Heart Rate AdJustment of the Time-hostage Identification of Coronary Heart Rate-Adjusted ST S
ST Segment Integral:
PETER
PAUL
M. OKIW,
New Yorlt. NIW
MD, FACC.
GEOFFREY
BERGMAN.
MD.
KLIGFIELD.
MD.
FACC
York
To assessthe etTprt of heart rate adjustment of the magnkude al the ST tntograt (ST+IR integral) on exercise ,PS,performance, the exerctre ekctrwrdtwgram
IECG) al 50 dintally
“armat subjects
and IW nattents with ki,mvn or suswcted c”r”nsrv arterv diiease was mm&d. At matc’,ed specifiriiy 01968 Hith.stand&d ECG criteria (20.1 IV al additional korbantat or downrtuping ST ~ment daprcszion,, ar “,,sdj”r:cd ST in,er,ra, partilian uf 16 pV.r tdenti6ed coronary disease in the 100 oalienls ni,b knawn or wspected dikeaw with a sensitivity of “nty 410. 3 ratw significantly
tower than the 59% sensitivity of standard ECG
criteria tp < 0.01) cad the 65% sensitivity al an ST depression partitttiriancl 130 pV tp < O.Wtt. Wawever, test prlormance of the ST integral was gea,,y improved by simple hear, rate adjustmonf: at P matched spwiRctty of %%, an ST.HR integral partition of 0.154 *V-rlbaat per min identified cmonary d&w in the tW patients with P renri-
Standard exercise elec,rocard;“graphic
(ECCI cri,eria baaed
“n the magnitude. time course and configuraion of ST segment depression have poor sensiwity and wboptmxai specificity for the detection of c”r”nary artery disease Estimation of the total area of repolariration abnormalily by
CI.!).
the time-voltage
inlegral
of ST regmm
depression t,be ST
integral) has been proposed to impmvs ,br prrformancc of criteria based 0.) the magnitude of ST depression a, a singIs measurement pain, for rhe exercire ECC identifica,iw “f cxonary obs,ru\:‘.,r ‘3). However. accurncy of ,he ST inlrgral for the aetec,,“n of cornnary dlrease has \aricd in subsequcn, repons (4-X). The magnitude of the ST m,egrai ha been shown 1” c”nela,e
with
changes
in hear,
rate during
sxerci~e
in
normal subjects (5.9) and t” improve tbr delcction “t twonary disease when compared wilh !he magnitude of the ST inlrgral a, similar heart mtes in normal subJecls (9). Bccus! hear, rate adjusrmen, of measured ST wgment deprrs,ion
an ~mpruvr tes, accuracy d,rca,e ,,“-21,. Ihe ““r~ow
for idenrifxaon “f c”r”nxy of this stud” was to assess the
abilily of hex, raw adj&men, 1” improbe the performance of ihe ST mtesral !n de,ec,mp. wnifican, c”r”nan oWruclion: Ihs ST &ml. alone and after hear, rhlr adjwmen!. wn c”mparc,, wth standard and recently described hew, n,e-adw,ed dcpresiion.
cnteria bawd “n measurement
of ST re%men,
Methods Study patients. wbjec,s and p&n&
One hundred fifty conseculively
studled
who me, the criteria t” .e described were
entered m,o one of three clinical
gmupa,
one wth
a lo*
likehhood of c”r”nxy disease and two wh a high (or ceniinl bkcbhood of dlwase. All rubjccts and patienls were assigned 1” one of the clmical groupa befurc cwminadon of the exercise ECG and no s,udy was excluded afw group assign. men,. None of ,he normal subjects “7 patie”,s
I” this rep””
wre included in the groups from which test criteria for ihe ST segment-hear, rate IST-HR) slope and ST.HR mdcx wcrc dcrwed and ,he,r perf”rmance for the ,den,,ficarion of COT”. nary dnease was described (191. Howver. a rubret of thir population was mcluded in a recent wdy (20) that examined
_
_
the effect of :he limme of ST secment measurement relative lo the J point on test performance of ST depression criteria. Normat subiects IGroup I). There were 50 normal subjecx 40 men aid IO worn& with a meanage (+SD)of48 * 10 yeara Them were no voluntet: objects in this group, and each wa\ referred bv a ohvsician for exercise electrocardiography ai part of a comprehensive screening evaluation or as a precautionary evaluatton before beginning an exercise program. All subjects were free of chest pain. had no history of cardiac direasr. were not taking medications. had a normal cardiac physical examination. normal blood pressure and normal rest ECG. In addition. all subjects were free of chest pain during treadmill exercise. On the basis of data of Diamond and coworkers (22.23). the age- and genderadjuatrd likelihood ofcoronary disease in lhts asympromat~ group can be estimaled as ~0.05. Patients with clinical angina (Group 2). There were 60 patients with clinical angina in this gruup of 35 men and 25 women with a mean age of 62 t 9 years. Each was referred bv a ohvsician for exercise electrocardioeraohv to evaluate ciinically stable retrostcrnal chest discorr;fo;t ihat was conrirtently provoked by exertion and relieved by rest. In addition. all aatients in this crouo develooed tvoical chest discomfort d&ing treadmill exe&. Patients wi;h left bundie branch block or mvocardial infarction within 8 weeks were not included. Five’patients had rest ECG evidence of a previous Q uwe myocardial mfarction. and six patients had KG evidence of left ventricular hypertrophy. There were 33 patient> who were not takine medications: amonr tbc mmkning 27 patients. 14 were-taking a beta-adr&rgic blocking drug, 10 were taking nitrates and I3 were taking a calcium channel blocker at the lime of exercise testing. The agr- and gender-adjusted likelihood of coronary dtsease in this erouo can be estimated as ~0.93 (22.23). Patients with roronsry disease (Group 3). There were 40 patients with coronaw disease proveh by catheterization. This group included 33 men and 1 women with a mean age of 59 i 10 years. All patienls in this group had stable, effort-related chest pan thal was relieved by rebt. and sach patient was referred by a physician for exercise testmg and coronary arteriaglaphy. Patients with left hundle branch block or recent myocardial infarction were excluded Three oatients had ECG evidence of orevious Q wave myocardtal infarction, and seven had evibence of left ventrtcular hypertrophy. Twelve patients were taktrtg no medications; among thr rcmaming 28 patrots. I9 were taking a beta-bloc:,er. I? were taking nitrates and 17 were takmg B calcium channel blacking drug P the lime of exercise evaluation. Exercise electroeardineraphv. Exercise EC& were performed on a treadmtll with’ (se of a Quinton lns~ru~enl QSOOOexercise system modified by the addmon of a bipolar lead CM,. AU patients exercised according to the Carncll p~Xoio1 (24.25) and an age-adjusted target heart rate was sought as the exercise end point for all studies. In addition. computer-calculated ST segment amplitude, measured to the I
.
_
.I
I
catb~terizatian.provcd
nearest IO UV at a point 60 ms after the I point with the end of the PR segment &a reference. was obtained in each lead after each stage of exercise and at peak exercise; the accuracy of this measurement has been previously validated in our laboratory (261. Exer&e tests were evaluated with use of standard ECG criteria based on the measured amoun. of ST segment depression on the peak exercise ECG (19). The test was considered positive in the presence of 0.1 mV (100 pVv) of additional horizontal or dawnslapir@ ST segment depression. It was considered negative in the presence of any amount of upsloping ST segment depression or
rapher
using
calipers Without knowledge
of clinical or cwr-
cise test dam as previously reported in detail (IS.191 Degree of stenosis wa, defined as the grates, percsn, reduction of diame,er in any view as compared wh the nearer, norex segment. For clwificallon of the numher of obstructed coronary arteries. d&arc was conwdered significant when 50% iuminal obstrucdon WIB~prsrcnt. Lrf, main coronary ar,ery narrowing ~50% uar scored as the equivalent of ,wo-vessel disease (15.19). Accordinr! ,o thrsc chteria there were 3 patients wth one-vcbbel d&are. IO ~a,,en,s with two-vessel disease and 27 oe,,en,s with ,hreevessel disease. Five patiems had left main coronary d~seune. including one with addi,ional two-vessel disease and four with additional three-vessel disease. Ma en&is sod statistical methods. Defin~tmnr of test sensitivitv and soecificitv conform lo slandard use 129). Test specifici,; for each method was defined in the SO clinically normal subjects (Group I). Because of ,he highly referred nature of our angiography popularion. and the known and poential effecls of refernl bias on selection for angiography (30.31). test sensitivity was assessed in all IW patienls wilh known or suspected coronary disease IGroups 2 and 31. Comparison uf ,es, sensitivity of standard ,es, criteria. ST ssgnent depression ii: peak cxercibe. the ST-HRIndex. the ST-HRslope and senple and heart ra,e+djur,cd ST II,:& criteriawasperformed usicgpartidona for eachcriterion that produced a matched specihcily of Yh% with tha, found for standard criteria. using McNemar’% modification of the chi-square melhod for pdired proportionr. Because benbili!‘. ily and specificlly al a ,a, arc dependent on Ihe parWon value chosen for test positivily. tcs, accuracy of rhe ST integral, ,he ST-HR inbegral. ST segmen, deprcwon. ,he ST-HR index and ,hs ST-HR &pr nere also compared using receiver-opera,ing characteris,ic curve analysis. The% curves were compared statistically by meanr of a uni\drMc Z score ,a, of Ihe difference beween Ihe area under I~O CurYes (32,. Recauar ,he accuracy of both rmndard and hear, rateadjusted ST segmen, criteria hsre been shown lo vary with the time of ST depresston meawrement rrla~e 10 ihe J point (20.21). ,he excrcw te\r d&a ofall 50 subjects and IOU
luminal
dnd 80 mi after the J pain, in each isad al peak exercise (201. The rrla:iun of end-cxerclse ST inleoral measuremen,~ 10 ST \egmen, dcprenion rneasuremimc~l 0. 60 and 80 ms alicr rhs J point %wb determined by linear regess;on analysir. Coeffxienrr of correlation (4 and clooec were calculated for rach r&don. and ewe compared rtauWally by using a two-laded Fisher‘, 2 transformation 031. Mean ialuet for all InJings are reported wilh the scandard de\wmn a\ rhe index of dispcnion. Comparison of mean YBIUSE betncen and among groups w,s performed by one-say an&~> of varianw with porthnc lesliug of individual grwp ddTerences by ScheWs merhod. Subgap propor,icnc riere compared by chi-square analysir with ~orr~uon ior ~~mu!,y. For alI compar~ons. a p value 4.05 ~a* reqwed for rejection of Ihe null hypmhesls.
Gr up charaeterislics and exercise performance [Table 1). Normal uh]ec,y here younger, exercised longer. achieved a higher hear, rxe and greater pent sgrwlic blood prrssure and had B !owr mean ST-HR ~Iox than did patiems wi,h ,yoirai angma (II pe.:w~,x wi,h proved coronary aRery d&e A&g p.&~Is wilh known or suspected coronary d&se. patirn,s referred for angmgraphy and found IO have coronary dneaw iGroup 3) were more commonly men. and achwed lower values for maximal hean raw. peak systolic blood pre\,ure and percent of ,argct hean rate and a higher vdluc for mean ST-tiR slow than did patients with fypxal angina IGroup 2) Howwx. sender d,s,r,buhon was simda! in &ally normal subjcrls and in pa,~n,s s,,h coronary dnease at anpmcnphy. snd palien,s uilh coronary disease uere wudar 10 pdlients ailh cbmcal angina with revec, 10 age and ewrcirc duratxon. Among all patien,~ and subjects, lhere was a strong r~la,~m oeween the maxImal ST integral and maximal ST ,egmen, deprewon measured a, hO III, &i the J polnf (I = 0.97. p < UNI1). In con,r?~,. Ihe correlalion was significandy lower bcween ,be maximal ST inlegral and exercw-
induced ST depression rceasured at the J point (r = 0.83. Q 4 0.001) or at 80 ,ns after the J point (r = 0.92. Q < 0.00,; each Q < 0.001 vs. correlation with ST depression at 60 ms after the J point). There wa\ a weak hot .tatirtically signifumt correlauon between the ST integral ad the change in heart rote dunng exercise among climcally normal subjects (r = 0.30. p < 0.05: elope = 0.084) and a similar weak correlal~on among patients with known or aspected corenary disease (r = 0.23. p < 0.05: slope = 0.107. p = NS VP. clinically normal whjectr) Identification of coronary artery disease by ST inlegal and standard ECC criteria (Table 21. A positive exercise ECG according to standard criteria had a specificity of 96% (48 of 50) in normal suhjccts and a sensitivity of 59% (59 of IUD) in patients with known or swpected coronary disease. At a matched specificity of 96%, an ST in:egral partition of 16 &V-s had a Fensitivity of only 41% (41 of IW), which was significantly lower than the 59% sensdivtty of standard criterialp 4 O.Ol)and the bS%h(65of IW) sensitivitvofan ST deprerric- partition of 130 pV (p < 0.001). R&dive test performance of the ST integral was -01 improved by using criteria derived from orevious studies 13.6.34.351: an ST integral partitmn of 7.s’~V.s had a specificity of 66% (33 of SO1 and a sensitivity of 86% (86 of 100l. whereas an ST integral partition of IO pV-s had a specificity of 82% (41 of SO) and B sensitivity of 69% (69 of 100); each sensitivity war ri~mficantly lower than Ihe 96% I96 of 100. Q < &lC51 renuivity ad 80% (80 of 100, Q < 0.005) sensitivity of mmple ST depression at matched specificities. Comparison of receiver-operating characterislic curves confirmed the superior overall test performance of simple ST depression alone. measured 60 ms aRer the J pomt relative to the ST mtegral (Fig. I). Heart rate adjustment of the ST mtegrat and ST seg,,,en, depression ITablc 21. The effects of heart rate adjustment of the ST integral and the magnitude of ST segment depression on lest perfommnce for the identification of coronary artery disease werr: examined (Fig. I). Test pelformance of ST integral measurement was improved by heart rate correction: at a matched specificity of96% in normal subject,, the rate-corrected ST-HR integral partition of 0.154 fiV.slbeat pw min identified coronary disease with a sensmvity of 90%
190 of 100). which was significantly higher than the 41% sensitivity of the simple ST integrav alone and the 65% sensitivitv of ST deoression alone (each o < 0.001). Comparison if receiver-operating charac&tic curves confirmed the superior overall Tess performance of the ST-HR integral as compared with that ojeither the ST integral alone or ST depression (Fig. I I. At matched specificity of%%, test sensitivity of an ST-HR slope partition of 2.08 PVlbeat per min (93%. 93 of IOOI and a simple ST-HR index partition of 1.41 PVibeat per min (91%. 91 of 100) were similar to that of the ST-HR inlegral leach p = NS). Although comparison of receiver-operating characteristic curves for the entire range of test specificities demonstrated no significant difference in overall performance of the ST-HR integral or ST-HR slop or ST-HR index for the identification of coronary disease (Fig. I). comparison of partial receiver-operating characteristic curve areas over the clinically relevant range of specilicity from 90% to IWO/o revealed a trend toward superior
Figure 1. Receiver-opraling characteristic CUN~D comparing the overall pelformanee of the timbvobage inlrgrat of ST segment depressionlthe ST integral). ST segment depression, Ihe ST segment~heartrate IST-HR) integral. the ST-M index and Ihe ST.HR slope for the identificationof coronary artery disease. l Q c 0.001 Y~US ST segment depression. ST.HR integral. ST-HR index and ST-HR slope: bp < o.WI ~crsos ST integral. ST-HR integral. ST-HR index and ST-HR slope.
performance of ,he ST.HR slope lpar,~al aiea 0 94 w 0 Hc. ,I = 0.06) and the ST-HR index Iparlial arca 0.92 ‘II 0 Xi, p = 0.W) a? compared wh ,he ST-HR intcsrral Discussion These dam demonstrate rha, the ST integral alone. wlh. ou, hear, rate adios,r,,en,. can degrade the performance of the exercise ECG relalive 10 standxd ST wgmcn, dews+n and configura!ian cntwa. However. simple hear; rate adjustmen, of the ST ~m,egral vgnificnntly improve\ I,\ performance for the Identification of coronary ar,er> dw ease. wnh accuracy approaching lha, ol borh \~mple and regression-based hear, vale-adjusted ST ~egmenl crilewa. These findings provide further strong ruppor, lor the value of hear, rate adjublmen, of repolanzauon msasurrmentr in diagnostic exercise elec,rocardlognphy. ST Intcp.ml and ST wgnenl dep:esrion. The prewn, dala support previous finding. (5.6) ,hn, ST m,egral cntrria alone do no, improve the performance of ST regmen, deprewun criteria for ldenlification of coronary obwuclwn during exercise. Simoons I51 reporled that the ST m,egml in Frank lead X had a rensilivily of 58% and a specificity of 94% for ,he delection of coronarv disease. YBIUC~ ,ha, are slirhllv lower than the 63% &i,ivi,y and Y5Si hpeciBci,y o? St depression measured a, 60 ms afler ,he J porn,,, S~mdxly. Ascoop e, al. (61 found Ihat sensilivily of the ST m,egrel varied berween 32% and 49% a, specificilic$ prrrtrr than WZ. whereas simole ST deoression criteria had ~ensdw,w of 44% to 61% at’similarly’high specificilies. Uwg sligh,ly different ST integrdl and ST depression meawrement !echniques based on lime af,er the peak of Ihe R wave, Sketch et al. (7) also found that the ST integnl provided no impmvemen, in accuracy for the detection of corop’I-- anery dwarc over visual as~essmen, of the magnitude of ST regmen, depression. Hen11 late adjustment of ST i,degraI and ST agmen, depression. Because ,he ST in,egrdl was found ,n b?>!e :: linear correlation with increxing kern rate during exercise in normal subjects 01. Snoons (5: adjusted the magm!udr of Ihe ST in,egral ai p?ak exerciw for ,he magimde of the Sl in&gal in normal subJec[r a, similar peak exerwe hear, rates. ‘This method OfadJuslmg Ihe S ior changes in hex, r~tr improved ienrilivity of ,he ST inlrgral for coronary disease from 58% io 81% wilh easemially no 1os5 of specificily. bu, doffers dis,mc,ly from ,hc hear, mle correc,ion method of the oresen, sIudv_ The orexn, method of simple hear, rale (Hd) correclion of th; ST integral iq analogous to tha, used for the simple ST-HR mdey and resuks in similarly nnproved test performance rela,we 10 unadjusled ST integral mearuremems. Although srnsmvily of ,he S’PHR inregral was nearly idcmical lo tha, of the ST-HR index and more complex ST-HR slope zt marched specificky of 96%. and although overall receiver-opcralmp characterwic corves were similar for the lhree methods. comparison of test performaccs belueen ,he hear, n,e-
adJw,ed ST deors,,ion and ST mregral cn,ena over the chmcall\ reier;m, ranae of ,a, ,oe⁣,\ from 90% 10 ,Ml’;i dem.~nrirawJ il ,renb,oward i&enor &ay of crwria bawd an ST ~rgmen, depression rather than on the ST lnlegral ‘The differences in ,es performance beween ST mrepral and Sl deprrwon measurements may be in par, accouxrd fur b) the mcor~orauoo of J pain, and 80 ms af,er rhe J WI, ST wmsn, mraruremen,, mto de,ermma,ion of ,he ST mregrdl IZ1).21.?71. Recent sedies have demonsrratcd rha, ST dep~ewon measured a, ,he J point 12021) and a, 80 ms dfter Ihc J pain1 130) significantly reduce ,hc performance of bolh vmplc and bear, rae-adjusted ST segment deprewon cnterril for ihc dc,rc,ion of coronar) dgceaw. Although ,he ET m,egrdl was found to correlate strongly ai,h ST meawremeo,\ mrde a, 60 mr after the J point. Ihe m,ega,ion of ST xgmen, deprasion beraeen 0 and 80 ms after the J point appeari 10 dscre.ae the d,scrimma,ory power of the ST ~ntsgrai me!hod rela,~ve IO ST segment dsprerrwn alone ibhcn mcawred ai 60 ms afler ,he J qoint. Furrher wdy in nr‘ce~~ary 10 doss ahelher redefioitions of the ST mapral tha, exclude ST measurementc made I, the J point and 80 ms afw ihe J porn,, m~prove Ihe perfomunce of this method. Clinjcal im,dica,i@ns. Several larce coooerative studies have demon&wd tha, the ST in,e&l &hod in ConJunG uon uilh t34.351 or independen, of 136.37) a porilivs ,a, responw by \?andard ECG eritena can s,a,ify the risk of future cardiac evenb. The prcsen, study. taken together w,h the recent finding tha, simple hear, rate adjustmen, of ST \egnpmen,deprrrsion more accurately stratifieF Ihe rirk of Rure coronary events than do ,,andard ECG cnrena (381. w_gg:c‘0 that the dngnoruc and prognostic aecumcy of the ST m,egr,d method may be improved by hear, rate correcbon of the mapni,ude of the end-ewrcise ST intpiral. The simplicily of ,ha type of hean rate correction lends i&elf to potential rclrorpective analysts of standard da,‘, bases that have meawed the ST mtegml.
1integral IuBCCOUI)I
dia. IR. Deconick F. ed. lnfo,mationP,uccEringinMcdiFal Imaging. The Hagur: Martinus NiihotT. 1984.412-45. 33. Za, IH. Biortalisliul Analysis. 2nd ed. En&wood CliRr. NJ. Preate. Hall. 1984:316-7. 01. Ekelund LG. Karen JM. et PI. Predictive value of the 34. Gordon exerc~re lolerance 1.51 for mortality in Nonh Amencan men: Ihe Lipid Rerearch Clinics Momhly FoIIoK-UP Study. Circulation IP86:74252-61. 15 EkeIt nd LG. Suchindrnn CM. McMahon RP. et al. Coronary hart dlsrasr morbidity and mvnalily in hywchokwotmic men Qrrdicled from an exercise lea: the Lipid Research Clinics Corooary Mmary Pievenlion Trial. J Am Call Cardiol 1559;14:556-63.
38.