Quantitative aortography for assessment of aortic regurgitation in the era of percutaneous aortic valve replacement

Paravalvular leak (PVL) is a shortcoming that can erode the clinical benefits of transcatheter valve replacement (TAVR) and therefore a readily applicable method (aortography) to quantitate PVL objectively and accurately in the interventional suite is appealing to all operators. The ratio between the areas of the time-density curves in the aorta and left ventricular outflow tract (LVOT-AR) defines the regurgitation fraction (RF). This technique has been validated in a mock circulation; a single injection in diastole was further tested in porcine and ovine models. In the clinical setting, LVOT-AR was compared with trans-thoracic and trans-oesophageal echocardiography and cardiac magnetic resonance imaging. LVOT-AR > 17% discriminates mild from moderate aortic regurgitation on echocardiography and confers a poor prognosis in multiple registries, and justifies balloon post-dilatation. The LVOT-AR differentiates the individual performances of many old and novel devices and is being used in ongoing randomized trials and registries.

Paravalvular leak (PVL) is a shortcoming that can erode the clinical benefits of transcatheter valve replacement (TAVR) and therefore a readily applicable method (aortography) to quantitate PVL objectively and accurately in the interventional suite is appealing to all operators. The ratio between the areas of the time-density curves in the aorta and left ventricular outflow tract (LVOT-AR) defines the regurgitation fraction (RF). This technique has been validated in a mock circulation; a single injection in diastole was further tested in porcine and ovine models. In the clinical setting, LVOT-AR was compared with transthoracic and trans-oesophageal echocardiography and cardiac magnetic resonance imaging. LVOT-AR > 17% discriminates mild from moderate aortic regurgitation on echocardiography and confers a poor prognosis in multiple registries, and justifies balloon post-dilatation. The LVOT-AR differentiates the individual performances of many old and novel devices and is being used in ongoing randomized trials and registries. KEYWORDS aortic regurgitation, paravalvular leak, videodensitometry, transcatheter aortic valve replacement, transcatheter aortic valve implantation, quantitative aortography Highlights • Aortic regurgitation following TAVR negatively affects patient outcomes. • Video-densitometry is an objective, accurate, and well-validated tool for aortic regurgitation adjudication. • LVOT-AR > 17% has a poor prognosis and justified further intervention. • Many TAVR devices have been evaluated using videodensitometry, and the technique is currently being used in ongoing trials and registries.
Quantitative aortography for regurgitation assessment after TAVR: an unmet need The clinical indications, vascular approaches, and technologies of transcatheter aortic valve replacement (TAVR) are continuously evolving, and as its use increases exponentially, the abrogation of major procedural shortcomings becomes imperative. Among the imperfect results, paravalvular leak (PVL) is one that contributes considerably to the erosion of clinical benefits with its adverse effects on mortality, morbidity, and reverse cardiac remodeling. Consequently, a simple method, readily applicable in the interventional suite to timely detect and quantitate PVL objectively, reproducibly, and accurately remains the wish and goal of the "minimalist" operator.
Trans-thoracic echocardiography (TTE) is a safe and convenient tool that provides real-time information on the severity of trans-or para-valvular aortic regurgitation (AR) and is thus the preferred method of assessing and monitoring PVL. However, TTE is not exempt from methodological drawbacks such as the significant influence of the imaging plane and a lack of consistency in the severity of PVL described between core labs and between imaging modalities (Supplementary Figure S1) (1).
In the era of minimalist TAVR the use of angiography has become more common as it provides PVL grading as the sum of all jets regardless of their number, level, or trajectory, furthermore as it is performed immediately post implant, any significant PVL can be corrected by device post-dilatation, repositioning, or implantation of a second valve. Interventional cardiologists are familiar with this technique, which is fast, readily available, and part of their technical environment, however, angiographic grading is qualitative and subjective. These issues highlighted the need for a quantitative assessment of PVL from aortography. In this comprehensive narrative review, we delve into the principles and history of video-densitometric assessment of aortic regurgitation, discussing its in vitro and in vivo validation, correlation with other imaging modalities, prognostic value, current limitations, and potential for future development. To find relevant articles for this narrative review, we searched MEDLINE and Embase for articles in English language with the following terms: "Paravalvular leak", "PVL", "aortic regurgitation", "AR", "Transcatheter aortic valve implantation", "TAVI", "transcatheter aortic valve replacement", "TAVR", "video-densitometry", "aortography", and "Quantitative Aortography".

Principle and history of video-densitometry
In the eighties, long before the TAVR era, aortography was dethroned by echocardiography for the iterative and non-invasive assessment of AR, however, aortography was actually already "quantitative" at the time, and several attempts had been made Abdelshafy et al. 10.3389/fcvm.2023.1161779 Frontiers in Cardiovascular Medicine to make the assessment of AR by aortography more objective, less categorical and more numerically quantitative (2)(3)(4)(5). At that time, video-densitometric analysis was used for the evaluation of native AR; moreover, it had also been used for the assessment of AR after balloon aortic valvuloplasty (Supplementary Figure S2) (2).

The concept of video-densitometry and phantom validation
The most common objection raised by clinicians regarding the methodology of video-densitometry is the potential for measurement errors due to an assumed parallax effect when the outflow tract is assessed from various angiographic views. However, the approach is aimed at circumventing this issue; a video-densitometric measurement is the constant product of an area in which the radiopaque contrast is detected, multiplied by the density (hemodilution) of the radiopaque contrast medium in that area/volume (6). This means that the x-ray attenuation of an object is virtually unaffected by the viewing direction (angulation) of the x-ray system. As illustrated in Figure 1, the error in the video-densitometric measurement, as demonstrated using an egg-shaped plastic phantom filled with angiographic medium and rotated in the radiation field, is <2.5% (maximal video-densitometric error) even at extreme parallax angles. Notably, this does not compromise the accuracy in assessing regurgitation values in clinical practice, which range from >6% for mild regurgitation to >17% for moderate/severe regurgitation.
In 2014, Schultz et al. (7) used video-densitometry and pioneered the concept of quantitative Seller's assessment for regurgitation (qRA index) by quantifying the density, duration, and depth of LV opacification. Ultimately this yields a continuous severity scale of regurgitation ranging from grade 0, which indicates no contrast leakage into the LV, to grade 4 that indicates contrast filling the entire left ventricle (LV), with the density of angiographic contrast medium higher in the LV than the aortic root.
However, the criteria for analyzability were only met in 15% (n = 44) of 285 retrospective aortograms and in 69% (n = 22) of 32 prospectively collected aortograms. Nevertheless, the visual grade of AR (Sellers) was highly correlated with the time-density measurements including the Relative Area Under the Curve (RAUC) and qRA index (r = 0.81 and 0.83, respectively, p < 0.001).
In 2016 Tateishi et al. (8) proposed changing the region of interest (ROI) as interrogation of the entire LV was not always feasible. Indeed, the apex was sometimes not visualized, or there were overlapping structures in LV ROI which influenced the video-densitometric analysis such as a contrast in the descending aorta, radiopaque objects such as the trans-esophageal echocardiography (TEE) probe, a high position of the diaphragm, and gastric or bowel gas etc. To avoid the interference of such radiopaque structures, the left ventricle outflow tract (LVOT) was selected as an alternative ROI.
Briefly, the software (CAAS A-valve, Pie Medical Imaging, Maastricht, the Netherlands) constructs two time-density curves assessed in two ROI: the aortic root where contrast is injected and the LVOT where the regurgitation is quantified. To define the LVOT area a line is drawn from the valve plane to the apex, subdivided in three equal regions; the upper region is the ROI for the video-densitometric assessment of regurgitation in the LVOT. The ratio between the areas under the two-time density Video-densitometric assessment of an egg-shaped plastic phantom filled with radiopaque contrast medium rotated in an x-ray field from short (0, 180 degree) to long axis view (90 degree). Frontiers in Cardiovascular Medicine curves of these regions over at least three cardiac cycles is the Regurgitation Fraction expressed in percentage. This new approach was called "LVOT-AR"-left ventricular outflow tract-aortic regurgitation ( Figure 2) -and its feasibility and reproducibility was investigated in 182 aortograms taken as part of the Brazilian TAVR registry. LVOT-AR was analyzable in 64.8% of aortograms vs. 29.7%, when using the entire LV gram and qRA index.

In vitro and vivo validation
Quantitative assessment of PVL by video-densitometry of aortograms correlates strongly with the actual regurgitation fraction (RF) as measured by a flow probe (Transonic 28PAU, with TS410 flowmeter; Transonic, Ithaca, NY, USA) in a controlled in vitro mock circulation model (9) and in a porcine model (10)

Using of synchronized injection to minimize the amount of contrast
One of the main limitations of aortography in TAVR patients -who commonly have some degree of kidney dysfunction-is the large amount of contrast medium injected during a conventional fluoroscopic acquisition, which may subsequently contribute to the risk of periprocedural acute kidney injury (AKI) portending a poor outcome after TAVR (11).
In clinical practice, aortography is performed with a contrast injection (15-25 ml) which typically lasts for 1-2 s, covering a few cardiac cycles. Given the fact that PVL is a diastolic event, a short injection during only diastole could avoid the wash-out of contrast by the stroke volume ejected into the aorta during systole (when the aortic valve is open) and, consequently, reduce the contrast volume required per acquisition down to 8 ml (12). This concept was tested in vitro by Miyazaki    Correlation of LVOT AR with clinical trans thoracic echocardiographic assessment post TAVR In 2017, LVOT-AR was quantitated in 228 consecutive patients enrolled in the TAVR Brazilian registry before, and after, TAVR by echocardiography, and by video-densitometric analysis of aortograms after TAVR (13). Post TAVR LVOT-ARs of 10.6 ± 8%, 13.6 ± 10% and 28.6 ± 14% were respectively measured in none-trace, mild and moderate-severe post-TAVR AR as defined by echocardiography (p < 0.001). An LVOT-AR of 17%corresponding to the Youden index on the area under the curve (AUC = 0.84, sensitivity 81% and specificity 72%), is to the best cut off criteria discriminating trace-mild AR from moderatesevere AR as diagnosed on TTE Figures 3A,B.
Based on previous validation, the following cut of points were used: LVOT-AR > 17% for moderate/severe AR, 6%≤ LVOT-AR ≤17% for mild AR and LVOT-AR < 6% representing the normal closing volume of the aortic valve.

Regurgitation fraction of 17% as the threshold criteria for post dilatation
In 2018 the Brazilian TAVR registry reported a significant reduction in LVOT-AR from 24.0 (18.0-30.5) % to 12.0 (5.5-19.0) % after balloon post-dilatation (BPD) in 61 patients with some degree of PVL (p < 0.001), among their cohort of 399 patients who had undergone TAVR. The relative delta of LVOT-AR after BPD ranged from −100% (improvement) to +40% (deterioration) and its median value was −46.2%. The frequency of improvement, no change, and deterioration were 70% (n = 43), 25% (n = 15) and 5% (n = 3), respectively. Significant AR (LVOT-AR > 17%) was observed in 47 patients (77%) before and in 19 patients (31%) after BPD. These results confirmed that LVOT-AR after TAVR provides a quantitative assessment of post-TAVR regurgitation and can help in deciding whether to perform a BPD and determine its efficacy Figure 4. Indeed, online videodensitometric assessment of AR in the cath lab has been proven feasible in over 92% of cases through multiple studies described in the following section (17,18). However, whether BPD, according to intra-procedural assessments of LVOT-AR > 17% will provide a better outcome for patients remains to be proven with prospective clinical trials. Indeed, the OVAL GUIDE trials, as described in detail in the following sections, are designed to fill this evidence gap.

Regurgitation fraction of 17% as a vital prognostic determinant
In 2016 Tateishi et al. (8) established the prognostic impact of an LVOT-AR > 17% when he reported that these patients had a significantly higher one-year all-cause mortality compared with those with LVOT-AR ≤17% (37.1% vs. 11.2%, p = 0.0008, Central Illustration A). The poor prognosis in this  , p = 0.14) and the prevalence of LVH (from74 to 87%, p = 0.23) evolved in the opposite direction and failed to show any significant change (13).
Regarding the long-term prognostic impact of mild PVL,assessed by video-densitometry (LVOT-AR)-, on the prediction of 5-year mortality, a preliminary analysis shows diverging Kaplan-Meyer curves and suggest a potential mortality discrimination between none/trace (LVOT-AR 0%-2%), mild (LVOT-AR 3%-16%) and moderate/severe (LVOT-AR >16%) regurgitation (statistically not significant p = 0.132) Central Illustration D. Of note the cut-off criteria of videodensitometric regurgitation vs. categorical echocardiographic assessment of none/trace, mild, and moderate/severe regurgitation has been specifically determined in a blind fashion (CORRIB lab) by C statistic and AUC in a cohort of patients studied at Bad Krozingen (Courtesy of Prof Neuman and dr Schoechlin) (19).

Feasibility of analysis, protocol of acquisition and yield of data
LVOT-AR by video-densitometry has been evaluated in comparison with TTE, TEE and CMR using a variety of THV devices. However, due to the retrospective nature of these analyses and the absence of any specific acquisition protocols, analyses were only feasible in 57.1% and 57.5% of the patients enrolled in the respective Brazilian TAVR Registry (8,20) and the RESPOND study (14), leaving a large proportion of patients unaccounted for.
The major issue impacting on the feasibility of retrospective LVOT-AR assessment is the overlap of the descending aorta with either the ROI (LVOT) or the reference area (aortic root). The software analyses the density changes over time in both regions, thus any other change in density in the background of the RX acquisition due to contrast passing through the descending aorta and overlapping with these regions impacts on the final result. Although the analyses are still technically possible in these situations, the results would not reflect the true regurgitation; therefore, they are considered non-analysable.
Methods to improve the feasibility of video-densitometry include using pre-procedural multi-slice computed tomography (MSCT) to optimize the angiographic projection, and this was first tested at the Yamaguchi University along with other directives to optimise the angiographic acquisition protocol for video-densitometry Table 1 and Figures 5, 6.
In 92 consecutive patients from the Yamagishi University, post-TAVR LVOT-AR was assessed in two sequential cohorts investigated either prior to (n = 54) or following (n = 38) implementation of a standardized acquisition protocol. The protocol involved using MSCT for planning the optimal angiographic projection, and its use led to the feasibility of video-densitometry improving from 57.4% to 100% Figure 6. In

Image acquisition environment
With ECG recording Image acquisition should take ≥2 beats before contrast injection for subtracting the static background Image acquisition should take ≥3-5 heartbeats after contrast injection Frontiers in Cardiovascular Medicine 69 analysable aortograms (69/92; 75%), LVOT-AR ranged from 3% to 28% (median 12%). Inter-observer agreement was high (mean difference ± SD, 1 ± 2%) with the two observers' measurements highly correlated (r = 0.94, p < 0.0001) (21). These promising results paved the way for the ASSESS-REGURGE (NCT03644784) registry, which was a multicontinental trial conducted in four centres in Asia (Japan), North America (Canada), and Europe (the Netherlands and Germany) enrolling 354 consecutive patients with Heart Team consensus in favour of TAVR over a median period of 12 months. In this registry, operators performed the final aortogram according to an angiographic projection which was pre-planned by CT (3mensio or Circle) or visually by fluoroscopy (Teng's rule) (22). An independent core laboratory analysed all images for feasibility and assessed the regurgitation. The acquisition protocol was followed in all 354 patients and all aortograms were analysed by the core lab, with analysis feasible in 95.5% [95% confidence interval (CI): 93.2% to 97.5%] of cases. No differences were observed among the different planning strategies (CT 96.5% vs. Teng's rule 93%, p = 0.159: or Circle 98.5% vs. 3mensio 95.8% vs. Teng's rule 93%, p = 0.247) (18) Figure 7.

Cooperation with patients and other staff
The next step was to conduct the OVAL trial (Online Videodensitometric Assessment of Aortic Regurgitation in the Cath-Lab; NCT04047082) to determine the feasibility of online assessment of regurgitation (percentage of analyzable cases) and the reproducibility of results between online and offline analysis by the core laboratory. One hundred consecutive patients with aortic stenosis and indications to undergo TAVR were enrolled. The planning of optimal views for video-densitometry was determined with the use of Heart Navigator software (Philips Healthcare, Best, the Netherlands) Supplementary Figure S8. All final aortograms were analyzed with online software in the catheterization laboratory and were also forwarded to an independent core laboratory for blinded offline assessment. The feasibilities of online and offline analysis by the core laboratory were identical (92%; 95% CI:86%-97%, Figure 8). Reproducibility assessment showed a high correlation between online and core laboratory evaluations (R 2 0.87, p < 0.001), with an intraclass correlation coefficient of 0.962 (95% CI: 0.942 to 0.975; p < 0.001) (17).

Comparison of aortic regurgitation following different TAVR devices
Taking advantage of our large network, we performed a multicentre retrospective corelab pooled analysis of aortograms  Figure 9. The results showed that the proportion of patients with moderate or severe regurgitation followed the same ranking order as for RF as a continuous variable. The rates of LVOT-AR > 17% are shown in Figure 9A with the lowest rate 1.7% (ACURATE neo2), and the highest rate 30.1% (CoreValve, χ 2 p < 0.001). Figure 9B shows the mean LVOT-AR for each valve with the lowest mean AR 3.5 ± 4.4% (Lotus) and the highest 13.7 ± 10.7% (CoreValve, ANOVA p < 0.001).
Post-hoc two-by-two testing showed that the Lotus valve had a significantly lower RF compared with each of the other valves except for the ACURATE neo2, which had a significantly lower RF compared with SAPIEN 3, Evolut R, SAPIEN XT, Venus-A, ACURATE neo, and CoreValve. Similarly, the first generation CoreValve had a significantly higher RF than all the other valves.
Apart from the CoreValve, Lotus, and ACURATE neo2, no other valve differed in the amount of RF from each other.
These result highlight that the Lotus valve has the lowest average RF and the ACURATE neo2 the lowest percentage of moderate/severe regurgitation, with the latter device performing significantly better than the ACURATE neo. Myval, VitaFlow and Venus-A are promising options in the THV armamentarium. In our latest report of 103 patients receiving the new Myval Octacor THV, moderate AR was basically eliminated in patients with tricuspid valve morphology (28). Of note, although the incidence of moderate/severe AR has regressed with contemporary THV, the incidence of mild AR remains noticeable, and assessment of its long-term prognosis still requires further investigation to establish whether it has a benign or malignant outcome and whether further refinements (e.g., outside skirt expanding with moisture) of THV will be able to eliminate para valvular regurgitation.

Video-densitometry limitations
Video-densitometric angiography is, however, not without limitations. Various radio-opaque objects intrathoracic or extracorporeal (guidewires, catheters, pacemaker lead, surgical wires, TEE probe, mitral annular calcification, and electrocardiogram electrodes) may interfere with the videodensitometric background of ROI's. Whenever possible the radio-opaque structures not permanently implanted should be removed prior to aortography.
Video-densitometric angiography is also sensitive to pulmonary diaphragm/patient/table motion. If the standardized protocol of acquisition avoiding the over-projection of the ascending and descending aorta over LVOT has dramatically improved the feasibility of analysis (see above), nevertheless it remains that the volume, rate, and time of injection could be further optimized and regulated by single diastolic injection triggered by ECG, thereby reducing the injected amount of contrast medium to a minimum of 8 ml (10,12). The lack of randomized head-to-head trials comparing video-densitometric angiography and traditional AR assessment is currently a limitation of video-densitometry. However, multiple prospective clinical trials are ongoing to fill the current evidence gap. For example, the multicentre, randomized LANDMARK trial (NCT04275726) will compare the post-TAVR AR of Meril's Myval transcatheter heart valve (THV) series vs. contemporary valve with Video-densitometry assessment and echocardiography.
Ongoing and future designed trials using video-densitometry for AR assessment Although still absent on European and American practice guidelines of TAVR, Quantitative aortography with videodensitometric assessment is increasingly recognized as a valid method for aortic regurgitation assessment. Indeed, videodensitometry was mentioned in the VARC III document (30) as an acceptable technique for the evaluation of aortic regurgitation post-TAVR and as such, has been used in ongoing registries and randomized trials ( Table 2), e.g., LANDMARK trial (NCT04275726) and Compare TAVR trial (NCT04443023), as well as in the ongoing registries evaluating the ACURATE neo2 (NCT04810195) (24). . Currently, the Pause-VD trial ( Table 2) investigates the reproducibility in the Cath lab of 2 sequential LVOT-AR performed at a time interval of 10-15 min to investigate the reproducibility shortly after the procedure.
The OVAL GUIDE trials are prospective, multicentre, observational, investigator-initiated studies, aimed at intraprocedural TAVR guidance using online video-densitometric angiographic AR quantification for determining whether a PVL requires further corrective measures. Design of both trial is described in the Supplementary Data Sheet and illustrated in Supplementary Figures S11, S12.

Future developments Mitral video-densitometry
Angiographic assessment of mitral regurgitation fraction (MRF) remains at best semi-quantitative (Sellers's method) and operator dependent. Recently an attempt was made to adapt the video-densitometric methodology to the mitral space. In vitro MRF and MRV were assessed in a mock circulation with transonic flow measurement at a cardiac output comparable as the one measured in vivo by thermodilution Supplementary Figure S13C (6).
The in vitro and in vivo MRF, MRV, and interobserver reproducibility for QMR analysis strongly correlated. There were also very strong correlations of in vivo MRF between 2 independent analysts, Supplementary Figure S13D. It was concluded that in vivo MRF using the novel software is feasible, accurate, and highly reproducible. These promising results have led us to initiate the first human feasibility study comprising patients undergoing percutaneous mitral valve edge-to-edge repair (6) or mitral valve replacement Figure 11.

Three-dimensional video-densitometry
One of the limitations of two-dimensional video-densitometry is the lack of three-dimensional localization of the regurgitation jet. Development of three-dimensional video-densitometric analysis from biplane aortography is ongoing (Supplementary Figure S14).

Video-densitometry and deep learning
Automatic identification of the aortic root and the LVOT during video-densitometric analysis by applying deep learning algorithms is considered the next step towards a seamless integration of quantitative video-densitometric assessment of aortic regurgitation into a minimalist TAVR approach.

Conclusions
Quantitative video-densitometric assessment of paravalvular leak by aortography is an objective, accurate, and reproducible tool which has been extensively vetted and validated in-vitro, invivo, and in clinical settings, for the assessment of AR following TAVR. Many TAVR devices have been evaluated using videodensitometry, and the technique is currently being used in ongoing trials and registries. Three-dimensional videodensitometric analysis for precise spatial localization of paravalvular jets might be the next frontier.

Author contributions
MA and PS contributed equally to the creation of this review. TT, PCR and SC improved the quality of the manuscript and submitted the article. The remaining authors were all major contributors to the development of videodensitometry technology and helped produce various parts of the review, and supervised the writing of the review. All authors contributed to the article and approved the submitted version.

Conflicts of interest
PWS reports personal fees from Philips/Volcano, SMT, Novartis, Xeltis, Merillife. PCR has no conflict of interest. Aben is an employee of Pie Medical Imaging. MC is employed by and holds shares of Xeltis BV CS is an employee of Philips Healthcare. MA-W reports other from Medtronic, other from Boston Scientific, outside the submitted work. NP is a consultant to Medtronic, Peijia, and MicroPort. RM is an employee of Example of QMR analysis post mitral valve replacement with highlife valve.

FIGURE 10
Example of LVOT-AR analysis pre and post TAVR using jenaValve.

Frontiers in Cardiovascular Medicine
Boston Scientific. AR reports grants and personal fees from Boston Scientific, personal fees from Edwards Lifesciences, outside the submitted work. KK reports personal fees from Abbott, personal fees from Boston Scientific, personal fees from Edwards Lifesciences, personal fees from Medtronic, personal fees from Meril Life Sciences, outside the submitted work. NVR reports grants and personal fees from Abbott, grants from Philips, grants from Biotronik, and personal fees from Microport, outside the submitted work. Rudolph is a proctor, speaker's honoraria of JenaValve. FJN reports personal fees from Amgen, personal fees from Boehringer Ingelheim, personal fees from Daiichi Sankyo, grants and personal fees from Pfizer, grants and personal fees from Biotronic, grants and personal fees from Edwards Lifesciences, grants from Medtronic, grants and personal fees from Bayer Healthcare, personal fees from Novartis, grants from GlaxoSmithKline, grants and personal fees from Boston Scientific, personal fees from Ferrer, outside the submitted work. IJA-S is a proctor for Medtronic, Boston Scientific and Meril Life. OS and YO report several institutional research grants, outside the submitted work.
The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.