- Email: [email protected]
Reproducibility of two in vivo tests of platelet function
THROMBOSIS RESEARCH 52; 507-515, 1988 0049-3848/88 $3.00 t .OO Printed in the USA. Copyright (c) 1988 Pergamon Press plc. All rights reserved.
REPRODUCIBILITY OF TWO IN VIVO TESTS OF PLATELET FUNCTION
Vincent T. Miller, Rahul Nath, Hau C. Kwaan, and Bing Nishiura Departments of Neurology and Medicine, VA Lakeside Medical Center, Chicago, Illinois, U.S.A. (Received 1.2.1988; accepted in revised form, 26.9.1988 by Editor N.U. Bang)
ABSTRACT We correlated repeat determinations of platelet aggregate ratios (PAR) and plasma beta-thromboglobulin (BTG) in the right and left arms of thirty-four subjects in order to measure test reproducibility and to evaluate the effects of venipuncture. The reproducibility of PAR on the same blood samples was relatively low but similar to that obtained when the right arm and left arm values were compared, thus indicating little variation secondary to venipuncture. For BTG, the reproducibility on identical blood samples was quite high, but fell significantly when plasma samples from the two sides were correlated. This finding and increased BTG values when obvious blood collection problems occurred, indicated that plasma BTG levels are quite sensitive to venipuncture technique. INTRODUCTION Circulating platelet aggregate ratios (PAR) and radioimmunoassay of the plasma beta-thromboglobulin (BTG), though based on quite different principals, are both purported to reflect the in vivo state of platelet activity. These tests have been criticized on several grounds, with one common concern being that platelet activation during venipuncture strongly influences the results (l-3). We determined the reproducibility of these tests in a group of healthy and hospitalized individuals by comparing them in the right and left arms, reasoning that if
Key words: Platelets, platelet function tests, beta globulins 507
508
IN VIVOPLATELET FUNCTION TESTS
Vol. 52, No. 6
they actually measure in vivo platelet activation values should be the same bilaterally. A discrepancy between sides would be attributable to variations in platelet activation during venipuncture. We specifically sought to determine if minor, inapparent differences in venipuncture technique would affect the measurements and whether obvious venipucture difficulties consistently influenced results. MBTHODS Blood was drawn from 11 normal volunteers and 24 hospitalized patients with predominantly strokes or neurodegenerative disorders. All subjects gave written informed consent. The hospitalized patients were included to obtain a broader range of test values. None of the healthy volunteers had ingested any platelet inhibiting drugs in the ten days prior to venipucture. Six of the patients were taking aspirin, but were included because, although aspirin may influence the absolute values of PAR and BTG, it is not known to interfere with the actual measurement of either. Blood Collection All subjects had venipucture performed first in the right arm, then in the left. A tourniquet was applied briefly while the needle was introduced into the vein, then released before blood was withdrawn. A 19-gauge needle was employed and 7 ml of blood drawn into a chilled 10 ml syringe, a quantity sufficient to allow duplicate determinations of PARS and BTGs. This was immediately transferred to iced BTG collection tubes supplied with the RIA kits (2.5 ml) and then to tubes in duplicate for PAR determination (4.0 ml). In each instance, any difficulty entering the vein or withdrawing blood was recorded as a bad draw and these results analyzed separately. Platelet Amte
RatiQ
We employed the modified method of Bowry et al. (4) using Sorenson’s buffer. The only variations were that we doubled the quantities of EDTA solution, saline and saline with formalin in each tube (0.8 ml total) and thus used twice the blood volume (1.Oml) per tube. To aid in rapidly transferring blood from the syringe into these tubes, marks were made on the side of the tubes to approximate one milliliter. The exact volume of added blood was determined by weighing the tubes before and after the addition of blood. Tubes then stood for 15 minutes with gentle inversion every five minutes until they were centrifuged at 150 g for 15 minutes. The top 0.5 ml of this platelet-rich supematant was then mixed and diluted into Isoton II (Curtin Matheson Scientific, Houston, Texas) at a dilution of 1:500 for counting in a Coulter Counter. Platelet counts were determined in triplicate and averaged. The platelet counts thus obtained were then corrected for the amount of added blood by the formula: ( AW + 0.8)/AW x platelet count = corrected platelet count where AW is the difference in tube weight before and after blood was added.
Vol. 52, No. 6
IN VIVOPLATELET FUNCTION TESTS
509
The corrected counts were used to calculate the platelet aggregate ratio as: corrected count (EDTA & formalin)/corrected count (EDTA) = PAR Platelet aggregate ratios were run in duplicate on each side resulting in values designated “PAR-1C” and “PAR-2C”. We additionally counted platelets in the supematant manually with a counting chamber to check the actual platelet counting as a source of error in the final PAR determination. These values were recorded as “PAR-1M” and “PAR-2M”. Beta-Thromboglobulin Beta-Thromboglobulin determinations were made with an radioimmunoassay kit supplied by Amersham Corporation (Arlington Heights, Illinois). Blood was added to the chilled collection tubes and cooled on ice for 20-30 minutes. Thereafter, centrifugation at 4’ C and 2000 g for 30 minutes yielded platelet poor plasma. We removed 0.5 ml of plasma just below the surface, divided it into two aliquots and froze them at -20” C until the assays were performed. The first aliquot from each blood draw was thawed and run in duplicate in an RIA within one month These values were designated “BTG-1”. The second aliquots were also run in duplicate with separate RIA kits within two months of freezing and labelled “BTG-2”. RIAs were run exactly as per the manufacturer’s instructions.
Mean values were compared with the two-tailed, paired or unpaired t test. Correlation coefficients were calculated in the standard fashion. To compare correlation coefficients between tests we calculated the z value as: z = [l/2 In (l+r,/l- r,) - l/2 In (l+ r2/1- r2)] / [l/(nr-3) + l/(n,-3)]‘n RESULTS Blood was drawn cleanly from both the right and left arms of 26 subjects (18 patients and 8 normal volunteers) who we labelled “Group I”. The values of BTG and PAR measured on these samples were used for side to side correlations. In two instances insufficient blood was available for a second PAR determination on these samples. In 8 subjects (6 patients and 2 normals) some difficulty was encountered in entering the vessel or getting blood flow started on one side only and these were designated “Group II”. In five cases, enough blood was collected to allow only one PAR determination in addition to the duplicate BTG measurements. Values from these eight individuals are
510
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
analyzed separately for the purposes of side to side comparisons. Finally, in one normal volunteer, difficulty with venipuncture occurred bilaterally and the values were discarded. Mean values for BTG, PAR-C and PAR-M did not differ significantly between healthy controls and hospitalized volunteers. In subsequent analysis, no differentiation between these subjects was made. TABLE I Test Reproducibility on Identical Blood Samples Test*
Mean + SEM
Range
n
BTG-1 (&ml)
26.9 + 2.0
10.1 - 64.3
34
BTG-2
26.2 Itr2.1
10.0 - 63.6
34
PAR- 1C
0.88 4.03
0.38 - 1.20
34
PAR-2C
0.85 + .04
0.45 - 1.35
30
PAR-1M
0.81 + .03
0.38 - 1.20
34
PAR-2M
0.86 + .03
0.43 - 1.19
30
t-test
Correlation
p = 82
r = .947
p = 57
r = SO8
p = .35
r = .373
* Right arm values, Group I & II subjects
Platelet Apgregate Ratio For blood drawn on Group I & II patients from the right arm, the mean values of PAR-1C and PAR-2C, duplicate determinations on the same samples, were nearly the same, being 0.88 + .03 (n = 34) (mean + SEM) and 0.85 + .04 (n = 30) respectively. These PAR-C determinations correlated with r = 0.508, (n = 30, p < .Ol) (Table I). In the 26 Group I subjects, the mean platelet aggregate ratios on the right were similar to those on the left. The correlation coefficient between the sides was 0.583 (n = 26, p < .Ol) (Table II), similar to the value for repeat determinations on the same samples. If Group II values were considered along with the Group I values in determining side to side correlation of PAR-IC, the value of r dropped to 0.48 1 (n = 34, p < .Ol). Side to side correlation between PAR-2C values was lower (Table 11). In the eight Group II individuals with unilateral venipuncture difficulty, the mean PAR-1C on the cleanly drawn side was 0.88 + .05 compared to 0.94 + .07 (n = 8, NS) on the poorly drawn side.
511
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
Manual counting of platelets did not improve the reproducibility on single samples or side to side correlation of PAR determinations (Tables I &II). The correlation of ratios on the right side determined by Coulter counting (PAR-1C) with those done by manual counting (PAR-1M) is reflected by r = 0.547 (n = 34, p c 0.01).
TABLE II Platelet Values In Group I Patients
Test
Right
(mean+ SEM)
Left
Paired t-test
Correlation
BTG- 1 (q/ml)
25.0 ). 1.5 (26)
27.9 + 2.3 (26)
p = .08
r = .722
BTG-2 (ng/ml)
24.0 * 1.6 (26)
24.1 + 1.5 (26)
p = .91
r = .747
PAR-1c
0.87 + .03 (26)
0.89 k .04 (26)
p = .61
r = .583
PAR-2C
0.84 f .04 (25)
0.92 + .04 (25)
p = .07
r = .341
PAR-1M
0.81 f .04 (26)
0.84 + .03 (26)
p = .31
r = .561
PAR-2M
0.86 + .04 (25)
0.92 zk.04 (25)
p= .ll
r = .283
Beta-Thromboglobulin The BTG-1 measurements in the right arm averaged 26.9 f 2.0 ng/ml, similar to the mean BTG-2 of 26.2 f 2.1 rig/ml. Reproducibility on the same blood sample was quite high, as reflected by the correlation coefficient of 0.947, (n = 34, p < .OOl)between BTG-1 and BTG-2 values (Table I). Considering only Group I subjects, mean values for BTG- 1 in the right arm did not differ significantly from the left (Table II). The correlation coefficient between the two sides (r = 0.722) (n = 26, p < .Ol) was significantly lower than the correlation coefficient for repeat determination on the same samples (z = 3.3 1, p c .OOl>.This correlation dropped further if Group II were also considered such that r = 0.280 (n = 34, NS). The mean BTG-2 values and side to side correlation were about the same as those calculated for BTG-1 determinations (Table II). In the eight Group II subjects, the mean value of BTG-1 on the poorly drawn side was 50.1 + 12.1 ng/ml as compared to 25.8 + 3.5 ng/ml on the cleanly drawn side. This difference didn’t quite reach significance (p = .07).
512
IN VIVOPLATELET FUNCTION TESTS
Vol. 52, No. 6
Analvsis The correlation (r = 0.947) of repeat BTG determinations on identical blood specimens was significantly greater than that of PAR-C (r = 0.508) or PAR-M (r = 0.373) (z = 4.72, p < .OOl, for BTG vs. PAR-C). Although side to side correlation in Group I patients for BTG-1 was somewhat higher than for PAR- 1C, the difference did not reach significance (z = 0.83). No significant correlation was observed between in Group I & II patients between right arm BTG-1 and PAR-1C (r = 0.165, NS) or BTG- 1 and PAR- 1M (r = 0.073, NS).
DISCUSSION Both platelet aggregate ratios and plasma beta-thromboglobulin levels continue to be reported in patients with a wide variety of vascular and thromboembolic diseases (5-7). Results have been inconsistent and sometimes, conflicting. Our results help to explain some of this discrepancy. We found the reproducibility of the PAR determination to be disturbingly low. The correlation of duplicate PARS from blood samples in the same syringe (r = 0.508). though significant, was not strong. The side to side correlation in Group I individuals with cleanly obtained blood samples was similar (r = 0.583). When Group II patients, with obviously suboptimal venipunctures, were included, the side to side correlation changed insignificantly (r = 0.48 1). Apparently, the quality of venipuncture itself is relatively unimportant in influencing the calculated PAR. Further evidence for this can be seen in the comparison of mean PAR values in Group II patients. Here, values on the poorly drawn side were insignificantly different than those on the well drawn side, contrary to what might have been expected if platelet aggregation were occuring as the result of suboptimal venipuncture. In sum, these results suggest that there is an inherent variability in the PAR determination that overshadows any lesser effects secondary to the quality of the venipuncture. Since the original description of platelet aggregate ratio determination (8), the emphasis has been on obtaining a test that yields values near the theoretical normal of 1.00 in a healthy population with relatively little variation. Mean values and standard error or deviation are usually reported. The mean values we calculated are entirely in keeping with the original norms (0.90 + 0.02 for normal subjects or 0.88 -+0.01 for patient controls reported by Wu and Hoak (8), However, test reproducibilty as measured by the correlation coefficient of duplicate samples was less than optimal in our study and has been largely ignored by others. In only two previous reports were repeat determinations made in the same individuals. One stated that the values were reproducible but reported only the averaged results (4). In the second, repeat determinations were made in ten normal subjects on blood drawn through the same needle (9). From these data, the correlation coefficient can be calculated to be 0.562, similar to what we achieved.
Vol. 52, No. 6
IN VIVO PLATELET FUNCTION TESTS
513
Whether platelet aggregatesareactually present in the circulation or are formed only during blood collection has been debated. Studies with light microscopy have failed to confii their in vivo existence (10,ll). Electron microscopic results suggest that although circulating aggregates may exist, they sometimes contain red and white blood cells and their numbers are not reflected by calculation of the platelet aggregate ratio (11). Rather than occurring because of platelet aggregates, it has been suggested that the difference in platelet counts measured between solutions of EDTA and EDTA/formalin results from an alteration in platelet mobility during centrifugation after exposure to formalin (10). In more recent modifications, the process of centrifugation has been eliminated by counting platelets in whole blood. This has resulted in ratios nearer to the theoretical normal value of 1.OO(1). The accuracy of platelet counting could obviously have a bearing on PAR calculation. In addition to using a Coulter counter, we employed a counting chamber, but found no improvement in the reproducibility of results. The relatively poor correlation between Coulter and manually counted determinations confirms that counting may be an important source of variation. The possibility that this weak correlation was due to a difference in the way the Coulter counter handled platelet aggregates was considered but rejected because analysis of the raw data showed that the correlation between manual and Coulter counts used to calculate the PARS was higher for the formalin-containing samples (with aggregates) than for those without. Other factors suspected of altering platelet aggregate ratios have been the composition of the buffer solution, blood flow through plastic tubing during collection, and the speed of blood flow into the syringe (1,4,9). We optimized or standardized all of these variables. Specifically, we employed the the buffer solution recommended by Bowry and counted platelets at a dilution identical to theirs (4). Blood was collected through a 19 gauge needle directly into a syringe to avoid contact with plastic tubing. Finally, as discussed, we did not find that moderate difficulty with blood collection resulted in spuriously low PAR values. In one report, when blood was intentionally drawn quite slowly (total venipuncture time > 100 sec.), the PARS were lower (8). None of our venipunctures were nearly that prolonged. We found the reproducibility of duplicate BTG determinations on the same sample to be excellent. However, the side to side correlation was significantly lower in the Group I patients. Mean values between the sides were similar, eliminating the possibility that a systemic effect of venipuncture in the right arm changed the values in the left. These data and the further drop in correlation if Group II patient are included suggest that venipucture plays an important role in decreasing the reproducibility by altering the measured BTG. More support for the importance of venipuncture is derived from our Group II individuals, where mean BTG values in the poorly drawn arm were almost twice those on the opposite side. The critical effect of venipucture technique in determining plasma BTG was recognized early (2,3), and this study serves to emphasize that even subtle variations in blood collection may influence measured values.
514
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
Given that PARS, in principal, reflect platelet aggregation while plasma BTG measures platelet alpha-granule release, it is perhaps not surprising that we found them not to correlate with one another. Previously, BTG levels were shown to correlate with platelet lifespan but not platelet turnover (12). It thus appears an oversimplification to regard abnormalities in these tests as indicative of increased platelet activity, an ill-defined concept. We conclude that PAR determinations, even with recently suggested modifications are poorly reproducible. This lack of reproducibilty is due mainly to the technical aspects of the methods used to preserve, separate and count aggregates and not to variations in venipuncture. BTG measurments are quite reproducible on given blood samples, but are strongly influenced by even minor differences in venipucture. ACKNOWLEDGEMENT Supported by Veteran’s Administration (Medical Research Service). We appreciate the secretarial assistance of Yolanda Vallejo. REFERENCES 1.
KOHANNA, FG., SMITH MH. and SALZMAN, EW. Do patients with thromboembolic disease have circulating platelet aggregates? Blood 64: 205209, 1984.
2. LUDLAM, CA., MOORE, S., BOLTON, AE., PEPPER, DS., and CASH, JD. The release of a platelet specific protein measured by radioimmunoassay. Thrombosis Research 6: 543-548, 1975. 3. LUDLAM, CA. and CASH, JD. Studies on the liberation of B-thromboglobulin from human platelets in vitro. Brit. J. Hematol. 33: 239-247, 1976. 4. BOWRY, SK., PRENTICE, CRM. and COURTNEY, JM. A modification of the Wu and Hoak method for the determination of platelet aggregates and platelet adhesion. Thrombosis and Haemostasis 53: 381-385, 1985. 5. HAY, CRM., WALLER, PC., CARTER, C., CAMERON, HA., PARNELL, L., RAMSAY, LE., PRESTON, FE. and GREAVES, M. Lack of effect of a 24-hour infusion of iloprost in intermittent claudication. Thrombosis Research 46: 317-324, 1987. 6. RIVAL, J., RIDDLE, JM. and STEIN, PD. Effects of chronic smoking on platelet function. Thrombosis Research 45: 75-85, 1987. 7. THEROUX, P., LATOUR, J-G., LEGER-GAUTHIER, C. and DE LARA, J. Fibrinopeptide A and platelet factor levels in unstable angina pectoris. Circulation 75: 156-162, 1987.
Vol. 52, No. 6
IN VIVOPLATELET FUNCTION TESTS
515
8. WU, KK. and HOAK, JC. A new method for the quantitative detection of platelet aggregates in patients with arterial insufficiency. Lancet 2: 924-926, 1974. 9. ROHRER, TF., PFISTER, B., WEBER, C., IMHOF, PR. and STUCKI, P. Validity of the Wu-Hoak method for the quantitative determination of platelet aggregation in vivo. Blut 36: 15-20, 1978. 10. RAPER, CGL. Circulating platelet aggregates. (letter) Thrombosis and Haemostasis 39: 537-538, 1978. 11. SANIABADI, AR., LOWE, GDO., MADHOK, R., SPOWART, K., SHAW, B., BARBENEL, JC. and FORBES, CD. A critical investigation into the existence of circulating platelet aggregates. Thrombosis and Haemostasis 56: 45-49, 1986. 12. LUDLAM, CA. Evidence for the specificity of B-thromboglobulin and studies on its plasma concentration in healthy individuals. Brit. J. Hematol. 41: 271-278, 1979.
REPRODUCIBILITY OF TWO IN VIVO TESTS OF PLATELET FUNCTION
Vincent T. Miller, Rahul Nath, Hau C. Kwaan, and Bing Nishiura Departments of Neurology and Medicine, VA Lakeside Medical Center, Chicago, Illinois, U.S.A. (Received 1.2.1988; accepted in revised form, 26.9.1988 by Editor N.U. Bang)
ABSTRACT We correlated repeat determinations of platelet aggregate ratios (PAR) and plasma beta-thromboglobulin (BTG) in the right and left arms of thirty-four subjects in order to measure test reproducibility and to evaluate the effects of venipuncture. The reproducibility of PAR on the same blood samples was relatively low but similar to that obtained when the right arm and left arm values were compared, thus indicating little variation secondary to venipuncture. For BTG, the reproducibility on identical blood samples was quite high, but fell significantly when plasma samples from the two sides were correlated. This finding and increased BTG values when obvious blood collection problems occurred, indicated that plasma BTG levels are quite sensitive to venipuncture technique. INTRODUCTION Circulating platelet aggregate ratios (PAR) and radioimmunoassay of the plasma beta-thromboglobulin (BTG), though based on quite different principals, are both purported to reflect the in vivo state of platelet activity. These tests have been criticized on several grounds, with one common concern being that platelet activation during venipuncture strongly influences the results (l-3). We determined the reproducibility of these tests in a group of healthy and hospitalized individuals by comparing them in the right and left arms, reasoning that if
Key words: Platelets, platelet function tests, beta globulins 507
508
IN VIVOPLATELET FUNCTION TESTS
Vol. 52, No. 6
they actually measure in vivo platelet activation values should be the same bilaterally. A discrepancy between sides would be attributable to variations in platelet activation during venipuncture. We specifically sought to determine if minor, inapparent differences in venipuncture technique would affect the measurements and whether obvious venipucture difficulties consistently influenced results. MBTHODS Blood was drawn from 11 normal volunteers and 24 hospitalized patients with predominantly strokes or neurodegenerative disorders. All subjects gave written informed consent. The hospitalized patients were included to obtain a broader range of test values. None of the healthy volunteers had ingested any platelet inhibiting drugs in the ten days prior to venipucture. Six of the patients were taking aspirin, but were included because, although aspirin may influence the absolute values of PAR and BTG, it is not known to interfere with the actual measurement of either. Blood Collection All subjects had venipucture performed first in the right arm, then in the left. A tourniquet was applied briefly while the needle was introduced into the vein, then released before blood was withdrawn. A 19-gauge needle was employed and 7 ml of blood drawn into a chilled 10 ml syringe, a quantity sufficient to allow duplicate determinations of PARS and BTGs. This was immediately transferred to iced BTG collection tubes supplied with the RIA kits (2.5 ml) and then to tubes in duplicate for PAR determination (4.0 ml). In each instance, any difficulty entering the vein or withdrawing blood was recorded as a bad draw and these results analyzed separately. Platelet Amte
RatiQ
We employed the modified method of Bowry et al. (4) using Sorenson’s buffer. The only variations were that we doubled the quantities of EDTA solution, saline and saline with formalin in each tube (0.8 ml total) and thus used twice the blood volume (1.Oml) per tube. To aid in rapidly transferring blood from the syringe into these tubes, marks were made on the side of the tubes to approximate one milliliter. The exact volume of added blood was determined by weighing the tubes before and after the addition of blood. Tubes then stood for 15 minutes with gentle inversion every five minutes until they were centrifuged at 150 g for 15 minutes. The top 0.5 ml of this platelet-rich supematant was then mixed and diluted into Isoton II (Curtin Matheson Scientific, Houston, Texas) at a dilution of 1:500 for counting in a Coulter Counter. Platelet counts were determined in triplicate and averaged. The platelet counts thus obtained were then corrected for the amount of added blood by the formula: ( AW + 0.8)/AW x platelet count = corrected platelet count where AW is the difference in tube weight before and after blood was added.
Vol. 52, No. 6
IN VIVOPLATELET FUNCTION TESTS
509
The corrected counts were used to calculate the platelet aggregate ratio as: corrected count (EDTA & formalin)/corrected count (EDTA) = PAR Platelet aggregate ratios were run in duplicate on each side resulting in values designated “PAR-1C” and “PAR-2C”. We additionally counted platelets in the supematant manually with a counting chamber to check the actual platelet counting as a source of error in the final PAR determination. These values were recorded as “PAR-1M” and “PAR-2M”. Beta-Thromboglobulin Beta-Thromboglobulin determinations were made with an radioimmunoassay kit supplied by Amersham Corporation (Arlington Heights, Illinois). Blood was added to the chilled collection tubes and cooled on ice for 20-30 minutes. Thereafter, centrifugation at 4’ C and 2000 g for 30 minutes yielded platelet poor plasma. We removed 0.5 ml of plasma just below the surface, divided it into two aliquots and froze them at -20” C until the assays were performed. The first aliquot from each blood draw was thawed and run in duplicate in an RIA within one month These values were designated “BTG-1”. The second aliquots were also run in duplicate with separate RIA kits within two months of freezing and labelled “BTG-2”. RIAs were run exactly as per the manufacturer’s instructions.
Mean values were compared with the two-tailed, paired or unpaired t test. Correlation coefficients were calculated in the standard fashion. To compare correlation coefficients between tests we calculated the z value as: z = [l/2 In (l+r,/l- r,) - l/2 In (l+ r2/1- r2)] / [l/(nr-3) + l/(n,-3)]‘n RESULTS Blood was drawn cleanly from both the right and left arms of 26 subjects (18 patients and 8 normal volunteers) who we labelled “Group I”. The values of BTG and PAR measured on these samples were used for side to side correlations. In two instances insufficient blood was available for a second PAR determination on these samples. In 8 subjects (6 patients and 2 normals) some difficulty was encountered in entering the vessel or getting blood flow started on one side only and these were designated “Group II”. In five cases, enough blood was collected to allow only one PAR determination in addition to the duplicate BTG measurements. Values from these eight individuals are
510
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
analyzed separately for the purposes of side to side comparisons. Finally, in one normal volunteer, difficulty with venipuncture occurred bilaterally and the values were discarded. Mean values for BTG, PAR-C and PAR-M did not differ significantly between healthy controls and hospitalized volunteers. In subsequent analysis, no differentiation between these subjects was made. TABLE I Test Reproducibility on Identical Blood Samples Test*
Mean + SEM
Range
n
BTG-1 (&ml)
26.9 + 2.0
10.1 - 64.3
34
BTG-2
26.2 Itr2.1
10.0 - 63.6
34
PAR- 1C
0.88 4.03
0.38 - 1.20
34
PAR-2C
0.85 + .04
0.45 - 1.35
30
PAR-1M
0.81 + .03
0.38 - 1.20
34
PAR-2M
0.86 + .03
0.43 - 1.19
30
t-test
Correlation
p = 82
r = .947
p = 57
r = SO8
p = .35
r = .373
* Right arm values, Group I & II subjects
Platelet Apgregate Ratio For blood drawn on Group I & II patients from the right arm, the mean values of PAR-1C and PAR-2C, duplicate determinations on the same samples, were nearly the same, being 0.88 + .03 (n = 34) (mean + SEM) and 0.85 + .04 (n = 30) respectively. These PAR-C determinations correlated with r = 0.508, (n = 30, p < .Ol) (Table I). In the 26 Group I subjects, the mean platelet aggregate ratios on the right were similar to those on the left. The correlation coefficient between the sides was 0.583 (n = 26, p < .Ol) (Table II), similar to the value for repeat determinations on the same samples. If Group II values were considered along with the Group I values in determining side to side correlation of PAR-IC, the value of r dropped to 0.48 1 (n = 34, p < .Ol). Side to side correlation between PAR-2C values was lower (Table 11). In the eight Group II individuals with unilateral venipuncture difficulty, the mean PAR-1C on the cleanly drawn side was 0.88 + .05 compared to 0.94 + .07 (n = 8, NS) on the poorly drawn side.
511
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
Manual counting of platelets did not improve the reproducibility on single samples or side to side correlation of PAR determinations (Tables I &II). The correlation of ratios on the right side determined by Coulter counting (PAR-1C) with those done by manual counting (PAR-1M) is reflected by r = 0.547 (n = 34, p c 0.01).
TABLE II Platelet Values In Group I Patients
Test
Right
(mean+ SEM)
Left
Paired t-test
Correlation
BTG- 1 (q/ml)
25.0 ). 1.5 (26)
27.9 + 2.3 (26)
p = .08
r = .722
BTG-2 (ng/ml)
24.0 * 1.6 (26)
24.1 + 1.5 (26)
p = .91
r = .747
PAR-1c
0.87 + .03 (26)
0.89 k .04 (26)
p = .61
r = .583
PAR-2C
0.84 f .04 (25)
0.92 + .04 (25)
p = .07
r = .341
PAR-1M
0.81 f .04 (26)
0.84 + .03 (26)
p = .31
r = .561
PAR-2M
0.86 + .04 (25)
0.92 zk.04 (25)
p= .ll
r = .283
Beta-Thromboglobulin The BTG-1 measurements in the right arm averaged 26.9 f 2.0 ng/ml, similar to the mean BTG-2 of 26.2 f 2.1 rig/ml. Reproducibility on the same blood sample was quite high, as reflected by the correlation coefficient of 0.947, (n = 34, p < .OOl)between BTG-1 and BTG-2 values (Table I). Considering only Group I subjects, mean values for BTG- 1 in the right arm did not differ significantly from the left (Table II). The correlation coefficient between the two sides (r = 0.722) (n = 26, p < .Ol) was significantly lower than the correlation coefficient for repeat determination on the same samples (z = 3.3 1, p c .OOl>.This correlation dropped further if Group II were also considered such that r = 0.280 (n = 34, NS). The mean BTG-2 values and side to side correlation were about the same as those calculated for BTG-1 determinations (Table II). In the eight Group II subjects, the mean value of BTG-1 on the poorly drawn side was 50.1 + 12.1 ng/ml as compared to 25.8 + 3.5 ng/ml on the cleanly drawn side. This difference didn’t quite reach significance (p = .07).
512
IN VIVOPLATELET FUNCTION TESTS
Vol. 52, No. 6
Analvsis The correlation (r = 0.947) of repeat BTG determinations on identical blood specimens was significantly greater than that of PAR-C (r = 0.508) or PAR-M (r = 0.373) (z = 4.72, p < .OOl, for BTG vs. PAR-C). Although side to side correlation in Group I patients for BTG-1 was somewhat higher than for PAR- 1C, the difference did not reach significance (z = 0.83). No significant correlation was observed between in Group I & II patients between right arm BTG-1 and PAR-1C (r = 0.165, NS) or BTG- 1 and PAR- 1M (r = 0.073, NS).
DISCUSSION Both platelet aggregate ratios and plasma beta-thromboglobulin levels continue to be reported in patients with a wide variety of vascular and thromboembolic diseases (5-7). Results have been inconsistent and sometimes, conflicting. Our results help to explain some of this discrepancy. We found the reproducibility of the PAR determination to be disturbingly low. The correlation of duplicate PARS from blood samples in the same syringe (r = 0.508). though significant, was not strong. The side to side correlation in Group I individuals with cleanly obtained blood samples was similar (r = 0.583). When Group II patients, with obviously suboptimal venipunctures, were included, the side to side correlation changed insignificantly (r = 0.48 1). Apparently, the quality of venipuncture itself is relatively unimportant in influencing the calculated PAR. Further evidence for this can be seen in the comparison of mean PAR values in Group II patients. Here, values on the poorly drawn side were insignificantly different than those on the well drawn side, contrary to what might have been expected if platelet aggregation were occuring as the result of suboptimal venipuncture. In sum, these results suggest that there is an inherent variability in the PAR determination that overshadows any lesser effects secondary to the quality of the venipuncture. Since the original description of platelet aggregate ratio determination (8), the emphasis has been on obtaining a test that yields values near the theoretical normal of 1.00 in a healthy population with relatively little variation. Mean values and standard error or deviation are usually reported. The mean values we calculated are entirely in keeping with the original norms (0.90 + 0.02 for normal subjects or 0.88 -+0.01 for patient controls reported by Wu and Hoak (8), However, test reproducibilty as measured by the correlation coefficient of duplicate samples was less than optimal in our study and has been largely ignored by others. In only two previous reports were repeat determinations made in the same individuals. One stated that the values were reproducible but reported only the averaged results (4). In the second, repeat determinations were made in ten normal subjects on blood drawn through the same needle (9). From these data, the correlation coefficient can be calculated to be 0.562, similar to what we achieved.
Vol. 52, No. 6
IN VIVO PLATELET FUNCTION TESTS
513
Whether platelet aggregatesareactually present in the circulation or are formed only during blood collection has been debated. Studies with light microscopy have failed to confii their in vivo existence (10,ll). Electron microscopic results suggest that although circulating aggregates may exist, they sometimes contain red and white blood cells and their numbers are not reflected by calculation of the platelet aggregate ratio (11). Rather than occurring because of platelet aggregates, it has been suggested that the difference in platelet counts measured between solutions of EDTA and EDTA/formalin results from an alteration in platelet mobility during centrifugation after exposure to formalin (10). In more recent modifications, the process of centrifugation has been eliminated by counting platelets in whole blood. This has resulted in ratios nearer to the theoretical normal value of 1.OO(1). The accuracy of platelet counting could obviously have a bearing on PAR calculation. In addition to using a Coulter counter, we employed a counting chamber, but found no improvement in the reproducibility of results. The relatively poor correlation between Coulter and manually counted determinations confirms that counting may be an important source of variation. The possibility that this weak correlation was due to a difference in the way the Coulter counter handled platelet aggregates was considered but rejected because analysis of the raw data showed that the correlation between manual and Coulter counts used to calculate the PARS was higher for the formalin-containing samples (with aggregates) than for those without. Other factors suspected of altering platelet aggregate ratios have been the composition of the buffer solution, blood flow through plastic tubing during collection, and the speed of blood flow into the syringe (1,4,9). We optimized or standardized all of these variables. Specifically, we employed the the buffer solution recommended by Bowry and counted platelets at a dilution identical to theirs (4). Blood was collected through a 19 gauge needle directly into a syringe to avoid contact with plastic tubing. Finally, as discussed, we did not find that moderate difficulty with blood collection resulted in spuriously low PAR values. In one report, when blood was intentionally drawn quite slowly (total venipuncture time > 100 sec.), the PARS were lower (8). None of our venipunctures were nearly that prolonged. We found the reproducibility of duplicate BTG determinations on the same sample to be excellent. However, the side to side correlation was significantly lower in the Group I patients. Mean values between the sides were similar, eliminating the possibility that a systemic effect of venipuncture in the right arm changed the values in the left. These data and the further drop in correlation if Group II patient are included suggest that venipucture plays an important role in decreasing the reproducibility by altering the measured BTG. More support for the importance of venipuncture is derived from our Group II individuals, where mean BTG values in the poorly drawn arm were almost twice those on the opposite side. The critical effect of venipucture technique in determining plasma BTG was recognized early (2,3), and this study serves to emphasize that even subtle variations in blood collection may influence measured values.
514
IN VIVO PLATELET FUNCTION TESTS
Vol. 52, No. 6
Given that PARS, in principal, reflect platelet aggregation while plasma BTG measures platelet alpha-granule release, it is perhaps not surprising that we found them not to correlate with one another. Previously, BTG levels were shown to correlate with platelet lifespan but not platelet turnover (12). It thus appears an oversimplification to regard abnormalities in these tests as indicative of increased platelet activity, an ill-defined concept. We conclude that PAR determinations, even with recently suggested modifications are poorly reproducible. This lack of reproducibilty is due mainly to the technical aspects of the methods used to preserve, separate and count aggregates and not to variations in venipuncture. BTG measurments are quite reproducible on given blood samples, but are strongly influenced by even minor differences in venipucture. ACKNOWLEDGEMENT Supported by Veteran’s Administration (Medical Research Service). We appreciate the secretarial assistance of Yolanda Vallejo. REFERENCES 1.
KOHANNA, FG., SMITH MH. and SALZMAN, EW. Do patients with thromboembolic disease have circulating platelet aggregates? Blood 64: 205209, 1984.
2. LUDLAM, CA., MOORE, S., BOLTON, AE., PEPPER, DS., and CASH, JD. The release of a platelet specific protein measured by radioimmunoassay. Thrombosis Research 6: 543-548, 1975. 3. LUDLAM, CA. and CASH, JD. Studies on the liberation of B-thromboglobulin from human platelets in vitro. Brit. J. Hematol. 33: 239-247, 1976. 4. BOWRY, SK., PRENTICE, CRM. and COURTNEY, JM. A modification of the Wu and Hoak method for the determination of platelet aggregates and platelet adhesion. Thrombosis and Haemostasis 53: 381-385, 1985. 5. HAY, CRM., WALLER, PC., CARTER, C., CAMERON, HA., PARNELL, L., RAMSAY, LE., PRESTON, FE. and GREAVES, M. Lack of effect of a 24-hour infusion of iloprost in intermittent claudication. Thrombosis Research 46: 317-324, 1987. 6. RIVAL, J., RIDDLE, JM. and STEIN, PD. Effects of chronic smoking on platelet function. Thrombosis Research 45: 75-85, 1987. 7. THEROUX, P., LATOUR, J-G., LEGER-GAUTHIER, C. and DE LARA, J. Fibrinopeptide A and platelet factor levels in unstable angina pectoris. Circulation 75: 156-162, 1987.
Vol. 52, No. 6
IN VIVOPLATELET FUNCTION TESTS
515
8. WU, KK. and HOAK, JC. A new method for the quantitative detection of platelet aggregates in patients with arterial insufficiency. Lancet 2: 924-926, 1974. 9. ROHRER, TF., PFISTER, B., WEBER, C., IMHOF, PR. and STUCKI, P. Validity of the Wu-Hoak method for the quantitative determination of platelet aggregation in vivo. Blut 36: 15-20, 1978. 10. RAPER, CGL. Circulating platelet aggregates. (letter) Thrombosis and Haemostasis 39: 537-538, 1978. 11. SANIABADI, AR., LOWE, GDO., MADHOK, R., SPOWART, K., SHAW, B., BARBENEL, JC. and FORBES, CD. A critical investigation into the existence of circulating platelet aggregates. Thrombosis and Haemostasis 56: 45-49, 1986. 12. LUDLAM, CA. Evidence for the specificity of B-thromboglobulin and studies on its plasma concentration in healthy individuals. Brit. J. Hematol. 41: 271-278, 1979.