| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Clinical Research Unit, Division of Cardiothoracic Surgery, Joseph B. Whitehead Department of Surgery, Rollins School of Public Health, Emory University School of Medicine, Atlanta, Georgia
b Department of Biostatistics, Rollins School of Public Health, Emory University School of Medicine, Atlanta, Georgia
Accepted for publication May 30, 2007.
* Address correspondence to Dr Lattouf, Crawford Long Hospital, 6th Floor, Medical Office Tower, Atlanta, GA 30308. (Email: omar.lattouf{at}emoryhealthcare.org).
Presented at the Forty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 29–31, 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Methods: From 2002 to 2005, 487 cardiac surgery patients with postoperative thrombocytopenia (50% drop in platelet count or absolute count < 100,000/µL) underwent at least one enzyme-linked immunosorbent assay for HIT platelet factor 4 antibodies. Risk factors and outcomes of patients with a positive HIT assay (HIT+) were compared with patients with thrombocytopenia, but without HIT antibodies (HIT-).
Results: 23.2% of patients (113 of 487) were HIT+. Multivariable predictors of HIT included previous percutaneous coronary interventions (odds ratio [OR] = 1.76, p = 0.03), class IV New York Heart Association heart failure (OR = 1.80, p = 0.012), and infectious endocarditis (OR = 3.66, p = 0.0123). Postoperative infections occurred more frequently in HIT+ patients, including sepsis (16.8% versus 9.9%, p = 0.0433) and pneumonia (46.9% versus 23.3 %, p < 0.001). The HIT+ patients also had a higher rate of renal failure requiring hemodialysis (23.0% versus 9.1%, p < 0.001) and acute limb ischemia (15.9% versus 4.3%, p < 0.001). Thirty-day mortality was significantly higher in the HIT+ group (24.8% versus 15.2%, p = 0.019). Postoperative HIT emerged as an independent predictor of renal failure (OR = 1.73, p < 0.001) and thromboembolic complications (OR = 2.39, p = 0.02).
Conclusions: Heparin-induced thrombocytopenia patients are at significantly greater risk of thrombosis, renal failure, and mortality in the postoperative setting. Greater awareness of this devastating problem may allow earlier detection of HIT, with prompt institution of appropriate anticoagulation therapy, which could potentially limit the associated morbidity and mortality.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Although the presence of HIT antibodies after cardiac surgery has been reported to be as high as 50% [8], the incidence of HIT with thrombocytopenia is only 2% to 3% among all cardiac surgery patients. However, among those with HIT and thrombocytopenia, clinically detectable thromboembolic events may develop in as many as 50% [4]. It is well known that patients with detectable HIT antibodies have increased morbidity and mortality after cardiac surgery, but few data exist on which patient population has the highest risk for postoperative HIT. The objectives of this study are to assess the impact of HIT on postoperative morbidity and mortality among cardiac surgery patients as well as to identify preoperative risk factors that may predispose patients to the development of HIT. In addition, we aim to determine if a correlation exists between measured HIT antibody titers and postoperative morbidity and mortality.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
From 2002 through 2005, 6,349 patients underwent coronary artery bypass grafting or valve surgery (including primary operations, reoperations, and concomitant procedures), or both, at Emory University Hospital and Crawford Long Hospital of Emory University. During this period, 518 patients (12.3%) who had at least one enzyme-linked immunosorbent assay (ELISA) test for HIT antibodies (PF4 Enhanced ELISA; GTI Diagnostics, Waukesha, Wisconsin) after cardiac surgery were identified from hospital laboratory records. The HIT assays were typically performed for persistent postoperative thrombocytopenia (platelet count < 100,000/µL), which did not begin to rise after the second postoperative day. Patients undergoing aortic surgery or cardiac transplantation were excluded from the analysis, as were patients who were identified to be HIT positive preoperatively. After these exclusions, the final patient cohort for the basis of this study is 487 patients.
Preoperative risk factors and demographics, intraoperative variables, and postoperative outcomes for these patients were tabulated from prospectively collected data from our institutional cardiac surgery database. The patients were divided into two groups according to their final HIT status: positive HIT assay (HIT+) or negative HIT assay (HIT-). Risk factors and outcomes of patients with a positive HIT assay were compared with patients with thrombocytopenia, but without HIT antibodies.
To identify risk factors for the development of HIT, univariate and multivariate analyses were performed with respect to preoperative risk factors and intraoperative variables utilizing logistic regression methods (Tables 1, 2, and 3).
Any single risk factor that was univariately associated with HIT status at the 
= 0.20 level was entered as an independent variable in a larger multivariable model (with HIT status as the dependent variable). The = 0.20 cutoff was instituted because of the small sample size to prevent the multivariable model from being saturated, thus reducing the risk of inestimable effects. The risk factors that were significant in the multivariable model were then considered independent risk factors for HIT development. Odds ratio (OR) were computed for significant risk factors of HIT.
|
|
|
The heparin-PF4 ELISA is reported with an optical density (OD) value that is indicative of the antibody titer. To identify the HIT titer that optimizes it as a binary predictor of adverse outcomes, an area under the receiver operating characteristic (ROC) curve analysis was performed. The area under the ROC curve is a measure of model discrimination, that is, how well a model separates dichotomies [9]. The area under the ROC curve is bounded by 0.5 and 1.0; values close to 1.0 indicate excellent discrimination and values closer to 0.5 indicate no model discrimination. To achieve this, the cutoff for the highest titer was continually shifted by 0.1 units and reevaluated for discrimination through a ROC curve within a univariate logistic regression model. The HIT titer cutoff (the independent variable in the model) that maximized the area under the ROC curve was determined in this manner. This was performed for three outcomes separately: mortality, thromboembolic complications, and renal failure.
Statistical analyses were performed using SAS Version 9.0 (SAS Institute, Cary, North Carolina). All analyses were preplanned, and primary aims were evaluated at an = 0.05 significance level. All authors had full access to the data under consideration and take responsibility for its integrity.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
|
Univariate Comparison of Outcomes
Infectious complications occurred more frequently in HIT+ patients, including sepsis (16.8% versus 9.9%, p = 0.04), pneumonia (46.9% versus 23.3 %, p < 0.001), and urinary tract infections (17.7% versus 10.7%, p = 0.05; Table 3). The HIT+ patients had a significantly higher incidence of acute limb ischemia (15.9% versus 4.3%, p < 0.001) and renal failure requiring hemodialysis (23.0% versus 9.1%, p < 0.001). Although not statistically significant, there was a trend toward increased incidence of cerebrovascular accident in the HIT+ group (9.7% versus 5.6%, p = 0.12). Thirty-day mortality was significantly higher in the HIT+ group (24.8% versus 15.2%, p = 0.019).
Multivariable Predictors of HIT and Adverse Outcomes
A multivariable logistic regression model was created to identify preoperative characteristics that were predictive of HIT (Tables 4 and 5).
Risk factors that emerged as independent predictors for the development of HIT included previous percutaneous coronary interventions (OR = 1.76, p = 0.03), class IV NYHA heart failure (OR = 1.80, p = 0.012), and infectious endocarditis (OR = 3.66, p = 0.012). To determine if HIT was an independent predictor of death, acute renal failure, and thromboembolic complications, a second multivariable logistic regression model was formed using HIT as an independent variable and adjusting for preoperative and operative covariates (Table 5). Based on this model, HIT was only marginally predictive of mortality (1.68, p = 0.06), but strongly associated with the development of renal failure (OR = 1.73 p < 0.001) and thromboembolic complications (OR = 2.39, p = 0.02).
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
One of the objectives of this study was to establish a preoperative risk factor profile that could be used to identify patients more likely to develop HIT. Preoperative identification of these patients could lead to heightened clinical suspicion and earlier confirmation of diagnosis, and could ultimately result in an early aggressive treatment strategy and reduced associated complications. Based on the multivariable analysis, previous percutaneous coronary intervention, NYHA class IV heart failure, and the presence of infectious endocarditis were independent predictors of HIT postoperatively. These patient subsets should potentially have a preoperative HIT assay as well as serial postoperative HIT antibody titers evaluated to facilitate early detection of HIT before development of clinical signs and symptoms.
Patients with previous percutaneous coronary intervention (PCI) had a nearly twofold increase in the incidence of HIT, likely related to periprocedural heparin exposure. Similarly, patients with NYHA class IV heart failure are potentially more likely to be hospitalized and more frequently undergo procedures (ie, pacemaker implantation or central venous catheterization) that may increase their exposure to heparin. Other studies have confirmed that patients exposed to heparin within 100 days of cardiac surgery have an increased risk of HIT [6]; however, prior PCI and class IV heart failure have not been previously identified as independent risk factors for HIT. As the number of cardiac surgery patients with previous PCI and symptomatic heart failure increases, surgeons must be aware of this potential risk and act promptly if any suspicion for HIT exists. Although it is intuitive that patients with previous PCI and class IV heart failure may have an increase in the incidence of HIT due to previous heparin exposures, it is unclear why endocarditis would be associated with HIT. Although not confirmed by data from this study, one possibility is that patients with endocarditis have a heightened immune response secondary to the infectious process and may be more likely to form HIT antibodies.
The majority of HIT-related research has focused on the associated hypercoagulable state and thromboembolic events. However, in addition to causing platelet activation and aggregation, HIT antibodies have also been shown to activate endothelial cells, neutrophils, and macrophages [12–15]. The expression of tissue factor from the endothelium as well as production of proinflammatory cytokines from activated leukocytes may lead to an inflammatory state as is seen with severe sepsis, ultimately causing multiorgan dysfunction [4]. The high incidence of renal and respiratory failure noted in this study could potentially be attributed to secondary end-organ damage resulting from platelet and endothelial cell dysfunction and the ensuing inflammatory response. Future investigations elucidating the role of the associated inflammatory state in HIT are warranted.
Heparin-induced thrombocytopenia remains predominantly a clinical diagnosis based on the presence of thrombocytopenia and thromboembolic events; however, laboratory confirmation of diagnosis is important to establish the condition and initiate treatment. Several assays are commonly used for this purpose: the PF4-heparin ELISA test used in this study, the serotonin release assay, heparin-induced platelet activation assay, and the platelet aggregation test [16]. In our series, we have found a serially rising antibody titer by the ELISA tests to be useful in making treatment decisions. Patients with borderline antibody titers initially may have a positive result on subsequent assays, and should therefore have serial assays performed until the diagnosis is firmly established or refuted. In this series, among patients with a negative initial assay who had a subsequent second, third, or fourth assay, 28% ultimately had a positive result on follow-up tests. Furthermore, 91% of theses positive results occurred on the second or third assay and 9% occurred on the fourth assay. Therefore, if the clinical suspicion for HIT exists, the assay should be performed a minimum of three times despite a negative initial result. In our practice, if we have a high clinical suspicion for HIT (ie, high pretest probability), we will initiate treatment with alternative anticoagulation despite a negative initial assay. This should be done cautiously, however, as the safety of the alternative anticoagulant medications available is unproven in the postoperative setting and may be associated with hemorrhagic complications.
The antibody titer is further useful in attempting to predict the likelihood of a HIT-related complication in case of a positive result [17]. The ROC curve analysis in this study demonstrated that those patients with an HIT antibody titer greater than 1.6 have a threefold increase in mortality, and those with a titer greater than 0.7 are at particularly increased risk of developing acute renal failure requiring hemodialysis (OR 2.69). Although the risk of HIT associated thrombosis is well established, the correlation between elevated HIT titers and renal failure has not been previously shown. Ideally, these patients should receive alternative anticoagulant therapy as soon as possible barring any absolute contraindications.
This study has several limitations that should be noted. Owing to the retrospective identification of patients from the laboratory database, it is possible that all HIT+ patients were not included in the analysis. Furthermore, although an HIT assay was typically performed for platelet counts below 100,000/µL, a strict definition of thrombocytopenia is not uniformly used among all surgeons in our group. Additionally, 28% of patients with a negative initial assay who had two or more tests performed had a positive result on subsequent tests. However, of the 407 patients with a negative initial assay, 291 had no further tests performed, and they could potentially have been HIT+ had subsequent tests been carried out. That may explain the high mortality rate (15%) seen in the HIT- group and underscores the importance of performing serial HIT assays despite a negative initial result. Finally, the heparin dose used for off-pump coronary artery bypass graft surgery at our institution are not standardized, but varies based on surgeon preference ranging from 10,000 units (with additional dosing to maintain an activated clotting time of greater than 300 s) to a full "pump dose" of 400 units/kg. Although we did not consider the impact of heparin dosing on HIT in this analysis, we plan to do so in a future study with a larger number of isolated coronary artery bypass graft surgery patients comparing on-pump to off-pump coronary artery bypass graft surgery with varying heparin doses.
Our initial observation that HIT was becoming more frequently noted in our patient population prompted us to undertake this analysis. The high morbidity and mortality rate in the HIT positive group in this study has prompted us to become increasingly vigilant with early investigation of postoperative thrombocytopenia. With the aging patient population, the likelihood of previous heparin exposure from prior hospitalizations, surgical procedures, and percutaneous coronary interventions, one anticipates that HIT will become increasingly problematic among patients undergoing cardiac surgery. Our awareness should be heightened so that the diagnosis of HIT can be confirmed as soon as possible and treatment with alternative anticoagulants initiated before the onset of associated thromboembolic complications.
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
DR KERENDI: What we have done in recent months, especially after we started looking at these numbers, is to make some changes in our practice. First, a negative initial assay does not necessarily mean that the patient is not going to be positive for HIT. Based on this study, 28% of patients with a negative initial assay who had a subsequent assay were ultimately positive, and so we concluded that a minimum of three assays should be sent.
Secondly, we have not proven this with data from this study, but I think trying to identify these patients at an earlier stage before the development of symptoms would probably be beneficial in terms of avoiding these complications in the postoperative setting.
DR PUSKAS: Will we see increasing numbers of HIT-positive patients given that more and more of our coronary patients, for instance, have had multiple exposures to heparin in the catheterization laboratory?
DR KERENDI: In examining the preoperative factors, previous percutaneous coronary interventions stood out as an independent predictor of HIT, and I think that would suggest that as the rate of previous PCI increases, we are likely to see more patients with HIT in the future.
DR DIMITRI NOVITZKY (Tampa, FL): Very nice presentation. Originally I had a patient who had an acute dissection of the ascending aorta. She is 87 years old. I did the repair, and she did very well. She woke up, she was moving all extremities, she was coherent, and on the third day she started becoming confused, and then small spots appeared in the periphery. So immediately my mind jumped up that this is HIT. So the test for HIT was negative. I called hematology. They said, no, do not start anything and continue persisting with your second test, third test, and then I called another hematologist, and eventually we started, with our hematology consult, argatroban. Because every time we were giving platelets, the platelets would go to 30,000, and the lesions in the periphery went up. The question is, in the absence of antibodies, would you start argatroban? How would you follow and monitor the patient? This patient became positive on the fifth or sixth test that we did. She had negative blood cultures, she developed renal failure, as you mentioned. But the main issue, I think, is once you start a drop of platelets and it is not explained by diffuse intravascular coagulation, it is not explained by the presence of sepsis, it is not explained by other events, I think it is extremely, extremely important to consider the patient as having HIT in the absence of positive antibodies.
DR KERENDI: I agree. The ELISA antibody test that we conduct at most centers is pretty sensitive and nonspecific, but occasionally we do get patients who don’t test positive, and I think if the suspicion is high enough in the absence of other causes for thrombocytopenia, it is reasonable to initiate treatment. That should be done with some trepidation, however, because in the postoperative setting, most of these anticoagulants are unproven and the risk of hemorrhagic complications is not minimal.
DR MAHMOOD MIRHOSEINI (Germantown, WI): Mr Chairman already alluded to the patients who we have to be aware of for HIT are the patients who have gone through multiple catheterizations, angioplasty, et cetera. Historically this was first described by Dr Hussey from our Medical College of Wisconsin, and we had a patient in the service of the late Dr LePley, who was a member of this Society, and this patient had multiple gangrene toes and I think postoperative, postcardiac surgery, valve surgery developed. As a result, Dr Hussey, who was the hematologist at the Medical College of Wisconsin, the service, Dr Quick did the Quick Test, et cetera, and discovered the sensitivity of heparin, the platelets to heparin, and it is postulated that the patients who take subcutaneous heparin become more sensitive to HIT. Thank you.
DR KERENDI: Thank you for those comments. It is true that patients with the use of subcutaneous heparin postoperatively are at increased risk of HIT, and it is slightly decreased with the use of low molecular weight heparin and other alternatives.
DR ALAN SPOTNITZ (New Brunswick, NJ): When you get a positive test preoperatively, what do you do intraoperatively to address the issue?
DR KERENDI: I think there are several ways to handle that. With our practice, what we have done is in patients who don’t necessarily need to have surgery in an urgent manner, we have put off those operations and reevaluated with a repeat assay in 3 months, and typically the antibody titers get to a low enough level where you can actually administer heparin safely to those patients. We have done several patients who have needed urgent operations with the use of bivalirudin or other agents, but as I stated earlier, those drugs are not proven, and we have had several complications with bleeding postoperatively in those situations.
DR SPOTNITZ: As a follow-up, when you have waited the 3 months and use heparin for the surgery, what do you do postoperatively for a patient who has a mechanical valve in place or some other indication for early anticoagulation?
DR KERENDI: There are relatively few patients like that, but generally we try to avoid heparin and start them on coumadin without any other anticoagulation.
DR KEITH B. ALLEN (Indianapolis, IN): I enjoyed your talk. Our group has had some experience with prospective studies using alternatives to heparin. Do you at Emory feel comfortable enough with your predictive models that you have made the leap in practice change to utilize alternatives to heparin like bivalrudin to try and avoid this complication that you are seeing in 15% to 20% of your patients?
DR KERENDI: We haven’t changed practice in that sense yet. I think what we might do before jumping to the alternative anticoagulants is to perhaps get a preoperative HIT assay, and in patients who are positive or marginally positive, perhaps consider those options, but we have not done that yet.
DR GLENN J. R. WHITMAN (Philadelphia, PA): Given the morbidity and the mortality associated with HIT, a lot of us have always thought that if we suspect HIT, we try as hard as we can to prevent it from becoming thrombotic thrombocytopenia, HITT, and regardless of the ELISA assay, if we suspect it, we do start a direct antithrombin inhibitor.
I have two questions. One, the ELISA assay is qualitative, not quantitative, and furthermore, it is expensive and time consuming. Does your institution simply repeat the ELISA assay or does it do a more specific assay for the antibody? Does it do it daily, because in many institutions it is only done once a week and sometimes, if you’re lucky, twice a week? And when you suspect HIT, do you wait for symptoms of HITT or do you have a different kind of protocol for when you start a direct thrombin inhibitor?
DR KERENDI: I will answer your last question first. There is no protocol in place for that. The ELISA assays are run three times a week at our institution. We do not do any of the more specific tests like the platelet activation assays or serotonin release assay. We do repeated ELISA assays. We do get a qualitative optical density value from the ELISA assay, which is what I referred to in the presentation, but we don’t know the usefulness of that yet. I think if we strongly suspect HIT, there are times when we initiate treatment with one of the direct thrombin inhibitors without having the laboratory confirmation. But again, for these patients in the postoperative setting, these drugs are not proven and the risk of hemorrhagic complications is unknown.
DR ARTHUR J. CRUMBLEY (Charleston, SC): It is my understanding that the platelet factor IV assay, which you call the HIT assay, is very sensitive but not specific at all, being positive in many inflammatory situations, endocarditis and some of your risk factors. How did you make the diagnosis of HIT without a confirming serotonin release assay, since your screening test is a test of platelet activation rather than antibody presence?
DR KERENDI: You are right, it is a nonspecific assay but it is sensitive. Unfortunately, we are limited at our institution by the assay that is run by our laboratory, and so I think if we get a positive assay in a patient with a high pretest probability, then we have used that as a confirmatory test and we proceed to anticoagulate those patients.
DR CRUMBLEY: And you screened your patients preoperatively before heparin exposure with a platelet factor IV assay?
DR KERENDI: No. These were all patients who postoperatively either had a decrease in their platelet count or had a thrombotic manifestation.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A V Mattioli, L Bonetti, U Carletti, G Ambrosio, and G Mattioli Thrombotic events in patients with antiplatelet factor 4/heparin antibodies Heart, August 15, 2009; 95(16): 1350 - 1354. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Shantsila, G. Y. H. Lip, and B. H. Chong Heparin-Induced Thrombocytopenia: A Contemporary Clinical Approach to Diagnosis and Management Chest, June 1, 2009; 135(6): 1651 - 1664. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Pappalardo, A. Greinacher, and S. Selleng Postoperative Complications After Cardiac Surgery and HIT: A Word of Caution Ann. Thorac. Surg., September 1, 2008; 86(3): 1054 - 1055. [Full Text] [PDF]
|
|
|
|
|
|
F. Kerendi, V. H. Thourani, J. D. Puskas, R. A. Guyton, and O. M. Lattouf Reply Ann. Thorac. Surg., September 1, 2008; 86(3): 1055 - 1055. [Full Text] [PDF]
|
|
|
|
 |
|
 K. Knobloch, A. Gohritz, M. Spies, and P. M. Vogt HIT in OPCAB surgery Interactive CardioVascular and Thoracic Surgery, June 1, 2008; 7(3): 382 - 382. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
