Dr Horne, Rm. 2C306, Building 10, NIH, Bethesda, MD 20892, USA (tel: 0–11–301–402–2457; fax: 0–11–301–402–2046; e-mail: firstname.lastname@example.org).
Abstract. Kadidal VV, Mayo DJ, Horne MK (National Institutes of Health, Bethesda, MD 20892, USA) Heparin-induced thrombocytopenia (HIT) due to heparin flushes: a report of three cases (Case Report). J Intern Med 1999; 246: 325–329.
Three cases of heparin-induced thrombocytopenia (HIT) are reported that were provoked by daily heparin flushes of central venous access devices. Each case had confounding features that delayed recognition of the problem. A review of the literature revealed only 29 previously reported cases of HIT secondary to heparin flushes. However, the true incidence of this problem is unknown. A high index of suspicion and confirmatory laboratory tests are necessary to make the diagnosis.
Heparin is the most common cause of immune-mediated thrombocytopenia related to drugs. One to 3% of patients who receive therapeutic intravenous doses of heparin develop this complication . Although much smaller doses of heparin are also recognized to cause thrombocytopenia, only a limited number of cases have been described in the literature, and the true incidence is unknown [2–12]. Recently, however, during a period of about 2 years we encountered three patients with both clinical and laboratory evidence of heparin-induced thrombocytopenia (HIT) secondary to daily heparin flushes of central venous catheters. Here we describe these cases and review the others that have been reported.
A 61-year-old woman with a myelodysplastic syndrome was hospitalized for treatment with antithymocyte globulin (ATG) according to an experimental protocol. On admission her platelet count was 209 000 µL–1. A double lumen Hickman™ (Davol Inc., Cranston, RI) catheter was inserted on hospital day 2 ( Fig. 1). Each lumen of the catheter was flushed with 2 mL of saline containing 100 U mL–1 porcine heparin each time the catheter was used, at least once daily. ATG was given continuously on hospital days 5, 6, 7, and 8. By day 8 the patient’s platelet count had dropped, and she received a platelet transfusion on day 9. Although transient thrombocytopenia had been anticipated in response to ATG, the platelet count failed to show recovery by day 10, and additional causes of thrombocytopenia were considered. The patient was afebrile and had no other signs of sepsis. At that time she was taking meperidine, diphenhydramine, famotidine, lorazepam, demerol, donnatol acetaminophen, benadryl, and the heparin flushes of her catheter. Serum obtained on day 10 was shown by ELISA and by a serotonin-release assay (SRA) to contain heparin related antibodies ( Table 1). Heparin was discontinued on day 11. Subsequently the platelet count began rising immediately. The patient was discharged on day 15 with a platelet count of 124 000 µL–1.
Table 1. Laboratory tests for the diagnosis of HIT in Cases 1, 2 and 3
A positive result requires the combination of >10% serotonin release with the low dose of heparin, <5% release with high dose, and <4% with the low dose plus the Fc-blocking antibody (see Laboratory methods).
b The limit of normal is 0.300 (see Laboratory methods).
A 44-year-old woman with metastatic ovarian carcinoma was hospitalized because of subclavian vein thrombosis associated with a Hickman™ catheter.
The catheter had been inserted 22 days prior to her admission and had been flushed daily with 2 mL of heparinized (porcine) saline (100 U mL–1) at least once a day ( Fig. .2). Two days after insertion of the catheter, the patient received her first dose of chemotherapy, which included cisplatin, paclitaxel, and cyclophosphamide. Eight days later her platelet count had fallen from 228 000 µL–1 to 111 000 µL–1 but returned to baseline within 11 days. On admission to the hospital the patient’s platelet count was 256 000 µL–1. Venography showed total obstruction of a short segment of the right subclavian vein proximal to the junction of the internal jugular vein, and the patient began a therapeutic dose of intravenous porcine heparin (25000 U day–1). Twenty-five milligrams of recombinant tissue plasminogen activator (Genentech, San Francisco, CA) was injected into the thrombus on the second and third hospital days . By the hospital day 4 the patient’s platelet count had fallen to 121 000 µL–1. Heparin was discontinued, and the catheter was removed. Blood drawn on admission (before therapeutic-dose heparin) and on hospital day 4 both tested positive for heparin related antibodies by SRA and ELISA ( Table 1). A follow-up venogram 4 days after the heparin was stopped showed that the subclavian thrombus had extended throughout the length of the right axillary and brachial veins.
A 63-year-old woman with metastatic ovarian carcinoma presented to the clinic complaining of right arm swelling. Two and one-half weeks prior to her visit a Hickman™ catheter had been inserted in her right subclavian vein, and she had received cisplatin, paclitaxel, and cyclophosphamide. During the subsequent 3 weeks she flushed her catheter daily with 2 mL of heparinized saline (100 U mL–1). Ten days after starting her chemotherapy the patient’s platelet count had fallen as expected from 329 000 µL–1 to 73 000 µL–1. However, it then continued falling to a nadir of 20 000 µL–1 on the day she presented with her arm swelling. Right subclavian vein thrombosis was documented by venography. A platelet transfusion was given and a therapeutic dose of intravenous heparin was started. However, because the platelet count failed to rise significantly after the transfusion, the possibility of HIT was considered. Heparin was stopped and replaced with danaparoid (Organon, Oss, the Netherlands). A serum sample obtained before starting therapeutic-dose heparin tested positive for heparin-related antibodies ( Table 1). On danaparoid the patient’s platelet count rose to over 100 000 µL–1 within 3 days.
Sera from all of the patients were tested for heparin-associated antibodies by an ELISA (Asserachrom HPIA; American Bioproducts, Parsippany, NJ) that detects IgG, IgA and IgM. The results of this assay are in units of absorbance at λ490. The upper limit of normal for this assay in our laboratory is 0.300, which is three standard deviations above the mean of values given by the sera of 20 normal volunteers.
The patient sera were also tested for heparin-dependent platelet stimulating activity with a 3H-serotonin-release assay (SRA) . For this, washed normal platelets are loaded with 3H-serotonin and incubated with 0.1 U mL–1 or 100 U mL–1 heparin plus patient serum that has undergone complement inactivation by heating at 56 °C for 30 min. In our laboratory a positive result consists of >10% serotonin release with the low dose of heparin and <5% release with the high dose. In addition, the stimulation with 0.1 U mL–1 must be blocked to <4% by a monoclonal antibody against the platelet Fc-receptor (IV.3; Medarex, West Lebanon, NH). Sera from 10 normal donors and from 30 thrombocytopenic patients who did not have HIT consistently caused <10% serotonin release in the presence of 0.1 U mL–1 or 100 U mL–1 heparin.
To determine whether the ATG administered to Case 1 could give a false positive result in either of these assays, a serum sample was obtained from another patient during ATG administration and tested in both the SRA and the ELISA. The results were negative.
The MEDLINE data base (United States National Library of Medicine, Bethesda, MD) was searched for cases of HIT due solely to heparin flushes reported between 1968 and July, 1998. We found 29 cases of HIT attributed to heparin flushes of venous or arterial catheters reported between 1986 and 1997 ( Table 2). Nineteen of these were included in reports of larger series of HIT cases that provided no details about the individuals who had received only heparin flushes [4, 5, 12]. In 25 patients the clinical diagnosis of HIT was confirmed with laboratory tests [3–6, 8, 10–12]. Only five reports included clinical details of laboratory-confirmed cases [3, 8, 10, 11].
Table 2. Literature review of HIT secondary to heparin flushes
During a period of approximately 2 years we saw three patients with both clinical and laboratory evidence of HIT due to heparin flushes of central venous access devices. All of the cases had confounding features. Thrombocytopenia in our Case 1 was initially an expected effect of the ATG she received for her myelodysplastic syndrome. It was only when the platelet count failed to recover promptly that additional causes were entertained. Similarly the initial thrombocytopenia in Cases 2 and 3 was consistent with an effect of chemotherapy. In Case 2 the platelet count actually recovered completely whilst the patient was receiving heparin flushes but then fell immediately when a higher dose of heparin was started.
Because the clinical manifestations of HIT can be misleading, laboratory tests are often important in confirming the diagnosis, as was true for our cases. The older tests for HIT are based upon the ability of antibody from HIT patients to stimulate either platelet aggregation or platelet release of radiolabelled serotonin in the presence of therapeutic heparin concentrations (e.g., 0.1–0.5 units mL–1) but not in the presence of excess heparin added in vitro (100 U mL–1) [14, 15]. Recently it was discovered that the target of the HIT antibody is actually a complex of heparin with platelet factor-4 (PF4), which is normally present in only nanogram concentrations in the plasma but is released from platelets when they become activated . The heparin-PF4-antibody complex attaches to platelets via its heparin component and stimulates the cells . High concentrations of free heparin prevent this attachment and therefore do not lead to aggregation or serotonin release in the in vitro assays. With the newly recognized role of PF4 in HIT pathogenesis ELISAs have been developed using heparin-PF4 complexes as the target antigen. These assays are much less cumbersome than the older tests and are much more widely available. They also appear to be more sensitive .
All three of our patients had demonstrable HIT antibodies by both an ELISA and the serotonin-release assay. Stored serum collected from Cases 2 and 3 immediately before they began therapeutic doses of heparin revealed that these antibodies were already present, the result of exposure to heparin flushes during the previous 3 weeks. This pre-sensitization was consistent with the immediate fall in platelet count following initiation of the higher dose of heparin in Case 2 .
The frequency with which heparin flushes lead to HIT is unknown. Although in one large series of cases 14% of the patients with HIT were reported to have received only heparin flushes, we could find only 10 case reports that included any clinical details, and only five of these were confirmed by laboratory tests ( Table 2). During the period when our three cases of HIT were identified, approximately 150 Hickman™ catheters were inserted at our institution and received daily heparin flushes. Although this evidence suggests a relatively low incidence of flush-related HIT, some cases may escape clinical detection, as our second case would have if she had not eventually received larger doses of heparin. The true incidence of flush-related HIT must be determined in prospective studies that follow not only the platelet count but also the results from ELISAs for the HIT antibody.
Even if these cases are rare, the widespread use of heparin flushes suggests that physicians are likely to encounter this problem. Furthermore, since patients who are receiving heparin flushes usually have serious medical conditions, the appearance of thrombosis or thrombocytopenia may not seem surprising. Therefore, a high index of suspicion and reliance upon confirmatory laboratory tests are important in establishing the diagnosis. Simply discontinuing heparin is often inadequate treatment for HIT, as our Case 2 illustrates by the extension of her thrombus . Fortunately, alternative anticoagulants are available and should be used to prevent progressive thrombosis related to HIT [21, 22].
The authors received expert technical assistance from Ms Ann Cuillinane, who performed the SRAs and ELISAs for the patients we have reported, and help from Dr Candido E. Rivera in conducting the literature search.
Received 29 September 1998; accepted 11 February 1999.