This analysis was designed to (1) examine the impact of heparin-induced thrombocytopenia (HIT) on contemporary cardiac surgical practice and (2) describe the results of a protocol designed for early identification of the presence of the immune mechanisms involved. Consecutive patients who underwent cardiac surgery were screened postoperatively for thrombocytopenia. Patients with thrombocytopenia were tested for antiplatelet factor 4 (PF4)/heparin antibodies by ELISA and clinical evidence of thrombosis sought. Demographics, co-morbidities, operative details, and outcomes were abstracted from the departmental registry. Of 14,415 consecutive patients undergoing cardiac surgery, 1,849 patients (13%) had thrombocytopenia. Of them, 277 patients (15%) had PF4/heparin antibodies and 76 patients (4%) had both antibodies and clinical thrombosis. Antibodies were more frequent: (1) in women (p = 0.01), (2) in patients with an increased body mass index (p <0.01), and (3) in patients with clinical heart failure before surgery (p <0.01). Thirty-day mortality was greatest among the 76 patients with the triad of thrombocytopenia, antibodies, and clinical thrombosis (30%). Of the 1,849 patients with thrombocytopenia, the presence of PF4/heparin antibodies was an independent predictor of 30-day mortality (odds ratio 2.09, 95% CI 1.46 to 2.49; p <0.001). HIT remains an infrequent but very serious complication of heparin therapy in contemporary cardiac surgical practice. The possibility that the presence of HIT antibodies in patients with thrombocytopenia independently increases operative mortality deserves further study.
Heparin-induced thrombocytopenia (HIT) is an infrequent but potentially fatal complication of heparin therapy. The drug binds to circulating platelet factor 4 (PF4). Patients undergoing cardiac surgery appear to be at high risk for HIT because the exposure to heparin with cardiopulmonary bypass (CPB) is nearly universal. This study has 4 purposes: (1) with HIT in a large, contemporary, “real world” cardiac surgery practice and with our screening strategy for early detection of HIT; (2) to ascertain the frequency of immunoglobulin (ImG) anti-PF4/heparin antibodies; (3) to identify preoperative and intraoperative risk factors for antibody formation; and (4) to analyze the excess mortality conferred by these antibodies.
Methods
The study protocol was approved by the Institutional Review Board of MedStar Research Institute. The standardized protocol for our department requires that all patients have a preoperative platelet count to be repeated after surgery at least every other day until discharge or, in the event of a prolonged postoperative stay, up to 14 days after surgery. Because of the expected substantial platelet count decrease after cardiac surgery, the highest platelet count in the postoperative period, rather than the preoperative count was designated as the reference level, and “thrombocytopenia” was defined as: either a decrease from this level of ≥50% or an absolute count <80,000 mm 2 . A patient in whom either occurs is considered to be an “HIT suspect.” In such patients, the presence of HIT antibodies was sought by means of a commercially available enzyme-linked immuno sorbent assay (ELISA; PF4 Enhanced assay; Gen-Probe, San Diego, California). In the face of a negative ELISA but a strong clinical suspicion of HIT, a repeat ELISA or a serotonin release assay was conducted.
Prophylactic use of neither low molecular weight heparin (LMWH) nor unfractioned heparin is routine in the postoperative period, but in selected patients, prophylactic low molecular weight or unfractioned heparin was given beginning on postoperative day 1. In patients with an ELISA test with positive results, an alternative was used.
We analyzed data from the comprehensive cardiac surgery registry maintained prospectively by the Department of Cardiovascular Surgery at the Washington Hospital Center. A data coordination center maintains this registry contemporaneously with the operative procedure as part of routine clinical practice. Preoperative, intraoperative, and postoperative data are collected. Patients are contacted at 30 days after hospital discharge by data center staff for clinical status. Data fields are defined in accordance with the Society of Thoracic Surgeons national database.
All patients operated in this institution from January 1, 2003, to June 30, 2011, were reviewed. Excluded from the analysis were 265 patients who had heart transplantation, placement of a ventricular assist device, tricuspid valve surgery, or endovascular surgery. The 14 patients who received no heparin in connection with their cardiac surgery were also excluded. Accordingly, the study included 14,415 consecutive patients who underwent coronary artery bypass grafting, valve surgery, combined coronary artery bypass grafting and valve surgery, or surgical repair of the aorta.
When the ELISA was positive (optical density >0.45), clinical signs of thrombosis (e.g., deep venous thrombosis, pulmonary embolus, and stroke) were specifically sought. Occult venous thrombosis in the lower extremities was sought by means of a Doppler ultrasound examination.
Unfractioned heparin (350IU/kg) was given to achieve an activated clotting time of ≥480 seconds in all patients. An additional 2,000 unit of heparin was administered if required. CPB was conducted at 2.4 L × min × m −2 with a nonpulsatile pump. Moderate systemic hypothermia (30°C to 34°C) was used. At the completion of CPB, protamine was administered to reverse the heparin effect. Protamine dosage was 1 mg per 100 units of heparin. Aprotinin was rarely used.
Continuous data are reported as mean and SD and compared by means of a 2-sided Student t test when normally distributed. Nonparametric data are reported as median and interquartile range and are compared by means of a Wilcoxon 2-sided test. Categorical data are reported as frequency and percentage and compared by means of a chi-square test. Statistical significance was set at p <0.05.
Multivariate logistic regression analysis was used. Two models were created. One was designed to identify the independent baseline clinical and operative characteristics of patients with thrombocytopenia that predicted the occurrence of HIT antibodies. The second was undertaken to assess the possibility that the appearance of HIT antibodies was independently associated with 30-day mortality. For both models, baseline characteristics univariately associated at a p <0.2 level with the dependent variable were entered into the model.
Results
Of the 14,415 patients, postoperative thrombocytopenia by protocol definition was detected in 1,849 patients (13%). Table 1 compares the baseline and operative details of the entire population with regard to whether or not thrombocytopenia by the protocol definition occurred. Numerous clinical and operative features were significantly different. Most of these differences can be associated with the presence of more serious co-morbid conditions or a more complex surgical procedure. In many, a longer period on CPB was identified.
Variables | Thrombocytopenia | Total (n=14,415) | p-value | |
---|---|---|---|---|
Yes (n=1,849) | No (n=12,566) | |||
Age (years, M±SD) | 69±12 | 64±11 | 65±12 | <0.01 |
Women | 766 (17%) | 3,819 (83%) | 4,585 | <0.01 |
Black | 562 (13%) | 3,631 (87%) | 4,193 | 0.3 |
Body mass index (kg/m 2 , M±SD) | 28±6.0 | 29±9.5 | 29±9.1 | <0.01 |
Renal insufficiency | 111 (21%) | 430 (80%) | 541 | <0.01 |
Diabetes mellitus | 605 (12%) | 4,430 (80%) | 5,035 | 0.1 |
Dyslipidemia | 1,249 (12%) | 8,964 (88%) | 10,213 | 0.2 |
Infectious endocarditis | 74 (4%) | 305 (2%) | 379 | <0.01 |
Corticosteroid therapy | 113 (6%) | 430 (3%) | 543 | <0.01 |
Peripheral artery disease | 358 (15%) | 1,975 (85%) | 2,333 | 0.01 |
Cerebrovascular disease | 324 (16%) | 1,720 (84%) | 2,044 | <0.01 |
Prior CABG or PCI | 587 (15%) | 3,458 (86%) | 4,045 | 0.01 |
Myocardial infarction | 627 (12%) | 4,673 (88%) | 5,300 | 0.1 |
Heart failure | 556 (21%) | 2,104 (79%) | 2,660 | <0.01 |
Ejection fraction (M±SD) | 44.6 ±13.3 | 48.0±11.7 | 47.9±11.8 | <0.01 |
LV EF <35% | 398 (18%) | 1,835 (82%) | 2,233 | <0.01 |
Prior Atrial fibrillation | 308 (23%) | 1,037 (77%) | 1,345 | <0.01 |
IABP used | 164 (19%) | 711 (81%) | 875 | <0.01 |
Platelet count (×1000, M±SD) | 187±76 | 228±70 | 222±72 | <0.01 |
Hematocrit (%, M±SD) | 37±5 | 38±5 | 39±7 | <0.01 |
Creatinine (mg/dl, M±SD) | 1.5±1.5 | 1.3±1.3 | 1.3±1.3 | <0.01 |
Heparin≤48hrs(UF+LMWH) | 384 (21%) | 2,084 (17%) | 2,468 | <0.01 |
Elective surgery | 807 (11%) | 6,465 (89%) | 7,272 | 0.5 |
On-pump surgery | 1,618 (17%) | 8,212 (84%) | 9,830 | <0.01 |
Isolated CABG | 758 (7%) | 9,646 (93%) | 10,404 | <0.01 |
CABG + valve surgery | 423 (32%) | 892 (68%) | 1,315 | <0.01 |
Valve surgery | 561 (25%) | 1,698 (75%) | 2,259 | <0.01 |
Aortic surgery | 107 (25%) | 330 (76%) | 437 | <0.01 |
Pump time (minutes, M±SD) | 98±47 | 80±38 | 83±40 | <0.01 |
Heparin in OR (unit×1000, M±SD) | 43±14 | 47±14 | 43±14 | <0.01 |
Protamine (mg, M±SD) | 402±144 | 399±130 | 400±132 | 0.3 |
All patients with thrombocytopenia were tested for anti-PF4/heparin antibodies. Of the 1,849 patients, antibodies were found by ELISA testing in 277 patients (15%; 2% of the total population). Table 2 compares baseline characteristics and operative details of the patients with thrombocytopenia with regard to the presence of such antibodies. They were more frequent: (1) in women (p = 0.01), (2) in patients with an increased body mass index (BMI; p <0.01), and (3) in patients with clinical heart failure before surgery (p <0.01). Those with antibodies on average received a slightly larger amount of heparin in the operating room (p = 0.01) and more often had urgent or emergent surgery (p <0.01). Moreover, operations directly related to the aorta were univariately associated with the postoperative presence of antibodies (p = 0.02).
Variable | ELISA Antibodies | |||
---|---|---|---|---|
Present (n=277) | No (n=1,572) | Total (n=1,849) | p-value | |
Age (years, M±SD) | 69±12 | 69±12 | 69±12 | 0.5 |
Women | 133 (48%) | 633 (40%) | 766 (41%) | 0.01 |
Black | 96 (35%) | 467 (30%) | 563 (30%) | 0.3 |
Body mass index (kg/m 2 , M±SD) | 29±7 | 27±6 | 28±6 | <0.01 |
Renal insufficiency | 17 (6%) | 94 (6%) | 111 (6%) | 1.0 |
Diabetes mellitus | 94 (34%) | 511 (33%) | 605 (33%) | 0.7 |
Dyslipidemia | 177 (64%) | 1,072 (68%) | 1,249 (68%) | 0.2 |
Endocarditis | 16 (6%) | 58 (4%) | 74 (4%) | 0.1 |
Corticosteroid therapy | 20 (7%) | 93 (6%) | 113 (6%) | 0.5 |
Peripheral art disease | 60 (22%) | 298 (19%) | 358 (19%) | 0.4 |
Cerebrovascular disease | 51 (19%) | 273 (17%) | 324 (18%) | 0.7 |
Prior CABG or PCI | 85 (31%) | 502 (32%) | 587 (32%) | 0.7 |
Myocardial infarction | 102 (37%) | 525 (33%) | 627 (34%) | 0.3 |
Heart failure | 105 (38%) | 451 (29%) | 556 (30%) | <0.01 |
LV EF <35% | 66 (24%) | 332 (21%) | 398 (22%) | 0.3 |
Ejection fraction (M±SD) | 45±13 | 46±13 | 46±12 | 0.04 |
Prior Atrial fibrillation | 59 (21%) | 249 (16%) | 308 (17%) | 0.03 |
Preoperative IABP insertion | 27 (10%) | 137 (9%) | 164 (9%) | 0.7 |
Platelet count (×1000, M±SD) | 220±119 | 183±64 | 187±75 | <0.01 |
Hematocrit (%, M±SD) | 37±6 | 37±5 | 37±5 | 0.9 |
Creatinine (mg/dl, M±SD) | 1.5±1.5 | 1.3±1.3 | 1.4±1.3 | <0.01 |
Prior Heparin&LMWH treatment ≤48 hours | 56 (20%) | 328 (21%) | 384 (20.8%) | 0.9 |
Elective surgery | 97 (35%) | 710 (46%) | 807 (44%) | <0.01 |
On-pump surgery | 234 (85%) | 1,384 (88%) | 1,618 (88%) | 0.1 |
Heparin in OR (unitsx1000,M±SD) | 44±14 | 42±13 | 42±13 | 0.01 |
Protamine (mg, M±SD) | 403±144 | 399±130 | 403±144 | 0.3 |
Isolated CABG surgery | 105 (38%) | 653 (42%) | 758 (41%) | 0.3 |
CABG + valve surgery | 72 (26%) | 351 (22%) | 423 (23%) | 0.2 |
Valve surgery | 75 (27%) | 486 (31%) | 561 (30%) | 0.2 |
Aortic surgery | 25 (9%) | 82 (5%) | 107 (6%) | 0.02 |
Notably, our data do not demonstrate an increased frequency of antibodies in patients with postoperative thrombocytopenia with regard to treatment with heparin either within the 48 hours preceding the procedure or with a history of a previous cardiac procedure during which an exposure to the drug could reasonably have occurred.
With regard to preoperative laboratory studies, those who were found to have antibodies had, on average, a lower preoperative platelet count (p <0.01). Renal insufficiency as evidenced by an elevated preoperative serum creatinine level was also more frequent (p <0.01).
After adjustment for all available potentially confounding factors ( Table 3 ), the increased prevalence of HIT antibodies remained significant only for greater BMI and lower preoperative platelet count.
Odds Ratio | 95% Confidence Limits | p-value | |
---|---|---|---|
Body Mass Index (per unit) | 1.05 | 1.03-1.07 | <0.01 |
Preoperative platelet count | 0.99 | 0.99-0.99 | <0.01 |
Heart failure | 1.38 | 0.98-1.94 | 0.1 |
Operation type | |||
Isolated Coronary grafting bypass surgery (reference) | 1.0 | ||
Isolated Valve Surgery | 1.46 | 0.95-2.25 | 0.1 |
Coronary Bypass + valve Surgery | 1.74 | 1.13-2.67 | 0.01 |
Surgery of Aorta | 2.86 | 1.49-5.50 | <0.01 |