Thirty subjects underwent sodium [18F]fluoride PET-CT and PET-MRI examinations as part of the Assessment of Risk in Thoracic Aortopathy using sodium [18F]fluoride study (AoRTAs; NCT04083118). All participants were under surveillance for bicuspid aortic valve disease (diagnosed with transthoracic echocardiography with or without adjunct CT or MRI) and were aged over 40 years. Subjects with previous aortic root replacement, previous aortic dissection or rupture, contrast allergy or pregnancy were excluded. The study was approved by the Scottish Research Ethics Committee (REC reference: 18/SS0136), the United Kingdom Administration of Radiation Substances Advisory Committee and local institutional review board. Written informed consent was obtained from all participants.

All participants were administered a target dose of 250 MBq sodium [18F]fluoride intravenously and were imaged after 60 min using a hybrid 128-slice PET-CT scanner (Biograph mCT, Siemens Healthineers, Germany). A low-dose attenuation correction CT scan was performed (100–120 kV, 40–50 mAs, 5/3 mm). PET data were acquired using electrocardiogram (ECG)-gating in list-mode in three 10-min bed positions ensuring coverage of the entire thoracic aorta and heart. PET-CT images were corrected for attenuation, dead time, scatter and random coincidences, using an optimised iterative reconstruction algorithm (ultra-HD; TrueX + Time of Flight, matrix 200, zoom 1; 5 mm Gaussian filter).

Upon completion of the PET-CT examination, the subjects immediately underwent PET-MRI on a hybrid PET-MRI scanner (Biograph mMR, Siemens Healthineers, Germany) allowing simultaneous MRI and PET data acquisition. Using PET compatible elements from a 12-channel body matrix and a spine matrix coils, all images were acquired using ECG-gating. Both Dixon (end-expiration, breath-held, 3-dimensional, dual-echo spoiled gradient-recalled echo [10]) and free-breathing radial GRE (gradient recalled echo; RadialVIBE [11, 12]) sequences were acquired. Following acquisition, PET data were reconstructed using the following parameters to provide optimum signal-to-noise and contrast-to-noise ratios: 344 matrix, 3 iterations, 21 subsets, 5-mm Gaussian filter. Three final attenuation correction maps were generated based upon MRI sequences performed: a standard Dixon map (segmentation into 4 tissue types: air, lung, soft tissue and fat) and two custom attenuation correction maps based on radial GRE (RadialVIBE-2 with segmentation into 2-tissue types [lung/air and soft tissue]) and RadialVIBE-4 with segmentation into 4-tissue types [soft tissue, lung, fat, background air]). All analyses were performed on the ordered subsets expectation maximisation (OSEM) algorithm image reconstructions.

Aortic valve morphology was assessed using aortic valve CINE images and classified using international consensus classification [13]. The aortic valve function was assessed using 2-dimensional phase contrast magnetic resonance images at the sinotubular junction (with the velocity encoded threshold set to the lowest value without aliasing). Aortic valve velocities and velocity gradients were assessed by transthoracic echocardiography. Valve stenosis was determined using the maximum aortic velocity and graded as none (< 2.0 m/s), mild (2.0-2.9 m/s), moderate (3.0-3.9 m/s) or severe (≥ 4.0 m/s) [14]. Valve regurgitation was assessed using cardiac MRI images and graded as none, mild, moderate or severe based on visual assessment of the valve, the presence or absence of flow reversal in the thoracic aorta and regurgitation fraction by an accredited consultant with specialist expertise in MRI blinded to the PET results (MRD) (calculated using 2-dimensional phase contrast sequences and stroke volume differential).

PET-MRI images were qualitatively assessed visually for similarity to PET-CT images and the presence of artefact at tissue type interfaces by one investigator (JN). Quantitative sodium [18F]fluoride uptake was assessed using FusionQuant v1.21.0421 software (Cedars-Sinai Medical Centre, Los Angeles). Images from PET-CT and each PET-MRI reconstruction were co-registered in three orthogonal planes using magnetic resonance angiography. Background blood pool activity was measured by drawing one 2.1-cm3 sphere of interest in both the right and left atria and calculating the average standardised uptake value divided by volume (background SUVmean).

Previously described method of determining the total standardised uptake value (total SUV), average aortic standardised uptake value (SUVmean) and the maximum standardised uptake value (SUVmax) were used for assessment of the ascending aorta. This method has been shown to be highly reproducible and repeatable in aortic PET quantification [15, 16] (Fig. 1).

Method of determining aortic sodium [18F]fluoride activity in positron emission tomography (PET) images. PET images are first co-registered with magnetic resonance angiography in three orthogonal planes. (a) The ascending aortic activity is measured by drawing a centreline from the sinotubular junction to the brachiocephalic artery (green line). Calculated values are total standardised uptake value (total aortic SUV), mean standardised uptake value (aortic SUVmean), maximum standardised uptake value (SUVmax) and aortic microcalcification activity (AMA). Blood pool estimates are obtained by drawing a 2.1-cm3 sphere of interest (green circle) in the left atrium (b) and right atrium (c)

In brief, the images were analysed using a centreline-based approach to create a volume of interest. Starting at the sinotubular junction and ending at the brachiocephalic artery, a centreline was created at a minimum of three levels within the ascending aorta using a multiplanar reconstruction viewer. The diameter was drawn to the maximal internal aortic diameter then increased by 4 mm. Contouring around the aorta was drawn using a variable radius function to accommodate any anatomical vessel tortuosity or focal dilatation and reduce partial volume effects. The variable radius function has been demonstrated to improve sodium [18F]fluoride quantification in abdominal aortic aneurysm analysis as described previously [16]. After drawing the target volume of interest, total SUV, SUVmean (total SUV within the aorta divided by its volume) and SUVmax (maximum voxel intensity with the drawn region of interest) were generated. Aortic microcalcification activity (AMA) was calculated as the cumulative voxel intensity in the region of interest divided by its volume and corrected for background sodium [18F]fluoride activity as described previously [16].

Statistical analysis was performed using the software package R (v4.0.2, R Foundation Statistical Computing, Vienna). Shapiro-Wilk tests were used to assess normality of distribution of results. Categorical variables were presented as number (percentage). Continuous variables with normal distribution were present as mean (± standard deviation), and non-normally distributed variables were presented as median [interquartile range]. Associations between sodium [18F]fluoride uptake in PET-CT and PET-MRI and associations between aortic valve mean pressure gradient and sodium [18F]fluoride uptake were evaluated as a continuous variable (Spearman’s correlation coefficient). Correlation coefficients (R) of 0-0.19 were regarded as very weak, 0.2–0.39 as weak, 0.40–0.59 as moderate, 0.6–0.79 as strong and 0.8-1 as very strong. Agreements between sodium [18F]fluoride uptake in PET-CT and each PET-MRI reconstruction were assessed using mean bias, 95% limits of agreement, intraclass correlation coefficient (consistency and two-way random effects model) [17] and Bland-Altman plots [18]. Intraclass correlation coefficient values was described as poor when less than 0.5, moderate when 0.5–0.75, good when 0.75–0.9 and excellent when greater than 0.9 [17]. Mean sodium [18F]fluoride uptake values as quantified by AMA, total aortic SUV, aortic SUVmean, aortic SUVmax and blood pool were compared using a one-way ANOVA and post-hoc paired t-tests. Statistical significance was set as a two-sided p < 0.05.