390

Activation of Neurohumoral Systems in Postinfarction LeftVentricular DysfunctionJEAN L. ROULEAU, MD, FACC, JACQUES DE CHAMPLAIN, MD, PHD, MARC KLEIN, MD,

DANIEL BICHET, MD, LEMUEL MOYE, MD, PHD, MILTON PACKER, MD, FACC,

GILLES R. DAGENAIS, MD, FACC, BRUCE SUSSEX, MD, FACC,

J. MALCOLM ARNOLD, MD, FACC, FRANCOIS SESTIER, MD, PHD, FACC,

JOHN O. PARKER, MD, FACC, PATRICIA McEWAN, MD, FACC,

VICTORIA BERNSTEIN, MD, T . EDWARD CUDDY, MD, FACC,

GERVASIO LAMAS, MD, FACC, STEPHEN S. GOTTLIEB, MD, FACC, JOHN McCANS, MD,

CLAUDE NADEAU, MD, FACC, FRANCOIS DELAGE, MD, PEGGY HAMM, PHD,

MARC A. PFEFFER, MD, PHD, FACC'"

Objectives. This study was conducted to evaluate the degree ofneurohumoral activation around the time of hospital dischargealter myocardial infarction.

Background. Because pharmacologic interventions that blockthe effects of neurohumoral activation improve the prognosis afterinfarction, we hypothesized that widespread neurohumoral acti-vation persists in some patients until at least the time of hospitaldischarge and that the determinants of activation very from onesystem to another .

Methods. Five hundred nineteen patients in the Survival andVentricular Enlargement Study (SAVE) bad plasma neurobor-mones measured before randomization at a mean of 12 days afterInfarction. All patients had left ventricular dysfunction (leftventricular ejection fraction s40%) but no overt heart failure .

Results. Although all neurohormones except epinephrine wereincreased compared with values In age-matched control subjects,plasma norepinephrine (301 t 193 vs . 2,22 :t 87 pg/ml, p < 0.001),

Activation of all neurohumoral systems occurs in the major-ity of patients during the 1st 72 h of an acute myocardialinfarction (1) . The degree of neurohumoral activation ap-pears to be largely related to the degree of left ventriculardysfunction and is amplified by the use of diuretic drugs . ByI week after infarction, previous studies (1) suggest thatcirculating levels of all neurohormones return to normal .The only exception appears to be patients who have overt

*A complete list of authors' affiliations appears in the Appendix . "hisstudy was supported by the Medical Research Council of Canada, Ottawa,Ontario, Canada and by Bristol-Myers-Squibb Laboratories, Montreal, Que-bec, Canada .

Manuscript received June 18, 1992 ; revised manuscript received Decem-ber 7,1992. accepted February 1, 1993 .

Address for conesrande,me: Jean L . Rouleau, MD, Service de Cardiol-ogic, Universitd de Sherbroke, Sherbrooke, Quebec, 11H 5N4 Canada .

®1993 by the American College of Cardiology

JACC Vol . 22, No . 2August 1993 :390-5

resin activity (3.0 t 3.7 vs. 1.2 t 1.2 ng/nil per h, p < 0 .001),arginine vasopressin (1 .9 t 6.9 vs. 0.7 t 0.3 ply/ml, p < 0.001)and atrial natriuretic peptide (75 ± 75 vs. 21 t 9 pg/ml, p <0.001) values ranged from normal to very high, indicating a widespectrum of neurohumoral activation. Activation of one systemdid not correlate with activation of another. The clinical andlaboratory variables rest closely associated with neurohuinoraiactivation were Killip dam, left ventricular ejection fraction, ageand use of diuretic drugs . The association between neurohumoralactivnicn and clinical and laboratory variables varied from oneneurohormone to another .

Conclusions. Neurohuntoral activation occurs in a significantproportion of patients at the time of hospital discharge afterinfarction. Which seurohormote is activated and which clinicaland laboratory variables determine this activation vary from oneneurobormone to another.

(J Am Coil Cardiol 1993;22:390-8)

heart failure or are receiving diuretic drugs, in whomplasma resin activity is increased, and patients with leftventricular dysfunction, in whom atria) natriuretic peptidelevels remain elevated . This relative lack of sustained neu-rohumoral activation appears to be at odds with data fromclinical studies documenting beneficial effects when treat-ment with beta-adrenergic blocking agents or angiotensin-converting enzyme inhibitors is started at the time of hospi-tal discharge .

Patients with left ventricular dysfunction but no overtheart failure after myocardial infarction are a high risksubgroup in which the use of therapeutic interventions thatblock the effects of neurohumoral activation are known toprolong survival and decrease cardiovascular morbidity(2,3) . The relative efficacy of angiotensin-converting enzymeinhibition in the Survival and Ventricular Enlargement study(SAVE) (3) in patients with a left ventricular ejection frac-

0735-1097/931$6.00

JACC Vol. 22, No . 2August 1993 :390-8

tion >_3l% and the lack of efficacy in this subgroup in theStudy of Left Ventricular Dysfunction (SOLVD) (4,5) wouldargue in favour of specific pathophysiologically importantconsequences of neurohumoral activation during the earlypostinfarction period. Also, the use of beta-blockers in thissubgroup of postinfarction patients has been found to beparticularly useful in preventing further morbidity and mor-tality, whereas the benefits of beta-blockers on survival ofpatients with chronic heart failure remains unproved (6) .Further studies to document the degree and range of neuro-humoral activati: n, previously thought to be generally ab-sent in these patients, as well as the factors driving thisactivation, are thus relevant and may be useful in identifyinpatients in this subgroup at particular risk of subsequentmorbidity or mortality .

In this study, 519 patients with left ventricular dysfunc-tion (left ventricular ejection fraction X40%) but withoutovert heart failure had circulatory neurohumoral levels mea-sured a mean of 12 days after myocardial infarction. Thedegree of neurohumoral activation was then correlated withthe clinical and laboratory characteristics of these patients .Our hypothesis was that widespread neurohumoral activa-tion persists in some patients until at least the time ofhospital discharge after infarction and that the determinantsof activation vary from one neurohumoral system to an-other. Our objectives were thus to better understand theextent of and the pathophysiologic mechanisms contributingto activation of a given neurohormone in the early (1 to 2weeks) postinfarction period .

Methods

Study patients. The 519 patients included in this studywere drawn from the Survival and Ventricular Enlargement(SAVE) study. That study is a randomized multicenter trialtesting the hypothesis that the angiotensin-converting en-zyme inhibitor captopril started 3 to 16 days after infarctionimproves survival and reduces the incidence of a majordeterioration in ejection fraction (?9% units) in survivors ofacute myocardial infarction. The method used to estimateleft ventricular ejection fraction was equilibrium multigatedradionuclide ventriculography performed after the labelingof red blood cells by technetium-99m. The SAVE radionu-elide ventricutography-left ventricular ejection fraction corelaboratory reviewed the findings of radionuclide ventriculog-raphy in patients randomly selected by the SAVE datacoordinating center at baseline, permitting an assessment ofradionuclide ventriculography and left ventricular ejectionfraction quality and confirming the presence of a low ejectionfraction . Only patients with left ventricular dysfunction (de-fined as a left ventricular ejection fraction s40% at the clinicalsites) and no overt heart failure or heart failure responsive todigitalis or diuretic drugs (defined as lack of pulmonary rales onphysical examination at the time of randomization despite theuse of digoxin or diuretic drugs, or both) were included in thestudy. Exclusion criteria included contraindications to capto-

ROULEAU ET AL .

391NEUROHUMORAL ACTIVATION AFTER INFARCTION

pril and concurrent medical problems such as renal insuffi-ciency (creatinine >2 .5 mgldl), severe valvular disease, refrac-tory hypertension, malignancy or other conditions thought tolimit survival, geographic problems or inability to give in-formed consent (3) . Patients with ischemic symptoms or with amarkedly positive exercise test were also excluded from thestudy unless cardiac catheterization and appropriate revascu-larization procedures were conducted before randomization .All patients gave written informed consent agreeing to haveextra blood samples drawn .

Ten Canadian and three American centers participated inthis substudy . Patient recruitment began on November 27,1987 and ended on January 28, 1990 . Patients were recruiteda mean of 12 days (median 12) after myocardial infarction .Their average age was 60 ± 11 years and there were 93women and 426 men . Thirty-four percent had a history ofhypertension, 19% a history of diabetes and 35% a history ofprior myocardial infarction .

The mean maximal increase in creatine kinase during theinfarction was 13 .4 times the upper limit of normal . TheKillip class during the acute phase of the myocardial infarc-tion was I in 57%, II in 39%, III in 8% and IV in 1%, whereKillip class I = no rates and no third heart sound, Killip classII = rates in 550% of the lung field or third heart sound ;Killip class III = rates in >50% of the lung field (pulmonaryedema); and Killip class IV = cardiogenic shock (clinicallydetermined) . The mean left ventricular ejection fraction wasof 31 ± 7% . Thirty-one percent of patients had receivedthrombolytic agents and 70 patients (3%) underwent a revas-cularization procedure (coronary artery bypass surgery in 31and coronary angioplasty in 39) during the interval betweentheir myocardial infarction and neurohumoral testing . Pa-tients who had a revascularization procedure had clinical,laboratory and neurohumoral values similar to those of otherpatients and were thus considered together with the rest ofthe group. At the time of neurohumoral sampling, 35% ofpatients were receiving diuretic drugs, 33% were receivingbeta-blockers, 43% were receiving nitrates, 28% were re-ceiving digoxin, 25% were receiving calcium channel block-ers and no patients were receiving angiotensin-convertingenzyme inhibitors or other vasodilators .

Thirty-eight age-matched people (57 = 7 years, l` womenand 24 men) were used as control subjects . These controlsubjects had no known disease and were receiving no medica-tion. Control subjects were volunteers wort ;^a -n the variousparticipating centers, and neurohumoral sans_-. ' were ob-tained and handled in a fashion similar to those of the SAVEpatients .

Neurohumoral measurements. After an overnight fast, anindwelling venous catheter was inserted and the patient waspermitted to rest in the supine position for 30 min . Blood wassampled for plasma norepinephrine, epinephrine, dopamine,renin activity, aldosterone, arginine vasopressin and atrialnatriuretic peptide . Blood was also drawn for measurementof serum electrolytes, blood urea nitrogen, creatinine, glu-cose and osmolality to verify sample quality . Mean arterial

392 ROULEAU ET AL .NEUROHUMORAL ACTIVATION AFTER INFARCTION

pressure (84 ± 11 nm Hg) and heart rate (77 ± 19 beats/min)were also measured. Samples were then frozen at -80°C andsent monthly to the central laboratory (H6pital du Sacrd-Coeur de Montreal) packed in dry ice . All sample transfersbetween peripheral and central lanoratories took <24 h . Nosample was delayed, hemolyzed or thawed . Neurohormoneswere measured monthly as soon as received to minimize theinevitable loss of atria) natriuretic peptide during storage (7) .The methods used to treasure neurohumoral levels havebeen previously described (8) .

Statistical analyses . All statistical analysis was performedusing the Statistical Analysis System (SAS) . Participantswho took part in the substudy of SAVE had data collectedon the standarG SAVE study forms ; control subjects haddata collected separately. The SAVE substudy participantshad descriptive statistics measures of demographic andclinical information. Participants in the control group hadtheir mean values for neurohormones (with standard devia-tion) computed. The neurohormones of the SAVE partici-pants in the substudy were compared with those in thecontrol subjects. Unpaired t tests were used to evaluate thestatistical significance of the differences in the mean valuesof the neurohormones in these two groups .

Subsequent analyses to identify relations between neuro-hormonal levels at the time of entry to SAVE and baselinedemographic and clinical variables were restricted to thoseparticipants of SAVE involved in the substudy . The univari-ate analyses consisted of the quantification of the relationbetween each of the neurohormonal levels and each of thefollowing demographic and clinical information : age, left ven-tricular ejection fraction, diuretic drug use, Killip class, use ofthrombolytic therapy, number of days between the indexmyocardial infarction and randomization, prior myocardialinfarction, peak creatine kinase, gender, diabetes, blood pres-

Table 1. Comparison of Neurohormones in Survival and Ventricular Enlargement (SAVE) StudyPatients and Normal Control Subjects

sure elevation requiring therapy before the myocardial infarc-tion, systolic blood pressure at the time of neurohumoralsampling and the use of beta-blockers, nitrates, calcium chan-nel blockers or digoxin. The variables involving medicationindicated the medication use at the time of neurohumoralsampling. These univariate relations were assessed using sim-ple linear regression procedures. All p values were based onthe value of the coefficients from the regression analysis andany p value < 0 .05 was considered statistically significant .

The multivariate analysis consisted of multiple linearregression . In each model, one and only one neurohormonewas the dependent variable . The independent variables werethe collection of demographic and clinical information asaddressed for univariate analysis. These independent vari-ables were analyzed all at once in the multiple regressionmodel. A set of independent variables that had a significantrelation with the individual neurohormones in the multivari-ate analysis was identified . Multicolinearity was assessed onthe basis of measurements of variance inflation factor,tolerance, condition index and condition number and foundnot to present a problem in our multiple regression models .As in the univariate analysis, p values from the multipleregression analyses were based on the value of the coeffi-cients from the regression equations and any p value < 0 .05was considered statistically significant .

ResultsBecause the majority of neurohumoral samples were

drawn 7 to 16 days after the myocardial infarction, the timingof neurohumoral sampling was found not to influence neu-rohumoral values in this study .

Plasma neurohormones . The increase in circulating neu-rohumoral levels was widespread, with SAVE patients hav-

Values are mean value ± SD. The p values compare SAVE participants with age-matched normal subjects .N/A = left ventricular ejection fraction not obtained in normal subjects.

IACC Vol . 22, No. 2August 1993 :390-8

SAVE Patients

NormalSubjects(n = 38)

NoDiuretic

Diuretic(n = 335)

p Value

(n = 184)

p Value

Norepinephrine (pg/ml) 221 t 87 278 ± 173

< 0.0001

343 ± 222

< 0.0001Epinephrine (pg/ml) 31 ± 16 35 ± 27 0.25

40 t 31 0.0143Dopamine (pg/ml) 21 ± 7 27 ± 22 0.001

31 ± 30 0.0002Aldosterone (ng/dl) 17 t 7 22 ± 22 0.029

38 t 35 0 .193Atria) natriuretic peptide (pg/m1) 21 ± 9 65 ± 64

< 0.0001

92 ± 90

< 0.0001Plasma renin activity (ng/ml per h) 1 .2 ± 1 .2 2 .4 t 1 .9

< 0.0001

4.0 ± 4.7 0 .0049Arginine vasopressin (pg/mi) 0.7 ± 0.3 1 .6 ± 8.2 0.05

2.3 ± 3.1 0 .001Age (yr) 57 ± 7 58 ± 10 NS

63 ± 9 0.01Left ventricular ejection fraction (°ro) N/A 33.2 ± 0.7 N/A

28.6 ± 0.7 N/ASerum sodium (mEq/liter) 143 ± 3 141 ± 7 0.0002

142 ± 10 0 .121Serum creatinine (mg/dl) 1 .0 t 0.2 1 .1 ± 0.4

< 0.0001

1 .3 ± 0.4

<0.0001Serum osmolality (mOsm/kg per H20) 294 ± 7 285 t 25

< 0.0001

289 t 16 0.096Serum glucose (mg/dl) 92 ± 20 122 ± 53

< 0.0001

138 ± 99

<0.0001

JACC Vol. 22, No . 2August 1993 :390-8

120

C 100AD

60Q

O 60I-A 40E7z 20

0

20 100 180 260 340 420 500 560 660 740

Plasma Norepinephrine (pg/ml)

I -a-.~nomd

PSAVE

< 0.0001

0.72 1.62 2.52 3.42 4.32 522 6 .12 7.02 7.92 8.82

Plasma Renin Activity (ng/ml/h)

ing an increase in all measured neurohormoneE (plasmanorepinephrine, 301 ± 193 vs. 222 ± 87 pg/ml ; plasma reninactivity, 3 .0 ± 3.7 vs. 1 .2 t 1 .2 nglml per h; argininevasopressin, 1 .9 ± 6.9 vs. 0.7 t 3 pg/ml, and atrial natri-uretic peptide, 75 ± 75 vs . 21 ± 9 pg/mI [all p < 0 .001])except plasma epinephrine, which was not increased (TableI, Fig. 1). Forty-eight percent of all patients had activation(>2 SD above the mean of age-matched control subjects) ofat least one of the vasoconstrictor neurohormones (plasmarenin activity, norepinephrine or arginine vasopressin) and61% had activation of the vasodilator neurohormone atrialnatriuretic peptide. However, despite this general increasein neurohumoral activation, a wide spectrum of neurohu-moral values could be found at all levels of left ventricularejection fraction, including many normal or low valueswhether or not diuretic drugs were being used (Fig . 2 and 3) .

Plasma norepinephrine values were distributed in a bell-shaped curve skewed to the right (Fig. IA). A total of 22% ofpatients had values >2 SD above the mean value of controlsubjects (expected 2 .5%). As plasma renin activities in-creased, the number of patients at each level of activationdecreased (Fig . 1B). Although many patients had normallevels, 22% had values >2 SD above the mean value incontrol subjects. This increase in plasma renin activityoccurred whether or not patients were receiving diurticdrugs (Table 1) . The pattern of increase in plasma aldoste-rone resembled that of plasma renin activity .

ROULEAU ET AL .

393NEUROHUMORAL ACTIVATION AFTER INFARCTION

0.4 1 .0

1 .8

2.6

3.4

4.2

5.0 5.8

Arginine Vasopressin (pg/ml)

m

ma,a.OI-aEz

F_p__i nomadSAVE

6.6 7.4

p < 0.0001

035 63

91 119 147 175 203 231 259

®

Atdal Natriuretic Peptide (pg/ml)

Figure 1 . Distribution of plasma neurohormone values for theSurvival and Ventricular Enlargement (SAVE) study patients .A, Plasma norepinephrine. Twenty-two percent of patients hadvalues >2 SD above the mean in normal control subjects ; 2% ofpatients had values greater than those shown . B, Plasma reninactivity . Twenty-two percent of patients had values >2 SD abovethe mean in normal control subjects ; 2% of patients had valuesgreater than those shown. The minimal measurable value variedbetween 0.5 and 0.6 ng/ml per h . C, Plasma arginine vasopressin .Twenty-seven percent of patients had values >2 SD above the meanin normal control subjects ; 2% of patients had values greater thanthose shown . The minimal measurable value was 0 .5 pg/ml . D, Plasmaatrial natriuretic peptide . Sixty-one percent of patients had values>2 SD above the mean in normal control subjects ; 2% of patientshad values greater than those shown . In A through D, each panelrepresents the mean value ± SD and the p value for age-matchedcontrol subjects and SAVE patients .

The wide range in values occurred with arginine vaso-pressin (Fig. IC). Again, as arginine vasopressin valuesincreased, the number of patients at each level of activationdecreased. The majority of patients had arginine vasopressinvalues within the normal range ; nevertheless, 27% of patientshad values >2 SD above the normal mean. This increase inarginine vasopressin in the SAV E patients is all the moreimpressive because it occurred despite a decrease in plasmaosmolality and in patients not receiving diuretic drugs (Table1). The increase in arginine vasopressin occurred in patientswith an osmolality within I SD of that in normal control

394

wO

EaZ

ROULEAU ET AL .NEUROHUMORAL ACTIVATION AFTER INFARCTION

-- o _4NOmfI

100 240 300 400 500 600 200 000 .500

ON DIURETICS

® NO DIURETIC

Figure 2 . Distribution of plasma norepinephrine values for theSurvival and Ventricular Enlargement (SAVE) study patients withthe patients classified according to left ventricular ejection fraction(LVEF) quartile. Patients are identified as to whether or not theywere receiving diuretic drugs . Values at the top of the figurerepresent mean value ± SD for age-matched control subjects .

subjects (1 .45 ± 1 .65 pg/ml, p < 0.001) and in patients with anosmolality >1 SD above that in control subjects (29 ±3.74 p&/ml, p < 0.001). Patients with an osmolality <1 SDbelow that in control subjects also tended to have a higherarginine vasopressin value (2.05 ± 10.4 pg/mI, p = 0.06) .

The greatest increase in any neurohormone occurred withatria) natriuretic peptide (Fig. ID). Values were distributedin a bell-shaped curve skewed largely to the right . A total of61% of patients had an increase in atrial natriuretic peptideto levels >2 SD above the normal mean .

Neurohumoral activation occurred whether or not di-uretic drugs were being used (Table 1, Fig . 2 and 3) . Patientsnot receiving diuretic drugs (about two thirds of the group)had better left ventricular (p < 0 .001) and renal (p < 0.001)function and were younger (p < 0 .001) . Nevertheless, theyalso had a significant increase in plasma renin activity,norepinephrine, arginine vasopressin and atrial natriureticpeptide compared with normal control subjects . Not surpris-ingly, a reasonably gdod correlation existed between levelsof the various plasma catecholamines (Table 2) . A similarlysignificant corrmlation was found between plasma reninactivity and aldosterone (Table 2) . Correlation, although

N

0)

I

50

40

20

a10

00

t0

20

30

40

50 a50

Plasma Renin Activity (ng/mlh)

Figure 3. Distribution of plasma renin activity values for the Sur-vival and Ventricular Enlargement (SAVE) study patients with thepatients classified according to left ventricular ejection fraction(LVEF) quartile . Patients are identified as to whether or not theywere receiving diuretic drugs . Values at the top of the figurerepresent mean value ± SD for age-matched control subjects . Theminimal measurable value varied between 0 .5 and 0.6 nglml per h .

weaker, were also found between plasma norepinephrineand plasma renin activity and aldosterone . Other correla-tions included atrial natriuretic peptide and norepinephrineand epinephrine, and arginine vasopressin and epinephrineand aldosterone. The use of diuretic drugs did not modifythese correlations in a major way .

Determinants of neurohnmoral activation (Table 3) . B_univariate analysis, plasma norepinephrine values were re-lated to age, Killip class, left ventricular ejection fractionand the use of diuretic drugs. The relation between Killipclass, norepinephrine and other neurohormones is illustratedin Figure 4, in which neurohumoral levels are plotted againstKillip class . Multivariate analysis revealed only Killip class,ejection fraction and age as important, with the use ofdiuretic drugs no longer being related to circulating norepi-nephrine levels (Table 3) . By multivariate analysis, plasmaepinephrine values were related to Killip class and ejectionfraction and plasma dopamine values were related to Killipclass, the use of thrombolytic therapy and diabetes (Table 3) .

By univariate analysis, plasma renin activity was relatedto the use of diuretic drugs, left

ventricular ejection fraction,

I ---Q---INormel

KINLVEF .35%

27% s LVEF 32%

0 1 .0 00 30 4,0 5.0 .50

JACC Vol . 22 . No. 2August 1993 :390-8

ON DIURETICS

® NO DIURETIC

JACC Vol . 22, No . 2August 1993 :390-8

Table 2. Summary of Cross-Correlations Between Neurhormones in Patients Taking (+) and NotTaking (-) Diuretic Drugs

Dopamine

Aldosterone0.338(<0 .001) 0.182(<0.001)0 .247 (< 0 .001) 0.243(< 0.001)

Atral natriureticpeptide

Plasma reninactivity

Killip class and the use of digoxin . However, by multivariateanalysis, only the use of diuretic drugs and ejection fractionwere important, with Killip class being marginal and the useof digoxin insignificant (Table 3) . By multivariate analysis,plasma aldosterone was related to the use of diuretic drugs,systolic blood pressure at the time of sampling and Killipclass (Table 3) .

By univariate analysis, plasma arginine vasopressin lev-els were related to seven variables, but by multivariateanalysis, only Killip class, age, diuretic drugs and diabeteswere important (Table 3) .

Plasma atrial natriuretic peptide had the largest increaseof al! neurohormones -a essured . By univariate analysis,atrial natriuretic peptide levels could be correlated with ninevariables. However, by multivariate analysis, only left ven-tricular ejection fraction, the use of thiombolytic agents,prior myocardial infarction or blood pressure requiring ther-apy before the SAVE-myocardial infarction were found tobe important (Table 3) . Tile relation between atrial natri-uretic peptide and Killip class was marginal .

DiscussionWe believe that our study is the first to indicate that

widespread neurohumoral activation occurs at the fine ofhospital discharge in patients with left ventricular dysfunc-tion but without overt heart failure after myocardial infarc-tion. In 48% of patients, significant activation (>2 SD abovethe mean of age-matched control subjects) of at least onevasoconstrictor neurohumoral system occurred . Also, 61%of patients had activation of the vasodilator neurohormone

ROtJLEAU ET AL .

395NEUROHUMORAL ACTIVATION AFTER 1NFARCT)ON

Values are r values for the correlation between neurohormones . with the p value in parentheses . Only significantcorrelations are shown .

atrial natriuretic peptide . Activation of one neurohumoralsystem did not correlate with activation of another, indicat-ing that factors regulating the activity of the various neuro-humoral systems are at least partially independent of oneanother.

Neurohumoral activation in the SAVE patients wasclearly greater than that in patients in the asymptomatic(prevention) arm of the Study of Left Ventricular Dysfunc-tion (SOLVD) study (9) even though the two groups had asimilar left ventricular ejection fraction. In the SAVE pa-tients, plasma renin activity was increased whether or notdiuretic drugs were used . whereas in the SOLVD preventiongroup, plasma renin activity was increased only in patientsreceiving diuretic drugs . Also, in SAVE, both plasma argi-nine vasopressin and atrial natriuretic peptide were in-creased to a greater extent than they were in patients in theSOLVD prevention arm. These results indicate that patternsof neurohumoral activation differ between patients with leftventricular dysfunction but without overt heart failure amean of 12 days after infarction and patients with similarasymptomatic left ventricular dysfunction who have neverhad or have long since recovered from an acute myocardialinfarction .

Compared with results in the SOLVD treatment or symp-tomatic group (9), the results of SAVE are variable, witharginine vasopressin being greater in SAVE patients, atrialnatriuretic peptide bei.i,~, similar ant{ htet!~ „hcm~ renirt

activity and plasma norepinephrine being somewhat !ewerthan values in the SOLV;u treatment group. The significanceof neurohumoral activation before hospital discharge afterinfarction and in patients with chronic heart failure may also

Epinephrine Dopamine Aldosterme

AtrialNatriureticPeptide

Plasma ReninActivity

ArginineVasopressin

Norepinephriue0.345 1< O .Otl11 0.411(<0 .001) 0.158

(0 .1:06) 0.115 10 .04)) 0 .2541< 0.N)l1+ 0 .480(<0 .001) 0.361(<0 .001) 0.220(<0.001) - -

Epinephrine0.419 ( < 0 .001) 0.3001< 0.001)

+ 0.3601< 0 .001) - 0.193 (0 .012)

Table 3. Variables Found to Influence the VariousNeurohormones by Multivariate Analyses

BP Req Rx = hypertension requiring therapy before the index myocardialinfarcdon in the Survival and Venricular Enlargement (SAVE) study pa-tients. Killip class is that obtain-A within 72 h after infarction (for details, seeMethods) ; LVEF = left ventricular ejection fraction: Ml = myocardialinfarction .

be different because the use of angiotensin-converting inhib-itors in the SAVE patients improved survival equally well inpatients with ?30% but <_40% and in patients with anejection fraction <30%, whereas in SOLVD, the benefit inpatients with an ejection fraction z30% was much less thanthat in patients with an ejection fraction <30% (3-5) .

Determinants of neurhwnoral activation . Seve1 al clinicalvariables and one laboratory test were identified as indepen-dent predictors of neurohumoral activation, the importanceof each varying from one neurohortnone to another . Time ofsampling did not significantly influence levels of any of theneurohormones measured. Because neurohumoral levelsvary widely during the first few days after infarction (I), atfirst glance this observation may appear surprising . How-ever, all previous studies have also indicated that the hyper-acute rise after infarction subsides by 7 days to attain morestable levels and because the overwhelitting majority ofsamples in this study were drawn between 8 and 16 days, atime when the hyperacute increase in neurohormones is

10 -a-v-t-

5-

8-a-

o-

e-

s-

4-

2-

0-

Normal T39 TIP-11 Ki81p . IYControl 174 n _ 45A=39

Figure 4. Relation between neurohormone levels and Killip class .Plasma norepinept-ine (p < 0.001), renin activity (p < 0 .01) andatria) natriuretic peptide (p < 0.001) but not arginine vasopressinwere related to Killip class . Values are mean value ± SD .

•

a-

•

4-

•

1-

past, the lack of time dependence in our study is understand-able .

Not surprisingly, left ventricular ejection fraction, a re-flection of left ventricular systolic function, appeared toinfluence the activation of most neurohormones . Neverthe-less, a large proportion of patients with the lowest quartile ofelection fraction had little or no neurohumoral activation(Fig. 2 and 3), indicating that neurohumoral activation is theresult not simply of left ventricular dysfunction but of theconvergence of numerous factors .

Ki:lip class a :@ assessed within 72 h of the myocardialinfarction was the clinical variable most consistently associ-ated with neurohumoral activation. Killip class is the resultnot only of systolic dysfunction but also of diastolic dysfunc-tion and of sodium and water retention. That it was the onlyvariable related to activation of all measured neurohormonesand that this relation was independent of left ventricularejection fraction underscore the importance of diastolicdysfunction in determining postinfarction neurohumoral ac-tivation. Alternatively, Killip class may simply identifypatients with the greatest initial neurohumoral activation

39G

ROULEAU ET AL.N1 UROHUMORAL ACTIVATION AFTER INFARCTION

JACC Vol . 22. No . 2August 1993 :390-8

Adjusted RegressionCoefficient P Value

NorepinephrineLVEF -5.5 < 0.001Killip class 62 < 0.001

Age 3 0.005

EpinephrineKillip class 4 .6 0 .01LVEF -0.33 0 .03

DopamineKillip class 5 .2 0.004T rombolytic therapy --4 .7 0.004Diabetes 6,8 0.03

ReninDiuretic 1 .3 < 0.001LVEF -0.07 < 0 .001Killip class 0,47 0 .08

AldosteroneDiuretic 14 < 0 .00'Systolic blood pressure 0 .2 0 .014Killip class 4 0 .05

Arginine vasopressinKillip class 0 .62 < 0 .001Age 0.03 < 0.001Diuretic 0 .69 0 .004Diabetes 0.52 0 .05

Atrial natriuretic peptideLVEF -1 .6 < 0 .001Thrombolysis -19 .4 0.01Prior M1 19 0.02BP Req Rx 17 0.04Killip class 9 0.09

JACC Vol. 22. No . 2August 7993 :390-8

and thus at greatest risk of sustained neurohumoral activa-tion.

The use of diuretic drugs was found to be related toactivation of the renin-angiotensin system and plasma argi-nine vasopressin . Other investigators (1) have described astrong relation between the use of diuretic drugs and activa-tion of the renin-angiotensin system in this type of patient .However, we believe that this is the first time that persistentactivation of the renin-angiotensin system has been shown tooccur without the use of these agents . Early after infarction,patients are known to retain fluid (10) and to be predisposedto development of hyponatremia. In this study, SAVEpatients had hypo-osmolality and thus had an inappropriateincrease in circulating arginine vasopressin levels . It istherefore tempting to speculate that the relation betweenthis neurohormone and the use of diuretic drugs is the resultof arginine vasopressin-induced fluid retention and over-load.

Age was found to be correlated with plasma norepineph-rine and arginine vasopressin . The increase in plasma nor-epinephrine with age has been shown to be largely the resultof an increase in spillover rate rather than of any alteration inextraction (11) . This increase in plasma norepinephrine withage combined with other poor prognostic signs such asdecreased compliance, contractility and glomerular filtrationrate known to occur with advancing age may help explainwhy older patients have a worse prognosis after infarction(12) . An increase in plasma arginine vasopressin with ad-vancing age has already been described in normal patients(13) and thus tais relation was not surprising. The mecha-nism for this increase is poorly understood . No relation wasfound to occur between age and plasma renin activity oratrial natriuretic peptide. Under normal circumstances sucha relation exists (14,15) . Why it was not found in thissituation remains speculative . Perhaps these relations weremasked or overshadowed by other factors.

Other variables independently related to neurohumoralactivation were the use of thrombolytic agents, prior myo-cardial infarction and hypertension requiring therapy beforethe SAVE myocardial infarction for atrial natriuretic pep-tide, and diabetes for arginine vasopressin . The inverserelation between the use of thrombolytic agents and atrialnatriuretic peptide after infarction has previously been doc-umented and is thought to be at least partially the result ofprevention of ventricular dilation by the use of thrombolytictherapy. The correlation between atrial natriuretic peptideand prior myocardial infarction and hypertension requiringtherapy before the SAVE myocardial infarction probablyreflects the myocardial compliance problems related to thesedisease processes . An exaggerated increase in arginine va-sopressin in response to stress in diabetes, although stillcontroversial, has previously been described (16) .

Certain variables previously shown to affect neurohu-moral activation were not found to be important in thisstudy. Medications such as beta-blockers (17,18), calciumchannel blockers (19,20), nitrates (21) and digoxin (22,23),

ROULEAU ET AL.

397NEUROHUMORAL ACTIVATION AFTER INFARCTION

which have been fund L modify neurohumoral levels undercertain circumstances, were found to be unrelated to neuro-humoral activation. This lack of efftct may have occurredbecause measurements were oLtained after these drugs werewithheld for 712 h or because their effects were over-whelmed by other more important factors .

Conclusions. This study indicates that widespread neuro-humoral activation occurs at the time of hospital discharge inmany patients with left ventricular dysfunction but no overtheart failure after infarction . This pattern differs from that ofpatients with chronic left ventricular dysfunction but withoutovert heart failure (SOLVD prevention arm) in whom lessactivation of the plasma renin activity, arginine vasopressinand atrial natriuretic peptide systems occurs (9), indicatingthat a unique profile of neurohumoral activation occurs at amean of 12 days after infarction in patients with left ventric-ular dysfunction but no overt heart failure . This activationvanes widely from patient to patient and is not wholly theresult of left ventricular d, sfunction or of the use of diureticdrugs. A significant number of patients who have a very lowejection fraction and are taking diuretic drugs have noneurohumoral activation, whereas an equally importantgroup of patients who have a higher ejection fraction and arenot taking diuretic drugs have marked activation . Killipclass, left ventricular ejection fraction, the use of diureticdrugs and age are the clinical and laboratory variables mostclosely related to neurohumoral activation .

We acknowledge the expert coordination of the project by Nicole Poitras,BSc ; the remarkably care of the patients by the nurses involved, GinetteGaudette, Jocelyne Fouquette, Gail Burton, Shirley Vorderbrugge, SusanMooney, Pat Courcelles. Diane Laforge, Kimberley Luon, Eileen Shaht,Linda Reid, CEline Roy, Bonnie Cochrane, Josephine Sallano, Mary-EllenCoglienese, Carole Chapin, Kathy Connors and Michele Weindberg ; theexcellent biochemical work by Michble Lonergan, Made-Frangoise Arthur,Nicole Ruelle and Marthe Parent, and the expert secretarial and editorial helpof Marguerite Cloutier . Finally, the very helpful comments and advice of Dr .Jean-Gil Joly, MD and of Laurent Harvey, BPharm of Bristol-Myers Squibbare appreciated.

Appendix

Affdiatlom of the Study AuthorsThe study authors are associated with the following institutions : Institut

de Cardiologie, Montreal, Quebec, Canada (Dr . Rouleau) ; Hdpital du SacraCoeur, Montreal, Quebec (Drs . de Champlain, Klein and Bichet) ; Universityof Texas School of Public Health, Houston, Texas (Drs . Maye and Hamm) ;College of Physicians and Surgeons, New York, New York (Dr. Packer) ;Institut de Cardiologie, Montreal, Quebec, (Dr . Dagenais) ; Health SciencesCenter, Memorial Hospital, St . John's, Newfoundland, Canada (Dr . Sussex) ;Victoria Hospital, London, Ontario, Canada (Dr. Arnold) ; Hdpital Notre-Dame de Montreal, Montreal (Dr. Sestier) ; Kingston General Hospital,Kingston, Ontario (Dr. Parker) ; The Wellesley Hospital, Toronto, Ontario(Dr. McEwan) ; University of British Columbia, V ;,irtcouver, British Colum-bia, Canada (Dr. Bernstein) ; University of Manitoba, Winnipeg, Manitoba,

Canada (Dr. Cuddy) ; Brigham and Women's Hospital, Boston, Massachusetts(Drs . Lamas and Pfeffer); University of Maryland School of Medicine,Baltimore, Maryland (Dr. Gottlieb); Jewish General Hospital, Montreal,Quebec (Dr. McCans) ; Hdpital Enfant-Jdsus, Quebec (Dr . Nadeau); Hdtel-Dieu de Levis, Quebec (Dr . Delage).

398

ROULEAU ET AL .NEUROHUMORAL ACTIVATION AFTER INFARCTION

References

1 . Rouleau JL, Moy6 LA, de Champlain J, et al. Activation of neurohumoralsystems following acute myocardial infarction. Am J Cardiol 1991 ;68:DOD-61).

2. The Norwegian Multicenter Study Group . Timolol-induced reduction inmortality and reinfarction in patients surviving acute myocardial inf :irc-lion. N Engl J Med 1981 ;304:301-7.

3 . Pfeffer MA . Braunwald E, Moy6 LA, et al . The effect of captopril onawrtality and morbidity in patients with left ventricular dysfunctionfollowing myocardial infarction . N Engl J Mad 1992 ;327 :669-77.

4. The SOLVD Investigators . Effect of enalapril on survival in patients withreduced left ventricular ejection fractions and congestive heart failure.N Engl J Med 1991 ;325:293-302.

5. The SOLVD Investigators. Effect of enalapril on mortality and thedevelopment of heart failure in asymptomatic patients with reduced leftventricular ejection fractions. N Engl J Mod 1992 ;327 :685-91 .

6. Chadda K, Goldstein S, Byington R, Curb JD . Effect of propranolol afteracute myocardial infarction in patients with congestive heart failure.Circulation 1986;73 ;503-10.

7 . Nelesen RA, Dimsdale JZ, Zisgler MG . Plasma atrial natriuretk peptideis unstable under most storage conditions . Circulation 1992:86:463-6.

8. Mettauer B, Rouleau JL, Bichet D, at al . Differential long-term intrarenaland neurohormonal effects of captopril and prazosin in patients withchronic congestive heart failure : importance of initial plasma reninactivity . Circulation 1986:73 :492-502 .

9. Francis GS, Ben•• t C, Johnstone DE, at al . Comparison of neuroen-docrine activatio' . in patients with left ventricular dysfunction with andwithout congestive heart failure . Circulation 1990 ;82 :1724-9.

10 . Col J, Petain M, Van Eyll C, Cheron P, Chattier AA . Pouleur H. Earlychanges in sodium and water balances in patients with acute myocardialinfarction: relationship to haemodynamics and creatine kinase . Eur J ClinInvest 1984;14:24754.

II . MacGilchrist Al. Hawksby C, Howes LG, Reid JL . Rise in plasmanoradrenaline with age results from an increase in spillover rate. Geron-tology 1989;35:7-13 .

JACC Vol . 22, No. 2August 1993:390-8

12. Cairns JA, Singer J, Gent M, et al . One year mortality outcomes of allcoronary and intensive care unit patients with acute myocardial infarc-tion, unstable angina or other chest pain in Hamilton, Ontario, a city of375,000 people. Can J Cardiol 1989;5 :239-46.

13 . Os 1 . Kjeldsen SE . Aakesson I, et al. Evidence of age-related variation inplasma vasopressin of normotensive men. Scand J Clin Lab Invest1985 ;45:263-8.

14 . Meade TW, Imeson JD, Gordon D . Peart WS. The epidemiology ofplasma resin. Clin Sri 1983 ;64:273-80.

15 . Haller BGD, ZOst H, Shaw S, Gnidinger MP, Uehlinger DE, WeidmannP. Effects of posture and aging on circulating atria) natriuretic peptidelevels in man. I Hypertension 1987 ;5:551-6.

16. Zerbe RL, Vinicor F, Robbertson GL . Regulation of plasma vasopressinin insulin-dependent diabetes mellitus . Am J Physiol 1985 ;12 :E317-25.

17. Hansen JF, Hesse B, Christensen NJ . Enhanced sympathetic nervousactivity after intravenous propranolol in ischaemic heart disease . Eur JClin Invest 1978 ;8 :31-6.

18 . Tsai RC, Yamaji T, Ishibashi M, et al . Effect of p-adrenergic blockade onplasma levels of atrial natriuretic peptide during exercise in humans .I Cardiovasc Pharnacol 1988 ;11 :614-8.

19 . Loft FC, Aronoff GR, Sloan RS, Fineberg NS, Weinberger MH . Calciumchannel blockade with nitrendipine : effect on sodium homeostasis, therenin-angiotetisin system, and the sympathetic nervous system in hu-mans . Hypertension 1985 ;7:438-42 .

20. Kiowski W, Boili P, Eme P, Hulth'en UL, M'uiler FB, BOhler FR.Hemadynamic and counterregulatory effects of calcium antagonists inhypertension . J Cardiovasc Pharmacol 1988 ;12 :539-43 .

21 . Olivari MT, Carlyle PF, Levine TB, Cohn JN . Hemodynamic andhormonal response to transdermal nitroglycerin in normal subjects and inpatients with congestive heart failure . J Am Coll Cardiol 1983 ;2:872-8 .

22 . Covit AB, Schaer GL, Sealey JE, Laragh JH, Cody RJ . Suppression ofthe renin-angiotensin system by intravenous digoxin in chronic congestiveheart failure. Am J Mad 1983 ;75:445-7 .

23 . Antonello A. Cargnielli G, Ferrari M, Melacini P, Montanaro D. Effect ofdigoxin on plasma renin activity in man. Lancet 1976 ;2 :850 .