Results
Platelet reactivity was tested in 466 consecutive patients with VN and LTA ADP 5 and 20 μM. Baseline characteristics stratified by hematocrit ≥39% are presented in Table I. To evaluate the influence of hematocrit upon platelet function assays, further analysis was done by dividing the patients into quartiles (Q1: 25–35.9, Q2: 36–38.9, Q3: 39–41.9, Q4: 42–51) based upon baseline hematocrit (percentages). The patients in Q1 and Q2 were considered anemic, and Q3 and Q4 were considered nonanemic.
Platelet Function Assays
In the overall study population, PRU as measured by VN assay strongly correlated with device-reported percent inhibition (r = 0.90, P < .001), MPA measured by LTA ADP 5 μM (r = 0.71, P < .001), and MPA as measured by LTA ADP 20 μM (r = 0.77, P < .001). Mean PRU and iso-TRAP BASE value (obtained with iso-TRAP as agonist) varied significantly with changing hematocrit (Figure 1A and B ). Compared between Q1 and Q4, the mean PRU (209 ± 119 vs 143 ± 91, P < .001) and mean BASE value (345 ± 51 vs 264 ± 52, P < .001) were significantly different. The device reported percent inhibition did not vary with hematocrit (P = .34) (Figure 1C). No differences were noted in the mean platelet reactivity among the 4 hematocrit quartiles as measured by LTA ADP 5 μM (P = .23) and LTA ADP 20 μM (P = .21). (Figure 1D) As shown in Figure 2, lower hematocrit was associated with a higher rate of HTPR as measured by VN assay (P < .001) but not with LTA ADP 5 μM ≥46% (P = .79) and LTA ADP 20 μM ≥59% (P = .37).
(Enlarge Image)
Figure 1.
Distribution of the results of PRU base (A), PRU (B), device-reported percent inhibition (C), and LTA ADP 20 μM (D) according to hematocrit (percentages) quartiles (Q1: 25–36; Q2: 36–39; Q3: 39–42; Q4: 42–51). The horizontal line in the middle of each box indicates the median; the top and bottom borders of each box indicate the 25th and 75th percentile value, whereas whiskers denote lowest and highest values.
(Enlarge Image)
Figure 2.
Percentage of patients with HTPR according to hematocrit (percentages) quartiles. A clear influence of hematocrit in defining the patients as HTPR by traditionally used cutoffs was seen only by PRU but not by LTA ADP 5 μM ≥46% and LTA ADP 20 μM ≥59%.
Correlation Between Hematocrit and Platelet Function Assays
Pearson correlation coefficients showed moderate to good correlation between hematocrit and iso-TRAP BASE (−0.52, P < .001) and VN PRU (−0.24, P < .001) but poor correlation with LTA ADP 20 μM (−0.08, P = .09). In a multivariable logistic regression model (Table II), lower baseline hematocrit was independently associated with HTPR by PRU ≥208 (odds ratio [OR] 0.92, 95% CI 0.86–0.97, P = .005). Hematocrit was not independently associated with HTPR by LTA ADP 5 μM ≥46% (OR 1.0, 95% CI 0.95–1.06, P = .88) or LTA ADP 20 μM ≥59% (OR 1.03, 95% CI 0.97–1.09, P = .39).
Clinical Outcomes
One-year follow-up was available for 375 patients (81%). Patients with anemia had higher composite MACE (8.8% vs 3.6%, P = .03) compared with those without anemia. Receiver operating characteristic curves were analyzed for each platelet function assay for prediction of 1-year MACE. VerifyNow P2Y12 PRU (AUC 0.69, P = .002), LTA ADP 5 μM (AUC 0.62, P = .03), and LTA ADP 20 μM (AUC 0.63, P = .04) assays were able to discriminate the patients with or without 1-year MACE. Table III shows the discriminatory power of a logistic model of the relevant clinical variables alone and after adding platelet function assay results, hematocrit, and the interaction between hematocrit and PRU. The addition of PRU, hematocrit, and the interaction between hematocrit and PRU had a significant effect upon the AUC for prediction of MACE compared with baseline clinical variables (0.63 vs 0.76, P = .006). In a similar analysis using LTA ADP 20 μM, the addition of LTA, hematocrit, and interaction between hematocrit and LTA had no influence on the AUC for prediction of MACE compared with baseline clinical variables (0.64 vs 0.74, P = .13).