Article Text

Anti-heparin platelet factor 4 antibodies in systemic lupus erythaematosus are associated with IgM antiphospholipid antibodies and the antiphospholipid syndrome
  1. D Alpert1,
  2. L A Mandl,
  3. D Erkan1,
  4. W Yin2,
  5. E I Peerschke2,
  6. J E Salmon1
  1. 1
    Department of Rheumatology, Hospital for Special Surgery, New York, USA
  2. 2
    Department of Pathology and Laboratory Medicine, Weill Medical College of Cornell University, New York, USA
  1. D Alpert, Jersey Shore University Medical Center, 1945 State Route 33, Neptune, New Jersey 07753, USA; alperd01{at}


Objective: To investigate the prevalence and clinical correlates of anti-heparin platelet factor 4 antibodies (anti-HPF4) in systemic lupus erythaematosus (SLE) patients with and without antiphospholipid antibodies (aPL).

Methods: Sera and clinical data were obtained from the Hospital for Special Surgery Autoimmune Disease Registry for 78 aPL-positive and 91 aPL-negative SLE patients without heparin-induced thrombocytopenia (HIT). Controls were 90 blood donors of comparable age and sex. Sera were assayed for anti-HPF4, IgG/IgM antiphospholipid antibodies (APhL), and IgG/IgM anti-β2-glycoprotein 1 antibodies (anti-β2GP1). Serotonin release assays (SRAs) were performed for subjects with positive anti-HPF4.

Results: Positive anti-HPF4 was seen in 9% of aPL-positive SLE patients, 4% of aPL-negative SLE patients and 1% of controls (p = 0.026, aPL-positive SLE vs controls). Two of 12 subjects with positive anti-HPF4 had reactive SRAs. In SLE patients, anti-HPF4 significantly correlated with IgM APhL, IgM anti-β2GP1, and inversely with complement C4. In immunoabsorption experiments, there was partial cross-reactivity of IgM anti-HPF4 with IgM APhL, but not with IgM anti-β2GP1. SLE patients with positive anti-HPF4 had increased odds of the antiphospholipid syndrome (APS; odds ratio (OR) 4.5, p = 0.019), and APS with arterial thrombosis (OR 6.1, p = 0.007). In multivariate linear regression analyses, APS and IgM APhL were independently associated with anti-HPF4.

Conclusions: Anti-HPF4 is detectable in SLE patients with and without aPL in the absence of HIT, and is most prevalent in aPL-positive SLE patients. In this SLE cohort, anti-HPF4 correlates with IgM APhL, IgM anti-β2GP1 and inversely with C4, and is associated with manifestations of APS.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Heparin-induced thrombocytopenia (HIT) is an immune-mediated disorder characterised by thrombocytopenia during or shortly following heparin therapy, leading to thromboses in at least a third of cases.1 It is associated with antibodies against platelet factor 4 (PF4) complexed with heparin or other glycosaminoglycans (anti-HPF4), detected by immunoassay.2 Serotonin release assays (SRAs) are functional assays for the ability of heparin-dependent IgG to activate platelets in vitro, and are more specific for clinical HIT than anti-HPF4.3

The incidence of HIT is 1–5% in all patients exposed to heparin.1 Isolated anti-HPF4 in the absence of HIT have been documented in patients recently exposed to heparin, with a prevalence ranging from 7.5–50%,4 and have occasionally been reported in patients without heparin exposure.5

The antiphospholipid syndrome (APS) is defined as the presence of persistent antiphospholipid antibodies (aPL) in the setting of thrombosis and/or pregnancy morbidity.6 Often, APS is associated with systemic lupus erythaematosus (SLE). Antibodies found in APS are heterogeneous and generally recognise anionic phospholipids complexed with phospholipid binding proteins, such as β2-glycoprotein 1 (β2GP1).7 APS has also been associated with thrombocytopenia, with an estimated prevalence of 20–40% and diverse postulated aetiologies.7 8

HIT and APS are antibody-mediated disorders associated with thrombocytopenia and/or thrombosis.9 Several small studies have demonstrated the presence of anti-HPF4 in 6–15% of heterogeneous patients with aPL.1014 Interestingly, most of these subjects did not have clinical HIT, and many did not have prior documented heparin exposure.

It is unclear whether the increased prevalence of anti-HPF4 in aPL-positive patients is significant compared with the general population, or clinically relevant. Some authors have postulated that anti-HPF4 in aPL-positive patients may represent an additional pro-thrombotic risk factor for APS,11 15 whereas a recent report finds no association of anti-HPF4 with APS in 72 aPL-positive patients.13 In the present study, anti-HPF4 prevalence in well-defined cohorts of aPL-positive and aPL-negative SLE patients was compared to prevalence in blood donors from the general population. Associations of anti-HPF4 with serologic and clinical manifestations of SLE and APS were assessed.


Study population

Patients enrolled in the Hospital for Special Surgery (HSS) Autoimmune Disease Registry were screened for inclusion in this retrospective study. Included subjects were 18–75 years of age and met American College of Rheumatology (ACR) criteria for SLE.16 SLE patients with APS had a history of IgG or IgM anticardiolipin antibody (aCL) levels ⩾20 units, anti-β2GP1 IgG or IgM levels ⩾20 units, and/or positive lupus anticoagulant (LA) test, documented on two or more occasions at least 6 weeks apart, and met Sapporo clinical criteria.6 SLE patients having aPL without APS met the above laboratory criteria, but not Sapporo clinical criteria. APL-negative SLE patients had at least one documented negative LA and negative (<20 units) IgG, IgM and IgA aCL, with no documented positive aPL. Exclusion criteria were pregnancy or concurrent heparin treatment on the date of serum collection, or history of HIT.

A test serum aliquot for each SLE patient, collected between 1995 and 2006 and stored at–80°C, was obtained for analysis. Demographic and clinical data for each patient were extracted from the registry and supplemented by systematic chart reviews. Sera from blood donors of comparable age and sex were obtained commercially as controls (Innovative Research, Southfield, Michigan, USA). The HSS Institutional Review Board approved this study.


Anti-HPF4 was detected using the PF4 Enhanced immunoassay kit (GTI, Waukesha, Wisconsin, USA), which includes a microtitre plate coated with PF4-polyvinylsulfonate (PVS) as substrate, and a mixture of anti-IgG/IgM/IgA secondary antibodies for total anti-HPF4 detection. Optical densities at 405 nm (OD 405) were assayed in duplicate, along with standardised positive and negative controls. Inter-assay and intra-assay coefficients of variation averaged 6.6% and 2.6%, respectively. An OD 405 ⩾0.400 was used to define anti-HPF4 positive samples, as per the manufacturer. This cut-off also represented 2.8 standard deviations above the mean OD 405 of our control samples. For some experiments, separate anti-human IgG, IgM, and IgA secondary antibodies (GTI) were used. The proportion of a given anti-HPF4 isotype in each sample was defined as the OD 405 for that isotype, divided by the sum of OD 405 values for all individual isotypes.

Levels of aPL and anti-β2GP1 antibodies were determined for all serum samples. IgG and IgM aPL were assayed by a commercial immunoassay kit (Louisville APL Diagnostics, Seabrook, Texas, USA), using a proprietary phospholipid mixture (APhL) and β2GP1 as substrate coated on polystyrene plates. In our laboratory, APhL levels correlate strongly with aCL levels (data not shown), and may be more specific for APS.17 18 IgG and IgM anti-β2GP1 were assayed using a commercial immunoassay kit (Inova Diagnostics, San Diego, California, USA). Optical densities were measured at 450 nm (OD 450), with results expressed as lgG or lgM phospholipid units (GPL and MPL, respectively) for APhL,19 or standard IgG (SG) or IgM (SM) units for anti-β2GP1.

Serum immunoabsorption experiments

Sera from four aPL-positive SLE patients with elevated IgM anti-HPF4 and IgM APhL levels were diluted 1:50 in phosphate-buffered saline containing 0.05% sodium azide, and incubated overnight at 4°C on PF4-PVS, APhL, or β2GP1-coated microtitre plates. Control diluted samples were incubated overnight at 4°C in Eppendorf tubes or uncoated polystyrene plates, and both methods produced comparable results. Control and immunoabsorbed samples were then assayed on each of the above microtitre plates to test for potential cross-reactivity among IgM anti-HPF4, APhL and anti-β2GP1.

Serotonin release assays

Blood from a healthy donor, known to be reactive in the assay, was collected in 3.8% sodium citrate (1:9 anticoagulant/blood ratio). Platelet-rich plasma was incubated with 33 nCi/ml 14C-serotonin (Amersham Biosciences, Piscataway, New Jersey, USA) for 45 min at 37°C. Platelets were centrifuged following addition of 13 mM EDTA and resuspended at 300 000 platelets/μl in 0.01 M Hepes-buffered modified Tyrode solution.20 Washed platelet suspensions were stirred with 20 μl heat-inactivated (56°C, 30 min) test serum in the presence of unfractionated porcine heparin (Elkin Sims, Cherry Hill, New Jersey, USA) at final concentrations of either 0.1 U/ml or 100 U/ml. After 60 min, the reaction was stopped with 100 μl of 0.5% EDTA in 0.01 M phosphate-buffered saline. Supernatant radioactivity was quantified by liquid scintillography, and 14C-serotonin release was calculated as radioactivity in platelet supernatants relative to total platelet serotonin uptake.21 Sera were considered reactive if induced serotonin release was ⩾20% in the presence of 0.1 U/ml heparin, and decreased by ⩾50% in the presence of 100 U/ml heparin.22 23 Results were confirmed in two independent experiments using different platelet donors.


Analyses were performed using SPSS version 14 (SPSS Inc., Chicago, Illinois, USA) and GraphPad Prism version 4.03 (San Diego, California, USA). Either χ2 or Fisher exact tests were performed for frequency comparisons. Continuous variables were analysed by unpaired t tests and ANOVA, or Mann–Whitney U tests and Kruskall–Wallis tests (for non-parametric variables). For certain analyses, non-parametric variables were log-transformed to allow for parametric testing. Correlations between variables were determined using Spearman rank tests. Multivariate linear regression analyses were used to determine covariates that were independently associated with log-transformed anti-HPF4 levels. All analyses were two-tailed and p values ⩽0.05 were considered significant.


A total of 402 consecutive patients from the HSS Autoimmune Disease Registry were screened for this study; 169 SLE patients (91 aPL-negative and 78 aPL-positive) meeting inclusion and exclusion criteria were ultimately analysed. Most SLE patients were excluded for incomplete or inconsistent aPL serologies (see Supplementary data).

APL-positive SLE patients were more likely to be white, and trended towards older age (table 1). Compared with aPL-negative SLE patients, aPL-positive SLE patients were more likely to have livedo, recent heparin exposure, a lower platelet count, a higher erythrocyte sedimentation rate (ESR), and less likely to be taking antimalarials. There were no significant differences in medical comorbidities including hypertension, diabetes, hypercholesterolemia, coronary artery disease, thyroid disease, or osteonecrosis between SLE groups (data not shown).

Table 1 Demographic, clinical and laboratory characteristics of all subjects at date of test serum

Of the 78 aPL-positive SLE patients, 27 had aPL without APS, and 51 had APS by Sapporo criteria.6 Of the SLE patients with APS, 41 had at least one thrombosis (32 arterial and 17 venous), and 23 experienced at least one pregnancy morbidity.

Prevalence of anti-HPF4

There was an increased prevalence of positive anti-HPF4 in aPL-positive SLE patients compared with controls (9% vs 1%, p = 0.026; fig 1A). Additionally, anti-HPF4 levels were significantly higher in aPL-positive SLE patients, compared with aPL-negative SLE patients and controls (table 1 and fig 1B). Exclusion of controls (n = 10) and aPL-negative SLE patients (n = 9) whose test sera showed low to moderate IgG or IgM APhL levels, and one aPL-positive SLE patient whose test serum was negative for APhL and anti-β2GP1, did not significantly alter the analysis results (data not shown).

Figure 1 Prevalence and levels of anti-heparin–platelet factor 4 complex (HPF4) in controls and systemic lupus erythaematosus (SLE) patients. (A) Percentage of subjects in each group with positive anti-HPF4 levels. Number (%) is indicated on top of each bar. (B) Anti-HPF4 levels among subjects in each group. A positive anti-HPF4 is defined as OD 405⩾0.400 (above dotted line).

There were 12 subjects with positive anti-HPF4 levels (table 2). Among the seven aPL-positive SLE patients, all had APS and six of seven patients had arterial thromboses. None of the four aPL-negative SLE patients had a history of vascular or pregnancy events. Of note, the one control patient with positive anti-HPF4 had a low positive IgM APhL level. Two of the 12 subjects also had positive SRAs; both were aPL-positive SLE patients.

Table 2 Characteristics of the 12 subjects with positive anti-heparin–platelet factor 4 (HPF4) levels (OD 405⩾0.400)

Clinical associations of anti-HPF4

There was no significant correlation between anti-HPF4 and age, platelet counts, ESR, dsDNA levels, C3 levels, number of ACR SLE criteria, or duration of SLE (data not shown). Total anti-HPF4 levels significantly correlated with IgM APhL and IgM anti-β2GP1, less strongly correlated with IgG APhL, and inversely correlated with C4 levels (table 3). To further characterise these correlations, specific isotypes of anti-HPF4 were determined for all SLE patients with anti-HPF4 levels ⩾0.200 (n = 58). For these patients, the median proportion of IgM anti-HPF4 (56%) was significantly higher than IgG (22%) and IgA (14%) anti-HPF4 (p value <0.001). The correlations between IgM anti-HPF4 and IgM APhL, IgM anti-β2GP1, and C4 were more robust compared to each of these correlations with total anti-HPF4 (table 3). No significant correlations were seen between IgG anti-HPF4 and any IgG aPL or C4.

Table 3 Correlations of anti-HPF4 levels with APhL, anti-β2GP1 and complement C4 levels in systemic lupus erythaematosus (SLE) patients

To examine the underlying biology of these correlations, we performed immunoabsorption experiments to determine whether IgM anti-HPF4 cross-react with IgM APhL or IgM anti-β2GP1. We analysed sera from four aPL-positive SLE patients with positive IgM anti-HPF4 and elevated IgM APhL; all but patient 1 also had elevated IgM anti-β2GP1 (table 2). A serum sample from each patient was immunoabsorbed onto PF4-PVS, APhL or β2GP1-coated microtitre plates, or not immunoabsorbed (controls), and then analysed on each of the immunoassay plates. Neither APhL-absorbed or anti-β2GP1-absorbed sera showed decreased reactivity in an IgM anti-HPF4 immunoassay (fig 2A). By contrast, anti-HPF4-absorbed sera showed decreased reactivity in an IgM APhL immunoassay for each patient (fig 2B), but not in an IgM anti-β2GP1 immunoassay (fig 2C). The cross-reactivity between IgM anti-HPF4 and IgM APhL was specific for anti-HPF4 absorption; when sera from four additional IgM APhL-positive SLE patients with negative anti-HPF4 (OD 405 <0.200) were immunoabsorbed onto PF4-PVS plates, there was no significant decrease in reactivity in IgM HPF4 or IgM APhL immunoassays (data not shown).

Figure 2 Solid-phase immunoabsorption assays for cross-reactivity among anti-heparin–platelet factor 4 complex (HPF4), antiphospholipid antibodies (APhL) and anti-β2-glycoprotein 1 (β2GP1) IgM antibodies. Diluted serum samples were incubated overnight at 4°C (controls, black bars), or immunoabsorbed overnight at 4°C on PF4-polyvinylsulfonate (PF4-PVS; dotted bars), APhL (hatched bars) or β2GP1 (white bars) microtitre plates. Immunoabsorbed samples were then aliquotted equally and assayed on (A) a PF4-PVS plate, (B) an APhL plate, and (C) a β2GP1 plate, using anti-IgM secondary antibodies. Results are presented as mean (SD) for two independently performed experiments.

Table 4 demonstrates the characteristics of anti-HPF4 positive vs anti-HPF4 negative SLE patients. Anti-HPF4 positive SLE patients were more likely to have APS (OR 4.5, 95% CI 1.3–16.3), APS with thrombosis (OR 4.2, 95% CI 1.2–14.6), and particularly APS with arterial thrombosis (OR 6.1, 95% CI 1.7–21.5). There was no significant difference between anti-HPF4 positive and anti-HPF4 negative SLE patients in the prevalence of aPL without APS, APS with venous thrombosis or APS with pregnancy morbidity (data not shown). Anti-HPF4 positive SLE patients were also more likely to have lower C4 levels, and trended towards increased prevalence of neurological disease, livedo, hospitalisations within the preceding year, and increased IgM APhL and anti-β2GP1 levels.

Table 4 Demographic, clinical and laboratory characteristics of anti-HPF4 positive vs anti-HPF4 negative SLE patients at date of test serum

In a multivariate linear regression model including age, sex, race, IgM APhL, IgM anti-β2GP1, APS with arterial thrombosis, recent heparin exposure and recent hospitalisation, only APS with arterial thrombosis (β = 0.28, p = 0.001) and IgM APhL (β = 0.29, p = 0.016) were significantly associated with anti-HPF4 levels in SLE patients (R2  = 0.19, p value <0.001). Similar results were observed when either APS (R2  = 0.17, p value <0.001) or APS with thrombosis (R2  = 0.17, p = 0.001) were substituted for APS with arterial thrombosis in this model.


Herein we describe the largest reported series defining the prevalence of anti-HPF4 in SLE patients with and without persistent aPL. Others have characterised the presence of anti-HPF4 in heterogeneous aPL patients, including primary APS, secondary APS (typically associated with SLE) and aPL without APS.1014 We found the 9% prevalence of anti-HPF4 in aPL-positive SLE patients to be significantly greater than the 1% prevalence in blood donor controls. The 4% prevalence of anti-HPF4 in aPL-negative SLE patients was non-significantly elevated compared with controls. However, this study was likely underpowered to detect differences in prevalence between these groups.

Although the association of anti-HPF4 with HIT is well described,1 the presence of anti-HPF4 without HIT has also been documented, and may be clinically important.5 1014 24 In one study, patients with acute coronary syndrome had pre-existing anti-HPF4 without prior heparin exposure, which imparted increased risk of thrombosis following cardiac catheterisation.5 Another group reported that 13% of patients undergoing cardiac surgery had positive pre-operative anti-HPF4, which was an independent risk factor for major postoperative complications.24 In aPL-positive patients without HIT, it is unclear whether anti-HPF4 are irrelevant bystanders,13 or a potential risk factor for thrombosis.11 15

Here, we demonstrate an association between anti-HPF4 and APS, and particularly APS with arterial thrombosis in our SLE cohort. These associations persist after controlling for demographic characteristics, aPL levels, recent heparin exposure, and recent hospitalisation using multivariate linear regression models. Recently, Martin-Toutain et al did not find a significant difference in the prevalence of anti-HPF4 in patients with primary or secondary APS (20%) compared with aPL without APS (6%).13 However, this study included fewer patients than our study, perhaps explaining the lack of significant difference. In our cross-sectional study, we cannot determine whether anti-HPF4 precede or result from APS, or relate to other genetic or environmental factors associated with APS. Further longitudinal studies are warranted to address this.

Our study also demonstrates significant correlations between IgM anti-HPF4 and IgM aPL, which were preserved in multivariate linear regression models. We found no evidence of cross-reactivity between IgM anti-HPF4 and IgM anti-β2GP1 in four patients, although we did observe partial cross-reactivity of IgM anti-HPF4 with IgM APhL (fig 2). Martinuzzo et al report no evidence of cross-reactivity between anti-HPF4 and IgG aCL or IgG anti-β2GP1 in five patients with positive anti-HPF4 and aCL,14 whereas Bourhim et al report a partial cross-reactivity between IgM anti-HPF4 and IgM anti-β2GP1 in a single patient.10 Since we performed APhL instead of aCL immunoassays, we cannot comment on any potential correlation or cross-reactivity between aCL and anti-HPF4. Our observed cross-reactivity of IgM anti-HPF4 with IgM APhL may represent charge-related cross-recognition of polyanionic substrates. Overall, the association between IgM anti-HPF4 and IgM aPL may be due partly to cross-reactivity, as well as reflect similar endovascular inflammatory microenvironments driving production of these antibodies. We could not determine the contribution of LA activity to anti-HPF4 in our cohort, as platelet-poor plasma samples were unavailable for analyses.

Unlike clinical HIT, we and others find no association of anti-HPF4 with thrombocytopenia in SLE or aPL patients without HIT.13 HIT results from IgG binding to heparin-PF4 complexes, causing platelet activation via cross-linking FcγRII receptors.25 Most of our SLE patients with elevated anti-HPF4 had predominantly IgM anti-HPF4, which do not interact with platelet FcγRII receptors. Although isolated IgM and IgA HPF4 have been reported with HIT,26 recent studies have not corroborated this, and the clinical relevance of non-IgG anti-HPF4 to HIT remains unclear.3 27 28 However, IgM is a potent activator of complement, and we observe an inverse correlation between IgM anti-HPF4 and C4 (but not C3) levels in SLE patients. Although not considered crucial to HIT pathogenesis, complement activation is necessary for aPL-mediated thrombophilia and foetal loss in murine models,2931 and may be pathogenic in patients with APS.3234 A potential role for complement activation by IgM anti-HPF4 contributing to APS merits further exploration.

Interestingly, we found that recent heparin exposure was not associated with positive anti-HPF4 in univariate or multivariate analyses. It is unlikely that anti-HPF4 arising in such patients is due to more distant heparin exposure, given the transient nature of anti-HPF4 in clinical HIT.35 We did observe a trend towards association between recent hospitalisation and anti-HPF4 (table 4). Besides being a surrogate for potential heparin exposure, recent hospitalisation may represent acute illness. This could set up an endovascular inflammatory milieu favouring anti-HPF4 development due to platelet activation and PF4 release in the setting of endogenous heparinoids.2 36 To this effect, four cases of “spontaneous HIT” in the absence of heparin exposure have been reported in patients with recent antecedent inflammatory or invasive events.37

Of the 12 anti-HPF4-positive subjects, two had moderately positive SRAs. Both were aPL-positive SLE patients with prevalent IgG and IgM anti-HPF4. Neither was definitely exposed to heparin within the preceding year. It is intriguing to speculate that subsequent heparin exposure could predispose to thrombosis in these patients.

There are mechanistic similarities between HIT and APS,9 and we postulate that anti-HPF4 and aPL are interrelated, perhaps acting in concert to contribute to APS in SLE patients. Further longitudinal studies are warranted to test this hypothesis, and clarify whether heparin products are safe therapeutic options for SLE patients with aPL and anti-HPF4.


The authors wish to thank Michelle Perna and Shaziya Assur for their assistance with registry data and samples, and Dr Margaret Peterson (supported in part by NIH P30 AR046121) for assistance with statistical analysis.


Supplementary materials


  • Additional data are published online only at

  • Funding: This study was supported by the Mary Kirkland Center for Lupus Research (DA, JES) and the Barbara Volcker Center for Women and Rheumatic Disease (DE). LAM is supported by NIH K23 AR050607-01 and a Clinical Arthritis Investigator Award from the New York State and National Chapter of the Arthritis Foundation. WY is supported by a fellowship from the American Heart Association, Heritage Affiliate. EIP is supported by NIH NHLBI HL067211.

  • Competing interests: None.