Working towards risk stratification for ascending aortic dilatation in pediatric Turner syndrome patients: results of a longitudinal echocardiographical observation

Since 2013, all our pediatric TS patients of eastern Austria under surveillance and treatment in our endocrinology department have regularly been examined. For all patients registered, phenotypical apparent TS was confirmed by karyogram. Inclusion criteria for this longitudinal echocardiographic evaluation were (1) age between 0 and 20 years at baseline and (2) a minimum of two echocardiographic examinations with a follow-up of a minimum of 1 year. Examinations were included at minimum intervals of 1 year to prevent overrepresentation of severely affected patients. Exclusion criteria were (1) lack of written informed consent from all patients and/or their relatives, (2) incomplete or insufficient examinations, (3) cardiac operations other than correction of CoA, (4) CoA repair less than six months before examination, and (5) other connective tissue disorders such as Marfan, Ehlers-Danlos, or Loeys-Dietz syndrome or inflammatory changes of the aorta such as Takayasu arteritis.

The study protocol was approved by the institutional review board of the Vienna University Hospital committee on human research (approval reference number 1197/2018).

All examinations were carried out by the same investigator (CP). The echocardiographic assessment — performed with a Vivid E9 GE Medical device (Horton, Norway) — included measuring diameters of the aortic valve (AV), aortic root (AoR), sinutubular junction (STJ), and ascending aorta (AoA) perpendicular to the aortic flow in the parasternal long axis using the inner-edge to inner-edge method. All values were normalized to the current BSA of the patient, and data is presented as corresponding Z-score values according to the dataset by Warren et al. [20]. Z-scores ≥ 2 were considered as indicating dilatation. Baseline data of all four aortic measurement levels were as well expressed in the TS specific Z-score percentile graphs by Quezada et al. [19]. The elastic properties of the AoR were evaluated by M-mode measurements in parasternal long axis view. Offline calculations of mean systolic and diastolic diameters were performed for the calculation of stiffness and distensibility as previously described [21]. Blood pressure was measured simultaneously on the right upper arm using a Philips SureSignsVS2 (Andover, MA, USA); in patients with CoA, blood pressure was measured at all four limbs.

Data analysis was performed with IBM Statistics, version 26.0 (SPSS, Inc., Chicago). Kolmogorov–Smirnov tests were used to test for parametricity. For continuous variables, Student’s t-test, and for nonparametric parameters, Kruskal Wallis/Man-Whitney U computations were carried out. Correlation structures were tested using Pearson’s and Spearman’s correlation coefficient, as applicable. Multiple regression analyses were used to quantify the impact of various presumed risk factors. Data is given as two-sided p-values, which were considered statistically significant if < 0.05.

Forty-nine patients of our outpatient clinic fulfilled the inclusion criteria and all approved their participation in this study. Demographic baseline data is depicted in Table 1, showing a median age at baseline with 9.7 ± 5.9 (range 0–19.8) years with a balanced age distribution with 12 patients ≤ 5 years, 13 patients each between 5 to ≤ 10 years and 10 to ≤ 15 years, and 11 patients > 15 years of age. Patients were followed-up for on average 2.8 ± 1.0 examinations (range 2–6) and a median follow-up time of 3.4 ± 1.5 (range 1.5–7.6) years. While on average, body mass index (BMI) percentiles were within the normal ranges, eight patients were deemed overweight (> 85th percentile) and one obese (> 96th percentile). Blood pressure measurements taken during the echocardiographic examinations showed normal average percentiles. Fourteen patients (28.6%) were diagnosed with BAV but no case of aortic valve stenosis or more than mild aortic regurgitation was noted. Seven of the 14 patients with a BAV presented with additional CoA and one further patient had isolated coarctation (2.0%), adding up to 16.3% of patients with CoA. Seven of the eight patients with CoA were operated on within 10 days after birth. Four patients received an angiotensin converting enzyme inhibitor, three of these four only during the first months after CoA repair.

The genetic results showed 27 patients (53%) with a complete monosomy 45X; the remaining 22 patients had different forms of mosaicism. Of the 14 patients with BAV, 11 had monosomy 45X, as well as six out of seven patients with combined BAV and CoA.

Patients with BAV did not show significant differences in baseline parameters concerning age, time of follow-up, auxology, or blood pressure measurements compared to the subgroup with tricuspid aortic valves (TAV), but were slightly younger and were examined significantly more frequently (Table 1).

In total, 78% of patients received growth hormone and 61% estrogen therapy. At the onset of growth (GH) and sex hormone replacement therapy (SHRT), patients were on average 6.7 ± 3.2 and 12.7 ± 1.6 years of age, respectively. Patients with complete monosomy were slightly younger at the start of GH replacement therapy compared to those with mosaicism (6.2 ± 3.4 vs. 7.1 ± 2.9 years, n.s.), SHRT was introduced at similar age (complete monosomy 12.8 ± 1.7 vs. mosaicism 12.6 ± 1.6 years, n.s.).

Aortic diameter measurements as well as Z-scores of the entire cohort are summarized in Table 2. Eight patients (16.3%) showed a Z-score ≥ 2 at baseline at one or more measuring points (AV:0, AoR:5, STJ:4, AoA:6).

Comparing the baseline measurements of our 49 patients with the percentile curves of Quezada et al. resulted in equal Z-scores at the AV level. Distally, the Z-scores according to the Warren data were consistently + 0.5 SD (AoR) and + 1 to + 1.5SD (STJ, AoA) higher.

In regard to this, none of our patients reached a Quezada-Z-score level ≥ 3 (i.e., cut-off level for medical treatment) even though there were three patients with a Z-score of 3 to 3.5 and one with a Z-score of 4.6 according to the Warren data.

Over the entire observation period, all parameters showed a net progression, but these changes per se were not statistically significant.

The fourteen patients with BAV — half of which also had CoA — did not show significant differences in baseline demographics or average length of follow-up compared to those with TAV (Table 1). Baseline aortic diameter Z-scores were markedly different especially at the more distal ascending aortic levels; however, significance was only reached at the level of the AoA (Table 2).

At the end of the observation period, in comparison to patients with TAV, Z-scores of those with BAV were significantly greater at the level of the STJ and AoA with 0.10 ± 1.04 compared to 1.27 ± 1.20, and −0.33 ± 1.28 compared to 1.14 ± 1.70, with p = 0.005 and p = 0.002, respectively. At the directly supravalvular measuring levels, median Z-scores were approximately +0.5 higher in those with BAV, although this did not show statistical significance. Patients with TAV were most likely to develop aortic dilatation at the level of the AoR, while for patients with BAV, dilatation presented more distally. In total, the TAV group exhibited rather low Z-score progressions over the observation period between 0.03 ± 0.26/year (STJ) and 0.16 ± 0.43/year (AV). The BAV group showed relatively similar progression rates, but reaching significantly higher progression rates at the STJ level (0.26 ± 0.35/year). Accordingly, Spearman’s test showed moderate correlation of BAV with aortic dimensions at STJ and AoA levels despite the small subgroup sizes (STJ: r = 0.35; AA: r = 0.41; both p < 0.001).

At the end of follow-up, the number of patients with a Z-score of ≥ 2 at any level of the ascending aorta had increased from eight to 14 (28.6%). Three (8.6%) of TAV versus five (35.7%) of BAV patients showed dilatation at baseline compared to seven (20%) of TAV and seven (50%) of BAV at the end of follow-up, with one patient with BAV having an entirely dilated aorta at all four measuring points. This corresponds to a fourfold increased risk of at least one Z-score ≥ 2 (odds ratio 4.15) in BAV TS patients (Fig. 1).

At all four measurement levels of the ascending aorta, patients with monosomy 45X tended to show larger aortic diameters at baseline as well as at end of follow-up, however, statistical significance was not reached.

Spearman correlation equations revealed a distinct correlation of age and aortic diameter z-scores (AV r = 0.41, p < 0.001; AoR: r = 0.45, p < 0.001; STJ: r = 0.39, p < 0.001; and AoA: r = 0.25; p = 0.003). Therefore, we divided our TS population into age subgroups of < 10 and > 10 years of age in order to investigate whether changes can be pinpointed to a certain age range, the start of puberty or the start of estrogen therapy. Both groups were of the same size with n = 25 vs. n = 24, and the remaining demographic data at baseline did not differ, although length of follow-up was shorter in the older age group. Aortic Z-scores showed marked differences at all points of measurement, the younger subgroup showed on average rather smaller aortic diameters compared to the older subgroup with −0.30 ± 1.07 vs. 0.46 ± 0.61 at the AV (p = 0.004), −0.12 ± 0.98 vs. 0.78 ± 0.96 at the AoR (p = 0.002), −0.33 ± 1.00 vs. 0.75 ± 0.90 at the STJ (p < 0.001), and −0.53 ± 1.58 vs. 0.48 ± 1.78 at the AoA (p = 0.041). When only including patients with TAV in this sub-analysis, the results were similar. Due to the small subgroup of BAV patients of these different age subgroups, we refrained from a sub-analysis of these patients.

At the end of follow-up, with the younger group having aged on average 3.8 years, results were still markedly different to the data of the older age group at baseline but did not reach statistical significance with Z-scores of 0.2 compared to 0.81 at the level of the AV, 0.29 compared to 0.83 at AoR, −0.16 compared to 0.73 at the STJ, and −0.29SD compared to 0.46SD at the level of the AoA (p = 0.054, p = 0.088, p = 0.096, p = 0.092, respectively) (Fig. 2).

In graphical representations, a decelerating increase in Z-scores was observed at all measuring points, starting approximately at twelve years of age (Fig. 3).

Aortic elasticity parameters were routinely calculated. With a stiffness index of approximately 4.5, pathological levels were reached both at baseline and end of follow-up (Table 2). Subdivision according to the presence of BAV, age group, and karyotype also showed similar values over the entire observation period. Robust correlation analysis showed that elasticity parameters did not correlate with aortic diameter growth and did not predict absolute aortic size.

Assumed risk factors BAV, COA, age, genotype, and SHRT were integrated by multivariate regression. At all levels of measurement, age was the leading predictor of Z-score outcomes (AV adjusted R² 21.6%, p < 0.001; AoR 29.9%, p < 0.001; STJ 30.6%, p < 0.001; AoA 11,8%; p = 0.016). BAV improved the model significantly at all levels except for the AV (AoR 36%, p = 0.040; STJ 40.8%, p = 0.007; AoA 43%, p < 0.001). The change was most pronounced at the level of the AoA; here, the presence of BAV increased Z-scores by 2.7 (p < 0.001). Other presumed risk factors did not lead to a significant alteration of the model and SHRT could not be investigated due to age-dependency.