Mechanical prophylaxis is a heparin-independent risk for anti–platelet factor 4/heparin antibody formation after orthopedic surgery

Heparin-induced thrombocytopenia (HIT) is caused by platelet-activating antibodies (HIT antibodies), mostly against platelet factor 4 (PF4)/heparin complexes.¹  When heparin makes a complex with PF4 in an optimal stoichiometric ratio, heparin induces conformational changes in PF4, thereby exposing neoantigens that trigger immune responses, which in turn generate anti-PF4/heparin antibodies.^2,3  Thus, the frequency of anti-PF4/heparin antibody formation depends on pharmacologic factors such as the type,⁴  exposure duration,⁵  and plasma concentration^2,3  of heparin. However, recent studies have also demonstrated that nonpharmacologic factors, such as the type of surgery³  and extent of trauma,⁶  also influence risk of anti-PF4/heparin immunization. An additional issue is that certain nonheparin polyanions, such as bacterial surfaces, nucleic acids, and hypersulfated chondroitin sulfate, can also induce anti-PF4/polyanion antibodies with properties similar to those of HIT antibodies.^7,8  Indeed, spontaneous HIT syndrome and fondaparinux-associated HIT can occur in patients with infections or recent major orthopedic surgery, both of which can generate sources of polyanions (such as bacterial surfaces and nucleic acids) from major tissue damage and the breakdown of bacteria, viruses, and blood cells, without any exposure to unfractionated heparin (UFH) or low-molecular-weight heparin (LMWH).^9,10  Because spontaneous HIT syndrome and fondaparinux-associated HIT have been reported most often in patients who have undergone total knee arthroplasty (TKA), major orthopedic surgery itself might be capable of triggering an anti-PF4/heparin immune response.⁹  However, the frequency of, and risk factors for, anti-PF4/heparin antibody formation without any heparin exposure remains unclear.

We conducted a multicenter, prospective cohort study of patients who underwent TKA or total hip arthroplasty (THA) to assess the effectiveness of various thromboprophylactic regimens.¹¹  Approximately half of the patients were treated with either UFH, LMWH, or fondaparinux. The other half received only mechanical thromboprophylaxis that was dynamic (intermittent plantar compression device [foot pump] or intermittent pneumatic compression device [IPCD]), static (graduated compression stockings [GCSs]), or both. Predefined, prospective serologic assessment of anti-PF4/heparin antibodies permitted us to clarify the frequency of, and risk factors for, antibody formation without any heparin exposure.

We performed a multicenter, prospective cohort study (the Clinical Study of Prevention and Actual Situation of Venous Thromboembolism After Total Arthroplasty, University Hospital Medical Information Network Clinical Trials Registry #UMIN000001366) involving patients undergoing elective TKA (n = 1294) or THA (n = 868) at 34 Japanese National Hospital Organization (NHO) hospitals, as described elsewhere.¹¹  The study protocol was approved by the NHO Central Institutional Review Board. Written informed consent was obtained from each patient. The study was conducted in accordance with the Declaration of Helsinki. The primary aim of the study was to assess the safety and effectiveness of various thromboprophylactic regimens. Patients were not randomly allocated to the various treatments. Thromboprophylaxis for each patient was at the treating physician’s discretion, according to his or her usual clinical practice. Approximately half of the patients received postoperative pharmacologic thromboprophylaxis with UFH, LMWH (enoxaparin), fondaparinux, or an antiplatelet drug. The other half received only mechanical thromboprophylaxis, either dynamic (foot pump or IPCD) or static (GCSs), or both. Details on each thromboprophylactic strategy (eg, dosage and duration) were described previously.¹¹  In this predefined substudy, duration of anticoagulation therapy was defined as the interval between the day that anticoagulation was initiated and the day that anticoagulation was terminated or that a blood sample was collected to measure anti-PF4/heparin antibodies, whichever came first. It was unlikely that patients treated with only mechanical thromboprophylaxis were exposed to heparin because heparin is not routinely used during TKA and THA in Japan, including for flushing of intraoperative catheters.

As a predefined substudy, we collected blood samples to test for anti-PF4/heparin antibodies before surgery and on postoperative day (POD) 10. Anti-PF4/heparin antibodies were evaluated using a commercially available enzyme-linked immunosorbent assay (ELISA) for detecting immunoglobulin (Ig)G, IgA, and IgM (Asserachrom HPIA; Diagnostica Stago, Asnières-sur-Seine, France) at a central laboratory by personnel blinded to all clinical information. The cutoff value was ∼0.4 to 0.5 optical density (OD) units, depending on the assay kit. Each kit has a specific reference standard to determine the cutoff value. We defined seroconversion as a positive test result on POD 10 corresponding to a negative result before surgery. In addition, we considered patients who tested positive before surgery to be seroconverters if their blood sample on POD 10 was positive with a twofold-or-higher increase in OD, as defined in a previous study.³  In some analyses, we grouped the immunoassay data into 3 categories based on OD units: less than the cutoff value (negative), between the cutoff value and 1.4 (weakly positive), and greater than or equal to 1.4 (strongly positive), based on a previous study that suggested that ≥1.4 OD units is possibly associated with a high (>50%) probability of the presence of HIT antibodies, which strongly induces platelet activation at therapeutic concentrations of heparin.¹² 

We scheduled blood samples to be collected on POD 10, but some samples were collected at different times. Previous studies have demonstrated that OD values peak and that anti-PF4/heparin seroconversion occurs on POD 8 or later in most patients who undergo major orthopedic surgery.¹³  Therefore, patients whose postoperative blood samples were not available or were taken on or before POD 7 were excluded from this study. Blood samples collected on or after POD 8 were treated the same as those collected on POD 10.

Statistical analysis was performed using SPSS, version 19 (IBM SPSS, Chicago, IL). Summary data for continuous variables are presented as medians and interquartile ranges. Categorical variables are summarized as frequencies and percentages, and were compared using the χ² test. When the sample size of a cell in a 2-by-2 table was <10, Fisher’s exact test was used instead.

The primary objective of this predefined substudy was to clarify the frequency of, and risk factors for, anti-PF4/heparin antibody formation without any heparin exposure.

Multivariate logistic regression was performed to identify independent risk factors for anti-PF4/heparin antibody seroconversion. Variables with a P value <.2 with the χ² test or Fisher’s exact test in the univariate logistic regression analysis were selected for the multivariate model. To calculate odds ratios (ORs) and 95% confidence intervals (95% CIs) for anti-PF4/heparin antibody seroconversion while controlling for potential confounders, variables selected by univariate logistic regression analysis were included into a multivariate logistic regression model with the stepwise forward selection method, with forced entry of gender, surgery type, and each pharmacologic prophylaxis type, which have been identified as risk factors for anti-PF4/heparin antibody formation in previous studies.^3,14-17  A 2-tailed P value <.05 was considered to indicate statistical significance.

The results of the multivariate logistic regression analysis showed that dynamic mechanical thromboprophylaxis (DMT; ie, foot pump or IPCD) was a major heparin-independent risk factor for anti-PF4/heparin antibody formation, as described in detail in “Results.” Because thromboprophylaxis for each patient was at the treating physician’s discretion, propensity-score matching was performed to (1) minimize any effects of confounding caused by nonrandomized assignment to thromboprophylaxis strategy, (2) reduce the impact of selection bias, and (3) allow relevant variables to be balanced among patients who received or did not receive DMT. All the variables in the study were included in the logistic regression model to estimate the propensity score, which represents the probability of DMT use. A Markov chain Monte Carlo procedure was used for multiple imputation of missing values for numerical variables such as body mass index and estimated glomerular filtration rate. A mode imputation procedure was used to impute missing values for categorical variables such as comorbidities. The multiple imputation and mode imputation were performed using the XLSTAT-Base 2015 (Addinsoft, Paris, France). We matched the patients who received or did not receive DMT on a 1:1 basis using nearest-number matching with a caliper of 0.02. After matching, we compared seroconversion rates and the proportion of patients with high OD values (≥1.4 units) for anti-PF4/heparin antibodies using the McNemar test.

The seroconversion rates and the proportion of patients who tested strongly positive (ELISA values ≥1.4 OD units) were determined separately for patients receiving UFH, LMWH, fondaparinux, an anti-platelet drug, or only mechanical thromboprophylaxis after TKA or THA. The seroconversion rates and the proportion of patients who tested strongly positive (ELISA values ≥1.4 OD units) were compared between the patients treated with or without DMT in each group using the χ² test or Fisher’s exact test, whichever was appropriate. P values were not corrected for multiple hypothesis tests.

A total of 2069 patients (n = 1244 for TKA and n = 825 for THA) were ultimately eligible, after excluding patients whose postoperative blood samples were not available or were taken on or before POD 7 (n = 50 for TKA and n = 43 for THA). The proportion of patients excluded from the analysis due to unavailable postoperative samples or samples taken on or before POD 7 in each treatment group is shown in Figure 1. Pharmacologic thromboprophylaxis in patients who underwent TKA consisted of fondaparinux in 334 patients (26.8%), LMWH (enoxaparin) in 217 (17.4%), UFH in 70 (5.6%), and antiplatelet agents in 45 (3.6%). There were 578 patients (46.4%) who received mechanical thromboprophylaxis without any antithrombotic drugs. Among patients who underwent THA, 241 (29.2%) were treated with fondaparinux, 143 (17.3%) with LMWH (enoxaparin), 32 (3.9%) with UFH, 43 (5.2%) with antiplatelet agents, and 366 (44.4%) with only mechanical thromboprophylaxis.

Univariate analysis identified several variables that differed significantly between patients who developed anti-PF4/heparin antibodies and those who did not experience seroconversion (Table 1). Multivariate analysis revealed that higher seroconversion rates were significantly associated with female gender, TKA surgery, spinal anesthesia, DMT (foot pump or IPCD) use, and fondaparinux therapy (Table 2). The presence of rheumatoid arthritis and the use of GCSs were significantly associated with a lower seroconversion rate. Female gender¹⁴  and TKA surgery³  were previously shown to be risk factors. Among pharmacologic prophylaxis regimens, only fondaparinux was significantly associated with a higher risk of developing anti-PF4/heparin antibodies (OR, 1.48; 95% CI, 1.08-2.02; P = .014). In contrast, the administration of LMWH was not associated with anti-PF4/heparin antibody formation (OR 1.00; 0.67-1.48; P = .999). Patients treated with UFH had a higher seroconversion rate (OR 1.76; 95% CI, 0.98-3.15; P = .059), which was not statistically significant, which may be explained by the small sample size for that subgroup. These findings were inconsistent with previous studies,^3,16  which suggested that anti-PF4/heparin antibodies are generated at similar frequencies in patients treated with fondaparinux or enoxaparin (LMWH) who underwent major orthopedic surgery. Of interest, we found for the first time that DMT is an independent risk factor (OR, 2.01; 95% CI, 1.34-3.02; P = .001) for anti-PF4/heparin seroconversion, which may lead to different seroconversion rates across studies.

To further explore the role of DMT in anti-PF4/heparin antibody formation, seroconversion rates and the proportion of patients who tested strongly positive (ELISA values ≥1.4 OD units) were compared among patients who underwent TKA (Figure 2A) or THA (Figure 2B). Among patients who underwent TKA, those who received fondaparinux plus DMT had a significantly higher seroconversion rate than those who were treated with fondaparinux but no DMT (21.3% vs 5.7%, P = .025). Surprisingly, the frequency of seroconversion in patients treated with DMT but no anticoagulation (15.4%) appeared to be similar to that in patients treated with UFH or LMWH but no DMT (14.8% or 13.7%, respectively) and significantly higher than that in patients treated with only GCSs (15.4% vs 6.5%, P = .002), as shown in Figure 2A. In addition, Figure 2A suggests that the proportion of patients with high OD values (≥1.4 units) was higher among patients treated with any anticoagulant plus DMT than among patients treated with any anticoagulant without DMT, although a significant difference was observed between patients treated with and without DMT only when no anticoagulant was administered (4.1% vs 1.0%, P = .041). A total of 46 patients tested strongly positive with high OD values (≥1.4 units) on POD 10 (Figure 2). However, only 1 patient in the TKA group treated with LMWH plus DMT tested positive before surgery, suggesting that in most of these patients, strong anti-PF4/heparin antibodies were induced by stimuli during or after surgery, independent of the presence of anti-PF4/heparin antibodies before surgery. These results suggest that DMT is a major heparin-independent risk factor for antibody formation.

Similar trends were observed for patients who underwent THA (Figure 2B); however, we cannot show any statistically significant difference across thromboprophylaxis strategies because the seroconversion rates were low in this patient population and the number of patients who did not receive DMT was small.

As shown in Table 3, there were 365 pairs of patients matched on a 1:1 basis according to DMT status. We confirmed that DMT was an independent risk factor for seroconversion (OR, 1.99; 95% CI, 1.17-3.37; P = .018) and for high absorbance values (≥1.4 OD units; OR, 10.3; 95% CI, 1.31-80.5; P = .012) after adjusting for differences in risk with propensity-score matching (Table 4). As shown in Table 3, most variables were balanced after matching, although there was a significant imbalance in LMWH treatment status. Nonetheless, LMWH treatment was not a risk factor for anti-PF4/heparin antibody formation in both univariate and multivariate logistic regression analyses (Table 2). Thus, it is very unlikely that this imbalance could have influenced the conclusion that DMT was an independent risk factor for anti-PF4/heparin antibody formation.

The most significant finding of this study is that major orthopedic surgery, particularly TKA, independent of heparin exposure, can induce substantial anti-PF4/heparin antibody formation, which is strongly associated with postoperative DMT consisting of a foot pump or IPCD. We reviewed the literature of 15 cases of spontaneous HIT syndrome, a thromboembolic disorder resembling HIT that occurs in patients without preceding heparin exposure.¹⁰  Among them, 7 occurred after orthopedic surgery (all but 1 occurred after TKA). Based on these lines of evidence, it has been suggested that orthopedic surgery itself could trigger an anti-PF4/heparin immune response, although direct evidence to confirm this hypothesis is lacking and its mechanisms remain uncertain.⁹  To the best of our knowledge, the present study clearly shows for the first time that major orthopedic surgery can induce formation of anti-PF4/heparin antibodies independent of heparin exposure, and that this effect is enhanced by subsequent DMT therapy. We speculate that the mechanical stimulation and subsequent tissue damage associated with DMT induces anti-PF4/heparin antibodies through the production of polyanions such as glycosaminoglycans and nucleic acids. Several studies, including the present one, have indicated that the seroconversion rate is significantly higher in TKA vs THA patients,^3,18  probably due to local tissue ischemia related to tourniquet use in TKA.³  DMT and tourniquet use may have similar effects. PF4 release from platelets activated by mechanical stimulation with DMT might be another possible mechanism. Further studies are needed to determine the precise mechanisms by which DMT is involved in the immune response against PF4/heparin complexes.

Multivariate logistic regression analysis revealed that only thromboprophylaxis with fondaparinux was significantly associated with an increased risk of seroconversion among the pharmacologic agents studied (Table 2). The administration of LMWH was not associated with anti-PF4/heparin antibody formation. There was a statistically nonsignificant increase in the seroconversion rate in patients treated with UFH (OR 1.76; 95% CI, 0.98-3.15; P = .059), which may be explained by the small sample size for that subgroup. Our findings are inconsistent with the results of previous studies,^3,16  which found that patients undergoing elective TKA or THA with fondaparinux and enoxaparin (LMWH) as postoperative thromboprophylaxis have similar frequencies of anti-PF4/heparin antibody generation. Variations in drug-related factors such as plasma concentration, which is affected by dose and body mass index, and duration of treatment may have contributed to differences in results.^2,3,5  In addition, our novel finding that DMT is an independent risk factor for anti-PF4/heparin antibody formation may be a key factor in explaining the differences observed. As shown in Figure 2A, we found the frequency of seroconversion in patients treated with UFH or LMWH but no DMT to be relatively high (14.8% or 13.7%, respectively) and similar to that in patients treated with DMT but no anticoagulation (15.4%), which was the largest patient group in the present study (n = 370). This surprising finding is a possible reason why administration of LMWH or UFH was not an independent risk factor for anti-PF4/heparin antibody formation in the present study, unlike previous reports of TKA patients,^15,18,19  which found that UFH and LMWH are involved in the generation of anti-PF4/heparin antibodies.

Most cases of fondaparinux-associated HIT, a disorder resembling HIT that occurs during fondaparinux therapy without any exposure to UFH or LMWH, have been reported in patients undergoing TKA.⁹  A substantial number of patients who underwent major orthopedic surgery and received fondaparinux thromboprophylaxis developed anti-PF4/heparin antibodies.^3,16  Therefore, it has been suggested that fondaparinux can induce anti-PF4/heparin antibodies. In contrast, several studies have indicated that anti-PF4/heparin antibodies did not react against PF4/fondaparinux complexes,²⁰  including those in sera obtained from patients who formed antibodies during fondaparinux therapy.¹⁶  Previous studies indicated that 11 to 12 saccharides constitute the minimal structure length necessary to complex with and induce antigenicity in PF4 to react with HIT antibodies.^21-24  In this context, fondaparinux, a pentasaccharide, is too small to strongly react with HIT antibodies. These observations suggest that fondaparinux would not be associated with antibody formation. In the present study, the seroconversion rate in patients treated with fondaparinux but no DMT was much lower than that in patients treated with fondaparinux plus DMT, and was nearly identical to the rate in patients treated with only GCSs (Figure 2A). Our results support the hypothesis that fondaparinux alone is not associated with antibody formation.

HIT has several features that are atypical for an immune-mediated disease: heparin-naïve patients can develop IgG antibodies as early as day 4,¹³  as in a secondary immune response; evidence for an anamnestic response on heparin re-exposure is lacking²⁵ ; and HIT antibodies are relatively short-lived (50-85 days), unlike those in a true secondary immune response.²⁶  To explain this atypical response, previous studies have proposed that the pattern of antibody formation might be more compatible with a non–T cell–dependent immune reaction, which is induced efficiently when an antigen is presented in a repetitive rigid form.^5,27-29  The antigenic PF4/heparin complex can expose such repetitive epitopes, which are similar to repetitive viral epitopes known to cause T cell–independent B-cell activation.^5,30  A recent study of anti-PF4/heparin immunization in patients after cardiac surgery suggested that perioperative inflammation affects the immune response.³¹  An inflammatory stimulus can break down PF4/heparin-specific B-cell tolerance, resulting in spontaneous production of PF4/heparin-specific antibodies, especially IgG antibodies.^5,28  DMT-associated inflammation may modify the immune system as an additional danger signal to trigger B cells, probably primitive marginal zone B cells,^29,32  with specificity for PF4/polyanion complexes. In the present study, the proportion of patients with high OD values (≥1.4 units) appears to be higher in patients treated with any anticoagulant plus DMT than in patients treated with any anticoagulant but no DMT, although a significant difference was observed between TKA patients treated with and without DMT only when no anticoagulant was administered (Figure 2). These results suggest that DMT might also play a costimulatory role in immune response induction.

Several studies suggested that heparins such as UFH, LMWH, and fondaparinux have anti-inflammatory effects.³³  These anti-inflammatory effects could attenuate the inflammation induced by DMT, which triggers the immune response against PF4/heparin, although UFH or LMWH itself provides antigenicity against PF4/heparin. The difference in anti-inflammatory effects and antigenicity among heparins might contribute to a variety of costimulatory effects of DMT in inducing anti-PF4/heparin antibody formation. This hypothesis may explain why seroconversion rates were significantly different between patients treated with and without DMT only when they were treated with fondaparinux or no anticoagulant, although further studies are needed to test this hypothesis.

To date, several studies have investigated the anti-PF4/heparin antibody seroconversion rate and risk factors for an immune response in TKA and THA patients who received thromboprophylaxis with UFH, LMWH, or fondaparinux.^{3,15,16,18,19}  These studies showed that female gender and TKA surgery (vs THA surgery) were risk factors for an immune response, consistent with our results. However, the seroconversion rate varied among the studies, even those in which the patient population had the same surgery and postoperative anticoagulant. Although most papers did not state whether the patients were treated with DMT, 2 papers clarified that the use of intermittent pneumatic compression was prohibited in their studies.^3,16  Rates of seroconversion, assessed using an immunoassay for anti-PF4/heparin IgG/A/M, appear to be lower in these studies than among patients in our study treated with DMT who underwent the same surgery and received the same anticoagulant. These findings might also support our hypothesis that DMT plays a crucial role in the immune response against PF4/heparin, resulting in different seroconversion rates across studies.

This study has several limitations. Patients were not allocated to prophylactic regimens on a random basis; thus, there could be unrecognized confounders even after propensity-score matching. An imbalance in the proportion of patients who were excluded from the analysis due to postoperative samples being unavailable or taken on or before POD 7 was observed among treatment groups, as shown in Figure 1. This might have resulted in residual confounding. The number of patients in each group is relatively small, especially the group treated with any anticoagulant but no DMT. Variable duration of treatment of each anticoagulant might be a possible confounder. We did not study the seroconversion rates for anti-PF4/heparin IgG, IgA, or IgM separately, or antibodies with platelet-activating properties, which might be reasons for the differences from previous studies.

In conclusion, we propose a novel hypothesis that DMT is involved in the immune response against PF4/heparin independent of heparin exposure, and that DMT increases the risk of PF4/heparin seroconversion and presumably the associated risk of developing clinical HIT, spontaneous HIT syndrome, or fondaparinux-associated HIT. Our findings provide insight into the pathogenesis of HIT.

This study was supported by a grant from the NHO (Multicenter Clinical Studies for Evidenced-Based Medicine).

Contribution: S.B., S. Miyata, K.M., M.N., and S. Motokawa participated in the design of the study; K.S., T.S., T.T., M.S., Y.S., K.K., C.K., S.N., T. Mori, K.I., S.O., T. Minamizaki, S.Y., and N.S. collected the clinical data; S.B., S. Miyata, K.M., M.K., and S. Motokawa analyzed the data; and S.B., S. Miyata, K.M., and S. Motokawa wrote the manuscript. All authors had access to the complete clinical data set and read and approved the final manuscript.

Conflict-of-interest disclosure: S. Miyata has received research support and speaker honoraria from Daiichi Sankyo Co., Ltd (Tokyo, Japan), Mitsubishi Tanabe Pharma Corporation (Osaka, Japan), and CSL Behring K.K. (Tokyo, Japan). M.N. has received remuneration from Daiichi Sankyo Co., Ltd (Tokyo, Japan) and Bayer Yakuhin, Ltd (Tokyo, Japan). The remaining authors declare no competing financial interests.

Correspondence: Shigeki Miyata, Division of Transfusion Medicine, National Cerebral and Cardiovascular Center, 5-7-1 Fujishirodai, Suita-city, Osaka 565-8565, Japan; e-mail: smiyata@ncvc.go.jp.