We examined the concordance of heparin levels measured by a chromogenic anti-Xa assay and the activated partial thromboplastin time (APTT) during unfractionated heparin therapy (UFH) and the biochemical basis for differences between these measures. We also investigated the endogenous thrombin potential (ETP) as a possible measure of anticoagulation. Paired measures of anti-Xa and APTT were performed on 569 samples from 149 patients on UFH. The anti-Xa values and the APTT were concordant in only 54% of measurements. One hundred twelve samples from 59 patients on UFH were assayed for APTT, anti-Xa, factor II, factor VIII, and ETP. Supratherapeutic APTT values but therapeutic anti-Xa results had decreased factor II activity. Subtherapeutic APTT but therapeutic anti-Xa values had high factor VIII activity. ETP correlated with anticoagulation status and UFH dose. In conclusion, factor II and VIII activity contributes to discordance between APTT and anti-Xa results. ETP measurements may provide an additional assessment of anticoagulation status.
Unlike international normalized ratio testing, there is no standardization of reagents used for activated partial thromboplastin time (APTT) testing despite its use in monitoring unfractionated heparin therapy (UFH) since the early 1970s.1 Not only does variation in reagents and coagulometers affect APTT results, but also lot-to-lot variation of a given manufacturer’s reagents produces significantly different results due to differences in reagent sensitivity to UFH.2 For these reasons, the College of American Pathologists (CAP) requires a validation of therapeutic ranges with reagent lot changes and/or instrument changes.3 Anti-Xa monitoring has been suggested as an alternative to APTT because the anti-Xa assay is based on enzymatic inhibition, which can be accurately measured spectrophotometrically using well-defined chemical reagents that are not biologically derived.4–6 In addition, use of the anti-Xa assay to monitor UFH does not require reestablishment of the therapeutic range with each new lot of reagent as is necessary for APTT because the recommended anti-Xa therapeutic range of 0.3 to 0.7 IU/mL does not change.3 A stable therapeutic range is appealing because it eliminates confusion among clinicians as to the correct therapeutic ranges for UFH therapy and avoids the need to update electronic order sets and to retrain members of the health care team.
Despite these advantages, the anti-Xa assay also has disadvantages when used for UFH monitoring. Anti-Xa assays do not reflect all of the anticoagulant properties of UFH.5 Although thrombin and factor Xa are the most sensitive to UFH-antithrombin–mediated inhibition, the UFH-antithrombin complex also inhibits several other serine proteases.7 Thus, the anti-Xa assay only reflects part of UFH’s anticoagulant activity. In addition, anti-Xa assay results do not reflect changes in the activity level of other coagulation factors such as prothrombin, factor IX, factor VIII, and antithrombin, which could influence an individual patient’s overall anticoagulant response to UFH.8–10 Although APTT may serve as a better surrogate marker of the global level of anticoagulation, it can be affected by acquired inhibitors such as lupus anticoagulants. Thus, both APTT and the anti-Xa assay have limitations when used for monitoring UFH therapy. During early stages of a proposed transition to anti-Xa–based UFH monitoring, we noted significant discrepancies between APTT and anti-Xa results performed on blood samples drawn from patients who were receiving UFH therapy. The hypothesis that we tested was that factor VIII (FVIII) and factor II (FII) activity changes were probable causes of discordance of APTT and anti-Xa assay results. We also evaluated the utility of endogenous thrombin potential (ETP) as a measure of anticoagulant status during UFH therapy and in cases of discordance.11,12 On the basis of these studies, we suggest that it may be useful to measure APTT and the anti-Xa assay in patients with coagulopathy on UFH, with ETP providing an additional assessment of coagulation status that deserves further study.
Materials and Methods
For 1 week, we compared the results of anti-Xa UFH levels with APTT, performed on the same blood sample for all inpatients receiving UFH. Providers ordered UFH using electronic order sets based on a weight-based nomogram. These patients were adults admitted to the medical and surgical services of Johns Hopkins Hospital for suspected acute thromboembolic disorders, including stroke, myocardial infarction, pulmonary embolism, and deep venous thrombosis. The blood samples were collected in the first 72 hours of the initiation of heparin, and no patients had yet started warfarin therapy.
Blood samples were collected into 3.2% citrated glass Vacutainers from Becton Dickinson (Franklin Lakes, NJ). All APTT measurements and anti-Xa levels were measured within 45 minutes of collection after double centrifugation of the sample to ensure adequate removal of platelets. After aliquoting into 200-μL samples, the remaining plasma was frozen at −70°C until further investigational testing, as approved by our institutional research board.
We used a single lot number of Actin FSL to measure APTT (Siemens, Marburg, Germany) using a BCS-XP coagulometer (Siemens). When the APTT exceeds 199 seconds, the results are reported as greater than 200 seconds. To measure UFH, we used a chromogenic anti-Xa assay, the Berichrom Heparin Assay (Siemens). This assay ensures replacement of antithrombin levels and other clotting factors so that the assay is specific for the inhibitory effect of UFH on factor Xa.4
We used CAP guidelines to estimate the therapeutic range using freshly collected plasma from 100 adult patients who had normal prothrombin times (PTs) and baseline APTT measurements.3 The APTT that corresponded to 0.3 to 0.7 IU of UFH per milliliter by regression analysis was 49 to 82 seconds. We measured thrombin generation using the ETP assay (Siemens) with the BCS-XP coagulometer as described by the manufacturer. The activation of coagulation was started by incubation of plasma with phospholipids and recombinant human tissue factor (Innovin; Siemens). The final concentration of tissue factor was approximately 300 pmol/L. Thrombin generation was measured with a chromogenic substrate and corrected for the activity of α2-macroglobulin–bound thrombin. The results of this assay are expressed as ETP (area under the curve).
FII activity measurements were conducted using Innovin recombinant human thromboplastin (Siemens) with the BCS-XP coagulometer as described by the manufacturer. FVIII activity was measured using a 1-stage assay with Actin FSL on the BCS-XP coagulometer as described by the manufacturer. Because patients were receiving UFH, heparinase treatment of the plasma was done prior to performing the FVIII activity according to package insert instructions of the heparinase reagent (Siemens). The study protocol prevented our collecting blood beyond what was needed for monitoring UFH, and we were restricted to using the plasma that was left over for any additional tests. To test whether reduced FII activity leads to increased APTT values compared with the anti-Xa assay results, we first tested in vitro prepared samples with different FII levels. To prepare the test plasmas, we mixed normal pooled plasma with FIIdeficient plasma to get different final FII activities (ie, 25%, 50%, 75%, and 100%). UFH was then added to each sample to a final concentration of 0.2 to 0.8 IU/mL, followed by measurement of APTT. We used the same experimental approach to test the impact of different FVIII activity levels on APTT results by serially diluting normal plasma with FVIII-deficient plasma.
After obtaining institutional review board approval, we correlated the results of the factor activity assays and ETP with the APTT, PT, and anti-Xa levels. Differences between measurements of factor activities and ETP were determined with the Student t test. A P value less than .05 was considered significant. Correlation between ETP, APTT, and anti-Xa was measured by the Pearson correlation coefficient (r).
During the study period, 569 paired blood samples were drawn from 149 patients. In Figure 1, we show the distribution of anti-Xa UFH levels vs APTT. The mean APTT was 69.7 seconds (median, 57.7 seconds; range, 29.7–203.7 seconds). The mean anti-Xa level was 0.37 IU/mL (median, 0.3 IU/mL; range, 0–1.3 IU/mL). For any given anti-Xa value, a wide range of APTT values was measured. For example, when the anti-Xa level was 0.5 IU/mL, the corresponding APTT values ranged from 30 to 200 seconds. The correlation between APTT and anti-Xa levels is relatively low (r = 0.61; 95% confidence interval, 56–66 from 27–200 seconds). To reduce the potential bias of including multiple specimens from patients with discordance, we averaged values of multiple specimens and used single values for each individual patient. When single values of anti-Xa and APTT from individual patients were analyzed, there was still low correlation (r = 0.42 for single values vs r = 0.61 for multiple values). The fact that there was not a significant bias from multiple specimens may reflect that multiple specimens were included from individuals with discordance and concordance.
In Table 1, we show the concordance of test results according to the validated therapeutic range as previously described. There was poor concordance between the measures when the anti-Xa concentrations were within the target range. There was equally poor concordance between the values when considering APTT as the standard.
To test whether reduced FII activity leads to increased APTT values compared with the anti-Xa assay results, we first tested in vitro prepared samples with different FII levels. We used the same experimental approach to test the impact of different factor VIII activity levels on APTT results by serially diluting normal plasma with FVIII-deficient plasma. The results are shown in Figure 2. These experiments show that in the UFH therapeutic range of 0.3 to 0.7 IU/mL, decreasing the level of FII and FVIII substantially increases the APTT result for a given concentration of UFH. For instance, APTT is 55 seconds in the presence of 0.4 IU/mL UFH when measured on a sample with 100% FII activity, but APTT is as high as 100 seconds when the FII activity is 50%.
We then determined the effect of FII and FVIII activity on the frequency of concordance between the APTT and anti-Xa levels during UFH therapy. These results are shown in Figure 3. We had 205 samples from 109 patients on which to measure the FII and FVIII activity. In samples with a discordantly high APTT for anti-Xa measurement, there was a statistically significant decrease in FII activity compared with samples with concordant APTT and anti-Xa levels. The percentage of samples with an FII activity of 50% or less was notably increased in this group of samples with discordantly high APTT for anti-Xa activity levels. Conversely, it is notable that 89% of the plasma samples (24 of 27) with discordantly low APTT values (ie, subtherapeutic APTT with therapeutic/high anti-Xa or therapeutic APTT with supratherapeutic anti-Xa) had an elevated FVIII activity (≥150%). Only 4% (1 of 27) of these samples had an FII activity of 50% or less. These data demonstrate that low FII activity is associated with discordantly high APTT values relative to anti-Xa levels, whereas high FVIII activity is associated with discordantly low APTT values relative to anti-Xa levels. We did not investigate other clotting factors, and these too may have contributed to discrepancies.
Given the discrepancies between APTT and anti-Xa levels, it was unclear whether these patients were under- or over-anticoagulated. Therefore, we investigated whether ETP might provide additional information about the degree of thrombin inhibition and measured ETP in 124 specimens from 57 patients on UFH. To address the potential bias of including multiple samples from patients with discordance, we averaged values of multiple specimens and reported single values from each individual patient. As shown in Figure 4A and Figure 4B, we found that ETP for patients receiving UFH varied inversely with both APTT and anti-Xa level, and there was relatively low correlation between these measures (r = −0.44 for APTT and r = −0.63 for anti-Xa). We also examined ∆ETP (patient’s baseline ETP minus ETP after heparin) and found a moderate correlation between the suppression of ETP with UFH, with UFH dose as measured by the anti-Xa assay (r = 0.69) Figure 4C. However, this correlation between UFH dose and ETP was relatively high in the subset of patients with normal FII and FVIII activity (r = 0.83) Figure 4D.
As previously shown, the APTT and anti-Xa UFH levels are discordant in approximately 50% of patient samples. In these cases, it is unclear whether thrombin generation is significantly decreased. To address this question, we examined ETP in both concordant and discordant groups as a measure of bleeding or thrombotic risk. In patients with subtherapeutic anticoagulation as determined by concordantly low anti-Xa and APTT values, ETP was significantly higher (P < .0001) than ETP measured in patients with therapeutic anticoagulation Figure 5A. Conversely, when the anticoagulation was supratherapeutic, as determined by concordantly elevated anti-Xa UFH levels and APTT values, ETP was significantly lower than ETP measured in patients with therapeutic anticoagulation (P < .0023). Thus, ETP measurements correlated with the degree of anticoagulation in samples with concordant APTT and anti-Xa measurements. Next, we examined ETP in patients with discordant anti-Xa levels and APTT values. As shown in Figure 5B, for patients with a therapeutic anti-Xa and subtherapeutic APTT, ETP was significantly increased (P = .0009) compared with those in the therapeutic APTT range. Similarly, as shown in Figure 5C, ETP was also increased in patients with therapeutic APTT values and subtherapeutic anti-Xa levels (P = .046). These data suggest that when either APTT or anti-Xa was low, anticoagulation was subtherapeutic. Conversely, when APTT was high and anti-Xa therapeutic, ETP was depressed compared with ETP in patients with therapeutic anticoagulation, suggesting increased anticoagulation; however, the difference in ETP was not statistically significant (P = .26).
In a sample of 149 consecutive patients on UFH, we noted that discordance is frequent between the APTT and anti-Xa assays and that alterations in FII and FVIII activity are associated with this discordance. We show both in vitro and in vivo that low FII activity results in a discordantly high APTT for a given anti-Xa activity level; conversely, discordantly low APTT values were noted for a given anti-Xa UFH activity level in the presence of elevated FVIII activity. These findings may have clinical relevance as reduced FII activity may be seen in liver dysfunction, vitamin K deficiency, and disseminated intravascular coagulation, whereas FVIII is an acute-phase reactant, and elevations are seen commonly in inflammatory states.13 These findings provide a biochemical basis for discordant APTT and anti-Xa measurements that are encountered frequently in hospitalized patients treated with UFH and underscore the limitations of currently available assays for UFH monitoring. A limitation to our study is that we did not investigate other coagulation factors.
Although APTT is considered a global assessment of coagulation status, its use to monitor UFH therapy with concomitant conditions that prolong APTT may be problematic.5,9 The anti-Xa assay has been proposed by some as a better assay for heparin monitoring.8,14 Consequently, a number of medical centers have switched to the anti-Xa assay for UFH monitoring. Although anti-Xa testing offers the advantages of not being affected by coagulation factor deficiencies other than antithrombin, being less affected by preanalytical factors such as underfilled tubes, not being affected by high FVIII activity, and circumventing the need to establish an APTT therapeutic range with each new lot of reagent, limited published studies document the effectiveness of anti-Xa assays for monitoring UFH. Moreover, there is some uncertainty about what constitutes the ideal approach to measuring anti-Xa levels.
Commonly available anti-Xa assays, including our own, use reagents that provide supplemental antithrombin (to prevent low antithrombin levels from influencing the UFH measurement) as well as dextran sulfate (to strip UFH from plasma proteins).4,5 Proponents of adding exogenous antithrombin suggest that low endogenous antithrombin levels could lead to underestimation of UFH concentrations and over-anticoagulation.9 A theoretical argument against reconstitution is that the addition of antithrombin might overestimate the true biologic effect of UFH present in vivo.9 Similarly, addition of dextran sulfate may lead to a relative overestimation of the in vivo activity of UFH. To date, there have not been any studies comparing clinical outcomes with the different anti-Xa assay methods.
Our results indicate that a global assay of coagulation, such as ETP, might be worthy of further investigation. Alternative global assays for thrombin generation, such as ETP, have been used to assess bleeding and thrombotic risk in a variety of diseases. ETP has been evaluated in healthy subjects receiving heparin, but its utility in monitoring anticoagulation status is not well studied in patients.14 Some conditions that result in APTT prolongations, such as liver failure, have been shown to be associated with increased ETP, which correlates with a prothrombotic state.15 On the other hand, other coagulopathic conditions with a prolonged APTT have been shown to have decreased ETP, which may be associated with an increased risk of bleeding.15–17 Thus, if the goal of anticoagulation monitoring is to have a measure of thrombin inhibition, ETP may be a more direct measure. We found a significant correlation between the intensity of anticoagulation and ETP in patients treated with UFH. Although there was poor correlation between ETP and either APTT or anti-Xa levels, there was higher correlation between UFH dose and the degree of ETP suppression with UFH treatment. We also found that in samples with either therapeutic anti-Xa or APTT measurements but discordantly low corresponding APTT or anti-Xa, ETP was high, suggesting under-anticoagulation. These interesting results warrant further investigation to determine the utility of ETP in monitoring UFH.
In conclusion, we found that both APTT and anti-Xa have limitations when used for UFH monitoring and may not accurately assess anticoagulant status in selected situations. Alterations in FII and FVIII activity may underlie discordance between APTT and anti-Xa activity during UFH therapy. It remains unclear whether APTT, anti-Xa, or ETP provides the most accurate measurement of the global coagulation status of patients receiving UFH. Further studies are warranted to assess the utility of these different assays in UFH monitoring and to guide safe and effective UFH therapy.