Study population

A total of 71 consecutive patients with symptomatic severe AS, who qualified for TAVI, were prospectively enrolled in this single‑center, observational study from July 2017 to January 2020.^9,10 Our research protocol included a baseline visit (2 days before TAVI) and follow‑up clinical visits at 1, 6, 12, and 24 months, accompanied by transthoracic echocardiography (TTE) examinations (Figure 1). PET/CT and transesophageal echocardiography (TEE) were performed at the 6‑month visit, and initially planned for the final, 24‑month follow‑up visit; however, due to the COVID‑19 pandemic, some patients experienced delays. Consequently, a large proportion of clinical follow‑up with TTE was also abandoned in this high‑risk population. FDG‑labeled PET/CT examinations were delayed in patients diagnosed with thrombosis until valve morphology was normalized. The final PET/CT results were then compared with the initial PET/CT data.

Echocardiography

All patients underwent TTE, encompassing 2‑dimensional (2D), 3‑dimensional (3D), and Doppler imaging, at baseline and during follow‑up. Acquisition settings were meticulously optimized to ensure maximal image quality. To enhance visualization of the TAVI valve leaflets, 2D and 3D TEE were conducted at the 6‑month and final follow‑up visits. TTE and TEE examinations were performed using Vivid E9 and Vivid E95 devices (GE Healthcare, Waukesha, Wisconsin, United States). Postprocessing evaluations were carried out on a dedicated workstation (EchoPAC, GE Healthcare), with linear measurements taken with virtual calipers. TTE examinations and measurements adhered to the recommendations for assessing AS and TAVI valves.^11,12 TEE visual assessments of the aortic valve morphology and leaflet movements were performed. All measurements were independently validated by a core laboratory (KCRI, Kraków, Poland).

Positron emission tomography computed tomography

Electrocardiography (ECG)-gated and breathless PET/CT scans of the TAVI valves were performed using ¹⁸F‑fluorodeoxyglucose (¹⁸F‑FDG) and ¹⁸F‑sodium fluoride (¹⁸F‑NaF) with the GE Discovery 690 VCT scanner (GE Healthcare).^{9,10 18}F‑NaF and ¹⁸F‑FDG were administered intravenously at a dose of 4 MBq per kg of body mass (followed by a 60‑minute rest in a quiet environment), with acquisition performed as described above. An attenuation‑corrected CT scan (nonenhanced, ECG‑gated, low‑dose 120 kV with modulated rays [Smart mA] range 80 to 220 mAs) was performed, followed by PET imaging covering a 15 cm single‑bed position centered over the valve in 3D mode for 16 minutes. The PET data were reconstructed using the GE reconstruction algorithm (matrix size, 128 × 128, Vue Point FX: 24 subsets, 2 iterations with a 12 mm cutoff Cardiac 3D filter for static images; Vue Point FX: 24 subsets, 5 iterations with a 4.3 mm cutoff Cardiac 3D filter for gated images). Corrections were applied for attenuation, dead time, scatter, and random coincidences. Upon PET reconstruction, we obtained 8‑bin CT‑gated PET images. CardIQ Fusion PET software (GE Healthcare) was used for the image analysis.

All image analyses were performed on fused PET/CT datasets. For both ¹⁸F‑FDG PET/CT and ¹⁸F‑NaF PET/CT scans, measurements were performed as described below (Figure 2A–2D). ECG‑gated PET/CT scans were used to increase reliability of the results, with measurements taken during diastole, when the valve leaflets had the highest chance of being closed. 3D multiplanar reconstruction mode was used to visualize the valve in the coaxial short‑axis view, for more precise delineation of the regions of interest (ROI) around the valve. In all cases, the mediastinal blood pool subject (MBPS) in the right atrium (RA), maximal standardized uptake value (SUV), mean SUV, and mean density in Hounsfield units were measured. The initial measurements were performed at the level of the closed leaflets (informed by CT stent visualization and TAVI valve manufacturers). The ROIs (according to the stent shape) inside and outside of the stent were drawn, followed by the ROIs on 2 adjacent slices (up and down) to increase the chance of the entire valve examination. The term “inner area” is used to define the area inside the valve (including its leaflets but excluding the valve stent and the aortic annulus), and the outer area contains the valve leaflets, the aortic annulus, and the native leaflets (Figure 2A–2D). Mean and maximal SUVs were calculated for each slice, the inner and outer region, and the mean value was determined for all levels of the valve as well as for the inner and outer ROIs. However, SUV measurements in vascular structures are influenced by intraexamination variations of ¹⁸F‑FDG and ¹⁸F‑NaF activity in the blood pool. Therefore, subtraction SUV values of the ROIs, as well as background tracer uptake in MBPS and blood pool in the RA were measured and analyzed according to the suggestions from previously published papers.¹³ The gathered data included mean and maximum tissue‑to‑background ratios (TBRs).¹³ These ratios help minimize the influence of individual SUVs (SUV_max and SUV_mean) on the results.

The study received ethics approval from the Institutional Review Board of the Jagiellonian University Medical College (1072.6120.90.2017), and written informed consent was obtained from all participants. The study protocol followed the ethical standards set out in the 1975 Declaration of Helsinki.

Statistical analysis

Categorical variables are presented as numbers and percentages, while continuous variables are expressed as either mean (SD) or median (interquartile range [IQR]), as appropriate. The Shapiro–Wilk test was used to assess normality, and the Levene test was applied to evaluate the equality of variances. For the normally distributed variables, differences between the baseline and follow‑up parameters were compared using the mixed effect models. The Pearson χ² test or the Fisher exact test was used to compare the categorical variables, with the latter applied if 20% of cells had an expected count of less than 5 (the Monte Carlo simulation was utilized for the Fisher test in the cases with dimensions exceeding 2 × 2 Tables). Intraclass correlation coefficients were computed to evaluate the intraobserver variability in the analysis of ¹⁸F‑FDG and ¹⁸F‑NaF activity. Two‑sided P values below 0.05 were considered significant. All analyses were performed using R 4.2.3 software (R Foundation for Statistical Computing, Vienna, Austria), with “lme4” package, version 1.1–30 and “emmeans” package, version 1.8.2.

General characteristics

We recruited 71 patients with TAVI valves, and the final ¹⁸F‑NaF and ¹⁸F‑FDG PET/CT scans were performed for 31 of them. All the patients were followed up with clinical visits and TTE examinations (Figure 1). Due to the COVID‑19 pandemic‑related constraints, the median (IQR) time interval from the baseline PET scan, conducted 6 months post‑TAVI, to the follow‑up PET/CT scan was 18.1 (15.5–21.8) months for ¹⁸F‑NaF, and 18.5 (15–26.7) months for ¹⁸F‑FDG. Baseline clinical characteristics of the patients are presented in Table 1, and ECG data can be found in Table 2. In our study group, we identified 5 types of aortic valves used during TAVI: 71% of the patients had the Evolut R valve (Medtronic, Minneapolis, Minnesota, United States), 16.1% had the Accurate Neo valve (Boston Scientific SciMed Inc., Maple Grove, Minnesota, United States), 6.4% had the SAPIEN 3 valve (Edwards Lifesciences, Irvine, California, United States), 6.4% had the Evolut PRO valve (Medtronic), and 12.8% had the Portico valve (St. Jude Medical, Minneapolis, Minnesota, United States). We observed a 9.6% mortality rate at 1 month, 13.7% at 6 months, and 45% at 68 months. Eight patients refused to attend the follow‑up visits due to the fear of contracting COVID‑19. Asymptomatic valve thrombosis was diagnosed at 6 months in 8.4% of patients (6 cases), and the baseline PET/CT examinations were postponed until the thrombotic lesions resolved. Control PET/CT were ultimately performed in only 3 out of 6 cases. Seven patients declined the 6‑month on‑site follow‑up visit due to the COVID‑19 pandemic.

Echocardiography

We observed a decrease in both maximal and mean transvalvular gradients, as well as an increase in the effective orifice area and ejection fraction after TAVI. No changes in the valvular parameters were observed at 6, 12, and 24‑month follow‑up (Table 2). At 6 months, TEE revealed thickening and a slight reduction of the leaflet motion in 6 patients; however, anticoagulant therapy resolved these abnormalities. Thrombosis of the TAVI valve was noticed in 1 patient with the Accurate Neo, 2 patients with the Edwards SAPIEN 3 valve, and 1 patient with the Evolut R valve at 6 months. PET/CT examination was postponed until the valve parameters normalized. Similarly, ECG parameters assessed after the resolution of thrombotic features were included in the analysis.

¹⁸F‑sodium fluoride and ¹⁸F‑fluorodeoxyglucose positron emission tomography / computed tomography

The reliability analysis demonstrated an excellent agreement between the observers for both mean and maximal ¹⁸F‑FDG and mean and maximal ¹⁸F‑NaF activity, with the intraclass correlation coefficients ranging from 0.986 to 0.993. Analysis of microinflammation assessed by ¹⁸F‑FDG uptake showed similar or increasing values on follow‑up PET/CT (Tables 3 and 4). Differences between absolute SUVs were observed only for SUV_mean, while SUV_max remained similar on both PET/CTs. The microinflammatory process was comparable in the inner areas of the leaflets of the bottom valve segments and middle valves segments (Table 3). An increase in absolute SUV_mean of ¹⁸F‑FDG uptake was observed for the outer area in the middle segments (Table 3). The averaged absolute SUV_max and SUV_mean for the TAVI valves counted at baseline and follow‑up PET/CT did not show differences; however, the analysis of averaged corrected SUV_max values : SUV_max TBR‑RA showed differences between the baseline and follow‑up PET/CT. This relationship was not observed for the averaged corrected SUV_mean (Table 4).

The analysis of microcalcification evaluated by ¹⁸F‑NaF PET/CT showed no differences between the absolute SUV_max or SUV_mean of the inner and outer areas, independently of the mapping level (Table 3). The averaged absolute SUV_max and SUV_mean of the ¹⁸F‑NaF uptake for the TAVI valves counted at baseline and follow‑up PET/CT did not show differences. In addition, there were no differences between baseline and follow‑up PET/CTs, when comparing the averaged values of corrected ¹⁸F‑NaF uptake in the TAVI valves (Table 4).

Conclusions

In conclusion, our study identified an increased uptake of ¹⁸F‑FDG, potentially indicative of local inflammation that may precede the onset of calcification as detectable by ¹⁸F‑NaF uptake. However, the significance of this phenomenon and its connection with valve degeneration remains to be definitively established. These findings underscore the potential of PET/CT imaging as a valuable tool for monitoring local inflammation, which could represent an early stage of the degenerative process in TAVI valves, and for guiding future research toward development of novel therapies to impede the progression of AS. However, given the limitations of the imaging method described and the size of the study group, additional studies involving larger patient populations and extended follow‑up periods are necessary to confirm our findings and ascertain the clinical implications of PET/CT imaging in assessing the TAVI valve degeneration.