THERAPY AND PREVENTIONCORONARY REOCCLUSION

Aspirin and dipyridamole in the prevention of acutecoronary thrombosis complicating coronaryangioplastyELLIOT S. BARNATHAN, M.D., J. SANFORD SCHWARTZ, M.D., LYNNE TAYLOR, M.S.,WARREN K. LASKEY, M.D., J. PATRICK KLEAVELAND, M.D., WILLIAM G. KUSSMAUL, M.D.,AND JOHN W. HIRSHFELD, JR., M.D.

ABSTRACT To test the hypothesis that pretreatment with adequate antiplatelet therapy reduces thelikelihood of acute coronary thrombosis during routine percutaneous transluminal coronary angioplasty(PTCA), we reviewed, blinded to treatment group, the films and records of 300 consecutive initiallysuccessful PTCAs. Films before PTCA, immediately after, and at least 30 min after the last ballooninflation were assessed for the presence of any thrombus at the PTCA site. We excluded 37 patientswho received streptokinase before PTCA or who had 100% occlusion or thrombus on pre-PTCA films.New thrombi were classified as clinically significant (defined as causing 100% occlusion or requiringemergency surgery or streptokinase therapy) or as not significant (not causing an acute problem orrequiring intervention). Patients were classified into three groups, based on the type and extent ofantiplatelet therapy received. Group 1 (no aspirin, n = 121) consisted of patients who did not receiveaspirin either before admission or in hospital before PTCA (with or without dipyridamole). Group 2(standard treatment, n = 110) received aspirin with or without dipyridamole but did not receive bothdrugs before admission and in hospital before PTCA. Group 3 (maximal treatment, n = 32) receivedboth aspirin and dipyridamole before admission and in hospital before PTCA. New thrombi weredetected at 39 (14.8%) PTCA sites, of which 15 (5.7% of all PTCA sites) were considered clinicallysignificant. Group 1 had the highest incidence of both thrombus (21.5%) and clinically significantthrombus (10.7%). A reduction was seen in group 2 in thrombus (11.8%; p = .07) and in clinicallysignificant thrombus (1.8%; p = .005). Group 3 had no thrombus (p = .001) and no clinicallysignificant thrombus (p = .04). In addition to inadequate pretreatment with antiplatelet therapy,univariate analyses demonstrated several other risk factors for thrombus: higher percent diameterstenosis before PTCA (p < .008), higher platelet count (p = .013), and current smoking (p = .03).Only higher platelet count (p < .001) and inadequate pretreatment (p = .001) were associated withclinically significant thrombus. Stepwise logistic regression analysis demonstrated that for thrombus,the lack of effective antiplatelet therapy was the most discriminatory variable, followed by currentsmoking, higher percent diameter stenosis, and dissection. For clinically significant thrombus, oncethe lack of pretreatment with effective antiplatelet therapy was considered, no other factors addedsignificant discriminatory information. Thus, antiplatelet therapy administered before PTCA is associ-ated with a decreased incidence and significance of acute coronary thrombosis complicating PTCA.The lack of effective antiplatelet therapy before PTCA, as well as current smoking, higher percentdiameter stenosis, and dissection, identify patients at risk for this complication.Circulation 76, No. 1, 125-134, 1987.

From the Cardiovascular and General Medicine Sections, Depart- ACUTE CORONARY THROMBOSIS is a potentiallyment of Medicine, and the Leonard Davis Institute of Health Econom- serious complication of percutaneous transluminalics, University of Pennsylvania, Philadelphia. coronary angioplasty (PTCA)1-1 and may occur in spite

Dr. Barnathan is a recipient of American Heart Association Fellow- coronary tapay Ta)lo andu may cculr inj sp itships, Northeastern and South Central Pennsylvania Chapters. of heparin therapy. Balloon-induced vascular injury iS

Address for correspondence: Elliot Barnathan, M.D., Cardiovascular a potent stimulus for platelet activation, which maySection, Hospital of the University of Pennsylvania, 3400 Spruce St., lead to thrombosis. Antiplatelet therapy given beforePhiladelphia, PA 19104.

Received Jan. 28, 1987; revision accepted March 26, 1987. noncoronary angioplasty was associated with de-Presented in part at the American College of Cardiology Scientific creased deposition of platelets at the angioplasty site in

Sessions, Atlanta, 1986, and at the American Heart Association Scienti-fic Sessions, Dallas, 1986. pigs given low-dose aspirin or aspirin and dipyrida-

Vol. 76, No. 1, July 1987 125

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

BARNATHAN et al.

mole6 or in man given aspirin alone.' Other factors,such as smoking, may alter endothelial competenceand promote platelet reactivity.8-10 We hypothesizedthat pretreatment with effective antiplatelet therapywould decrease the incidence or severity of thrombusformed at the angioplasty site. We designed the fol-lowing retrospective study to assess the association ofpretreatment with aspirin (with or without dipyrida-mole) with a decreased incidence of acute coronarythrombosis complicating routine PTCA. In addition,we attempted to identify those clinical and angiograph-ic variables that were predictive of this complicationand to develop a predictive model of acute thrombosiscomplicating PTCA using stepwise logistic regressionanalysis.

MethodsPatients. The study group was drawn from all PICAs per-

formed at the Hospital of the University of Pennsylvania be-tween April 20, 1980, and April 20, 1985 (402 attempted dila-tions in 369 patients). During this 5 year period, there was nouniform policy regarding the use of antiplatelet therapy beforePTCA. The decision to institute prophylactic antiplatelet ther-apy in these patients was left to the referring physician and thehospital-based cardiologist. Concerns over excess bleedingproblems in the event of emergency bypass surgery causedseveral physicians to avoid antiplatelet pretreatment. More than95% of PTCA patients received at least 10,000 U of heparinintravenously in the laboratory (mean dose 10,213 U). After1981, all patients received additional heparin as needed to keepthe activated clotting time over 300 sec.

For inclusion in this study, we required that (1) the stenosis becrossed and dilated at least once, (2) a detailed medicationhistory and in hospital nursing medication records be available,and (3) that films before, immediately after, and at least 30 minafter dilation be available for review. Our laboratory's policy isto routinely obtain angiograms in all patients 30 min after thelast balloon inflation.

Our major aim was to study the effect of aspirin (with orwithout dipyridamole) on balloon-induced new thrombus for-mation. Accordingly, a total of 139 sites from 139 differentpatients were excluded (figure 1). There were inadequate filmsor records for 57 patients (14%). No dilation was performed in

Group 3: 32Maximal Rx"

45 patients. Of the remaining 300 sites, 13 had visible thrombusbefore PTCA and were analyzed separately. Since there was noway to exclude thrombus in patients receiving streptokinase orhaving 100% occlusion before PTCA, 21 sites meeting either ofthese criteria also were excluded. In addition, three patientstaking antiplatelet agents other than aspirin or dipyridamolewere excluded, leaving 263 sites from 220 patients as our studygroup.

Antiplatelet therapy. Each patient's medication history wastaken from the hospital record. A patient was considered to havebeen taking aspirin or dipyridamole before admission if it wasrecorded in the admission history, regardless of the dose. Apatient was considered to have taken aspirin or dipyridamole inthe hospital if the medication record indicated administration ofthe drug before PTCA, regardless of the dose or administrationfrequency. Since these criteria include patients with minimaltherapy in a treatment group, they would be expected to bias theresults against the identification of a treatment effect if the effectcorrelates positively with size of dose and frequency ofadministration.The 263 sites were divided into three groups representing

three increasing levels of pretreatment with antiplatelet drugs(figure 1). Group 1 ("no aspirin," 121 sites) consisted of 101sites from 99 patients with no antiplatelet pretreatment and 20sites from 19 patients treated with dipyridamole alone. Group 2("standard treatment," 110 sites) comprised 106 patients treatedwith aspirin either before admission or in hospital before PTCAor both. Dipyridamole was also taken in 71% of these cases ateither or both times. Group 3 ("maximal treatment," 32 sites)was composed of 28 patients who were treated with both aspirinand dipyridamole both before admission and in hospital beforePTCA. Group 3 was distinguished from group 2 by the require-ment that the patients must have received both drugs over bothperiods. These and other clinical characteristics (table 1) wereobtained from the medical record without knowledge of theangiographic results.Film review. All films were reviewed by one reviewer (J. W.

H.) blinded to patient identity and treatment group. Films werescored for the presence of thrombus before, immediately after,and at least 30 min after the last balloon inflation. All films werescored as either definite thrombus, no thrombus, or uncertainfor thrombus. The same three-tiered scale was used for allangiographic findings. A PICA site was considered to havedeveloped a thrombus only if a postinflation film was scored ashaving a definite thrombus. The reviewer attempted to differen-tiate between dissection and thrombus and to score thrombus asdefinite only when this distinction seemed clear angiographical-ly. Films also were scored for the presence or absence of the

: 12

FIGURE 1. Selection of the study population.

CIRCULATION

TOTAL SITES ATTEMPTED402

Included Excluded263 139

Inadequate Data: 57

F \\\\ Not Dilated: 45

Thrombus Pre-PTCA: 13

Other Antiplatelet Rx: 3

Streptokinase Pre-PTCA

100% Stenosis: 9

126

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

THERAPY AND PREVENTION-cORONARY REOCCLUSION

TABLE 1Clinical and angiographic characteristics

Group 1 Group 2 Group 3"No aspirin" "Standard Rx" "Maximum Rx"

Characteristic (n = 121) (n = 110) (n = 32) p value

ClinicalDiabetes mellitus (%) 14.9 13.6 18.8 .77Sex (% M) 79.3 78.2 84.4 .75,3-Blocker (%) 82.6 76.4 75.0 .72Prior MI (%) 36.3 33.6 25.0 .48Hyperlipidemia (%) 18.2 26.4 31.3 .45Platelet count (X 1031mm3) 269 ± 8A 265 ± 7 247 ± 13 .36Calcium antagonist (%) 71.9 77.3 78.1 .29Unstable angina (%) 59.5 51.8 43.8 .22Hypertension (%) 49.6 31.8 37.5 .02Year (% pre-1983) 24.8 11.8 9.4 .02Current smoker (%) 49.6 30.0 37.5 .01Dilation in bypass graft (%) 0.8 2.7 12.5 .003Prior bypass surgery (%) 2.5 6.4 21.9 .0004

Angiographic% Stenosis pre-PTCA 78.8 ± 1 78.3 ± 1 78.5 ± 2 .95Eccentric stenosis (%) 47.9 41.8 37.5 .48Ulcerated plaque (%) 28.1 23.6 15.6 .33

MI = myocardial infarction.AMean ± SEM.

following angiographic findings: ulcerated plaque, eccentricstenosis, dissection, luminal irregularity, sluggish flow, peri-vascular haziness, and filling defect. The percent diameter ste-nosis was measured before and after PICA.Thrombus was defined as any combination of the following

angiographic findings that the reviewer deemed secondary todefinite thrombus and not a result of dissection or spasm: (1) theappearance of a new, nonlinear filling defect, (2) new perivas-cular haziness, (3) sluggish flow, or (4) new luminal irregulari-ties. More than one thrombus at a single site was counted onlyonce, although two thrombi at two separate sites were countedseparately.

Routine laboratory procedure included angiography immedi-ately after the last balloon inflation and angiography 30 minlater. Angiography was performed between these two times ifwarranted by the patient's symptoms. A change in the angio-graphic appearance of the angioplasty site between the immedi-ate and 30 min postinflation angiogram was instrumental inidentifying thrombi. Only 15% of thrombi were identified onthe immediate postinflation angiogram, leaving 85% that wereidentified only on the 30 min postinflation angiogram.

Angiographically detected thrombi were further subdividedinto those that were clinically significant and those that werenot. A clinically significant thrombus was defined as one caus-ing 100% occlusion of the vessel or requiring emergency bypasssurgery or streptokinase therapy. Vascular occlusion caused bydissection or spasm was not considered to represent thrombosis.A nonsignificant thrombus was defined as any angiographicallydetected thrombus that did not meet the criteria for clinicallysignificant thrombus. Thrombi discovered at subsequent cath-eterization or acute occlusions thought to be thrombotic occur-ring after initially leaving the catheterization laboratory werenot included in this study. Therefore thrombus and clinicallysignificant thrombus refer to acute events only.

Statistical analyses. For dichotomous variables, chi-squareanalysis was used unless the expected value for a cell was less

Vol. 76, No. 1, July 1987

than 5, in which case Fisher's exact test was used. A two-tailedStudent's t test was used for continuous variables. A p value<.05 was considered significant for the univariate analyseswithout correction for multiple comparisons. The relative riskof thrombosis was estimated from the odds ratio. For stepwiselogistic regression analyses, the BMDP statistical package'1was used with the F to enter a criterion of?j < .10 and the F toremove a criterion of p > .15. Continuous variables were divid-ed into four ranges for the purpose of the multivariate analyses.Loosely associated variables (p <.15) from the univariate anal-yses were entered into the stepwise logistic regression analyses.The contribution of antiplatelet therapy was further assessed byperforming multivariate logistic regression analyses after con-trolling for potentially confounding variables. Data were ana-lyzed with both the PTCA site and the patient as the units ofanalysis. Results for both groups were comparable. Unless oth-erwise specified, the results reported refer to the data analysisperformed with PTCA site as the unit of analysis.

ResultsOverall. Of 263 sites dilated, 39 (14.8%) were iden-

tified as having developed a definite thrombus. Of the39 thrombi, 15 were considered clinically significant(5.7% of total sites dilated). Six of these were treatedsuccessfully with intracoronary streptokinase alone,whereas eight required emergency bypass surgery.One complete occlusion was treated conservativelywithout immediate intervention (figure 2). Thusthrombi were responsible for eady failure of the PTCAprocedure at nine sites (3.4%).

Clinical characteristics. Clinical characteristics weresimilar in all three groups (see table 1). Group 3 had a

127

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

BARNATHAN et al.

L100% OCCLUSIONMEDICAL Rx

1(0.4%)

FIGURE 2. Overall outcome of 263 sites qualifying for the study.

disproportionately high number of bypass graft dila-tions, since most patients are treated with aspirin anddipyridamole after bypass surgery. There were morecurrent smokers (p - .01) and patients with hyperten-sion (p = .02) in groups 1 and 3 than in group 2.Because the use of antiplatelet pretreatment becamemore common during the time of the study, there was agreater proportion of patients from the 1980 to 1982period in group 1 than in the 1983 to 1985 period(p = .02).

Univariate analysis. Figure 3 demonstrates the inci-dence of thrombus and clinically significant thrombusbroken down by antiplatelet treatment groups. Com-pared with group 1 (10.7%), there were significantreductions in clinically significant thrombus for bothgroup 2 (1.8%; p - .005) and group 3 (0%; p = .04).Group 1 had a 21.5% incidence of thrombus, whereasgroup 2 had a lower rate (1 1. 8%; p = .07) and group 3had no thrombus (vs group 2, p = .04; vs group 1, p =

.001). Combining groups 2 and 3 yields a group that

GROUP 1 GROUP 2

NO STANDARDASPIRIN RxN=121 N-11O

= compared to Group 1

D: NONSIGNIFICANT-= SIGNIFICANT

P=.005 ,P=.001Vp=.04

GROUP 3

MAXIMALRx

N -32

_4= "999 v 2

FIGURE 3. Effect of increasing antiplatelet pretreatment on the inci-dence of any thrombus and of clinically significant thrombus.

128

comprises all patients who received aspirin at somepoint before PTCA. When this group is compared withgroup 1 (no aspirin), the incidence of thrombus wasreduced by more than half to 9.2% (p< .01), and theincidence of clinically significant thrombus was re-duced sixfold to 1.4% (p = .001).As seen in figure 4, the incidence of thrombus and

clinically significant thrombus also were analyzed bystrict categories of no treatment, dipyridamole alone,aspirin alone, and aspirin and dipyridamole, with anydoses being acceptable. The results were similar tothose reported above. The rates of thrombus were asfollows: no treatment 21.8%, dipyridamole alone 20%(p = .56), aspirin alone 12.5% (p = .19), and aspirinand dipyridamole 8.2% (p = .01). When analyzed forclinically significant thrombus in this way the resultsalso were similar: no treatment 11.9%, dipyridamolealone 0%(p - . 10), aspirin alone 3.1% (p = .13), andaspirin and dipyridamole 0.9% (p = .001). All p val-ues represent comparisons with the no treatmentgroup.

Several other outcome variables in addition tothrombus and clinically significant thrombus also wereassessed for the three treatment groups. These are sum-marized in table 2. Emergency bypass surgery for allreasons (p - .04) and streptokinase administration (p

.02), as well as luminal irregularity (p = .04), werenegatively correlated with antiplatelet pretreatment.Importantly, the incidence of dissection was uniformlydistributed in the groups, as would be expected if itoccurred independently of antiplatelet therapy. Thelack of significance of the correlation between fillingdefect and treatment group is probably because dissec-

25- NONSIGNIFICANT

E 20- SIGNIFICANT0

s15- p=.01

10 /1 20 r 0 P °° 'LIJ

V) -

NO Rx DP ASA ASA &ALONE ALONE DP

N-101 N-20 N=32 N:110

ASA = ASPIRIN DP = DIPYRIDAMOLE

p values compared to NO Rx

FIGURE 4. Effect of specific antiplatelet pretreatment regimens on theincidence of any thrombus and of clinically significant thrombus.

CIRCULATION

SITES DILATED263

NO THROMBUS THROMBUS224(85%) 39(15%)

NONSIGNIFICANT CLINICALLY SIGNIFICANT24 (9%) L 15(6%)

STREPTOKINASE1 EMERGENCY SURGERYALONE +X- STREPTOKINASE6(2%) 8 (3%)

V1

0

I

I-

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

THERAPY AND PREVENTION-CORONARY REOCCLUSION

TABLE 2The association of several outcome variables with increasing antiplatelet pretreatment

Group 1 Group 2 Group 3"No aspirin" "Standard Rx" "Maximal Rx"

Outcome variable (n = 121) (n = 110) (n = 32) p value

Dissection 33 (27.3) 35 (31.8) 9 (28.1) .79Sluggish flow 12 (9.9) 8 (7.3) 1 (3.1) .42Luminal irregularity 63 (52.0) 39 (35.5) 14 (43.8) .04Filling defect - any cause 37 (30.6) 25 (22.7) 5 (15.6) .15Perivascular haziness 38 (31.4) 22 (20.0) 8 (25.0) .14Streptokinase administration all causes 12 (9.9) 2 (1.8) 1 (3.1) .02Emergency bypass surgery - all causes 12 (9.9) 3 (2.7) 0 (0) .04Any thrombus 26 (21.5) 13 (11.8) 0 (0) <.005Clinically significant thrombus 13 (10.7) 2 (1.8) 0 (0) <.005

Data in parentheses are percentages.

tion-induced filling defects were included in thiscategory.

Univariate analyses also were performed with allclinical and angiographic variables to assess their de-gree of association with thrombus and clinically sig-

nificant thrombus. As seen in table 3, in addition to thelack of pretreatment with aspirin and/or dipyridamole,higher platelet count (p = .013), current smoking (p= .03), and higher percent diameter stenosis beforePTCA (p = .008) were positively associated with

TABLE 3Association of clinical and angiographic variables with any thrombus and clinically significant thrombus

Any thrombus Clinically significant thrombus

Yes No p value Odds ratioA Yes No p value Odds ratioA

Dichotomous variablesDrug pretreatment

Aspririn before admission (%) 15 33 .04 0.4 (0.1-0.9) 7 32 .03 0.2 (0.02-0.9)Aspirin in hospital (%) 26 45 .04 0.4 (0.2-0.9) 13 44 .02 0.2 (0.04-0.9)Dipyridamole before admission (%) 10 29 .02 0.3 (0.1-0.9) 0 28 .008 0Dipyridamole in hospital (%) 31 45 .14 0.5 (0.2-1.2) 7 45 .002 0.1 (0.01-0.6)No aspirin (%) 67 42 <.01 2.7 (1.5-5.9) 87 44 .001 8.3 (1.7-33.7)No dipyridamole (%) 67 48 .04 2.2 (1.0-4.8) 93 48 .0004 15.2 (2.1-82.6)No aspirin or dipyridamole (%) 56 35 .02 2.4 (1.1-5.0) 87 35 .0001 11.8 (2.5-44.0)Calcium antagonist (%) 67 76 .28 0.6 (0.3-1.4) 67 75 .65 0.7 (0.2-2.3)fl-Blocker (%) 85 78 .48 1.5 (0.6-4.9) 80 79 .61 1.1 (0.3-1.7)

ClinicalSex (% M) 79 79 .88 1.0 (0.4-2.6) 73 80 .37 0.7 (0.2-2.7)Year (% pre-1983) 23 17 .44 1.5 (0.6-3.7) 33 17 .19 2.5 (0.7-8.6)Diabetes mellitus (%) 8 16 .13 0.4 (0.1-1.6) 13 15 .61 0.9 (0.3-4.4)Hyperlipidemia (%) 21 24 .19 0.8 (0.3-2.0) 13 24 .28 0.5 (0.1-2.4)Current smoking (%) 56 37 .03 2.2 (1.1-4.6) 67 38 .06 3.2 (1.0-11.2)Prior MI (%) 28 35 .53 0.9 (0.4-1.9) 20 35 .19 0.5 (0.1-1.9)Prior bypass surgery (%) 0 8 .06 0 0 7 .36 0Unstable angina (%) 56 54 .91 1.1 (0.5-2.3) 60 54 .43 1.3 (0.4-4.2)

AngiographicDissection (%) 41 27 .12 1.9 (0.9-4.0) 40 29 .52 1.7 (0.5-5.4)Eccentric stenosis (%) 59 42 .06 2.0 (1.0-4.3) 47 44 .95 1.1 (0.4-3.5)Ulcerated plaque (%) 39 22 .05 2.2 (1.0-4.7) 27 25 .90 1.1 (0.3-4.0)

Continuous variables% Stenosis pre-PICA 83 1B 78 + 1 .008 82 2 79 1 .22Platelet count (X 103/mm3) 293 16 260±5 .013 329±29 261±5 <.001

MI = myocardial infarction.A95% confidence limits in parentheses.BMean ± SEM.

129Vol. 76, No. 1, July 1987

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

BARNATHAN et al.

thrombus. Only the lack of pretreatment with aspirinand/or dipyridamole and higher platelet count (p<.001) were associated with clinically significantthrombus. Trends toward association with thrombuswere noted for ulcerated plaque (p = .05), eccentricstenosis (p - .06), and dissection (p = .12). Forclinically significant thrombus, a trend was noted forcurrent smoking (p = .06). A trend toward a negativeassociation was noted with prior bypass surgery (p =

.06), although virtually all of these patients were treat-ed with aspirin and/or dipyridamole. There were onlyeight dilations in bypass grafts, and when these wereexcluded the results were unchanged.The relative risk of developing a thrombus during

PTCA was estimated by the odds ratio. Without aspirinpretreatment the relative risk for developing anythrombus was 2.7 (95% confidence limits = 1.5-5.9)(table 3). There was an eightfold increased risk ofdeveloping a clinically significant thrombus withoutaspirin and a 15-fold increased risk without dipyrida-mole. There was approximately a twofold increasedrisk of any thrombus for current smoking, ulceratedplaque, eccentric stenosis, and dissection.

The year the procedure was performed is a potentialconfounding variable. The fraction of pretreated pa-tients increased in the later years. Additionally, experi-ence was greater and technique improved in the latterhalf of the study period. If the lack of thrombus pro-duction was a function of operator experience and im-proved technique, an association between year ofPTCA and thrombus formation would be expected.However, if the production of thrombi were insteadrelated only to the balloon injury (which presumablyhas not appreciably changed), then the year of theprocedure should not be associated with thrombus. Asshown in figure 5, there was no significant differencebetween the early period (1980 to 1982) and the lateperiod (1983 to 1985) in the incidence of either throm-bus or clinically significant thrombus for the aspirin-treated or the untreated patients.

Thrombus before PTCA. We analyzed separately thedata from the excluded group that had thrombus beforePTCA. Of the 13 sites, nine still had visible thrombusafter PTCA (69%) and three had a clinically significantthrombus (23%). Two of these three patients requiredemergency bypass surgery and one was treated withstreptokinase. These three patients were among thefive that received no aspirin or dipyridamole at anypoint before PTCA, thus yielding a clinically signifi-cant thrombus rate of 60% in patients with preexistingthrombus who were not pretreated with antiplateletdrugs.

, 25

0 20I

I 151-

(. 10.LU

o 5.-

p=NS1

1

71 '/

/ //7,

/7

80-82 83-85N-30 N-91

NO ASPIRIN

W:1 NONSIGNIFICANT

-= SIGNIFICANT

p= NS

nS80-82 83-85N-16 N-126

ASPIRINFIGURE 5. Rates of thrombosis in the 'early' (1980 to 1982) vs the"late" (1983 to 1985) period for aspirin-treated and untreated patients.

Multivariate analyses. Using stepwise logistic regres-sion analyses, we evaluated several models for predict-ing thrombus and clinically significant thrombus. Ini-tially we evaluated the seven variables identified by theunivariate analyses as being associated with thrombusor clinically significant thrombus: antiplatelet pretreat-ment status, platelet count, current smoking, percentdiameter stenosis before PTCA, ulcerated plaque, ec-centric stenosis, and dissection. The "lack of treat-ment" variable was entered as a graded treatment vari-able with no treatment = group 1, intermediatetreatment = group 2, and maximal treatment = group3. Platelet count was entered as four groups in relationto the mean: more than 1 SD below the mean, from 1SD below to the mean, from the mean to 1 SD above,and over 1 SD above. The percent diameter stenosiswas entered as four groups as well: under 70%, 70% to79%, 80% to 89%, and 90% to 99%. All other varia-bles were dichotomous.

For thrombus, "lack of treatment" entered first, withan improvement in the chi-square p value of .001,followed by current smoking (p = .06), percent diam-eter stenosis (p = .03) , and dissection (p = .12).Eccentric stenosis, ulcerated plaque, and platelet countdid not add any additional discriminating information.If "lack of treatment" was redefined as simply "noaspirin" and the rest of the model was kept the same,"lack of treatment" still entered first (p = .006), fol-lowed by eccentric stenosis (p = .06), dissection (p =.10), and current smoking (p - .06). For clinicallysignificant thrombus, "lack of treatment" defined ei-ther way entered first (p = .001) with no other variableentering and staying in the model, suggesting that noother variable added any significant discriminatory in-

CIRCULAFION

L, / .' / / / A--

L, 11 / 11 1/ /1

130

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

THERAPY AND PREVENTION-coRONARY REOCCLUSION

formation after controlling for the lack of pretreatmentwith aspirin.To address the question of nonrandomization of

clinical characteristics in our treatment groups, we ranthe stepwise logistic regression analysis with the non-randomly distributed variables entered first, thus con-trolling for them. The year the procedure was per-formed was made into a dichotomous variable with the"early" period being 1980 to 1982 and the "late" periodbeing 1983 to 1985. This variable, along with currentsmoking, hypertension, and prior bypass surgery, wasforced into the regression model first. Then the re-mainder of the associated variables from the univariateanalysis were allowed to move in or out. With throm-bus as the dependent variable, percent diameter steno-sis entered first (p = .01), followed by "lack of treat-ment" (p = .007) and ulcerated plaque (p = .10). Noother variables entered the model. For clinically sig-nificant thrombus, "lack of treatment" entered first (p= .01), followed by platelet count (p = .12) anddissection (p = .16).We attempted to control for nonrandomly distribut-

ed variables one last way by running nonstepwise lo-gistic regression with all variables forced into the equa-tion simultaneously. In this case, the size of the "F toremove" for a variable is a measure of its relativeimportance in the model. As shown in table 4 for boththrombus and clinically significant thrombus, the F toremove is highest for "lack of treatment."

DiscussionThis study demonstrates that in patients undergoing

elective PTCA without angiographic evidence of

TABLE 4Nonstepwise (forced) logistic regression analysis

Clinicallysignficant

Any thrombus thrombus

F to F to

Variable remove p value remove p value

Inadequate treatment 77.1 <.0001 3.1 <.05

Ulcerated plaque 3.2 .07 0.2 .63% Stenosis 2.2 .09 0.7 .58Hypertension 2.4 .11 2.4 .13Dissection 2.1 .14 1.5 .22Year 1.7 .15 1.3 .26Current smoking 2.0 .16 1.9 .17Platelet count 1.5 .21 2.2 .09Eccentric stenosis 0.6 .45 0.0 .95Prior bypass surgery 0.1 .83 0.0 .85

Vol. 76, No. 1, July 1987

thrombus before the procedure, the lack of pretreat-ment with effective antiplatelet therapy was the mostpowerful predictor of acute thrombosis complicatingthe procedure. Moreover, it was an even better predic-tor of clinically significant thrombosis. Increasing lev-els of antiplatelet therapy were associated with reducedstreptokinase administration and, more importantly,with less emergency bypass surgery for all reasons.The observation that thrombi can form during PlTCA

despite apparently adequate heparin therapy is notnew.' Heparin used alone in PTCA has theoreticalreasons for being inadequate and possibly detrimentalin terms of preventing platelet activation at the angio-plasty site. It has several effects on platelets in vitro,'2some inhibitory,'3 and others resulting in increasedactivation.'3-15 Heparin has been shown to neutralizethe antiaggregatory activity of prostacyclin.'6 Plateletinhibitory drugs may overcome heparin-dependent ag-gregation. 15 Thus aspirin may attenuate this potentiallynegative effect.PICA causes a pronounced, immediate, localized

endothelial injury that has been demonstrated graphi-cally by scanning electron microscopy.'1818 The vesselthat is denuded of its endothelium exposes proaggrega-tory substances in the subendothelial layers such ascollagen. Without the intact, functioning endothelium,a host of natural anticoagulant, fibrinolytic, and anti-aggregatory mechanisms may be lost. In addition, en-dothelial injury may lead to the release of procoagulantsubstances such as tissue factor. '9 Platelets have beenshown in vitro to augment this response. 9 20 Severalinvestigators have shown in animals that, in the ab-sence of antiplatelet therapy, platelets are rapidly de-posited at an angioplasty site.6' 17 18 21,22 This has beencorroborated in man by indium-ill platelet scintig-raphy.23 Pretreatment with aspirin in peripheral an-gioplasty in man has been shown to decrease plateletdeposition at the angioplasty site with the same tech-nique.7 Endothelial injury also may lead to decreasedendogenous synthesis of tissue plasminogen activator(t-PA) and increased release of an inhibitor of plasmin-ogen activators.2426 These two effects tend to be addi-tive in diminishing the natural fibrinolytic mecha-nisms, thereby impairing the vessel's ability to lysedeveloping fibrin thrombi. Endothelial cells also haveone or more receptors for t-PA that may modulate localfibrinolytic function.27 28 Balloon-induced endothelialinjury may alter local t-PA binding. Actual fibrinstrands have been seen shortly after PTCA by angios-copy.29 Recently, activated human platelets in vitrohave been shown to release an inhibitor of t-PA.30 Thuslocal platelet activation also may help protect fibrin-

131

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

BARNATHAN et al.

containing thrombi from the body's natural fibrinolyticmechanisms. Although this is part of the normal hemo-static process, it also may contribute to the pathogene-sis of acute thrombosis complicating PTCA.The finding that higher platelet counts before PTCA

were associated with a higher thrombosis rate is notsurprising. Platelet count has been shown to be directlyrelated to the size of a thrombus in a thrombus growthmodel in vitro.31 That it failed to be an independentpredictor after controlling for antiplatelet therapy sug-gests that the negative effect of having more platelets ispartially or completely ameliorated by rendering thempartially ineffective.We demonstrated an association between current

smoking and thrombosis complicating PTCA. Nico-tine has been shown to inhibit vascular prostacyclin,but not platelet thromboxane production.'0 Platelet ag-gregation is increased in chronic smokers, and smok-ing makes platelets more responsive to various proag-gregatory stimuli over the short term .8 In a ratpreparation, chronic smoking caused aortic endothelialdenudation, subendothelial platelet adhesion, and al-tered endothelial ultrastructure.9 Thus a smoker whoalready may be in a hypercoagulable state may requireaggressive antiplatelet and/or anticoagulant therapy toprevent a thrombotic complication when subjected tothe prothrombotic stress of PTCA.

Clearly, dissections also produce powerful stimulifor platelet adherence and thrombus formation."8 Thepresence of an eccentric stenosis, an ulcerated plaque,a very tight stenosis, or an intimal dissection mayincrease the risk of developing small thrombi. Ourstudy suggests that in the face of adequate or maximalantiplatelet therapy, these lesion characteristics are notassociated with an increased risk of major thrombusformation. However, our group sizes are small and theresulting power of not being able to demonstrate anassociation is low. Correcting for nonrandomly dis-tributed characteristics suggested a possible predictivevalue for a higher percent diameter stenosis, dissec-tion, and ulcerated plaque. Further evaluation of thepotential association of these characteristics with pooroutcomes is warranted.

Support for the contention that total occlusion fre-quently may be related to platelet activation comesfrom Peterson et al.,32 who demonstrated that therewas no rise in coronary sinus thromboxane B2 in sevenof seven uncomplicated PTCAs. However, both pa-tients with total occlusion requiring bypass surgery andnot responding to nitroglycerin had markedly elevatedlevels of thromL,.ne B2. A patient with occlusionsecondary to spasm had no elevation. This suggests

that platelet activation is either a consequence of or a

contributing causative factor in acute occlusion.Previous studies. Six large series of patients undergo-

ing PTCA report an incidence of total occlusion be-tween 1.6% and 12.2%.2 4 5 Of these six series,four reported their typical antiplatelet regimen. Inthree series in which aspirin with or without dipyrida-mole was administered routinely before PTCA, themean total occlusion rate was 3% in 1394 patients. Incontrast, the remaining study in which dipyridamolewith or without aspirin was administered4 reported a

12.2% total occlusion rate in 238 patients. These fig-ures do not represent thrombosis rates. Since total oc-clusion may be a result of spasm, dissection, thrombo-sis, or a combination of these, it is hard to comparethese results directly with ours. Nevertheless, ouroverall emergency surgery rates are comparable tothose series in which patients were treated with similarregimens.

Limitations. There are several caveats to be consid-ered when interpreting our data. The angiographicmethod of detecting thrombi is neither very sensitivenor entirely specific. Filling defects may be caused byfactors other than thrombi (e.g., dissection). Never-theless, preliminary angioscopic results suggest thatangiography underestimates the incidence of throm-bus, being more specific than it is sensitive.36 37 Whena vessel becomes totally occluded, it is often difficultto be sure whether the cause is purely thrombotic,purely secondary to dissection, or a combination ofboth factors. In one series, all six patients whosePTCAs failed, who required emergency surgery, andwho underwent angioscopy had thrombus, althoughonly two were recognized angiographically.37 We re-

garded all sites scored as "uncertain" for thrombus as

having no thrombus, thereby probably underestimat-ing the true incidence of thrombus. This was true fordissection as well. Thrombi that are layered uniformlyagainst the vessel wall likely will be missed, as will bethose microthrombi that are too small to be seen on an

angiogram.The overall incidence of thrombi detected in our

laboratory is greater than that reported in other se-

ries.2, 4 We believe this difference is in part due to our

laboratory's policy to routinely perform angiogramson all patients at least 30 min after the last ballooninflation. Only 15% of the 39 thrombi detected were

seen on the immediate postinflation films. Thus, if a

patient leaves the laboratory shortly after the last bal-loon inflation, most thrombi will not be detected andthe incidence of thrombus formation will beunderestimated.

132 CIRCULATION

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

THERAPY AND PREVENTION-CORONARY REOCCLUSION

This study was retrospective with no randomizationrelative to treatment groups. We attempted to correctfor this by forcing those nonrandomly distributed var-iables into the multivariate analysis first, thereby con-trolling for them. When doing so, our results wereessentially unchanged in that the lack of adequate anti-platelet pretreatment was still a powerful predictor ofboth any thrombus and clinically significant thrombus.Similarly, it was the most important variable when allindependent variables were forced into a regressionwith either thrombus or clinically significant thrombusas the dependent variable.

There was marked heterogeneity within our drugtreatment groups, due in part to changes in the waysthese drugs have been used over the past 5 years. Toqualify for the maximal treatment group, one only hadto take both drugs before admission (any dose qualifiedfor entry) and to receive at least one dose of each in thehospital. For dipyridamole, one dose likely would rep-resent a suboptimal regimen, and this was often thecase. Thus we do not believe that our study suggests anoptimal regimen. Arguments could be made for lessaspirin (to minimize the effects on endothelial prosta-cyclin production) or for more aspirin (to reduce totalthromboxane A2 production as much as possible).Even the timing of the dose relative to the time of theprocedure theoretically could matter, with a longerinterval between aspirin administration and PTCA al-lowing for more complete return of vascular prostacy-clin production.Our data suggest that some aspirin is probably better

than none and that long-term as well as short-termdipyridamole before the procedure may add additionalbenefit in terms of reducing the incidence of anythrombus at the PTCA site. This needs to be investigat-ed further by prospective controlled trials before fur-ther conclusions can be drawn.

We thank Dr. Robert Centor for his thoughtful review of thedata analyses and Ms. Eileen Eckstein for her photographicexpertise.

References1. Schoffer J, Krebber H, Bleifeld W, Mathey DG: Acute coronary

artery occlusion during percutaneous transluminal coronary angio-plasty: reopening by intracoronary streptokinase before emergencycoronary artery surgery to prevent myocardial infarction. Circula-tion 66: 1325, 1982

2. MacDonald RG, Feldman RL, Conti CR, Pepine CJ: Thromboem-bolic complication of coronary angioplasty. Am J Cardiol 54: 916,1984

3. Ischinger T, Zack P, Aker U: Acute coronary occlusion duringballoon angioplasty due to intracoronary thrombus and coronaryspasm: a reversible complication. Am Heart J 107: 1271, 1984

4. Mabin TA, Holmes DR, Smith HC, Vlietstra RE, Bove AA,Reeder GS, Chesebro JH, Bresnahan JF, Orszulak TA: Intracoro-

Vol. 76, No. 1, July 1987

nary thrombus: role in coronary occlusion complicating percutane-ous transluminal coronary angioplasty. J Am Coll Cardiol 5: 198,1985

5. Shiu MF, Silverton NP, Oakeley D, Cumberland D: Acute coro-nary occlusion during percutaneous transluminal coronary angio-plasty. Br Heart J 54: 129, 1985

6. Steele PM, Chesebro JH, Stanson AW, Holmes DR, Badimon L,Fuster V: Balloon angioplasty: effect of platelet-inhibitor drugs onplatelet-thrombus deposition in a pig model. J Am Coll Cardiol 3:506, 1984

7. Cunningham DA, Kumar B, Siegal BA, Gilula LA, Totty WG,Welch MJ: Aspirin inhibition of platelet deposition at angioplastysites: demonstration by platelet scintigraphy. Radiology 151: 487,1984

8. Klein L: Cigarette smoking, atherosclerosis and the coronaryhemodynamic response: a unifying hypothesis. J Am Coll Cardiol4: 972, 1984

9. Pittilo RM, Mackie JJ, Rowles PM, Machin SJ, Woolf N: Effectsof cigarette smoking on the ultrastructure of rat thoracic aorta andits ability to produce prostacyclin. Thromb Haemost 48: 173, 1982

10. Wenmalm A, Alster P: Nicotine inhibits vascular prostacyclin butnot platelet thromboxane formation. Gen Pharmacol 14: 189, 1983

11. BMDP Statistical software package, Berkeley, 1981, University ofCalifornia Press

12. Ham JM, Lawrence JC: Heparin, heparin-activated enzymes andplatelets. Haemostasis 6: 26, 1977

13. Saba HI, Saba SR, Morelli GA: Effect of heparin on platelet aggre-gation. Am J Hematol 17: 295, 1984

14. Salzman EW, Rosenberg RD, Smith MH, Lindon JN, Favreau L:Effect of heparin and heparin fractions on platelet aggregation. JClin Invest 65: 64, 1980

15. Brace LD, Issleib S, Fareed J: Heparin-induced platelet aggrega-tion is inhibited by antagonists of the thromboxane pathway.Thromb Res 39: 533, 1985

16. Saba HI, Saba SR, Blackburn CA, Hartman RC, Mason RG: Hepa-rin neutralization of PGI2: effects upon platelets. Science 205: 499,1979

17. O'gara PT, Guerrero JL, Feldman B, Fallon JT, Block PC: Effectof dextran and aspirin on platelet adherence after transluminalangioplasty of normal canine coronary arteries. Am J Cardiol 53:1695, 1984

18. Steele PM, Cheseboro JH, Stanson AW, Holmes DR, DewanjeeMK, Badimon L, Fuster V: Balloon angioplasty: natural history ofthe pathophysiological response to injury in a pig model. Circ Res57: 105, 1985

19. Tannenbaum SH, Finko R, Cines DB: Antibody and immune com-plexes induce tissue factor production by human endothelial cells. JImmunol 137: 1532, 1986

20. Brox JH, Osterud B, Bjorklid E, Fenton JW: Production and avail-ability of thromboplastin in endothelial cells: the effects of throm-bin, endotoxin, and platelets. Br J Hematol 57: 239, 1984

21. Wu KK, Chen YC, Fordham E, Ts'ao CH, Rayudu G, MatayoshiD: Differential effects of two doses of aspirin on platelet-vesselwall interaction in vivo. J Clin Invest 68: 382, 1981

22. Buchanan MR, Dejana E, Gent M, Mustard JF, Hirsch J: Enhancedplatelet accumulation onto injured carotid arteries in rabbits afteraspirin treatment. J Clin Invest 67: 503, 1981

23. Pope CF, Ezekowitz MD, Smith EO, Rapoport S, Glickman M,Sostman HD, Zaret BL: Detection of platelet depostion at the site ofperipheral balloon angioplasty using indium- 111 platelet scintigra-phy. Am J Cardiol 55: 495, 1985

24. Tannenbaum S, Barnathan E, Vanderkeyl H, Kuo A, Murray S,Cines DB: Aggregated IgG stimulates the fibrinolytic inhibitoryactivity of human endothelial cells. Blood 68: 119A, 1986

25. Emeis JJ, Kooistra T: Interleukin I and lipopolysaccharide inducean inhibitor of tissue-type plasminogen activator in vivo and incultured endothelial cells. J Exp Med 163: 1260, 1986

26. Nachman RL, Hajjar KA, Silverstein RL, Dinarello, CA: Interleu-kin I induces endothelial synthesis of plasminogen activator. J ExpMed 163: 1595, 1986

27. Haijar KA, Harpel PC, Nachman RL: Binding of tisuue plasmino-gen activator to cultured human endothelial cells. Circulation74(suppl II): 11-233, 1986 (abst)

28. Barnathan ES, Kuo A, Vanderkeyl H, McCrae K, Cines DB: Bind-

133

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

BARNATHAN et al.

ing of plasminogen activators to human endothelial cells. Clin Res35: 596A, 1987 (abst)

29. Myler RK, Stertzer SH, Millhouse F, Crew J, McMorrow J, Hidal-go B: Intravascular angioscopy (IVA) and human peripheral angio-plasty. Circulation 70(suppl II): 11-265, 1984 (abst)

30. Erikson LA, Ginsberg MH, Loskutoff DJ: Detection and partialcharacterization of an inhibitor of plasminogen activator in humanplatelets. J Clin Invest 74: 1465, 1984

31. Hanson SR, Harker LA: Studies of suloctidil in experimentalthrombosis in baboons. Thromb Haemost 53: 423, 1985

32. Peterson MB, Machaj V, Block PC, Palacios I, Philbin D, WatkinsWD. Thromboxane release during percutaneous transluminal coro-nary angioplasty. Am Heart J 111: 1, 1986

33. Holmes DR, Vleitstra RE, Mock MB, Reeder GS, Smith HC, BoveAA, Bresnahan JF, Piehler JM, Schaff HV, Orszulak TA: Angio-graphic changes produced by percutaneous transluminal coronaryangioplasty. Am J Cardiol 51: 676, 1983

34. Hollman J, Gruentzig AR, Douglas JS, King SB, Ischinger T,Meier B: Acute occlusion after percutaneous transluminal coronaryangioplasty a new approach. Circulation 68: 725, 1983

35. Dorros G, Cowley MJ, Simpson J, Bentivoglio LG, Block PC,Bourassa M, Betre K, Cosselin AJ, Gruentzig AR, Kelsey SF,Kent KM, Mock MB, Mullin SM, Myler RK, Passamini ER,Stertzer SH, Williams DO: Percutaneous transluminal coronaryangioplasty: report of complications from the National Heart,Lung, and Blood Institute PTCA registry. Circulation 67: 723,1983

36. Grundfest WS, Litvack F, Sherman T, Lee ME, Carroll R, ChauxA, Kass R, Matloff J, Forrester JS: Definition of new pathophysio-logic mechanisms and altered decisions: an outcome of intravascu-lar angioscopy. J Am Coll Cardiol 7: 153A, 1986

37. Hickey A, Litvack F, Grundfest W, Sherman T. Lee M, Chaux A,Kass R, Blanche C, Matloff J, Swan HJC, Forrester, JS: In vivoangioscopy following balloon angioplasty. Circulation 74(supplII): 11-458, 1986 (abst)

CIRCULATION134

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from

Hirshfeld, JrE S Barnathan, J S Schwartz, L Taylor, W K Laskey, J P Kleaveland, W G Kussmaul and J W

coronary angioplasty.Aspirin and dipyridamole in the prevention of acute coronary thrombosis complicating

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1987 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation doi: 10.1161/01.CIR.76.1.125

1987;76:125-134Circulation. 

http://circ.ahajournals.org/content/76/1/125the World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer information about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. FurtherEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculationpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on March 22, 2018

http://circ.ahajournals.org/D

ownloaded from