## Background

Hypoplastic left heart complex (HLHC) is characterized by a mitral valve or an aortic valve annular *Z *score < −2, antegrade flow in the ascending aorta, ductal dependency, coarctation or aortic arch hypoplasia, and absence of significant (sub)valvar stenosis. The *Z *scores of the mitral and aortic valve annuli are major determinants of HLHC. Therefore, the algorithm for *Z *-score calculation is essential for diagnosis. However, no single universal method of calculation is in use. In the scientific literature addressing HLHC, various *Z *-score calculation methods have been applied. The aim of this study was to evaluate *Z *scores derived from two-dimensional echocardiographic dimensions in patients with HLHC.

## Methods

To compare the different published methods using two-dimensional echocardiographic measures for *Z *-score calculation, a cohort of 18 newborns diagnosed with HLHC was retrospectively evaluated. In addition, the methods to determine body surface area in newborns were evaluated.

## Results

Three *Z *-score calculation methods were included and compared. Using the method of Daubeney *et al *. to calculate *Z *scores in our cohort illustrated a lack of correlation beyond a *Z *score < 0, compared with the methods of Zilberman *et al *. and Pettersen *et al *. *Z *scores calculated using Zilberman *et al *.’s and Pettersen *et al *.’s methods were fairly consistent. The equations used by Pettersen *et al *. are based on the largest population of neonates.

## Conclusion

Although the different methods for calculating *Z *scores for mitral and aortic valve dimensions correspond fairly well in the normal range, *Z *scores < −2 diverge substantially. A useful scientific comparison of published data and outcomes of patients with HLHC remains elusive. The *Z *-score calculation algorithms used by Pettersen *et al *. appear to be the most appropriate for use in an evaluation of HLHC. Because these different methods can yield different values, reporting the method as well as the *Z *score is essential for an accurate diagnosis. Similarly, the method used to determine body surface area should be reported.

The term *hypoplastic left heart complex* (HLHC) has been recently introduced to characterize significant underdevelopment of the left heart–aorta complex, absence of significant valvar (aortic and mitral) stenosis, a mitral valve (MV) or aortic valve (AoV) annular *Z* score < −2, coarctation or aortic arch hypoplasia, and no significant ventricular septal defect. Functionally, HLHC demonstrates holosystolic antegrade flow in the ascending aorta and ductal dependency with bidirectional ductal flow.

Hypoplastic left heart syndrome (HLHS) encompasses a spectrum of cardiac malformations, characterized by significant underdevelopment of the components of the left heart and the aorta, including the left ventricular (LV) cavity. At the severe end of the spectrum lies the combination of aortic and mitral atresia. Here, the left ventricle can be virtually nonexistent, yielding a functionally univentricular heart. The preferred therapeutic option is the Fontan circulation, which is considered a palliative approach. At the mild end of the spectrum, hypoplasia of the AoV and MV (but without intrinsic valvar stenosis or atresia), and milder degrees of LV hypoplasia, can be present.

Biventricular circulation in neonates with HLHC is considered superior to univentricular circulation. The surgical approach consists primarily of reconstruction of the aortic arch and, if necessary, creation of an atrial fenestration. Reported mortality is low. Nevertheless, no cutoff point has been determined to indicate whether left heart morphology and size are predictive of a successful outcome.

Ventricular size has been the dominant measure of imbalance between the right and left ventricles in neonatal pediatric cardiology. However, assessment of ventricular size using echocardiography is prone to inaccuracy. Furthermore, LV size does not express categorical functional incompetence in a neonate with reduced preload during the fetal and postnatal periods. In cases of severe coarctation, or total anomalous pulmonary venous return, morbidity is principally unrelated to ventricular size. Therefore, the inflow and outflow of a relatively small ventricle determine LV competence. In patients with unbalanced atrioventricular septal defects, the sizes of the valves as expressed in the atrioventricular valve index also appear to be relevant predictors of outcomes.

The dimensions and *Z* scores of the MV and AoV annuli are the major determinants in the identification of HLHC. This is analogous to the published use of tricuspid and pulmonary valve *Z* scores in patients with pulmonary atresia with intact ventricular septum. However, the use of *Z* scores is a relatively recent addition to the literature. Different algorithms are available for *Z* -score calculation for the MV and AoV annuli. In all publications addressing the outcomes of patients with HLHC, the *Z* -score calculation methods used are too diverse to be comparable. This can even lead to the implausible suggestion that biventricular repair in HLHC with an AoV *Z* score of −10.8 is feasible. The *Z* -score calculation algorithm fundamentally influences clinical decisions. The aim of this study was to identify the most appropriate *Z* -score calculation method for the evaluation of MV and AoV annuli in patients with HLHC.

## Methods

## Literature Search

A literature search was conducted in the MEDLINE, Cochrane, and Embase databases for all *Z *-score calculation methods and the use of *Z *scores in the treatment of patients with HLHC. All identifiable methods of calculating *Z *scores were collected and checked for validity. All collected methods were then compared on the basis of the regression equation and the standard values for cardiac dimensions used.

## Study Design and Study Population

All patients with HLHC who underwent biventricular repair between 2004 and 2010 at the University Medical Center Utrecht/Wilhelmina Children’s Hospital (Utrecht, The Netherlands) were included and studied retrospectively.

The preoperative diameters of the MV annulus and AoV annulus were measured, and the *Z *scores of the MV and AoV were calculated using all *Z *-score calculation methods published between 1993 and 2008. Height and weight were recorded at the time of the echocardiographic evaluation; body surface area (BSA) was calculated using the method of Haycock *et al *.

## Echocardiography

Recordings of routine cross-sectional echocardiographic and M-mode studies were performed in all patients preoperatively and postoperatively using a Vivid 7 (GE Vingmed Ultrasound AS, Horten, Norway) system with phased-array transducers (7 and 10 MHz). Measurements were conducted following published standards and using a recommended methodology, such as the leading-edge method. The MV annulus was measured in the apical four-chamber projection during ventricular end-diastole. AoV annular diameter was obtained during end-systole from a parasternal long-axis view. All measurements were performed three times by the same investigator (M.W.F.). The mean of these three measurements was subsequently used in this analysis.

## Definitions

The *Z *score was defined as the number of standard deviations away from the mean diameter for a normal population of neonates, indexed to BSA. A *Z *score < −2 (“more negative” than −2) means that the score is farther from the mean. This also means that the corresponding measured value is smaller than the value corresponding exactly with a *Z *score of −2. A *Z *score > −2 describes the opposite (that the score is closer to the mean, or “less negative”).

## Statistical Analysis

Analyses were performed using IBM SPSS Statistics version 20 for Windows (IBM, Armonk, NY). *Z *scores were calculated for the MV and AoV using the equations provided by Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. For Zilberman *et al *., the gender-specific equations were applied. Agreement among the *Z *scores was analyzed using intraclass correlation coefficients and Bland-Altman plots. Graphs were constructed calculating the difference between the two measurements, against the average of the two measurements. The limits of agreement involving the mean difference, and the standard deviation of the difference, were computed. Plots were constructed using GraphPad Prism 5.04 for Windows (GraphPad Software, San Diego, CA).

## Results

## Literature Search

Five different methods were identified to calculate *Z *scores of the MV and AoV annuli in HLHC: Kirklin and Barratt-Boyes (1993), Serraf *et al *. (1999), Daubeney *et al *. (1999), Zilberman *et al *. (2008), and Pettersen et al. (2008). Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. all used normal values derived from two-dimensional echocardiography; therefore, these methods were used for further evaluation in the present study. In addition, publications discussing HLHC, neonatal coarctation, and neonatal AoV stenosis were reviewed to identify the methods used to measure dimensions and to calculate *Z *scores (see Table 1 ).

Study | Number of patients | Type of patients | MV Z score |
AoV Z score |
Comment ^{∗ } |
---|---|---|---|---|---|

Tchervenkov et al . (1998) |
11 | HLHC | Yes | Yes | Z -score calculation according to Kirklin and Barratt-Boyes, LV/RV long-axis ratio, LVEDD |

Blaufox et al . (1998) |
7 | HLHC/CoA | No | No | Indexed LV volume, indexed MV area, LV/RV long-axis ratio, LVEDD |

Alboliras et al . (1999) |
8 | HLHC/CoA | No | No | Indexed MV area, indexed aortic annulus, indexed LV length, LVEDD |

Serraf et al . (1999) |
20 | HLHC | Yes | Yes | Own Z -score calculation as described in text, LVEDD |

Tani et al . (2000) |
63 | CoA | No | No | MV and AoV annuli, Rhodes score, LV/RV long-axis ratio, LVEDD |

Grosse-Wortmann et al . (2008) |
12 | HLHC | Yes | Yes | Z score Daubeney et al ., LVEDD, MRI-derived LV and RV volumes |

Puchalski et al . (2004) |
55 | CoA/HLHC | Yes | Yes | Z score Daubeney et al ., LV/RV long-axis ratio, LVEDD |

Phoon and Silverman (1997) | 22 | RV volume overload | No | No | RV and LV volume (echocardiography derived) |

Rhodes et al . (1991) |
65 | Valvar AS | No | No | Diameter of AoV annulus (mm), indexed MV area, indexed LV length and volume |

McElhinney et al . (2005) |
113 | Valvar AS | Yes | Yes | MV area Z score, Z score calculated on the basis of data obtained at their institution, Z -score formula not mentioned; LV/RV long-axis-ratio, LVEDD |

Han et al . (2007) |
53 | Valvar AS | Yes | Yes | Z score Daubeney et al ., LVEDD, LV length |

∗ Mode of *Z *-score calculation is explained, and any determinations of LV length or volume are mentioned.

## Regression Equations

Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. indexed mean mitral and aortic diameters to the natural logarithm of BSA as the predictor. To prevent heteroscedasticity (the larger the BSA, the larger the variance), the authors stabilized the variances by performing a logarithmic transformation on the measurements, using the natural logarithm. In all three publications, the following equation was used to calculate the *Z *score: Z = [ln(measured valve diameter) – ln(mean normal value of valve diameter)]/v(mean square error), where ln(measured valve diameter) is the natural logarithm of the measured valve diameter (in centimeters), and ln(mean normal value of valve diameter) and the root mean square error are calculated values. These values were calculated using the equations published by Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. These equations are summarized in Table 2 .

Calculation method | ln(mean normal value) | Root mean square error |
---|---|---|

MV | ||

Daubeney et al . |
0.9651 + 0.4658 × ln(BSA) | 0.09167 |

Zilberman et al . |
Male: 0.765 + 0.425 × ln(BSA) | 0.169 |

Female: 0.733 + 0.408 × ln(BSA) | 0.180 | |

Pettersen et al . |
−0.271 + 2.446 × BSA − 1.700 × BSA ^{2 }+ 0.425 × BSA ^{3 } |
0.148 |

AoV | ||

Daubeney et al . |
0.5183 + 0.5347 × ln(BSA) | 0.06726 |

Zilberman et al . |
Male: 0.472 + 0.492 × ln(BSA) | 0.141 |

Female: 0.437 + 0.461 × ln(BSA) | 0.127 | |

Pettersen et al . |
−0.874 + 2.708 × BSA − 1.841 × BSA ^{2 }+ 0.452 × BSA ^{3 } |
0.100 |

Both Daubeney *et al *. and Zilberman *et al *. calculated ln(mean normal value) using an intercept and ln(BSA). Zilberman *et al *. made gender-specific equations. Pettersen *et al *. evaluated several nonlinear models and selected a polynomial model to the third power as their regression model for the prediction of ln(mean normal value).

The normal values of Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. are all based on two-dimensionally derived echocardiographic data. Daubeney *et al *.’s, Zilberman *et al *.’s, and Pettersen *et al *.’s normal values are based on 125, 748, and 785 healthy infants and children, respectively. Pettersen *et al *. included 82 healthy subjects with BSAs ≤ 0.25 m ^{2 }. Daubeney *et al *. and Zilberman *et al *. did not specify the numbers of tested neonates.

## Comparison of *Z *-Score Equations

To express the differences of the three *Z *-score calculations (Daubeney *et al *., Zilberman *et al *., and Pettersen *et al *. ), the *Z *scores for a given BSA of 0.22 m ^{2 }(the average size of our neonate patient group) for both the MV and the AoV were plotted ( Figure 1 ).

The varying results obtained using the three regression equations (Daubeney *et al *., Zilberman *et al *. [male equation], and Pettersen *et al *. ) revealed disagreement for both the MV and AoV, as shown in Figure 1 . All three methods corresponded well in the range of normal (BSA, 0.22 m ^{2 }) neonatal MV (1.1–1.3 cm) and AoV (0.7–0.8 cm) annular diameters. However, Figure 1 demonstrates that differences in *Z *scores begin to emerge below a *Z *score of −2, which has important clinical implications. *Z *scores fall significantly lower using Daubeney *et al *.’s *Z *-score calculations than with the other two equations, at *Z *scores < −1 for the AoV and < 0 for the MV. Furthermore, extreme MV *Z *scores in patients with HLHC can vary enormously when using different *Z *-score calculations for a given small annular diameter. One annular diameter can yield an MV *Z *score of −4 using Pettersen *et al *.’s method, while producing a score of nearly −7 using Daubeney *et al *.’s method.

## Application of *Z *-Score Equations in Clinical Practice

## Patients

Between 2004 and 2010, 18 patients were diagnosed with HLHC and underwent biventricular repair. Patient characteristics are provided in Table 3 . *Z *scores were calculated using the equations provided by Zilberman *et al *., Daubeney *et al *., and Pettersen *et al *.

Patient | Gender | Diagnosis | BSA (Haycock) | Mitral ring (cm) | Z score |
Aortic ring (cm) | Z score | ||||
---|---|---|---|---|---|---|---|---|---|---|---|

Zilberman et al . |
Daubeney et al . |
Pettersen et al . |
Zilberman et al . |
Daubeney et al . |
Pettersen et al . | ||||||

1 | Male | Borderline HLHS, CoA, LSVC | 0.22 | 0.68 | −3.05 | −7.14 | −3.93 | 0.45 | −3.80 | −7.70 | −4.45 |

2 | Male | Borderline HLHS, CoA | 0.22 | 0.60 | −3.75 | −8.43 | −4.73 | 0.59 | −1.82 | −3.54 | −1.67 |

3 | Male | Borderline HLHS, CoA | 0.22 | 0.90 | −1.39 | −4.08 | −2.04 | 0.52 | −2.77 | −5.55 | −3.00 |

4 | Female | HLHC, LPSVC, VSD, LVOTO, coarctation | 0.19 | 0.70 | −2.31 | −6.03 | −3.34 | 0.50 | −2.91 | −4.89 | −2.75 |

5 | Male | Hypoplastic aortic arch, CoA, VSD, PFO | 0.23 | 0.80 | −2.20 | −5.59 | −2.95 | 0.70 | −0.82 | −1.48 | −0.22 |

6 | Male | Unbalanced complete AVSD, MS, AS, HAA, CoA | 0.23 | 0.85 | −1.81 | −4.86 | −2.50 | 0.57 | −2.22 | −4.42 | −2.21 |

7 | Female | Borderline HLHS, CoA | 0.23 | 0.82 | −1.86 | −5.25 | −2.75 | 0.64 | −1.64 | −2.70 | −1.06 |

8 | Female | Unbalanced partial AVSD, hypoplastic LV, MV, HAA | 0.23 | 0.59 | −3.63 | −8.73 | −4.91 | 0.48 | −3.82 | −6.80 | −3.83 |

9 | Male | Borderline HLHS, CoA | 0.15 | 0.64 | −2.43 | −5.83 | −3.44 | 0.34 | −4.43 | −8.78 | −5.76 |

10 | Female | CoA, HLHC | 0.21 | 0.74 | −2.25 | −5.98 | −3.23 | 0.43 | −4.49 | −7.99 | −4.69 |

11 | Male | HLHC | 0.22 | 0.84 | −1.74 | −4.71 | −2.44 | 0.48 | −3.25 | −6.54 | −3.69 |

12 | Female | HLHC | 0.19 | 0.65 | −2.65 | −6.68 | −3.76 | 0.40 | −4.55 | −7.95 | −4.85 |

13 | Male | HLHC | 0.19 | 0.70 | −2.50 | −6.07 | −3.36 | 0.50 | −2.53 | −4.94 | −2.77 |

14 | Male | HLHC | 0.20 | 0.62 | −3.35 | −7.66 | −4.30 | 0.50 | −2.71 | −5.36 | −2.98 |

15 | Male | Borderline LV, CoA | 0.23 | 0.69 | −2.98 | −7.02 | −3.85 | 0.55 | −2.40 | −4.77 | −2.47 |

16 | Female | HLHC | 0.34 | 0.62 | −4.29 | −10.28 | −5.80 | 0.64 | −3.05 | −5.80 | −3.00 |

17 | Female | HLHC, HAA, CoA | 0.21 | 0.82 | −1.64 | −4.76 | −2.50 | 0.48 | −3.56 | −6.21 | −3.52 |

18 | Male | HLHC | 0.19 | 0.58 | −3.60 | −8.08 | −4.61 | 0.47 | −3.39 | −5.80 | −3.36 |