Platelet dense granule release assays are recommended for diagnosing platelet function disorders and are commonly performed by Lumi-Aggregometer (Chrono-Log, Havertown, PA) assays of adenosine triphosphate (ATP) release. We conducted a prospective cohort study of people tested for ATP release defects to assess bleeding symptoms. Reduced release, with 1 or more agonists, was more common among patients with bleeding disorders than among healthy control subjects (P < .001). The respective likelihood (odds ratio [95% confidence interval]) of a bleeding disorder or an inherited platelet function disorder were high when release was reduced with 1 or more agonists (17 [6–46]; 128 [30–545]), even if aggregation was normal (12 [4–34]; 105 [20–565]). ATP release had high specificity and moderate sensitivity for inherited platelet function disorders, with most abnormalities detected by the combination of 6 μmol/L epinephrine, 5.0 μg/mL collagen, and 1 μmol/L U46619. Platelet ATP release assays are useful for evaluating common bleeding disorders, regardless of aggregation findings.
Platelet secretion defects are common, heterogeneous, inherited bleeding disorders that impair the release of platelet dense granule contents, with or without associated defects in platelet aggregation.1–6 To diagnose these conditions, experts recommend testing for platelet dense granule release when evaluating people for platelet function disorders.1,3–5,7 However, because few diagnostic laboratories offer testing for platelet release defects,8,9 this testing is often done as a secondary investigation, after testing for coagulation defects, thrombocytopenia, and von Willebrand disease (vWD).
Among the clinical laboratories that assess platelet release, most use bioluminescent assays that measure stored adenosine triphosphate (ATP) release that, unlike serotonin release assays, do not require radioisotopes.1,3,4,8–12 The bioluminescent assays are typically performed using a Lumi-Aggregometer (Chrono-Log, Havertown, PA) and platelet-rich plasma, although some laboratories test whole blood.1,4,7–10,12 The assay is performed by adding a luciferin-luciferase reagent, along with an agonist (to stimulate the release of platelet dense granule contents), while stirring the sample at low shear to promote platelet activation and aggregation.1,12 The ATP released from platelets reacts with the luciferin-luciferase reagent, resulting in light emission that is usually quantified by a Lumi-Aggregometer, relative to an ATP standard.1,12 While ATP release assays are typically used to assess people suspected of having a platelet disorder, their sensitivity and specificity have never been reported.8,9 Furthermore, uncertainties exist about the number and type of agonists that are needed to detect impaired platelet ATP release due to common platelet function disorders.
The goals of our study were to determine the following: (1) the incidence and spectrum of platelet ATP release abnormalities, detected by a panel of agonists and a Lumi-Aggregometer, in nonthrombocytopenic people who were referred for further assessment after testing for aggregation defects, dense granule deficiency, vWD, and coagulation defects; (2) the diagnostic usefulness of the assay for bleeding disorders (including inherited platelet function disorders) in this population; and (3) the usefulness of different agonists for detecting impaired platelet ATP release due to common inherited platelet function disorders.
Materials and Methods
This study was conducted with institutional ethics review approval, in accordance with the recently revised Declaration of Helsinki on human subject research, with written informed consent from all subjects.
Patients were part of a cohort who consented to participate in a research study on bleeding disorders. Each completed a bleeding history questionnaire (clinical history assessment tool [CHAT]) and underwent laboratory testing to determine their platelet count and to assess for coagulation defects, vWD, platelet aggregation defects by light transmission aggregometry (LTA), and platelet dense granule deficiency by electron microscopy.13 The cohort evaluated for release defects was the subset of CHAT study subjects13 who had been referred for ATP release testing by a hematologist to further evaluate or exclude a bleeding disorder. LTA findings for these subjects (evaluated with a different type of aggregometer and not measured simultaneously with ATP release) were included in a previous report.13 Healthy control subjects were volunteers without a history of bleeding problems who were not taking drugs that inhibited platelet function.
Determination of Diagnoses
Bleeding disorder diagnoses were established using the recorded opinion of the consulting hematologist, obtained by independent chart reviews (including review of laboratory test results) by 2 physicians (M.P. and C.H.; and for some records, K.W.), with discrepancies resolved through consensus.13 Because there are no validated methods to diagnose platelet function disorders based on the bleeding history or to assess their clinical severity, the diagnosis for each patient tested for platelet function defects was categorized, as described,13 based on the opinion of the consulting hematologist. Briefly, subjects were classified as having one of the following: (1) no bleeding disorder (defined as a negative bleeding history and normal laboratory findings); (2) a possible bleeding disorder (defined as an equivocal bleeding history with inconsistent or nondiagnostic laboratory findings), or (3) a definite bleeding disorder (defined as a significant bleeding history) that was further subcategorized into undefined (diagnostic tests normal) or diagnosed categories, based on whether there were laboratory findings consistent with vWD, a coagulation factor deficiency, a platelet disorder (defined as a history consistent with defective primary hemostasis, plus ≥1 of the following: abnormal LTA with 2 or more agonists; dense granule deficiency; reduced or absent ATP release; a drug and bleeding history consistent with a drug-induced platelet function abnormality; or documented bone marrow disorder), or another cause of bleeding (eg, liver disease or renal disease).
Platelet disorders were further subcategorized as acquired or inherited (based on a positive family history and/or bleeding since childhood without other causes) and according to type of abnormality. Briefly, platelet disorders were categorized as follows: (1) a secretion defect if the subject had reduced or absent ATP release with 1 or more agonists with or without abnormal LTA; (2) thromboxane generation defect if LTA showed impaired or absent aggregation with arachidonic acid and a normal response to thromboxane analogue; (3) dense granule deficiency as determined by whole mount electron microscopy; or (4) an undefined type of platelet function defect if the patient had an uncharacterized type of platelet function disorder associated with abnormal LTA results but normal ATP release.
All sample collections and investigations were done centrally, by the Regional Specialized Coagulation Laboratory, Hamilton Regional Laboratory Medicine Program, Hamilton, Canada, in accordance with the laboratory’s standardized operating procedures and published guidelines on platelet function testing.13–15 All subjects completed a pretest drug questionnaire. Patients and control subjects who had taken a nonsteroidal anti-inflammatory drug or a thienopyridine in the past 7 days were deferred for later testing. There were no other exclusions for medication use.
ATP release was quantified using a Lumi-Aggregometer, according to the methods recommended by the manufacturer with minor modifications in agonist panel. Briefly, assays were performed using 400 μL of platelet-rich plasma, which was prepared as described and adjusted to 250 × 109 platelets/L with autologous platelet-poor plasma13,16; 50 μL of agonist; and 50 μL of 0.2 nmol/L luciferin-luciferase reagent (Chrono-Lume, Chrono-Log; concentration reduced to 0.16 nmol/L if the result with 2 nmol/L ATP standard exceeded the measurable range). The panel of agonists (in final assay concentrations; sources as previously described13 unless otherwise stated), was as follows: 1 U/mL thrombin (Chrono-Log); 1.25 and 5.0 μg/mL Horm collagen; 5.0 μmol/L adenosine diphosphate (ADP); 1.6 mmol/L arachidonic acid; 1.0 μmol/L thromboxane analogue U46619; and 6 and 100 μmol/L epinephrine. Samples were tested with all agonists, unless there was limited sample volume. Results were reported as nmol/L of ATP release.
Reference intervals (RIs) for ATP release with each agonist were determined by nonparametric analysis of all test results from the 78 control subjects to determine the limits for 95% of the data, similar to methods previously described.16 Results were considered abnormal if the measured value was less than the lower limit of the RI.
Anonymous, coded information on diagnoses and laboratory results was entered into a database and analyzed using SAS, version 9.2 (SAS Institute, Cary, NC). When a sample had multiple determinations to verify an abnormality, the final reported value was used for data analyses. When a subject had multiple determinations on different dates, the first set of ATP release tests with all agonists was selected for odds ratio (OR) analysis. Analysis of variance was performed on all control data to determine the amount of within-individual variation and the amount of between-individual variation for ATP release. Intrarun assay variability was estimated by analysis of repeated measures on a representative sample from a healthy control subject who had been tested multiple times and consented to donate a larger sample.
Wilcoxon signed rank tests were performed on control data for agonists tested at multiple concentrations (epinephrine and collagen) to determine if the ATP release was significantly different. The Pearson χ2 test or Fisher exact test (when the expected values in any of the cells of the contingency table were <5) was used to compare the proportion of abnormal ATP release for patients with bleeding disorders with that for healthy control subjects. ORs, with 95% confidence intervals (CIs), were used to estimate the association of ATP release findings with the presence of a definite bleeding disorder (defined and undefined causes combined), based on the hematologist’s opinion of the subject’s bleeding history and not on the laboratory findings. In addition, ORs were estimated for the association of ATP release findings with a definite bleeding disorder or an inherited platelet disorder (with or without LTA abnormalities, based on results for the full agonist panel for both tests). For comparisons of agonist results for subjects with bleeding disorders and healthy control subjects, all ORs that were infinite were reexpressed as adjusted ORs that were calculated after adding the value of 1 to all data cells.
Logistic regression analysis with Firth bias correction was used for stepwise elimination of agonists with the smallest contributions to the prediction of an inherited platelet function disorder. Receiver operator characteristic (ROC) curves17 were generated to evaluate the sensitivity and specificity of ATP release, determined using data for subjects tested with the full panel of informative agonists (those with a lower RI boundary that did not extend to zero) or panels reduced to the combinations of the best 6, 5, 4, and 3 agonists that were most predictive of an inherited platelet function disorder. All statistical tests were 2-sided, and we used a .05 level of significance. P values were reported to 3 decimal places, with values less than .001 reported as < .001.
Diagnosis of Study Subjects
Study participants included 76 patients (7 men [9%]; 69 women [91%]; age range, 18–88 years; median age, 43 years) and 78 healthy control subjects (15 men [19%], 63 women [81%]; age range, 23–59 years; median age, 41 years). Because there were no patients or control subjects younger than 18 years, analyses for pediatric patients were not possible.
There was good initial agreement on bleeding disorder diagnoses, and consensus resolved the discrepancies in final diagnosis for the remaining 7 patients (9%), who all had a clinical history compatible with a definite bleeding disorder. These 7 patients included 2 people with abnormal LTA but normal ATP release findings (final classification, undefined platelet function disorder) and 5 patients with nondiagnostic findings (final classification, undefined bleeding disorder). The final consensus opinion on bleeding disorder diagnoses for the patient cohort is summarized in Table 1. Most patients referred for release testing had a definite bleeding disorder (71/76 [93%]), due to a diagnosed (40/76 [53%]) or an undefined cause (31/76 [41%]). Most patients with a diagnosed bleeding disorder had an inherited platelet function disorder (38/40 [95%]), commonly with impaired secretion. None had dense granule deficiency or severe platelet function defects (such as Bernard-Soulier syndrome or Glanzmann thrombasthenia). All patients classified as having an inherited platelet disorder had a clinical history compatible with a mild bleeding disorder.
Diagnostic Usefulness of ATP Release
Table 2 summarizes information on RIs for each agonist and for within- and between-subject variability for control subjects. The RI derived from all tests performed on the 78 control subjects indicated that ATP release with 5.0 μmol/L ADP was not informative because some had no release and accordingly the lower boundary of the RI extended to zero (Table 2). The strong agonists in the panel (thrombin and collagen) had highest values for the lower RI limit of ATP release (Table 2). The RIs overlapped for the agonists tested at multiple concentrations (epinephrine and collagen; Table 2), but the higher concentration of both agonists triggered significantly more ATP release than did the lower concentration (P < .001; Wilcoxon signed rank test).
The precision of the assay with different agonists was studied for control subjects. Repeated measures for a representative control subject indicated that the intra-assay variability ranged from 4.1% to 29.9% of the measured value for different agonists and was lowest for the strong agonists, such as thrombin and 5 μg/mL of collagen. The tests with thrombin and 5 μg/mL collagen also showed the least within-and between-subject variability, whereas arachidonic acid showed the most variability (Table 2). Analysis of variance indicated that between-individual variation accounted for 38% to 55% of the total variation in ATP release for each agonist.
Further analyses were done to compare the data for all subjects tested with the full panel of agonists, which included 74 patients and 76 control subjects. Reduced ATP release with 1 or more informative agonists was more common among patients with definite bleeding disorders (diagnosed and undefined categories combined) than among healthy control subjects (42/74 [57%] vs 6/76 [8%]; P < .001), consistent with detection of platelet function abnormalities. Accordingly, reduced ATP release with 1 or more of the informative agonists was strongly associated with a bleeding disorder (OR, 17; 95% CI, 6–46), regardless of whether the LTA result was normal (OR, 12; 95% CI, 4–34) or abnormal (OR, 101; 95% CI, 11–907).
Among the subjects classified as having an inherited platelet disorder, there was an increased likelihood of reduced ATP release with each of the informative agonists compared with healthy control subjects Table 3. The patients also had an increased likelihood of reduced ATP release with 1 or more agonists in the panel (OR, 128; 95% CI, 30–545) compared with healthy control subjects (Table 3). Reduced ATP release with 1 or more agonist was associated with an inherited platelet disorder regardless of whether the LTA result was abnormal (OR, 101; 95% CI, 11–907) or normal (OR, 105, 95% CI, 20–565) Table 4. When only a single agonist abnormality was present (compared with no abnormality), the OR was still significantly predictive of an inherited platelet disorder (OR, 23; 95% CI, 4–127). Accordingly, changing the criteria for abnormal ATP release from abnormalities with 1 or more agonists to abnormalities with 2 or more agonists or with 2 or more different agonists had minimal effect on the association between abnormal ATP release and an inherited platelet disorder (Table 4).
With the full panel, using the criterion of impaired release with 1 or more agonists gave 8% false-positives and 8% false-negatives, whereas using the criterion of abnormalities with 2 or more agonists reduced the false-positives to 1% but increased the false-negatives to 22% Table 5.
Usefulness of Different Agonists for Detection of Abnormal ATP Release
ROC curves were constructed to assess the sensitivity and specificity of impaired ATP release with different agonist panels for inherited platelet function disorders Figure 1. Impaired ATP release with 1 or more components of the full panel of informative agonists (1 U/mL thrombin; 1.25 μg/mL collagen; 5.0 μg/mL collagen; 1.6 mmol/L arachidonic acid; 6 μmol/L epinephrine; 100 μmol/L epinephrine; and 1.0 μmol/L thromboxane analogue U46619) had high specificity and moderate sensitivity for inherited platelet function disorders.
Regression modeling with stepwise elimination was used to identify the most useful agonists for the detection of inherited platelet function disorders. Figure 1 compares the ROC curves for the full panel and successively smaller panels, using the criterion of reduced ATP release with 1 or more agonists to define abnormal results. The ROC curves for the full panel and panels of the best 5 or 6 agonists were identical, indicating that testing collagen and epinephrine at multiple concentrations is not necessary (Figure 1). Reducing the test panel to the 3 most informative agonists (5.0 μg/mL collagen, 6 μmol/L epinephrine, and 1 μmol/L U46619) gave comparable results to the full panel (Figure 1 and Table 5).
The uncertainties about the value of testing for ATP release defects when diagnosing common bleeding disorders led us to formally evaluate the diagnostic usefulness of a standardized ATP release assay performed with multiple agonists. Although there was some variability in the findings for repeated tests, reduced ATP release with 1 or more agonists was strongly associated with bleeding disorders among patients already determined not to have thrombocytopenia, vWD, and coagulation defects, whereas false-positives were uncommon in healthy control subjects. Another major finding of our study was that impaired platelet ATP is associated with inherited platelet function disorders, even when the LTA result is normal. By using logistic regression and ROC analysis, we found that most of the diagnostic information from ATP release testing for inherited platelet function disorders was provided by just 3 agonists tested at single concentrations: 1 μmol/L U46619; 6 μmol/L epinephrine; and 5.0 μg/mL collagen.
Our study has several strengths. First, it relied on a large and complete database of patient symptoms and laboratory findings for a prospective cohort of patients referred for bleeding problem assessments, and it incorporated a group of healthy control subjects. Second, the ATP release assays were done with a large agonist panel that included almost all of the agonists used to assess platelet function in diagnostic laboratories. This led us to identify that some weak agonists are not informative for testing platelet ATP release (eg, 5 μmol/L ADP).8
The findings of our study have important implications for laboratory professionals who want to improve their strategies for the detection of common bleeding disorders. They provide evidence in support of expert opinion that platelet secretion testing is useful to evaluate bleeding disorders,1,3–5,7 including some conditions that are not detected by aggregometry. This major finding agrees with a recent study of platelet carbon 14–serotonin release (a less commonly performed method) that concluded that an evaluation of platelet dense granule release is helpful for diagnosing common platelet function disorders.11
One of the challenges that we faced in our study was that there are no “gold standards” or well-accepted diagnostic criteria for diagnosing platelet secretion defects or quantifying their severity. These conditions are recognized to be heterogeneous, and few have a known genetic cause.4,5,7 Bleeding scores have been developed for some bleeding disorders, but they have not been validated as suitable tools for the diagnosis of platelet secretion defects in prospective studies.18 Expert opinion recommends diagnosing platelet function disorders by using a “composite reference standard” that takes into account the clinical and laboratory findings, as done in our study.4,5,7,19 The accuracy of our subject diagnoses was optimized by having 2 additional hematologists independently review patient charts to determine diagnoses, with discrepancies resolved through consensus.13 We recognize that using the information from ATP release assays to subcategorize bleeding disorder diagnoses introduces some bias and may have increased agreement. Nevertheless, defective platelet release was associated with a higher OR for the presence of a bleeding disorder, which was a clinical diagnosis, not dependent on the results of ATP release assays.
Our future goals are to determine the bleeding risks associated with defective platelet ATP release, which may differ from the bleeding risks associated with other types of platelet abnormalities (eg, Quebec platelet disorder).20 The subjects with secretion defects in our study had a history that hematologists thought was compatible with a mild bleeding disorder. Now that we have established that impaired ATP release is associated with a bleeding disorder, we plan to use the data collected from study subjects using the CHAT questionnaire to estimate bleeding risks for common platelet disorders, including secretion defects (manuscript in preparation).
Our study primarily provides information on the diagnostic usefulness of ATP release assays as second-line investigations for subjects with a high pretest probability of a bleeding disorder (Table 1). This is a common situation in clinical practice, given that few diagnostic laboratories offer testing for platelet dense granule release defects and the laboratories that perform this testing typically evaluate release when bleeding problems are suspected.8–10 Our study provides important evidence that the ATP release assays can help determine if a bleeding disorder is present, even when the LTA result is normal. The assays can also help clarify whether an abnormal LTA result is associated with a dense granule release defect when there are normal numbers of platelet dense granules shown by electron microscopy. While our findings suggest that ATP release is more sensitive than LTA to common platelet function disorders, it is important to recognize that both tests are useful because some patients in our cohort had impaired aggregation but not dense granule release, in keeping with other reports on platelet disorders.4,5,7,8,11,13,16,19
Given that ATP release testing detected abnormalities in many subjects with bleeding disorders in our study, it would be interesting to explore the diagnostic usefulness of ATP release as an initial investigation for bleeding disorders. This would require testing on a Lumi-Aggregometer (capable of measuring aggregation and release), currently made by only 1 manufacturer. Another disadvantage is the increased cost and complexity of initial hemostasis testing. In our study, we were unable to assess the diagnostic usefulness of simultaneously measuring ATP release and LTA or to explore the relationship between release and LTA findings because our standardized procedure was to record LTA findings using a different, multichannel instrument (from a different manufacturer) with higher throughput, a validated RI, and known diagnostic usefulness.7–9 Unfortunately, the LTA RIs for the other instrument were not transferable to the Lumi-Aggregometer (unpublished observations). This suggests a need to compare the diagnostic usefulness of LTA for common bleeding disorders using different commercial instruments, which has never been undertaken.
The numbers and concentrations of agonists needed to diagnose common platelet function disorders by aggregation and ATP release assays have been uncertain, and practices vary.1 There was significant interindividual and intraindividual variability in test findings for ATP release among healthy control subjects tested on multiple occasions, which influenced the RI and probably also the diagnostic usefulness of the test. This variability likely reflects the combination of biologic variation and sample preparation–dependent effects on platelet function, in addition to measurement imprecision. Nevertheless, many of the tested agonists were helpful for evaluating ATP release, with the exception of 5.0 μmol/L ADP, which had a lower RI boundary that extended to zero.
It is interesting that weak agonists, like 6 μmol/L epinephrine, do not induce ATP secretion until aggregation occurs.13 However, 6 μmol/L epinephrine was among the most informative agonists for detecting ATP release, and it is also helpful for diagnosing common platelet function disorders by LTA.9 ROC curves confirmed that there was no added benefit to testing a higher concentration (100 μmol/L) of epinephrine, with an RI that overlapped that of a more typically evaluated (6 μmol/L) epinephrine concentration. Three agonists, tested at single concentrations (1 μmol/L U46619, 6 μmol/L epinephrine, and 5.0 μg/mL collagen, which was the higher concentration of collagen tested) had very good sensitivity and specificity for common inherited platelet function disorders. We recommend that these 3 agonists be included when assessing for dense granule release defects. Although our data suggest that it may be possible to limit test panels to only these 3 agonists, this approach increased the rate of false-negatives. In addition, a larger study would help exclude the possibility that other agonists are helpful to detect platelet function disorders that impair ATP secretion.
There have been no large studies on ATP release findings in patients with dense granule deficiency, and some laboratories assess dense granules by electron microscopy in addition to evaluating for defective release.13 We suspect that the reasons why none of the CHAT study subjects with dense granule deficiency were tested for ATP release testing in our present study is because their diagnosis was already established by electron microscopy, which was done at the same time as LTA.1,4,12 Like others, we have observed that patients with dense granule deficiency typically have impaired ATP release with many agonists, including strong agonists, such as thrombin.1,4,5 However, similar abnormalities can also be seen in patients who have defective secretion but normal numbers and contents of dense granules.7 It is reasonable for medical laboratories that assess platelet ATP release, but do not assess dense granules by electron microscopy or other methods, to report that impaired platelet ATP release can reflect a secretion defect or a dense granule deficiency.4 It is interesting that the higher prevalence of ATP release defects due to causes other than dense granule deficiency among the entire cohort of CHAT study subjects suggests that performing tests of ATP release detects more bleeding disorders than does electron microscopy assays for dense granule deficiency.
Our study demonstrates that ATP release assays, performed by a standardized method, have important diagnostic usefulness for common bleeding disorders, regardless of whether platelet function abnormalities are detected by aggregometry. In the future, the findings from this study may be helpful to formulate evidence-based recommendations on how to use ATP release assays in diagnostic algorithms for common bleeding disorders. More widespread implementation of validated testing for platelet secretion disorders would likely reduce the numbers of patients who are categorized as having an “undefined” bleeding disorder after testing for common bleeding problems, which is currently a significant proportion.3,13 The improved detection of platelet function disorders would help clinicians decide whether platelet transfusions should be considered if desmopressin therapy fails to control serious bleeding, as generally recommended for the treatment of platelet function disorders.4 A standardized approach to ATP release testing would also facilitate characterization of the clinical features, genetic causes, and bleeding risks associated with defective platelet release.
At this time, we recommend that laboratories consider measuring dense granule release by assays of ATP or serotonin release in patients with a high pretest probability of a bleeding disorder, even when LTA findings and the number of platelet dense granules per platelet are normal, using an assay with a validated RI. We also recommend testing using a panel that includes informative agonists (including 1 μmol/L U46619, 6 μmol/L epinephrine, and 5.0 μg/mL Horm collagen) and possibly other agonists, such as 1 U/mL thrombin, to help detect platelet function disorders. Given that there is significant interindividual and intraindividual variability for ATP release stimulated by different agonists, we also recommend that abnormal findings be confirmed by repeated testing to help exclude preexamination and analytic artifacts. Guideline development might be a helpful next step to improve practices for testing platelet function by dense granule release assays, as recently done for LTA.15
We thank the participants in this study and the staff of the Special Coagulation Laboratory, Hamilton Regional Laboratory Medicine Program for their assistance.
Upon completion of this activity you will be able to:
describe the test principle for bioluminescent assays of platelet adenosine triphosphate (ATP) release.
name the agonists that help detect common platelet function disorders by Lumi-Aggregometer assays of platelet ATP release.
discuss the role of ATP release assays in evaluating patients for common bleeding disorders.
The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.
The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.
Questions appear on p 477. Exam is located at www.ascp.org/ajcpcme.