prevent its recurrence. The FDA is reviewing possible deficiencies in the manufacturer s reagent package labeling.

Transcription

1 science [coagulation and hematology] Laboratory Variables That May Affect Test Results in Prothrombin Times (PT)/International Normalized Ratios (INR) David L. McGlasson, MS, CLS/NCA, H, ASCP 59th Clinical Research Squadron/MSRL, Lackland AFB, TX DOI: /RT9D02JXCHY8V9HM 124 Monitoring patients on oral anticoagulant therapy (OAT) requires constant monitoring by physicians using highly sophisticated coagulation testing. The World Health Organization (WHO) put forth the idea for prothrombin time standardization based on a mathematical formula known as the International Normalized Ratios (INR) Because manufacturers provide limited guidelines for instruments, the laboratory should validate its values by performing local on-site International Sensitivity Index (ISI) calibration. In a MMWR report dated August 24, 2001 the following circumstances on Adverse events and deaths associated with laboratory errors at a hospital-pennsylvania, 2001 were reported. On August 3, 2001, the Center for Disease Control (CDC) was contacted by the Pennsylvania Department of Health (PADOH) to assist with an investigation of laboratory errors that may have contributed to the deaths of at least 2 persons taking the anticoagulant drug warfarin (Coumadin, DuPont Pharmaceuticals M, Wilmington DE). The Food and Drug Administration (FDA) also was investigating this incident. The monitoring of warfarin s anticoagulant effect is accomplished by performing 2 laboratory assays: the prothrombin time (PT) and the INR. The INR is a numeric value calculated from the PT, a clotting assay result, and the ISI, which indicated the sensitivity of the reagent used to perform the assay. The WHO recommends that INR values be used by physicians to compare laboratory results between reagent/instrument combinations at different institutions. During June 4 through July 25, 2001 the hospital laboratory in Pennsylvania reported 2,146 test with correct PT results but with incorrectly calculated INR results. The mathematical formula required for calculating the INR uses a reagent-specific number, the ISI, which is usually furnished by the manufacturer of the reagent. In June, the hospital laboratory did not verify the new reagent used in the coagulation analyzer for performing the PT assay. As a result, the ISI used to calculate the INR was incorrect for the reagent used. For approximately 7 weeks, the reported INRs were falsely low. Several physicians who interpreted these results increased their patients doses of coumadin. In several cases, this led to the patient s OAT regimen being compromised. 1 One subject previously had coronary bypass surgery and was having the INR followed on an outpatient basis. He started having bleeding from the gums and was bruising easily. At the original hospital laboratory the INR result was 2.62, which is in the therapeutic range. He then went to another hospital to have the INR repeated. That result yielded an INR of 5.7, which was very high. His cardiologist directed the patient to discontinue the coumadin dosing for 4 days. 2 Investigations are also under way by PADOH, Centers for Medicare and Medicaid Services (CMS), and CDC to identify other patient morbidity and mortality associated with the error, its possible causes, and the steps needed to prevent its recurrence. The FDA is reviewing possible deficiencies in the manufacturer s reagent package labeling. Discussion Monitoring patients on OAT requires constant monitoring by physicians using highly sophisticated coagulation testing. The most frequently used assay is the PT. Standardization of the PT test results has proven to be difficult because of the abundance of manufacturer thromboplastins and instrumentation available throughout the world. Physicians monitor patients on OAT by frequently performing repeat measurement of the PT assay. By doing this they can adequately monitor the subjects warfarin intake. They are then able to maintain each subject in a designated therapeutic range. There are many variables in the performance of the PT that may affect the test result. These include the specimen collection, processing of the sample, instrumentation used to perform the assay, and the sensitivity of the thromboplastin used to perform the PT assay. Each thromboplastin s sensitivity is determined by the individual ISI. Manufacturers individual thromboplastin reagent sensitivity to the Vitamin K dependent coagulation factors affected by warfarin have a wide range of variability. This has created problems for physicians when trying to monitor patients in different institutions with different reagent/instrument combinations. A subject monitored with a low sensitivity reagent (high ISI) may give a low PT time of 14 seconds. While the same patient when using a high sensitivity reagent (low ISI) may give a time of 18

2 seconds. These results may drastically affect the interpretation of the PT assay and the warfarin dosage in the individual patients. 3 The WHO in 1977 introduced a standardized thromboplastin that was touted as an international reference preparation (IRP). The hope was that each laboratory could develop a reference range for PT testing that was to be compared to the IRP range. Taking this initial step further the WHO in 1983 then put forth the idea for PT standardization based on a mathematical formula (INR) that uses the elements of the PT assay. The INR then would be the PT result that a laboratory would obtain if the test were performed using the standardized WHO reference thromboplastin reagent with an assigned ISI value of 1.0. Each assigned thromboplastin ISI could then compare the individual reagent sensitivity to an IRP that has been calibrated with a WHO reference plasma. The INR is calculated using the following formula 3 : INR = (patient PT) ISI Mean normal PT Each individual laboratory would supply the mean normal PT. The INR method would then allow the physician to compare the PT results with different laboratories reagent/instrument combinations. Manufacturers now produce a wide variety of thromboplastin with ISI values ranging from 0.9 to 3.0. This wide range of ISI will produce a huge difference of PT results. 3 The INR was designed to correct for this difference. Therefore, if a patient were traveling from one city to another the physician would have the advantage of expecting a standardized INR value to maintain a patient s OAT range. As observed in the case report improper assignment of ISI values with different reagent/instrument combinations can result in clinically significant inaccurate values. This difference in ISI values resulted in improper management of a large group of patients on OAT with catastrophic consequences. Huge differences in INR values have been reported when there are significant differences between low versus high ISI sensitivities in thromboplastins. 4,5 Preanalytic Variables of INR Testing Citrate Concentration Prothrombin time assays are performed on citrated platelet-poor plasma. The concentration of the sodium citrate anticoagulant can affect the PT time thus affecting the INR value. 6-8 In the United States, until recently, approximately 80% of the PT assays were performed in plasma that was collected using vacutainer tubes that contained 3.8% (129 mol/l) sodium citrate. In 1998, NCCLS Subcomittee on Coagulation started to recommend that a 3.2% (109 mol/l) sodium citrate concentration be used for most coagulation studies. This concentration was selected because the WHO was calibrating thromboplastins using the 3.2% concentration. 9 Other international standardization organizations also recommend using the 3.2% citrated anticoagulant. The osmolality of the 3.2% concentration is apparently closer to plasma. Higher INR values have been found in some thromboplastin reagents in plasma collected with 3.8% citrated vacutainer tubes. In one direct comparison study between the 2 citrate concentrations using a low ISI thromboplastin, the INR difference in 18.0% of the specimens showed a INR difference of 0.7. In another study, the difference on all specimens gave an average difference of 19.0%. The lower the ISI value is the more difference the citrate concentration makes in the INR results. 9 In platelet studies and testing of blood bank products such as cryoprecipitate, the 3.8% citrate concentration is still recommended. Specimen Handling Improper specimen collection can also drastically affect the PT assay thus affecting the INR value. This area has become controversial due to a number of studies that discuss the length of time and how a specimen can be stored be performing a PT assay. One study showed that storage of specimens for up to 6 hours at room temperature did not affect the PT results. 10 In another study, it was found that specimens from patients who were on low dose warfarin could be shipped overnight at room temperature without significantly changing the INR values. 11 Another study looked at specimens stored as whole blood. It found that after 3 days stored at room temperature that there was no significant change in the INR. 12 NCCLS guidelines from 1998 stated that uncentrifuged specimens for the PT/INR test or centrifuged plasma left of top of the separated cells stored at various temperatures from 2 to 24 C should be run within 24 hours from the time the sample was obtained from the test subject. 13 The PT results may vary from the time of collection of the specimen and storage of the sample if the subject was receiving heparin and was on OAT. The laboratory must be sure that the thromboplastin has a heparin neutralizer to counteract the effects of heparin. Evidence has also been presented that if specimens are stored at approximately 4 C that the PT/INR results may be shortened due to cold activation of FVII. 9 Our recommendation is that all specimens should be treated as if they were the last specimens that could be obtained from the test subject without having to recollect the sample. If another assay is required, such as an APTT or factor assay, the prolonged storage times of >4 hours would possibly render the specimen compromised and required the patient to have a second venipuncture. This increases time and expense for all concerned. More recently it has been discovered that the coagulation tube has been found to be a clinically significant source of variability in coagulation testing. There has been controversy in using partial-draw tubes which required less vacuum to manufacture which resulted in 1.8 ml or 2.7 ml blood draws versus the conventional 4.5 ml draw with the blood to anticoagulant ratio of 9:1. Since the partial-draw tubes filled more slowly than the so-called full-draw tubes to platelets may have been activated due to lengthy exposure to shear forces arising from exposure to shear forces arising from drawing blood into the increased headspace. The activated platelets then may release platelet-factor 4 or phospholipids which could compromise coagulation testing

3 Instrumentation There are over 300 different methods of performing a PT assay in the United States alone. Because of this wide variety of testing it is almost impossible for manufacturers to provide accurate reagent/instrument specific ISI values for each reagent/instrument combination. The WHO protocol for assigning ISI values to thromboplastin reagents is determined by performing a manual tilt-tube method in quadruplicate. 15 Laboratories then have to depend on manufacturers to accurately assign the proper ISI of their reagents based on some adaptation of the WHO protocol. 16 Therefore, the instrumentation can vastly affect the PT test results. If a coagulation laboratory is using a reagent from one company and an instrument from another supplier they have to rely on the reagent companies generic assignment of mechanical or photo-optic ISI values. Laboratories may not be aware that different automated coagulation systems can directly affect the ISI and the INR result. There may be distinct differences between mechanical and photo-optical coagulation systems with different reagents. 16,17 In our institution, we performed a protocol with 7 different reagent/instrument combinations on the same specimens. 16 In many specimens, we saw clinically significant differences of INR values between the different reagent/ combinations. T1 displays some of these INR result differences. Observe that the specimen #1180 has an INR value of 3.95 with one reagent/instrument combination and 7.16 with another system. This is an 81.3% difference. The 3.95 value might be considered a high therapeutic range value that would probably require no intervention. The 7.16 INR might cause some institutions to consider this result a panic value. Many other variables can affect the coagulation instrument ISI sensitivity used in the PT analysis. These are temperatures of the system and reagents, accuracy of volume of the reagents being dispensed, and the citrate concentration of the specimen collected. 15,16 Looking for a simpler method to locally calibrate ISI values we performed a protocol at our institution that was sponsored by Dade Behring M (Deerfield, IL) using a calibration system supplied by Precision-Biologic M (Dartmouth, NS, Canada). Precision Biologic has proposed a simplified method of locally calibrating the reagent ISI using a 5-day ISI calibration protocol. The method used a set of 5 OAT frozen plasmas and 1 normal frozen plasma that are assayed against a standard (WHO IRP) thromboplastin. The frozen OAT calibrators and the normal plasma encompassed the 4 therapeutic categories used in OAT (<2.0; ; ; >4.5). Two recent papers discussed in detail how any laboratory can use the frozen calibrator plasma method to locally calibrate their reagent/instrument combinations and how they would affect subject and manufacturers suggestions for ISI values with many different thromboplastins and instruments. 15,17 Laboratories that run PT/INR assays can then set up there own ISI calibration of the locally used thromboplastin. This method may be the most practical for improving inter-laboratory INR agreement. See the results in T2 and T3 on differences in local and manufacturer calibrations of ISIs that were used in our study. 14 Heparin Effects on PT/INR Testing Unfractionated heparin contamination can adversely affect the results of a PT/INR assay. Many subjects starting on OAT are coming off heparin therapy, which could give physicians problems with falsely elevated INR values when trying to establish baseline INR ranges for patients. Each manufacturer s thromboplastin has an individual sensitivity to the effect of heparin. Some reagents have a heparin neutralizer present such as polybrene or protamine sulfate that can counteract the effects of heparin in the therapeutic range so the PT/INR results are not compromised. However, reagents without a heparin neutralizer present have been found to have PT results prolonged in some subjects by as long as 10 seconds. This could lead to Comparison of Selected INR Differences of Human Subject s Plasma T1 Human MLA 900C CI+ MLA 900C T-D MLA 900C T-R MLA 1600C T-D STA CI+ STA T-D STA T-R Sample Done on the STA and MLA 900 C with thromboplastins: Neoplastine CI+, Diagnostica Stago (CI+); Thromboplastin D, Pacific-Hemostasis (T-D); Recombiplastin, Hemolance (T-R) and the MLA 1600C with T-D only.

4 Summary of Mean INR Results on 25 Normal Patient Samples Before and After Local ISI Calibration T2 Instrument Reagent Mean INR Using Mean INR Using Difference in Vendor-Assigned ISI Locally Calibrated ISI INR Means (%) Diagnostica Stago/STA Neoplastine CI Thromborel S Thromboplastin C Innovin Sysmex CA 540 Neoplastine CI Thromborel S Thromboplastin C Innovin Dade-Behring BCS Neoplastine CI Thromborel S Thromboplastin C Innovin Summary of Mean INR Results on 95 Patient OAT Samples Before and After Local ISI Calibration T3 Instrument Reagent Mean INR Using Mean INR Using Difference in Results >10% Vendor-Assigned ISI Locally Calibrated ISI INR Means (%) Difference Diagnostica Stago/STA Neoplastine CI Thromborel S Thromboplastin C Innovin Sysmex CA 540 Neoplastine CI Thromborel S Thromboplastin C Innovin Dade-Behring BCS Neoplastine CI Thromborel S Thromboplastin C Innovin serious patient mismanagement during OAT. There is no current evidence that the heparin neutralizers may neutralize the presence of low molecular weight heparin. Each institution should check the effect of a therapeutic dose of unfractionated heparin against their reagent/instrument system. Possible Effect of a Lupus Anticoagulant (LA) on PT/INR The recommended therapeutic range of INR for OAT in patients with the presence of a LA is currently 2.5 to 3.5. This is still a controversial subject. It has been suggested that using the INR to monitor these patients may be inadequate due to interference by the presence of a LA on the clot-based PT assay. Some investigators have suggested using the prothrombin-proconvertin assay in lieu of the PT/INR for patients with this disorder since it doesn t appear to be affected by a LAs inhibitory actions. 3 Others have suggested measuring coagulation factors II and X by either chromogenic or 1-stage clotting assays based on at least 3 dilutions to lessen the LA inhibitory effect. 3 Other investigators found little effect of the presence of a LA on PT/INR results using different thromboplastins and instruments. One small exception was a subgroup of 6 patients in which a recombinant thromboplastin (Innovin, Dade-Behring) was used. 3 One recent study performed a protocol to see if subjects who have the presence of anticardiolipin antibodies without the presence of an LA to see if the INR assay was affected using local specific ISIs with 11 different thromboplastins. No effect on the INR results was observed. 18 In another study using 11 different LA positive, 11 negative LA subjects, and 7 different thromboplastins, the presence of an LA did not disturb the laboratory tests for monitoring subjects on OAT. 19 Each institution should check their own reagent/instrument combination with positive LA plasmas to determine if the LA interference is an issue in their laboratory. 127

5 128 Reporting of PT/INR results There continues to be no consensus on the method of reporting INR values for conditions other than OAT. Many laboratories report only the INR for all PT measurements since the INR is simply a mathematical conversion of the PT using the ISI. 9 Many laboratories have been reluctant to convert to low ISI thromboplastins because the prolonged PTs are thought to be confusing to physicians used to shorter PTs with high ISI (less sensitive) reagents. This may be true when a PT is ordered for situations other than monitoring OAT. 9 A situation such as a preoperative coagulation screen or liver diseases can confuse the issue of INR reporting. It has been suggested by some clinicians to use 2 systems for reporting PT results. One result for patients on OAT and another for subjects not on OAT has been instituted at some facilities. There is data that suggests that the INR value is appropriate for use in patients beginning anticoagulation as well as in subjects with other coagulation disorders (eg, liver impairment). The range of PT prolongation appears to be greater for patients on OAT than for patients with liver disease or disseminated intravascular coagulation, but the PT and INR has been shown to correlate in a linear fashion when reagents of varying ISI were compared. Some experts have proposed reporting the INR in lieu of the PT for all patients, but suggested the best way to accomplish this would be to have an international consensus conference. This has still not been accomplished as recently as the 2001 International Society of Thrombosis and Hemostasis Conference in Paris, France. In the subcommittee conference that dealt with this issue, there was still no consensus on how to use the INR for other than those subjects on OAT. Some notable clinicians and researchers have publicly stated that the INR is recommended as the most appropriate means for reporting PT results in patients who are stable, orally anticoagulated. They have suggested that the INR is not appropriate for use in other clinical situations (eg, liver disease, hereditary factor deficiencies, routine screening, preoperative testing, and detecting vitamin K deficiency). There is little or no information available on the use of the INR reporting of drug-induced coagulation defects in subjects on OAT. Does a single coagulation factor deficiency, such as a factor FVII deficiency, have a significant effect on the INR? 9 This could present problems in clinical settings where the INR only is reported. Fairweather and colleagues 9 presented a further argument against the use of reporting INRs only. The presence of a mildly prolonged PT may be the only clue to a clinically significant coagulopathy. However, the upper range of normal is obscured by the exponential nature of the INR calculation. The example they used discussed the upper limit of normal for 2 reagents, which correspond to a PT ratio of 1.2. This would correspond to an INR of 1.2 for a reagent with an ISI of 1.0 and an INR of 1.4 for a reagent with an ISI of 2.0. The problem would then be in interpreting the INR values in the range of 1.3 to 1.4. One local hospital in our area went to INR value reporting because the surgeons would misinterpret what they perceived to be mildly prolonged PTs. Consequently, they would order freshfrozen plasma for surgeries unnecessarily. Final Notes The accuracy and precision of the INR is dependent on the PT assay and the ISI of thromboplastin. Other researchers have noted that since the ISI is the exponent of the INR equation, the higher the ISI assigned to thromboplastins the greater the imprecision of the INR as a result of the mathematical outcome. 20 In support of this information we discovered, in a study, that when the INR approaches 3.0 or greater the discrepancy in the INR value was between different reagent/instrument systems. Therefore the large difference in assigned ISI values is probably the most important variable that influences the poor correlation seen in different laboratories reagent/instrument combinations. Thromboplastins with an ISI less than 1.2 produce a wider range of values in the PT and PT ratio. Consequently, the precision of the INR is similarly improved. Since the calculation of the INR requires using the ISI exponentially, the farther the ISI value is from 1.0, the less accurate the INR values will be. Manufacturer assigned ISIs, not specific to the local reagent/instrument set-up, may introduce even more inaccuracy. Instrumentation can also greatly influence the INR values. Because manufacturers provide limited guidelines for instrument assigned ISIs, the laboratory should validate its coumadin influenced INR values by performing local on-site ISI calibration. If the laboratory in Pennsylvania had a protocol for locally calibrating their ISI values, the disaster that occurred with the inaccurate ISI being used may have been averted. 1. Adverse events and deaths associated with laboratory errors at a hospital-pennsylvania, MMWR. 2001: McCullough, Marie. Lab error deaths may now total five. The Inquirer (Philadelphia, PA), August 3, McGlasson DL. A review of variables affecting PTs/INRs. Clin Lab Sci. 1999;12: Cunningham MT, Johnson GF, Pennell BJ, et al. The reliability of manufacturer-determined, instrumentspecific international sensitivity index values for calculating the international normalized ratio. AJCP. 1994;102: Ts ao C, Swedlund J, Neofotistos D. Implications of use of low international sensitivity index thromboplastins in prothrombin testing. Arch Pathol Lab Med. 1994;118: Duncan EM, Casey CR, Duncan BM, et al. Effect of concentration of trisodium citrate anticoagulant on calculation of the international normalized ratio and the international sensitivity index of thromboplastin. Thromb Haemost. 1994;72: Adcock JM, Kressen DC, Marlar RA. Effect of 3.2% vs. 3.8% sodium citrate on routine coagulation testing. Am J Clin Pathol. 1997;107: Dwyre AL, Giles AR, Key LA, et al. The effects of sodium citrate anticoagulant concentration, storage on or off cellular matrix, and specimen age on prothrombin time INR using a low ISI (1.06) rabbit brain thromboplastin. Thromb Haemost. 1995;73: Fairweather RB, Ansell J, Anton M, et al. College of American Pathologists Conference XXXI on laboratory monitoring of anticoagulant therapy: Laboratory monitoring of oral anticoagulant therapy. Arch Pathol Med. 1998;122: NCCLS H21-A3. Collection, transport, and processing of blood specimens for coagulation testing and general performance of coagulation assays; approved guideline-3rd ed. 1998;18: Van den Besselaar AMHP, Halem-Visser LP, Loeliger EA. The use of evacuated tubes for blood collection in oral anticoagulant control. Thromb Haemost. 1983;50: Raskob GE, Durica SS, Owen WL, et al. Monitoring low-dose warfarin therapy be a central laboratory and implications for clinical trials and patient care: the coumadin aspirin reinfarction (CARS) pilot study group. Am J Cardiol. 1996;78: Baglin T, Luddington R. Reliability of delayed INR determination: Implication for decentralized anticoagulant care with off-site blood sampling. Br J. Haematol. 1997;96: Foubister Vida. Quick on the draw-coagulation tube response. CAP Today. 2002;16:38-42.

6 15. McGlasson DL, Hickman JR, More LE, et al. Discrepancies in international normalized ratios (INR) in swine and humans. Clin Lab Sci. 1998;11: McGlasson DL. A Comparison of international normalized ratios after local calibration of thromboplastin international sensitivity indexes. Accepted for Publication in Clinical Laboratory Science. 17. Adcock DM, Johnston M. Evaluation of frozen plasma calibrants for enhanced standardization of the international (INR): A multi-center study. Thromb Hemost. 2002;87: Mant MJ, Stang L, Etches WS. Warfarin monitoring in patients with anticardiolipin antibodies, but without lupus anticoagulants. Thrombosis Research. 2000;99: Bijsterveld NR, Middeldorp S, Berends F, et al. Monitoring therapy with vitamin K antagonists in patients with lupus anticoagulant: effect on different tests for INR determination. J Thrombosis Thrombolysis. 2000;9: Van den Besselaar AMHP. Field study for lyophilized plasmas for local prothrombin time calibration in the Netherlands. J Clin Pathol. 1997;50: