid at position 131 could also explain differences in platelet susceptibility to antibodies.25
It is thus possible that, together with acquired factors related to the underlying disorders, other genetic parameters play a role in the development of heparin-dependent antibodies, their pathogenicity, and the occurrence of HIT.
We are grateful to the members of the Department of Cardiac Surgery of Hôpital Trousseau (Tours) for providing the patient samples We also thank Pascal Garaud for statistical assistance and Doreen Raine for editing the English.
- Received November 18, 1998.
- Revision received January 29, 1999.
- Accepted February 16, 1999.
- Copyright © 1999 by American Heart Association
Kelton J, Smith J, Warkentin T, Hayward C, Denomme G, Horsewood P. Immunoglobulin G from patients with heparin-induced thrombocytopenia binds to a complex of heparin and platelet factor 4. Blood. 1994;83:3232–3239.
Khuri S, Michelson A, Valeri C. Effect of cardiopulmonary bypass on hemostasis. In: Loscalzo J, Schafer A, eds. Thrombosis and Hemorrhage. Boston, Mass: Blackwell Scientific Publications; 1994:1051–1073.
Bauer TL, Arepally G, Konkle BA, Mestichelli B, Shapiro SS, Cines DB, Poncz M, McNulty S, Amiral J, Hauck WW, Edie RN, Mannion JD. Prevalence of heparin-associated antibodies without thrombosis in patients undergoing cardiopulmonary bypass surgery. Circulation. 1997;95:1242–1246.
Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 1988;16:1215.
Arepally G, McKenzie SE, Jiang XM, Poncz M, Cines DB. Fc gamma RIIA H/R(131) polymorphism, subclass-specific IgG anti-heparin/platelet factor 4 antibodies and clinical course in patients with heparin-induced thrombocytopenia and thrombosis. Blood. 1997;89:370–375.
Denomme GA, Warkentin TE, Horsewood P, Sheppard JAI, Warner MN, Kelton JG. Activation of platelets by sera containing IgG1 heparin-dependent antibodies: an explanation for the predominance of the Fc gammaRIIa “low responder” (his131) gene in patients with heparin-induced thrombocytopenia. J Lab Clin Med. 1997;130:278–284.
Carlsson LE, Santoso S, Baurichter G, Kroll H, Papenberg S, Eichler P, Westerdaal NAC, Kiefel V, van de Winkel JGJ, Greinacher A. Heparin-induced thrombocytopenia: new insights into the impact of the FcγRIIa-R-H131 polymorphism. Blood. 1998;92:1526–1531.
Manger K, Repp R, Spriewald B, Rascu A, Geiger A, Wassmuth R, Westerdaal N, Wentz B, Manger B, Kalden J, van de Winkel J. Fc gamma receptor IIa polymorphism in Caucasian patients with systemic lupus erythematosus: association with clinical symptoms. Arthritis Rheum. 1998;41:1181–1189.
Polgar J, Eichler P, Greinacher A, Clemetson KJ. Adenosine diphosphate (ADP) and ADP receptor play a major role in platelet activation/aggregation induced by sera from heparin-induced thrombocytopenia patients. Blood. 1998;91:549–554.
Norris CF, Pricop L, Millard SS, Taylor SM, Surrey S, Schwartz E, Salmon JE, McKenzie SE. A naturally occurring mutation in Fc gamma RIIA: a Q to K-127 change confers unique IgG binding properties to the R-131 allelic form of the receptor. Blood. 1998;91:656–662.