Risk Factors, Angiographic Patterns, and Outcomes in Patients With Ventricular Septal Defect Complicating Acute Myocardial Infarction
Background—Ventricular septal defect (VSD) complicating acute myocardial infarction has been studied primarily in small, prethrombolytic-era trials. Our goal was to determine clinical predictors and angiographic and clinical outcomes of this complication in the thrombolytic era.
Methods and Results—We compared enrollment characteristics, angiographic patterns, and outcomes (30-day and 1-year mortality) of patients enrolled in the Global Utilization of Streptokinase and TPA for Occluded Coronary Arteries (GUSTO-I) trial with and without a confirmed diagnosis of VSD. Univariable and multivariable analyses were used to assess relations between enrollment factors and the development of VSD. In all, 84 of the 41 021 patients (0.2%) developed VSD, a smaller percentage than reported in the prethrombolytic era. The median time from symptom onset to VSD diagnosis was 1 day. Enrollment factors most associated with this complication were advanced age, anterior infarction, female sex, and no previous smoking. The infarct artery was more often the left anterior descending and more likely to be totally occluded in patients who developed VSD. Mortality at 30 days was higher in patients with VSDs than in those without this complication (73.8% versus 6.8%, P<0.001). Patients with VSDs selected for surgical repair (n=34) had better outcomes than patients treated medically (n=35; 30-day mortality, 47% versus 94%).
Conclusions—Compared with historical control subjects, patients who undergo thrombolysis within 6 hours of infarction onset may have a reduced risk of later VSD. If patients develop this mechanical complication, however, it typically occurs sooner than described in the prethrombolytic era. Despite improvements in medical therapy and percutaneous and surgical techniques, mortality with this complication remains extremely high.
Ventricular septal defect (VSD) complicating acute myocardial infarction (MI) is uncommon with a reported incidence in the prethrombolytic era of ≈1% to 2%.1 2 It typically occurs in the first week after infarction, with a mean time from symptom onset of 3 to 5 days.1 3 4 Previous investigations have found age and female sex to be risk factors for its development5 6 ; such patients also commonly have no prior angina or MI.1 7 8 Angiographically, patients with VSD have been noted to have total occlusion of the infarct-related artery with minimal collaterals.6 9 10
In the prethrombolytic era, outcomes after the development of VSD were extremely poor, with an in-hospital mortality of ≈45% in surgically treated patients and ≈90% in those managed medically.1 2 11 12 13 Poor prognostic factors in this patient population included the development of cardiogenic shock, right ventricular dysfunction, and inferior infarct location.4 5 14 15
Postinfarction VSD has not been well studied in the thrombolytic era, and most prior studies have been relatively small. The large patient population from the Global Utilization of Streptokinase and TPA for Occluded Coronary Arteries (GUSTO-I) trial—41 021 patients enrolled at 1081 institutions in 15 countries—provides a unique opportunity to explore this topic more fully.16 17 The goals of the present study were to establish the incidence of VSD after acute MI in patients treated with thrombolytic therapy and to identify the enrollment characteristics and angiographic patterns associated with its occurrence. The association of this complication with patient outcomes also was to be evaluated. Finally, we wanted to identify enrollment characteristics associated with increased mortality in patients who develop VSD.
As described,16 enrollment criteria for the GUSTO-I trial included presentation to a participating hospital within 6 hours after the onset of symptoms, with chest pain lasting ≥20 minutes and accompanied by ≥0.1-mV ST-segment elevation in ≥2 limb leads or ≥0.2 mV in ≥2 contiguous precordial leads. Exclusion criteria included previous stroke, active bleeding, and recent trauma or major surgery. Informed consent was obtained before enrollment.
Patients were randomized to receive streptokinase with subcutaneous heparin, streptokinase with intravenous heparin, accelerated alteplase with intravenous heparin, or the combination of alteplase and streptokinase with intravenous heparin.16 Subcutaneous heparin (12 500 U twice daily) was continued for 7 days or until discharge; intravenous heparin (5000-U bolus, 1000 U/h, adjusted to maintain an activated partial thromboplastin time of 60 to 85 seconds) was given for 48 hours or longer at the investigator’s discretion. Chewable aspirin (≥160 mg) was given as soon as possible and daily thereafter (160 to 325 mg/d). Patients without contraindication to β-blockers were treated with atenolol (10 mg IV given in 2 doses), followed by 50 to 100 mg orally daily. All other medications (including calcium channel antagonists, nitrates, ACE inhibitors, and antiarrhythmics) and procedures (including coronary angiography, angioplasty, and bypass surgery) were at the discretion of the attending physician.
Identification of Patients With VSD
Patients with VSDs were first identified by review of case report form data (check box for VSD). Ancillary questionnaires were then distributed to sites with potential VSD patients to obtain further information on their diagnosis and treatment. These forms, along with all available source documents, were then reviewed. VSD typically was diagnosed by echocardiography (disrupted ventricular septum with evidence of left-to-right shunt by color Doppler), cardiac catheterization (evidence of left-to-right shunt by ventriculography), and/or pulmonary artery catheterization (increase in oxygen saturation from the right atrium to the right ventricle, quantification of which was unavailable for most patients). Patients thought to possibly have VSDs on the basis of cardiac arrest or pulseless electrical activity were excluded. Data for the development of VSD were complete for all 41 021 patients. For the purposes of this article, patients were divided into 2 groups: those with confirmed VSDs and those without VSD confirmation.
In patients not randomized to the angiographic substudy, angiography was performed per institutional protocols. Patients enrolled in the angiographic substudy underwent angiography as described.17 Ejection fractions were calculated by the area-length method.18
The primary end point of this study was diagnosis of VSD as defined earlier. Of primary interest were the relations of enrollment clinical factors to the occurrence of this outcome. Also considered were the relations between angiographic factors and VSD. Second, we investigated the effect of VSD on other prospectively defined, in-hospital clinical outcomes: reinfarction, shock, stroke, ischemia, and congestive heart failure or pulmonary edema. Mortality 30 days and 1 year after randomization also was assessed.
Each categorical factor is described as the number and percentage of patients with that characteristic. Continuous factors are described as medians with 25th and 75th percentiles. With only 84 events, it was necessary to diminish as much as possible the chance of finding spurious results. All formal statistical analyses were therefore reserved for the prospectively specified analysis of baseline (enrollment) predictors of VSD. The enrollment factors identified as having potential prognostic significance were age, sex, infarct location, smoking history, previous MI, heart rate, current smoking, hypertension, systolic and diastolic blood pressures, Killip class, previous bypass surgery, and time from symptom onset to treatment. Other factors, including the angiographic results, are presented for descriptive purposes only.
Logistic regression modeling was used for the predictive analysis. Spline transformations of each continuous factor versus outcome were evaluated to determine the appropriateness of the linearity assumption. Linear splines were applied to systolic and diastolic blood pressures.
A full model containing all the potential predictors was first fitted. Then, the relative significance of each factor in the full model was plotted as the Wald χ2 for each factor minus the corresponding degrees of freedom. A reduced model was determined by use of stepwise variable-reduction techniques. This model contains factors that are all multivariably significant at P<0.05. The predictive ability of the model was expressed by inclusion of a C index, the area under the receiver-operator characteristic curve.
The 1-year event rates for patients with and without VSDs and for VSD patients who did and did not undergo surgical repair were calculated by use of Kaplan-Meier survival estimates. Log-rank statistics were computed to determine the significance of the differences in the 1-year curves.
A total of 84 patients were identified as having confirmed VSDs (0.2% of the total GUSTO-I population). In 43 of 50 patients (86%), VSD was suspected by physical examination. The methods of diagnosis are shown in Table 1⇓. VSD was diagnosed most often by echocardiography alone (21 patients). The median time from MI symptom onset to VSD diagnosis was 1 day (range, 0 to 47 days); 94% of the cases were diagnosed within 1 week.
Patients who developed VSDs were more likely to be older, female, and hypertensive; have no history of smoking; and have anterior infarction, an increased heart rate, and a worse Killip class at admission (Table 2⇓). The relations between enrollment systolic and diastolic blood pressures and VSD also was significant but not linear. The likelihood of VSD decreased with increasing systolic pressure up to ≈130 mm Hg. Beyond 130 mm Hg, the likelihood of VSD increased with increasing systolic pressure. Similarly, the likelihood of VSD decreased with increasing diastolic blood pressure to ≈75 mm Hg and then increased with increasing diastolic pressure >75 mm Hg. The time from symptom onset to thrombolytic administration did not differ between groups.
Angiographic data were available for 50 of 84 patients (60%) diagnosed with VSDs (Table 3⇓). These patients underwent first angiography at a median of 21 hours after onset of the index infarction, in contrast to the median of 95 hours for GUSTO-I patients who did not develop VSDs. Patients with VSDs were more likely to have TIMI grade 0 or 1 flow at first angiography, and 57% had total occlusion of the infarct artery. The infarct-related artery was more often the left anterior descending in patients with VSDs than in patients with no VSDs. More than 50% of the patients with VSDs had 2- or 3-vessel coronary disease. The median ejection fraction was lower in patients with VSDs.
Univariable and Multivariable Modeling
The complete set of variables included in the model to predict VSD and their relative importance are shown in Table 4⇓ and the Figure⇓. Increasing age, anterior infarction, and female sex were the most important multivariable predictors of VSD. Current smoking and prior MI also were significant in being associated with not developing this complication. A reduced version of the model (containing only the significant multivariable predictors) is shown in Table 5⇓. The reduced model performed well with a C index of 0.774.
In-Hospital Procedures and 30-Day and 1-Year Outcomes
Patients who developed VSDs showed a much greater use of in-hospital procedures than those who did not; they also were more likely to develop shock and congestive heart failure (Table 6⇓). Patients with VSDs also had significantly higher mortality at 30 days and 1 year. Data regarding treatment of VSD were available for 69 of the 84 patients. In all, 34 underwent surgical repair a median of 3.5 days (95% CI, 1 to 7) after MI symptom onset, including 3 (10%) who underwent repair between 30 days and 1 year after enrollment. Patients who underwent surgical repair had lower mortality at 30 days and 1 year than the 35 patients who were treated medically: 47% versus 94% at 30 days (P<0.001) and 53% versus 97% at 1 year (P<0.001).
Patients who developed VSDs and died within 30 days were more likely to be female and to have an inferior infarction than those who survived (Table 7⇓). All VSD patients who had pulmonary congestion (Killip class III or IV) at admission died within 30 days; the mortality rate was 27% among VSD patients who were in Killip class I or II.
VSD remains an infrequent but devastating complication of acute MI. Its incidence in GUSTO-I was lower than in previous reports, which may reflect 2 factors. First, all patients in this trial were treated with thrombolytics within 6 hours of symptom onset. Reperfusion therapy, particularly if begun early, may prevent the extensive myocardial necrosis typically associated with mechanical complications.19 Second, we included only patients with confirmed diagnoses of VSD to better define enrollment characteristics and overall outcomes associated with this complication. Although this may have caused an underestimation of the incidence of VSD (by excluding patients who died before the diagnosis could be confirmed), it may be more useful clinically because it provides information on outcomes once the diagnosis has been made.
The mean time from infarction to development of VSD was found to be 1 day, shorter than the 3 to 5 days reported from the prethrombolytic era.1 3 4 Becker and colleagues,20 21 who reported a time similar to that of the present study, postulated that this may be due to different pathophysiological mechanisms. Thrombolytic therapy may prevent extensive transmural necrosis, a prerequisite for cardiac rupture in the prethrombolytic era. However, thrombolysis also may cause myocardial hemorrhage during the “lytic state,” so that if VSD occurs, its time course would be accelerated.
Advanced age, anterior infarct location, female sex, and no current smoking were found to be the most important predictors of VSD. Unlike previous studies, hypertension21 and no previous MI or angina7 were found to be less helpful when all other variables were considered. There was a bidirectional association of systolic and diastolic blood pressures at enrollment with the incidence of VSD. The positive correlations (increase in the incidence of VSD as systolic blood pressure increased to >130 mm Hg and the diastolic blood pressure to >75 mm Hg) reflect the association between hypertension and VSD. Extensive MI and right ventricular involvement, both known risk factors for VSD, may cause hypotension and cardiogenic shock on admission. The negative correlations between enrollment systolic (≤130 mm Hg) and diastolic (≤75 mm Hg) blood pressures with the incidence of VSD probably reflect the incidence of hemodynamic compromise associated with extensive MI or right ventricular infarction.
Our angiographic data, consistent with previous studies, show that patients who develop VSDs after acute MI are more likely to have total occlusion of the infarct artery.6 10 22 This suggests that the pathophysiology of acute VSD involves sudden, severe ischemia, leading to extensive myocardial necrosis, and that patients who do not reperfuse with thrombolysis are at increased risk of mechanical complications. We also found that this patient population had extensive coronary artery disease (51% with 2- or 3-vessel disease) and poor left ventricular function.
Mortality with this complication remains extremely high in the thrombolytic era, despite improvements in medical therapy and percutaneous and surgical techniques. However, the mortality rates for all patients with VSDs were similar at 30 days and 1 year (74% and 78%). This suggests that if the patient survives the initial admission, the long-term prognosis is relatively good.
Although this is the largest study of VSD in a prospective trial of thrombolysis, the number of patients with VSDs was insufficient to fully determine enrollment characteristics associated with survival. However, it appeared that patients with inferior infarcts and VSDs tended to have a worse outcome than those with anterior infarcts. This is most likely related to factors previously noted by Edwards and colleagues.3 In their necropsy study, inferior infarcts were more likely to be associated with complex VSDs (multiple, irregular, and/or variable interventricular connections) located in the inferobasal portion of the septum and therefore were more difficult to approach surgically. Anterior infarcts were more commonly associated with simple, through-and-through defects in the apical septum, which tend to be more easily repaired. Right ventricular infarction and dysfunction, more commonly associated with inferior infarcts, also have been shown to be poor prognostic factors in patients with VSD.4 9 14 15 All patients in our study with Killip class III or IV at presentation died. Nonsurvivors also were older and more likely to have reduced blood pressure at enrollment.
Patients with VSDs in GUSTO-I selected for surgical repair had better outcomes than those treated medically. However, those who did not undergo surgical repair may have been more critically ill; these patients had a worse Killip class, tended to be slightly older, and had a lower enrollment blood pressure, all of which have been shown to correlate with mortality.23 It is worth noting that the 30-day mortality of patients not having surgical repair was 94%.
The relation of cardiac rupture to timing of thrombolytic administration is controversial. Some studies have shown an increased risk with late therapy (ie, >12 hours after symptom onset),18 but more recent evidence does not support this finding.21 We did not find a significant difference in the timing of thrombolytic administration between those who developed VSD and those who did not (3.1 versus 2.8 hours). However, enrollment criteria for GUSTO-I included a 6-hour limit from symptom onset; therefore, we cannot exclude the possibility of increased risk with thrombolytic administration after that period.16
The number of patients in the present study, although 1 of the largest series reported, is still relatively small with limited statistical power. Additionally, the diagnosis of VSD was not made according to prospective criteria but rather at each investigator’s discretion. Thus, some patients may have been misclassified as either having or not having VSDs. However, after review of source documents and ancillary questionnaires, it was clear that most patients had been diagnosed with standard cardiac catheterization, pulmonary artery catheterization, or echocardiography. Although there was a mortality difference between patients with VSDs treated surgically and those treated medically, this may have been due at least partly to confounding factors. We could not explore this further because of the small VSD sample.
In conclusion, the incidence of VSD after acute MI appears to have declined in the thrombolytic era, most likely because of improved reperfusion and myocardial salvage. Although VSD is occurring with decreased frequency, the timing appears to be accelerated in the thrombolytic era, possibly related to intramyocardial hemorrhage. Advanced age, anterior location of infarction, female sex, and no history of smoking are the most important predictors of this complication. After VSD has developed, advanced age and inferior location of infarction may be the most important prognostic factors in this patient population. Outcome remains extremely poor, with a mortality rate of ≈50% in patients undergoing surgical repair and ≈95% in those treated medically. More effective ways to predict, prevent, and treat this devastating complication are needed.
This study was funded by grants from Genentech, South San Francisco, Calif; Bayer Corporation, New York, NY; CIBA-Corning, Medfield, Mass; ICI Pharmaceuticals, Wilmington, Del; and Sanofi Pharmaceuticals, Paris, France.
- Received May 27, 1999.
- Revision received August 3, 1999.
- Accepted August 5, 1999.
- Copyright © 2000 by American Heart Association
Moore CA, Nygaard TW, Kaiser DL, Cooper AA, Gibson RS. Postinfarction ventricular septal rupture: the importance of location of infarction and right ventricular function in determining survival. Circulation. 1986;74:45–55.
Feneley MP, Chang VP, O’Rourke MF. Myocardial rupture after acute myocardial infarction: ten year review. Br Heart J. 1983;49:550–556.
Skehan JD, Carey C, Norrell MS, DeBelder M, Balcon R, Mills PG. Patterns of coronary artery disease in post-infarction ventricular septal rupture. Br Heart J. 1989;62:268–272.
Pohjola-Sintonen S, Muller JE, Stone PH, Willich SN, Antman EM, Davis VG, Parker CB, Braunwald E, for the MILIS Study Group. Ventricular septal and free wall rupture complicating acute myocardial infarction: experience in the multicenter investigation of limitation of infarct size. Am Heart J. 1989;117:809–816.
Radford MJ, Johnson RA, Daggett WM, Fallon JT, Buckley MJ, Gold HK, Leinbach RC. Ventricular septal rupture: a review of clinical and physiologic features and an analysis of survival. Circulation. 1981;64:545–553.
Cummings RG, Reimer KA, Califf R, Hackel D, Boswick J, Lowe JE. Quantitative analysis of right and left ventricular infarction in the presence of postinfarction ventricular septal defect. Circulation. 1988;77:33–42.
Bolooki HB. Emergency cardiac procedures in patients in cardiogenic shock due to complications of coronary artery disease. Circulation. 1989;79(suppl I):I-137–I-148.
Becker RC, Gore JM, Lambrew C, Weaver WD, Rubinson RM, French WJ, Tiefenbrunn AJ, Bowlby LJ, Rogers WJ, for the National Registry of Myocardial Infarction Participants. A composite view of cardiac rupture in the United States National Registry of Myocardial Infarction. J Am Coll Cardiol. 1996;27:1321–1326.
Lee KL, Woodlief LH, Topol EJ, Weaver WD, Betriu A, Col J, Simoons ML, Aylward PE, Van de Werf F, Califf RM, for the GUSTO I Investigators. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction: results from an international trial of 41 021 patients. Circulation. 1995;91:1659–1668.