- Email: [email protected]
Usefulness of the chest x-ray for predicting abnormal left ventricular function after acute myocardial infarction
Usefulness of the chest x-ray for predicting abnormal left ventricular function after acute myocardial infarction We evaluated 229 patients discharged after a definite acute myocardial infarction. Pulmonary venous congestion determined from chest x-ray films during the hospitalization and at discharge and the cardiothoracic ratio at discharge were compared to the left ventricular ejection fraction measured at discharge by a gated radionuclide technique. During hospitalization, pulmonary venous congestion was found on at least one x-ray frame in 94 patients (41%). At discharge 134 patients (59%) had abnormal ejection fraction (<0.51) and 35 had pulmonary venous congestion (15%). The sensitivity of the x-ray for detecting an abnormal ejection fraction was 20% when pulmonary venous congestion was observed on the discharge x-ray film (specificity 92% and predictive value 77%), 52% if pulmonary venous congestion was present on any x-ray film during the hospitalization (specificity 74% and predictive value 73%), and 47% if the cardiothoracic ratio was abnormal (r0.50) on the discharge x-ray film (specificity and predictive value 66%). We conclude that an abnormal x-ray film at discharge or during the hospitalization will identify approximately one-half of the abnormal ejection fractions at the time of hospital discharge. Therefore, to reliably assess left ventricular function, either for prognostic or therapeutic purposes in the individual patient, a more direct measure of left ventricular function such as radionuclide angiography must be obtained. (AM HEART J 108:1431, 1984.)
Erling Birk Madsen, M.D., Elizabeth Gilpin, M.S., Robert A. Slutsky, M.D., Staffan Ahnve, M.D., Hartmut Henning, M.D., and John Ross, Jr., M.D. San Diego, Calif.
Assessment of left ventricular function after acute myocardial infarction (AMI) is important because prognosis is less favorable when left ventricular function is depressed.le4 Direct evaluation of left ventricular function after myocardial infarction is possible by direct contrast ventriculographf or, more recently, by means of radionuclide techniques.‘-*’ 6-8 Previous studies have reported that plain chest roentgenograms provide fair correlations between
From the Division of Cardiology, University of California, San Diego. Supported by National Institutes of Health Research Grant HL 17682; by Ischemic Heart Disease SCOR grant awarded by the National Heart Lung, and Blood Institute, United States Public Health Service, Department of Health, Education, and Welfare, No. CA 26666; by International Research Fellowships 1 F05 TWO 31544-01 and 1 F05 TWO 03308-01, Fogerty International Center; by the United States Public Health Service, Department of Health, Education, and Welfare; and by the Danish Heart Foundation. Received for publication Jan. 23, 1984; revision received Apr. 25, 1984; accepted May 21, 1984. Reprint requests: John Ross, Jr., M.D., Dept. of Medicine, M-013B, University of California, San Diego, La Jolla, CA 92093.
radiologic criteria for left ventricular failure and hemodynamic findings, particularly the pulmonary capillary wedge pressure measured during AMI?-” Chest x-ray abnormalities have also been employed for prognosis after AMI.” The possibility that indirect assessment of left ventricular function might be obtainable from standard chest x-ray films would be of both economic and clinical importance. Therefore, the purpose of this paper was to investigate the relationship between roentgenographic findings and left ventricular function assessed by a predischarge left ventricular ejection fraction (EF) study in patients with AMI. METHODS Patient
population. Our population consisted of 229 patients discharged after definite AMI. Data concerning these patients were available in a data base maintained by the Specialized Center of Research (SCOR) for Ischemic Heart Disease at the University of California, San Diego, Medical Center. The patients were recruited over a 3-year period from 1979 to 1982 from the University of California, San Diego, Medical Center, the Veterans Administrations Hospital, and the United States Naval Regional 1431
December,
1432
Madsen et al.
American
Heart
1984 Journal
1. Use of dischargechest x-ray pulmonary venous congestiongrade and cardiothoracic ratio to predict abnormal EF
Table
EF <0.51
EF >0.51
134 27
95 8
194
107
a7
219 92
129 61
90 31
127
68
59
Total Pulmonary Total Grade I-IV Grade
venous
0
Cardiothoracic Total >0.50
Absent
Specificity
CICCID-UCY
Predictive value
20.1
91.6
49.8
77.1
47.3
65.6
54.8
66.3
52.7
63.3
57.1
67.3
ratio
<0.50 Combination Total Present
congestion 229 35
Percent Sensitivity
(grade
I-IV 219 101 118
and/or
ratio 70.50) 129 68 61
90 33 57
Medical Center, all in San Diego; and from Vancouver General Hospital in British Columbia, Canada. All patients were admitted to the hospital within 24 hours after onset of symptoms. The diagnosisof AM1 wasestablishedby at least two of the following criteria: (1) characteristic chest pain, (2) elevation of creatine kinase, and (3) ECG changeswith evolution of Q waves (transmural infarction). Nontransmural infarction wasdiagnosedby typical ST segmentor T wave changesaccompaniedby at least criterion (2). No patient in this study had intracoronary or intravenous streptokinase therapy, coronary angioplasty, or coronary bypass surgery between hospital admission and discharge. The age of the patients ranged from 30 to 95 years (mean age 63 years) and 198 of the patients were men (86%). A previous history of myocardial infarction (MI) was registered in 74 patients (32%), previous hypertension wasnoted in 103patients (45%), and previous heart failure was registered in 36 patients (16%). The location of the AM1 was anterior in 68 patients (30%), inferior in 81 patients (35%), indeterminate in 30 patients (35%), and subendocardialin 50 patients (22%). Chest roentgenograms. All patients had at least two chest x-ray examinations done during the hospitalization-one early in the acute phaseshortly after admission and the other just before hospital discharge.Some of the earlier x-ray films were taken with the patient in the supine position, but all the predischargex-ray films were performed with the patient upright and in the usual posterior-anterior projection. The initial x-ray films were exposed after inspiration of about 1 L from functional residual capacity and were timed at end diasto1e.l”The predischargex-ray film was supplementedwith a lateral study. All radiographs were read by individuals unaware
of the radionuclide data. In addition, the radionuclide data were assessedby individuals not familiar with the patient or his clinical course. Pulmonary venous congestion on the chest x-ray film was graded as we have previously reported.” Grade 0 equals no pulmonary venous congestion. Grade I equals pulmonary venous hypertension with redistribution of pulmonary blood flow, defined asa greater diameter of the upper comparedwith the lower lobe pulmonary vessels.If this finding was the only abnormal finding on a supine x-ray film, the study was interpreted as normal (no pulmonary venous hypertension). In the supine position, pulmonary vascular redistribution and either peribronchial cuffing or lossof the right hilar angle were required to consider the radiography grade I. Grade II equals interstitial pulmonary edema,defined by lossof definition of the pulmonary vascular markings in associationwith Kerley’s B lines. Grade III equalslocalized alveolar edema defined as confluent alveolar infiltrates in perihilar areas and lower lung fields. Grade IV equals diffuse alveolar edema defined as diffuse confluent alveolar infiltrates throughout most areasof both lung fields. Cardiomegaly was assessedfrom the discharge x-ray film by the cardiothoracic ratio,12defined as the relation between the transverse diameter of the heart and the internal diameter of the chest. The transverse diameter of the heart wasmeasuredasthe sumof the widest portion of the heart from the right to the left border of the cardiac silhouette at the midline. The internal diameter of the chest wasmeasuredat the level of the highest point on the left hemidiaphraghm. Cardiothoracic ratios above or equal to 0.50 were consideredabnormal.‘* This measurement wastechnically possiblein only 219 patients. Left ventricular EF. Shortly before discharge,left ventricular EF was measuredby radionuclide ventriculogra-
Volume Number
106 6
0.80-
Chest x-ray and LV function
*
... ::* ... ........... .... 0.60- ............ .::.. ..... *:::: ................... ............ 0.50- .....:-:* ............... ........... ..... 0.40_ ....... ..::. .......... ..::.. .::. 0.30.: .:. 0.20 t 0.10
l0.8C
.. I0.70 * “...
’
. :
*-
*
*
0.60I E 0.50 ~ F 2E 0.40 E 5 0.30 2
.I .
. .
0.20
1
1
I
,
0
I 28
P 6
III
194
.
.
.
* t
0.00
1433
A
0.70-
g 2 8 x F Ei 3
in AMI
1
0.00 -
0.30
.
. . .
0.40
0.50
0.60
CARDIOTHORACICRATIO
Fig.
tion for each group. Grade 0: no pulmonary venous congestion. Grade I: pulmonary venous hypertension. Grade II: interstitial pulmonary edema.Grade III: localized alveolar edema.
:.: . ..
.: .-.. * . - . *. . *. --
0.10
PULMONARY CONGESTIONGRADE 1. Ejection fraction (EF) related to chest x-ray pulmonary venous congestiongrade at discharge.Vertical bars with cross-hatchingshow mean and standard devia-
..:
2. Relationship betweenEF and cardiothoracic ratio from dischargechest x-ray film. N = 219 patients. Correlation coefficient = -0.24.
Fig.
RESULTS
phy with the patient in the supine position. After in vivo labeling of red cells with 15 to 20 mCi technetium-99m, multigated equilibrium angiogramswere obtained in the left anterior oblique position for 5 minutes, as previously reported from our laboratory.13,14 Left ventricular EF was calculated from a computer-generated background corrected time-activity curve using a variable region of interest as: (end-diastolic counts minus end-systolic counts)/(end-diastolic counts). An EF below or equal to 0.51wasconsideredabnorma1.3Usually both the discharge chest x-ray examination and the radionuclide ventriculogram were performed the day before discharge. Statistics. The relationship between the chest x-ray findings and the EF values was assessedby analysis of variance with multiple comparisonsamong group means and linear regressionanalysis.For eachset of observations of pulmonary venous congestionand abnormal cardiothoracic ratio we defined: (1) sensitivity equals percent of patients with abnormal EF who had a positive x-ray finding; (2) specificity equals percent of patients with normal EF who had a negative x-ray finding; (3) accuracy equalspercent of patients with true positive (both abnormal EF and abnormal x-ray) and true negative (both normal EF and normal x-ray) findings; and (4) predictive value of abnormal x-ray equalspercentageof patients with an abnormal x-ray who have an abnormal EF.
Discharge EF and chest x-ray. The discharge EF was 0.48 + 0.13 (mean + SD). It was abnormal (
December,
1434
Madsen
American
0.60
[
-
p 0.50 0 $ LL 0.40 zi i= u ; 0.30 -
... ::.: ...
.....:.. ...... ::;::. ..:. ...... .1.. .....::: .......... ..::. ..... ..... ..I;:’ ... ... .i. ::. ..
; 0.10 I
.::.2. ::.:. ...i. ::.:. ;I
DISCUSSION
0.20
0.00’
;
135 -0.005-0.0005-
I
1
I
It
57
23
I
III-l.K 14
0.03 A I
1994 Journal
highest grade of pulmonary venous congestion from any x-ray film during the hospitalization, the mean discharge EF was 0.51 f 0.12 in 135 patients with grade 0, 0.45 + 0.13 in 57 patients with grade I, 0.44 _+ 0.11 in 23 patients with grade II, and 0.36 f 0.14 in 14 patients with grade III to IV (Fig. 3). The difference between patients with grade 0 compared to patients with grade I, II, or III to IV was significant (analysis of variance), and patients with grade III to IV had significantly lower EF compared to patients with grade I.
0.80 0.70
Heart
0.03-
PULMONARY CONGESTIONGRADE Fig. 3. EF related to highest chest x-ray pulmonary
venouscongestiongrade during the entire hospitalization. Vertical bars with cross-hatching show mean and standard deviation for each group. Grade 0: no pulmonary venous congestion.Grade I: pulmonary venous hypertension. Grade II: interstitial pulmonary edema. Grade III: localized alveolar edema. Grade IV: diffuse alveolar edema.
discharge chest x-ray examination (Fig. 1). This shows the considerable overlap between EF values for patients in the grade groups. The mean EF was 0.49 + 0.13 for patients with grade 0, 0.44 + 0.14 for patients with grade I, and 0.40 + 0.14 for patients with grade II (no significant differences). Fig. 2 presents the linear regression between the cardiothoracic ratio and EF. The correlation coefficient was only -0.24. No breakpoint values in either cardiothoracic ratio or EF could be determined from this plot. Discharge EF and worst chest x-ray during hospitalization. Pulmonary venous congestion was present on
at least one chest x-ray film during the hospitalization in 94 patients (41%). This finding yielded a higher sensitivity (52% ), a lower specificity (79%), higher accuracy (61%), and nearly unchanged predictive value (73%) compared with the discharge x-ray examination (20%, 92%, 50%, and 77%, Tables II and I, respectively). When patients were grouped according to the
The prognostic value of the presence of pulmonary venous congestion on the initial chest x-ray film both for early (30-day) and later (l-year) mortality has been established previously.12 Also, assessment of left ventricular EF has been used prognostically after AMI.‘e6 Therefore we endeavored to evaluate the relationship of an abnormal chest x-ray examination and left ventricular EF in AM1 patients. Radiographic evidence of pulmonary venous congestion has been shown to reflect the pulmonary capillary wedge pressure during AMI.s-” Although the correlation between pulmonary congestion and the pulmonary wedge pressure has been good, in some patients a phase lag is evident in which the films remain abnormal for several days after the wedge pressure has returned to norma1.l’ The cardiothoracic ratio has been reported to yield 70% accuracy for correctly predicting a combined measurement incorporating left ventricular mass, enddiastolic volume, and left atria1 volume obtained from quantitative biplane angiography, with a correlation coefficient of 0,64.15 Discharge
EF and chest
x-ray in the present
study.
This study has shown that whereas many patients (59%) had an abnormal EF at hospital discharge, in only one of five patients was it associated with pulmonary venous congestion on the discharge chest x-ray film. In 9 out of 10 patients with a normal EF, the x-ray film showed no pulmonary venous congestion. In the relatively
few patients in whom the
discharge chest x-ray film showed pulmonary venous congestion, three out of four also had an abnormal EF (predictive value 77 % ). If the x-ray film with the highest grade of pulmonary venous congestion from the hospitalization was utilized, the sensitivity increased from 20% to 52% but was still unsuitably low. Also, the specificity fell and the predictive value was not improved. The combination of pulmonary venous congestion and abnormal cardiothoracic ratio did not improve
Volume
106
Number
6
Chest x-ray
and LV function
II. Use of highest chest x-ray pulmonary venous congestion grade during the entire hospitalization abnormal EF .
Table
Total
EF <0.51
EF >0.51
Total Grade I-IV
229 94
134 69
25
Grade 0
135
65
70
Sensitivity
Specificity
Percent accuracy
in AMI
1435
to predict Predictive value
95 51.5
the ability to predict an abnormal EF appreciably (Fig. 2). The correlation coefficient was only -0.24 compared to previously reported correlation coefficients of -0.435 and -0.29.7 This is, however, not too surprising since the cardiothoracic ratio not only takes the transverse cardiac diameter (which includes both ventricles and the right atrium) into account, but also the transverse diameter of the chest as an indicator of body build. This measurement thus introduces a variable that is totally independent of whether there is cardiac disease or not. We defined a cardiothoracic ratio of ~0.50 to be abnormal, a value generally used. A higher value of 0.60 would have decreased the specificity but would not have improved the predictive value for an abnormal EF in these patients (Fig. 2). Our results indicate that pulmonary venous congestion on a chest x-ray film from the hospitalization or at discharge is indicative of abnormal EF (predictive values of 73 % and 77 % , respectively). However, most patients with an abnormal discharge EF will have normal discharge chest x-ray films (Table I, Fig. 1). In the patients studied, a normal cardiothoracic ratio on the discharge chest x-ray film was often accompanied by an abnormal EF, and a higher ratio often was present in patients with a normal EF.5 The latter finding may be explained partly by a high incidence of previous hypertension in our patients (45 % ). Thus, cardiomegaly on x-ray examination correctly indicated depressed left ventricular function in 66% of the patients. EF, chest x-ray, and heart failure. Previous comparisons between clinical heart failure c1ass,16 which included chest x-ray evidence of pulmonary venous congestion criteria, have shown that the EF measured in AMI patients early after hospital admission was significantly lower in patients with left heart failure and pulmonary edema compared to patients without left heart failure.‘, 3 Most patients who developed shock or died within 3 months had a low EF.6 Our findings at discharge show a significantly lower mean EF in patients with grade I, II, or III to IV pulmonary congestion on chest x-ray film compared to patients with grade 0, and in patients with
73.1
60.7
73.4
grade I compared to patients in grade III to IV. However, as shown previously,‘v3 although the differences in the mean EF values are significant, the overlap of values among the grade groups is great. This makes prediction of an abnormal EF for the individual patient unreliable. In recent studies, we have noted that a phase lag may occur acutely after an experimentally induced elevation in left atrial pressure.17* l8 Nonetheless, sustained left atrial hypertension will invariably result in interstitial and alveolar pulmonary edema, both experimentally and clinically.‘% lg An abnormal radiographic finding coupled with normal cardiac function and size is confusing. It has been recently proposedzO that this may in part reflect noncardiogenie pulmonary edema during acute coronary occlusion. This leaky membrane-induced edema during AM1 was independent of the degree of pulmonary microvascular hydrostatic pressure, and could be substantially minimized by the administration of indomethacin. Thus, if the etiology of radiographic changes in AM1 is not solely hemodynamic, it is less surprising that it is not possible to consistently predict the underlying hemodynamic status from a chest x-ray film. Conclusions. The value of the chest x-ray examination for the assessment of left ventricular function estimated by EF in AM1 patients is limited. Correct assessment of left ventricular function at hospital discharge requires more direct measurements than chest x-ray findings.
REFERENCES
1. Schelbert HR, Henning H, Ashburn WL, Verba JW, Karliner JS, O’Rourke RA: Serial measurements of left ventricular ejection fraction by radionuclide angiography early and late after myocardial infarction. Am J Cardiol 38:407, 1976. 2. Schulze RA, Strauss HW, Pitt B: Sudden death in the year following myocardial infarction. Am J Med 62:192, 1977. 3. Battler A, Slutsky R, Karliner J, Froelicher V, Ashburn W, Ross J Jr: Left ventricular ejection fraction and first third ejection fraction early after acute myocardial infarction: Value for predicting mortality and morbidity. Am J Cardiol 45:197, 1980. 4. Shah PK, Pichler M, Berman D, Singh BN, Swan HJC: Left ventricular ejection fraction determined by radionuclide ven-
December,
1436
5.
6.
7.
8.
9.
10.
11.
12.
Madsen et al.
triculography in early stages of first transmural myocardial infarction. Am J Cardiol 45:542, 1980. Feild BJ, Russell RO, Moraski RE, Soto B, Hood WP, Burdeshaw JA, Smith M, Maurer BJ, Rackley CE: Left ventricular size and function and heart size in the year following myocardial infarction. Circulation %X331. 1974. Rigo P, Murray M, Strauss HW, Taylor D, Kelly D, Weisfeldt M, Pitt B: Left ventricular function in acute myocardial infarction evaluated by gated scintiphotography. Circulation 50:678, 1974. Dewhurst NG, Hannan WJ, Muir AL: Prognostic value of radionuclide ventriculography after myocardial infarction. Q J Med 49:479, 1981. Nemerovski M, Shah PK, Pichler M, Berman DS, Shellock F, Swan HJC: Radionuclide assessment of sequential changes in left and right ventricular function following first acute transmural myocardial infarction. AM HEART J 104:709, 1982. Lassers BW, George M, Anderton JL, Higgins MR, Philip T: Left ventricular failure in acute myocardial infarction. AM HEART J 25:511, 1970. McHugh TJ, Forrester JS, Adler L, Zion D, Swan HJC: Pulmonary vascular congestion in acute myocardial infarction: Hemodynamic and radiologic correlations. Ann Intern Med 76:29, 1972. Kostuk W, Barr JW, Simon AL, Ross .J Jr: Correlations between the chest film and hemodynamics in acute myocardial infarction. Circulation 48:624, 1973. Battler A, Karliner JS, Higgins CB, Slutsky R, Gilpin E, Froelicher VF, Ross J Jr: The initial chest x-ray in acute myocardial infarction. Circulation 61:1004, 1980.
American
1984
Heart Journal
13. Ashburn WL, Schelbert HR, Verba JW: Left ventricular ejection fraction-A review of several radionuclide angiographic approaches using the scintillation camera. Prog Cardiovasc Dis 20:267, 1978. 14. Pfisterer ME, Ricci DR, Schuler G, Swanson SS, Gordon DG, Peterson KE, Ashburn WL: Validity of left ventricular ejection fractions measured at rest and peak exercise by equilibrium radionuclide angiography using short acquisition times. J Nucl Med 20:484, 1979. 15. Clover L, Baxley WA, Dodge HT: A quantitative evaluation of heart size measurements from chest roentgenograms. Circulation 47:1289, 1973. 16. Killip T, Kimball JT: Treatment of myocardial infarction in a coronary care unit. Am J Cardiol 20:457, 1967. 17. Slutsky R, Higgins C: Intravascular and extravascular pulmonary fluid volumes. II. The response to rapid left atria1 hypertension and the theoretical implications for radiographic and radionuclide images. Invest Radio1 181:33, 1983. 18. Slutsky R, Higgins C: Intravascular and extravascular pulmonary fluid volumes. III. Response to sustained left atria1 hypertension. Cardiovasc Res 17:113, 1983. R, Slutsky R, Shabatai R, Higgins C: Extravascular 19. Grover lung water in patients with congestive heart failure. A comparison of patients with acute and chronic left heart disease. Radiology 147:659, 1983. 20. Richeson JF, Paulshock C, Yu P: Non-hydrostatic pulmonary edema after coronary ligation in dogs: Protective effect of indomethacin. Circ Res 50:301, 1982.
Erling Birk Madsen, M.D., Elizabeth Gilpin, M.S., Robert A. Slutsky, M.D., Staffan Ahnve, M.D., Hartmut Henning, M.D., and John Ross, Jr., M.D. San Diego, Calif.
Assessment of left ventricular function after acute myocardial infarction (AMI) is important because prognosis is less favorable when left ventricular function is depressed.le4 Direct evaluation of left ventricular function after myocardial infarction is possible by direct contrast ventriculographf or, more recently, by means of radionuclide techniques.‘-*’ 6-8 Previous studies have reported that plain chest roentgenograms provide fair correlations between
From the Division of Cardiology, University of California, San Diego. Supported by National Institutes of Health Research Grant HL 17682; by Ischemic Heart Disease SCOR grant awarded by the National Heart Lung, and Blood Institute, United States Public Health Service, Department of Health, Education, and Welfare, No. CA 26666; by International Research Fellowships 1 F05 TWO 31544-01 and 1 F05 TWO 03308-01, Fogerty International Center; by the United States Public Health Service, Department of Health, Education, and Welfare; and by the Danish Heart Foundation. Received for publication Jan. 23, 1984; revision received Apr. 25, 1984; accepted May 21, 1984. Reprint requests: John Ross, Jr., M.D., Dept. of Medicine, M-013B, University of California, San Diego, La Jolla, CA 92093.
radiologic criteria for left ventricular failure and hemodynamic findings, particularly the pulmonary capillary wedge pressure measured during AMI?-” Chest x-ray abnormalities have also been employed for prognosis after AMI.” The possibility that indirect assessment of left ventricular function might be obtainable from standard chest x-ray films would be of both economic and clinical importance. Therefore, the purpose of this paper was to investigate the relationship between roentgenographic findings and left ventricular function assessed by a predischarge left ventricular ejection fraction (EF) study in patients with AMI. METHODS Patient
population. Our population consisted of 229 patients discharged after definite AMI. Data concerning these patients were available in a data base maintained by the Specialized Center of Research (SCOR) for Ischemic Heart Disease at the University of California, San Diego, Medical Center. The patients were recruited over a 3-year period from 1979 to 1982 from the University of California, San Diego, Medical Center, the Veterans Administrations Hospital, and the United States Naval Regional 1431
December,
1432
Madsen et al.
American
Heart
1984 Journal
1. Use of dischargechest x-ray pulmonary venous congestiongrade and cardiothoracic ratio to predict abnormal EF
Table
EF <0.51
EF >0.51
134 27
95 8
194
107
a7
219 92
129 61
90 31
127
68
59
Total Pulmonary Total Grade I-IV Grade
venous
0
Cardiothoracic Total >0.50
Absent
Specificity
CICCID-UCY
Predictive value
20.1
91.6
49.8
77.1
47.3
65.6
54.8
66.3
52.7
63.3
57.1
67.3
ratio
<0.50 Combination Total Present
congestion 229 35
Percent Sensitivity
(grade
I-IV 219 101 118
and/or
ratio 70.50) 129 68 61
90 33 57
Medical Center, all in San Diego; and from Vancouver General Hospital in British Columbia, Canada. All patients were admitted to the hospital within 24 hours after onset of symptoms. The diagnosisof AM1 wasestablishedby at least two of the following criteria: (1) characteristic chest pain, (2) elevation of creatine kinase, and (3) ECG changeswith evolution of Q waves (transmural infarction). Nontransmural infarction wasdiagnosedby typical ST segmentor T wave changesaccompaniedby at least criterion (2). No patient in this study had intracoronary or intravenous streptokinase therapy, coronary angioplasty, or coronary bypass surgery between hospital admission and discharge. The age of the patients ranged from 30 to 95 years (mean age 63 years) and 198 of the patients were men (86%). A previous history of myocardial infarction (MI) was registered in 74 patients (32%), previous hypertension wasnoted in 103patients (45%), and previous heart failure was registered in 36 patients (16%). The location of the AM1 was anterior in 68 patients (30%), inferior in 81 patients (35%), indeterminate in 30 patients (35%), and subendocardialin 50 patients (22%). Chest roentgenograms. All patients had at least two chest x-ray examinations done during the hospitalization-one early in the acute phaseshortly after admission and the other just before hospital discharge.Some of the earlier x-ray films were taken with the patient in the supine position, but all the predischargex-ray films were performed with the patient upright and in the usual posterior-anterior projection. The initial x-ray films were exposed after inspiration of about 1 L from functional residual capacity and were timed at end diasto1e.l”The predischargex-ray film was supplementedwith a lateral study. All radiographs were read by individuals unaware
of the radionuclide data. In addition, the radionuclide data were assessedby individuals not familiar with the patient or his clinical course. Pulmonary venous congestion on the chest x-ray film was graded as we have previously reported.” Grade 0 equals no pulmonary venous congestion. Grade I equals pulmonary venous hypertension with redistribution of pulmonary blood flow, defined asa greater diameter of the upper comparedwith the lower lobe pulmonary vessels.If this finding was the only abnormal finding on a supine x-ray film, the study was interpreted as normal (no pulmonary venous hypertension). In the supine position, pulmonary vascular redistribution and either peribronchial cuffing or lossof the right hilar angle were required to consider the radiography grade I. Grade II equals interstitial pulmonary edema,defined by lossof definition of the pulmonary vascular markings in associationwith Kerley’s B lines. Grade III equalslocalized alveolar edema defined as confluent alveolar infiltrates in perihilar areas and lower lung fields. Grade IV equals diffuse alveolar edema defined as diffuse confluent alveolar infiltrates throughout most areasof both lung fields. Cardiomegaly was assessedfrom the discharge x-ray film by the cardiothoracic ratio,12defined as the relation between the transverse diameter of the heart and the internal diameter of the chest. The transverse diameter of the heart wasmeasuredasthe sumof the widest portion of the heart from the right to the left border of the cardiac silhouette at the midline. The internal diameter of the chest wasmeasuredat the level of the highest point on the left hemidiaphraghm. Cardiothoracic ratios above or equal to 0.50 were consideredabnormal.‘* This measurement wastechnically possiblein only 219 patients. Left ventricular EF. Shortly before discharge,left ventricular EF was measuredby radionuclide ventriculogra-
Volume Number
106 6
0.80-
Chest x-ray and LV function
*
... ::* ... ........... .... 0.60- ............ .::.. ..... *:::: ................... ............ 0.50- .....:-:* ............... ........... ..... 0.40_ ....... ..::. .......... ..::.. .::. 0.30.: .:. 0.20 t 0.10
l0.8C
.. I0.70 * “...
’
. :
*-
*
*
0.60I E 0.50 ~ F 2E 0.40 E 5 0.30 2
.I .
. .
0.20
1
1
I
,
0
I 28
P 6
III
194
.
.
.
* t
0.00
1433
A
0.70-
g 2 8 x F Ei 3
in AMI
1
0.00 -
0.30
.
. . .
0.40
0.50
0.60
CARDIOTHORACICRATIO
Fig.
tion for each group. Grade 0: no pulmonary venous congestion. Grade I: pulmonary venous hypertension. Grade II: interstitial pulmonary edema.Grade III: localized alveolar edema.
:.: . ..
.: .-.. * . - . *. . *. --
0.10
PULMONARY CONGESTIONGRADE 1. Ejection fraction (EF) related to chest x-ray pulmonary venous congestiongrade at discharge.Vertical bars with cross-hatchingshow mean and standard devia-
..:
2. Relationship betweenEF and cardiothoracic ratio from dischargechest x-ray film. N = 219 patients. Correlation coefficient = -0.24.
Fig.
RESULTS
phy with the patient in the supine position. After in vivo labeling of red cells with 15 to 20 mCi technetium-99m, multigated equilibrium angiogramswere obtained in the left anterior oblique position for 5 minutes, as previously reported from our laboratory.13,14 Left ventricular EF was calculated from a computer-generated background corrected time-activity curve using a variable region of interest as: (end-diastolic counts minus end-systolic counts)/(end-diastolic counts). An EF below or equal to 0.51wasconsideredabnorma1.3Usually both the discharge chest x-ray examination and the radionuclide ventriculogram were performed the day before discharge. Statistics. The relationship between the chest x-ray findings and the EF values was assessedby analysis of variance with multiple comparisonsamong group means and linear regressionanalysis.For eachset of observations of pulmonary venous congestionand abnormal cardiothoracic ratio we defined: (1) sensitivity equals percent of patients with abnormal EF who had a positive x-ray finding; (2) specificity equals percent of patients with normal EF who had a negative x-ray finding; (3) accuracy equalspercent of patients with true positive (both abnormal EF and abnormal x-ray) and true negative (both normal EF and normal x-ray) findings; and (4) predictive value of abnormal x-ray equalspercentageof patients with an abnormal x-ray who have an abnormal EF.
Discharge EF and chest x-ray. The discharge EF was 0.48 + 0.13 (mean + SD). It was abnormal (
December,
1434
Madsen
American
0.60
[
-
p 0.50 0 $ LL 0.40 zi i= u ; 0.30 -
... ::.: ...
.....:.. ...... ::;::. ..:. ...... .1.. .....::: .......... ..::. ..... ..... ..I;:’ ... ... .i. ::. ..
; 0.10 I
.::.2. ::.:. ...i. ::.:. ;I
DISCUSSION
0.20
0.00’
;
135 -0.005-0.0005-
I
1
I
It
57
23
I
III-l.K 14
0.03 A I
1994 Journal
highest grade of pulmonary venous congestion from any x-ray film during the hospitalization, the mean discharge EF was 0.51 f 0.12 in 135 patients with grade 0, 0.45 + 0.13 in 57 patients with grade I, 0.44 _+ 0.11 in 23 patients with grade II, and 0.36 f 0.14 in 14 patients with grade III to IV (Fig. 3). The difference between patients with grade 0 compared to patients with grade I, II, or III to IV was significant (analysis of variance), and patients with grade III to IV had significantly lower EF compared to patients with grade I.
0.80 0.70
Heart
0.03-
PULMONARY CONGESTIONGRADE Fig. 3. EF related to highest chest x-ray pulmonary
venouscongestiongrade during the entire hospitalization. Vertical bars with cross-hatching show mean and standard deviation for each group. Grade 0: no pulmonary venous congestion.Grade I: pulmonary venous hypertension. Grade II: interstitial pulmonary edema. Grade III: localized alveolar edema. Grade IV: diffuse alveolar edema.
discharge chest x-ray examination (Fig. 1). This shows the considerable overlap between EF values for patients in the grade groups. The mean EF was 0.49 + 0.13 for patients with grade 0, 0.44 + 0.14 for patients with grade I, and 0.40 + 0.14 for patients with grade II (no significant differences). Fig. 2 presents the linear regression between the cardiothoracic ratio and EF. The correlation coefficient was only -0.24. No breakpoint values in either cardiothoracic ratio or EF could be determined from this plot. Discharge EF and worst chest x-ray during hospitalization. Pulmonary venous congestion was present on
at least one chest x-ray film during the hospitalization in 94 patients (41%). This finding yielded a higher sensitivity (52% ), a lower specificity (79%), higher accuracy (61%), and nearly unchanged predictive value (73%) compared with the discharge x-ray examination (20%, 92%, 50%, and 77%, Tables II and I, respectively). When patients were grouped according to the
The prognostic value of the presence of pulmonary venous congestion on the initial chest x-ray film both for early (30-day) and later (l-year) mortality has been established previously.12 Also, assessment of left ventricular EF has been used prognostically after AMI.‘e6 Therefore we endeavored to evaluate the relationship of an abnormal chest x-ray examination and left ventricular EF in AM1 patients. Radiographic evidence of pulmonary venous congestion has been shown to reflect the pulmonary capillary wedge pressure during AMI.s-” Although the correlation between pulmonary congestion and the pulmonary wedge pressure has been good, in some patients a phase lag is evident in which the films remain abnormal for several days after the wedge pressure has returned to norma1.l’ The cardiothoracic ratio has been reported to yield 70% accuracy for correctly predicting a combined measurement incorporating left ventricular mass, enddiastolic volume, and left atria1 volume obtained from quantitative biplane angiography, with a correlation coefficient of 0,64.15 Discharge
EF and chest
x-ray in the present
study.
This study has shown that whereas many patients (59%) had an abnormal EF at hospital discharge, in only one of five patients was it associated with pulmonary venous congestion on the discharge chest x-ray film. In 9 out of 10 patients with a normal EF, the x-ray film showed no pulmonary venous congestion. In the relatively
few patients in whom the
discharge chest x-ray film showed pulmonary venous congestion, three out of four also had an abnormal EF (predictive value 77 % ). If the x-ray film with the highest grade of pulmonary venous congestion from the hospitalization was utilized, the sensitivity increased from 20% to 52% but was still unsuitably low. Also, the specificity fell and the predictive value was not improved. The combination of pulmonary venous congestion and abnormal cardiothoracic ratio did not improve
Volume
106
Number
6
Chest x-ray
and LV function
II. Use of highest chest x-ray pulmonary venous congestion grade during the entire hospitalization abnormal EF .
Table
Total
EF <0.51
EF >0.51
Total Grade I-IV
229 94
134 69
25
Grade 0
135
65
70
Sensitivity
Specificity
Percent accuracy
in AMI
1435
to predict Predictive value
95 51.5
the ability to predict an abnormal EF appreciably (Fig. 2). The correlation coefficient was only -0.24 compared to previously reported correlation coefficients of -0.435 and -0.29.7 This is, however, not too surprising since the cardiothoracic ratio not only takes the transverse cardiac diameter (which includes both ventricles and the right atrium) into account, but also the transverse diameter of the chest as an indicator of body build. This measurement thus introduces a variable that is totally independent of whether there is cardiac disease or not. We defined a cardiothoracic ratio of ~0.50 to be abnormal, a value generally used. A higher value of 0.60 would have decreased the specificity but would not have improved the predictive value for an abnormal EF in these patients (Fig. 2). Our results indicate that pulmonary venous congestion on a chest x-ray film from the hospitalization or at discharge is indicative of abnormal EF (predictive values of 73 % and 77 % , respectively). However, most patients with an abnormal discharge EF will have normal discharge chest x-ray films (Table I, Fig. 1). In the patients studied, a normal cardiothoracic ratio on the discharge chest x-ray film was often accompanied by an abnormal EF, and a higher ratio often was present in patients with a normal EF.5 The latter finding may be explained partly by a high incidence of previous hypertension in our patients (45 % ). Thus, cardiomegaly on x-ray examination correctly indicated depressed left ventricular function in 66% of the patients. EF, chest x-ray, and heart failure. Previous comparisons between clinical heart failure c1ass,16 which included chest x-ray evidence of pulmonary venous congestion criteria, have shown that the EF measured in AMI patients early after hospital admission was significantly lower in patients with left heart failure and pulmonary edema compared to patients without left heart failure.‘, 3 Most patients who developed shock or died within 3 months had a low EF.6 Our findings at discharge show a significantly lower mean EF in patients with grade I, II, or III to IV pulmonary congestion on chest x-ray film compared to patients with grade 0, and in patients with
73.1
60.7
73.4
grade I compared to patients in grade III to IV. However, as shown previously,‘v3 although the differences in the mean EF values are significant, the overlap of values among the grade groups is great. This makes prediction of an abnormal EF for the individual patient unreliable. In recent studies, we have noted that a phase lag may occur acutely after an experimentally induced elevation in left atrial pressure.17* l8 Nonetheless, sustained left atrial hypertension will invariably result in interstitial and alveolar pulmonary edema, both experimentally and clinically.‘% lg An abnormal radiographic finding coupled with normal cardiac function and size is confusing. It has been recently proposedzO that this may in part reflect noncardiogenie pulmonary edema during acute coronary occlusion. This leaky membrane-induced edema during AM1 was independent of the degree of pulmonary microvascular hydrostatic pressure, and could be substantially minimized by the administration of indomethacin. Thus, if the etiology of radiographic changes in AM1 is not solely hemodynamic, it is less surprising that it is not possible to consistently predict the underlying hemodynamic status from a chest x-ray film. Conclusions. The value of the chest x-ray examination for the assessment of left ventricular function estimated by EF in AM1 patients is limited. Correct assessment of left ventricular function at hospital discharge requires more direct measurements than chest x-ray findings.
REFERENCES
1. Schelbert HR, Henning H, Ashburn WL, Verba JW, Karliner JS, O’Rourke RA: Serial measurements of left ventricular ejection fraction by radionuclide angiography early and late after myocardial infarction. Am J Cardiol 38:407, 1976. 2. Schulze RA, Strauss HW, Pitt B: Sudden death in the year following myocardial infarction. Am J Med 62:192, 1977. 3. Battler A, Slutsky R, Karliner J, Froelicher V, Ashburn W, Ross J Jr: Left ventricular ejection fraction and first third ejection fraction early after acute myocardial infarction: Value for predicting mortality and morbidity. Am J Cardiol 45:197, 1980. 4. Shah PK, Pichler M, Berman D, Singh BN, Swan HJC: Left ventricular ejection fraction determined by radionuclide ven-
December,
1436
5.
6.
7.
8.
9.
10.
11.
12.
Madsen et al.
triculography in early stages of first transmural myocardial infarction. Am J Cardiol 45:542, 1980. Feild BJ, Russell RO, Moraski RE, Soto B, Hood WP, Burdeshaw JA, Smith M, Maurer BJ, Rackley CE: Left ventricular size and function and heart size in the year following myocardial infarction. Circulation %X331. 1974. Rigo P, Murray M, Strauss HW, Taylor D, Kelly D, Weisfeldt M, Pitt B: Left ventricular function in acute myocardial infarction evaluated by gated scintiphotography. Circulation 50:678, 1974. Dewhurst NG, Hannan WJ, Muir AL: Prognostic value of radionuclide ventriculography after myocardial infarction. Q J Med 49:479, 1981. Nemerovski M, Shah PK, Pichler M, Berman DS, Shellock F, Swan HJC: Radionuclide assessment of sequential changes in left and right ventricular function following first acute transmural myocardial infarction. AM HEART J 104:709, 1982. Lassers BW, George M, Anderton JL, Higgins MR, Philip T: Left ventricular failure in acute myocardial infarction. AM HEART J 25:511, 1970. McHugh TJ, Forrester JS, Adler L, Zion D, Swan HJC: Pulmonary vascular congestion in acute myocardial infarction: Hemodynamic and radiologic correlations. Ann Intern Med 76:29, 1972. Kostuk W, Barr JW, Simon AL, Ross .J Jr: Correlations between the chest film and hemodynamics in acute myocardial infarction. Circulation 48:624, 1973. Battler A, Karliner JS, Higgins CB, Slutsky R, Gilpin E, Froelicher VF, Ross J Jr: The initial chest x-ray in acute myocardial infarction. Circulation 61:1004, 1980.
American
1984
Heart Journal
13. Ashburn WL, Schelbert HR, Verba JW: Left ventricular ejection fraction-A review of several radionuclide angiographic approaches using the scintillation camera. Prog Cardiovasc Dis 20:267, 1978. 14. Pfisterer ME, Ricci DR, Schuler G, Swanson SS, Gordon DG, Peterson KE, Ashburn WL: Validity of left ventricular ejection fractions measured at rest and peak exercise by equilibrium radionuclide angiography using short acquisition times. J Nucl Med 20:484, 1979. 15. Clover L, Baxley WA, Dodge HT: A quantitative evaluation of heart size measurements from chest roentgenograms. Circulation 47:1289, 1973. 16. Killip T, Kimball JT: Treatment of myocardial infarction in a coronary care unit. Am J Cardiol 20:457, 1967. 17. Slutsky R, Higgins C: Intravascular and extravascular pulmonary fluid volumes. II. The response to rapid left atria1 hypertension and the theoretical implications for radiographic and radionuclide images. Invest Radio1 181:33, 1983. 18. Slutsky R, Higgins C: Intravascular and extravascular pulmonary fluid volumes. III. Response to sustained left atria1 hypertension. Cardiovasc Res 17:113, 1983. R, Slutsky R, Shabatai R, Higgins C: Extravascular 19. Grover lung water in patients with congestive heart failure. A comparison of patients with acute and chronic left heart disease. Radiology 147:659, 1983. 20. Richeson JF, Paulshock C, Yu P: Non-hydrostatic pulmonary edema after coronary ligation in dogs: Protective effect of indomethacin. Circ Res 50:301, 1982.