Author + information
- Received March 31, 1997
- Revision received July 25, 1997
- Accepted August 14, 1997
- Published online November 15, 1997.
- ↵*Dr. Colin K. Phoon, Pediatric Echocardiography Laboratory, New York University Medical Center, 530 First Avenue, Suite 9U, New York, New York 10016.
Objectives. We modeled the utility of preoperative potential left ventricular (LV) volume in predicting postoperative volume in conditions causing LV compression.
Background. With right ventricular (RV) overload lesions, LV “hypoplasia” may be primarily due to compression by reverse septal bowing. If so, preoperative potential LV volume should correspond 1:1 with postoperative volume. The potential volume for a given endocardial circumference can be calculated from the maximal potential cross-sectional area (where A = circumference2/4π) and LV length.
Methods. We studied echocardiographic variables from 22 patients with RV overload lesions perioperatively.
Results. Preoperative LV volume was 15.0 ± 7.1 ml/m2(59% of patients had a volume <15 ml/m2); potential volume was 20.0 ± 9.8 ml/m2. Postoperative volume increased to 28.2 ± 8.6 ml/m2(100% of patients had a volume >15 ml/m2). Preoperative potential volume correlated well with, but generally underestimated, postoperative volume (r = 0.75, p < 0.0001). Postoperative increases in both LV circumference and length contributed to this discrepancy.
Conclusions. In RV overload lesions, LV “hypoplasia” is primarily due not to compression; rather it is due to underfilling. Even “hypoplastic” ventricles can achieve an adequate cavity after operation normalizes loading conditions. Both true and potential preoperative volume can predict postoperative volume well. However, potential volume, which is less prone to underestimating ventricular adequacy, may better help to determine suitability for biventricular repair in lesions of RV overload associated with a “hypoplastic” LV.
In conditions with a small left ventricle (LV) considered for biventricular repair, one critical factor determining outcome is the adequacy of the ventricular volume. Usually discussed in the context of critical aortic stenosis or hypoplastic left heart syndrome [1–6], a small LV may also be associated with other conditions, such as unbalanced atrioventricular septal defect (AVSD) [7–10]and total anomalous pulmonary venous connection (TAPVC) [11–15]. In these latter conditions, the LV may not be truly “hypoplastic” . Moreover, its geometry is distinctly different from that of typical hypoplastic left heart syndrome. Instead of being an underdeveloped and small, globular ventricle , the LV is more typically flattened or even crescentic, compressed by a dilated, pressure- and volume-overloaded right ventricle (RV) (Fig. 1) [2, 11–15].
In such conditions of distorted LV geometry, small preoperative LV volumes may not accurately predict the feasibility of biventricular repair. Indeed, certain defects, such as TAPVC, are generally amenable to complete (biventricular) repair [16–19]. However, other defects, such as severely unbalanced AVSD, may sometimes require univentricular palliation . For congenital cardiac defects with RV pressure and volume overload, we hypothesized that apparent LV hypoplasia is primarily due to compression of the ventricular cavity by reverse bowing of the interventricular septum. The septum should assume a normal position after operation normalizes the loading conditions. Thus, we hypothesized that the postoperative capacity may be better reflected by preoperative potential LV volume, which is that volume if the septal position were normal. In a small series of patients with right-dominant AVSD, we have shown that a preliminary model of preoperative potential LV volume may better reflect postoperative capacity [9, 10].
In this study, we developed a more sophisticated model of potential LV volume in patients with lesions with RV overload undergoing biventricular repair at our institution. This model tested the hypothesis that preoperative potential LV volume predicts postoperative volume better than does preoperative true volume; our compression model also predicted a 1:1 relation between preoperative potential LV volume and postoperative volume.
If we assume that the LV is a compressible but otherwise noncompliant chamber (like a plastic bag), then the relative loading states of the RV and LV will determine the interventricular septal position and therefore the true LV cavity volume. No matter what the ventricular geometry, however, there is a potential spacein the LV cavity that depends only on the endocardial surface area, as with any cavity.
Because the LV under normal loading conditions is circular in cross section , the cross-sectional area is at the maximal area for a given endocardial circumference. At a given long dimension, then, the normal conical LV is at the maximal volume for a given endocardial surface area (Fig. 2). For any given circumference of a compressed LV in cross section, the potentialLV area can be obtained when that circumference is translated into a circle (maximal area for a given circumference). For a circle,then, and the area of the circle or potential area is described by
LV volume has been well estimated by the “bullet” formula [5, 20, 21]:
In our theoretic model, the determinant of intracavitary volume is the position of the interventricular septum; we assume the ventricle is noncompliant and length does not change. Thus, endocardial surface area and therefore cross-sectional circumference are expected to remain the same preoperatively and postoperatively, and preoperative potential volume should correlate 1:1 with postoperative volume. Preoperative true LV volume should not correlate well with postoperative volume because it depends on the degree of reverse septal bowing and would vary, despite a theoretically fixed potential volume.
To model the utility of preoperative potential LV volume, we chose the lesions TAPVC, cor triatriatum and right-dominant AVSD. These lesions exhibit at least some degree of LV compression (often severe) by a pressure- and volume-overloaded RV, and all the patients with these lesions have undergone biventricular repair at our institution in recent years, with normalization of septal bowing postoperatively. Preoperative true and potential LV volumes could therefore be measured and correlated with postoperative volumes.
1.2 Patient Group
All patients found in the echocardiography data base at the University of California San Francisco with TAPVC or cor triatriatum, or both, between January 1992 and April 1996 were considered for this study. Inclusion criteria included 1) complete (biventricular) repair; 2) successful discharge from the hospital; and 3) availability of both preoperative and postoperative echocardiograms. Seventeen patients met all three criteria. No patient had heterotaxy or an isomerism syndrome or significant associated cardiac lesions. In addition, five patients with right-dominant AVSD whose results have been preliminarily analyzed [9, 10]were reanalyzed according to the revised theoretic model described earlier and were included in this study. During the study period, all patients with this lesion underwent biventricular repair .
1.3 Echocardiographic Measurements
Preoperative echocardiography was typically performed on admission to the hospital. All postoperative measurements were made within several days of the operation (7 ± 5 days postoperatively) to ensure that any increase in LV volume was due to geometric changes and not to growth. To blind the observers to the outcome variables, preoperative variables were measured before postoperative variables. The following were measured: aortic annulus (parasternal long-axis view); LV circumference and area (by planimetry) at the level of the papillary muscles just below the mitral valve (parasternal short-axis view) [5, 20, 21]; and mitral and tricuspid annulus diameters and LV long dimension (apical four-chamber view). For right-dominant AVSD, the crux of the heart was estimated from the atrioventricular valve and septal crest positions, and measurements were made accordingly . Also, the position of the interventricular septum in the parasternal short-axis view was arbitrarily defined as normally positioned, flattened or reverse bowed. From the apical four-chamber view, the RV was defined as apex forming or not.
LV volume was calculated using the “bullet” formula , and preoperative potential LV volume was calculated using the potential area, as described under “Theory.”
Measurements were made on an off-line, frame-grabbing package (Freeland, Prism Imaging Systems). At end-diastole (defined as the onset of the Q wave of the electrocardiogram), each variable was measured at three different frames, and the values were averaged. Ectopic or postectopic beats were excluded. To test interobserver variability, LV circumferences, aortic annulus diameters and tricuspid valve diameters were measured in 10 random patients and repeated by each author, who had no knowledge of the other’s results. Variability was expressed as the difference from the mean of the results obtained by the two observers. To test variability of measurements within an individual patient made by one observer (C.K.L.P.), the first and third measurements of these same echocardiographic variables were averaged in 10 random patients. Variability was expressed as the difference from the mean of these two observations.
The data are expressed as mean value ± SD, unless otherwise stated. Comparisons of variables within individual patients (for instance, preoperative and postoperative measurements) were made using the two-tailed, paired Student ttest. The relations between preoperative and postoperative LV volumes (true and potential) were sought using simple linear regression analysis. The significance level was set at p < 0.05.
2.1 Patient Demographics
Fifteen patients had TAPVC, one had cor triatriatum, one had both TAPVC and cor triatriatum and five had right-dominant AVSD (Table 1). Ages ranged from 0 days to 6.5 months (median 21 days). One patient (Patient 21) with a right-dominant AVSD died 71 days postoperatively after severe mitral insufficiency necessitated valve replacement, which was complicated by valve thrombosis. Despite the mitral insufficiency, her LV was able to sustain the systemic circulation [9, 10].
2.2 LV Geometry
Interobserver and intrapatient/observer variabilities for echocardiographic measurements were 5.5% and 6.0%, respectively. Indicators of preoperative and postoperative LV size are shown in Table 1. Preoperatively,the LV appeared small in most patients, with an RV-forming apex in 21 of 22 patients. The mitral valve was substantially smaller (p < 0.0001) than the tricuspid valve annulus (mitral:tricuspid annulus ratio of 0.71 ± 0.19), and septal position was flat or reversed in 16 of 22 patients. Notably, despite the small LV sizes, aortic annuli were typically within the normal range (diameter range 5.0 to 8.8 mm, Z scores −0.62 ± 0.94); mitral annuli tended to be smaller but were also mostly within the normal range (diameter range 4.9 to 15.9 mm, Z scores −0.84 ± 1.82). Postoperatively,these variables became largely normalized. The mitral:tricuspid valve annulus ratio increased to 0.96 ± 0.11, so that the annuli were not significantly different (p > 0.05). Septal position became normalized in 19 of 22 patients; only 3 patients, all with TAPVC or cor triatriatum, exhibited a flattened septal position postoperatively. Indeed, the postoperative cross-sectional areas were 91.1 ± 4.9% of the theoretic maximal circular areas for their given circumferences; thus, most patients demonstrated a circular or nearly circular (normal) LV geometry postoperatively. Notably, LV length increased by 15.0 ± 12.3% (p < 0.0001) and LV circumferences by 21.3 ± 19.2% (p < 0.0001) postoperatively.
The preoperative and postoperative LV circumferences and true volumes and preoperative potential volumes are shown in Table 2. Preoperative LV volumes were small, with a mean indexed volume of 15.0 ± 7.1 ml/m2, which increased to 28.2 ± 8.6 ml/m2postoperatively (p < 0.0001). Preoperatively, more than half of our patients (13 of 22) had true LV volumes <15 ml/m2. Postoperatively, all patients had volumes >15 ml/m2postoperatively. In contrast, the preoperative indexed potential volume was 20.0 ± 9.8 ml/m2, which was higher than the true volumes (p < 0.0001), as expected, but 9 of 22 patients still showed potential volumes <15 ml/m2.
The various relations between preoperative true or potential LV volumes and postoperative LV volumes are shown in Table 3. For all patients, correlations between true preoperative and postoperative volumes and preoperative potential and postoperative volumes were similarly significant (Eqs. (1)-(3)in Table 3Figs. 3 and 4). ⇓Notably, preoperative potential volume was lower than postoperative true volume in 18 of 22 patients, underestimating the postoperative LV volume by 25% to 30% on average (p < 0.0002).
The relations for patients with and without right-dominant AVSD were probably not significantly different, although the numbers of AVSD were small (Eqs. (4) and (5)in Table 3). Although the correlation coefficients for AVSD were high, the range of volumes was wide for the relatively few points. Preoperative potential and postoperative LV volumes did not correlate 1:1 in patients with either right-dominant AVSD or other lesions, mainly TAPVC (Eqs. (4)-(7)in Table 3).
We then reexamined the degree of compression to assess whether those patients with smaller preoperative true volumes showed an improved correlation between potential and postoperative volumes (Eqs. (8) and (9)in Table 3). In those 10 patients whose preoperative true volumes were <75% of the potential volumes (septum flat or reversed, see AppendixAppendix A), the relations showed only fair correlation. The correlation between preoperative potential LV volume and postoperative volume did not improve in this subset of patients with smaller preoperative ventricles, nor was potential volume better able than preoperative true volume to predict the postoperative volume.
In the 12 patients whose preoperative true volumes were ≥75% of the potential LV volumes (less significant compression), relations showed an excellent correlation with slopes of ∼1. However, postoperative volumes were still underestimated by ∼1.5 ml (Eqs. (10) and (11) in Table 3), which represents some 35% to 40% of the cavity volume.
We have introduced the concept of ventricular potential space, which we believe may be important in lesions producing distorted LV geometry. In this study of RV overload and LV potential space, preoperative potential LV volume did not predict postoperative volume any better than did preoperative true LV volume. Potential volume also did not exhibit the 1:1 relation with postoperative volume predicted by our model, typically underestimating postoperative volume. Nevertheless, preoperative potential volume, as expected, was consistently higher than true volumes and therefore was less prone to underestimate the eventual postoperative LV size. Importantly, both potential and true volumes predicted postoperative volumes in a highly significant manner.
3.1 Prediction of Postoperative LV Volume by Preoperative Volume Measurements
That both preoperative true and potential volume predicted postoperative LV volume equally well surprised us. We concluded that septal bowing and LV compression are not the primary determinants of a small LV cavity and that, in fact, the ventricle’s volume status is far more important, overriding any effect of ventricular compression. The relation between preoperative true volume and postoperative volume (Eq. (2)in Table 3) suggests that, with a slope of 1, the preoperative LV is simply underloaded by ∼3 to 4 ml. Because a typical preoperative LV volume is 4 ml, this degree of underloading would certainly outweigh the effects of compression, because each 25% reduction of cavity volume by compression is equivalent to only 1 ml.
There are other possible reasons for the failure of our model to correlate even better with postoperative LV volume and for its underestimation of postoperative volume. Myocardial distensibility and variability in postoperative intravascular volume status (and therefore LV volume status) likely contributed to our model’s failure to predict postoperative volume in a 1:1 relation. The model assumes (incorrectly) a noncompliantventricle whose walls can be distorted but are otherwise not distensible; that is, it assumes only changes in the septal position, not accounting for postoperative increases in length. These factors are corroborated by the 15% to 20%, but variable, increase in both the LV circumference and long dimension seen postoperatively.
In addition to uncontrollable variability of postoperative volume status, the influence of preoperative LV volume status on the value of the potential space concept was also studied. If an LV is not at all compressed or compressed to the same degree in all patients, then preoperative potential and true volume should correlate equally well, and highly, with postoperative volume. Even in patients with less significant compression, however, postoperative LV volume was underestimated by ∼1.5 ml (some 35% to 40% of the cavity volume). In contrast, if an LV is completely compressed (zero volume), then preoperative potential volume and postoperative volume should correlate well, whereas true preoperative and postoperative volumes should not correlate at all. Because the degree of compression varied, we expected an improvement in the predictive value of potential over true preoperative volumes, as compression of the LV by the RV became greater. However, the correlation between preoperative potential volume and postoperative volume did not improve, nor did potential volume correlate any better than preoperative true volume with postoperative volume. Again, it is clear that compression plays a minor role, if any, in the preoperative LV cavity size.
Despite these confounding factors that cannot be accounted for in our patients, the relations were still highly significant.
3.2 “Hypoplastic” LVs
Several studies have now demonstrated that 15 to 20 ml/m2is the minimal LV volume necessary to support the systemic circulation [1, 4, 6]. Our results show that even a compressed and underloaded ventricle with a “hypoplastic” cavity (<15 to 20 ml/m2) can expand to an adequate volume after surgical repair normalizes loading conditions. Despite small true preoperative volumes in most patients, most patients also exhibited acceptable aortic and mitral annulus sizes. Their relatively normal sizes suggest that the LV has not been chronically “hypoplastic”; it is likely that LV volumes are normal or nearly so in fetal life, with RV overload and LV compression occurring only postnatally. Thus, even in the presence of a very small LV cavity (<15 ml/m2), because of the potential volume and the capacity to expand after normalization of loading conditions, biventricular repair should not be dismissed on the grounds of a very small LV cavity alone.
3.3 Echocardiographic Measurements
Cross-sectional circumferences and areas can be easily obtained and volumes thus estimated with current technology. Even with distorted geometry, preoperative true volumes must be smaller than potential volumes, and we believe that measurements taken in orthogonal planes provide a reasonable estimate of LV volume. The advantages of estimating potential LV volume are twofold. Because it assumes a circular cross-sectional area, potential volume is likely to be estimated accurately by the “bullet” formula. Also, potential volume is less prone than preoperative true volumes to underestimating the postoperative LV adequacy.
We have developed a theoretic model of LV geometry in RV overload conditions that assesses the LV potential space for a given endocardial surface area. Our results show that both true and potential preoperative LV volumes can predict postoperative volumes in a highly significant manner, a finding important in the decision-making process for surgical management. Our model also indicates that small preoperative LV volumes are not primarily due to compression by an overloaded RV, but rather to underfilling. Thus, even a “hypoplastic” LV (<15 to 20 ml/m2) can expand to an adequate size after surgical repair normalizes loading conditions. Potential LV volume may be particularly valuable preoperatively because it is likely to be more accurately estimated by existing echocardiographic formulas and is less prone to underestimating LV adequacy than is preoperative true volume. Thus, potential volume may provide additional useful information in some patients with distorted LV geometry and may be useful in determining suitability for biventricular repair when the LV is apparently “hypoplastic.”
A.1 Geometry of Compression by a Flat Septum
In the theoretic case of a flattened interventricular septum (Fig. A1), our model assumes that only the septal position changes and that the internal (endocardial) circumference does not change.
For a normal LV (Fig. A1, left), the short-axis area (A) and circumference (C) are those described for a circle. Therefore, C= 2πR and A= πR2.
For the semicircle of a compressed LV (Fig. A1, right), the circumference C′ = R′ + R′ + 1/2 (2πR′) = (2 + π)R′. According to our model, the circumference has not changed (C= C′), so that 2πR= (2 + π)R′. Therefore, R′ = (2πR) ÷ (2 + π) = 1.222R. Area A′ of the semicircle, then, is described by A′ = 1/2 π (R′)2= 1/2 π (1.49328)R2= 0.747πR2= 0.747A.
Thus, the area A′ of the compressed LV is 75% of the area A of the noncompressed LV. Because our model also assumes that LV length has not changed, LV volume is directly proportional to the short-axis area. Thus, according to our model, the volume of a compressed LV with an entirely flat septum will be reduced 25%.
This work was done in part during the tenure of a research fellowship from the American Heart Association, California Affiliate (Dr. Phoon) and was presented in part at the 46th Annual Scientific Session of the American College of Cardiology, Anaheim, California, March 1997.
↵1 All editorial decisions for this article, including selection of referees, were made by a Guest Editor. This policy applies to all articles with authors from the University of California San Francisco.
- atrioventricular septal defect
- left ventricle, left ventricular
- right ventricle, right ventricular
- total anomalous pulmonary venous connection
- Received March 31, 1997.
- Revision received July 25, 1997.
- Accepted August 14, 1997.
- The American College of Cardiology
- Ludman P,
- Foale R,
- Alexander N,
- Nihoyannopoulos P
- Parsons MK,
- Moreau GA,
- Graham TP,
- Johns JA,
- Boucek RJ
- Rhodes LA,
- Colan SD,
- Perry SB,
- Jonas RA,
- Sanders SP
- Hoffman JIE
- Cohen MS,
- Jacobs ML,
- Weinberg PM,
- Rychik J
- Phoon CK,
- van Son JAM,
- Haas GS,
- Silverman NH
- Graham TP,
- Jarmakani JM,
- Canent RV
- Nakazawa M,
- Jarmakani JM,
- Gyepes MT,
- Prochazka JV,
- Yabek SM,
- Marks RA
- Bando K,
- Turrentine MW,
- Ensing GJ,
- et al.
- Wyatt HL,
- Heng MK,
- Meerbaum S,
- et al.
- Triulzi MO,
- Wilkins GT,
- Gillam LD,
- Gentile F,
- Weyman AE