The prognostic impact of vascular calcification on abdominal aortic aneurysm progression

1 Objective: The maximal aortic diameter is currently the only clinically applied predictor of 2 abdominal aortic aneurysm (AAA) progression. It is known that risk of rupture is associated with 3 aneurysm size, hence accurate monitoring of AAA expansion is crucial. Aneurysmal vessel wall 4 calcification and its implication on AAA expansion are insufficiently explored. We evaluated the 5 vascular calcification using longitudinal computed tomography angiographies (CTA) of AAA 6 patients and its association with AAA growth. 7 Methods: We conducted a retrospective study of 102 AAA patients with a total number of 389 8 abdominal CTAs at six-month intervals, treated and followed-up at the Division of Vascular 9 Surgery, Department of General Surgery, Medical University of Vienna. Digitally stored CTAs 10 were reviewed for vascular calcification (volume and score) of the infrarenal aorta and common 11 iliac arteries as well as for morphometric AAA analysis. In the prognostic setting, slow versus 12 fast AAA progression was defined as < 2 or ≥ 2 mm increase in AAA diameter over six months. 13 In addition, to analyze the association of vascular calcification and AAA growth rate with 14 longitudinal monitoring data, a specifically tailored log-linear mixed model was employed. 15 Results: An inverse relation of increased abdominal vessel wall calcification and short-term 16 AAA progression was detected. Compared to fast progressing AAA, the median calcification 17 volume of the infrarenal aorta (1225.3 vs 519.8 mm3, P = 0.003), the median total calcification 18 volume (2014.1 vs 1434.9 mm3, P = 0.008) and the median abdominal total customized Agatston 19 calcium (cAC) score (1663.5 vs 718.4, P = 0.003) were significantly increased in slow 20 progressing AAA. Importantly, a log-linear mixed model efficiently predicted AAA expansion 21 based on current diameter and abdominal total cAC score (P = 0.042). 22 Jo ur al Pr e-p roo f

1 methods of assessing aortic diameter and vessel wall calcification are described in the current 2 literature, with many of them lacking in assessment reproducibility. 15,16 As a result, there is no 3 agreement on whether or not a calcification score can be used as a reliable AAA evaluation 4 parameter. [17][18][19] Based on our CTA-based assessment accuracy (six-month intervals), we 5 hypothesize that AAA progression is associated with the degree of vessel wall calcification and 6 increased vascular calcification may stabilize the aortic aneurysmal wall protecting against 7 progressive AAA expansion.  Patient demographics were recorded by a structured questionnaire and all study participants 12 underwent serial blood withdrawing and CTA every six months. 13 Morphometric CTA and calcification analyses 14 Individual morphometric AAA analysis was performed by analyzing CTAs with syngo.via 15 (version VB40B, Siemens Healthineers, Forchheim, Germany) and impax EE (Agfa-Gevaert, 16 Mortsel, Belgium) imaging software. Multiple measurements of the same CTA image (maximal 1 Multiplanar reconstructions were done by using images reconstructed in 1 or 2 mm slices. 2 We measured the vascular calcification applying the "CTA syngo Calcium Scoring" tool of 3 syngo.via software. First, a Hounsfield unit threshold was set individually for each CTA in order 4 to exclude the contrast agent within the vessel lumen next to the calcification plaques. All 5 distributions of pixels with a density above the defined threshold were detected and manually 6 color marked for each calcification plaque. The semiautomatic software allows accurate 7 visualization and quantification of calcified arterial lesions. For the calcification measurements, 8 four anatomic regions (infrarenal aorta, common iliac arteries and combined segments) were 9 defined separately and collectively (Figure 1). Subsequently, from the selected areas, the 10 software automatically calculated the segmental calcification volume and calcification score -11 customized Agatston calcium (cAC) score. 21

12
Statistical analysis 13 The results are presented as median and interquartile range (IQR) for continuous variables and 14 counts with percent sample group for nominal variables. For assessment of statistical 15 significance, non-parametric tests were employed (Mann-Whitney U test for group comparisons 16 of continuous variables). Slow versus fast AAA disease progression was characterized as a < 2 17 mm or ≥ 2 mm increase in AAA diameter over the next 6 months, reflecting previously 18 published cut-offs in the prognostic setting of 4 mm per year. 22

19
To analyze the association of vascular calcification and AAA growth rate with longitudinal 20 monitoring data, a specifically tailored log-linear mixed model was employed (incorporating , and thus projects AAA size (y) in a certain time period 1 (years) based on the current aneurysm diameter and calcification level at baseline. The natural 2 logarithm of two subsequent maximal AAA diameter measurements was calculated through the 3 time interval between the two measurements (in years) and the interaction of time interval and 4 vascular calcification (volume or score) as measured at the beginning of the time interval. The 5 effect of time interval was modelled as normally distributed random effect at patient level to 6 account for correlations of repeated observations within the same patient. Calcification values 7 were centered at the sample mean before entering the model to allow for direct estimation of the 8 average growth rate. No intercept was included, because at a theoretical time interval of length 9 zero, the AAA growth is zero by definition. Baseline maximal AAA diameter was not included 10 as predictor, because the model explains relative change, and hence, the initial maximal AAA 11 diameter of each time interval is included in the log transformed maximal diameter ratio which is 12 the outcome variable of the model.

13
Multivariable binary logistic regression was conducted to assess the biomarker value of vascular 14 calcification to predict fast AAA progression when adjusting for other calcification parameters, 15 comorbidities, and medication. During the course of data analysis, only variables with p < 0.1 in 16 univariate analysis were included in multivariable binary logistic regression analysis. Two-sided

1
Patient Collective 2 This retrospective study analyzed 389 CTAs of 102 AAA patients, who were followed-up with 3 serial blood withdrawing and CTA analyses in six-month intervals. The main objective of this 4 study was to determine the prognostic impact of vascular calcification on AAA progression.

5
Patient demographics are listed in Table 1. The majority of study participants were men, i.e. 89 6 (87.3%) and 13 (12.7%) women were included. The median age at study entry was 71 years for  Table 1. 21 Table 2). 8 Furthermore, we evaluated the biomarker value of vascular calcification volume and cAC score 9 to predict AAA progression in a log-linear mixed model. The total cAC score (infrarenal and 10 common iliac arteries combined) allowed a reliable AAA growth prediction and confirmed the 11 prognostic potential of vascular vessel wall calcification for AAA progression ( Table 3)  logistic regression analysis to identify fast progressors. After adjusting for morphological and 2 blood parameters as metric variables as well as comorbidities and patient medication as 3 categorical variables (Table 4), the total cAC score did not hold independent prognostic 4 information for AAA growth. Maximal AAA diameter (P = 0.017) was confirmed as 5 independent predictor of disease progression in multivariable analysis.

1
This study demonstrates a protective effect of increased arterial vessel wall calcification on AAA 2 progression, based on accurate, serial CTA analyses (six-month intervals) of substantial sample 3 size. Further, we confirmed the predictive potential of vascular calcification for AAA growth in a 4 log-linear mixed model, accounting for multiple measurements in patients over time. In 5 multivariable analysis (binary logistic regression) of AAA patients, the total cAC score did not 6 prevail and only the maximal AAA diameter was independently associated with faster AAA 7 progression over six months. It is known that the AAA growth rate is associated with the initial 8 aneurysm size and could therefore also be associated with the degree of calcification. 16 However, 9 AAA expansion represents a surrogate parameter and was nevertheless included in multivariable 10 binary logistic regression analysis with risk of overadjustment. Yet, confounding factors could be 11 present, although data acquisition and analysis make severe selection bias most unlikely. The Our results are in line with another study reporting that increased calcification reduces AAA 3 progression. However, this prior study achieved an interobserver variability of 1.4 mm and 4 vascular calcification was measured via ultrasonography. 18 In contrast, in our study, multiple 5 measurements of the same CTA image (maximal AAA diameter) were performed by two 6 independent experts, resulting in a mean intra-and interobserver variability ranging at 0.14 mm 7 and 0.20 mm, respectively. A study with reasonable selection bias, but yet calcification analysis 8 of AAA by multidetector-row computed tomography and calcification index, reported results 9 consistent to our findings. 24 Contrarily, a study, which investigated growth models of AAA 10 between 40 and 49 mm of maximal diameter found no significant influence of vascular 11 calcification on AAA growth patterns. However, this study did not quantify vascular 12 calcification parameters (volume and score) like in our study, but classified depending on the Originally, both, the Callister and Agatston methods are described to be able to only assess 20 calcification in non-contrast enhanced computed tomography images, because the signal 21 generated by the intravascular contrast agent interferes with that generated by the calcification of 22 the vessel wall. 19 However, the preferred imaging modality used to study AAA is CTA. 1 Of note, 23 J o u r n a l P r e -p r o o f we applied a method in our study to quantify the vascular calcification with a computerized 1 method, however, we set the Hounsfield units threshold individually for each CTA in order to 2 adjust for contrast agent and to be able to measure the amount of calcification as precise as 3 possible. Since CTAs can also be used for standardized quantification of arterial vessel wall 4 calcification, we applied the "CTA syngo Calcium Scoring" tool of syngo.via software (Siemens 5 Healthineers, Forchheim, Germany), which is usually used for quantification of calcified 6 coronary lesions. 31, 32 The calcium scoring tool allows accurate visualization and quantification 7 by automatic selection and subsequent essential manual defining for the calcified plaques. The 8 software provides a comprehensive analysis of calcification volume (mm³) and score (Agatston 9 method), if analyzed, also of the abdominal arteries (cAC score). 21

10
There are several limitations, which must be addressed. First, it should be noted that the data for 11 this study was collected retrospectively. The "CTA syngo Calcium Scoring" tool of syngo.via  patients. Compared to current literature with predominantly x-ray or ultrasonography 6 measurements of AAA calcification, we provide a many times higher sample size in addition to 7 increased accuracy of CTA analyses. 15 However, a further prospective study seems favorable and 8 necessary to confirm and add evidence of validity to identify fast AAA progression via vascular 9 calcification. Afterwards, the method to efficiently predict AAA expansion based on current 10 diameter and abdominal total cAC score is expected to be applicable in clinical practice. In 11 addition, after demonstrating the efficacy of computerized clinical decision support with regards 12 to vascular calcification on abdominal aortic aneurysm progression, it is conceivable that the 13 improved prognostic value of aneurysm growth has several positive effects on patient care. This 14 concerns improved patient education and diminished anxiety because of previously more 15 unpredictable AAA growth. We believe that it also improves the assessment accuracy of 16 aneurysm growth or AAA-size related rupture risk by surgeons. Considering AAA as 17 multifactorial disease, it seems logical that the more factors that are included in the growth rate 18 prognosis, the more precise it will be. If the intervals of extensive and costly CTAs can be 19 prolonged (or safely alternated with sonography) as well as surgical AAA repair can be delayed 20 based on this prognosis, patients and health care systems might further benefit.