A Percent Correction Formula for Evaluation of Mixing Studies

Transcription

1 Coagulation and Transfusion Medicine / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES A Percent Correction Formula for Evaluation of Mixing Studies Sheng-hsiung Chang, MD, Veronica Tillema, MT(ASCP), and Doris Scherr, MT(ASCP) Key Words: Prothrombin time; Activated partial thromboplastin time; Patient plasma; Citrated normal plasma; Mixing studies Abstract The study examined a percent correction formula for evaluating mixing study results comparing a 1:1 mix with a new 4:1 mix of patient plasma with citrated normal plasma for a prolonged activated partial thromboplastin time (aptt) and/or prothrombin time (PT). The study also examined 3 suggested definitions of correction for evaluating mixing study results for comparison. Applicability of percent correction for evaluating the aptt 4:1 mix testing with and without incubation also was studied. Our results showed that percent correction of the aptt or PT 4:1 mix had an overall good sensitivity and specificity for detecting anticoagulant and factor deficiency and was better than that of the aptt or PT 1:1 mix. The 3 suggested definitions of correction all had a poor sensitivity for detecting anticoagulant. The percent correction of the aptt 4:1 mix testing after incubation had better sensitivity and specificity that that of testing immediately. Nevertheless, these procedures were complementary to each other. The percent correction using the aptt or PT 4:1 mix seemed to offer a simple, objective, and effective criterion for evaluating mixing study results. In 1994, our laboratory began to use a percent correction of a prolonged prothrombin time (PT) or activated partial thromboplastin time (aptt) in a 1:1 mix of patient plasma (PP) with citrated normal plasma (CNP) for evaluation of mixing study results after a retrospective study. 1 The percent correction is expressed in the following formula: Percent Correction = PP PT (or aptt) 1:1 Mix PT (or aptt) 100 PP PT (or aptt) CNP PT (or aptt) For aptt mixing studies, more than 70% correction suggested the presence of a factor deficiency, less than 58% correction suggested the presence of a circulating anticoagulant, and a 58% to 70% correction was considered a borderline result. For PT mixing studies, more than 75% correction suggested the presence of a factor deficiency, less than 70% correction suggested the presence of a circulating anticoagulant, and 70% to 75% correction was considered a borderline result. Based on the percent corrections, we prioritized follow-up testing to workup prolonged PT, aptt, or both with fairly successful results. However, with reagent changes and a change in instrumentation, the classification and testing scheme did not seem to perform as well. We had noticed the 1:1 mix of PP with CNP frequently seemed to overcorrect (false high percent correction); samples with a lupus anticoagulant were misclassified as a factor deficiency. We undertook a reevaluation of the procedure correlated with follow-up testing results. Because it has been suggested a 4:1 mix of PP with CNP is more sensitive for the detection of a lupus anticoagulant, 2 we incorporated a 4:1 mix of PP with CNP into the study. We also studied 3 previously suggested definitions of correction for evaluating mixing study results in the aptt 1:1 62 Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

2 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE mix of PP with CNP for comparision. 3 In addition, applicability of percent correction for evaluating the aptt 4:1 mix testing after incubation also was studied to ascertain the described time-dependent inhibitor effect of some anticoagulants in mixing studies. 3 Materials and Methods Instrument Used in the Study We used the AMAX CS-190 coagulation analyzer (Sigma Diagnostics, St Louis, MO). The following reagents were used: (1) for PT, ThromboMax (Sigma Diagnostics; International Sensitivity Index [ISI], 1.89 [mechanical] except for cases for which a new reagent lot was used, ISI, 1.78); (2) for aptt, Sigma Diagnostics APTT reagent; (3) for CNP, CryoCheck (Precision Biologic, Dartmouth, Nova Scotia); (4) factor-deficient plasma (factors II, V, VII, VIII, IX, X, XI, XII; Sigma Diagnostics); (5) Accuclot Reference Plasma, Normal (for preparation of standard curve for factor assay; Sigma Diagnostics); (6) Accuclot Reference Plasma, Abnormal (for preparation of mild factor-deficient plasma with the aforementioned factor-deficient plasma; Sigma Diagnostics); (7) DRVVT (dilute Russell viper venom time; for lupus anticoagulant), LA- Screen and LA-Confirm (Gradipore, North Ryde, Australia); and (8) PTT-LA (for lupus anticoagulant; StaClot LA, Diagnostica Stago, Asnieres-Surseine, France). Coagulation Testing Prothrombin Time The test was performed using ThromboMax and the automated AMAX CS-190 coagulation analyzer in the mechanical mode. The reference range was 11.0 to 13.0 seconds (except for cases for which a new reagent lot was used; reference range, seconds). Activated Partial Thromboplastin Time The test was performed using Sigma Diagnostic APTT reagent and the automated AMAX CS-190 coagulation analyzer in the mechanical mode. The reference range was 23.3 to 29.7 seconds. Dilute Russell Viper Venom Time The test was performed initially using LA-Screen reagent and the automated AMAX CS-190 coagulation analyzer in the mechanical mode. The reference range was 32.4 to 41.4 seconds. If the result was within reference range, no further testing was performed. The sample was considered negative for lupus anticoagulant. If the result was higher than the reference range, the test was performed with LA-Confirm reagent. The final result was expressed as a ratio of LA-Screen/LA-Confirm clotting time. The ratio of more than 1.3 was considered positive for lupus anticoagulant, 1.3 was borderline, and less than 1.3 was negative. PTT-LA The test was performed initially using StaClot LA reagent with reagent 1 (buffer) and the automated AMAX CS-190 coagulation analyzer in the mechanical mode. The reference range was.1 to 50.4 seconds. If the result was within reference range, no further testing was performed. The sample was considered negative for lupus anticoagulant. If the result was higher than the reference range, the test was performed with StaClot LA reagent with reagent 2 (hexagonal phase phospholipid). The final result was expressed as the difference between the clotting time result with reagent 1 and the result with reagent 2. A difference of more than 3.2 seconds was considered positive for lupus anticoagulant, a difference of less than 3.2 seconds was considered negative, and a difference of 3.2 seconds was considered borderline. Factor Assays Factor assays were performed using the automated AMAX CS-190 coagulation analyzer. Assays for factors II, V, VII, and X used a PT-based assay, while assays for factors VIII, IX, XI, and XII used an aptt-based assay. For each assay, a standard curve was prepared with Accuclot normal reference plasma and factor-deficient plasma. The clotting times of dilutions of the plasma to be tested were obtained, and the factor activity of each dilution was determined from the standard curve. The reference range was 50% to 150% ( ). Plasma Samples We used 3 sources of plasma samples: (1) patient plasma samples with an abnormal PT and/or aptt submitted for routine 1:1 mixing studies and eventually worked-up for a factor deficiency or circulating anticoagulant; (2) selected plasma samples from patients receiving oral anticoagulant therapy; and (3) factor-deficient plasma samples routinely used to perform factor assays were used as severe factor deficiency plasma samples; 1:1 mixtures of a specific factor-deficient plasma sample with the abnormal factor-deficient control plasma (Accuclot Reference Plasma, Abnormal) were used as mild factor-deficient plasma samples. The mixtures had specific factor levels of approximately 25% to 37% ( ). Plasma samples (from sources 1 and 2) were obtained via atraumatic venipuncture and collected into a 3.2% buffered sodium citrate solution with a ratio of 9 parts whole blood to 1 part anticoagulant. The citrated whole blood was centrifuged at no less than 2,500g for 15 minutes to obtain platelet-poor plasma with a platelet count of less than /µl ( /L). American Society for Clinical Pathology Am J Clin Pathol 2002;117:

3 Chang et al / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES Table 1 Summary of Results for Plasma Samples (Part 1 Study) * Case No. PT aptt Group Pertinent Laboratory Data 1 N DRVVT+ 2 N Indeterminate 3 N PTT-LA+ 4 N PTT-LA Factor VII, 14% (0.14); factor V, 34% (0.34); factor X, 54% (0.54) DRVVT+; PTT-LA DRVVT N 1 Factor VII, 39% (0.39) PTT-LA+ 10 N PTT-LA+ 11 N 30 3 PTT-LA+ 12 N Indeterminate Factor VII, 1% (0.01); factor X, 11% (0.11); factor V, 96% (0.96) Factor VII, 5% (0.05); factor X, 19% (0.19); factor V 120% (1.20) 15 N DRVVT+; PTT-LA+ 16 N PTT-LA+ 17 N Indeterminate N 1 Factor VII, 14% (0.14) N 1 Factor VII, 48% (0.48) N 1 Factor VII, 21% (0.21) Factor VII, 6% (0.06); factor X, 47% (0.47); factor V, 101% (10.1) 22 N PTT-LA+ 23 N DRVVT+; PTT-LA Factor VII, 32% (0.32); factor X, 43% (0.43); factor V, 75% (0.75) Factor V, % (0.); factor VII, 65% (0.65); factor X, 65% (0.65) DRVVT+; PTT-LA DRVVT Indeterminate N 1 Factor VII, 32% (0.32) 30 N PTT-LA INR, INR, INR, INR, INR, N Factor VIII, <1% (<0.01) 37 N Factor VIII, approximately 26% (0.26) 38 N Factor IX, <1% (<0.01) 39 N 34 1 Factor IX, approximately 32% (0.32) N Factor XII, <1% (<0.01) 41 N Factor XII, approximately 35% (0.35) N 1 Factor VII, <1% (<0.01) N 1 Factor VII, approximately 32% (0.32) Factor II, <1% (0.01) Factor II, approximately 30% (0.30) Factor V, <1% (<0.01) Factor V, approximately 37% (0.37) Factor X, <1% (<0.01) Factor X, approximately 33% (0.33) 50 N Factor XI, <1% (<0.01) 51 N Factor XI, approximately 25% (0.25) +, positive for lupus anticoagulant; aptt, activated partial thromboplastin time; DRVVT, dilute Russell viper venom time; INR, international normalized ratio; N, normal; PT, prothrombin time; PTT-LA, PTT-LA (for lupus anticoagulant): StaClot LA, Diagnostica Stago, Asnieres-Surseine, France. * Reference ranges: PT, seconds; for cases 18-30, seconds (new reagent lot); aptt, seconds. Group 1, factor deficiency; group 2, indeterminate; group 3, anticoagulant. The plasma samples were classified into 3 groups representing a factor deficiency, circulating anticoagulant, or indeterminate, based on the follow-up results and available information about the samples. Cases with any factor assay result of less than 50% (<0.50) were classified in the factor-deficiency group (group 1). Cases with a positive DRVVT (LA- Screen and LA-Confirm) and/or positive PTT-LA (StaClot LA) were classified in the circulating anticoagulant group (group 3). One case with factor VIII inhibitor was classified in the anticoagulant group. Cases were classified in the indeterminate group (group 2) if available factor assay results showed more than 50% (>0.50) and were negative for circulating anticoagulants with the DRVVT and PTT-LA tests. The Part 1 Study The study involved 51 samples Table 1. Included were 30 patient samples received for mixing studies (cases 1-30), 5 samples from patients receiving oral anticoagulant therapy 64 Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

4 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE Table 2 Summary of Results for Plasma Samples (Part 2 Study) * Case No. PT aptt Group Pertinent Laboratory Data DRVVT+; PTT-LA DRVVT Indeterminate DRVVT DRVVT+; factor VII, 34% (0.34); factor V, 106% (1.06); factor X, 106% (1.06) 57 N DRVVT+; PTT-LA+ 58 N Factor XII, 43% (0.43); factor VIII, 105% (1.05); factor IX, 86% (0.86); factor XI, 97% (0.97) INR, INR, INR, INR, INR, Factor V, <1% (<0.01) Factor V, approximately 37% (0.37) 66 N Factor VIII, <1% (<0.01) 67 N Factor VIII, approximately 26% (0.26) 68 N Factor IX, <1% (<0.01) 69 N Factor IX, approximately 32% (0.32) 70 N Factor VIII, <1% (<0.01); factor VIII inhibitor, 12.4 Bethesda U 71 N 34 3 DRVVT+ 72 N PTT-LA PTT-LA+ 74 N DRVVT+; PTT-LA DRVVT+; PTT-LA+ +, positive for lupus anticoagulant; aptt, activated partial thromboplastin time; DRVVT, dilute Russell viper venom time; INR, international normalized ratio; N, normal; PT, prothrombin time; PTT-LA, PTT-LA (for lupus anticoagulant): StaClot LA, Diagnostica Stago, Asnieres-Surseine, France. * Reference ranges: PT, seconds; aptt, seconds. For a description of the groups, see Table 1. with international normalized ratios ranging from 2.1 to 4.0 (cases 31-35), and 16 artificially prepared factor-deficient samples that included severe and mild deficiencies of factors II, V, VII, VIII, IX, X, XI, and XII (cases 36-51). An initial PT and aptt were performed on all samples. Mixing studies were performed on samples with a prolonged PT and/or aptt result using a 1:1 mixture and a 4:1 mixture of PP/CNP. A percent correction was calculated based on the original formula and the following formula, respectively: Percent Correction = PP PT (or aptt) 4:1 Mix PT (or aptt) 100 PP PT (or aptt) CNP PT (or aptt) The aptt or PT 1:1 and 4:1 mix results, CNP aptt or PT results, their respective percent correction, and group classification (as defined in Plasma Samples ) were compiled. For the aptt or PT 1:1 mix, the original criteria 1 of percent correction cutoff values were used for evaluation. Sensitivity and specificity for detecting anticoagulant and factor deficiency were calculated based on the cutoff values. For the aptt or PT 4:1 mix, a dot plot was made for initial correlation of percent correction with the groups. 4 An optimal cutoff then was selected for differentiation of factor deficiency and anticoagulant. Sensitivity and specificity for detecting factor deficiency and anticoagulant were calculated based on the selected cutoff for evaluation. 5 The 3 definitions of correction suggested by Brandt et al 3 were as follows: (1) aptt 1:1 mix result less than or equal to the upper limit of normal, (2) aptt 1:1 mix result less than or equal to the CNP aptt plus 5 seconds, and (3) a Rosner index of 15 or less. The Rosner index was calculated from the following formula: Index = 1:1 Mix aptt CNP aptt 100 PP aptt For definitions 1 and 2, the actual aptt 1:1 mix results and the CNP aptt results were used for evaluation. The Rosner index for each case was calculated from the preceding formula. Sensitivity and specificity for each category then were calculated for comparison. The Part 2 Study The study involved 24 samples Table 2. Included were 7 samples received for mixing studies (cases 52-58); 5 samples from patients receiving oral anticoagulant therapy with international normalized ratios ranging from 2.5 to 3.6 (cases 59-63); 6 artificially prepared factor-deficient samples that included severe and mild deficiencies of factors V, VIII, and IX (cases 64-69); and 6 patient samples with circulating anticoagulant, including 1 with a factor VIII inhibitor (case 70) and 5 samples with a documented lupus anticoagulant (cases 71-75). American Society for Clinical Pathology Am J Clin Pathol 2002;117:

5 Chang et al / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES Table 3 Results for aptt 1:1 and 4:1 Mixes and Percent Corrections and Group Classification * Case No. aptt CNP 1:1 Mix % Correction 4:1 Mix % Correction Group aptt, activated partial thromboplastin time; CNP, citrated normal plasma. * All data given in seconds except % correction. aptt reference range, seconds. For a description of the groups, see Table 1. For this part of the study, mixing studies were performed only on samples with a prolonged aptt. The mixing studies used a 4:1 mixture of PP/CNP and were tested immediately and after 60 minutes of incubation in a 37 C water bath. The CNP was incubated in the same water bath used for incubating the test mixture. The percent correction for the immediate testing and percent correction after incubation then were calculated using the same percent correction formula as for the aptt 4:1 mix. The aptt 4:1 mix results and the CNP aptt results for testing immediately and testing after incubation, the immediate percent correction and the incubated percent correction, and group classification (as defined in Plasma Samples ) were compiled. For the aptt 4:1 mix tested immediately, the same percent correction cutoff of 50% or more in the part 1 study of the aptt 4:1 mix was used for evaluation. For the aptt 4:1 mix tested after incubation, a dot plot was made for initial correlation and selection of an optimal cutoff value. Sensitivity and specificity for detecting factor deficiency and anticoagulant were calculated based on the selected cutoff value for evaluation. Results The Part 1 Study In the part 1 study, 44 samples had a prolonged aptt and 30 samples had prolonged PT. The results of the aptt 1:1 and 4:1 mixes, percent correction, and group classification are shown in Table Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

6 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE Percent Correction Percent Correction Group Group Figure 1 Results for activated partial thromboplastin time 1:1 mix, correlation of percent correction with groups, cutoffs at >70% and <58% (dotted lines, original criteria). Group 1, factor deficiency (n = 25); group 2, indeterminate (n = 4); group 3, anticoagulant (n = 15). Figure 2 Results for activated partial thromboplastin time 4:1 mix, correlation of percent correction with groups, cutoff at 50% or more (dotted line, new criterion). For a description of the groups, see Figure 1. The correlation of the aptt 1:1 mix percent correction with group classification is shown in Figure 1. The original criteria 1 of percent correction cutoff at more than 70% and at less than 58% were used in the evaluation. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the criteria are shown in Table 4. There was a slight discrepancy in case number between cases correctly classified and sensitivity because there were 2 cutoff values in the original criteria. The cases with percent correction values between 70% and 58% were considered borderline and were excluded in the calculation of sensitivity and specificity. All indeterminate cases also were excluded in all calculations of sensitivity and specificity. The correlation of the aptt 4:1 mix percent correction with group classification is shown in Figure 2. The percent Table 4 Sensitivity and Specificity for Detecting Factor Deficiency and Anticoagulant of aptt 1:1 and 4:1 Mixes at Defined Percent Correction Cutoffs aptt Mix Group Cases Correctly Classified Sensitivity Specificity (%) 1.1 % Correction cutoff at >70% and at <58% Factor deficiency 22/25 (88%) 22/22 (100%) 33 (original criteria) Anticoagulant 4/15 (27%) 4/12 (33%) 100 4:1 % Correction cutoff at 50% or more Factor deficiency 22/25 88% 100 Anticoagulant 15/15 100% 88 aptt, activated partial thromboplastin time. * Percent correction cutoff >70% positive for factor deficiency; <58% positive for anticoagulant; <70% and >58% considered borderline, not included in calculation of sensitivity and specificity. As a result there is a slight discrepancy in case number between cases correctly classified and sensitivity. American Society for Clinical Pathology Am J Clin Pathol 2002;117:

7 Chang et al / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES Table 5 Results for PT 1:1 and 4:1 Mixes and Percent Correction and Group Classification * Case No. PT CNP 1:1 Mix % Correction 4:1 Mix % Correction Group CNP, citrated normal plasma; PT, prothrombin time. * All data given in seconds except % correction. PT reference range: seconds; cases 18-30, seconds (new reagent lot). For a description of the groups, see Table 1. correction of 50% or more was the selected cutoff value. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the cutoff of 50% or more also are shown in Table 4. The results of the PT 1:1 and 4:1 mixes, percent correction, and group classification are shown in Table 5. The correlation of the PT 1:1 mix percent correction with the group classification is shown in Figure 3. The original criteria 1 of percent correction cutoff at more than 75% and at less than 70% were used in the evaluation. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the criteria are shown in Table 6. There was a slight discrepancy in case number between cases correctly classified and sensitivity for the same reason described for the aptt 1:1 mix. The correlation of the PT 4:1 mix percent correction with the group classification is shown in Figure 4. The percent correction of more than % was the selected cutoff value. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the cutoff of more than % also are shown in Table 6. The sensitivity and specificity using the 3 suggested definitions for correction for evaluating the aptt 1:1 mix based on patient aptt, CNP aptt, and 1:1 mix aptt results in Table 3 are shown in Table 7. The Part 2 Study The part 2 study involved 24 samples with a prolonged aptt. The results of the aptt 4:1 mix with immediate testing or testing after a 60-minute incubation, percent correction, and group classification are shown in Table 8. The correlation of the aptt 4:1 mix percent correction for immediate testing with group classification is shown in Figure 5. The same percent correction cutoff of 50% or more as in the part 1 study of the aptt 4:1 mix was used in the evaluation. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the cutoff of 50% or more are shown in Table 9. The correlation of the aptt 4:1 mix percent correction for testing after incubation with group classification is shown in Figure 6. The percent correction of more than 10% was the selected cutoff value. The sensitivity and specificity for detecting factor deficiency and anticoagulant based on the percent correction cutoff of more than 10% are also shown in Table Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

8 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE Percent Correction 60 Percent Correction Group Figure 3 Results for prothrombin time 1:1 mix, correlation of percent correction with groups, cutoffs at >75% and <70% (dotted lines, original criteria). Group 1, factor deficiency (n = 24); group 2, indeterminate (n = 1); group 3, anticoagulant (n = 5) Group Figure 4 Results for prothrombin time 4:1 mix, correlation of percent correction with groups, cutoff at >% (dotted line, new criterion). For a description of the groups, see Figure 3. Discussion Mixing studies have long been used for the evaluation of an unexplained prolongation of PT and aptt. If a prolonged PT or aptt of patient plasma is corrected to normal when mixed with CNP, a factor deficiency is indicated. 6,7 If a prolonged PT or aptt is not corrected to normal when mixed with CNP, a circulating anticoagulant is indicated. The principle of the mixing study seems simple, but results often are difficult to interpret in practice. 8 There is no uniform agreement as to what criteria should be used to judge correction. 3 The normal range usually is used as the guide for correction. However, it often is inadequate because a circulating anticoagulant with a minimally prolonged PT or aptt may be corrected to normal in a 1:1 mix of PP with CNP. 8 In contrast, factor deficiencies with a markedly prolonged PT or aptt may not be corrected to normal in a 1:1 mix of PP with CNP. 3 As to the mixing study procedure itself, 1:1 mix of PP with CNP often is used. 2,3,6-8 A 4:1 mix of PP with CNP also Table 6 Sensitivity and Specificity for Detecting Factor Deficiency and Anticoagulant of PT 1:1 and 4:1 Mixes at Defined Percent Correction Cutoffs PT Mix Group Cases Correctly Classified Sensitivity Specificity (%) 1.1 % Correction cutoff at >75% and at <70% Factor deficiency 21/24 (88%) 21/22 (95%) 50 (original criteria) Anticoagulant 2/5 (%) 2/4 (50%) 95 4:1 % Correction cutoff at >% Factor deficiency 23/24 96% 100 Anticoagulant 5/5 100% 96 PT, prothrombin time. * Percent correction cutoff >75% positive for factor deficiency; <70% positive for anticoagulant; <75% and >70% considered borderline, not included in calculation of sensitivity and specificity. As a result there is a slight discrepancy in case number between cases correctly classified and sensitivity. American Society for Clinical Pathology Am J Clin Pathol 2002;117:

9 Chang et al / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES Table 7 Sensitivity and Specificity for Detecting Factor Deficiency and Anticoagulant of aptt 1:1 Mix Using Three Suggested Definitions 3 of Correction * aptt 1:1 Mix Group Cases Correctly Classified Sensitivity (%) Specificity (%) Upper limit of normal Factor deficiency 22/ Anticoagulant 2/ CNP aptt + 5 seconds Factor deficiency 23/ Anticoagulant 1/ Rosner index 15 or less Factor deficiency 25/ Anticoagulant 1/ aptt, activated partial thromboplastin time; CNP, citrated normal plasma. * Based on data in Table 3. See the text. Table 8 Results for aptt 4:1 Mix, Immediate or Incubated, and Percent Correction and Group Classification * Case No. aptt CNP 4:1 Mix % Correction CNP 4:1 Mix % Correction Group aptt, activated partial thromboplastin time; CNP, citrated normal plasma. * All data given in seconds except % correction. aptt reference range, seconds. For a description of the groups, see Table minute incubation. has been described. 2,3,7,8 However, it was not clear from the publications which of the 2 procedures was preferred. Kaczor et al 2 suggested that the 4:1 mix of PP with CNP might be more sensitive for detecting a lupus anticoagulant, while Brandt et al 3 found that the 4:1 mix was ineffective for correcting the aptt of the mixture and that it might misclassify samples with factor deficiency as anticoagulant. Our results for the aptt 1:1 mix with percent correction cutoff at more than 70% and at less than 58% (original criteria) (Figure 1 and Table 4) showed a very poor sensitivity (33%) for detecting anticoagulant, thus confirming our impression of overcorrection in the mixing studies resulting in frequent misclassification of samples with lupus anticoagulant as samples with factor deficiency. Results of the aptt 4:1 mix with percent correction cutoff at 50% or more (Figure 2 and Table 4) showed markedly improved sensitivity (100%) for detecting anticoagulant while maintaining a good specificity (88%). In addition, there was a better separation of percent correction of anticoagulant from that of factor deficiency in the 4:1 mix procedure, ie, 1 cutoff value was reasonably sufficient in classifying the anticoagulant and factor deficient groups, while the 1:1 mix procedure required 2 cutoff values owing to more overlapping borderline percent correction values between the anticoagulant group and the factor deficiency group in the original study in As shown in Table 4, at a defined percent correction cutoff value, the sensitivity for detecting factor deficiency 70 Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

10 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE Percent Correction Percent Correction Group Figure 5 Results for activated partial thromboplastin time 4:1 mix, immediate (part 2 study), correlation of percent correction with groups, cutoff at 50% or more (dotted line, new criterion). Group 1, factor deficiency (n = 12); group 2, indeterminate (n = 1); group 3, anticoagulant (n = 11) Group Figure 6 Results for activated partial thromboplastin time 4:1 mix, incubated (part 2 study), correlation of percent correction with groups, cutoff at >10% (dotted line, new criterion). For a description of the groups, see Figure 5. coincided with the specificity for detecting anticoagulant and vice versa. Therefore, no further discussion of sensitivity and specificity for detecting factor deficiency at the same cutoff value is necessary. There were 2 interesting observations: (1) Plasma samples with more severe factor deficiency tended to have higher percent corrections. They usually had percent corrections much higher than the cutoff of 50%. Samples with mild factor deficiency (eg, cases 5, 37, 39, 49 in Table 3, aptt 4:1 mix) tended to have borderline percent correction. Occasional cases with mild factor deficiency (eg, cases 24 and 25 in Table 3, aptt 4:1 mix) had percent correction below the cutoff of 50%. (2) One sample with a lupus anticoagulant (case 30) had a percent correction of 13% resulting from a more prolonged aptt on the mixture than on the patient plasma alone. This phenomenon has been termed the lupus cofactor effect. 8,9 Results of the PT 1:1 mix with percent correction cutoffs at more than 75% and at less than 70% (original criteria) (Figure 3 and Table 6) also showed a poor sensitivity (50%) Table 9 Sensitivity and Specificity for Detecting Factor Deficiency and Anticoagulant of aptt 4:1 Mix, Immediate and Incubated, at Defined Percent Correction Cutoff aptt 4:1 Mix Group Cases Correctly Classified Sensitivity (%) Specificity (%) Immediate % Correction cutoff at 50% or more Factor deficiency 9/ Anticoagulant 10/ Incubated % Correction cutoff at >10% Factor deficiency 12/ Anticoagulant 11/ aptt, activated partial thromboplastin time American Society for Clinical Pathology Am J Clin Pathol 2002;117:

11 Chang et al / A PERCENT CORRECTION FORMULA FOR EVALUATION OF MIXING STUDIES for detecting anticoagulant. As in the aptt 4:1 mix, the results for the PT 4:1 mix with a percent correction cutoff at more than % showed markedly improved sensitivity (100%) for detecting anticoagulant while maintaining very good specificity (96%). In addition, as in the aptt 4:1 mix, the PT 4:1 mix had a better separation of samples with factor deficiency from samples with anticoagulant. Only 1 cutoff value was needed in the PT 4:1 mix compared with 2 cutoff values in the PT 1:1 mix procedure for the same reason described for the aptt 4:1 mix. Like the aptt 4:1 mix, the samples with more severe factor deficiency tended to have higher percent correction. As shown in Table 7, all 3 suggested definitions of correction for evaluating the aptt 1:1 mix 3 showed poor sensitivity (13%, 7%, and 7%, respectively) for detecting anticoagulant. Brandt et al, 3 in their evaluation of these 3 definitions of correction, also showed less than satisfactory results. The aptt 4:1 mix tested immediately in the part 2 study was the same procedure as used for the aptt 4:1 mix in the part 1 study. We applied the same percent correction cutoff of 50% or more established in the part 1 study to the second set of samples of the part 2 study for evaluation (Figure 5 and Table 9). The sensitivity and specificity of the aptt 4:1 mix tested immediately using the percent correction cutoff of 50% or more for detecting anticoagulant were 91% and 75%, respectively, compared with 100% and 88% in the part 1 study (Table 4). They had fairly comparable sensitivity and specificity values. Furthermore, review of Figures 2 and 5 showed a similar pattern of percent correction distribution. Both the part 1 study and the part 2 study showed a few cases of factor deficiency with false-negative results (below the cutoff of 50% or more). Those cases generally had mild factor deficiency, such as cases 24 and 25 in the part 1 study and cases 65, 67, and 69 in the part 2 study. One case stood out in the anticoagulant group in the part 2 study of the aptt 4:1 mix with immediate testing (Figure 5 and Table 8). Case 70 with factor VIII inhibitor had a percent correction of more than 50% and was classified as factor deficiency on immediate testing. For aptt 4:1 mix testing after incubation, the selected percent correction cutoff of 10% was much lower than that for 4:1 mix immediate testing. However, the cutoff had a good separation of the anticoagulant group from the factordeficient group (Figure 6 and Table 9). The sensitivity and specificity of the aptt 4:1 mix tested after incubation with the percent correction cutoff at more than 10% were each 100% for detecting anticoagulant compared with 91% and 75% for the aptt 4:1 mix tested immediately with the percent correction cutoff of 50% or more. This was a substantial improvement. Table 10 Interpretation of Activated Partial Thromboplastin Time 4:1 Mixing Studies Immediate % Incubated % % Correction Correction Correction Results Suggest 50% or more >10% Factor deficiency <50% >10% Mild factor deficiency 50% or more 10% or less Factor inhibitor <50% 10% or less Lupus anticoagulant Three cases (cases 65, 67, and 69) with mild factor deficiency were misclassified as anticoagulant by immediate testing (immediate percent correction <50%). They were classified correctly as factor deficiency on testing after incubation (incubated percent correction >10%). Case 70 with factor VIII inhibitor, initially classified as factor deficiency according to the immediate percent correction (percent correction >50%) was classified correctly as circulating anticoagulant on testing after incubation (percent correction <10%). A similar incubation effect in an aptt mixing study of factor VIII inhibitor has been described. 3,10 Although the percent correction of the aptt 4:1 mix testing after incubation seemed to have better sensitivity and specificity for detecting anticoagulant and factor deficiency than that of the aptt 4:1 mix with immediate testing, both incubated and immediate testing procedures were complementary to each other for interpretation of mixing study results Table 10. An additional interesting observation was that in the part 2 study, 8 of 10 cases (cases 52, 53, 55, 57, 71, 72, 73, and 74) with lupus anticoagulant showed the lupus cofactor effect after incubation, while none of those cases and only 1 case (case 30) of 15 cases with lupus anticoagulant in the part 1 study showed the lupus cofactor effect on immediate testing. Incubation of the mixture seemed to enhance the lupus cofactor effect and made samples with a lupus anticoagulant more distinct from samples with a factor deficiency. In addition to good sensitivity and good specificity, the percent correction formula in combination of aptt or PT 4:1 mixing procedure offers an additional advantage in that it gives an objective and quantitative measure for differentiating samples with anticoagulant from those with factor deficiency, even samples with minimally or markedly prolonged aptt or PT results as demonstrated in our study cases (Tables 3, 5, and 8). Whether the percent correction procedure is applicable to other combinations of instruments and reagents is an open question. Our experience of adapting the procedure to our new instrument and new reagents suggested that the procedure could be applied to other instruments and reagent 72 Am J Clin Pathol 2002;117:62-73 American Society for Clinical Pathology

12 Coagulation and Transfusion Medicine / ORIGINAL ARTICLE combinations as well. However, like most laboratory procedures, each laboratory must establish its own reference cutoff values to derive meaningful interpretations. Twenty to 30 samples each of anticoagulant and factor deficiency cases probably suffice for study to establish cutoff values. Conclusion The study showed that the percent correction for the aptt or PT 4:1 mix had an overall good sensitivity and specificity for detecting anticoagulant and factor deficiency. The percent correction for the aptt or PT 4:1 mix had better sensitivity and specificity than that for the aptt or PT 1:1 mix. The 3 definitions of correction suggested by Brandt et al 3 showed a poor sensitivity for detecting anticoagulant in our study. The percent correction of the aptt 4:1 mix tested after incubation showed an improvement in sensitivity and specificity for detecting anticoagulant and factor deficiency compared with the percent correction of the aptt 4:1 mix tested immediately. Nevertheless, these procedures were complementary for interpretation of mixing studies. Therefore, both the immediate and the incubated percent corrections are useful. The interpretation of percent correction results of the aptt 4:1 mix are summarized in Table 10. No incubation study of PT was performed. The percent correction of the aptt or PT 4:1 mix of PP with CNP testing immediately and testing after incubation (aptt only) seemed to offer a simple, objective, and effective criterion for evaluating mixing study results. From the Hematology Laboratory, Marshfield Clinic, Marshfield, WI. Address reprint requests to Dr Chang: Dept of Pathology, Marshfield Clinic, 1000 N Oak Ave, Marshfield, WI References 1. Chang S, Tillema V. Mixing study procedure changes. Laboratory News Update Marshfield Laboratories. March 8, 1994;17(3): Kaczor DA, Bickford NN, Triplett DA. Evaluation of different mixing study reagents and dilution effect in lupus anticoagulant testing. Am J Clin Pathol. 1991;95: Brandt JT, Barna LK, Triplett DA. Laboratory identification of lupus anticoagulant: results of the Second International Workshop for Identification of Lupus Anticoagulant. Thromb Haemost. 1995;74: Krieg AF, Beck JR, Bongiovanni MB. The dot plot: a starting point for evaluating test performance. JAMA. 1988;260: Annesley TM. Predictive value theory. In: McClatchey KD, ed. Clinical Laboratory Medicine. Baltimore, MD: Williams & Wilkins; 1994: Miletich JP. Activated partial thromboplastin time. In: Beutler E, Lichtman MA, Coller BS, et al, eds. Williams Hematology. 5th ed. New York, NY: McGraw-Hill; 1995:L85-L Bockenstedt PL. Evaluation of a prolonged prothrombin or activated partial thromboplastin time. In: Loscalzo J, Schafer AI, eds. Thrombosis and Hemorrhage. 2nd ed. Baltimore, MD: Williams & Wilkins; 1998: Triplett DA. Laboratory diagnosis of lupus anticoagulant. Semin Thromb Hemost. 1990;16: Rivard GE, Schiffman S, Rapaport SI. Cofactor of the lupus anticoagulant. Thromb Diath Haemorrh. 1974;32: Kasper CK. Laboratory tests for factor VIII inhibitors, their variation, significance, and interpretation. Blood Coagul Fibrinolysis. 1991;2(suppl 1):7-10. American Society for Clinical Pathology Am J Clin Pathol 2002;117: