CARDIOVASCULAR IMAGING Year : 2020  Volume : 30  Issue : 2  Page : 170176 Determining the normal effective diameter of thoracic aorta in pediatric population of India Sunita V Kale, Shah Alam Department of Radiology, Seth G.S. Medical College and K.E.M Hospital, Parel, Mumbai, Maharashtra, India Correspondence Address: Background: It is imperative to establish normative ranges of aortic diameter to diagnose various aortic pathologies. There have been very few studies establishing the normal aortic diameter on crosssectional imaging, and none pertaining to the Indian pediatric population. The objective of this study was, therefore, to establish the normal effective diameter of thoracic aorta at multiple levels using computed tomographic data, calculate zscores, and plot reference curves. Subjects and Methods: The effective thoracic aorta diameters (average of anteroposterior and lateral diameters) were measured at predefined levels (aortic root, ascending aorta at the level of right pulmonary artery, aortic arch, proximal descending aorta, and aorta at the level of diaphragmatic hiatus) on doubleoblique reconstructed computed tomography (CT) images perpendicular to the direction of the vessel. Multiple functional forms relating the effective diameter to subjects' age were evaluated with least square regression methods, and further R^{2} was used to ascertain the best model. Agebased formulas to derive normal aorta diameters and mean squared errors (MSEs) were established. Results: Two hundred and seven contrastenhanced CT (CECT) thorax studies of children without known cardiovascular disease were studied. The polynomial regression model relating the effective diameter that included linear, quadratic, and cubic age terms as independent variables were found to the best statistical model. The z scores were calculated, and normative curves were plotted. Conclusions: We have established normative effective diameters of the thoracic aorta at multiple levels in Indian children of different age groups. Measurements outside of the normal ranges are indicators of ectasia, aneurysm, hypoplasia, or stenosis.
Introduction The normal standards for the aortic diameter at various levels have been established for the adult population and can be used to determine aneurysm formations or stenosis.[1] In contrast, similar standards for infants (1 month to 1 year old), children (1–12 years old), and adolescents (13–17 years old) are not as well established,[2] and such standards pertaining to Indian pediatric population have not yet been published in the literature. A prerequisite for identifying abnormal is to establish the normal. Effective aortic diameter assumes significance in early detection of diffuse aortic hypoplasia in conditions such as Williams syndrome or aneurysmal dilatation in children with connective tissue disorder.[3] Anomalies of the aorta in pediatric age group include coarctation, residual findings after catheterguided interventions or surgery, connective tissue diseases such as Marfan syndrome and dilatation of the aortic root associated with aortic valve anomalies, nonspecific aortoarteritis or postsurgical in patients with congenital heart diseases.[4] Though echocardiography is the standard method for determining the size of the thoracic aorta in children, a recent review of echocardiographic methods showed a general lack of standardization in technique.[2] The evaluation of thoracic aorta on echocardiography relies on planar measurements rather than on transverse measurements. Crosssectional imaging using computed tomography (CT) and magnetic resonance imaging (MRI) with multiplanar reconstructions does overcome these limitations. Effective aortic diameter is average of the transverse and anteroposterior diameters of the aorta, this method of measurement nullifies errors due to obliquity and has been previously used for similar studies.[2],[5] Subjects and Methods This was a single institutional crosssectional observation study. The study included children and young adults of age group 'zero' to eighteen years who underwent contrastenhanced CT (CECT) thorax scans at our institute during the 13month period between January 2016 and January 2017. Exclusion criteria included a) History of congenital heart disease/dysmorphisms. b) History of cardiovascular disease or cardiothoracic surgery. c) Patients who are being evaluated for cardiac diseases. In patients where multiple CT examinations were performed during the study period, only the first of such CT scans were included in the study. The study was approved by our Institutional Ethics Committee. During the study period of 13 months, a total of 321 CT studies were evaluated, out of which 207 satisfied the abovedescribed inclusion and exclusion criteria. Out of the excluded studies (n = 114), prior history of cardiovascular diseases or surgeries and subjects under evaluation for cardiac diseases (n = 68) formed the majority followed by repeat examinations in the study window (n = 27) and scans excessively degraded by motion artifacts (n = 19). All the CT scans included in the study were performed on a 64slice CT scanner (Philips Brilliance 64slice CT, Koninklijke Philips N.V). Nonionic iodinated contrast agent with an administration rate of 12.5 mL and a dose of 12 mL/kg [not exceeding 100 mL] along the peripheral venous route, followed by a saline chaser of 1020 cc was used. CT data were obtained in keeping with the as low as reasonably achievable (ALARA) principle with a weightbased variable dose parameters (80120 kVp, 20150 mAs) with scans performed from thoracic inlet to the level L1L2. Image data were analyzed on a workstation (Terarecon AQI viewer) after image reconstruction of 1mm slice thickness. Multiplanar reformations (MPR) were created using a workstation. All MPR with doubleoblique reconstructions were obtained perpendicular to the aorta [Figure 1].{Figure 1} The effective diameter at each level was determined by averaging the anteroposterior and lateral diameter measurements. Measurements were obtained by using an electronic cursor at the outer widest diameter of the vessels. The measurements were obtained at the following five predefined locations: Aortic root, ascending aorta at the level of the right pulmonary artery, aortic arch, proximal descending aorta (distal to the aortic arch where the descending aorta obtains a cranialcaudal orientation), and aorta at the level of diaphragmatic hiatus. Statistics The effective diameters at various levels were tabulated against the subject's age. Descriptive statistics were employed to calculate the mean, standard error (SE), and standard deviation (SD) of the aortic diameter at various levels for different age groups separately for both boys and girls. Regression analysis was used to describe the relationship between the aortic diameter (dependent variable) and the subject's age (independent variable). Multiple regression models as described by previous studies[2],[5],[6] were analyzed to determine the best fit model. Linear, logarithmic, exponential, and polynomial regression models with quadratic, cubic, and linear terms were evaluated using the R2 value to determine the best fit functional form. The intercepts for linear, cubic, and quadratic terms were determined and were tested for significance. Scatter plots were used to determine the equation of independent variables at various levels. The best regression model was used to plot the trend line in the scatter plot. From the slope estimates of the best fit model, formulas were specified for the predicted diameters along with R2 for each of them. The effect of gender on the aortic diameter was determined by comparing the means of the diameter in male and female subjects. From these regression formulas, estimated mean squared error (MSE) was calculated. Predicted estimates of the aortic diameter were then calculated using the regression models. The z scores were calculated and then used to plot charts that can be used to determine the normal aortic diameter within the confidence interval of 95% (z = 2). These statistical analyses were performed on Excel (Microsoft) and Statistical Package for the Social Sciences (SPSS) (IBM) software. Results The age and sex distributions are summarized in [Table 1]. The youngest patient included in the study was a 10dayold infant, and the oldest patient was 18 years old. The median age of the study population was nine years.{Table 1} The descriptive statistics data of the effective diameter of aorta and multiple locations has been summarized in [Table 2], and further subgroup analyses have been made and specified. [Table 3] summarizes the mean, SD, and the SE of the effective aortic diameter at different levels in gender subgroups.{Table 2}{Table 3} On regression analysis, the best model was the polynomial regression model of an effective diameter that included linear, quadratic, and cubic terms as independent variables. An example of regression models employed for selection of the best fit model is provided in [Table 4] along with respective R2 scores.{Table 4} For all levels, the intercept and linear, quadratic, and cubic terms were significant (all P < 0.05). The formulae for calculating the predicted diameters along with R2 for the model used (polynomial regression model of order three) are tabulated in [Table 5].{Table 5} Predicted diameters were calculated for each level and age group using the polynomial regression models with cubic terms determined previously. MSE was also calculated for each of the models. z scores were then calculated using the following formula, z = (observed diameterpredicted diameter)/√MSE. The z scores calculated are of approximate normal distribution, they have a mean of zero and SD of one. They represent how many SDs above or below the observation is in relation to the mean (predicted regression line). A z value of 1 signifies that the observed value is 1 SD above the estimated mean of that level at that agegroup, whereas a z of 1 signifies that the value is 1 SD below the mean. Assuming a normal distribution, approximately 68.3% of the population will fall within the mean ±1 SD interval. Whereas 95.4% of the population is within the mean ±2 SDs. This data has been plotted in the form of graphs [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6] that do not require any complex calculations to determine the normal. These graphs contain the mean for age group and ±2 z score barricade lines.{Figure 2}{Figure 3}{Figure 4}{Figure 5}{Figure 6} Discussion It is imperative to acquire a complete and thorough knowledge of normality and its variants to study and diagnose abnormalities and pathologies with certainty. Although the normal standards for the diameter of thoracic aorta have been established for adults, such standards are not well established in pediatric population. Though echocardiography is the standard method for determining the size of the thoracic aorta in children. A recent review of echocardiographic methods showed a general lack of standardization in technique.[1] Crosssectional imaging (CT & MRI) standards of the normal aortic diameter in children are not established. Our study aims to establish the normal aortic diameter at various levels of the thoracic aorta on CECT thorax studies. We analyzed CECT studies in 207 children who had no history of cardiovascular disease or cardiothoracic surgery. Effective diameters of the thoracic aorta were measured by doubleoblique reconstructions perpendicular to the aorta. Measurements were acquired at the aortic root, ascending aorta, arch of the aorta, proximal descending aorta, and the descending aorta at diaphragmatic hiatus. The effective diameter is the average of anteroposterior and transverse measurements. The youngest subject of our study was a 10dayold infant and the eldest was 18 years of age. Regression analyses of the data were done, multiple regression models like linear, logarithmic, exponential, polynomial with quadratic, and cubic terms were analyzed and the best fit functional form for our data was selected by comparing the R2 values for each model. The bestfit form was found to be polynomial regression with cubic terms at all the levels studied and had R2 values of more than 0.9. Fitzgerald et al.[7] in 1987 studied thoracic aortic diameter in 97 children aged between 2 weeks and 19 years and found a linear relationship with thoracic aortic diameter at various levels with age and with thoracic vertebral body width. Like our study, they did not find a significant distinction between male and female groups. However, they used only axial CT images with 510 mm thick axial sections without multiplanar reconstructions perpendicular to the aorta. Akay et al.[6] reviewed CECT chest scans of 133 pediatric patients to measure descending and ascending thoracic aortic diameter. They found that the ratio of the aortic diameter to that of the thoracic vertebral diameter is a constant, about 1.1 at the level of ascending aorta. Wolak et al.[8] determined the aortic diameter at various levels on noncontrast cardiac CT and defined the normal limits in relation to age, sex, and body surface area (BSA). However, pediatric population was not included in the study. Kaiser et al.[4] assessed the normal values for aortic diameters in 53 children and adolescents by contrastenhanced cardiovascular magnetic resonance (CMR)angiography, with doubleoblique maximum intensity projections perpendicular to the aorta. Their study found a linear relationship between the crosssectional aortic diameter with the square root of BSA. However, their study lacked any data on children aged less than 2 years. Mohiaddin et al.[9] measured the normal dimensions of the thoracic aorta in 70 healthy volunteers on MR imaging. They used enddiastolic spinecho images in oblique planes through the ascending aorta, transverse aorta, and the descending aorta. They correlated these measurements with the BSA and found a linear correlation. However, the youngest subject of the study was 10 years old, and the study had no information on children aged younger than 10 years. The youngest subject of our study was 10 days old. Hegde et al.[2] determined the normal effective diameter at various levels of the aorta on CECT studies in children. They included 88 thoracic and 110 abdominal scans in the study. They measured the average of the anteroposterior and the lateral diameters of the thoracic and abdominal aorta at various levels on 1 mm collimation double oblique reconstructions perpendicular to the course of the vessel. They calculated the z scores at each level for a particular age group. As with our study, they derived a polynomial regression model with cubic terms relating to the aortic diameters and log BSA. They found a significant sex difference in the study population. Bayindir et al.[10] evaluated thoracic CECT studies, and measured the diameters of ascending aorta, descending aorta, main pulmonary artery, and right and left pulmonary arteries. They concluded that the diameters of the thoracic vascular structures increased with age and found a significant statistical difference among the age groups and genders, with higher dimensions in male children. However, the study measured aortic dimensions at two locations and did not attempt regression analysis of the statistical data. Limitations of our study As the scans included in the study were done for noncardiac indications, electrocardiographic gating was not routinely performed. This resulted in significant cardiac motion artifacts in some cases, which could have introduced error in measurements, especially at the aortic root. The measurements acquired are neither endsystolic nor enddiastolic measurements. The measurements were neither the true maximum nor minimum but rather intermediate effective diameters. A major limitation of the study was the small number of the study population and the fact that it was carried out at a single institution which might not be a true representation of the normal population. Conclusions Effective aortic diameter increases with age, however, their relationship with age is not linear. A polynomial regression model with cubic terms is the best fit functional form to describe the relation between aortic diameter and age, at all the levels studied. The R2 values of the study model were high (>0.9) and significant at all levels. The range of normal effective diameters of the aorta at multiple levels, the predicted mean and the ±2 SDs values were determined and plotted on graphs. The knowledge of these normal ranges and the use of graphs can aid the radiologist in diagnosing abnormalities like ectasia, aneurysm, stenosis, hypoplasia, etc. Acknowledgement The authors would like to show our gratitude to Dr. Ravi Ramakantan, ExHead Dept of radiology, GSMC and KEMH, for his guidance and comments that greatly improved the manuscript. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent forms. In the form the patient(s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest. References


