- Top of page
Background: Platelet dysfunction contributes to the pathophysiology of bleeding complications during and after cardiac surgery. In most surgical institutions, no peri-operative point-of-care monitoring of platelet function is used. We evaluated the usefulness of the Multiplate® platelet function analyser based on impedance aggregometry for identifying groups of patients at a high risk of transfusion of platelet concentrates (PC).
Methods: Platelet function parameters were determined in 60 patients before and after routine cardiac surgery. Impedance aggregometry measurements were performed on Multiplate® using ADP (ADPtest), collagen (COLtest) and thrombin receptor activating peptide (TRAPtest) as platelet activators. The correlations between the aggregometry results and the transfusion of PC were calculated. The results of the aggregation tests were also divided into tertiles and the differences in PC transfusion between the low and the high tertile were assessed.
Results: Low aggregometry delimited groups of patients with significantly higher PC transfusion. In the receiver operating characteristic curve, low pre-operative aggregation in the ADPtest identified patients with high total transfusion of PC (area under the curve 0.74, P=0.001), while the ADPtest performed at the end of the operation identified patients with high PC transfusion on the intensive care unit (ICU) (area under the curve 0.76, P=0.002).
Conclusions: Near-patient platelet aggregation may allow the identification of patients with enhanced risk of PC transfusion, both pre-operatively and upon arrival on the ICU.
In the early post-operative period following cardiac surgery, platelets dysfunction represents an important cause of excessive bleeding and increased allogeneic blood products transfusion.1,2 Given the associated risks of infection and of severe immune reactions,3 as well as the scarcity of the resources,4,5 several possibilities to reduce transfusion have been investigated in recent years. One of the newer approaches highlights the importance of quicker and more specific laboratory and point-of-care coagulation analyses. The application of methods that assess platelet function in whole blood has been investigated, with controversial reports on their predictive value for clinical endpoints such as bleeding or transfusion requirements.6–14
Multiple electrode platelet aggregometry (MEA) is a whole-blood impedance aggregometry method that allows the assessment of platelet function both in the laboratory and in the near-patient setting. The method requires no centrifugation steps and the measurements are performed in single-use test cells.15 To date, MEA has proven sensitive for platelet inhibition induced by aspirin and clopidogrel, as well as for the effects of cardiopulmonary bypass (CPB) and of hypothermia on platelet aggregation.14,16–18 MEA usefulness in identifying groups of patients at a high risk of transfusion of platelet concentrates (PC) has not been investigated so far. Therefore, we performed a pilot study evaluating the predictive value of perioperative MEA measurements for PC transfusion in cardiac surgery.
- Top of page
Following Research Ethics Committee approval (reference number 4213), we assessed a sequential cohort of 60 patients undergoing routine cardiac surgery. Before the operation, the patients were interviewed on the intake of aspirin, non-steroidal antiinflammatory drugs, glycoprotein IIb–IIIa antagonists, thienopyridines or antibiotics known to influence platelet aggregation, and their written informed consent was obtained.
The operative and anaesthetic management was similar in all patients. Extracorporeal circulation was maintained by a roller peristaltic pump with a heparin-coated oxygenator and arterial filter. A bolus of heparin 400 IU/kg was administered before institution of CPB and 2 million kallikrein inhibiting units of aprotinin were given on CPB. After the initial anticoagulation, additional doses of heparin were given to maintain activated clotting time above 480 s. After CPB, heparin was neutralized with protamine sulfate, 1 mg protamine/100 U of the total heparin dose.
Blood products transfusion was performed according to the local protocol. The intra-operative trigger for red blood cell (RBC) concentrates was a haematocrit value below 24%. In case of diffuse bleeding after weaning from CPB, neutralization of heparin and completion of the surgical haemostasis, fresh-frozen plasma (FFP) and/or PC were transfused. PC were used as first-line therapy in case of recent intake of platelet aggregation inhibitors; otherwise, FFP were applied as first-line haemostatic therapy. Intra-operative haemostatic transfusion was considered sufficient when the bleeding clinically observed on the mediastinal operation field was reduced to a secure level. Haemostatic therapy on intensive care unit (ICU) was performed according to the chest tube drainage volume. PC, RBC and FFP administered during surgery and in the first 24 h post-operatively as well as the volume of chest tube drainage in the first 24 h on the ICU were documented. The results of the platelet function analysis were not provided to the surgeons, the anaesthetists and the intensivists in charge of the patients.
Blood samples were drawn at the beginning of the operation (before induction of the anaesthesia) and at the end of the operation from a radial artery catheter (20 G) into commercially available pre-filled Monovette® collection tubes (Sarstedt, Nuembrecht, Germany) that contained heparin, citrate or EDTA as an anticoagulant.
MEA was performed on the Multiplate® analyser (Dynabyte medical, Munich, Germany),15 an instrument with five channels for parallel aggregometry measurements and an internal computer system for real-time analysis and documentation. Blood and reagents were pipetted by means of an electronic pipette into single-use test cells containing four metal, silver-covered electrodes that formed two independent sensor units. Each aggregometry test was performed with 300 μl of saline and 300 μl of heparinized blood pipetted into the test cell and stirred using a magnetic stirrer. After an incubation period of 3 min at 37 °C, the agonist was added and the recording started. During the following 6 min, the ability of platelets to adhere to the metal sensors was detected. The adhesion and aggregation of platelets was measured by the change of electrical resistance between two sensor wires. The impedance change caused by the adhesion of the platelets onto the sensor surfaces was plotted against time and the area under the aggregation curve was used to measure the aggregation response, quantified in arbitrary aggregation units (U). Since the change in impedance was measured simultaneously on two sensor units, the results of each test represented the mean value of the two aggregation curves obtained. The following agonists were used for the analysis (test name and final concentration in blood in the parentheses): ADP (ADPtest, 6.5 μM/ml), TRAP-6 (thrombin receptor activating peptide, TRAPtest, 32 μM/ml) and collagen (COLtest, 3.2 μg/ml). The analysis was performed in a near-patient setting, in or in close proximity to the operating theatre.
The standard laboratory tests were run by the hospital laboratory. Activated partial thromboplastin time (aPTT) (APTT Kaolin, Stago Diagnostica, Asnieres, France) and prothrombin time (PT) (Neoplastin, Stago Diagnostica), were determined on the STAR analyser (Stago Diagnostica, Asnieres, France). Platelet count and haematocrit were measured on Sysmex XE-2100 (Roche Diagnostics, Mannheim, Germany).
Demographic and standard laboratory data are presented as mean and standard deviation. Aggregometry and transfusion data are presented as median (minimum, maximum). The distribution of transfusion data was found to be considerably skewed. In order to evaluate whether the results of the assays predicted the transfusion of PC, the aggregation values for the ADPtest, COLtest and TRAPtest were divided into tertiles (0–33°, 34–66° and 67–100° percentile). For each tertile obtained, the median (minimum, maximum) for intra-operative and 24-h post-operative PC transfusion, the primary endpoint, were calculated, as well as for RBC and FFP transfusion and for 24-h post-operative drainage volume. The differences in transfusion, as well as in baseline and operative characteristics between the low and the high tertile were assessed using the Mann–Whitney U-test, and P<0.05 was considered significant. Correction for multiple comparisons was performed using the Bonferroni–Holm correction, with significant P≤0.05.
Patients were also divided into transfused and non-transfused with PC, and receiver operating characteristic (ROC) curves were calculated for the three aggregometry tests regarding the correlation with intra- and post-operative PC transfusion. In addition, the values of the tests performed pre-operatively were compared with the 95% reference interval assessed by the manufacturer of the Multiplate® device in normal subjects (60.7–96.3 for the ADPTtest, 55.4–103.1 for the COLtest and 86.8–147.3 for the TRAPtest). PC transfusion was compared between the patients with normal and with decreased pre-operative values, using the Mann–Whitney U-test, followed by the Bonferroni–Holm correction for the significant P values obtained. The correlation of pre-operative aggregation values with total PC transfusion and with platelet count, as well as the correlation of the post-operative aggregation values with PC transfusion on ICU, with platelet count, CPB time and aortic clamp time were assessed using Pearson's product–moment coefficient r. Data analysis was performed with SPSS 11 and Microsoft Excel 2003.
- Top of page
For the 60 patients included in the study, the mean age was 63.6 (15.4) years, the mean body weight was 78 (16) kg and the body mass index was 26 (4) kg/m2. Eighteen patients were female, and none of the patients had previously undergone heart surgery. Twenty-one patients underwent aorto-coronary bypass surgery (ACB), eight ACB and additional interventions, 12 single-valve and five double-valve replacement, 13 thoracic aorta repair and one pulmonary thrombendarterectomy. The median operation time was 214 min (177, 246 min), the median CPB time was 108 min (78, 143) and the median aortic cross-clamp time was 54 min (41, 84). Aspirin had been discontinued for more than 4 days before surgery in 55 patients. Clopidogrel had been discontinued in 58 patients for at least 1 week before surgery.
Median pre-operative platelet aggregation was within the normal range indicated by the manufacturer of the device (Table 1). Median post-operative aggregation represented 59% of the baseline value for the TRAPtest, 35% for the COLtest and 42% for the ADPtest (Table 2). Haematocrit and platelet count decreased to 73% and 49%, respectively, while PT was prolonged with 15% and aPTT with 7% of the baseline values (Table 3).
Table 1. Pre-operative aggregometry and transfusion parameters.
|Aggregometric test||Aggregometric values for all 60 patients|| ||Aggregometric values for each tertile||Units of transfused blood products for each tertile|
|TRAPtest (U)|| ||First tertile||85.8 (51.2, 95.1)||2 (0, 9)||5 (0, 38)||4.5 (0, 38)|
|Reference interval 87–147||101.5 (51.2, 147.5)||Second tertile||101.5 (95.4, 114.9)||0 (0, 11)||4 (0, 31)||3 (0, 41)|
| ||Third tertile||124.2 (115.3, 147.5)||0 (0, 3)||4 (0, 9)||3.5 (0, 7)|
|COLtest (U)|| ||First tertile||43.4 (17.7, 63.1)||2 (0, 11)*||6 (0, 38)||4.5 (0, 41)|
|Reference interval 55.4–103||71.4 (17.7, 111.1)||Second tertile||71.4 (66.1, 85.8)||0 (0, 3)||3 (0, 8)||3 (0, 8)|
| ||Third tertile||92 (86.1, 111.1)||0 (0, 2)||4 (0, 7)||3 (0, 6)|
|ADPtest (U)|| ||First tertile||49 (19.6, 60.6)||2 (0, 11)*||5.5 (0, 38)||5.5 (0, 41)|
|Reference interval 61–96||67.9 (19.6, 106.1)||Second tertile||67.9 (61.3, 77.7)||0 (0, 2)||3.5 (0, 9)||3 (0, 6)|
| ||Third tertile||88.3 (77.7, 106.1)||0 (0, 2)||4 (0, 8)||3 (0, 8)|
Table 2. Post-operative aggregometry and transfusion parameters.
|Aggregometric test||Aggregometric values for all 60 patients|| ||Aggregometric values for each tertile||Units of transfused blood products for each tertile|
|TRAPtest (U)||59.9 (5.7, 160.9)||First tertile||30 (5.7, 43.5)||0 (0, 7)||0 (0, 15)||0 (0, 25)|
| ||Second tertile||59.9 (43.7, 81.5)||0 (0, 8)||1 (0, 35)||0 (0, 35)|
| ||Third tertile||99.2 (84.1, 160.9)||0 (0, 1)||0.5 (0, 5)||0 (0, 4)|
|COLtest (U)|| ||First tertile||10.1 (2, 16.7)||0 (0, 8)||0 (2, 35)||2 (0, 35)|
|25.2 (2, 111.3)||Second tertile||25.2 (16.8, 29.2)||0 (0, 3)||0 (0, 2)||0 (0, 3)|
| ||Third tertile||53.6 (31.7, 111.3)||0 (0, 1)||0.5 (0, 5)||0 (0, 4)|
|ADPtest (U)|| ||First tertile||13 (1.8, 18.2)||0 (0, 7)*||1 (0, 15)||1.5 (0, 25)|
|28.2 (1.8, 87)||Second tertile||28.2 (18.4, 40.2)||0 (0, 8)||0 (0, 35)||0 (0, 35)|
| ||Third tertile||56.1 (40.3, 87)||0 (0, 1)||1 (0, 5)||0 (0, 4)|
Table 3. Standard laboratory analyses.
|Haematocrit (%)||39.6 (4.1)||29.2 (3.1)|
|Platelet count (1000/ml)||239 (56)||116 (33)|
|PT (s)||14.6 (2.3)||16.8 (1.2)|
|aPTT (s)||31.1 (4.7)||33.4 (6.6)|
The median values for the blood products transfusion within each tertile are presented in Tables 1 and 2. The distribution of the results for the aggregometry tests and the respective requirements of PC transfusion are shown in Fig. 1 (pre-operative analysis vs. transfusion requirements for PC) and Fig. 2 (end of operation analysis vs. 24-h post-operative transfusion requirements for PC). Patients within the low tertile of the pre-operative ADPTtest and COLtest received significantly more PC transfusion than patients within the high tertile (P=0.006 for the ADPtest and P=0.03 for the COLtest, Mann–Whitney U-test, Table 1). For the post-operative aggregometric measurements, patients with the ADPtest within the low tertile received significantly more PC transfusion during the first 24 h on ICU than patients in the high tertile (P=0.028, Mann–Whitney U-test, Table 2). From the 20 patients within the low post-operative ADPtest tertile, 13 belonged to the low pre-operative ADPtest tertile and six to the middle pre-operative ADPtest tertile. The other seven patients from the low pre-operative ADPtest tertile had post-operative ADP test values within the middle tertile. The low tertile in the pre-operative TRAPtest and the post-operative COLtest and TRAPtest also defined groups of patients with higher PC transfusion (P=0.058 for each test).
Figure 1. Total transfusion of platelet concentrates within each tertile of the pre-operative aggregometry tests: low aggregometric tertile (0–33° percentile), middle aggregometric tertile (33–66° percentile) and high aggregometric tertile (67–100° percentile). ADPtest, COLtest and TRAPtest are expressed in aggregation units (U). The differences in platelet transfusion between the low and the high tertile were assessed with the Mann–Whitney U-test, with #significant P≤0.05. The boxes represent the median and the interquartile range, the line represents the median and the whiskers represent the 10–90% interval within the tertile.
Download figure to PowerPoint
Figure 2. Post-operative transfusion of platelet concentrates within each tertile of the post-operative aggregometry tests: low aggregometric tertile (0–33° percentile), middle aggregometric tertile (33–66° percentile) and high aggregometric tertile (67–100° percentile). ADPtest, COLtest and TRAPtest are expressed in aggregation units (U). The differences in platelet transfusion between the low and the high tertile were assessed with the Mann–Whitney U-test, with #significant P≤0.05. The boxes represent the median and the interquartile range; the line represents the median and the whiskers represent the 10–90% interval within the tertile.
Download figure to PowerPoint
No significant differences between the low and the high tertile were noted with respect to 24-h drainage volume, transfusion of FFP, transfusion of RBC and post-operative haematocrit value for any of the three aggregation tests (data not shown). There were no significant differences in the baseline parameters or in the CPB and aortic cross-clamp time between the patients in the low and the high tertile of the pre-operative ADPtest and COLtest. Patients in the low tertile of the post-operative ADPtest had a significantly longer CPB time and aortic cross-clamp time than the patients in the high tertile (P=0.004 and P=0.018, Mann–Whitney U-test).
For the ROC curve of the pre-operative ADPtest, the value of the area under the curve was 0.74 (P=0.001), with 0.77 sensitivity and 0.63 specificity for total PC transfusion at the cut-off value of 66 U (Fig. 3). For the pre-operative COLtest, the value of the area under the curve was 0.69 (P=0.01), with 0.7 sensitivity and 0.63 specificity for total PC transfusion at the cut-off value of 69 U. Regarding the aggregometry analyses performed post-operatively, for the ADPtest the value of the area under the ROC curve was 0.76 (P=0.002), with 0.71 sensitivity and 0.67 specificity for post-operative PC transfusion at the cut-off value of 22 U (Fig. 3). For all other aggregometry tests, the area under the ROC curve was below 0.69 (0.68 for the post-operative COLtest, 0.63 for the pre-operative TRAPtest and 0.67 for the post-operative TRAPtest).
Figure 3. Receiver operating characteristics (ROC) curves for the pre-operative ADPtest in relation to total platelet concentrates transfusion, with a value of 0.74 for the area under the curve and for the post-operative ADPtest in relation to platelet concentrates transfusion on the intensive care unit, with a value of 0.76 for the area under the curve.
Download figure to PowerPoint
Regarding the comparison with the normal range, patients with low pre-operative aggregometry in the ADP test had significantly higher transfusion of PC than patients within the normal range (20 patients compared with 36 patients, P<0.0002). The differences were also significant for the COLtest (17 patients compared with 41 patients, P=0.0003). For the groups with the low, respectively, normal TRAPtest, the differences in PC transfusion were not significant (11 patients and 48 patients, P=0.157).
Pearsons's product–moment r coefficient for the correlations between platelet transfusion and aggregometry measurements is presented in Table 4. The r coefficient for the correlation between pre-operative aggregometry values and platelet count was 0.41 for the ADPtest (P=0.0044), 0.24 for the COLtest (P>0.05) and 0.26 for the TRAPtest (P>0.05). In the post-operative measurements, the r coefficient to the correlation to platelet count was 0.57 for the ADPtest (P=0.0005), 0.39 for the COLtest (P=0.0057) and 0.28 for the TRAPtest (P>0.05). Regarding the post-operative tests, ADPtest showed the highest correlation to CPB time and aortic cross-clamp time (r−0.49, with P<0.0001, and r−0.38, with P=0.003, respectively). The COLtest showed correlations of −0.4 and −0.3 CPB time and aortic clamp time, while the TRAPtest showed correlations of −0.39 and −0.27 with the same parameters. No significant correlation was found either between the pre-operative platelet count and total PC transfusion, or between the post-operative platelet count and PC transfusion on ICU. No significant correlations were found between the CPB time or aortic cross-clamp time and post-operative transfusion.
Table 4. Correlations of PC transfusion with the aggregometric measurements.
| ||Total PC transfusion||PC transfusion on ICU|
| r || P || r || P |
- Top of page
The present study evaluated the utility of a new whole-blood platelet function analyser for monitoring platelet function before and after cardiac surgery performed on CPB. Low pre-operative aggregometry defined groups of patients with significantly higher total transfusion of PC, while decreased post-operative aggregometry correlated with high PC transfusion on ICU.
Because transfusion of blood products has a negative effect on the outcome of surgical interventions, efforts have been made to improve the management of haemostatic therapy according to the findings of standard and additional laboratory methods. In heart surgery, most disturbances in platelet activity are caused by factors related to extracorporeal circulation, like haemodilution, use of heparin and contact between blood and non-physiologic surfaces.19 Among the methods of assessment of platelet function, platelet aggregation performed in whole blood or in plasma was predictive for bleeding and transfusion requirements.12–14 Good correlation between the pre-operative values of ADP assessed with the Multiplate® device and the total blood transfusion during the intra-operative and the 24-h post-operative period was found in the study of Mengistu et al.14 In the study of Ray et al.,13 only patients with reduced post-operative platelet aggregation in LTA, and not in WBA, required significantly higher transfusion of blood products. No reports were made in any of these studies with regard to the type of transfused blood products.
Concerning post-operative bleeding, the results of our study confirm the findings of previous studies showing no correlation between decreased peri-operative platelet aggregometry and blood loss in the 24-h period after surgery.11–13 Because platelet dysfunction is among the main causes of early post-operative bleeding, but the therapy based on drainage volume is most often performed promptly and the platelet function usually recovers 12–24 after surgery,20 it may be more useful to assess the correlation between platelet function and bleeding during the early (3–4 h) post-operative period. For this time interval, the study of Ray showed pre-operative WBA and LTA, as well as post-operative LTA, to correlate with blood loss,13 a finding that could not be reproduced by Dietrich et al.11 in their study on the predictive value of WBA for blood loss. Other methods of assessment of platelet function, like lumi-aggregometry6 and PFA-100® (Dade Behring, Miami, FL), also showed limited or no predictive value for post-operative blood loss.8–10 PFA-100®, an in vitro simulation of the bleeding time,7 showed only limited predictive value for bleeding related to platelet dysfunction in cardiac surgery. This may be explained by the method being dependent both on the concentration and activity of plasmatic factors, such as the von Willebrand factor, and on platelet activity.8 This correlates with the findings of Slaughter et al.9 and Cammerer et al.,8 who evaluated the usefulness of PFA-100® in identifying patients unlikely to benefit from platelet transfusion.
Regarding the characteristics of the method applied in the present study, haematocrit and platelet count are among the factors known to influence platelet aggregometry.11,15,21 However, we found no significant differences in haematocrit values between the low and the high tertile of the aggregometric tests at any time point. Haematocrit may have had an effect on the aggregometric measurements but it could not be identified using this study design. On the other hand, a moderate correlation between platelet count and whole-blood aggregometry was shown at both time points, and particularly for the post-operative measurements, in agreement with other studies.11,13,15 At the same time, no correlation was found between platelet count and the requirements for PC transfusion, indicating that platelet count and platelet activity do not offer the same information with regard to haemostatic disturbances and transfusion requirements in this setting.
MEA performed on the Multiplate®device has several advantages that recommend it as a point-of-care method for platelet function assessment: it uses small amounts of diluted whole blood, standardized pipetting steps and single-use test cells. Most importantly, it supports the assessment of platelet function in the presence of other cellular components, like red blood cells, that directly promote platelet aggregation, and monocytes, that induce transcellular prostanoid formation.21,22 For the MEA analyses, the blood samples should be collected in tubes containing an anticoagulant with no or only a minor influence on the blood calcium concentration, i.e. with hirudin, Melagatran or heparin, but not with citrate, which interferes with platelet function by binding 98% of ionized calcium in the sample.23 Heparin was preferred as a blood anticoagulant for the present study, as the use of a non-standard blood collection tube may complicate the clinical application of impedance aggregometry. Although in vivo heparinization was shown to impair macroaggregation in WBA, the study of Belcher et al.24 offered evidence that ex vivo heparinization with low doses of heparin does not affect macroaggregation significantly.
The current study has several limitations. First, impedance aggregometry in single-use test cells has not been established as the standard method to detect changes in platelet activity. This platelet aggregation test does not reflect the complex interaction between thrombin, fibrinogen and platelets, and the end result of the coagulation process, the formation of the clot. Second, surgery-related factors may have influenced the activity of platelets and the resulting requirements for transfusion. All types of cardiac surgery were included in the study, in an attempt to characterize a platelet activity measurement applicable to many CPB patients, which led to high variations in the CPB and aortic cross-clamp time. A significant difference was observed between the low and the high tertile of the post-operative ADPtest with regard to CPB time and aortic cross-clamp time, but these parameters did not correlate with the post-operative PC transfusion. Low-dose aprotinin, known to have a protective effect on platelets,25 was administered in all patients, but its effect could not be isolated in the present study setting. Third, the study was designed as a pilot study, and so it lacks a sample size calculation. The absence of significant differences in transfusion between the tertiles of the TRAPtest after the Bonferroni–Holm correction may be a type II error. However, based on the present data, a prospective study may be designed to assess the predictivity of aggregometry tests for parameters of bleeding and transfusion.
In conclusion, our platelet aggregation data offer the possibility of an early estimation of the risk of transfusion. MEA performed at the beginning of a surgical intervention or upon arrival on ICU may improve the assessment of the necessity for PC transfusion. Whether the application of MEA may result in lower PC transfusion and lower operation costs remains to be studied.