Author + information
- Received May 26, 2016
- Revision received March 6, 2017
- Accepted March 23, 2017
- Published online June 19, 2017.
- Mark C.K. Hamilton, MBChBa,∗ (, )
- Jonathan C.L. Rodrigues, MBChBa,
- Robin P. Martin, MBChBb,
- Nathan E. Manghat, MDa and
- Mark S. Turner, PhDb
- aDepartment of Radiology, Bristol Royal Infirmary, University Hospitals Bristol, Bristol, United Kingdom
- bDepartment of Cardiology, Bristol Royal Infirmary, University Hospitals Bristol, Bristol, United Kingdom
- ↵∗Address for correspondence:
Dr. Mark C.K. Hamilton, Department of Radiology, Marlborough Street, Bristol Royal Infirmary, Bristol BS28HW, United Kingdom.
Objectives The aim of this study was to define the dynamic in vivo morphology of post-infarct ventricular septal defect (PIVSD), which has not been previously described in living patients.
Background PIVSD is a devastating complication of acute myocardial infarction.
Methods The anatomic features of PIVSD, as demonstrated by computed tomography or magnetic resonance imaging, were retrospectively reviewed.
Results Thirty-two PIVSDs were assessed, 16 left coronary artery and 16 right coronary artery PIVSDs. PIVSDs were large (mean maximum dimension 26.5 ± 11.5 mm, mean area 5.2 ± 4.2 cm2) and oval (mean eccentricity index 1.7 ± 0.5), with thin margins (diastolic mean thickness 5 mm from the edge of the PIVSD 6.4 ± 3.0mm), and only 22% of PIVSDs were entirely confined to the septum. The defects could be larger in diastole or systole. The stem of the largest available Amplatzer occluder stem (St. Jude Medical, St. Paul, Minnesota) filled only 50% of defects. Patients with small defects may survive without closure. Without closure, those with large defects die. If accepted for closure, PIVSD size and coronary territory did not predict survival >1 year (overall 60%).
Conclusions This is the first detailed anatomic description of PIVSD in living patients. Defects may be larger in systole or diastole, meaning that single-phase measurement is unsuitable. Its complex nature means that the most commonly available occluder device is frequently unsuitable. Successful closure leads to prolonged survival and should be attempted where possible. This study may lead to improved patient selection, closure techniques, and device design.
Post-infarct ventricular septal defect (PIVSD) is a devastating complication of acute myocardial infarction (MI). In the pre-reperfusion era, the prevalence was 1% to 3%, dropping to 0.2% with reperfusion, which although it reduces incidence probably accelerates development (1–4). PIVSD is associated with large infarcts (5) and without anatomic correction has a very high mortality rate, 94% medical versus 47% surgical 30-day mortality (1,6). Inferior septal defects are reported as carrying a particularly high risk for mortality even with surgery, which is associated with more right ventricular (RV) involvement (7). Early surgical intervention is often avoided because it is associated with a high mortality rate (>70%) (8).
Transcatheter closure of PIVSD is feasible (4,9) with 1 widely used device, the Amplatzer PIVSD occluder (St. Jude Medical, St. Paul, Minnesota), which can be a successful alternative to surgery (10). Percutaneous closure gives similar results to surgery, with a 30-day mortality rate of 65% when early closure was attempted (11) and 42% mortality at discharge in a series including some prior attempted surgical closures (12). There was a 30-day mortality rate of 28% in a group including patients who underwent early (≤6 days post-MI) and late (≥95 days post-MI) closure (13), and a small series of percutaneous closures suggested that delayed intervention may be better (14). Some have proposed the use of the Amplatzer device for smaller PIVSDs (<15 mm on transthoracic echocardiography) (15) and considered PIVSDs >24 mm unsuitable (13). Neither of these papers described the anatomy of PIVSDs in detail.
We describe the dynamic in vivo anatomy of acute PIVSDs, which represents the pathology of patients presenting for treatment.
Consecutive patients from 2006 to 2016 with PIVSDs presenting to our tertiary referral institution were reviewed. After diagnosis by transthoracic echocardiography, tomographic imaging (with magnetic resonance imaging [MRI] or computed tomography [CT]) was performed to assess morphology and suitability for closure (16).
Initially, MRI (Symphony and Avanto 1.5-T, Siemens Medical Solutions, Erlangen, Germany) was used for assessment of PIVSDs. Breath-held steady-state free precession short-axis cine images of the ventricles were obtained, with long- and short-axis imaging through the defect. Typical parameters were field of view 340 mm, echo time 1.12 ms, repetition time 39.45 ms, and voxel size 1.8 × 1.8 × 8 mm.
Electrocardiographically gated CT
Speed and ability to image patients in unstable condition meant that CT became the preferred imaging modality. Of the 26 computed tomographic examinations acquired locally, 9 were acquired with 1-mm voxels (Sensation 16, Siemens Medical Solutions), 16 with 0.65- to 0.7-mm voxels (Definition AS+, Siemens Medical Solutions), and 1 with 0.5-mm voxels (Aquilion One Vision, Toshiba, Tokyo, Japan). Breath holding was preferred, although gentle breathing was allowed. Patients were scanned regardless of heart rate without the administration of rate-control medication. Retrospective gating, mostly without dose modulation, was used to enable reconstruction of the highest quality datasets. Tube voltage of 100 to 120 kVp was used depending on patient body habitus. Iodinated contrast agent (80 to 100 ml, 300 to 400 mg/ml strength) was administered at 4 to 5 ml/s with bolus tracking in the ascending aorta (trigger 120 Hounsfield units) and a 6-s delay. When available, data were reconstructed at 10% intervals throughout the cardiac cycle. Radiation dose (millisieverts) was calculated by multiplying the dose-length product by the cardiac conversion factor (0.028) (17).
Anonymized data were retrospectively analyzed using the OsiriX Digital Imaging and Communications in Medicine viewer (Pixmeo, Geneva, Switzerland) by a cardiac radiologist with >10 years’ experience in cardiac CT and MRI. Whenever possible, all PIVSDs were evaluated at end-diastole and end-systole, as determined by maximum and minimum cavity size. With CT, long- and short-axis views of the PIVSD were obtained; an en face view was used to measure area and to adjust the biplane views so the margins of the defect were appropriately orientated. Biplane long- and short-axis dimension and en face area were assessed, where device closure was anticipated, on the left ventricular (LV) side of the septum (Figure 1). The measurements were repeated on the RV side of the septum, care being taken to ensure measurements were obtained where there was a circumferential closure zone. With MRI, long- and short-axis images of the defect were obtained (en face views were obtained but considered imperfect for comparative measurements). In all cases, insubstantial and trabeculated tissue was included as part of the defect. PIVSDs were assessed for segmental involvement, size, eccentricity, myocardial wall thickness, and intramyocardial dissection (ID) out of the septum.
The Amplatzer PIVSD device is made of nitinol wire, with 2 discs connected by a 10-mm stem (waist) covered with polyester fabric. The nominal size of the device is its waist diameter (18 to 24 mm), with ventricular discs 10 mm wider. Because the waist is round, an “eccentricity index” was calculated by dividing the maximum by the minimum PIVSD dimension; a value >1 implies that the PIVSD is not round.
Pragmatically, we considered that a 5-mm thickness of myocardium was likely to provide an adequate seal against an occluder with a 10-mm stem and give information about ventricular morphology for device design; this did not influence decision making regarding closure. Thus, the distance from the defect to 5-mm-thick myocardium was measured. In a thin apex, the measurement was extended around the apex. Myocardial thickness at a distance of 5 mm from the anterior, inferior, basal, and apical margins of the PIVSD was assessed, as this would be covered by the disc.
PIVSDs have been previously described as simple if there is a direct connection between the 2 ventricles and complex if they have a serpiginous course (18); this is subjective and unhelpful for detailed comparative analysis. Reflecting the challenges of closure, we defined septal rupture as complex if not wholly surrounded by septum. In complex PIVSD, the septum is torn off the external wall of the heart, which remains intact or the tear passes into adjacent myocardium, which is defined as ID (Figure 2), and therefore represents pathology that could progress to rupture into the pericardium. This is identified by following the plane of the septum (Figure 3) through the defect margins.
The infarct-related artery defined whether the defect was a right coronary artery (RCA) PIVSD or left coronary artery (LCA) PIVSD.
The Shapiro-Wilk test was used to assess distribution. Normally distributed continuous data were interrogated using either paired Student t tests (e.g., when comparing end-diastolic and end-systolic measurements in the same patient) or unpaired Student t tests (e.g., when comparing LCA vs. RCA PIVSD) as appropriate. Skewed data were interrogated using the Mann-Whitney U or Wilcoxon signed rank test, depending on whether the data were paired, as appropriate. Statistical significance was set at a 2-tailed p value <0.05. Pearson’s correlation coefficient (R) was used to measure the strength of a linear association between 2 variables. Univariate and multivariate logistic regression analysis was performed to determine if baseline parameters could predict 30-day and 1-year survival post-closure of PIVSDs.
There was a total of 36 patients. Thirty-one had cross-sectional imaging available for review. Of these 31 patients, 5 underwent MRI and 26 underwent CT. One LCA PIVSD had a single orifice on the left but a double orifice on the right side of the septum (considered a single defect), and another had a double orifice on both sides (considered 2 defects), resulting in 32 PIVSDs (16 post–LCA infarction, 16 post–RCA infarction) in 31 patients.
Three of 26 computed tomographic studies had only diastolic-phase imaging and were excluded from data comparing systole and diastole. Thus, 28 patients (23 CT and 5 MRI) with 29 PIVSDs (15 left, 14 right) had comparable systolic and diastolic imaging. In 3 of the cases with CT, the diastolic images chosen for morphological assessment were mid to late diastolic (diastolic phase <80%).
Morphological data encompass all 31 patients. Table 1 excludes a 72-year-old man with a 17-mm chronic RCA PIVSD successfully closed percutaneously 523 days post-infarct.
The mean radiation dose was 18.4 ± 8.45 mSv. The highest heart rate imaged was 123 beats/min. All studies were diagnostic, with no known scan-related complications.
Myocardial segmental involvement
Of the 32 PIVSDs, none involved the basal anterior septum, 11 (34%) involved the basal inferior septum (11 RCA) with 1 (3%) confluent with the mitral valve annulus, 5 (16%) involved the mid anterior septum (5 LCA), 17 (53%) involved the mid inferior septum (16 RCA, 1 LCA), and 13 (41%) involved the apical septum (13 LCA) (Figure 4). LCA PIVSDs usually involved the apical septum (81%). RCA PIVSDs usually involved the basal (69%) and/or mid inferior (100%) septum.
Dimension, area, and sphericity
Overall, the 32 PIVSDs were large on the LV side of the septum and oval (Table 2). The LCA PIVSDs were significantly smaller by dimension and area than the RCA PIVSDs. The LV side of the defects was significantly larger than the RV side.
Over the cases with comparative systolic and diastolic data, there was no significant change in size of either the LV or RV side of the defects between systole and diastole. Although overall of no statistical significance, it was noted that the maximum diastolic dimension of the defect on the LV side of the septum was larger than the systolic dimension in 60% of LCA PIVSDs and 79% of RCA PIVSDs.
PIVSDs were spherical and overall rounder in systole. No defect was round in both systole and diastole.
The orifice of the PIVSD was smaller on the RV (20 of 32) or LV (8 of 32) side of the septum and equal in 4 of 32. The RV side had the smaller orifice in 8 of 16 LCA PIVSDs (50%) and 12 of 16 RCA PIVSDs (75%). The LV side had the smaller orifice in 4 of 16 LCA PIVSDs (25%) and 4 of 16 RCA PIVSDs (25%). Other defects had equal RV and LV orifices. The orifices were not smaller centrally.
Wall thickness at 5 mm from the 4 margins of the defects
Overall, PIVSD myocardial thickness was similar in diastole (Table 3). LCA PIVSDs had thinner margins than RCA PIVSDs. The difference was highly significant at the anterior and apical margins of the defects.
Distance from the 4 margins to 5-mm-thick myocardium
LCA PIVSDs had a greater distance to reach a 5-mm thickness of myocardium than RCA PIVSDs. This did not vary significantly between systole and diastole. The distance to 5-mm myocardium in LCA PIVSDs was greater anteriorly and apically.
Extension of septal defect
The diastolic data for all 32 defects is presented as intraseptal (surrounded by septum) or extraseptal where merged with free wall (septum torn off) or with the plane of the PIVSD seen to be passing beyond the septum with ID (Table 4). LCA PIVSDs were significantly more likely to be entirely intraseptal and less likely to be extraseptal or associated with ID.
Device size compared with PIVSD
None of the PIVSDs were round throughout the cardiac cycle, meaning that the waist of the device may not conform to the PIVSD. Considering the maximum left-sided dimension alone, the existing maximum Amplatzer device waist diameter (24 mm) theoretically occluded only 16 of 32 (50%) of the left side of the PIVSDs in both systole and diastole; for LCA PIVSDs, this was 11 of 16 (69%) compared with RCA PIVSDs with 5 of 16 (31%) (p = 0.076).
The current maximum 34-mm-diameter disc would reach the margins of 24 of 32 defects (75%) in both systole and diastole. For LCA PIVSDs, this was 15 of 16 (94%), and for RCA PIVSDs, it was 9 of 16 (56%) (p = 0.037).
Eighteen of 30 patients with acute PIVSDs (60%) survived 30 days, and 17 of 30 (57%) survived 1 year (Tables 1 and 5). Two patients with LCA PIVSDs had small defects (maximum 13 and 15 mm) not clinically requiring closure, and these patients survived >1 year. Three patients with RCA PIVSDs were turned down for percutaneous and surgical closure and died within 30 days (maximum defect size 30, 44, and 54 mm). Percutaneous closure was attempted earlier than surgical (mean 5 days vs. 16 days post-MI), meaning that the efficacy of the therapies cannot be easily compared.
Sixteen patients underwent percutaneous closure. Nine of 16 (56%) survived 30 days, and all these patients also survived >1 year. Five of 9 survivors had defects that on the left side of the septum had a maximum (systolic or diastolic) dimension of 27 to 30 mm. These had maximum right side of septum dimensions of 15 to 31 mm (3 <25 mm). Of the 2 larger defects, 1 had a residual shunt.
Nine patients underwent surgical closure; 7 of 9 (78%) survived 30 days and 6 of 9 (67%) survived >1 year.
Among all those accepted for closure, 1-year survival was 60%.
Predictions of survival were hampered by small sample size, but of those accepted for closure, size of defect did not predict survival. The only significant independent predictor of survival was an RCA PIVSD at 30 days.
This is the first study to assess the dynamic morphology of PIVSDs in a series of living patients. The previous pathological research describes PIVSDs in those who did not survive the initial rupture, as well as those who died later (18), failed repairs of PIVSDs (19), and at necropsy (20). In post-mortem studies, serpiginous dissection tracts were more likely to occur in RCA PIVSDs (21).
There are limited data on the in vivo anatomy of PIVSDs: 1 case report described the use of 3-dimensional echocardiography (22), 2 reports suggested that CT is better at showing the precise anatomy than transthoracic echocardiography (23,24), and several reports have described ID using echocardiography (25–28).
Our study may differ from necropsy studies whereby the heart’s geometry may be altered by the absence of pressure and volume loading, the act of fixation, and effects of cell death. For example, the images in the necropsy studies frequently show thickened myocardium, implying LV hypertrophy, when computed tomographic studies have shown that this can be an effect of death (29).
Despite imaging being undertaken in critically ill patients, all computed tomographic studies were considered diagnostic. CT has superior spatial resolution, faster acquisition time, better patient tolerance, and safer monitoring than MRI, making it more appropriate. Three-dimensional models of the defects were not created, as it was too slow to be practical for clinical decision making. However, these techniques may become useful for visualizing closure or designing new devices.
PIVSDs are highly variable (Figure 5), not round, and also vary in size and shape during the cardiac cycle (Online Video 1). Unlike congenital muscular ventricular septal defects (which get smaller in systole because of contraction of muscle in the margins), the edges of PIVSDs can be thin and akinetic. Some get larger and others smaller in systole, meaning that single-phase imaging (systolic color Doppler or mid-diastolic CT) may not adequately assess size.
The thickness of the margins of an individual defect can vary substantially. We show that a full-dose retrospective computed tomographic protocol maximizes the anatomic information, which may be important in selecting the appropriate management strategy in these patients, where radiation dose is not a significant concern.
RCA PIVSDs more often have complex anatomy with associated ID (Online Video 2). Although this has been shown to occur in those who survive cardiac rupture and also in a postmortem series (30), the reason why RCA PIVSDs are more likely to be complex is unclear. The presence of more interdigitating fibers in the posterior interventricular groove (31) may contribute, as may the orientation of myocardial fibers to direct ID along the inferior heart.
LCA PIVSDs do not involve the basal septum in patients presenting for treatment. This may reflect that infarcts capable of causing this are not survivable. LCA PIVSDs have thinner margins, probably because the nature of LV blood supply means that the PIVSDs are remote from normal RCA-supplied myocardium.
Patients were thought to be candidates for intervention if their condition was satisfactory (with a balloon pump if needed): reasonable ventricular function, some urine output, absence of severe infection, and no major comorbidities that were likely to cause death if they survived PIVSDs.
Percutaneous intervention was undertaken when the initial imaging suggested the maximum diameter of the defect to be <25 mm on either side of the septum. Patients with larger defects were referred for a surgical opinion. Despite this, 2 patients with larger defects underwent successful percutaneous closure, with >1-year survival. These patients had maximum dimensions of 27 and 30 mm (left side) and 30 and 31 mm (right side), respectively. One was left with a residual shunt.
The absence of a rim to the inferior wall did not appear to be a problem when placing the occluder. The 5-mm overlap segment of the LV disc is compliant and adapted to the LV anatomy. On the RV side, the presence of trabeculations and insubstantial septal remnants meant that the RV disc did not always appear in its unrestrained shape, but this did not seem to prevent a good result. Thus, the central waist often abutted the free wall without any apparent problem, and the discs conformed to the anatomy well.
We oversized the devices when possible, with waist diameter up to 150% of the measured PIVSD diameter, without any sizing-related complication. Our reasoning was to aid device deployment, allow for oval defects, and, as we intervened acutely, to potentially deal with any subsequent defect expansion due to tissue necrosis and/or remodeling.
When implanting the Amplatzer PIVSD occluder, there is angulation between the delivery catheter and septum. If the device approaches the septum under tension and obliquely, it presents a smaller profile. A skill in placing the device is to keep it as coaxial to the defect as possible in order to maximize the presented diameter; oversizing helps this process. Sometimes we deploy the device far into the left ventricle so that it is fully formed. At this stage, recapturing the RV disc allows the LV disc to maintain its orientation, allowing a more coaxial approach to the septum and a more anatomic deployment.
Among patients who underwent closure (percutaneous or surgical), there were no clear predictors of long-term survival, suggesting that whenever possible, closure should be attempted.
We investigated possible reasons why percutaneous closure is not always successful. 1) The Amplatzer stem fills only 50% of the defects. 2) The disc reached 5-mm thickness of myocardium in systole and diastole in all 4 planes in only 10 of 29 of all PIVSDs: 2 of 15 LCA PIVSDs (13%) and 8 of 14 RCA PIVSDs (57%). This may promote device instability or peridevice shunting. 3) The plane of the left side of the defect is usually curved and is adjacent to damaged myocardium. Therefore, the left disc of the device needs to be compliant enough to conform to the left side of the septum, and soft enough to not cause further damage. More substantial overlap may be desirable to allow continued coverage in the setting of adverse remodeling.
There is a need for devices with shorter, thicker stems and larger discs (Online Video 3). Oval stems may be unnecessary if compliant enough to conform to the defect.
It may not be practical to design devices to fit each morphology, and deploying such devices in a specific orientation would be challenging. A wider diameter LV disc will allow successful closure of larger PIVSDs. Our maximum dimension data suggest that devices with LV disc diameters of 40, 50, or 60 mm would cover 94%, 100%, and 100% of LCA PIVSDs, respectively, and 63%, 88%, and 100% of RCA PIVSDs, with a 5-mm overlap.
Alternatively, as 63% of PIVSDs have a smaller orifice on the RV side of the septum, device closure may remain possible if a compliant waist can fill this.
The existing Amplatzer PIVSD device has an RV disc that is identical to the left; we suggest that this is more suitable for the LCA PIVSDs, which are more likely to be wholly intraseptal. For RCA-PIVSDs, the disc may jeopardize the dissected inferior heart, although we have not seen any late ventricular rupture.
Although beyond the scope of this study, device manufacturers may find manipulation of 3- and 4-dimensional data and indeed more advanced 3-dimensional printing techniques useful for the design of new occluders.
A potential limitation is that despite this being the most detailed anatomic description of PIVSDs ever reported, the relatively small number of patients may limit the conclusions that can be made. We hope that this study will encourage other units to undertake cross-sectional imaging of PIVSDs to assist decision making for closure so that multicenter data can add to the information offered.
This is the first detailed in vivo description of the dynamic anatomy of PIVSDs in a series of living patients. Those who survive the initial rupture often have defects that are not suited to the Amplatzer PIVSD occluder. Defects are oval and often are large. Thirty-eight percent of LCA PIVSDs and 81% of RCA PIVSDs were ≥25 mm. Margins are thin and with a 5-mm thickness of myocardium some distance away.
PIVSDs are often complex, with 44% of LCA PIVSDs and 100% of RCA PIVSDs being extraseptal. RCA PIVSDs are frequently associated with ID (75%).
Overall 1-year survival for acute PIVSD was 57% (including the 5 patients where closure was not attempted). PIVSDs can be closed to a smaller right-sided orifice. Of those who underwent either percutaneous or surgical closure, there was 60% 1-year survival.
The incidence of PIVSD has fallen with better reperfusion therapies. However, late presentation, an aging population, and lack of access to primary revascularization in some parts of the world mean that it will remain a serious clinical problem.
Given that PIVSD still has such high mortality, the pursuit of treatment evolution remains justifiable. It is our hope that this paper will not only augment the description of the dynamic morphology of PIVSD but also improve therapeutic selection and technique and ultimately facilitate improvements in device design.
WHAT IS KNOWN? PIVSDs have a poor mortality rate, however they are treated. There is limited antemortem dynamic anatomic knowledge.
WHAT IS NEW? PIVSDs are highly variable in morphology; they vary in size between systole and diastole, have thin margins, and, particularly with RCA PIVSDs, are frequently associated with ID out of the septum. LCA PIVSDs have poorer 30-day survival than RCA PIVSDs, though this was not significant at 1 year.
WHAT IS NEXT? Design of better therapeutic devices and techniques will aid in the management of PIVSDs.
For supplemental videos and their legends, please see the online version of this article.
Dr. Martin has a consultancy agreement for proctoring and educational activity with St. Jude Medical. Dr. Turner has served as a consultant and proctor for St. Jude Medical (now Abbott Vascular). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- computed tomography
- intramyocardial dissection
- left coronary artery
- left ventricular
- myocardial infarction
- magnetic resonance imaging
- post-infarct ventricular septal defect
- right coronary artery
- right ventricular
- Received May 26, 2016.
- Revision received March 6, 2017.
- Accepted March 23, 2017.
- 2017 American College of Cardiology Foundation
- Crenshaw B.S.,
- Granger C.B.,
- Birnbaum Y.,
- et al.
- Skehan J.D.,
- Carey C.,
- Norrell M.S.,
- et al.
- Jones B.M.,
- Kapadia S.R.,
- Smedira N.G.,
- et al.
- Cummings R.G.,
- Reimer K.A.,
- Califf R.,
- et al.
- Radford M.J.,
- Johnson R.A.,
- Daggett W.M.,
- et al.
- Moore C.A.,
- Nygaard T.W.,
- Kaiser D.L.,
- Cooper A.A.,
- Gibson R.S.
- Schlotter F.,
- de Waha S.,
- Eitel I.,
- et al.
- Turner M.S.,
- Hamilton M.,
- Morgan G.J.,
- Martin R.P.
- Thiele H.,
- Kaulfersch C.,
- Daehnert I.,
- et al.
- Calvert P.A.,
- Cockburn J.,
- Wynne D.,
- et al.
- ↵National Patient Safety Agency, National Research Ethics Service. Defining Research. Available at: www.hra.nhs.uk/documents/2016/06/defining-research.pdf. Accessed April 10, 2017.
- Halpern D.G.,
- Perk G.,
- Ruiz C.,
- et al.
- Paul J.F.,
- Macé L.,
- Caussin C.,
- et al.
- Hajj-Chahine J.,
- Jayle C.,
- Tomasi J.,
- et al.
- Galrinho A.,
- Loureiro J.,
- Magalhães M.P.
- Okuma H.,
- Gonoi W.,
- Ishida M.,
- et al.
- Becker E.,
- Caruso G.