Objective. Periaortic fat tissue (PaFT) has been implicated in the progression of abdominal aortic aneurysms (AAAs). Therefore, its quantification as a prognostic marker for aneurysm expansion has attracted clinical interest. Most existing research on PaFT, however, is based on unenhanced aortic CTscans, whereas the CT diagnosis of aortic aneurysms is usually performed with enhanced CT angiographies. The objective of this study is to examine the feasibility of measuring abdominal periaortic fat

more »

... tissue in enhanced aortic CT-scans using a new method based on the OsirixMD post-processing software and evaluate any methodological issues/considerations arising from it, in order to reliably quantify periaortic fat tissue from enhanced and unenhanced CT-scans. Methods . In a derivation cohort (n= 101), PaFT Volume and PaFT mean HU value were measured within a 5 mm-wide periaortic ring in arterial phases and compared to the same values from native scans. Fat tissue was defined within the range of -45 to -195 Hounsfield Units (HU). After testing their correlation, fat tissue values from both CT phases underwent linear regression through the origin to define a correction factor (slope of the line of best fit), allowing the conversion of arterial back to native scores. This conversion factor was then applied to fat tissue values in a different validation cohort (n=47) and the agreement of the corrected fat tissue values and values in the native scans was examined using Bland-Altman plots and Passing-Bablok regression. In a secondary study the pooled date sets from both studies (n=148) were stratified in an AAA and non-AAA group and the average fat tissue values for both groups (with PaFT volumes adjusted for aortic size) were calculated using both native and corrected arterial values. Results. In the derivation cohort, periaortic fat tissue Volume and mean HU value showed very high correlations between arterial and native scans (r> .99 and r= .95 respectively, p< .0001 both). Linear regression defined a conversion factor of 1.1057 for arterial periaortic fat tissue Volume and 1.0011 for arterial periaortic fat tissue mean HU. Potential confounding factors (mean intraluminal contrast density, aortic wall calcification, longitudinal contrast dispersion, aortic diameter, CT-tube voltage, slice thickness, image noise) showed no significant impact in multivariate regression. Application of the conversion factors in arterial scans of the validation study resulted in corrected arterial fat tissue values that showed very good agreement with PaFT values in native scans. Bland Altman analysis showed the following mean differences [95% confidence interval]: 0.36 [-0.01 to 0.73] for periaortic fat tissue Volume and 0.83 [-1.08 to 0.1] for periaortic fat tissue mean HU. Passing-Bablok regression confirmed minimal/no residual bias. Median periaortic fat tissue size-adjusted PaFT Volumes and Mean HU values from the Mann-Whitney test showed no significant difference between the AAA and non-AAA groups. Conclusion. Periaortic fat tissue Volume and mean HU values demonstrate only minimal variation between arterial and native scans and can be measured in enhanced aortic CT scans with very high reliability. Periaortic fat tissue Mean HU value, unlike Volume, is independent from the presence of paraaortic organs. Certain issues, like non-circular aortic discs, histological boundaries of periortic fat tissue and dependence from Body Mass Index and other fat tissue depots need to be explored further. Summary 2 1. ZUSAMMENFASSUNG Ziel. Das periaortale Fettgewebe spielt bei der Progression von Aortenaneurysmen eine Rolle, so dass seine Quantifizierung als prognostischer Marker für die Aneurysmaprogression von besonderem klinischem Interesse ist. Die aktuelle Forschung ist basiert jedoch fast ausschließlich auf nativen CTs, während Aortenaneurysmen üblicherweise nur mittels kontrastmittelverstärkten CT angiographien dargestellt werden. Das Ziel dieser Studie ist die methodische Überprüfung der Bestimmung vom abdominalen periaortalen Fettgewebe in kontrastmittelverstärkten CTs mit der frei verfügbaren OsirixMD Softwareanwendung und die Evaluation von potenziellen Faktoren, die eine zuverlässige periaortale Fettgewebsquantifikation in nativen und kontrastverstärkten CTs ermöglichen. Methodik. In einer Derivationsgruppe (n=101), wurde das Fettgewebsvolumen und die HU Mittelwerte innerhalb von einem 5 mm breiten periaortalen Ring in arteriellen CTs bestimmt und die Werte wurden mit entsprechenden Werten aus nativen CTs verglichen. Das Fettgewebe wurde als HU Werte -45 bis -195 HU definiert. Die Fettgewebswerte von beiden CT-Phasen wurden auf Korrelation überprüft und anschließend einer linearen Regressionsanalyse unterzogen, wobei ein Konversionsfaktor bestimmt wurde, um arterielle in nativen Fettgewebswerten zu konvertieren. Der Konversionsfaktor wurde danach in einer zweiten Validierungsgruppe (n=47) angewendet. Sodann wurde die Übereinstimmung von korrigierten arteriellen und nativen Fettgewebswerten mittels Bland-Altmann Plots und Passing-Bablok Regressionsanalyse überprüft. In einer Sekundärstudie, wurden die gepoolten Datasets beider Studien (n=148) in einer Bauchaortenaneurysma-und einer Nichtbauchaortenanerysmagruppe stratifiziert, um die Mittelwerte von Fettgewebsvolumen (adjustiert für Aortengröße) und HU Mittelwert in beiden Gruppen zu bestimmen. Ergebnisse. In der Derivationsgruppe, zeigte das Fettgewebsvolumen und der HU Mittelwert eine sehr hohe Korrelation zwischen kontrastverstärkten und nativen CTs (r > 0,99 und r= 0,95 entsprechend, p< 0,0001 für beide). Die lineare Regressionsanalyse ergab einen Konversionsfaktor von 1,1057 für das Fettgewebsvolumen und 1,0011 für den Fettgewebs-HU Mittelwert. Potenzielle Störfaktoren (intraluminale Kontrastmitteldichte, Aortenwandkalzifikation, longitudinale Kontrastmittelverteilung, Aortendiameter, CT-Röhrenspannung, Slicestärke, Größe der intraluminalen Kontrast-ROI, Bildrauschen) zeigten keinen signifikanten Einfluss in der multiplen Regressionsanalyse. In der Validierungsgruppe, zeigten die mittels Konversionsfaktor korrigierten Fettgewebswerte der arteriellen Phase eine sehr hohe Übereinstimmung mit den Fettgewebswerten der nativen CT-Phase. Die Bland-Altman Analyse ergab folgende mittlere Differenzen [95% Konfidenzintervall]: 0,36 [-0,01 bis 0,73] fürs Volumen und 0,83 [-1,08 bis 0,1] für den HU Mittelwert. Die Passing-Bablok Regressionsanalyse bestätigte ein minimales bzw. kein residuales Bias. In der Sekundärstudie, zeigten die Mediane der Fettgewebswerte aus dem Mann-Whitney Test keinen signifikanten Unterschied zwischen der BAA und nicht-BAA Gruppe. Schlussfolgerung. Periaortales Fettgewebsvolumen und HU-Mittelwert zeigen eine minimale Variation zwischen arteriellen und nativen CTs und lassen sich in kontrastverstärkten Aorten-CTs sehr zuverlässig bestimmen. Der Fettgewebsmittelwert ist von der Präsenz anderer paraaortale Organe unabhängig. Gewisse Faktoren, z.B. nicht-zirkuläre aortalen Scheiben, histologische Grenzen des periaortalen Fettgewebes und seine Abhängigkeit vom Body Mass Index und anderen Fettgewebskompartimenten benötigen eine weitere Analyse. Effect of periarterial fat tissue on the arterial wall Blood vessels are surrounded by adventitial perivascular fat tissue that may regulate vascular functions. 5 Increased perivascular fat volume in animals is associated with pronounced inflammation, higher oxidative stress, and vascular smooth muscle cell proliferation. 6-8 Similarly, PaFT adjacent to human atherosclerotic aortas is characterized by extensive macrophage infiltration and a shift in adipokine expression. 9,10 Therefore, perivascular fat tissue seems to exert a local paracrine effect, potentially contributing to atherosclerotic alterations of the vessel wall. 11,12 This, however, is not a one-way effect and a "crosstalk" between the perivascular adipose tissue and the vessel wall has been proposed. Perivascular fat tissue promoting vascular wall inflammation Cultured mice arteries show inflammatory factors secreted from the perivascular fat tissue that promote significant pro-oxidative and proinflammatory phenotypic alterations in the vascular wall. 7 This points to a localized perivascular adipose tissue inflammation causing an exacerbation of vascular oxidative stress which leads to increased macrophage infiltration. The latter causes a significant proinflammatory shift in the cytokine/chemokine profile secreted from periarterial fat. 7 Corroborating Possible role of periaortic fat tissue in AAA pathophysiology -Histological evidence PaFT regulates aortic function through a large number of vasocrine molecules, like cytokines, which contribute to vascular inflammation in atherosclerosis, as well as adipokines, e.g. interleukin-6, whose plasma levels are increased in AAA patients. 37 Specimens from human AAAs, periaortic adipose tissue, and fat tissue surrounding peripheral arteries collected during AAA surgical repair were examined for necrotic adipocytes, type of infiltrating leukocytes and expression of proteases. 38 PaFT was found in AAAs to larger extent compared with control aortas from healthy organ donors and an increased presence of inflammatory cells was found in the PaFT of AAAs. 38 The findings included high numbers of neutrophils, macrophages, mast cells, and T-cells in PaFT around AAAs as well as close to necrotic fat tissue. 38 The basic concept for PaFT quantification in non-enhanced MDCT was introduced by Schlett et al, who in 2009 quantified aortic and thoracic PaFT in 100 patients and showed that both abdominal and thoracic PaFT were correlated with both visceral and subcutaneous abdominal fat, waist circumference and BMI. 40 The authors concluded that "standardized semiautomatic CT-based volumetric quantification of PaFT is feasible and highly reproducible." 40 The intra-observer agreement was excellent for abdominal and thoracic PaFT (Intraclass correalation coefficient-ICC= .970 and .986) as