Methods

Patient Population. This human subjects research was approved by the Cleveland Clinic Institutional Review Board (IRB #16-900, approved 8/1/2016). From 8/2016 to 5/2020, 98 adults undergoing elective replacement of ascending aorta aneurysms of any etiology were prospectively enrolled. Written informed consent for data collection and publication was obtained from patients or family members. Exclusion criteria were history of mediastinal radiation, aortitis, aortic valve endocarditis, and prior surgical intervention to zone zero.

Echocardiographic Strain. Intraoperative transesophageal echocardiographic images of the aortic root and distal ascending aorta with concurrent electrocardiogram (EKG) tracings were obtained during induction by the attending anesthesiologist. Proximal and middle ascending views could not be reliably obtained due to echocardiographic interference by the trachea. Continuous blood pressure monitoring was achieved via arterial line and obtained from the anesthesia record.

Circumferential global peak strain was calculated by (software) over the course of one cardiac cycle as determined by EKG tracing.Stiffness index and distensibility were subsequently calculated via the equations:

\[\mathrm{Stiffness Index} = \frac{100*\ln(\frac{SBP}{DBP})}{Strain}\]
\[\mathrm{Distensibility} = \frac{2*Strain}{SBP-DBP}\]

All strain images were reviewed by one of four reviewers, and images with subpar quality were reviewed by one cardiac anesthesiologist. A total of 125 images were included in this analysis from 69 patients. 51 (40.8)% were from the root and 74 (59.2)% from the distal ascending aorta.

Tissue Processing. Aortic tissue specimens were collected from the operating room and transported in sterile room temperature saline. When same day testing could not be performed, tissue was fast-frozen, preserved at -80°C, and thawed overnight in 4°C normal saline. For each region, a steel punch was used to create full-thickness circumferential dogbone specimens 15mm in length. Specimen thickness was measured using a custom caliper machine accurate to three thousandths of a millimeter and taken as the average of three locations.

Tissue Samples. A total of 153 samples were generated from 87 patients undergoing elective replacement of ascending thoracic aortic aneurysm. Samples were classified by region as previously described, yielding 33 proximal, 88 middle, and 32 distal ascending aortic samples, respectively. A total of 71% of samples were from patients with a tricuspid aortic valve and 82% from patients with bicuspid aortic valve.

Statistical Analysis. A random forest model

Results

Characteristic N = 98
Age (Years) 61.1 (51.4, 68.8)
Disease
Aneurysm 98 (100%)
Sex
Female 24 (24%)
Male 74 (76%)
Body Mass Index (kg/m2) 28.6 (25.2, 32.4)
Aortic Valve Type
TAV 43 (44%)
BAV 55 (56%)
Bicuspid Classification
I-RL 37 (70%)
I-RN 3 (6%)
II 7 (13%)
0 4 (8%)
Unknown 2 (4%)
Aortic Regurgitation
None/Trivial 36 (37%)
Mild 21 (21%)
Moderate 23 (23%)
Severe 18 (18%)
Aortic Stenosis
None/Trivial 75 (77%)
Mild 6 (6%)
Moderate 9 (9%)
Severe 8 (8%)
Max Aortic Diameter (cm) 5.0 (4.7, 5.4)
Aortic Area-to-Height Index (cm2/m) 11.1 (10.1, 13.1)
Aneurysm Phenotype
Small 11 (12%)
Root 19 (20%)
Ascending 42 (45%)
Extended 22 (23%)
Hypertension 63 (64%)
Stroke 4 (4%)
Diabetes 8 (8%)
Chronic Obstructive Pulmonary Disease 4 (4%)
Hyperlipidemia 48 (51%)
Coronary Artery Disease 40 (41%)
Chronic Kidney Disease 9 (9%)
Heart Failure 7 (7%)

Random Forest Estimates of Failure Stress

A random forest regression model was constructed for failure stress. The calculated r-squared for this model was 0.125 and error rate 0.21.

Table. Estimates of Variable Importance for Estimation of Failure Stress
Variable VIMP 5% CI 95% CI p-value Significance
Patient ID 2.128 0.437 3.820 0.007 **
Age (Years) 3.891 0.799 6.983 0.007 **
Height (m) -0.846 -1.517 -0.174 0.993
Body Mass Index (kg/m2) -1.845 -3.312 -0.379 0.993
Sex -0.227 -0.407 -0.047 0.993
Aortic Valve Type . . . .
Bicuspid Classification 0.019 0.004 0.035 0.007 **
Aortic Regurgitation -0.078 -0.141 -0.016 0.993
Aortic Stenosis 6.832 1.403 12.260 0.007 **
Hypertension -0.084 -0.151 -0.017 0.993
Stroke . . . .
Diabetes . . . .
Chronic Obstructive Pulmonary Disease . . . .
Hyperlipidemia 0.005 0.001 0.010 0.007 **
Coronary Artery Disease 0.078 0.016 0.140 0.007 **
Chronic Kidney Disease 0.022 0.004 0.039 0.007 **
Heart Failure . . . .
Region . . . .
Aortic Diameter (cm) -0.421 -0.756 -0.086 0.993
Max Aortic Diameter (cm) -0.183 -0.329 -0.038 0.993
Aneurysm Phenotype -0.192 -0.344 -0.039 0.993
Tissue Thickness 5.570 1.143 9.996 0.007 **
Global Peak Strain (%) -0.083 -0.149 -0.017 0.993
Stiffness Index 2.541 0.522 4.560 0.007 **
Distensibility 3.331 0.684 5.978 0.007 **

Random Forest Estimates of Failure Strain

A random forest regression model was constructed for failure stress. The calculated r-squared for this model was 0.652 and error rate 164.94.

Table. Estimates of Covariate Effects
Variable \(\beta\) 5% CI 95% CI p-value Significance
Patient ID 0.658 0.135 1.181 0.007 **
Age (Years) 47.875 9.829 85.920 0.007 **
Sex -0.017 -0.031 -0.004 0.993
Height (m) 0.094 0.019 0.168 0.007 **
Body Mass Index (kg/m2) -0.160 -0.288 -0.033 0.993
Aortic Valve Type . . . .
Aortic Regurgitation 0.014 0.003 0.025 0.007 **
Aortic Stenosis 2.603 0.534 4.671 0.007 **
Hypertension -0.055 -0.099 -0.011 0.993
Stroke . . . .
Diabetes . . . .
Chronic Obstructive Pulmonary Disease . . . .
Hyperlipidemia 0.061 0.012 0.109 0.007 **
Coronary Artery Disease 0.317 0.065 0.569 0.007 **
Chronic Kidney Disease 0.084 0.017 0.151 0.007 **
Heart Failure . . . .
Bicuspid Classification -0.029 -0.052 -0.006 0.993
Max Aortic Diameter (cm) 2.544 0.522 4.566 0.007 **
Aortic Area-to-Height Index (cm2/m) 3.255 0.668 5.841 0.007 **
Aneurysm Phenotype 0.382 0.079 0.686 0.007 **
Region . . . .
Aortic Diameter (cm) 2.381 0.489 4.273 0.007 **
Tissue Thickness 0.173 0.035 0.310 0.007 **
Global Peak Strain (%) 0.907 0.186 1.628 0.007 **
Stiffness Index 1.369 0.281 2.457 0.007 **
Distensibility 1.925 0.395 3.455 0.007 **

Variable importance (VIMP) for each covariate are found in Figure . The covariate with the great contribution to failure strain was age (\(\beta\)=47.875, (9.829-85.92)),