Christopher M. Lehman, MD, 1,3 Jonathan A. Rettmann, MD, 2 Lori W. Wilson, MS, MT(ASCP), 3 and Boaz A. Markewitz, MD 2. Abstract

Transcription

1 Coagulation and Transfusion Medicine / ANTI-XA HEPARIN ASSAYS IN MICU PATIENTS Comparative Performance of Three Anti Factor Xa Heparin Assays in Patients in a Medical Intensive Care Unit Receiving Intravenous, Unfractionated Heparin Christopher M. Lehman, MD, 1,3 Jonathan A. Rettmann, MD, 2 Lori W. Wilson, MS, MT(ASCP), 3 and Boaz A. Markewitz, MD 2 Key Words: Unfractionated heparin; Activated partial thromboplastin time; aptt; Anti factor Xa assay; Antithrombin; Medical intensive care patients DOI: /8E3U7RXEPXNP27R7 Abstract The availability of automated anti-xa heparin assays provides the opportunity to manage patient unfractionated heparin levels directly, rather than by the activated partial thromboplastin time. Because critically ill patients can acquire an antithrombin deficiency, we compared the performance of 3 anti-xa heparin assays, 1 with and 2 without antithrombin supplementation, by analyzing in vitro aliquots of plasma with defined antithrombin levels and specimens from intensive care patients receiving intravenous heparin therapy. Heparin concentration recovery, in vitro, was dependent on the plasma antithrombin concentration for all 3 assays. The antithrombinsupplemented assay demonstrated improved heparin recovery in direct correlation to the heparin concentration in the plasma. The greatest effect of antithrombin supplementation occurred when the antithrombin level dropped below 40%, a level present in only 5% of the patient specimens. Analysis of patient specimens demonstrated significant correlation among the 3 assays. Classification of the clinical adequacy of patient heparin levels showed agreement of 80% or more between the antithrombin-supplemented and nonsupplemented assays. The antithrombinsupplemented assay did not significantly improve clinical usefulness. With the introduction of new anticoagulants with predictable, weight-based dosing regimens that do not require routine monitoring (eg, low-molecular-weight heparin) and that are not readily reversed, the indications for intravenous heparin therapy ultimately may be limited to clinical environments in which rapid reversal of anticoagulant effect is required, eg, the medical intensive care unit (MICU). Anticoagulation with intravenous heparin requires regular monitoring to ensure adequate anticoagulation and avoidance of bleeding complications. 1 The activated partial thromboplastin time (aptt) has been the standard assay for monitoring intravenous heparin therapy for decades because assays that measure the plasma heparin concentration have not been routinely available. However, the availability of modern anti activated factor X (anti-xa) heparin assays on automated coagulation analyzers provides the opportunity to bypass the indirect aptt measure and to monitor heparin levels more directly by assessing the activity of the heparin-antithrombin complex on the amidolytic activity of factor Xa. Because current recommendations for validating local PTT therapeutic ranges rely on measurement of heparin levels using anti-xa assays, 1 direct measurement of heparin levels should avoid the known error inherent in deriving populationbased aptt therapeutic ranges and applying them to individuals. 2,3 In fact, 1 study suggests that monitoring heparin concentrations results in more rapid attainment and maintenance of therapeutic heparin anticoagulation than the aptt and may be cost-effective. 4 However, that study did not focus on ICU patients, who may be more difficult to manage and who have anticoagulant protein deficiencies, including antithrombin, a necessary component of anti-xa heparin assays. 5 Acquired antithrombin deficiency may result in an underestimate of the 416 Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7

2 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE heparin concentration present in a patient 6 ; however, supplementation of antithrombin in an anti-xa assay may obviate this potential interference. We compared the performance of 3 anti-xa heparin assays, 1 with and 2 without antithrombin supplementation, on heparin recovery in spiked, antithrombin-deficient plasma at heparin concentrations that span the therapeutic range and in specimens from MICU patients receiving intravenous heparin therapy. Materials and Methods Participants and Specimen Collection Consecutive patients older than 18 years who were admitted to the MICU from June 2003 through November 2003, and who received intravenous, unfractionated heparin were eligible for the study. Twenty-five eligible patients were entered into the study according to an institutional review board approved protocol. All specimens (n = 149) were obtained by peripheral phlebotomy or line-draw into evacuated tubes containing 3.2% sodium citrate. Specimens submitted for aptt analysis were processed routinely into plateletpoor plasma and analyzed by the aptt and (Diagnostica Stago, Parsippany, NJ) assays within 2 hours of collection. The remainder of each specimen was frozen at 70 C and analyzed by the (Diagnostica Stago) and ACA (Dade Behring, Deerfield, IL) assays in batches. Specimens were included in the study only if they had been analyzed by all 3 heparin assays. Assays and Heparin-Spiked, Antithrombin-Deficient Plasma Preparation The ACA antithrombin and unfractionated heparin assays were run on the ACAstar analyzer (Dade Behring) according to the manufacturer s recommendations. All antithrombin concentrations are expressed as the percentage of normal human plasma. The limit of detection (LOD) of the ACA antithrombin assay is 5.8%. The and unfractionated heparin assays were run on the Sta-Compact analyzer (Diagnostica Stago) according to the manufacturer s recommendations. The stated LODs of the and ACA assays are 0.1 and 0.02 U/mL, respectively. The assay was performed with the 30-second incubation with factor Xa that results in a stated LOD of 0.2 U/mL. The aptt assay was performed on the Sta-Compact analyzer with STA- PTT A reagent (Diagnostica Stago). Antithrombin-deficient plasma samples of varying concentrations were prepared by diluting pooled normal plasma (PNP) (PrecisionBioLogic, Dartmouth, Canada) with antithrombin-depleted plasma (American Diagnostica, Stamford, CT) with a claimed antithrombin antigenic content of less than 1% antithrombin. Analysis of the antithrombin-depleted plasma with the ACA antithrombin assay returned a concentration of 6.3%. Therefore, antithrombin concentrations of mixed PNP and antithrombin-depleted plasma samples were calculated assuming an antithrombin concentration of 6.3% in the antithrombindepleted plasma. Unfractionated heparin at a concentration of 1,000 U/mL (American Pharmaceutical Partners, Schaumburg, IL) was spiked into PNP or antithrombin-deficient plasma to achieve the desired heparin concentrations. Statistics Deming linear regression was performed using Cbstat 5.0 (AACC, Washington, DC), assuming equal variances for the 3 heparin assays being compared. All other analyses were performed using SPSS 14.0 (SPSS, Chicago, IL). Two-sided Pearson correlations were calculated for antithrombin and heparin levels. Mean heparin recoveries were compared by using a 2-sided Student t test. Stepwise (forward) multiple linear regression analysis of the relationships among the aptt, heparin concentration, and antithrombin used a probability of F of.05 for entry into the equation and a probability of.10 to remove a variable. Results Heparin recovery was dependent on antithrombin and heparin concentrations for all 3 assays Figure 1. For the and ACA assays (without antithrombin supplementation), spiked heparin at concentrations of 0.2, 0.4, and 0.6 U/mL became undetectable at antithrombin concentrations of approximately 30%, 20%, and 6%, respectively. The effect of antithrombin supplementation to maintain heparin recovery in the face of decreasing plasma antithrombin concentrations also was heparin concentration dependent. At a spiked heparin concentration of 0.2 U/mL, the approximate LOD for the antithrombin-supplemented assay, heparin became undetectable at an antithrombin concentration of about 60%. At a spiked heparin concentration of 0.4 U/mL, heparin recovery progressively declined with decreasing antithrombin concentrations, reaching a nadir of 30% of the expected concentration at an antithrombin concentration of 6%. At a spiked heparin concentration of 0.6 U/mL, a recovery of 80% of the expected heparin concentration was obtained at the 6% antithrombin concentration (Figure 1). Because the decrease in heparin recovery seemed to be a continuous function across antithrombin concentrations, we attempted to determine the antithrombin concentration at which the recovered heparin concentration differed significantly from the measured heparin concentration at 100% antithrombin. For this experiment, we chose a concentration Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7 417

3 Lehman et al / ANTI-XA HEPARIN ASSAYS IN MICU PATIENTS A C ACA ACA AT Concentration (%) AT Concentration (%) in the middle of the therapeutic range, 0.5 U/mL, and compared the assay (without added antithrombin) with the assay (with added antithrombin). Statistically significant decreases in recovery (P <.01) occurred at an antithrombin concentration of 90% for both assays, and recovery continued to decrease with decreasing antithrombin concentrations. However, the rate of decline of recovery was slower for the assay, with significantly increased recoveries (P <.01) by the assay occurring at antithrombin levels of 70% or less Figure 2. To compare the performance of the 3 heparin assays in patients admitted to an MICU who were receiving intravenous heparin therapy, we prospectively evaluated 149 specimens from 25 patients for heparin levels by each of the 3 assays. Nineteen men and 6 women, ranging in age from 27 to 87 years, were entered into the study. The indications for heparin therapy included thrombosis (n = 16), acute coronary B ACA AT Concentration (%) Figure 1 Fraction of spiked heparin recovered by the 3 anti-xa heparin assays vs antithrombin (AT) concentration (percentage of normal human plasma). Varying concentrations of AT in plasma were prepared by diluting pooled normal plasma (PNP) with AT-depleted plasma. PNP was defined as containing 100% AT. Calculations of the AT concentration in diluted plasma assumed an AT concentration of 6.3% in the ATdepleted plasma. Each point represents the mean recovery from 2 separate experiments. A, Plasma spiked with 0.2 U/mL of unfractionated heparin. B, Plasma spiked with 0.4 U/mL of unfractionated heparin. C, Plasma spiked with 0.6 U/mL of unfractionated heparin. syndrome (n = 6), prosthetic cardiac valve (n = 1), continuous dialysis (n = 1), and vascular injury (n = 1). Deming regression of the results on the ACA and results demonstrated highly significant linear relationships. The and assays showed no proportional bias (slope ± SE = 0.95 ± 0.07) and a statistically significant y-intercept of 0.07 U/mL with R 2 = 0.87 (P <.001) Figure 3A. The assay demonstrated a proportional bias against the ACA assay (slope ± SE = 1.23 ± 0.08), also with a statistically significant y-intercept of 0.05 U/mL and R 2 = 0.83 (P <.001) Figure 3B. The median heparin concentrations measured by the 2 non antithrombinsupplemented assays were 0.27 U/mL for the ACA assay and 0.37 U/mL for the assay. For the antithrombinsupplemented, assay, the median concentration was 0.28 U/mL. Approximately 15% of the specimens had heparin levels below the manufacturer s stated LOD of the 418 Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7

4 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE AT Level (%) Figure 2 Fraction of 0.5 U/mL spiked, unfractionated heparin recovered by the and assays vs antithrombin (AT) concentration (percentage of normal human plasma). Varying concentrations of AT in plasma were prepared by diluting pooled normal plasma (PNP) with ATdepleted plasma. PNP was defined as containing 100% AT. Calculations of the AT concentration in diluted plasma assumed a concentration of 6.3% in the AT-depleted plasma. Each point represents the mean recovery (and 95% confidence interval) of 10 replicate analyses of each plasma specimen. For each assay, a drop below 100% AT resulted in a significant (P <.01) decrease in recovery. The assay showed significantly greater recovery (P <.01) at AT concentrations of 70% or less (* indicates a significant difference between the and means at an α level of.01). ACA assay (0.02 U/mL), whereas 20% had heparin levels less than the LOD (0.1 U/mL). For the assay, 32% of the specimens were below the manufacturer s LOD (0.2 U/mL). Based on a therapeutic range of 0.30 to 0.70 U/mL for unfractionated heparin, 1 classification of patient levels as subtherapeutic, therapeutic, and supratherapeutic by the and ACA assays agreed 84% of the time. The and demonstrated 80% agreement. Antithrombin levels were determined on 141 of the patient specimens. The levels were distributed normally with a mean concentration of 75% and an SD of 23%. The lowest concentration measured was 14%, and the maximum was 121% (antithrombin reference interval, 80%-120%). Approximately 4% of the specimens tested had antithrombin concentrations less than 30%, approximately 5% had concentrations less than 40%, approximately 20% had concentrations less than 60%, and approximately 60% had concentrations less than 80%. There were weak but statistically significant correlations between heparin level and antithrombin level for all 3 assays (, r = 0.39; P <.001; ACA, r = 0.48; P <.001;, r = 0.31; P <.001). Specimens with antithrombin levels less than 30% generally were below the limit of heparin detection for all 3 assays. Antithrombin levels alone did not correlate (r = 0.02; P =.87) with the aptt; however, multiple regression of antithrombin level and anti-xa level on the aptt resulted in a small but significant contribution of antithrombin level (P <.001) to the regression equations of each of the anti-xa assays. Therefore, the antithrombin level was predictive of the aptt only in the context of an associated heparin level, presumably owing to the relative dependence of the anti-xa activity of each heparin assay on the antithrombin level in an individual specimen. Regression of the heparin level and antithrombin level on the aptt produced the highest multiple correlation coefficient (R 2 = 0.45; P <.001), with the least dependence on antithrombin level (increase in R 2 of 0.06 units due to inclusion of the antithrombin level in the model). heparin levels had the next highest multiple correlation coefficient (R 2 = 0.37; P <.001) and an increase in the R 2 of 0.09 units due to inclusion of the antithrombin level in the model. The ACA assay had the lowest multiple correlation coefficient (R 2 = 0.32; P <.001) and the greatest dependence on the antithrombin level (R 2 increase of 0.13 units due to inclusion of the antithrombin level in the model). Discussion The current clinical standard for monitoring intravenous heparin therapy is the aptt. The recommended method for validating the aptt therapeutic range relies on simultaneous anti-xa heparin and aptt measurements to identify the aptt values that correspond to heparin concentrations of 0.3 and 0.7 U/mL. 1 The routine availability of anti-xa heparin assays on automated coagulation analyzers provides the opportunity to manage patient heparin levels directly, rather than through the aptt. It has been suggested that this approach may be more efficient and, therefore, cost-effective. However, the study that supports this approach enrolled patients with antithrombin concentrations in the normal range. 4 Many critically ill patients have decreased antithrombin levels secondary to their illnesses, 5 and anti-xa heparin assays require heparin to be bound to antithrombin to measure heparin activity. Of our patients, 60% had antithrombin levels less than 80%, the lower limit of the reference interval; 20% of patients had antithrombin levels less than 60%, and 5% had antithrombin Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7 419

5 Lehman et al / ANTI-XA HEPARIN ASSAYS IN MICU PATIENTS A B (U/mL) (U/mL) levels less than 40%. Our in vitro studies suggest that the steepest decrease in recovery of heparin levels for all of the assays begins when the antithrombin level drops below 60% to 70% activity, and the greatest benefit of antithrombin supplementation to the assay occurs when the antithrombin level drops below 40% (Figure 1). Therefore, in our population, only about 5% of the patient population would be expected to demonstrate a significant antithrombin supplementation effect on their heparin levels, and only at the mid to upper end of the therapeutic range. Consequently, it is not surprising that the contribution of the antithrombin level to the value of the heparin level for predicting the aptt was statistically significant, but small. In addition, the high degree of correlation among assays and the approximate 80% agreement in patient classification between the nonsupplemented assays and the supplemented assay confirms the minor role of antithrombin supplementation in heparin measurement in our overall patient population. It is not clear why the antithrombin supplementation to the assay did not have a greater effect to maintain heparin recovery in antithrombin-deficient plasma. It may be that the source and final concentration of the added antithrombin and/or the incubation conditions are not optimal. The manufacturer recommends increasing the incubation time with factor Xa from 30 to 240 seconds to increase the analytic sensitivity of the assay from 0.2 to 0.1 U/mL. Our study used the shorter incubation time. It may be that the increased incubation time also would increase the effect of the supplemented antithrombin. Further study will be required to determine if that is the case ACA (U/mL) Figure 3 Plot of vs (A) and ACA (B) heparin concentrations from analysis of 149 specimens obtained from 25 medical intensive care unit patients receiving intravenous, unfractionated heparin therapy. A line of identity is plotted on each graph. (U/mL) It has been suggested that antithrombin supplementation to a heparin assay actually may be detrimental, because it could overestimate the biologic effect of unfractionated heparin in an antithrombin-deficient patient. 6 This is not necessarily the case, however, because patients with heterozygous, congenital antithrombin deficiency who are receiving heparin therapy and who routinely have antithrombin levels of 40% are not resistant to standard-dose heparin therapy, as measured by the aptt response. 7 These patients would be expected to have decreased recovery of heparin levels, as measured by a nonsupplemented, anti-xa assay, owing to their antithrombin deficiency, although they achieve an appropriate biologic response (aptt) from their actual heparin level. In these patients, antithrombin supplementation might provide heparin levels more consistent with the clinical response. Furthermore, in vitro experiments with antithrombin-deficient plasma suggest that antithrombin deficiency must drop below approximately 25% before the aptt becomes resistant to heparin-induced prolongation. 8,9 Heparin concentration recovery, in vitro, was dependent on plasma antithrombin concentrations for all 3 assays. The antithrombin-supplemented assay demonstrated improved heparin recovery in direct correlation to the heparin concentration in the plasma. The greatest effect of the antithrombin supplementation occurred when the antithrombin level dropped below 40%, a level present in only 5% of the patient specimens. Analysis of patient specimens demonstrated a high degree of correlation among the 3 assays. Classification of the clinical adequacy of patient heparin levels showed agreement of 80% or more between the antithrombin-supplemented and 420 Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7

6 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE nonsupplemented assays. The antithrombin-supplemented assay studied did not significantly improve clinical usefulness in the MICU patient population. From the Departments of 1 Pathology and 2 Internal Medicine, Division of Respiratory, Critical Care, and Occupational Pulmonary Medicine, University of Utah Health Sciences Center; and 3 ARUP Institute for Clinical and Experimental Pathology, Salt Lake City. Address reprint requests to Dr Lehman: Dept of Pathology, UUHSC, 5C124 SOM, 30 N 1900 E, Salt Lake City, UT References 1. Büller HR, Agnelli G, Hull RD, et al. Antithrombotic therapy for venous thromboembolic disease: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest. 2004;126(3 suppl):401s-428s. 2. Rosborough TK. Comparison of anti factor Xa heparin activity and activated partial thromboplastin time in 2,773 plasma samples from unfractionated heparin-treated patients. Am J Clin Pathol. 1997;108: Baker BA, Adelman MD, Smith PA, et al. Inability of the activated partial thromboplastin time to predict heparin levels: time to reassess guidelines for heparin assays. Arch Intern Med. 1997;157: Rosborough TK. Monitoring unfractionated heparin therapy with antifactor Xa activity results in fewer monitoring tests and dosage changes than monitoring with the activated partial thromboplastin time. Pharmacotherapy. 1999;19: Messori A, Vacca F, Vaiani M, et al. Antithrombin III in patients admitted to intensive care units: a multicenter observational study. Crit Care. 2002;6: Holm HA, Abildgaard U, Larsen ML, et al. Monitoring of heparin therapy: should heparin assays also reflect the patient s antithrombin concentration? Thromb Res. 1987;46: Schulman S, Tengborn L. Treatment of venous thromboembolism in patients with congenital deficiency of antithrombin III. Thromb Haemost. 1992;68: Krulder JW, Strebus AF, Meinders AE, et al. Anticoagulant effect of unfractionated heparin in antithrombin-depleted plasma in vitro. Haemostasis. 1996;26: Ofosu FA, Blajchman MA, Modi G, et al. Activation of factor X and prothrombin in antithrombin-iii depleted plasma: the effects of heparin. Thromb Res. 1981;23: Am J Clin Pathol 2006;126: DOI: /8E3U7RXEPXNP27R7 421