![ORCID Icon]({"type":"image","src":"/templates/jsp/_style2/_tandf/images/orcid.svg"} "ORCID Icon"){kind=small}
http://orcid.org/0000-0002-1729-4063
Abstract
Abstract
Low platelet counts and hematocrit levels hinder whole blood point-of-care testing of platelet function. Thus far, no reference ranges for MEA (multiple electrode aggregometry) and PFA-100 (platelet function analyzer 100) devices exist for low ranges. Through dilution methods of volunteer whole blood, platelet function at low ranges of platelet count and hematocrit levels was assessed on MEA for four agonists and for PFA-100 in two cartridges. Using (multiple) regression analysis, 95% reference intervals were computed for these low ranges. Low platelet counts affected MEA in a positive correlation (all agonists showed r2 ≥ 0.75) and PFA-100 in an inverse correlation (closure times were prolonged with lower platelet counts). Lowered hematocrit did not affect MEA testing, except for arachidonic acid activation (ASPI), which showed a weak positive correlation (r2 = 0.14). Closure time on PFA-100 testing was inversely correlated with hematocrit for both cartridges. Regression analysis revealed different 95% reference intervals in comparison with originally established intervals for both MEA and PFA-100 in low platelet or hematocrit conditions. Multiple regression analysis of ASPI and both tests on the PFA-100 for combined low platelet and hematocrit conditions revealed that only PFA-100 testing should be adjusted for both thrombocytopenia and anemia. 95% reference intervals were calculated using multiple regression analysis. However, coefficients of determination of PFA-100 were poor, and some variance remained unexplained. Thus, in this pilot study using (multiple) regression analysis, we could establish reference intervals of platelet function in anemia and thrombocytopenia conditions on PFA-100 and in thrombocytopenia conditions on MEA.
Abbreviations
ADP, adenosine diphosphate; ASPI, arachidonic acid activation; AUC, area under the curve; COL, collagen; COL/ADP, collagen/adenosine diphosphate; COL/EPI, collagen/epinephrine; CT, closure time; DDAVP, desmopressin; HT, hematocrit; LTA, light transmission aggregometry; MEA, multiple electrode aggregometry; PFA-100, platelet function analyzer 100; PLT, platelet count; PPP, platelet poor plasma; PRP, platelet-rich plasma; TRAP, thrombin receptor activating peptide-6; vWF, von Willebrand factor; vWD, von Willebrand disease.
Introduction
Testing of platelet function comprises a variety of validated methods based on the use of platelet-rich plasma (PRP) or whole blood (point-of-care methods). Whole blood testing has the convenience of simple and fast operating procedures to complete a test and ease of interpreting obtained results. Multiple electrode aggregometry (MEA, Multiplate) is one of these tests in which whole blood is warmed and incubated for 3 min and after that tested for 6 more minutes. Multiple agonists can be used to start platelet activation and aggregation after the incubation phase. Platelet function is based on aggregation of platelets on two rods immerged in the incubated whole blood. This aggregation changes the impedance between these two rods, and this change in impedance is plotted as a graph. In the measuring cuvette, two pairs of rods are available, thus having an internal control with each measurement. PFA-100 (platelet function analyzer 100) testing is different in comparison with MEA and other methods [1]. The PFA-100 assesses platelet function while applying shear stress. Whole blood is added to a cuvette containing a small opening lined with collagen and either epinephrine or adenosine diphosphate (ADP); the blood flows under high shear stress (5000–6000 s−1) through the opening, which will close when sufficient platelets have aggregated; the time till occlusion (and thereby termination of blood flow) is defined as the closure time (CT).
One pitfall known in whole blood platelet function testing using the MEA or PFA-100 method is that in order to obtain reliable results, a sufficient amount of platelets is needed [2–8]. Platelet function testing is unreliable in low platelet ranges where thrombocyte transfusion is not yet required (20–100 × 109/L platelet range) [2–8]. Another pitfall is the presence of low hematocrit levels, especially in PFA-100 testing [9,10]. A low hematocrit will prolong the closure time, and a hematocrit of at least 0.25 L/L is needed for reliable results [9,11]. However, for instance, during cardiothoracic surgery, when platelet function testing may be warranted in case of bleeding, often lower hematocrit levels are present (a hemoglobin of 7 g/dL being a global transfusion trigger of red cells, equals to roughly a hematocrit of 0.20–0.21 L/L) [12], which hinders reliable platelet function testing [9,11].
When using light transmission aggregometry (LTA), the current gold standard in platelet function testing using PRP, platelet levels are not to be adjusted before testing [13]. In the past, Hayward and colleagues [14] have established reference intervals for thrombocytopenic patients on LTA using regression analysis for various agonists. Thus by defining reference intervals, LTA became suitable for measuring platelet function in a wide range of platelet counts. And because LTA is a plasma-based analysis, hematocrit is irrelevant in this in vitro test.
For MEA and PFA-100 testing, research has been done in the past, hereby verifying that both these tests rely on adequate platelet counts and hematocrit levels [2,9]. However, so far, no reference intervals have been established for either measuring device, which could guide users when encountering subjects with low platelet counts and/or hematocrit levels. Ideally, both devices would give the user reference intervals depending on these pre-analytical variables.
Our goal of this pilot study is to establish such reference intervals, similar to what has been done statistically for LTA [14], for each whole blood platelet function testing method by dilution of blood of healthy volunteers.
Methods
Blood sampling
Blood of healthy volunteers was used to obtain baseline and subsequent diluted whole blood samples for PFA-100 (Dade-Behring, Liederbach, Germany) and MEA (Dynabyte Medical, Munich, Germany) testing. Volunteers had to be at least 18 years of age. Volunteers were excluded if any medication, which could interfere with thrombocyte function, was used, or when they suffered from a bleeding diathesis. All blood donors volunteered and gave their consent for participation in this study. Twenty healthy volunteers were selected for each condition (PFA and low hematocrit, PFA and low platelets, MEA and low hematocrit, and MEA and low platelets). This was done because of budgetary reasons, time constraints, and manpower. Subsequently, a patient database search using all our MEA or PFA results was conducted in order to serve as positive control. Patients who had hematocrit levels below 0.25L/L and/or platelet counts below 150 × 109/L were selected. This study was carried out according to The Code of Ethics of the World Medical Association (Declaration of Helsinki). Our local institutional review board (azM/UM) has approved this study.
The basic principle was to use the volunteers’ own blood to make dilutions of different hematocrit and thrombocyte levels, a technique similar to what Harrison et al performed in 1999 [15]. Blood was drawn via a clean venipuncture in 3-ml hirudin tubes (Verum Diagnostica, Munich, Germany) for MEA testing and in 4.5-ml 3.2% (w/v) citrate tubes (Becton Dickinson BV, Breda, the Netherlands) for PFA-100 testing. A 5-ml EDTA tube (Becton Dickinson BV, Breda, the Netherlands) was collected first and used to calculate baseline hematocrit and thrombocyte counts on a Sysmex XE-5000 (Sysmex Nederland B.V., Etten-Leur, The Netherlands).
Blood preparation
For testing the influence of thrombocyte counts, the following was done. One tube was tested without any centrifuging steps; four tubes were spun down for 10 min at 170 g to acquire platelet-rich plasma (PRP) on top of the buffy coat and red cell layer. Four more tubes were spun down for 5 min at 2 500 g to acquire platelet poor plasma (PPP) on top of the platelets, buffy coat, and red cell layer. A second centrifugation step was done at 10 000 g with the PPP for 10 min to acquire the final PPP solution. By substituting different amounts of PRP with PPP, thrombocyte counts were reduced while retaining original hematocrit levels. After substitution, tubes were resuspended to obtain “whole blood.” We aimed for thrombocyte levels of 150, 100, 50, and 20 × 109/L when diluting the samples. Of all samples, exact thrombocyte counts and hematocrits levels were measured. These were corrected for the amount of citrate (if applicable).
For determining the influence of hematocrit levels on platelet function, a similar but slightly different approach was used as follows: instead of substituting PRP with PPP, these time equal volumes of the red cell layer were substituted with PPP, thus lowering the hematocrit levels while retaining thrombocyte levels. Hematocrit levels we aimed for were 0.35, 0.30, 0.25, and 0.20 L/L by diluting the samples. Again thrombocyte counts and hematocrit levels were measured and corrected for dilution by citrate in the citrate containing tubes.
Platelet function analysis
After these steps, closure time (CT) was analyzed on PFA-100 using the epinephrine and ADP cartridge and we measured the area under the curve (AUC expressed in U; 1 U equals 10 AU × min) on MEA using the TRAP (thrombin receptor activating peptide-6, final concentration (f.c.) 32 μmol/L), COL (collagen, f.c. 3.2 μg/mL), ADP (f.c. 6.5 μmol/L), and ASPI (arachidonic acid, f.c. 0.5 mmol/L) reagents using standard protocols as provided by the manufacturer. Results were compared to previously in house defined upper and lower limit reference intervals as advised by the device manufacturers.
As a control, the effect of the centrifugation steps on thrombocyte function was assessed by first testing a whole blood sample on both machines using all cartridges and reagents, then centrifuging the sample, and then resuspending and testing the centrifuged sample once more.
Statistics
The results before and after centrifugation were compared using a paired Student’s t-test. A p-value of <0.05 was considered to be significant. A (multiple) regression analysis was performed using (non-)linear curve fitting and in order to calculate coefficients of determination and predicted 95% reference intervals. A best fit was used in determining a linear or non-linear curve. Scatterplots of the (log transformed) standardized predicted values were used to see whether assumptions of homogeneity of variance and linearity were met. All statistics and regression analyses were done using SPSS Statistics v23 (IBM Corp, Armonk, NY, USA).
All results were plotted in a graph using GraphPad Prism (GraphPad Prism version 5.0a for Windows, GraphPad Software, San Diego, CA, USA). Using a (non-)linear fit, the 95% prediction intervals were plotted as well using GraphPad Prism for each graph. Three-dimensional plots and contour plots were made online using WolframAlpha® (Wolfram Alpha LLC. 2009. Wolfram|Alpha (accessed on 17 March 2016). Available on http://www.wolframalpha.com).
Results
The effect of centrifugation on MEA and PFA-100 testing
An alteration in platelet function due to centrifugation could not be found for ADP (p = 0.30), COL (p = 0.13), and ASPI (p = 0.74) for MEA testing (n = 20). For the TRAP reagent, there was a statistically significant difference (p = 0.02), with a mean lower AUC after centrifugation of 7.4 U (Figure 1). For the PFA-100, no alteration in platelet functionality due to centrifugation could be found for both cartridges (COL/EPI p = 0.29, COL/ADP p = 0.57).
Published online:
09 January 2017Figure 1. The effect of centrifugation on MEA and PFA-100 testing. The effect of centrifugation is shown here. Only in TRAP, a significant drop of 7.4 U was observed after centrifugation (p = 0.02). This was marked with an asterisk. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0001_b.gif"})
Figure 1. The effect of centrifugation on MEA and PFA-100 testing. The effect of centrifugation is shown here. Only in TRAP, a significant drop of 7.4 U was observed after centrifugation (p = 0.02). This was marked with an asterisk. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
Influence of platelet counts on MEA and PFA-100
Scatterplots of 20 subjects were made for each reagent separately with platelet counts on the X-axis and either the AUC or CT on the Y-axis (Figure 2). The initial results without dilution are shown in blue squares. For each individual, four dilutions were made (squares in orange), bringing the total to 100 data points. In our patient database, 14 patients were found with platelet counts below 150 × 109/L for the MEA and 13 for the PFA-100. These are indicated in black dots.
Published online:
09 January 2017Figure 2. Influence of platelet count on MEA and PFA-100 testing. Blue squares are the original results of the healthy volunteers without dilution. The orange dots represent separate data points of the dilution steps. Black dots are patients from our MEA and PFA-100 database with platelet counts below 150 × 109/L. Solid orange lines define the best fitting curve from regression analysis and the dotted orange lines its 95% reference intervals. Gray areas are according to the manufacturers advise in house defined upper and lower limits of normality. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; L, liter; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; PLT, platelet count; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0002_oc.jpg"})
Figure 2. Influence of platelet count on MEA and PFA-100 testing. Blue squares are the original results of the healthy volunteers without dilution. The orange dots represent separate data points of the dilution steps. Black dots are patients from our MEA and PFA-100 database with platelet counts below 150 × 109/L. Solid orange lines define the best fitting curve from regression analysis and the dotted orange lines its 95% reference intervals. Gray areas are according to the manufacturers advise in house defined upper and lower limits of normality. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; L, liter; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; PLT, platelet count; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
For the TRAP (r2 = 0.75) and COL (r2 = 0.76) reagents, best fit was acquired non-linear. For the ASPI (r2 = 0.80) and ADP (r2 = 0.77) reagents, a linear fit was best. On the PFA-100, both reagents (COL/ADP r2 = 0.47 and COL/EPI r2 = 0.37) were best fitted non-linearly. Standardized residual plots (after log transformation) showed that assumptions of normality, homoscedasticity, and linearity were met.
As can be observed from the MEA graphs, a positive correlation is apparent between platelet count and AUC, while an inverse correlation is present for the CT on the PFA-100.
Previously in house calculated 95% reference ranges (gray areas), determined as advised by the manufacturer of each device, are also shown in the graphs. For PFA-100, platelet counts lower than approximately 90 × 109/L result in prolonged CT for both cartridges (the upper boundary of the manufacturer’s recommended ranges is crossed), while for the MEA, the lower limit of platelet count is around 175 × 109/L for ADP, ASPI, and TRAP (here, the lower limit of the manufacturer’s recommended ranges is crossed). For COL, the lower limit is around 50 × 109/L.
For each of the reagents, a formula was calculated to compute the AUC for MEA testing or CT for PFA-100 analysis using regression analysis. Reference intervals (95% prediction bands) were also computed and are given underneath each formula. 
Influence of hematocrit on MEA and PFA-100
Again scatterplots of 20 subjects were made for each reagent separately with hematocrit on the X-axis and either the AUC or CT on the Y-axis (blue squares in Figure 3). The diluted samples are shown in orange’s squares bringing the total once more to 100. The black dots represent the patients from our database with a hematocrit below 0.25L/L. We found three patients with PFA-100 results (one of these was not tested with the COL/EPI cartridge) and three with MEA results.
Published online:
09 January 2017Figure 3. Influence of hematocrit levels on MEA and PFA-100 testing. Blue squares are the original results of the healthy volunteers without dilution. The orange dots represent separate data points of the dilution steps. Black dots are patients from our MEA and PFA-100 database with hematocrit below 0.25 L/L. Solid orange lines define the best fitting curve from regression analysis and the dotted orange lines its 95% reference intervals. Gray areas are according to the manufacturers advise in house defined upper and lower limits of normality. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; HT, hematocrit; L, liter; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0003_oc.jpg"})
Figure 3. Influence of hematocrit levels on MEA and PFA-100 testing. Blue squares are the original results of the healthy volunteers without dilution. The orange dots represent separate data points of the dilution steps. Black dots are patients from our MEA and PFA-100 database with hematocrit below 0.25 L/L. Solid orange lines define the best fitting curve from regression analysis and the dotted orange lines its 95% reference intervals. Gray areas are according to the manufacturers advise in house defined upper and lower limits of normality. Abbreviations: ADP, adenosine diphosphate; ASPI, arachidonic acid; AUC, area under the curve; COL, collagen; CT, closure time; HT, hematocrit; L, liter; MEA, multiple electrode aggregometry; PFA-100 COL/ADP, platelet function analyzer 100 collagen/adenosine diphosphate cartridge; PFA-100 COL/EPI, platelet function analyzer 100 collagen/epinephrine cartridge; TRAP, thrombin receptor activating peptide-6; U, 10 arbitrary units × min.
For the TRAP, ADP, and COL reagent, no fit could be made, as r2 was ≤0.01 in all cases. For the ASPI, a weak non-linear fit could be made with r2 = 0.14. On the PFA-100, both reagents (COL/ADP r2 = 0.37 and COL/EPI r2 = 0.31) were best fitted non-linearly. Again, (log transformed) standardized residual plots showed that the assumptions of homogeneity of variance, normality, and linearity were met.
As can be observed from the MEA graphs, a positive correlation is apparent between hematocrit and AUC for ASPI, while an inverse correlation is present once more for the CT on the PFA-100.
Previously in house calculated 95% reference ranges (grey areas), determined as advised by the manufacturer of each device, are also shown in the graphs. For PFA-100, hematocrit around 0.22 L/L and below results in prolonged CT for both cartridges (by crossing the upper limit of the manufacturer’s recommended ranges), while for the MEA, the lower limit of hematocrit is approximately 0.30 L/L for ASPI (at this HT values drop below the manufacturer’s recommended ranges).
Calculated formulas using regression analysis for each reagent are the following. Reference intervals (95% prediction bands) are again computed and given underneath each formula.
Multiple regression analysis of hematocrit and platelet count on MEA and PFA-100
For interpretation of the AUC of the ASPI test and CT of both the PFA-100 tests, we corrected the results for platelet counts as well as hematocrit levels. This resulted in the following standard equation for MEA: 
For the PFA-100 the following equation was used: 
Adjusted r2 values of each model were 0.49 for ASPI, 0.30 for COL/ADP, and 0.25 for COL/EPI.
This results in the following final equations for each of the three tests (reference intervals are below each formula): 
Three-dimensional plots and contour plots of the three equations above were made online using WolframAlpha® (Figures 4, 5, and 6). On the X-axis, platelet count is shown, while on the Y-axis, hematocrit levels are used. The resultant, AUC in U for ASPI testing on the MEA or CT in seconds for COL/ADP and COL/EPI on the PFA-100, can be observed with a color gradient and the height on the Z-axis. These 3D and contour plots show that a combination of low platelet count together with low hematocrit results in either a low AUC for MEA (Figure 4) or a prolonged CT for PFA-100 (Figures 5 and 6). A high platelet count combined with high hematocrit results in either a high AUC for MEA (Figure 4) or a shortened CT for PFA-100 (Figures 5 and 6).
Published online:
09 January 2017Figure 4. Influence of platelet count and hematocrit levels on MEA (ASPI) testing. 3D-graph for ASPI (4a) and a top-down view (4b). The darker the color, the lower the AUC on MEA. x = PLT (×109/L), y = HT (L/L), z = AUC (U).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0004_oc.jpg"})
Figure 4. Influence of platelet count and hematocrit levels on MEA (ASPI) testing. 3D-graph for ASPI (4a) and a top-down view (4b). The darker the color, the lower the AUC on MEA. x = PLT (×109/L), y = HT (L/L), z = AUC (U).
Published online:
09 January 2017Figure 5. Influence of platelet count and hematocrit levels on PFA-100 (COL/ADP) testing. 3D-graph for COL/ADP (5a) and a top-down view (5b). The darker the color, the shorter the CT on PFA-100. x = PLT (×109/L), y = HT (L/L), z = CT (s).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0005_oc.jpg"})
Figure 5. Influence of platelet count and hematocrit levels on PFA-100 (COL/ADP) testing. 3D-graph for COL/ADP (5a) and a top-down view (5b). The darker the color, the shorter the CT on PFA-100. x = PLT (×109/L), y = HT (L/L), z = CT (s).
Discussion
In this pilot study, we defined reference ranges for platelet function testing on MEA and PFA-100 at low platelet counts and hematocrit levels using regression analysis and defined formulas for these reference intervals.
In contrast to previous research, we tried to approach the clinical setting as much as possible, aiming for a condition of “pure” thrombocytopenia and anemia, of variable degree, respectively. Stissing et al researched the effects of low platelet counts on MEA by diluting whole blood samples with PRP and PPP, hereby effectively also lowering hematocrit levels [5]. Furthermore, they made grouping variables for the platelet counts and did not show standard deviations of their actual platelet counts for each group. Because of this, one cannot judge individual results of platelet counts in comparison with the AUC. Hanke and colleagues did similar experiments in MEA testing using platelet adjusted whole blood while retaining original hematocrit. Again they showed non-continuous data and used categories of low platelet count instead [8]. Another research group also assessed the relation between platelet counts on MEA results by dilution with a crystalloid solution instead of autologous components. Their work, unfortunately, had to be retracted [16]. What these previous authors evaluated for MEA, Harrison et al did for the PFA-100 [15]. Blood of healthy donors was used to assess the influence of thrombocytopenia and anemia. They came to similar results as we did, but again could not give reference intervals. Besides that, they used blood of only three volunteers. Prolongation of CT by lowering platelet counts or hematocrit in vitro, shown by work of Kundu et al. [11], was consistent to our study and an in vivo thrombocytopenia study by Carcao and colleagues proved the same relation between platelet count and the CT [17]. Cho et al, however, stated that they could not find a relation between platelet counts and hematocrit and the CT in their reference interval finding study among healthy volunteers [18].
In our search for reference intervals, we first assessed if our method of lowering platelet counts and hematocrit levels by centrifugation and autologous blood mixing itself would affect our measurement on both devices. This could not be found except for the TRAP reagent on the MEA. This 7.4 U diminishment due to centrifugation seems irrelevant when comparing this to the reference range of 94–156 U. Nevertheless, all our TRAP results should be interpreted as having a slightly false-low value when a sample was centrifuged.
We found that the AUC in MEA testing changes in a linear or a log-linear fashion (depending on the reagent) with lowering platelet counts. The coefficients of determination for all four tests were good as all showed an r2 ≥ 0.75. We think that this non-linear relation could be explained by a ceiling effect of the agonist. TRAP as a strong activator of thrombocytes might induce a maximum activation of thrombocytes, hereby completely covering both rods in MEA. ADP is a much weaker activator of platelets, which might not be maximal in low to normal ranges, but eventually in high platelet ranges, it could be maxed out. Reference intervals, according to the manufacturer’s recommendations, also point us in this direction. The upper limit of normality is lower in ADP (122 U) than it is for TRAP (156 U). For the PFA-100, thrombocyte counts have an inverse log-linear relationship with the CT for both cartridges. The coefficient of determination was acceptable for COL/ADP, but with an r2 < 0.40, it was poor for COL/EPI. When looking at hematocrit levels, variation in hematocrit does not seem to be of relevance for the AUC in MEA testing, except for the ASPI test. Though the coefficient of determination was really low for ASPI, the CT on the PFA-100 fluctuates heavily with an inverse log-linear relation at different hematocrit levels.
The combined models all showed lower coefficients of determination, than when a single regression analysis was done. For MEA, this changed the r2 from 0.80 for platelet counts and 0.14 for hematocrit level to 0.49 when combined. As a result, if measurements are to be done on the MEA, one should only correct them for platelet counts. Hematocrit does not seem to have a big influence on the AUC either alone or in conjunction with platelet count when testing platelet function on MEA. This is also apparent on the top-down view of the 3D graph for the ASPI (Figure 4b): The isolines are hardly affected by differences in hematocrit level as they are plotted almost parallel to the hematocrit axis. When viewing the top-down views of both the 3D-graphs for the PFA-100 (Figures 5b and 6b), one can appreciate that hematocrit does influence the CT together with platelet count. Thus, for the PFA-100, one should adjust the results accordingly to these pre-analytical variables.
Published online:
09 January 2017Figure 6. Influence of platelet count and hematocrit levels on PFA-100 (COL/EPI) testing. 3D-graph for COL/EPI (6a) and a top-down view (6b). The darker the color, the shorter the CT on PFA-100. x = PLT (×109/L), y = HT (L/L), z = CT (s).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/iplt20/2017/iplt20.v028.i07/09537104.2016.1257782/20171030/images/medium/iplt_a_1257782_f0006_oc.jpg"})
Figure 6. Influence of platelet count and hematocrit levels on PFA-100 (COL/EPI) testing. 3D-graph for COL/EPI (6a) and a top-down view (6b). The darker the color, the shorter the CT on PFA-100. x = PLT (×109/L), y = HT (L/L), z = CT (s).
Using our MEA and PFA-100 database, we found several patients that met the criteria of low hematocrit and thrombocyte counts (black dots in Figures 2 and 3). The AUC of most of the patients with low thrombocyte counts was below the gray standard reference ranges, but some of the results were within our newly defined reference intervals. This would mean a normal platelet function in these cases. On the other hand, some patients are below both reference areas. These patients, besides being thrombocytopenic, might have platelet defects due to the nature of their disease or due to antiplatelet drugs. This could not be resolved from the data we had in our database. Similar results were found for the PFA-100 in the thrombocytopenic patients. In the patients with anemia (hematocrit below 0.25 L/L), this was not so evidently. This is probably due to the fact that only three patients could be selected based on their hematocrit.
For both multiple regression models, however, there is still a lot of variance to be explained. It is known from literature that results on the PFA-100 are dependent on von Willebrand factor (vWF) [4,9]. In a study by Van Vliet et al. [19], type-1 von Willebrand disease (vWD) patients were treated with DDAVP (desmopressin) or Haemate-P (a factor VIII/vWF concentrate) and the effects were measured with the PFA-100 COL/EPI cartridge. This study showed a marked effect of changes in vWF which were assessed by relatively high coefficients of determination for vWF:CB (vWF:collagen binding), vWF:Ag (vWF:antigen), vWF:RCo (vWF:ristocetin cofactor), and factor VIII (all above 0.5 except for vWF:Ag in the Haemate-P pre- and post-treatment group (r2 = 0.42)). Ideally, we would like to have measured vWF activity or antigen in our diluted samples to see its effect. This effect of vWF could explain why in PFA-100 testing we have a non-linear regression. Due to a synergistic effect of vWF adhesion to platelets and vice versa, they enhance each other in an exponential fashion. Unfortunately, we do not have plasma remaining and correction of the PFA-100 for vWF activity or antigen would nullify the use of the PFA-100 as a screening tool for vWD.
One drawback of this study is that because we lowered the platelet count and the hematocrit level separately, we did not mimic the situation of a combined thrombocytopenia and anemia condition. Therefor contrast is lost in this combined thrombocytopenia and anemia range. However, it is very unlikely that a low platelet count in combination with a low hematocrit level would promote improved platelet function.
In this pilot study, we have shown that MEA testing of platelet function is dependent on platelet count and that PFA-100 testing of platelet function is dependent on both platelet count as well as hematocrit levels. For both testing methods, we were able to define variable reference intervals using (multiple) regression analysis and proposed formulas to guide users in interpreting MEA and PFA-100 results in case of thrombocytopenic and anemic patients. These data, however, need further validation in a new set of healthy volunteers and/or in more patients with anemia or thrombocytopenia.
Declaration of interest
The authors report no declarations of interest.
References
- Berkowitz SD, Frelinger 3rd AL, Hillman RS. Progress in point-of-care laboratory testing for assessing platelet function. Am Heart J 1998;136(4Pt 2 Su):S51–65. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Bochsen L, Johansson PI, Kristensen AT, Daugaard G, Ostrowski SR. The influence of platelets, plasma and red blood cells on functional haemostatic assays. Blood Coagul Fibrinolysis 2011;22(3):167–175. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Rubak, P, Villadsen K, Hvas AM. Reference intervals for platelet aggregation assessed by multiple electrode platelet aggregometry. Thromb Res 2012;130(3):420–423. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Seyfert UT, Haubelt H, Vogt A, Hellstern P. Variables influencing Multiplate(TM) whole blood impedance platelet aggregometry and turbidimetric platelet aggregation in healthy individuals. Platelets 2007;18(3):199–206. [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
- Stissing T, Dridi NP, Ostrowski SR, Bochsen L, Johansson PI. The influence of low platelet count on whole blood aggregometry assessed by Multiplate. Clin Appl Thromb Hemost 2011;17(6):E211–E217. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Toth O, Calatzis A, Penz S, Losonczy H, Siess W. Multiple electrode aggregometry: a new device to measure platelet aggregation in whole blood. Thromb Haemost 2006;96(6):781–788. [PubMed], [Web of Science ®], [Google Scholar]
- Wurtz M, Hvas AM, Kristensen SD, Grove EL. Platelet aggregation is dependent on platelet count in patients with coronary artery disease. Thromb Res 2012;129(1):56–61. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Hanke AA, Roberg K, Monaca E, Sellmann T, Weber CF, Rahe-Meyer N, Görlinger K. Impact of platelet count on results obtained from multiple electrode platelet aggregometry (Multiplate). Eur J Med Res 2010;15(5):214–219. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Favaloro EJ. Clinical application of the PFA-100. Curr Opin Hematol 2002;9(5):407–415. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Hayward CP, Harrison P, Cattaneo M, Ortel TL, Rao AK; Platelet Physiology Subcommittee of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis. Platelet function analyzer (PFA)-100 closure time in the evaluation of platelet disorders and platelet function. J Thromb Haemost 2006;4(2):312–319. [Google Scholar]
- Kundu SK, Heilmann EJ, Sio R, Garcia C, Davidson RM, Ostgaard RA. Characterization of an In Vitro Platelet Function Analyzer, PFA-100™. Clin Appl Thromb/Hemost 1996;2(4):241–249. [Crossref], [Web of Science ®], [Google Scholar]
- Likosky DS, FitzGerald DC, Groom RC, Jones DK, Baker RA, Shann KG, Mazer CD, Spiess BD, Body SC. The effect of the perioperative blood transfusion and blood conservation in cardiac surgery Clinical Practice Guidelines of the Society of Thoracic Surgeons and the Society of Cardiovascular Anesthesiologists upon clinical practices. J Extra Corpor Technol 2010;42(2):114–121. [PubMed], [Google Scholar]
- Cattaneo M, Cerletti C, Harrison P, Hayward CP, Kenny D, Nugent D, Nurden P, Rao AK, Schmaier AH, Watson SP, et al. Recommendations for the standardization of light transmission aggregometry: a consensus of the working party from the platelet physiology subcommittee of SSC/ISTH. J Thromb Haemost 2013. [Crossref], [Web of Science ®], [Google Scholar]
- Hayward CP, Moffat KA, Pai M, Liu Y, Seecharan J, McKay H, Webert KE, Cook RJ, Heddle NM. An evaluation of methods for determining reference intervals for light transmission platelet aggregation tests on samples with normal or reduced platelet counts. Thromb Haemost 2008;100(1):134–145. [PubMed], [Web of Science ®], [Google Scholar]
- Harrison P, Robinson MS, Mackie IJ, Joseph J, McDonald SJ, Liesner R, Savidge GF, Pasi J, Machin SJ. Performance of the platelet function analyser PFA-100 in testing abnormalities of primary haemostasis. Blood Coagul Fibrinolysis 1999;10(1):25–31. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Retraction: Whole-blood aggregometry: are there any limits with regard to platelet counts? Acta Anaesthesiol Scand 2011;55(7):903. [Web of Science ®], [Google Scholar]
- Carcao MD, Blanchette VS, Stephens D, He L, Wakefield CD, Butchart S, Christie DJ, Rand ML. Assessment of thrombocytopenic disorders using the Platelet Function Analyzer (PFA-100). Br J Haematol 2002;117(4):961–964. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]
- Cho YU, Jang S, Park CJ, Chi HS. Variables that affect platelet function analyzer-100 (PFA-100) closure times and establishment of reference intervals in Korean adults. Ann Clin Lab Sci 2008;38(3):247–253. [PubMed], [Web of Science ®], [Google Scholar]
- van Vliet HH, Kappers-Klunne MC, Leebeek FW, Michiels JJ. PFA-100 monitoring of von Willebrand factor (VWF) responses to desmopressin (DDAVP) and factor VIII/VWF concentrate substitution in von Willebrand disease type 1 and 2. Thromb Haemost 2008;100(3):462–468. [Crossref], [PubMed], [Web of Science ®], [Google Scholar]