Evaluation of a whole blood remote platelet function test for the diagnosis of mild bleeding disorders

Abstract

Background

Mild platelet function disorders (PFDs) are complex and difficult to diagnose. The current gold standard test, light transmission aggregometry (LTA), including lumi-aggregometry, is time- and labour-intensive and blood samples must be processed within a limited time after venepuncture. Furthermore, many subjects with suspected PFDs do not show a platelet abnormality on LTA.

Objective

To assess the diagnostic potential of an easy-to-use remote platelet function test (RPFT) as a diagnostic pre-test for suspected PFDs.

Methods

RPFT was compared to lumi-aggregometry in participants recruited to the Genotyping and Phenotyping of Platelets study (GAPP, ISRCTN 77951167). For RPFT, whole blood was stimulated with platelet agonists, stabilized with PAMFix and returned to the central laboratory for analysis of P-selectin and CD63 by flow cytometry.

Results

In the 61 study participants (42 index cases and 19 relatives) there was a good agreement between lumi-aggregometry and RPFT with diagnosis being concordant in 84% of cases (kappa=0.668, p<0.0001). According to both tests, 29 participants were identified to have a deficiency in platelet function and 22 participants appeared normal. There were 4 participants where lumi-aggregometry revealed a defect but RPFT did not, and 6 participants where RPFT detected an abnormal platelet response that was not identified by lumi-aggregometry.

Conclusion

This study suggests that RPFT could be an easy-to-use pre-test to select, which participants with bleeding disorders would benefit from extensive platelet phenotyping. Further development and evaluation of the test are warranted in a wider population of patients with excessive bleeding and could provide informative screening tests for PFDs.

Introduction

Mild platelet function disorders (PFDs) are complex, poorly understood, difficult to diagnose and require extensive laboratory testing limited to specialised laboratories [1,2]. Lumi-aggregometry, where dense granule secretion is assessed in parallel with traditional LTA [3-5] is informative, but is time- and labour-intensive. It is performed on platelet-rich plasma and requires a considerable volume of a fresh blood sample. In our previous study we demonstrated using lumi-aggregometry that approximately 60% of participants with excessive clinical bleeding and a suspected underlying inherited PFD have a demonstrable abnormality in platelet responses [5]. This suggests that 40% of patients may have a platelet abnormality that is not detectable by lumi-aggregometry or may have a defect in other components of the haemostatic pathway or a combination of mild defects, which leads to a bleeding phenotype. These findings are in agreement with an earlier study by Quiroga et al where approximately 60% of a slightly different patient population with mild bleeding had no identified defect when assessed by LTA and serotonin and ATP release measurements [6]. Accordingly, a simple and easy-to-perform preliminary test might be useful to select patients with excessive bleeding for whom extensive platelet phenotyping using traditional methodologies may be required for further characterisation. At present, there is no reliable easy-to-use and sufficiently sensitive global platelet function test available for use as a screening tool in subjects with suspected PFD [7].

We have developed a remote platelet function test (RPFT), in which a blood sample is manipulated in a simple way that does not require specialised staff or equipment at the test site. The sample is stabilised utilising a fixing solution, PAMFix, and sent to a flow cytometry fascility for analysis [8], where the expression of the α-granule marker P-selectin and the dense granule marker CD63 is measured. The samples treated with PAMFix are stable for up to 9 days at ambient temperature. CD63 expression specifically evaluates dense granule secretion; analysis of P-selectin expression is used as a marker of α-granule secretion and also as a general indicator of platelet reactivity. Here we describe the evaluation of this new approach against traditional lumi-aggregometry in patients with suspected inherited PFDs.

Methods

Study population

To assess the diagnostic potential of the RPFT in the diagnosis of PFDs, results were compared to lumi-aggregometry in participants recruited to the Genotyping and Phenotyping of Platelets study (GAPP, ISRCTN 77951167) from April 2012 to March 2013. We studied 61 participants (42 index cases and 19 recruited relatives), 13 males (median age 13, range 5-45) and 48 females (median age 39, range 2-77); all had a history of excessive bleeding with a suspected underlying inherited PFD. These participants had normal coagulation parameters and a normal platelet count and were referred from UK Comprehensive Care Haemophilia Centres. Patients with an existing diagnosis of a known platelet defect were not included in this study. Bleeding symptoms in adult patients (n=42) were assessed using the ISTH/SSC bleeding assessment tool (BAT) [9]. All the participants presented with a significant bleeding history with a median bleeding score of 13 (interquartile range: 8 – 16), whereas the 95^th percentile within the healthy volunteer population was 4 [10].

Platelet function was also assessed in 41 healthy volunteers recruited as controls, who had no history of bleeding and had refrained from taking any medication known to affect platelet function. All blood samples were either taken at or delivered by courier from several referring centres (maximum distance 100 miles) to the central laboratory as previously described [5]. This study was approved by the National Research Ethics Service Committee West Midlands – Edgbaston (ref 06/MRE07/36). All participants provided written informed consent.

Assessment of platelet function

For the RPFT measurements, whole blood samples collected into sodium citrate were prewarmed and stimulated at 37°C for 5 min without mixing. They were then fixed using PAMFix (Platelet Solutions Ltd, Nottingham). Samples were then transported from the central laboratory to the flow cytometry facility (60 miles) and routinely analysed within 3 days. For platelet stimulation blood samples were treated with 1) saline alone to provide a baseline test; 2) adenosine diphosphate (ADP) with U46619, a stable thromboxane A₂ analogue (ADP/U4 10μM and 1μM respectively); 3) thrombin receptor activating peptide (TRAP [SFLLRN] 20μM) and 4) arachidonic acid with epinephrine (AA/EPI, 0.5mM and 100μM respectively). U46619 and epinephrine were used as potentiating agents and were chosen based on previous studies where we investigated optimal conditions for platelet activation in whole blood [11]. Intra- and inter-assay coefficients of variation are 4.4% and 5.6% respectively.

Lumi-aggregometry was performed as previously described [5]. The results were considered abnormal when they fell below the 5^th percentile with reference to a bank of local healthy volunteers (n=68).

Statistical analysis

Results of platelet function measurements are presented as individual data in median fluorescence units (mf). To determine the ability of the RPFT to discriminate between subjects with and without abnormal platelet responses and to set optimal cut-off levels, receiver operator characteristic (ROC) curve analysis was performed. For P-selectin measurements ROC curves were generated using the subjects with and without any defect, as determined by lumi-aggregometry, and for CD63 measurements, subjects with and without a secretion defect. Agreement between the two tests was assessed using two-by-two analysis and Cohen’s kappa statistic. Analyses were performed using the software packages SPSS 14.0 and GraphPad Prism 6.

Results

According to lumi-aggregometry, 33 of the 61 participants recruited to GAPP study were considered to have an abnormality in platelet responses. The majority of these 33 participants could be subdivided into three categories on the basis of their lumiaggregation profile [5] – defects in dense granule secretion (n=9), G_i signalling (n=14) and the TxA₂ pathway (n=3). The remaining 7 participants had either a complex defect that was difficult to assign to any specific group, presenting with reduced responses via several pathways (n=4), or a defect in the response to collagen (n=1) or TRAP (n=2).

The area under the ROC curve for P-selectin (Figure 1a) and CD63 measurements (Figure 1b) with different agonists varied between 0.72 and 0.99 (p<0.004). This confirmed that RPFT can discriminate between individuals with excessive bleeding with and without abnormal platelet responses as detected by lumi-aggregometry. Based on the ROC analysis, appropriate cut-off levels for each parameter were chosen to provide the highest possible sensitivity of the RPFT to minimise false negative results. Applying the chosen cut-offs in the group of healthy controls indicated that 5 out of 41 participants presented with somewhat reduced platelet responses on RPFT (Figure 2a). This small degree of overlap in platelet responsiveness between populations of patients and healthy controls is also seen in other platelet function tests in addition to other tests of haemostasis [6].

Figure 1. Receiver operating characteristic curves for P-selectin (a) and CD63 (b) expression measurement with RPFT.

Receiver operating characteristic curves demonstrating the relationship between defects in platelet function as determined by lumi-aggregometry and a) P-selectin and b) CD63 expression measurements with three conditions of platelet stimulation.

Figure 2. Performance of the RPFT in healthy subjects and in participants with platelet function defects as determined by lumi-aggregometry.

a) P-selectin and CD63 measured with RPFT in healthy volunteers at baseline and after stimulation. For P-selectin cut-offs were 899 (ADP/U4), 991 (TRAP) and 1006 (AA/EPI), and for CD63 – 103 (ADP/U4), 120 (TRAP) and 112,5 (AA/EPI), expressed as median fluorescence units; b) dense granule secretion defect (the two highlighted participants are discussed in the text); c) defect in the TxA₂ pathway and d) defect in G_i signalling. Horizontal lines indicate cut-offs determined for each parameter using ROC curve analysis.

In the patient group, the overall agreement between lumi-aggregometry and RPFT was good, with the diagnosis being concordant in 84% of cases (kappa=0.668, p<0.0001) (Table). All participants who presented with a dense granule secretion defect (Figure 2b) or a defect in the TxA₂ pathway (Figure 2c) on lumi-aggregometry were also classified as having a platelet abnormality on the RPFT with at least one parameter falling below the normal range. Out of the 14 participants who presented with a defect in G_i signalling based on lumi-aggregometry (Figure 2d), 11 also showed abnormal platelet responses on RPFT, while 3 subjects were considered normal. It should be noted that presentation of the G_i-type platelet defect on lumi-aggregometry is variable, overlaps considerably with corresponding measurements in healthy volunteers (Figure 3) and is based on interpretation of kinetic information not available in endpoint assays. The remaining case that was missed on RPFT presented with a reduced response to collagen on lumiaggregometry and was considered to have a potential defect in GPVI.

Table. Two-by-two analysis of the agreement between lumi-aggregometry and the RPFT.

«Positive» indicates the presence of an abnormality in platelet function and «negative» indicates normal platelet responses. PPV and NPV stand for positive and negative predictive values respectively.

	Lumi-aggregometry positive	Lumi-aggregometry negative
RPFT positive	29	6	PPV = 83%
RPFT negative	4	22	NPV = 85%
	Sensitivity = 88%	Specificity = 79%

Figure 3. Platelet aggregation assessed by LTA in healthy volunteers (filled symbols) and in patients with a defect in G_i signalling (open symbols).

Platelet aggregation was assessed by LTA in response to ADP (10 and 30μM, final aggregation) and epinephrine (10 and 30μM) in participants with defect in G_i signalling. Horizontal lines indicate cut-offs determined as 5^th percentile of the corresponding measurement in healthy volunteers.

Six participants with normal responses on lumi-aggregometry presented with an abnormal response on RPFT suggesting that this test may have an increased sensitivity. One of these patients had normal aggregation and dense granule secretion on both tests, but markedly reduced P-selectin expression in response to all agonists, indicative of impaired α-granule secretion.

Discussion

Here we evaluated the performance of a novel remote platelet function test alongside lumi-aggregometry in patients presenting with bleeding symptoms suggestive of an inherited PFD. The new test directly assesses both α- and dense granule secretion in response to platelet stimulation with several agonists in whole blood. The RPFT was able to discriminate between patients with and without an abnormality in platelet function as detected by lumi-aggregometry, and showed good levels of agreement with this reference test.

All the participants identified by lumi-aggregometry to have a defect in dense granule secretion or in the TxA₂ pathway also presented with an abnormality on RPFT. Among the participants with a defect in dense granule secretion, one had almost absent ATP release on lumiaggregometry and low CD63 expression on RPFT and was later identified to have Hermansky-Pudlak syndrome-7 [12]. Interestingly, two other participants from the same group of defects (highlighted in Figure 2b) presented with reduced P-selectin expression, but elevated CD63 expression even at baseline, and were later identified to have Hermansky-Pudlak syndrome-2, which is known to be associated with impaired sorting of CD63 and its accumulation in the plasma membrane rather than granule membrane [13].

Three individuals who were diagnosed with a defect in G_i signalling with lumi-aggregometry appeared normal on RPFT. Their phenotype was relatively mild and the diagnosis was based mainly on the kinetic information, with transient rather than sustained aggregation being observed, or with delayed aggregation or secretion.

The RPFT identified 6 participants with reduced platelet reactivity that was not detected by lumi-aggregometry. One of these participants had reduced P-selectin but normal CD63 expression in response to all agonists, suggesting impaired α-granule secretion, which is reminiscent of Gray Platelet Syndrome, a defect that would not be detected with lumi-aggregometry. These results confirm that the absence of a platelet abnormality on traditional testing cannot eliminate the possibility of a platelet defect.

There is considerable variability in methodology of LTA with a historical lack of standardisation of aggregation [14]. Recently, the Platelet Physiology Scientific and Standardization Committee of the International Society of Thrombosis and Haemostasis has begun to standardise light transmission aggregometry [15] to improve the comparability of results between laboratories. However, there remain many analytical and pre-analytical variables, which render the application of lumi-aggregometry to a large population challenging. The main advantage of the RPFT is the use of whole blood. Furthermore, the analysis is performed on fixed samples and thus can be done remotely without the need for immediate processing of the blood sample. This offers the potential for standardisation and improved accessibility of platelet function testing, which is currently restricted to specialised laboratories.

The performance of the RPFT in this study indicates that if it had been used as a pre-test, the time and labour intensive lumi-aggregometry approach could have been avoided in over one third of participants in whom no abnormality was detected using the traditional testing. Further development of the RPFT should include the measurement of glycoprotein receptors and their function, including GPVI, and platelet procoagulant function, which could improve the sensitivity of this test and would add to the measurements that are lacking in traditionally available methodologies [2]. These additions would require extra flow cytometry analyses, but would not add complexity at the point-of-care.

We conclude that, while the RPFT would not be suitable for the comprehensive diagnosis of PFDs, due to the limited number of agonists and absence of kinetic information, we believe it has a role as a pre-test to identify subjects with a potential defect who would benefit from further detailed platelet phenotyping. In subjects with a minimal history of excessive bleeding, it may also help to identify those who are unlikely to have a defect in platelet function detectable by lumi-aggregometry and thus do not require such testing. Further development of the RPFT and research in the utility of this assay as a first-line screening test in a wider population of individuals with bleeding symptoms is warranted.