be quite deceiving and may not reflect the true severityof the intraluminal atherosclerotic disease. Moreover,angiographic severity does not determine plaque vul-nerability because most acute coronary syndromeshave been shown to occur in arteries with angio-graphically insignificant stenosis. Current concepts re-garding plaque instability imply a significant role ofinflammation. Our results agree with this notion andindicate that neopterin, a marker of macrophage ac-tivity and immune activation, may help identify pa-tients with vulnerable plaques, irrespective of the se-verity of their coronary narrowing. The high event ratein this patient group is somehow surprising, but allsubjects had risk factors, were symptomatic, and re-quired angiography. Moreover, our results agree withthe findings of Suwaidi et al7 who reported that 14%of patients with chest pain and endothelial dysfunctionhad serious adverse events despite having mild CAD.In our study, CRP and neopterin levels were higher inpatients who developed events, but only with multi-variate analysis did neopterin remain an independentpredictor of cardiovascular adverse events. Whetherthis finding indicates that neopterin is a better prog-nostic marker than CRP in this particular study pop-ulation requires further investigation.

The main limitation of this study was the relatively

small size of the patient population investigated, andthus a lack of statistical power may explain why CRPlevels did not differ significantly at follow-up betweenpatients with and without events. Moreover, the smallpopulation size did not allow us to identify a prognos-tic cut-off level for neopterin, and this is an issue thatrequires further work. Our results nevertheless suggestthat neopterin may be useful in identifying patientswho are at a higher risk among those with hyperten-sion, stable angina, and nonobstructive CAD.

1. Huber C, Batchelor JR, Fuchs D, Hausen A, Lang A, Niederwieser D,Reibnegger G, Swetly P, Troppamir J, Wachter H. Immune response-associatedproduction of neopterin. Release from macrophages primarily under control ofinterferon-gamma. J Exp Med 1984;160:310–316.2. Gupta S, Fredericks S, Schwartzman RA, Holt DW, Kaski JC. Serum neopterinin acute coronary syndromes. Lancet 1997;349:1252–1253.3. Garcia-Moll X, Coccolo F, Cole D, Kaski JC. Serum neopterin and complexstenosis morphology in patients with unstable angina. J Am Coll Cardiol 2000;35:956–962.4. Garcia-Moll X, Cole D, Zouridakis E, Kaski JC. Increased serum neopterin: amarker of coronary artery disease activity in women. Heart 2000;83:346–350.5. Braunwald E. Unstable angina. A classification. Circulation 1989;80:410–414.6. Auer J, Berent R, Lassnig E, Weber T, Eber B. Prognostic significance ofimmune activation after acute coronary syndromes. J Am Coll Cardiol 2002;39:1878.7. Suwaidi JA, Hamasaki S, Higano ST, Nishimura RA, Holmes DR Jr, LermanA. Long-term follow-up of patients with mild coronary artery disease andendothelial dysfunction. Circulation 2000;7;101:948–955.

Predicting Mortality in Patients With ST-ElevationMyocardial Infarction Treated With Primary

Percutaneous Coronary Intervention (PAMI Risk Score)

Srinivas Addala, MD, Cindy L. Grines, MD, Simon R. Dixon, MBChB,Gregg W. Stone, MD, Judith A. Boura, MS, Anthony B. Ochoa, MD,

Gregory Pellizzon, MD, William W. O’Neill, MD, and Joel K. Kahn, MD

We performed a pooled analysis of the Primary An-gioplasty in Myocardial Infarction (PAMI) trials toexamine predictors of death after primary percuta-neous coronary intervention. Using these data, wedeveloped a risk score with a range of 0 to 15 points.The PAMI risk score was found to be a strong predic-tor of late mortality. �2004 by Excerpta Medica,Inc.

(Am J Cardiol 2004;93:629–632)

Randomized clinical trials of patients with ST-segment elevation myocardial infarction

(STEMI) comparing primary percutaneous coronaryintervention (PCI) with thrombolytic therapy have

shown the superiority of PCI in terms of epicardialflow and outcome.1–3 Therefore, primary PCI has be-come the preferred reperfusion strategy for STEMI atmany centers.4 However, clinical outcome and sur-vival varies greatly according to the baseline riskprofile. Risk scores have been described for patientswith STEMI in general5–8 and for patients eligible forand treated with thrombolytic agents.9,10 Few data areavailable for patients treated with PCI.10 A reliableearly method to predict risk of death in patientspresenting with STEMI who are eligible for PCImay be valuable in selecting high-risk patients fornovel or aggressive therapies and in counselingfamilies on risk. Therefore, we analyzed a largepatient group to develop a clinical risk score thatcould be applied early to patients treated with PCIfor STEMI.

• • •We analyzed the baseline characteristics and clin-

ical outcome of 3,252 patients who underwent pri-mary PCI therapy for STEMI enrolled in various

From the William Beaumont Hospital, Royal Oak, Michigan; andLenox Hill Heart and Vascular Institute, New York, New York. Dr.Kahn’s address is: Division of Cardiology, William Beaumont Hospi-tal, 3601 West 13 Mile Road, Royal Oak, Michigan 48073. E-mail:kahn642@aol.com. Manuscript received July 16, 2003; revisedmanuscript received and accepted November 3, 2003.

629©2004 by Excerpta Medica, Inc. All rights reserved. 0002-9149/04/$–see front matterThe American Journal of Cardiology Vol. 93 March 1, 2004 doi:10.1016/j.amjcard.2003.11.036

Primary Angioplasty in Myocardial Infarction (PAMI)trials (PAMI-1 195 patients; PAMI-2 trial and registry1,100 patients; No-Surgery On Standby (No-SOS)trial and registry 499 patients; Stent PAMI trial andregistry 1,458 patients).1,11–14 In all the PAMI trials,patients were �18 years old, had ST-segment eleva-tion of �1 mm in �2 contiguous leads and/or a leftbundle branch block, and symptom onset within 12hours. Patients were excluded if thrombolytic agentswere given for the index STEMI, if they were incardiogenic shock, had a stroke within a month, wereof child bearing potential, had end-stage renal disease,or had a life expectancy from a noncardiac conditionof �1 year. Aspirin and heparin were administered toall patients in the emergency center. Institutional re-view boards at each participating center approved thePAMI protocols.

The primary end point of this study was death at 6months. Data were also calculated for in-hospital mor-tality and 1-month and 1-year intervals. Length ofhospital stay was also assessed. The pooled data setwas examined for the significance of 18 variables,taken from histories, physical examinations, or elec-trocardiograms that were collected at presentation, inpredicting mortality at 6 months using univariate lo-

gistic regression analysis. A p value of �0.05 wasconsidered significant. We then performed step-downlogistic regression analysis of mortality at 6 months.The least significant covariates were dropped at eachstep until only those covariates with a p value of�0.05 were kept. The independent predictors of mor-tality at 6 months were assigned a risk score based ontheir odds ratios, and a total risk score was thencalculated for each patient with a range of 0 to 15points (Table 1). The final model had a C statisticof 0.784 and a R2 of 0.16. After identifying very fewpatients at some scores and very little predictive dif-ference between adjacent scores, some scores weregrouped together for a final risk scale. The p valueswere calculated using a chi-square test for categoricalvariables when appropriate; otherwise Fisher’s exacttest was used. One-month mortality was available in�98% of patients, and 1-year mortality data wereavailable in 93% of patients. The PAMI risk score wasalso applied to survivors of hospitalization. The PAMIrisk score was then tested in an external data seton patients enrolled in recent PAMI trials (AIRPAMI, Local Med study).15,16 SAS version 8.0(SAS Institute, Cary, North Carolina) was used forall statistics.

Baseline characteristics of the 3,252 patients ana-lyzed are listed in Table 2. The overall in-hospitalmortality rate was 3.1%. The PAMI risk scoreshowed a strong association with mortality at 6months; there was a 33-fold graded increase in mor-tality between those with a risk score of 0 and thosewith a score of �9 (p �0.0001) (Figure 1). The riskscore was also a strong predictor of in-hospital,1-month, and 1-year mortality after PCI (Figure 2).The PAMI risk score was also applied to length ofhospital stay. The risk score had a highly significantrelation to hospital stay in patients with a score of 0whose hospitalization averaged 5.3 � 3.6 days,whereas those with a score �9 stayed an average of8.9 � 8.0 days (p � 0.0009 for trend). Moreover,among survivors of hospitalization, the PAMI riskscore continued to be a powerful predictor of mortality(p �0.0001 for trend) (Figure 3). The risk score alsopredicted mortality in the external validation set (Fig-ure 4).

• • •We used information readily available at presenta-

tion, which was derived from a large number of pa-tients treated at multiple centers with PCI for STEMI,to develop a simple bedside risk score that could beapplied to predict mortality over various time points,up to 1 year. The PAMI risk score described a widegradient of mortality risk using 6 rapidly available andindependent predictors of death. The prognostic abil-ity of the PAMI risk score was stable at multiple timepoints up to 1 year and performed well in an externaldata set of patients with STEMI who were treated withPCI.

Risk modeling for patients with acute myocardialinfarction has been described by a number of previousresearchers.5–8 The Thrombolysis In Myocardial In-farction (TIMI) risk score for STEMI in patients eli-

TABLE 1 Independent Predictors of Death Six Months AfterPrimary Percutaneous Coronary Intervention (PCI) for AcuteMyocardial Infarction

OR 95% CI p Value Risk Score

Age �75 yrs 7.4 4.9–11.2 �0.0001 7Age 65–75 yrs 2.9 1.9–4.5 �0.0001 3Killip class �1 2.2 1.5–3.2 �0.0001 2Heart rate �100 beats/min 1.8 1.3–2.5 0.0008 2Diabetes 1.7 1.1–2.4 0.0078 2Anterior myocardial infarction

or left bundle branch block1.7 1.2–2.4 0.0011 2

CI � confidence interval; OR � odds ratio.

TABLE 2 Baseline Characteristics of Patients Treated WithPrimary Percutaneous Coronary Intervention (PCI) for AcuteMyocardial Infarction

Study Population(n � 3,252)

Age (mean) 61 � 12Age �75 yrs 476 (15%)Age 65–75 yrs 835 (26%)Women 877 (27%)Diabetes mellitus 544 (17%)Systemic hypertension 1,515 (47%)Hyperlipidemia 933 (38%)Current smoker 1,317 (41%)Prior myocardial infarction 478 (15%)Prior coronary angioplasty 312 (10%)Prior angina pectoris 472 (18%)Killip class 2–4 440 (14%)Received glycoprotein IIb/IIIa inhibitors 251 (9%)Anterior infarct or left bundle branch block 1,332 (41%)Systolic blood pressure �100 mm Hg 967 (31%)Heart rate �100 beats/pmin 830 (26%)

630 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 93 MARCH 1, 2004

gible for and treated with thrombolytic agents in ran-domized trials has been described.9 In that analysis, 10baseline variables accounted for the predictive capac-ity of the multivariate model. The TIMI risk score waslater applied to an unselected community-based pop-ulation treated with thrombolytic therapy that wasentered into the National Registry of Myocardial In-farction-3 (NRMI-3).10 The risk score showed a strongprognostic capacity in patients treated with thrombo-lytic therapy (c � 0.79).

Patients with STEMI selected for PCI generallyhave fewer exclusion criteria compared with thosetreated with thrombolytic agents and may represent apatient population with different baseline characteris-tics. Furthermore, the higher efficacy of PCI in estab-lishing vessel patency and brisk epicardial flow hasresulted in lower mortality rates in randomized trials

for patients with STEMI comparedwith patients who receive thrombo-lytic therapy.1–4 Therefore, riskmodeling in PCI-treated patientsmay differ from thrombolytic-treated patients. The PAMI riskscore is the first predictor of mortal-ity developed specifically from a co-hort of patients treated with PCI forSTEMI. Compared with the patientsused to develop the TIMI risk score,patients in the PAMI cohort weremore commonly women, hyperten-sive, diabetic, and had a higherprevalence of heart rate of �100beats/min and systolic blood pres-sure of �100 mm Hg. Despite thesehigh-risk features, the in-hospitalmortality rate in the PAMI trialswas lower than that described in thecohort for the TIMI risk score andmay reflect the high rates of TIMI 3flow achieved.

The PAMI risk score identifiedage�75 years as the most powerful predictor of mor-tality, and accounted for 7 of the total of 15 pointspossible. Age �75 years was a more powerful predic-tor of mortality in PAMI group of patients treated withPCI than for those described in the TIMI risk scoreanalysis, where only 3 out of a possible 14 points wereassigned for age �75 years.9 Although there wasoverall lower mortality observed in the PAMI cohort,age appears to account for a larger proportion ofmortality risk. Heart rate, Killip class 2 to 4, anteriorSTEMI and/or left bundle branch block, and diabeteswere of generally similar predictive value in the TIMIand PAMI risk scores. Unlike the TIMI risk score,weight, hypertension, history of angina, and time totreatment were not independent predictors of mortalityrisk in the PAMI data set. This is consistent with theknown risk of life-threatening bleeding in smallerpatients17 and reduced reperfusion rates with delayedadministration of thrombolytic agents, which has notbeen observed after PCI.

The PAMI risk score is a clinically useful tool topredict risk early after patient presentation. It does notincorporate information derived in the catheterizationlaboratory during acute PCI or from ancillary nonin-vasive testing. Thus, the PAMI risk score for mortalityrisk for patients with STEMI who were treated withPCI may be rapidly applied at the bedside early afterpresentation. It may prove useful to triage and managepatients in terms of selection for standard or morenovel and aggressive catheterization laboratory-basedreperfusion therapies. Among survivors of hospital-ization, patients with higher risk scores continue to beat high risk of mortality, thus reinforcing the need foraggressive risk factor modification in patients with ahigh-risk score. It may also prove useful in designingfuture clinical trials, allowing patients with highermortality risk to be entered into studies with a smallersample size to determine benefit.

FIGURE 1. Development of the PAMI risk score using 6-month mortality from pooledPAMI data.

FIGURE 2. Application of the PAMI risk score to predict mortalityat various time intervals in patients presenting with STEMI.

BRIEF REPORTS 631

1. Grines CL, Browne KF, Marco J, Rothbaum D, Stone GW, O’Keeje J, OverlieP, Donohue B, Chelliah N, Timmis GC, et al, for The Primary Angioplasty inMyocardial Infarction Study Group. A comparison of immediate angioplasty withthrombolytic therapy for acute myocardial infarction. N Engl J Med 1993;328:673–679.2. Gibbons RJ, Holmes DR, Reeder GS, Bailey KR, Hopfenspirger MR, GershBJ. Immediate angioplasty compared with the administration of a thrombolyticagent followed by conservative treatment for acute myocardial infarction. N EnglJ Med 1993;328:685–691.

3. Zijlstra F, de Boer MJ, Hoorntje JC, Reiffers S,Reiber JH, Suryapranata H. A comparison of immediateangioplasty with intravenous streptokinase in acutemyocardial infarction. N Engl J Med 1993;328:680–684.4. DeGeare VS, Dangas G, Stone GW, Grines CL.Interventional procedures in acute myocardial infarc-tion. Am Heart J 2001;141:15–24.5. Killip T, Kimball JT. Treatment of myocardial in-farction in coronary care unit: a two-year experiencewith 250 patients. Am J Cardiol 1967;20:457–464.6. Norris RM, Brandt RW, Caughey DE, Lee AJ, ScottPJ. A new coronary prognostic index. Lancet 1969;1:1274–1278.7. Peel AAF, Semple T, Wenge I, Lancaster MW, DallJLG. A coronary prognostic index for grading severityof infarction. Br Heart J 1962;24:745–760.8. Jacobs DR, Kroenke C, Crow R, Deshpande M, GuDF, Gatewood L, Blackburn H. PREDICT: a simplerisk score for clinical severity and long term prognosisafter hospitalization for acute myocardial infarction orunstable angina. Circulation 1999;100:599–607.9. Morrow DA, Antman EM, Chalesworth A, Cairns R,Murphy SA, De Lemos JA, Giugliano RP, McCabe CH,Braunwauld E. TIMI risk score for ST elevation myo-cardial infarction: a convenient, bedside, clinical scorefor risk assessment at presentation. Circulation 2000;102:2031–2037.10. Morrow DA, Antman EM, Person L, De Lemos JA,Cannon CP, Giugliano RP, McCabe CH, Baron HV,Braunwauld E. Application of TIMI risk score for STelevation myocardial infarction in the National Registryof Myocardial Infarction 3. JAMA 2001;286:1356–1359.11. Grines CL, Cox DA, Stone GW, Garcia E, MattosLA, Giambartolomei A, Brodie BR, Madonna O, Eij-gelshoven M, Lansky AJ, O’Neill WW, Morice MC, forthe Stent Primary Angioplasty in Myocardial InfarctionStudy Group. Coronary angioplasty with or withoutstent implantation for acute myocardial infarction.N Engl J Med 1999;341:1949–1956.12. Stone GW, Marsalese D, Brodie BR, Griffin JJ,Donohue B, Costantini C, Balestrini C, Wharton T,Esente P, Spain M, et al. A prospective, randomizedevaluation of prophylactic intraaortic balloon counter-pulsation in high risk patients with acute myocardialinfarction treated with primary angioplasty. J Am CollCardiol 1997;29:1459–1467.13. Grines LL, Wharton T, Balestrini C, et al. Shouldhigh risk acute myocardial infarction patients admittedto non surgical hospitals be transferred for primaryPTCA or receive it on-site? (abstr). Circulation 2000;102:II-386.14. Grines CL, Marsalese DL, Brodie B, Griffin J,Donohue B, Constantini CR, Balestrini C, Stone G,Wharton T, Esente P, et al. Safety and cost effectivenessof early discharge after primary angioplasty in low riskpatients with acute myocardial infarction. PAMI-II In-vestigation of Primary Angioplasty in Myocardial In-farction. J Am Coll Cardiol 1998;31:967–972.15. Esente P, Kaplan AV, Ford JK, Martin JL, Ayres M,Kosinski EJ, Lasorda DM, Graham M, Gallant P,Grines LL, Grines CL. Local intramural heparin deliv-

ery during primary angioplasty for acute myocardial infarction: results of thelocal PAMI pilot study. Cathet Cardiovasc Interv 1999;47:237–242.16. Grines CL, Westerhausen DR Jr, Grines LL, Hanlon JT, Logemann TL,Niemela M, Weaver WD, Graham M, Boura J, O’Neill WW, Balestrini C. Arandomized trial of transfer for primary angioplasty versus on-site thrombolysisin patients with high-risk myocardial infarction: The Air Primary Angioplasty inMyocardial Infarction Study. J Am Coll Cardiol 2002;39:1713–1719.17. Reeder GS, Gersh BJ. Modern management of acute myocardial infarction.Curr Probl Cardiol 2000;25:677–782.

FIGURE 3. Use of the PAMI risk score to predict mortality among hospital survivors ofSTEMI.

FIGURE 4. Verification of the PAMI risk score in the AIR-PAMI and Local Med trials.

632 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 93 MARCH 1, 2004